Composites in Civil Engineering

International Institute for FRP in Construction (IIFC)

CICE 2006
December 13-15, 2006
Miami, Florida, USA

Editors:
Amir Mirmiran, Florida International University
and Antonio Nanni, University of Miami
 Copyright © 2006
To contact the editors:
Amir Mirmiran
Department of Civil and Environmental Engineering
Florida International University
10555 West Flagler Street, Suite 3600
Miami, Florida 33174
Tel (305) 348-3055
Fax (305) 348-2802
E-mail: mirmiran@fiu.edu

All rights reserved, including rights of reproduction and use in any form or by any means,
including making of copies by any photo process, or by any electronic or mechanical device,
printed or written or oral, or recording for sound or visual reproduction or for use in any
knowledge or retrieval system or device, unless permission in writing is obtained from the editors.

The papers in this volume have been peer reviewed by individuals who are expert in the subject
areas of the papers. However, the statements or opinions expressed in each paper are solely
those of the individual authors. The editors are not liable for any loss or damage caused by any
potential error or omission in each paper, whether such error or omission is the result of
negligence or any other cause. Any and all such liability is disclaimed herein.

The proceedings includes a copy protected CD attached on the back cover. The CD contains the
conference papers. Adobe Acrobat Reader® is needed. No other special installation is required.
More detailed information is available on the CD as readme text file.

Printed in the United States of America

First Printing: November 2006

ISBN: 0-615-13586-2
 Preface

Applications of fiber reinforced polymer (FRP) composites in civil engineering has

increased significantly in recent years, both for the strengthening of existing
structures and for new construction. The Third International Conference on
Composites in Civil Engineering (CICE 2006) is the official conference of the
International Institute for FRP in Construction (IIFC). The aim of this conference
is to provide an international forum for all concerned with the application of FRP
composites in civil engineering to exchange recent advances in both research
and practice. The first of this series of conferences was held in Hong Kong in
2001, followed by one in Australia in 2004.

This conference is also sponsored by the American Concrete Institute (ACI), the
American Society of Civil Engineers (ASCE) and its Structural Engineering
Institute (SEI), the Canadian Society for Civil Engineers (CSCE), and the
Intelligent Sensing for Innovative Structures (ISIS) Canada Research Network.

A total of 179 papers are included in this volume, covering a range of topics
including bond and development, bridge applications, bridge decks, composite
and hybrid systems, confinement issues, creep and sustained loads, fatigue and
durability issues, design guides, FRP reinforcing bars, health assessment,
masonry structures, modeling, novel applications, prestressing applications,
repair of columns, repair techniques, retrofit of slabs, seismic applications, shear
retrofit, and strengthening of concrete and metal structures. The technical papers
in this volume not only address analytical and experimental work, but also cover
field applications, and design and construction guidelines.

The papers in this volume are authored by experts in the field from 28 different
countries around the world, including Australia, Belgium, Brazil, Canada, Chile,
China, Czech Republic, Denmark, France, Germany, Hong Kong, Iran, Ireland,
Israel, Italy, Japan, Malaysia, the Netherlands, Oman, Poland, Portugal, Saudi
Arabia, South Korea, Sweden, Switzerland, Turkey, United Kingdom, and the
US.

This volume of valuable reference articles would not have been possible without
the help, dedication, and collaboration of numerous individuals. First and
foremost, we thank the authors for meeting our various submittal deadlines,
allowing this document to represent the most current work on the subject from
around the world. We would also like to thank the Organizing Committee and the
International Advisory Committee of the conference for their relentless efforts in
making this conference a reality. It is also important to thank the Executive
Committee and the Council of the International Institute for FRP in Construction
(IIFC) for their support throughout this process.

The conference is co-organized by the Florida International University and the

University of Miami. Sincere thanks to Ms. Dora Hernandez, Ms. Lilia Silverio,
Ms. Rui Zheng, Mr. Bin Li, and Mr. Yilei Shi for their assistance with the various
aspects of the daunting task of putting together this conference and the
proceedings.

The Editors
Amir Mirmiran, Florida International University
and Antonio Nanni, University of Miami

Miami, Florida
December 2006
 Conference Organization

Conference Chairs

Amir Mirmiran Florida International University Miami, Florida

Antonio Nanni University of Miami Miami, Florida

Organizing Committee

M. Arockiasamy Florida Atlantic University Boca Raton, Florida

S. Azhar Auburn University Auburn, Alabama
American Composites Manufacturers
J. Busel Arlington, Virginia
Association
R. El-Hacha University of Calgary Calgary, Canada
A. Fam Queen's University Kingston, Canada
H.R. Hamilton University of Florida Gainesville, Florida
R. Sen University of South Florida Tampa, Florida
Tallahassee,
M. Shahawy SDR Engineering, Inc.
Florida
Tallahassee,
L. Spainhour Florida State University
Florida
L. Zhao University of Central Florida Orlando, Florida
Z.Y. Zhu URS Corporation Miami, Florida
R.F. Zollo University of Miami Miami, Florida

International Advisory Committee

B. Bakht JMBT Structures Research Inc. Canada

C.E. Bakis Pennsylvania State University USA
L.C. Bank University of Wisconsin-Madison USA
N. Banthia University of British Columbia Canada
J.F. Chen University of Edinburgh UK
E. Cosenza Università Federico II Italy
M. Erki Royal Military College Canada
H. Fukuyama Building Research Institute Japan
K. Ghavami Pontificia universidade Catolica/PUC-Rio Brazil
I. Harik University of Kentucky USA
L. Hollaway University of Surrey UK
V.M. Karbhari University of California, San Diego USA
T. Keller Swiss Federal Institute of Technology Lausanne Switzerland
P. Labossière Université de Sherbrooke Canada
Z.T. Lu Southeast University China
A. Machida Hokkaido University Japan
U. Meier European Maritime Pilots´ Association Switzerland
M. Mottavalli University of Tehran Iran
A. Mufti University of Manitoba Canada
K. Neale Université de Sherbrooke Canada
D.J. Oehlers University of Adelaide Australia
S. Rizkalla North Carolina State University USA
S. Saiidi University of Nevada, Reno USA
L. Taerwe Ghent University Belgium
B. Taljsten Skanska Company Sweden
K.H. Tan National University of Singapore Singapore
J.G. Teng Hong Kong Polytechnic University China
T. Triantafillou University of Patras Greece
T. Ueda Hokkaido University Japan
G. Van Erp University of Southern Queensland Australia
P. Waldron University of Sheffield UK
Z.S. Wu Ibaraki University Japan
L.P. Ye Tsinghua University China
National Diagnosis and Rehabilitation of
Q.R. Yue China
Industrial Building Research Center
X.L. Zhao Monash University Australia
 TABLE OF CONTENTS

Preface
Conference Organization

Keynote Papers
The Prognosis of FRP Composites in Construction from a Worldwide Perspective 1
Len Hollaway

Future of FRP in Far EAST 11

Ueda Tamon

Part I. Bond and Development

Bond Tests on Concrete and Masonry Blocks Externally Bonded with FRP 17
Francesca Ceroni and Marisa Pecce

Cohesive-Bridging Zone Model of the FRP-Concrete Interface Debonding 21

Jialai Wang

Constitutive Model for Time Dependent Bonding and Debonding along FRP-
Concrete Interface 25
Hesham M. Diab and Zhishen Wu

Debond Characteristics of Carbon Fiber Laminates for Bridge Rehabilitation 29

M. Arockiasamy, M. Sivakumar and M. Shahawy

Debonding of CFRP Laminates Externally Bonded to Concrete Specimens at Low

and High Temperatures 35
Ernst Lucas Klamer, Dick Hordijk and Cees S. Kleinman

Effects of Bond Configurations on Flexural Response of RC Beams Externally

Strengthened with CFRP Sheets 39
Jiangguo Dai, Keisuke Sumiyoshi, Tamon Ueda, Hiroshi Yokota and Yasuhiko
Stao

Environmental Sensitivity and Defect Criticality in FRP Bond to Concrete Through

a Fracture Mechanics Approach 43
Rajiv Navada and Vistasp M. Karbhari

Improving the Bonding and Ductile Behavior of FRP Strengthened RC Beams 47

H. Fatemi, A. Davaran and J. Esmaeili

Intermediate Crack Debonding Analysis of Reinforced Concrete Beams Strengthened

with Externally Bonded FRP Plates 51
Luciano Ombres
Evaluation of FRP Bond Specifications for Development Lengths 55
Nathan Newman, Ashraf Ayoub and Abdeldjeltl Belarbi

Practical Use of Pull-off Strength Testing (ASTM D4541) for Assessing Quality of
FRP-to-concrete Bond 59
Kent A. Harries

Prediction of Bond Failure of Concrete Prisms Bonded with FRP Composites 63

H. Toutanji, P. Saxena and L. Zhao

Strength Model for Intermediate Crack Debonding in FRP-Strengthened Concrete

Members Considering Adjacent Crack Interaction 67
J. F. Chen, J. G. Teng and J. Yao

Study on Bond Characteristics of CFRP/Steel Double-Lap Shear Joints at Subzero

Temperature Exposure 71
Ahmed Al-Shawaf, Riadh Al-Mahaidi and Xiao-Ling Zhao

The Mechanism of Effect of Structural Span on Intermediate Crack-induced 75

Debonding
Hemdan Okasha Ahmed Said and Zhishen Wu

Strain Rate Effects on Strength of Unidirectional FRP Fabrics and Bond to Concrete 79
M. Saiid Saiidi, Rita Johnson and E. Manos Maragakis

Bond Behavior of CFRP Strengthened Full-Scale Prestressed Concrete Bridge

Girders 83
O. Rosenboom and S.Rizkalla

Bond Characteristics of GFRP Post-Installed Anchors 87

Ehab A. Ahmed, Ehab F. El-Salakawy and Brahim Benmokrane

Effect of Adhesive Modulus on the Monotonic and Fatigue Behavior of Externally

Bonded CFRP Strips 91
Kent A. Harries, Andrew Zorn and Benjamin Reeve

Experimental Study on the Local Bond Behavior of NSM-FRP Bars to Concrete 95

Donatella Galati and Laura De Lorenzis

Part II. Bridge Applications

A Case Study of Life Cycle Cost Based on a Real FRP Bridge 99
Itaru Nishizaki, Nobufumi Takeda, Yoshio Ishizuka and Takumi Shimomura

Structural Rehabilitation of an Off-System Bridge Using External Bonded CFRP

Laminates 103
Nestore Galati, Andrea Rizzo and Antonio Nanni
Enhancement of Shallow Depth Patches for Concrete Bridges using FRP Overlays 107
Golrokh Nossoni, Ronald S. Harichandran and Rigoberto Burgueño

Experimental Study on Prestressed Concrete Girders Strengthened with CFRP

Tapes at Different Bridge Repair Stages 111
Arkadiusz Mordak and Zbigniew Manko

Feasibility Study of Thermoplastic Wrap for Bridge Protection 115

Nasim Uddin, John D Purdue and Uday Vaidya

Field Tests of Prestressed Concrete Beams of Road Bridge Spans Strengthened by

CFRP Tapes 119
Zbigniew Manko and Arkadiusz Mordak

Blast Mitigation of A Concrete Arch Bridge Using Composites 123

Achille Devitofranceschi, Domenico Asprone, Andrea Prota, Renato Parretti
and Antonio Nanni

Performance of FRP Retrofitted Bridges under Blast Loading 127

Rui Zheng and Amir Mirmiran

Design and Analysis a 10-m FRP Deployable Bridge 131

R. G. Wight, M.A. Erki, C. T. Shyu, R. Tanovic and A. Xie

Response of No-Name Creek FRP Bridge to Static Loads, Moving and Impact
Traffic Loads 135
Eric Zhou, Youqi Wang, David Meggers and Jerry Plunkett

Part III. Bridge Decks

Behaviors of New Generation of FRP Bridge Deck with Outside Filament-wound
Reinforcement 139
Peng Feng and Lieping Ye

Tests of a FRP Deck Panel at Very Cold Temperatures 143

Z. John Ma and Usha Choppali

Static and Fatigue Testing of Innovative All-Composite Bridge Deck Slab 147
René J. Roy, Mathieu Robert, Brahim Benmokrane, Ahmed S. Debaiky and
Hossein Borazghi

Rapid Construction of Concrete Bridge Deck Using Prefabricated FRP

Reinforcement 151
Fabio Matta, Antonio Nanni, Thomas E. Ringelstetter and Lawrence C. Bank

Testing of an FRP Deck System for Moveable Bridges 155

Lei Zhao, Jignesh S. Vyas and Marcus Ansley
Part IV. Composites Systems
Urban Light Transport (ultra) Guideway Project 159
Thanongsak Imjai, Maurizio Gaudagnini, Kypros Pilakoutas and Peter Waldron

Composite Sandwich Wall Panels for Building Applications 163

Waleed Shawkat, Hart Honickman and Amir Fam

Behavior of Jute/Epoxy Composite Curved I-Beam under Bending Loading 167

Asad A. Khalid

Load Carrying Capacities of Pultruded FRP Sheet Piling 171

Yixin Shao and Fabien Darchis

A Case Study on Life-cycle Assessment of Environmental Aspect of FRP

Structures 175
Hirokazu Tanaka, Hitoshi Tazawa, Morio Kurita and Takumi Shimomura

Part V. Confinement Issues

Axial Behavior of Circular Columns Confined with Fiber-Reinforced Polymer
Jackets 179
Azadeh Parvin and Aditya S. Jamwal

CFRP Confined Ultra High Strength Concrete Columns 183

Adnan R. Malik and Stephen J. Foster

Strengthening of Short Circular RC Columns with FRP Jackets: A Design Proposal 187
T. Jiang and Jin-Guang Teng

Experimental Study on Concrete Columns Strengthened with CFRP Sheet

Confinement under Eccentric Loading 193
Shuangyin Cao, Denghu Jing and Ning Sun

Influence of Radius of Corners in Concrete Columns Confined with FRP Sheets 197
Giuseppe Campione, Nunzio Miraglia and Maurizio Papia

Part VI. Creep and Sustained Loads

Serviceability of High Strength Concrete Beams with Internal FRP Reinforcement
Under Sustained Load 203
Shawn P. Gross, Joseph Robert Yost and Jonathan V. Crawford

Deformational Behavior of FRP Confined Concrete Under Sustained Compression 207

Rajeev Kaul, R. Sri Ravindrarajah and Scott T. Smith

Long-Term Behavior of CFRP Laminates: An Experimental Study 211

Francesco Ascione and Geminiano Mancusi
Creep Tests on GFRP Pultruded Shapes Stiffened with Carbon Fiber Sheets 215
Marina Bottoni, Claudio Mazzotti and Marco Savoia

Part VII. Design Guides

A Design Procedure of FRP Confining Systems for Upgrade R/C Columns 219
Angela Di Nardo, Ciro Faella and Roberto Realfonzo

Design and Manufacturing of Low Cost Thermoplastic Composite Bridge

Superstructure 223
Nasim Uddin, Abdul Moeed Abro and Uday Vaidya

Italian Design Guidelines for the Strengthening of Existing Civil Constructions

Using Externally Bonded Fiber Reinforced Polymers 227
Luigi Ascione, Andrea Benedetti, Antonio Borri, Angelo Di Tommaso,
Luciano Feo, Roberto Frassine, Gaetano Manfredi, Giorgio Monti, Antonio
Nanni, Maurizio Piazza, Carlo Poggi and Elio Sacco

Large-Size Reinforced Concrete Columns Strengthened with Carbon FRP:

Validation of Existing Design Guidelines 231
Silvia Rocca, Nestore Galati and Antonio Nanni

Torsion Design of CFRP Plated RC Members 235

Adrian K. Y. Hii and Riadh Al-Mahaidi

Analysis and Design of FRP Reinforced Concrete Columns 239

Ching Chiaw Choo and Issam E. Harik

Probabilistic Design of FRP Strengthening of Concrete Structures 243

Anders Carolin

Thresholds of Construction Anomalies in FRP Repair Systems 247

Baris Yalim, Serhat Kalayci and Amir Mirmiran

Static and Dynamic Performance of FRP Deck Bridges under Vehicle Loading
and Implications in Design 251
Y. Zhang and C. S. Cai

Load and Resistance Factor Design for FRP Strengthening of Concrete Structures 255
Rebecca A. Atadero and Vistasp M. Karbhari

Czech Guidelines for Strengthening of Concrete Bridges by Composites 259

Miroslav Černý
Part VIII. Durability Issues
Deterioration Mechanism of Bond between CFRP Plate and Concrete in Moist
Environment 263
Zhenyu Ouyang and Baolin Wan

Durability Approach for GFRP Rebars 267

Andre Weber

Durability Assessment of Adhesives and Reinforcement at the FRP-Wood Interface 271

Gary M. Raftery, Annette M. Harte and Peter D. Rodd

Durability of Steel Members Strengthened by CFRP Strips under Mechanical and

Environmental Loading 275
Pierluigi Colombi, Giulia Fava and Carlo Poggi

Effect of Surface Deterioration on Stress Transfer between Concrete and FRP

Laminate 279
Amnon Katz

Part IX. Fatigue Issues

Fatigue Behavior of Externally Bonded Steel Fiber Reinforced Polymer (SFRP)
for Retrofit of Reinforced Concrete 283
Kent A. Harries and Patrick L. Minnaugh

Fatigue Performance of Concrete Bridge Deck Slabs Reinforced with Glass FRP
Composite Bars 287
Amr El Ragaby, Ehab El-Salakawy and Brahim Benmokrane

Fatigue Process of Adhesively Bonded Joints from Pultruded GFRP Adherends 291
Ye Zhang and Thomas Keller

Prediction of the Extended Fatigue Life of Strengthened Bridge Deck with FRPs 295
Hongseob Oh, Jongsung Sim and Minkwan Ju

Static and Fatigue Investigation of Bridge Deck Cantilevers 299

Chad Klowak, Aftab Mufti and Baidar Bakht

Part X. FRP Bars

Environmental Considerations in Using FRP Rebars in Concrete Pavements 303
Amnon Katz

Evaluation of Crack Widths in Concrete Flexural Members Reinforced with FRP

Bars 307
Charles E. Bakis, Carlos E. Ospina, Timothy E. Bradberry, Brahim
Benmokrane, Shawn P. Gross, John P. Newhook and Ganesh Thiagarajan
The Performance of Curved Non-Ferrous Reinforcement for Concrete Structures 311
Thanongsak Imjai, Maurizio Gaudagnini, Kypros Pilakoutas and Peter Waldron

Flexural Testing of Spun-Cast Hollow Concrete Pile Sections Reinforced with

CFRP Grid and Bars 315
Antonis Petrou Michael and H. R. Hamilton III

Use of GFRP Bars as Reinforcement for Concrete Bridge Deck Slabs 319
Brahim Benmokrane, Ehab El-Salakawy, Sherif El-Gamal and Amr El-Ragaby

Optimization of Braided Reinforced Composite Rods 323

Cristiana Gonilho Pereira, Raul Fangueiro, Said Jalali and Mário de Araújo

Part XI. Health Assessment

Nondestructive Evaluation of FRP Bonding by Shearography 327
F. Taillade, M. Quiertant and C. Tourneur

Structural Health Monitoring in Cold Climate of a CFRP Strengthened Concrete

Hollow Box Girder Bridge 331
Björn Täljsten and Arvid Hejll

Structural Health Monitoring of Degrading Concrete Beams in a Laboratory

Environment 335
Markus Bergström and Björn Täljsten

Debonding Detection in CFRP Strengthened RC Beams Using Active Sensors 339

Seung Dae Kim, Chi Won In, Kelly E.Cronin, Hoon Sohn and Kent Harries

Part XII. Hybrid Systems

Experimental Study on FRP-Concrete Hybrid Beams 343
Tianhong Li, Peng Feng and Lieping Ye

Hybrid Bonding of FRP Laminates to Reinforced Concrete Structures 347

YuFei Wu and Yue Huang

Hybrid FRP-Concrete Sandwich Bridge Deck 351

Thomas Keller, Erika Schaumann and Till Vallée

Hybrid Square FRP/Steel Grid Tube Confined Concrete Cylinders 355

Manu John and Guoqiang Li

Structural Characterization of Composite Sandwich Panel with Hybrid FRP-Steel

Core for Bridge Decks 359
Hyo Seon Ji, Byung Jik Son and Z. John Ma
Part XIII. Masonry Structures
Characterization of Masonry Wallettes and Shear Triplet Specimens Retrofitted
with GFRP Compoistes 363
S. Suriya Prakash and P. Alagusundaramoorthy

FRP Composite Reinforcements on Masonry Vaults: Effectiveness and Reliability 367

Alessandro Baratta and Ottavia Corbi

On the Reinforcement of Masonry Walls by Means of FRP Provisions 371

Alessandro Baratta and Ileana Corbi

Strengthening Masonry Arches with Composites 375

Giulio Castori, Antonio Borri, Skip Ebaugh and Paolo Casadei

Prioritized FRP Research for Concrete and Masonry Structures 379

Max Porter and Kent Harries

Part XIV. Modeling

An Analytical and Numerical Investigation of Debonding Problems in Beams
Strengthened with Composite Materials 383
Domenico Bruno, Fabrizio Greco and Paolo Lonetti

Numerical Analysis of Two-way Concrete Slabs with Openings Strengthened

with CFRP 387
Piotr Rusinowksi, Ola Enochsson and Björn Täljsten

Numerical Modeling of FRP Shear Strengthened RC Beams Using Compression

Field Theory 391
Zhe Qu, Xinzheng Lu, Lieping Ye, Jianfei Chen and John Michael Rotter

Numerical Simulation of Bond Deterioration between CFRP Plate and Concrete

in Moisture Environment 395
Zhenyu Ouyang and Baolin Wan

Numerical Study on Masonry Load Bearing Walls Retrofitted with FRP

Composites 399
S. Suriya Prakash and P. Alagusundaramoorthy

Modeling of RC Square Hollow Piers Wrapped with CFRP 405

Gian Piero Lignola, Andrea Prota, Gaetano Manfredi and Edoardo Cosenza

Prediction of failure Loads of FRP Pultruded profiles Using Closed-form Equations 409
Linda M. Vanevenhoven, Carol K. Shield and Lawrence C. Bank
Theoretical Study on Spalling Resistance of FRP Sheets Bonded to Bent Concrete
Surface 413
Hong Yuan and Renhuai Liu

Theory of a New Method for Externally Anchored FRP Band Pre-Stressed

Reinforcement 417
Jing Zhuo and Tangning Li

Post-Buckling Behavior of FRP Composite Thin-Walled Members 421

Nuno Freitas Silva, Nuno Silvestre and Dinar Camotim

Fatigue Crack Growth Simulation for CFRP Bonded Steel Plates using Boundary
Element Method 425
Hongbo Liu, Xiao-Ling Zhao and Riadh Al-Mahaidi

Nonlinear Finite Element Modeling of RC Beams Strengthened with Different

CFRP Schemes 429
Rajai Z. Alrousan, Mohammad A. Alhassan and Mohsen A. Issa

Finite Element Analysis of FRP Debonding from Concrete Undergoing Global

Mixed Mode I/II Loading 433
Baolin Wan and Zhenyu Ouyang

Finite Element Modeling of FRP Shear Strengthened RC Beams 437

O. F. A. Otoom, S. T. Smith and S. J. Foster

A Finite Element Analysis on Shear and Normal Stresses in Adhesively-Bonded

Joints of Composite Structures 441
Luciano Feo and Francesco Ascione

Finite Element Modeling of FRP-Filament Winded Poles 445

R. Masmoudi and H. Mohamed

Part XV. Novel Applications

Advanced Grid Stiffened FRP Tube Encased Concrete Cylinders 449
Guoqiang Li and Dinesh Maricherla

Applicability of Fiber Reinforced Plastics to Hydraulic Gates 453

Tomonori Tomiyama and Itaru Nishizaki

Stressed Arch Modular Deployable Composite Shelters Concept and Development 457
T. Omar, G Van Erp and P. W. Key

Evaluation of Epoxy Encapsulated Timber Piles 461

Nakin Suksawang, Hani Nassif, Ozgul Ozkul, Ali Maher and Abbas Sarmad
Part XVI. Prestressing Applications
CFRP Prestressing Tendons: Solution Uptake and Swelling Effects 467
Paul Scott and Janet Lees

Experimental Study on Behaviors of Concrete Beams Strengthened with External

Prestressing CFRP Tendons 471
Zhi Fang and Hongfang Li

Post-strengthening of Prestressed Concrete Beams Using Unstressed and Pre-

stressed CFRP Strips 475
Mohammad Reza Aram, Masoud Motavalli and Christoph Czaderski

Reinforcement Technology by Prestressed Carbon Fiber Reinforced Plastics

(CFRP) 479
Shouping Shang, Yongjun Jin, Hui Peng, Ming Wang and Qiaojuan Leng

Wedge Anchorage for Loaded or Prestressed FRP 483

Stefan L. Burtscher

Part XVII. Repair of Columns

Evaluation of the Carrying Capacity of RC Columns Strengthened with Composite
Materials 487
Emmanuel Ferrier, Marc Quiertant and Patrice Hamelin

Large-size Reinforced Concrete Columns Strengthened with Carbon FRP:

Experimental Evaluation 491
Silvia Rocca, Nestore Galati and Antonio Nanni

Parametric Analysis of Shape Effects in Concrete Columns Confined with FRP 495
Ricardo Carrazedo and João Bento de Hanai

Slenderness Effects on Circular FRP-Wrapped Reinforced Concrete Columns 499

Jason Fitzwilliam and Luke A. Bisby

Theoretical Model for FRP-Confined Circular Concrete-filled Steel Tubes Under

Axial Compression 503
Jin-Guang Teng and Y. M. Hu

Part XVIII. Repair Techniques

Methodological Considerations for Reinforced Concrete Structures Repair Using
Posttensioned CFPR Strips 507
José Antonio Bellido de Luna del Rosario

Physico-chemical and Mechanical Behavior of Fiber Reinforced Repair Mortars 511

Amjad Mallat and Abdenour Alliche
Severely Damaged URM Walls Retrofitted with FRP Strips Subjected to Out-of-
Plane Loading 515
C. R. Willis, Q. Yang, R. Seracino, S. H. Xia and M. C. Griffith

Supporting Conditions Effects in the Out-of-plane Behavior of URM Walls

Strengthened with Externally Bonded FRP Strips 519
Ehab Hamed and Oded Rabinovitch

Two Dimensional Evaluation of Patch Repairs of Structural Elements 523

Sameer Hamoush, Maurice McNeal and Hisham Abdul-Fattah

Part XIX. Retrofit of Slabs

Analysis of Debonding Failures in FRP-Strengthened Concrete Beams and Slabs 527
Kenneth W. Neale, U. A. Ebead, W. Elsayed, H. Abdel Baky and A. Godat

Australian Generic Retrofitting Guideline for Plating RC Beams and Slabs 531
Deric John Oehlers

Indirect Crack Control Procedure for FRP-Reinforced Concrete Beams and

One-Way Slabs 535
Carlos E. Ospina and Charles E. Bakis

Strengthening of Precast Prestressed Hollow Core Slabs with CFRP Sheets 539
M. K. Rahman, A. H. Al-Gadhib, J. Mohiuddin and M .H. Baluch

Use of Carbon-Fiber Reinforced Polymers in Slab-Column Connection Repair 543

Widianto, Ying Tian, Jaime Argudo, Oguzhan Bayrak and James O. Jirsa

Part XX. Seismic Applications

Seismic Retrofit of Large Scale Circular RC Columns Wrapped with CFRP Sheets 547
G. Wu, Z. S. Wu, Z. T. Lv and D. S. Gu

Seismic Retrofitting of Reinforced Concrete Columns by Continuous Fiber Rope

with Concrete Jacket 551
Takumi Shimomura, Nguyen Hung Phong, Akihiro Matsumoto, Kenzo Sekijima
and Kyuichi Maruyama

Seismic Performance of Hybrid FRP-Concrete Pier Columns 555

Yilei Shi, Bin Li, Amir Mirmiran, and M. Saiid Saiidi

Experimental Study of FRP Wrapping for RC Bridge Under Seismic Loads 559
Éric St-Georges, Nathalie Roy, Pierre Labossiere, Jean Proulx and Patrick
Paultre

Use of CFRP to Strengthen Poorly Detailed Reinforced Concrete Beams 563

Insung Kim, James O. Jirsa and Oguzhan Bayrak
Residual Behavior of FRP Retrofitted RC Columns after being subjected to
Earthquake Damage 567
Bo Shan, Yan Xiao and Zhengyu Huang

Estimation of the Ductility of Web-bonded FRP Beams for Assessment of

Strengthened RC Exterior Joints 571
S. S. Mahini and H.R. Ronagh

Part XXI. Shear Retrofit

Parameters Influencing the Behavior of RC Beams Strengthened in Shear with
Externally Bonded FRP 577
Omar Chaallal and Abdelhak Bousselham

Calculating the Thickness of FRP Jacket for Shear and Torsion Strengthening of
RC T-Girders 581
A. Deifalla and A. Ghobarah

Calibration of Eurocode-like Equation for the Concrete Contribution to the Shear

Capacity of FRP RC Members 585
Raffaello Fico, Andrea Prota, Renato Parretti, Gaetano Manfredi and Antonio
Nanni

Ductility and Shear Strength Enhancement by Fiber Sheet with Large Fracturing
Strain 589
Ueda Tamon, Dhannyanto Anggawidjaja, Senda Mineo, Nakai Hiroshi and
Jiangguo Dai

Application of Newly Developed GFRP Stirrups for Shear Reinforcement of

Concrete Beams 593
Jongsung Sim, Cheolwoo Park, Sungjae Park and MinKwan Ju

Repair of Corrosion Damaged Concrete Beams without Shear Reinforcement

using Carbon Fiber Reinforced Polymers Sheets 597
Abdullah Al-Saidy

Shear Retrofit of Low Strength Reinforced Concrete Short Columns with GFRP
Composites 601
Alper Ilki, Idris Bedirhanoglu, Ismail Hakki Basegmez, Cem Demir and Nahit
Kumbasar

State-of-Practice of FRP Strengthened RC Girders in Shear 605

Abdeldjelil Belarbi, Ashraf Ayoub and Sang-Wook Bae

Strengthening of Concrete Beams in Shear with Mineral Based Composites

Laboratory tests and theory 609
Björn Täljsten, Katalina Orosz and Thomas Blanksvärd
Analytical Prediction of Debonding Failures in RC Beams Strengthened in Shear
with NSM FRP Reinforcement 613
Andrea Rizzo and Laura De Lorenzis

Part XXII. Strengthening

Flexural Behavior of Reinforced Concrete Beams Strengthened with Near Surface
Mounted CFRP Strips 619
Renata Korynia

Strengthening Choices for the Repair and Retrofit of Concrete Bridge Structures
with FRP 623
J. M. Eyre, T. J. Ibell and A. Nanni

Natural Fiber Composites for Strengthening of Glued-laminated Timber in Tension

Perpendicular to Grain 627
Alann André, Helena Johnsson and Anders Carolin

Prestressed SRP and CFRP Sheets for Strengthening Concrete Beams 631
Daniel B. Prentice and Gordon Wight

Flexural Strengthening of 48-year Old Pedestrian Bridge Reinforced Concrete

Girders 635
Raafat El-Hacha

Strengthening of RC Beams with Prestressed Near Surface Mounted CFRP Rods 639
Moataz Badawi and Khaled Soudki

Strengthening of Steel Bridges Under Low Frequency Vibrations 643

L. Zhang, L. C. Hollaway, J. G. Teng and S. S. Zhang

Structural Strengthening with Externally Bonded SCRP (Steel Cord Reinforced

Polymer) 649
Wine Figeys, Luc Schueremans, Dionys Van Gemert and Kris Brosens

Worlds First Bridge with Sprayed FRP Strengthening: Revisited 653

N. Banthia, R. C. Tennyson and A. Mufti

Analysis of Post-tensioned Concrete Road Bridge Beams Strengthened by CFRP

Tapes Using FEM Under Static Load 657
Arkadiusz Mordak and Zbigniew Manko

A Review of FRP-Strengthened RC Beam-Column Connections 661

S.T.Smith and R. Shrestha

Anchorage of Carbon Fiber Reinforced Polymer Sheets with and without Height
Transition 665
Sarah L. Orton, James O. Jirsa and Oguzhan Bayrak
Effect of FRP U-jacketing Anchorage on FRP-Concrete Interfaces under Fatigue
Loading 669
Kentaro Iwashita, Zhishen Wu, Takashi Ishikawa and Yasumasa Hamaguchi

FRP Upgrade of Concrete Girders to Support Flood Loads 673

Abdelhakim Bouadi, Eric Green and Narendra Gosain

Tests on FRP-strengthened Timber Joints 677

K. I. Crews and S. T. Smith

Recent Developments on the Use of Inorganic Polymer for High Temperature/

High Strength Composites 681
James Giancaspro and P.N. Balaguru

The Influence of Fatigue and Aggressive Environment on the Performance of

RC Beams Reinforced with CFRP Materials 685
Marco Arduini, Mariano Romagnolo and Antonio Nanni

Bond Strength of Carbon Fiber Reinforced Polymers on Concrete and Masonry 689
Umit Serdar Camli and Baris Binici

Compressive Stress-Strain Models of FRP-Confined Concrete 693

Mingxue Liu and Jiaru Qian

Accelerated Replacing and Construction of Concrete Bridge Decks Using Thin

CFRP Stiffened Plates 697
Lijuan Cheng

Part XXIII. Strengthening of Metals

Bending Strength of CFRP-strengthened Circular Hollow Steel Sections 701
J. Haedir, M. R. Bambach, X.L. Zhao and R. Grzebieta

Bond and Splice Behavior of High Modulus CFRP Materials Bonded to Steel
Structures 705
M. Dawood and S.Rizkalla

CRFP-Aluminum Alloy Composite Structures: A New Type of Composite

Structures in Future 709
Lieping Ye, Peng Qian and Peng Feng

Experimental Study on CFRP-Aluminum Alloy Composite Pipes Under Axial

Compressive Load 713
Peng Qian, Lieping Ye and Peng Feng

Rehabilitation of Cracked Aluminum Structures with GFRP Composites

Considering Fatigue 717
Chris P. Pantelides and Justin D. Nadauld
Effectiveness of Different Composite Materials for Repair of Steel Bridge Girders 721
Amr Shaat and Amir Fam

Experimental and Numerical Analysis of the Structural Response of FRP-

Strengthened Cold-Formed Steel Columns 725
Ben Young, Nuno Silvestre and Dinar Camotim

Experimental and Numerical Study of the Debonding Mechanisms in Steel Beams

Strengthened by CFRP Strips 729
Pierluigi Colombi, Paolo Dubini, Giulia Fava and Carlo Poggi

Experimental Study on Fatigue Behavior of Tensile Steel Plates Strengthened

with CFRP Plates 733
Yun Zheng, Lieping Ye, Xinzheng Lu and Qingrui Yue

Preliminary Bond-Slip Model for CFRP Sheets Bonded to Steel Plates 737
S. Fawzia, X .L. Zhao, R. Al-Mahaidi and S. H. Rizkalla

Flexural Strengthening of Composite Steel-Concrete Girders Using Advanced

Composite Materials 741
Raafat El-Hacha and Nora Ragab
 Keynotes*

*Keynote papers that are available at the time this volume went to press.
Third International Conference on FRP Composites in Civil Engineering (CICE 2006)
December 13-15 2006, Miami, Florida, USA

THE PROGNOSIS OF FRP COMPOSITES IN CONSTRUCTION FROM A

WORLD WIDE PERSPECTIVE
L C Hollaway
University of Surrey, Guildford, Surrey, UK.

ABSTRACT
Advanced polymer composites (APC) have hitherto been utilized predominately in the aerospace and marine
industries but for the last three decades there has been a growing awareness amongst civil/structural engineers of the
importance of the unique mechanical and in-service properties of these materials together with their customized
fabrication technologies. These extraordinary properties of APCs have enabled researchers and engineers to innovate
and produce techniques in manufacture and design that can be use in the civil engineering infrastructure ranging
from the strengthening of reinforced concrete, metallic materials, and the seismic retrofitting of bridges and columns
to the use in replacement bridges decks and in the new bridge and building structures. Their lightness, corrosion
resistance, potentially high overall durability, tailorability and high specific attributes enable them to be used in
areas where the conventional construction materials might be restricted. In spite of these excellent characteristics
there are many engineers who are still not convinced of the potential of advanced polymer composites.
This paper will illustrate the advantageous position in which the APC composites industry currently finds
itself and how this situation might be further exploited. The discussion will cover a review of APCs in terms of (i)
The innovation concept developments of the advanced polymer composite material over the last three decades. (ii)
The durability concerns and the reliability of field tests (iii) Addition of additives to improve performance of the
APCs in construction. There seems to be no justification for resistance to its use in construction providing the
material is understood and designed correctly.

KEYWORDS
Polymer composites, durability, additives, infrastructure rehabilitation, hybrid constructions.

1. INTRODUCTION

Thirty years ago the construction industry, generally, was considered as a low technology, low skill and labour
intensive compared with most other industries; this fact was highlighted in a UK report, Latham Report (1994). It is
therefore not surprising that the polymer/composites industry did not show interest in the civil engineering industry
where the possibility of utilizing new materials and therefore market opportunities were not visible to potential
investors. The technological revolution in materials and processing in all other sectors of the manufacturing industry
had largely by-passed the construction industry. The mismatch between research investment and construction
expenditure meant that construction proceeded on a scale with an inadequate understanding of many aspects such as
deterioration mechanism for structures and this often meant that due allowance was not made for practical repair and
maintenance. Consequently, when in the early 1970, the initial inception of fibre reinforced polymer (FRP)
composites into the building industry was introduced as a semi-load bearing structural material it was hailed by a
few engineers as a breakthrough for the FRP composites industry. However, it was reviled by many as an
architectural gimmick which would not last. Understandably, during that early period, the progress of the advanced
polymer composite (APC) material for civil engineering applications was slow but over the last two decades its
utilisation has gained momentum. Currently major structural systems are being constructed completely from APCs
or the APCs are used in combination with the more conventional civil engineering materials to form hybrid
structures where the component parts of the structural system are used advantageously. Many civil engineers are
now convinced of the value of the APC material but still a percentage of engineers require assuring of its potential
capabilities. In addition, they do not accept it as being a potential competitor to the conventional heavier
construction materials.

1
 All scientific and engineering inventors are keen to exploit their findings and will sometimes make far
reaching claims which, at the time, cannot be justified; the initial inventors/users of the 1970s structural FRP
systems were no exception. Currently, inventors within the composites field are more restrained in their claims, due
possibly to a greater understanding of the material and an upsurge in litigation attitudes. Unfortunately, the claims
made during the early 1970s, with no justification, did the FRP composites industry a great deal of harm which was
difficult to overcome. Furthermore, during this period small fabricators, consisting of one or two operatives, set
themselves up as experts, and were mainly involved in the manufacture of components for load bearing or semi-
load-bearing infill panels; these were the systems used at that time and made by the hand lay-up method which is
rather primitive by present day standards. These fabricators paid little or no attention to the curing procedure of the
composites and as a consequence these units were inadequately cured and within a short period of time, in practical
use, commenced to degrade. However, the larger and experienced fabricators manufactured large structural
buildings such as Mondial House on the banks of the River Thames and the American Express Building, Brighton,
UK, both built in 1974 and currently are showing no signs of deterioration.
This paper illustrates the advantageous position, in which the APC industry currently finds itself, but it
discusses some perceived problems. It indicates the skepticism of some civil engineers to the new developments in
FRP composites for construction, whether these skepticisms are justified and how this situation might be overcome
with further initiatives, exploitations and design codes in the future. The analysis covers the possibilities and
realization of APC material. These points are illustrated by introducing the material and potential in terms of (i)
Innovations and concept developments, (ii) Durability and long term testing requirements, (iii) Addition of additives
to improve a particular property.

2. INNOVATIONS AND CONCEPT DEVELOPMENTS.

2.1 Introduction

The utilization of FRP as a substitute for the more conventional materials must be resisted, unless there is a definite
advantage in its use. An example where the APC has taken the initiative is in the rehabilitation of structural
members; this topic will be raised late in the paper. Ideally, the FRP material should be designed and used in
situations best suited to its properties and material characteristics; this implies innovations and indeed throughout the
world a considerable effort is being exerted by researchers examining ways of utilizing composites either as ‘all
composite’ structures or in combination with the more conventional civil engineering materials. This section will
demonstrate some of the FRP composite concepts which have been developed over the past 30 years and which give
an indication of the methodology that should be applied to FRP composite concept developments if the material is to
become one of the leading materials of the 21st Century. Composites do have disadvantages and these, of course,
must be taken into account when developing a new system; often these disadvantages can be made into ‘stepping
stones’ in the development of a system which is not affected by these disadvantages. The disadvantages can be
broadly divided into three groups as:
(i) Costs compared with other materials used in construction. In this context it must be emphasized that
the volume and amount of FRP material used in a similar form of construction is small compared with the
conventional material of civil engineering, therefore, they cannot be compared on a weight basis.
Nevertheless, cost can still be a problem and innovative ways of forming structural systems must be
developed. Examples will be given later where it has been a decided advantage to use APC due to their
advantageous long term properties. The whole life costing must be considered in structural work.
(ii) Fire is a problem with civil engineering structures which are designed as thin systems. On off-shore oil
installations FRP are used as fire barriers because of the relative thick composite systems used; is there a
possibility here for civil engineering?
(iii) APCs must be used as tension components. Due to the thin laminates used in construction buckling is
a major problem.

2.2 The Building Block System

The gradual implementation of innovative research and construction programmes, since the 1970s, has successfully
demonstrated the potential of FRP composites technology for use in construction. This revolution in the use of FRP
structural material commenced in the mid-1970s with the building of ‘all-composites’ construction such as the wet
lay-up (hand laminated) randomly orientated glass fibre reinforced polyester polymer (GFRP) composite classroom
at Preston Lancashire, UK. The idea was to develop a system which could speed the construction process by using
light weight large building blocks to construct the structure. In this example an icoshedron geometrical shape was

2
 Figure 1a Manual fabrication of Building Block Figure 1b Mechanized fabrication of Building Block
GFRP Class Room Lancashire UK Bonds Mill Bridge Gloucester UK.
used for the overall structure and the flat surfaces of the icoshedron were again folded by using four number three
sided ‘pyramids’. This folded plate structure gained stiffness by the folded nature of its geometrical shape.
Experimental tests to destruction were undertaken on coupon specimens made from pristine GFRP composite
material and on the building blocks of the structure (the large pyramidal units). To resist fire, 50 mm thick integral
skin phenolic foam units were fitted into the building block units; fire tests at the Building Research Establishment,
UK were carried out on an assembled four building block unit. The tests showed that the structural unit had over one
half hour fire resistance. It was felt at the time that the material and the building block system were very satisfactory
for utilization in construction; indeed a whole school system was designed. At that time little thought had been given
to the durability of the material, however, field analysis tests undertaken on this structure over the past 35 years have
shown that the material has suffered no degradation. In the 1980s the building block system was mechanized when
the Maunsell Plank (the building block) was developed by Maunsell Structural Plastics (now Faber Maunsell) and
used to form the Aberfeldy bridge in Scotland over the river Tay. The Plank was manufactured by the pultrusion
technique and the only manual operation associated with the erection of the bridge was to effect the bonding
together of the building blocks and the fabrication of the bridge. The building block (Maunsell Plank) was also used
in box beam form (10 plank units) to fabricate Bonds Mill Bridge. This mechanized building block system
demonstrated that the idea was sound but it exposed a failing in attitude of civil engineers to readily accept new
ideas. From past experience new materials and structural concepts will take twenty or more years to become
accepted by the civil engineering industry.
It can be concluded that the building block technique is an ideal procedure to readily produce ‘all FRP
composite’ structures quickly, efficiently and utilizing only a limited number of different structural units.

2.3 The Improvements in Polymers for the Composite Construction

Since the 1970s polymers have been modified and improved for various uses in construction. To some extent this
has made the understanding of the material more difficult as every time a new formulation is introduced, a new set
of characterization curves have to be produced. Consequently, obtaining information on its long term characteristics
are difficult to obtain quickly. In the development of FRP composites for construction, very often the civil
engineering industry has drawn upon the knowledge and experience of the aerospace industry to improve the
composite materials and manufacturing technologies. Generally the material that has been used to form the civil
engineering construction components has been the two part resin and a curing agent system; the resins were initially
polyester but these have largely been replaced by the vinylester and epoxy resins. Recently, the pre-impregnated
(prepreg) material, which is widely used for patching in the aircraft industry, has been of interest to construction as it
has advantages over the two part component material. It is realized, however, that there are subtle differences
between the material requirements and the manufacturing technologies for these two industries, but it is the
ingenuity and innovation of the material scientist and civil engineer who have modified/changed the polymer
component of the composite, the curing procedure and the way in which the final product is utilized to suit the
particular civil engineering requirements. This characteristic of being able to tailor the material to meet any desired
specification at the time of designing the structural component is a decided advantage over the more conventional
civil engineering materials which are normally restricted to specific strength/stiffness values. A new technique for
the fabrication of a carbon fibre pre-impregnated with an epoxy polymer, (the material/system, (the prepreg)
manufactured by Advanced Composites Group Ltd. Heanor, Derbyshire, UK), was initially introduced into the
aerospace industry, the material has been modified considerably for use in the civil engineering industry This
innovative system with a film adhesive compatible with the matrix of the composite can be used to upgrade a

3
concrete or metallic structural member; the polymer is cured at 650C applied for 16 hours under a vacuum assisted
pressure of 1 bar. The advantage of this material and the curing system is that the adhesive film is used in
conjunction with the prepreg, both can be fabricated in the factory or on site and around members of any geometric
shape. The system has been used for the strengthening of metallic structures (Garden 2004) and for the fabrication of
the ‘Duplex’ composite/concrete Spanish bridge beam system, which will be described in Section 2.4.2. The
compaction (for minimum voids), the film adhesive system (curing simultaneously with the matrix material of the
composite for greater interpenetration contact) and factory or site fabrication method (to provide the upgrading
material with an ideal fit to the structure) implies that a high quality composite/bonding system can been employed.
Because of the structural advantages of the technique mentioned above it is possible that this system will be
the preferred one for retrofitting structural members in the future.

2.4 Combining Composite/Concrete Material

Concrete, steel and timber have been the prominent materials utilized in the construction industry for many decades
and this situation is likely to continue to be the case for the next decade or so, however, advanced materials and in
particular the advanced polymer composites have been combined with the more conventional materials for over a
decade and the combination have produced a new generation of composite structures which outperform structural
systems using individual conventional materials only. This section will discuss these combinations which have been
used and new system which are entering the construction industry; these combinations have been developed through
the imaginative skills of the civil engineers.

2.4.1 The rehabilitation of degraded bridges or the upgrading to improve the loading capacity of the bridge.

During the late 1980s research and innovation was directed towards systems using FRP composites in conjunction
with the more conventional civil engineering materials. In these systems the two dissimilar materials were used to
their greatest structural advantage. Initially, structures using this concept tended to be those requiring rehabilitation
because of (i) degradation to the parent material due the hostile environment in which the structure was situated, (ii)
a change of use of a structural building and (iii) an increase in loading capacity of a bridge or structural building. It
was found that the utilization of FRP material in this latter example helped to solve the numerous challenges facing
the civil engineer. It has been reported in publications many times that a large number of bridges throughout the
world have been classified as functionally obsolete or structurally deficient and require some form of maintenance
or major rehabilitation to restore them to their original structural condition. There are many examples of upgraded
structures that can be cited Hollaway (2007) and where FRP composites have been successfully used. Although this
technique, over the short period of time in which it has been in existence, has been shown to be satisfactory, the long
term durability of the joint has yet to be proven.

2.4.2 Structures manufactured from composites and concrete as new structural systems.

The procedures used to upgrade structures were the forerunners of a future technique to combine FRP composites
and concrete. This innovative idea was first introduced by Triantafillou and Meier (1992) and the basic concept of
this construction has been developed independently by a number of researchers Van Urp et al 2003, Canning et al
(2000), Hulatt et al (2003). It consists of placing the bulk concrete in the compression part of a rectangular or Tee
section beam and the FRP composite in the tension zone; this latter to resist flexural and shear forces on the beam.
As the FRP composite can only be used in relatively thin plate form the interior of the tension part of beams may
require a ‘permanent shuttering’ constructed from, say, a foamed polymer. The advantages of this new concept are a
considerable reduction in beam weight, high load carrying capacity and good fatigue behaviour. Researchers at the
University of Southern Queensland, Australia have expanded the development into the construction industry where
two or three bridges have been built using this technique. From the research work on the combined FRP
composite/concrete member (duplex system), undertaken at the University of Surrey UK, an example of a bridge
constructed using this system has been erected by NECSO Entrecanales Cubiertas, Madrid, Spain using the material
of Advanced Composites Group Ltd. Heanor, Derbyshire, UK. The beam element utilizes the high compressive
strength of the concrete and the high tensile strength of carbon fibre; the manufacturing method is by the pre-
impregnated FRP technology, using a cure temperature of 650C for 16 hours and a vacuum assisted pressure of 1
bar. The benefits of this system is in the significant cost savings provided due to the lower weight and reduced
through life cost of the beam. The opportunities for this technology are remote site installations and refurbishment of
infrastructure in developing countries or war regions.

4
 Concrete

Foam (Permanent
Shuttering)

CFRP/GFRP
Composite
Figure 2a Testing of 5.5 m exp. Beam Figure 2b Construction of Bridge
NECSO Entrecanales Cubiertas/ACG Bridge, constructed in Spain
Research engineers at the University of Southern Queensland have developed a fibre/polymer composite
system that can replicate the appearance and integrity of hardwood timber for the replacement of timber rail and
road bridges. The researchers claim that the material stores more carbon dioxide than it release, and it is virtually
maintenance-free; researches are now investigating the possibility of making the polymers used in the structure from
plant oils rather than fossil fuel oils thus providing a renewable source
Concrete filled steel tubes have been in existence for some time and a variation of this form is two
concentric steel tubes with the space between them filled with concrete. Hybrid members in the form of a double
skin tube composed of a steel inner tube, a GFRP composite outer tube and a concrete infill in the space between the
two tubes has recently been introduced by the research team at The Hong Kong Polytechnic University; the member
can be employed as a column or beam. The advantages of this hybrid member over the earlier versions are: (a) The
hybrid member exhibits a high degree of ductility. (b) The hybrid section has good corrosion protection, (c) The
positioning of the inner steel tube towards the tension zone of the beam provides substantial improvements in
flexural stiffness, ultimate load and crack resistance, (Yu et al 2006).

2.4.3 Reinforced concrete using FRP rebars.

Over recent years there have been published research papers that have indicated a number of potential problems with
this method of using FRP technology as internal reinforcement to concrete and in particular if the fibre used in the
composite is a glass fibre. It has been reported by Bank, L. C. and Gentry, R.T. (1995), Sen, et al (2002) and Bank et
al, (1998) that accelerated laboratory test results of GFRP in a simulated concrete pore water solution of high pH
values and at elevated temperatures up to 800C have indicated that there is a decrease in the tensile, shear and bond
strengths,; these results would suggest that there is a case for not using GFRP rebars in concrete, Uomoto (2000).
However, Tomosawa and Nakatsuji (1997) have shown that after 12 months exposure to alkaline solutions at
temperature between 200-300C, and Clarke and Sheard (1998), likewise, after 2 years exposure to a tropical climate
on a test platform off the Japanese coast, have reported that there had been no material or physical deterioration to
the GFRP composite. Furthermore, Sheard et al (1997) reported that the overall conclusions of the work of the
EUROCRETE project were that GFRP is suitable in a concrete environment. ISIS Canada Research Network of
Centers of Excellent, and associated Universities, in Canada, are undertaking long term field exposure analysis and
laboratory tests on samples of five GFRP reinforced concrete structures across Canada to provide information on the
reliability of GFRP rebar materials, which had been incorporated into a concrete situated in the natural environment
for 5 to 8 years. To date no degradation of the GFRP rebars have been recorded. This would indicate that the results
from accelerated tests of FRP composite materials should be treated with caution by engineers reading technical
papers on this topic and that a thorough understanding of the accelerated test exposure conditions, procedures and
reasons for the chemical reactions must be understood.
The prognosis for the utilization of FRP rebars looks encouraging and the indications are that FRP rebars can be
used safely over many years.

2.5 Bridge deck replacement.

Many bridges have superstructures of reinforced concrete or steel beams supporting a concrete deck. Typically a
concrete deck has a 25-40 year life span. In areas of the world where it is necessary to use deicing salts during the
winter period, salt solution enters cracks or percolates through the running surface of the bridge to the reinforced
concrete with the possibility of corroding the steel reinforcement. When this deck requires replacement an

5
innovative and cost effective way of undertaking this task is to replace the bridge deck by high quality factory made
FRP components which can be delivered to site and would then be fabricated in an area adjacent to the bridge. The
whole bridge deck component can then be lifted in place by a light crane. The advantages of the FRP deck over the
conventional one are cost, ease of application, lightweight, high strength and high performance, chemical and
corrosion resistant, rapid project delivery and largely factory manufactured. This system has been utilized in the
USA for a number of years and was introduced into Europe in 2004, when the old West Mill Bridge, Gloucester,
UK. was dismantled and replaced with FRP beams and GFRP ASSET deck system. It may be concluded that the
innovative deck system for small to medium span bridges will become one of the major replacements for bridge
decks within the foreseeable future.

2.6 Joining FRP composite materials to similar or dissimilar materials.

The chemistry of adhesive systems is generally not well understood by the civil engineer and advice is normally sort
from the adhesive manufacturers. Epoxy resins are the best known and most widely used resins for the manufacture
of the structural adhesives. There are only a few commercial epoxy resins but a wide range of curing agents,
including amines and acid anhydrides which can polymerize these resins. Structural adhesives can be both cold
cured, used on construction sites and hot cured, used in a factory environment; all are cross-linked. Ideally the cold
cure adhesives should be post cured at an elevated temperature for a limited time depending upon the value of the
temperature of cure. All amorphous polymers have a glass transition temperature (Tg) and during service it is
unacceptable for the temperature of the adhesive to oscillate above and below the Tg, indeed, the temperature of
service should not be taken beyond 200C below the Tg.
There is a lack of information on the durability and long term physical and mechanical characteristics of
adhesive polymers; this area must be investigated urgently.
Generally, the civil engineer does not use the adhesive film when designing large structural systems but a
unique technique to the civil engineering industry has been demonstrated by Photiou et al (2006) and Zhang et al
(2006) and has been shown practically by Garden, (2004) that an excellent bond can be achieved between an FRP
prepreg composite and the structural steel member (or other general civil engineering material) using a film adhesive
compatible with a prepreg matrix of the FRP composite (a composite system is discussed in Section 2.4.2); both of
which are cured simultaneously under a vacuum assisted pressure at a slightly elevated temperature. This fabrication
procedure and cure method provides an almost void free joint. The joint using the adhesive film has a greater degree
of quality control over that of the two part adhesive joint. The thickness of the joint and the curing procedure can
readily be controlled. In addition, the film adhesive, due to its thinness (approximately 0.1 mm), has a much
reduced free surface area for any moisture or salt solution to enter, thus giving it a better durability characteristic.
The optimum composite joint design is the one capable of distributing stresses over a wide area rather than
concentrating them at a point. Adhesively bonded joints can satisfy this requirement; however, most of the adhesive
joins have a brittle failure. There have been many attempts to join FRP composite tubular members together when
skeletal structures are being fabricated; these structures are in the form of a double layer skeletal roof system, a
crane, etc. Three examples are mentioned here of innovative systems which have been developed and involve
mechanical fixing for transmitting tensile and compressive forces, such as (i) the clip on fasteners developed by W.
Brandt Goldworthy & Associates, Inc. The Snap Joint technology is a similar concept to that used by Kliptico,
developed before the Second World War by an Australian firm, as a toy construction kit; structures were formed
from a series of steel members which clipped/snapped together. The pultruded composite member (or other
manufactured rigid member) which is to be joined to another member by the Snap Joint technology method has one
end shaped as a ‘fir-tree’, and therefore has a large load bearing area, is snapped into another structural member. The
‘fir tree’ end of the joint is slotted along its longitudinal length to provide enough lateral flexibility to be able to
compress when entering the part to be joined. A hole is drilled into the section at the end of the slot to inhibit the
crack propagation along the length of the pultruded member. In order to make this joint concept successful, the fibre
architecture of the ’fir tree’ end must be designed in such a way that the load bearing surfaces have high interlaminar
shear strength capacity. (ii) A specially designed exterior end cap with fins to mate with a slotted cover plate to
mechanically join other skeletal members at a node point of either an axially loaded member or at skeletal joints.
The interior of the end cap has a concentric tube threaded on its outside surface. The precise amount of adhesive is
poured between this threaded side and the threaded inside of the end cap thus enabling the adhesive to work its way
around the threads as the pultruded, or similar, member is inserted into the annulus (iii) The ‘energy’ loaded joint to
deploy a skeletal structure, this system is described in Fanning and Hollaway (1993) These fixings do not involve
bolting through the FRP composite member but uses adhesives to bond the end cap to the tensile/compressive
member (item 2) and the energy loaded joint to the structural member (item 3). Figure 3 illustrates the system.

6
 FRP Linkages
o
Link pin
Figure 3a Structure Half o
FRP tubes (parts
of skeletal
oo o
oo
Figure 3b Structure Fully
Deployed structure)
o
Deployed
FRP Cover tube o
o o
o o o
o
o Deploying spring o
o
o
o o
o o
FRP skeletal tubes
(b)
FRP Linkage Deploying spring
FRP Cover tube FRP Skeletal tube
(a) FRP tube
o o o o
o o
o o o o

(c)

Figure 3c Joint Mechanism

3 DURABILITY AND LONG TERM TESTING OF FRP COMPOSITES

The above mentioned examples of innovative FRP composite structural systems may be considered the early
successes of FRP in construction, but the composite material must resist environmental attack over the life of the
structure. With conflicting durability test results from research laboratories many of which are derived from
accelerated tests and unrealistic elevated temperatures to accelerate the degradation, there remains some uncertainty
over the durability of composites. Exposure to high and low temperature variations, moisture and salt solution
ingress, ultra-violet rays from the sun and fire will all lead to reduced mechanical performance. A further concern is
the durability of ambient-cured systems as these have a relatively low glass transition temperature and may not reach
their full polymerization if they have not been post cured before use, thereby, making them more susceptible to
degradation. It should be remembered, however, that all engineering materials are sensitive to environmental
changes in different ways and those adverse factors mentioned above will not be unique to the FRP materials.
Indeed, composite materials do offer some significant durability advantages over the more conventional construction
materials; these include superior corrosion and fatigue resistance. These advantageous properties must be exploited
further in the future utilization of composite materials.
The only reliable way of estimating durability of composites is to undertake monitoring of the structure
over time or field tests. In 2004, ISIS, Canada Research Network of Centres of Excellent, Mufti et al (2005) Mufti et
al (2005), Mufti et al (2005), commenced an investigation into five GFRP reinforced concrete structures to provide
information on the reliability of GFRP materials, which had been incorporated into a concrete situated in the natural
environment for 5 to 8 years. To undertake this task, core specimens of the GFRP reinforcement were removed form
the structures, which were located across Canada from east to west; giving a wide coverage of varying natural
environmental conditions. These specimens were analyzed to determine any degradation of the E glass fibre and
vinylester polymer rebars used in the concrete structures. The overall conclusions of the ISIS researchers were that,
up to that time, no degradation of the GFRP flexural tension reinforcement had taken place and that the material is
durable and compatible with concrete; the tests are continuing.

ADDITION OF ADDITIVES

The polymers that are exposed to natural weathering will deteriorate over time but the degree of deterioration will be
dependent upon a number of factors. These may be listed as (i) The type of resin used. (ii) The ultra-violet
component of the sunlight and the orientation of the composite relative to the rays of the sun. (iii) The environment

7
into which the polymer is situated. (iv) Additives which are mixed with the polymer. The additives in the form of
particles or chemicals, if used, are mixed with the resin at the time of manufacture of the polymer and the purpose of
adding these is to (a) impart fire resistance, (b) to resist the uptake of moisture, (c) to accelerate the cure of the
polymer. These additives are impurities with respect to polymer and will affect its mechanical properties.
Probably the most severe attack on the composite is the ingress of moisture. Composite systems can be
engineered to provide resistance against diffusion of moisture and aqueous solutions into them, but moisture
eventually will diffuse into all organic polymers leading to changes in mechanical and chemical characteristics. The
primary effect of diffusion is through hydrolysis and plasticization. This process may cause reversible and
irreversible changes to a polymer. The characteristic of the polymer, which may occur, is a lowering of the Tg value
and in a composite causing deleterious effects to the fibre/matrix interface bond resulting in a loss of integrity.
Possible solutions for the reduction of diffusion through the polymer is to :-
(i) apply a protective coating/gel coat on to the system,
(ii) cure the product at an elevated temperature,
(iii) utilize factory made composites, such as the pultrusion technique; these composites are manufactured at
temperatures of 1200C-1400C. This ensures low void content, virtually full cure, high levels of overall
integrity,
(iv) use an appropriate sizing/finishes on the fibres of the composites.
An innovative procedure which is being developed for enhancing the resistance of ingress of moisture and
aqueous solutions into the polymer is by the addition of nano-particles into it. This innovation by material scientists
and engineers could be a major enhancement to the utilization of advanced polymer composites in construction,
particularly adhesives.
Chemically treated layered silicates (clays) can be mixed with polymer matrix materials to form a nano-
composite in which clay layers are evenly distributed through the material. Research has shown that these high
aspect ratio clays alter the properties of the composite by a number of mechanisms, to increase the strength and fire
resistance and to reduce permeability. However, as with all new material systems claims and counter-claims have
been made and the literature shows that there are a number of conflicting reports as to the percentage improvements
that nano-composites can achieve. This might be due to different polymers or curing agents being used in the
research or different nano-composites being employed. As in all research it is vital to confirm (i) precisely the
materials used viz. the type of clay used, and the surfactant that is mixed into the clay, (ii) the temperature and
length of time of mixing of the clays to the polymer, (iii) the type of epoxy polymer that is specified, and (iv) that no
chemical additives have been introduced into the polymer before introducing the nano-composites into the resin,
[additives are often introduced into the resins by manufacturers of polymers for construction]. Before nano-
composite materials can be seriously used in the construction field there are still many questions to be answered, i.e.
what will be the extended life-time of the structure if the material is used? Will the material be relevant to on-site
wet lay-up processing, or will the use of it be restricted to factory produced products? Will the production of nano-
composites be financially viable to civil engineers? One of the advantages in the utilization of nano-composites in
civil engineering will be: (i) The improvement in the inter-laminar shear failure of adhesives, (ii) The barrier
property improvement of composites and polymers in the construction industry. Both of these items will be of
particular relevance to the rehabilitation of structural members. Although nano-particles in FRP composites will
reduce the permeability of the pristine polymer and thereby improve the boundary properties, eventually the ingress
of solutions will penetrate the FRP composites. It will be realized that all man-made materials do have a finite life
and FRP polymer composites are no exception but it is hoped that the improvements that can be made by the
addition of nano-particles to civil engineering composites will improve and extend their performance considerably.

THE EUROPEAN UNION SPHERE

The European Commission Joint Research Centre, 2006, has stated that precise EU market share figures for the
composite market are difficult to obtain due to its high fragmentation. Furthermore, the composites’ market is
included with the plastics sector; indeed European Union’s statistical data base, the Statistical Office of the
European Administration (EUROSTAT), does not collect segregated figures for the composites industrial sector.
This logic is also applied by individual Member States. A survey by French Ministerial Organization Service des
Ètudes et des Statistiques Industriels (SESSI, 2004), based on the EUROSTAT data, estimated that the composite
sector represents 5% (equivalent to Є7 billion) of the total EU plastics market, which is estimated to be worth Є140
billion. These figures equate to those issued by the Department of Trade and Industry, UK , (DTI, 2001) which
estimated the overall revenue of the EU composites industry at Є5 billion and was expected to grow by 3.8%
annually; these figures were based on figures obtained for 1998. Projecting this figure forward for the market
turnover in 2004 would be of the order of Є6 billion, a figure very similar to that of the French government. It is

8
estimated that the industrial sector uses 10% of FRP (e.g. Pressure vessels, piping, chimney, chemical plant
applications), and the transportation sector uses 32%; nearly 65% of the total composites market. However, the
SESSI report has stated that European market share for plastics exports during the years 1990s and into 2000 fell
from 62% to 48% and over the same period the Asian share of the market increased from 12% to 21%. By contrast
the USA market use of FRP in Transportation and Construction has grown substantially over the past two decades
which indicates a growing importance of the FRP market, in spite of the fact that the USA composites industry has
faced many of the problems with which its European counterparts have had to cope. A number of organizations in
Europe have been formed to promote the use of composites in construction, including COBRE (based in the
Netherlands) and the Network Group for Composites in Construction.(based in the UK) but their funding cannot be
compared with that of equivalent programmes currently running in the USA.
The slow development of codes of practice in Europe incorporating advanced composite materials in civil
engineering is still seen as a barrier for its more extensive use. The European Commission Joint Research Centre is
in the process of presenting to the EU Commission justification for drawing up new codes and standards for the use
of FRP composites in civil engineering. Currently, designers rely upon best practice design guidelines, developed by
bodies as (i) the International Federation for Structural Concrete (fib) Task Group 9.3, (ii) European Committee for
Standardisation, prEN 1504, (iii) CIRIA for the strengthening of steel members, (Cadei et al. 2004), or (iv) the
American codes issued by the ASCE and ACI. Without codes of practice there will be little motivation to expand the
use of FRP composite materials for the European construction sector with the result that the resistance to its use by
skeptics will continue.

OBSERVATIONS

The advanced polymer composite material is a unique system and must be designed in such a way as to take full
advantage of its extraordinary properties. This implies that innovative ideas must be employed to ensure that its
advantageous properties are used to the full. All the examples that have been given in this paper have been
dependent upon innovative ideas, which have related to the uniqueness of the material. Currently, and into the
foreseeable future, the paper has suggested that the most efficient way of using the APC material is in conjunction
with the more conventional materials. The structural component formed from this combination, then uses to the full
the advantageous properties of the composite and those of the joining material.
When the whole life costing of a structural unit is taken into account the manufactured composite system
will generally be more economical than that made by conventional materials and for the design/build/maintain
contracts, composites should be at an advantage. Construction materials have an important role to play in sustainable
development through their energy performance and durability, as this determines the energy demand of buildings
throughout their lifetime. By developing the use of materials and their combinations, significant improvements of
the environment and quality of life can be achieved.
There are currently, some limitations to the use of composites, some of these can be overcome with further
research and development (e.g. durability) and some which, at the moment, cannot be overcome (e.g. fire) but
research eventually may be able to solve this problem. For instance, it has been claimed by some researchers that a
large percentage increase can be achieved in fire properties of composites by incorporating exfoliated nano-particles
into the pristine polymer of the composite. If this is economically feasible for the construction industry then other
additives which are to be incorporated into the polymer can be included, after the addition of the nano-particles, to
impart other particular properties to the composite.
As more results from field tests become available, it is clear that the prognosis for the durability of
composite materials, over the life span of a civil engineering structure, shows that the APC is much more durable
than had been first realized from laboratory accelerated tests and in many instances is more durable than the
conventional civil engineering materials. There seems to be no logical justification for not using composites in
construction; indeed the examples given in this paper have indicated the material has considerable advantages in
certain applications. However, one must realize that it cannot be used for every type of construction, it must be used
sensibly and its limitations must be fully realised.

REFERENCES
Bank, L. C. and Gentry, R.T. (1995) ‘Accelerated test methods to determine the long-term behaviour of FRP composite structures:
Environmental effects’, Journal of reinforced Plastics and Composites, Vol 14, pp559-587.

Bank, L. C., Gentry, R.T., Barkatt, A., Prian, L., Wang, F and Mangla, S. R. (1998). ‘Accelerated aging of pultruded glass/vinylester rods’,
Proceeding 2nd International Conference on Fibre Composites in Infrastructure (ICCI), (1998), Vol 2, pp423-437.

Canning, L, Hollaway, L. and Thorne, A.M. (2000) ‘An investigation of the composite action of an FRP/concrete prismatic beam’, Jnl.
Construction and Building Materials, Vol. 13, Elsevier Ltd. pp417-426.

9
Cadei, J M C., Stratford, T J. Hollaway, L C. and Duckett, W G. (2004). ‘Strengthening Metallic Structures Using Externally Bonded Fibre-
reinforced Polymers’ CIRIA,RP 645, Publication No. C595, CIRIA, London.

Clarke, J. L. and Sheard P., (1998). ‘Designing durable FRP reinforced concrete structures’, Proceedings 1st International Conference on
Durability of Fibre reinforced polymer (FRP) Composite for Construction (CDCC 1998), Sherbrooke, Quebec, Canada, (1998), pp 13-24.

European Committee for Standardisation, prEN 1504, Products and systems for the protection and repair of concrete structures; Part ,1
General scope and definitions; Part 2; Surface protection; Part 3, Structural and non-structural repair; Part 4, structural bonding; Part 5,
Concrete injection; Part 6, Grouting to anchor reinforcement or to fill external voids; Part 7, Reinforcement corrosion prevention; Part
8,Quality control and evaluation of conformity; Part 9, General principles for use of products and system; Part 10, Site application of products
and systems and quality control of the work.

European Commission Joint Research Centre European Lab. for Structural Assessment Institute for the Protection and security of Citizen

Fanning, P. and Hollaway, L C (1993) ‘A case study in the design and analysis for a 5 m deployable composite antenna’. In Composite
Engineering Vol 3 No 11, pp 1007-10232

fib, (2001). ‘Externally bonded FRP reinforcement for RC structures’ Bull. No 14, Technical Report prepared by Task Group 9.3 Lausanne’

Garden, H. (2004). ‘Refurbishment of Steel Building Using FRP Material’ MSc Module, Department of Civil Engineering, University of Surrey,
April 2004.

Hollaway, L C. (2007) ‘Field Survey Applications’ Chapter 12 ‘Durability of Composites for Civil Structural Applications’ Ed. V Karbhari,
Publishers Woodhead Publishing Ltd. To be published in 2007.

Hulatt, J., Hollaway, L. & Thorne, A. (2003), ‘Short term testing of a hybrid T-beam made from a new prepreg material’ ASCE Journal of
Composites for Construction, Vol 7, No. 2 pp 135-145. ISSN 1090-0268

Latham Report (1994) Government / Industry Review of Procurement and Contractual Arrangements. In, The UK Construction Industry
[HMSO, London, 1994

Mufti, A., Benmokrane, B., Boulfiza, M., Bakht, B. and Breyy, P. (2005). ‘Field study on durability of GFRP Reinforcement’ International
Bridge Deck Workshop, Winnipeg, Manitoba, Canada. April 14-15, 2005.

Mufti, A., Onofrei, M., Benmokrane, B., Banthia, N., Boulfiza, M., Newhook, J., Bakht, B., Tadros, G. and Brett, P. (2005) ‘Durability of
GFRP reinforced concrete in field structures’, 7th International Symposium on Fiber Reinforcement for Reinforced Concrete Structures
(FRPRCS-7), New Orleans Lousiana, USA, November 7-10, 2005.

Mufti, A., Onofrei, M., Benmokrane, B., Banthia, N., Boulfiza, M., Newhook, J., Bakht, B., Tadros. G. and Brett, P. (2005) ‘Report on the
studies of GFRP durability in concrete from field demonstration structures’. Proceedings of the Composites in Construction 2005 - 3rd
International Conference, eds. Hamelin et al 2005, Lyon, France, July 11-13, 2005.

Photiou, N.K., Hollaway, L.C., and Chryssanthopoulos, M.K. (2006) ‘Selection of CFRP systems for steelwork upgrading’ In press ASCE
Journal of Materials in Civil Engineering, September/October.

Sen, R., Mullins, G. and Salem, T. (2002). ‘Durability of E-glass/Vinylester Reinforcement in alkaline solution’, ASI Structural journal Vol 99,
pp 369-375.

Tomosawa, F and Nakatsuji, T (1997) ‘Evaluation of ACM reinforcement durability by exposure tests’, Non-metallic (FRP) reinforcement for
concrete structures, Proceedings 3rd International Symposium, Sappoto, (1997) Vol. 2, pp139-146.

Triantafillou, T.C. and Meier, U. (1992) ‘Innovative design of FRP combined with concrete’, Proceedings 1st Int. Conf. Advanced Composite
Materials for Bridges and Structures (ACMBS). Sherbrooke, Que., pp491-499.

Sheard, P., Clarke, J.L., Dill, M., Hammersley, G. and Richardson, D. (1997). ‘EUROCRETE – Taking account of durability for design of
FRP reinforced concrete structures’- non-metallic (FRP) reinforcement, for concrete structures, Proceedings of 3rd International Symposuim,
Sapporo, (1997), Vol 2, pp 75-82.

Uomoto, T. (2000). ’Durability of FRP as reinforcement for concrete structures’ Advanced Composite Matrerials in Building and Structures 3rd
International Conference,Ottawa, Ontario, Canada, (2000) pp3-14.

Van Erp, G., Cattell, C and Heldt, T. (2003). ‘Fiber Composites in Civil Engineering – An opportunity for a novel approach to traditional
reinforced concrete concepts’ Proc. Conference in Civil Engineering – Challenges of Concrete Construction VI – Composite Materials in
Concrete, Held at University of Dundee 5-11 September 2003.

Yu, T., Wong, Y L, Teng, J G, Dong, S.L. and Lam, E S S. (2006) ‘Flexural behaviour of hybrid FRP-concrete-steel double skin tubular
members’ Accepted for publication Journal of Composites for Construction, ASCE.

Zhang L, Hollaway, L C, Teng J-G and Zhang, S S. (2006) ‘Strengthening of Steel Bridges under Low Frequency Vibrations’ Accepted for
presentation at CICE 2006, Florida, USA.

10
Third International Conference on FRP Composites in Civil Engineering (CICE 2006)
December 13-15 2006, Miami, Florida, USA

FUTURE OF FRP IN FAR EAST

UEDA Tamon
(Professor, Hokkaido University, Sapporo, Japan)

ABSTRACT
This paper briefly introduces the present situation and future of FRP in construction in the Far East, namely Japan,
Korea and China, which is a major contributor to FRP in construction in the world. The present situation includes
statistics, practical applications, major researches and guidelines/standards. Future directions of FRP in construction
in the Far East are presented.

KEYWORDS
FRP reinoforcement, FRP sahape, Far East, practical application, research, design guideline, standard

1. INTRODUCTION
The Far East including Japan, Korea and China is a major region in the world to produce fibers, which are used for
fiber reinforced polymer (FRP) in construction. Japan is a country to produce 17,600 ton of PAN type carbon fiber,
which is three quarters of the world production in 2003 (Ishihara and Shibaya 2005). Japan produced 23,000 ton of
para-aramid fiber, which is equivalent to USA production of 22,700 ton in 2003 (Ishihara and Shibaya 2005). In
2003 Far East produced 1,250,000 ton of glass fiber (350,000 ton in Japan, 250,000 ton in Taiwan, 150,000 ton in
Korea and 500,000 ton in China) which is more than one third of the world production, 3,200,000 ton (Tazawa
2004). China is a major country for FRP products which produced 1,740,000 ton in 2004, while Japan produced
342,000 ton in 2004 (JRPS 2006). FRP in construction shared 32 % of the FRP production in China and 18 % in
Japan. Japan is the pioneer country to implement guidelines and standards for FRP reinforcement for concrete
(JSCE 1997 and JSCE 2001), while Korea is a leading country to introduce FRP bridges (Kim 2005)

This paper briefly introduces the recent situation and future of FRP in construction in the Far East, namely Japan,
Korea and China. FRP in construction means FRP reinforcement for concrete and FRP shape. Types of fiber in
FRP include carbon, aramid, glass and some new types of fiber.

2. FRP IN JAPAN
As the pioneer country for application of FRP as reinforcement for concrete and geo structures, Japan has been
leading the world in terms of amount of FRP reinforcement used and number of application cases. The amount of
FRP rods and cables does not show increase recently. One of its popular applications is ground anchor. FRP sheet,
however, shows increase in its amount used and application cases (see Figure 1). Mostly they are with carbon fibers
and over 70% of the total cases are for seismic retrofitting of both bridges and buildings. Construction cost of
seismic retrofitting with FRP jacketing has become competitive to steel and concrete jacketing (Ueda 2005).

Photo 1 shows a typical example of seismic retrofitting by CFRP sheet, in which the reduction in construction cost
due to the lightness was the reason for its application. Recently FRP sheet application for repairing deteriorated
concrete surface to prevent concrete pieces from falling apart is getting popular. A good application example of
FRP reinforcement for concrete was seen in pedestrian bridge in the ocean (Photo 2). This concrete bridge is steel-
free due to severe corrosive environment. Application of FRP shape to actual structures in Japan can be seen in only
two cases. One is GFRP pedestrian bridge under severe corrosive environment due to salt attack in Okinawa (Photo
3), and the other is strengthening of existing reinforced concrete bridge deck by adding GFRP beams underneath.

11
 160
Carbon fiber sheet 140
140
Aramid fiber sheet 130
120
120 120
Amount (104 m2)

105
100 100 100
90
80
60
40
20
20
6 2.9 4.3 5.7 6.4
3 0.1 0.3 1.8 2.4 2.6
0 0.1 0
1992
1993
1994
1995
1996
1997
1998
1999
2000
2001
2002
2003
2004
Year

Figure 1: FRP Sheet Amount Used in Japan Photo 1: Seismic Retrofitting by CFRP Sheet

Photo 2: Steel-free Concrete Bridge with FRP

Reinforcement Photo 3: First FRP Bridge in Japan

Japan is also a pioneer to introduce guidelines and standards for design, materials and construction. The followings
are the first in the world for FRP rod/cable as concrete reinforcement and FRP sheet for retrofitting concrete
structures for which the English versions are available:
• Recommendation for design and construction of concrete structure using continuous fiber reinforcing materials,
which includes quality specifications and test methods (JSCE 1997)
• Recommendation for upgrading of concrete structures with use of FRP sheets which includes test methods
(JSCE 2001)
Japan is chairing ISO/TC71/SC6 “Non-traditional reinforcing materials for concrete structures”, which has been
drafting standard testing methods for FRP rod/cable/gird and sheet.

In Japan research on FRP in construction is not as active as in the past. However, there are still innovative studies
going on and the followings are examples:
• Prestressing technique with PBO sheet for strengthening bridges
• Seismic retrofitting by fiber sheet (PAF sheet and PET sheet) with high fracturing strain
• Hybrid FRP beam consisting of hybrid C-GFRP flanges and GFRP web as a government funded project

3. FRP IN KOREA
In Korea major practical applications of FRP in construction are external bonded FRP sheet and FRP bridge deck.
Disastrous accidents in early 1990’s, which were the collapse of Sung-Su Bridge over Han River and of Sam-Pung
Department Store, made engineers and researchers in civil engineering start paying serious attention to repair/retrofit
of infrastructures. Maintenance works has been keeping 26 % or more of the overall construction works since 2003.
One of the repair/retrofit methods used is external bonding with FRP.

Statistics on papers published in the Proceedings of Korea Concrete Institute since 1997 shows steady increase in
number of papers on FRP and reached 10.4 % of the total number in 2005. Among the papers on FRP 82.5 % were

12
related to strengthening, while papers on FRP structural member were 14.6 %. Most popular research target was
carbon fiber (59 %) followed by glass fiber (30 %) and FRP sheet (58 %) followed by FRP plate (18 %).

A loading test on actual concrete bridge girders, which was severely deteriorated and retrofitted by external bonded
FRP as shown in Photo 4, was carried out in order to verify the proposed standard guidelines. On the other hand,
seismic retrofitting technique for piers using FRP wrapping automation device is under development (Photo 5). The
wrapping and gluing glass fiber is done automatically by the device. Since this process secures the continuity of the
fiber, the confinement stress can be effectively provided. This technique has been applied in the seismic retrofit of
two bridges; Seomjin Bridge on a national highway and a bridge on the 88 expressway.

Photo 5: Seismic Retrofit of RC Highway Bridge Pier

Photo 4: Retrofit of RC Highway Bridge Girder by by Automated GFRP Wrapping Machine
External Bonded FRP
In Korea there are 8 bridges with FRP deck, two of which are under construction. Six of them are with steel girder,
while the remaining two are with concrete girder (see Photo 6). The first one, Beoncheon Bridge of Junbu
Expressway was constructed in 2001 with a span of 8 m, while Nucha Bridge in Busan under construction is the
largest in the world with a length of 300 m.

Photo 6: Gwanghyang Harbor Bridge (FRP Deck with Figure 2: FRP-Concrete Composite Deck
Steel Plate Girder)

There are various researches on FRP in construction as follows:

• Sheet of basalt fiber produced by molten basalt for better fire resistance
• Near surface mounted method using T-shape CFRP plate
• Wedge-type mechanical anchor for CFRP plate
• FRP-concrete composite deck as “BRIDGE 200” project from 2002 to 2006 (see Figure 2)
• Concrete-filled FRP composite column (cast-in-place type and precast type)

13
• Hybrid GFRP rod with fiber optic as sensing element and winded glass fiber for bonding (see Figure 3)

Connector for Analysis of

External load structural data
FOS

FOS system in FRP rod Data acquisition

system
(a) Hybrid GFRP rod
(d) Schematic view of smart monitoring for concrete structure

(b) Surface detail (c) Lead-line of FOS (e) Cross section of hybrid GFRP rod

Figure 3: Hybrid GFRP Rod with Fiber Optic Sensor (FOS)

4. FRP IN CHINA
In China, as the country where a FRP bridge was constructed for the first time in the world, there are various
practical applications. The first FRP bridge was constructed in Miyun, Beijing in 1982. It is a highway bridge
consisting of six GFRP sandwich beams with a span of 20.7 m and still in good service condition (see Photo 7). A
recent example of bridge with FRP is a cable stayed bridge with a length of 51.5 m where the stayed cables are
CFRP (Photo 8). The cable stayed bridge was constructed in Zhenjiang, Jiangsu in 2004, in which 8 pairs of CFRP
cables were used.

Other examples are working platforms and backwater panels. A 10,000 m2 GFRP working platform was completed
at a mine in Gansu Province, which consists of FRP frame structures with connecting steel bolts and 12m height.
GFRP backwater panels, which are pultruded GFRP sandwich panels, have been installed on river bank about 40 km
long in the flood season.

Photo 7: World First Highway FRP Bridge in Beijing Photo 8: CFRP Cable Stayed Bridge in Jiangsu
(photographed by Feng Peng and Ye Lieping) (photographed by Ye Zhang)

Typical researches on FRP are as follows:

• FRP-bonding behaviors of FRP strengthened RC elements

14
• Behaviors of FRP confined concrete columns
• Anchorage system and dynamic behavior of CFRP cables
• High-performance FRP bridge decks
• Long-span FRP woven web structures (Photo 9)
• FRP-aluminum hybrid structural elements (Photo 10)
• FRP-concrete hybrid bridge decks

Photo 9: Long-span FRP Woven Web Structures Photo 10: FRP-aluminum Hybrid Structural Element

Chinese market of FRP in construction is quite active recently. The production of FRP in construction shared 32 %
of the total FRP production of 1,740,000 ton in 2004, which was similar to that in USA and much greater than in
Japan. In the recent decade, approximately 10,000,000 m2 of FRP sheets were used to strengthen RC structures,
which is comparable to the amount used in Japan for the same period.

For more reliable application of FRP, governments in China have introduced and will introduce relevant guidelines
as follows:
• Technical specification for strengthening concrete structures with carbon fiber reinforced polymer laminate,
CECS 146:2003, published in 2003
• Design code for strengthening concrete structure, GB 50367-2006, published in 2006 (including the chapters of
FRP materials and FRP strengthening)
• Code for applications of advanced fiber reinforced polymer composites in construction (in drafting), including
FRP strengthening concrete elements and masonry structures, FRP bar reinforced concrete elements, FRP
prestressed concrete elements, and FRP-concrete hybrid elements

Further information is also available in previous papers (Ye 2003 and Ye 2004).

5. FUTURE FOR FRP IN CONSTRUCTION AND CONCLUDING REMARKS

Japan Society of Civil Engineers (JSCE) set up Committee on Utilization of Innovative Structural Materials in 2004
upon the request of a group of companies lead by Shimizu Corporation and Toray Industries Inc. The Committee,
chaired by the author, presented the final report in which the following points were presented for market
enhancement for FRP in the near future (JSCE 2006):
(1) Shield Tunnel Segments: Trial design calculation of FRP-concrete composite segment and FRP segment were
made for a small size tunnel for swage and a large size tunnel for highway. In order to utilize high strength of
FRP, limit state design method is preferable to allowable stress design method. Segment thickness can be
reduced by applying FRP in comparison with conventional RC and steel segments. Trial LCC calculation
showed that FRP-concrete composite segment with thin layer of FRP (0.2 mm) would be more economical than
RC segment for the small size sewage tunnel.
(2) Bridge and Pedestrian Decks: As a substitute of steel bridge deck, FRP bridge deck showed slight increase in
the initial cost but reduced the construction time. A FRP bridge deck substituting concrete bridge deck could
reduce the construction time significantly. A new type of hybrid FRP beam consisting of hybrid C-GFRP flanges
and GFRP web was considered for a pedestrian deck and showed possibility of significant reduction in

15
 construction time. For the cases where construction time is limited, FRP deck and beam would provide a feasible
solution.
(3) Coastal and Harbor Structures: CFRP sheet pile could be a feasible structural element for coastal and harbor
structures. However, strength under impact should be investigated further. For a life cycle shorter than 50 years
high initial cost with the CFRP sheet pile cannot be a better solution than conventional steel sheet pile even after
maintenance cost for the steel sheet pile is considered. Reduction in production cost of the CFRP sheet pile and
in construction cost due to the lightness of CFRP sheet pile should make it more feasible.
(4) LCC and Environmental Impact: Since FRP is a highly durable material under corrosive environment, trial
LCC comparison was made for a pedestrian bridge situated under severe salt attacking environment. A FRP
bridge with special features to reduce initial cost, which were obtained through the experience of constructing the
first FRP pedestrian bridge in Japan, gave a better LCC than a prestressed concrete bridge with epoxy coating of
steel reinforcements. Estimated emission of CO2 in the case of the FRP pedestrian bridge is less than that in the
case of the PC bridge because the emission during construction in the former case can be reduced due to the
smaller sub-structure and the lighter super-structure.

The high material cost in comparison with conventional materials, such as concrete and steel, is the primary
obstruction for FRP application in construction. In order to overcome this drawback, we need more precise cost
estimation with consideration of the lightness of material and LCC, which is different from the conventional cost
estimation way. Development of rational design method for FRP may be also a way to ease the material cost barrier.
Both precise cost estimation and rational design method could enlarge FRP market for durable structures and
architectural structures as seen in Korea and China. Another suggested direction is to explorer new inexpensive
fiber materials, which may not have high strength/stiffness but other good features such as high fracturing strain.

6. ACKNOWLEDGMENT
Chapter 3 was prepared based on the information provided by Prof SIM Jongsung of Hanyang University, Korea, Dr
PARK Cheolwoo of Kangwon National University, Korea and Dr KIM Byun-Suk of Korea Institute of Construction
and Technology, while Chapter 4 was prepared based on the information provided by Prof YE Lieping and Dr
FENG Peng of Tsinghua University, China. The author would like to show his sincere gratitude for kindly
providing him the related information for this paper.

7. REFERENCES
Ishihara, H. and Shibaya, M. (2005). “Present situation of super fiber and technology for production of textile”.
Journal of Textile Engineering, The Textile Machinery Society of Japan, Vol.58, No.1 (in Japanese).
Japan Reinforced Plastics Society (JRPS). http://www.jrps.or.jp/index.html (in Japanese).
JSCE Committee on Utilization of Innovative Structural Materials (2006). Utilization of Innovative Structural
Materials for Civil Engineering Fields, JSCE (in Japanese)
JSCE Research Committee on Continuous Fiber Reinforcing Materials (1997). “Recommendation for design and
construction of concrete structure using continuous fiber reinforcing materials”. Concrete Engineering Series, Japan
Society of Civil Engineers, No.23.
JSCE Research Committee on Upgrading of Concrete Structures with FRP Reinforcement (2001).
“Recommendation for upgrading of concrete structures with use of FRP sheets”. Concrete Engineering Series,
Japan Society of Civil Engineers, No.41.
Kim, B. S., Cho, J. R. and Kim, S. T. (2005). “Korean research and application status of FRP in bridges”.
Proceedings of KCI-ACI Joint Seminar.
Tazawa, H. (2004). Document prepared for JSCE Committee on Utilization of Innovative Structural Materials (in
Japanese).
Ueda, T. (2005). “FRP for Construction in Japan”, Proceedings of JSCE-CICHE Joint Seminar on Concrete
Engineering in Mongolia, 19 May 2005, Ulan Batar, pp.54-68.
Ye, L. P., Feng, P., Lu, X. Z., Qian, P., Lin, L., Huang, Y.L. and Hu, W.H. (2004). “FRP in China: the state of FRP
research and design guidelines in construction”. Proceedings of 2nd International Conference on FRP Composites
in Civil Engineering (CICE2004), Adelaide, Australia, pp.109-119.
Ye, L. P., Feng, P., Zhang, K., Lin, L. and Hong, W.H. (2003). “FRP in civil engineering in China: Research and
applications”. Proceedings of the 6th International Symposium on FRP Reinforcement for Concrete Structures
(FRPRCS-6), Singapore, World Scientific, pp.1401-1412.

16
Part I. Bond and Development
Third International Conference on FRP Composites in Civil Engineering (CICE 2006)
December 13-15 2006, Miami, Florida, USA

BOND TESTS ON CONCRETE AND MASONRY BLOCKS EXTERNALLY

BONDED WITH FRP
Francesca Ceroni
(Assistant Professor, Department of Engineering, University of Sannio, Benevento, Italy)

Marisa Pecce
(Full Professor, Department of Engineering, University of Sannio, Benevento, Italy)

ABSTRACT
The paper is a preliminary presentation of a study focused on definition of the bond properties of carbon and glass fibers
externally bonded to strengthen concrete and masonry elements. An experimental program is now in progress starting from
the analysis of previous bond tests: monotonic and cyclic loads will be applied on concrete and masonry specimens according
to a pull-out test procedure considering various configurations of FRP sheets, anchorage systems, environmental temperature.
Comparisons of available experimental tests with existing debonding formulations are developed to check their effectiveness.

KEYWORDS
FRP laminates, bond behavior and strength, testing procedure, concrete, masonry.

1. INTRODUCTION
Different strategies can be pursued in repair and strengthening of existing structures. Innovative techniques based on Fibre
Reinforced Plastic (FRP) materials appear to be interesting alternatives to traditional solutions; along with high structural
effectiveness, composite materials are light and easy to install, their application does not imply loss of space and, in some
cases, it can be performed without interrupting the use of the structure. However high strength of FRP can not be used since
interface failures is typical issue of FRP composites. Therefore strength and stress distribution of bond between FRP are
central aspects especially at the anchorage or at discontinuities as cracks in concrete structures or mortar joints in masonry
structures. Generally debonding occurs in the support materials characterized by a lower tensile strength, so that bond strength
is strictly related to the tensile strength of the support material and to the bond stress distribution at the interfaces, that depends
on stiffness of adhesive and fibres. Cracks and irregularities of the surface could represent weak points for the bond behavior
due to concentration of stresses. For this reasons the bond strength in applications on masonry elements is more difficult to
prevent due to greater variability and uncertainness of masonry characteristics, to the irregularities caused by mortar joints or
unevenness of the surface. In general no standard procedure has been developed to evaluate the bond strength, even if a lot of
schemes have been proposed and tested by various researchers (Yao et al., 2005). The bond law in terms of shear stress-slip at
the interface is usually assumed bilinear (Lu et al., 2005); the experimental calibration of the factors is essential and
particularly difficult for defining the descending branch of this bilinear law. Furthermore in seismic retrofitting cyclic bond
stresses at the interface have to be considered, therefore it is necessary to analyze the bond strength degradation and the
efficiency of anchorage systems under cyclic loads to evaluate the effective performances.
In this paper a summary of experimental results obtained by the authors with different loading patterns is reported, comparing
the experimental maximum loads and the theoretical ones given by various formulations. A wide experimental program in
progress is described, aimed to examine the bond behavior under cyclic loads and temperature effects, evidencing also the
effect of various type of anchorage systems, for FRP applications on concrete and masonry.

2. THE EXPERIMENTAL PROGRAM

2.1 Results of previous experimental tests
Experimental tests have been realized according to different test procedures (Pecce et al., 2003) to investigate the
bond behavior at the interface between masonry or concrete and FRP laminates. In the following the results of
experimental tests already carried out are summarized and the experimental program planned is presented.
2.1.1 Bond tests on concrete specimens
Tensile tests were realized on concrete blocks connecting two concrete prisms on two opposite sides by one carbon sheet
with various bonded lengths and a fixed width of 100mm. On one side the laminate is anchored with a transversal sheet in

17
order to have the peeling on the other side where strain gauges are glued. Tensile load is applied by gripping steel bars
embedded at the ends of the concrete prisms. Four carbon FRP wet lay up products were used: the fiber thickness was
variable between 0.130 and 0.165mm and the Young’s modulus of fibers was 230GPa, as declared by producers, while the
concrete strength defined by compressive tests on cubes was 25MPa. Beam tests were also realized according to the beam-
test scheme usually carried out to evaluate bond properties of steel bars in concrete elements. Two prisms were linked at
the bottom side by FRP sheets and the two blocks are connected at the top by a steel hinge allowing rotation. The load was
applied in two points. Two carbon FRP wet lay up products were used with fiber thickness of 0.130 and 0.165mm and
Young’s modulus of 230GPa, while the concrete strength defined by compressive tests on cubes was 29MPa. In Table 1
the following geometrical properties of specimens are reported: width, B, height, H, and length, L, of specimen, width and
bonded length of fibers, Bf and Lb. The experimental failure modes and the maximum loads are also reported and the last
ones are compared with various theoretical debonding loads (fib bulletin 14, 2001; Chen & Teng, 2001;
CNRDT200/2004) reported in the following in terms of maximum tensile load (N) or tensile stress (f):
1) fib bulletin 14 (2001): N fa ,max = αc1k c k b b E f t f f ctm α = 0.9, c1 = 0.64, kc = 0.67, c2 = 2
πL
2) Chen & Teng (2001): N fa , max = 0.427βpβL b f Le f 'c ; L e = E f t f ; bL=1 for Lb>Le, β L = sin for Lb<Le
f 'c 2L e
1 2E f ΓFK ( 2 − b f / b)
3) CNR DT200/2004 (2004): f fdd = ; ΓFK = 0.03k b f ckf ctm ; k b =
γ f ,d γ c tf (1 + b f / 400)
The safety factors for debonding failure, γf,d, and for concrete, γc, are assumed equal to 1 in order to compare the
provisions with the experimental results. For tensile tests a variability of failure loads is registered varying the
products used with a tendency to increase at growing the bonded length. The average test–to–predicted bond
strength ratio and its standard deviation are respectively 1.14 and 0.34, 1.01 and 0.44, 1.06 and 0.31 for formulations
1), 2) and 3) for tensile tests, while are 1.83 and 0.22, 1.57 and 0.18, 1.67 and 0.20 for beam tests. The experimental
results are strongly under estimated when beam test is used, due to type of concrete crashing involving a larger
volume of material. The considered formulas give comparable results, if the safety factors for design are not
considered; besides it would be specified if the average or characteristic values of parameters have to be introduced.
Table 1. Experimental tests on concrete specimens strengthened with CFRP
B H L Bf Lb Fmax Fib bulletin 14 Chen & Teng CNR DT-200/2004
Specimen Test Failure
B

[mm] [mm] [mm] [mm] [mm] [kN] [kN] [kN] [kN]

C5_1 Tensile 150 150 300 100 50 12.9 DB 10.0 12.0 10.8
C5_2 Tensile 150 150 300 100 50 7.3 P 10.0 12.0 10.8
C5_3 Tensile 150 150 300 100 50 7.3 DB/P 9.5 11.6 10.3
C5_4 Tensile 150 150 300 100 50 9.5 P 10.0 12.0 10.8
C10_1 Tensile 150 150 300 100 100 16.8 DB/P 12.5 15.8 13.5
C10_2 Tensile 150 150 300 100 100 18.3 P/CR 12.5 15.8 13.5
C10_3 Tensile 150 150 300 100 100 15.2 DB 11.1 14.0 12.0
C10_4 Tensile 150 150 300 100 100 14.3 DB/P 12.5 15.8 13.5
C15_1 Tensile 150 150 300 100 150 12.5 DB/P 12.5 15.8 13.5
C15_2 Tensile 150 150 300 100 150 15.3 DB 12.5 15.8 13.5
C15_3 Tensile 150 150 300 100 150 7.4 DB 11.1 14.0 12.0
C15_4 Tensile 150 150 300 100 150 22.5 P/CR 12.5 15.8 13.5
C10_1c Beam test 150 150 230 100 100 19.7 SH 11.7 14.6 12.8
C15_1 Beam test 150 150 230 100 150 25.0 SH 11.7 14.6 12.8
C15_1c Beam test 150 150 230 100 150 20.6 SH 11.7 14.6 12.8
C10_2 Beam test 150 150 230 100 100 21.8 SH 13.2 16.5 14.6
C10_2c Beam test 150 150 230 100 100 20.7 SH 13.2 16.5 14.6
C15_2 Beam test 150 150 230 100 150 27.3 SH 13.2 16.5 14.6
C15_2c Beam test 150 150 230 100 150 25.7 SH 13.2 16.5 14.6
DB = superficial debonding, P = debonding with concrete detachment, CR = concrete failure, SH = shear failure of block
2.1.2 Bond tests on tuff specimens
Various types of set-up were used to solve difficulties into applying a tensile load to tuff stones. First tensile tests
(specimen TC1-TC5 in Table 2) were performed gripping the tuff blocks through articulated systems of steel plates. In one
of the set-up, stones were in tension together with fibers (TC5), while in two different ones the laminates were in tension
and the stones were compressed using 2 (TC1 & TC2) or 1 blocks (TC3 & TC4). In tensile tests having the tuff blocks in
compression, debonding occurred with detachment of the not compressed stone thickness; when masonry was in tension
the tuff blocks crashed in tension. Finally beam-tests were performed, with the same scheme described above; in some

18
cases fibers were anchored with transversal sheets at both ends. The mean experimental compressive strength of tuff is
3.3MPa and the Young's modulus is 1850MPa. Carbon and glass wet-lay up systems were used having thickness of
0.167mm and 0.111mm respectively. The elastic modulus were 230 and 74GPa according to manufacturer’s indications.
In Table 2 the geometrical properties of specimens are reported: width, B, height, H, and length, L, of specimen, width and
bonded length of fibers, Bf and Lb, number of layers, n, and fiber type, anchorage system. The experimental failure modes
and loads are also reported and compared with the theoretical formulation suggested by (CNR DT200-2004):
1 2E f ΓFK ;
4) f fdd = ΓFK = c1 f mkf mtm c1=0.015. The safety factors for debonding failure, γf,d, and for
γ γ tf
f ,d m

masonry, γm, are assumed equal to 1 in order to compare the provisions with the experimental results. It is interesting to
noticed that the failure loads of tuff specimens are comparable with the values of concrete, even if the tensile strength of
tuff is about 10% of concrete: this leads to suppose that probably for masonry elements the distribution of stress and the
failure mode are able to involve greater amount of material. Moreover the theoretical provisions are too much conservative
respect to the experimental results, so that the main parameters affecting the strength should be reviewed.

Table 2. Experimental tests on tuff specimens strengthened with CFRP and GFRP
B H L Bf Lb Fmax CNR DT-200/2004
Specimen Set-up n Fiber Anchorage Failure
B

[mm] [mm] [mm] [mm] [mm] [kN] [kN]

TC1 Tensile test 200 260 400 100 200 1 C No 10 DB 3.5
TC2 Tensile test 200 260 400 100 200 1 C No 17 DB 3.5
TC3 Tensile test 240 110 400 100 200 1 C No 23 DB 3.5
TC4 Tensile test 240 110 400 100 200 1 C No 17 DB 3.5
TC5 Tensile test 240 110 400 100 200 1 C No 15 TM 3.5
TC10_1 Beam test 185 110 185 100 100 1 C No 11.8 SHT 3.5
TC15_1 Beam test 185 110 185 100 150 1 C No 14.9 SHT 3.5
TC20 Beam test 250 110 315 100 200 1 C No 17.8 SHT 3.5
TC10a_1 Beam test 185 110 185 100 100 1 C CFRP sheet 15.8 SHT 3.5
TC10_2 Beam test 190 110 175 100 100 1 C No 11 SHT 3.5
TC10a_2 Beam test 190 110 175 100 100 1 C CFRP sheet 14.7 SHT 3.5
TC15_2 Beam test 190 110 175 100 150 1 C No 13.5 SHT 3.5
TC15a Beam test 190 110 175 100 150 1 C CFRP sheet 16.0 SHT 3.5
TG10_1 Beam test 185 110 185 100 100 1 G No 10.5 SHT 1.6
TG10a_1 Beam test 185 110 185 100 100 1 G GFRP sheet 18.2 SHT 1.6
TG10_2 Beam test 190 110 175 100 100 1 G No 12.6 SHT 1.6
TG15 Beam test 190 110 175 100 150 1 G No 16.7 SHT 1.6
TG10a_2 Beam test 190 110 175 100 100 1 G GFRP sheet 15.0 SHT 1.6
TG15a Beam test 190 110 175 100 100 1 G GFRP sheet 18.6 SHT 1.6
TG20 Beam test 250 110 315 100 200 1 G No 9.6 TF 1.6
SHT = Shear failure of stone, TF = Tensile fracture of fiber, DB = superficial debonding, TM = Tensile fracture of masonry, C = Carbon; G = Glass

2.2 The experimental program in progress

The experimental program in progress comprises concrete and masonry blocks externally bonded with carbon and glass
FRP sheets having bonded lengths of 100 and 150mm. The load scheme is an asymmetric pull-out test where the tensile
load is transferred to the laminate by gripping the end of FRP sheets and applying compression to the concrete/masonry
block included in a steel frame (Figure 1). This loading pattern seems to be the best one to calibrate the bond strength and
law (Yao et al., 2005). Carbon and glass wet-lay systems will be used having thickness of 0.165mm and 0.343mm
respectively. The elastic modulus are 230 and 81GPa according to manufacturer’s indications. In Table 3 the following
geometrical properties of specimens are reported: width, B, height. H, and length, L, of specimen, width and bonded
length of fibers, Bf and Lb, number of layers, n, and fiber type, anchorage system and special treatment before or during
B

the test. Effects of width, bonded length, anchorage system, number of layers, temperature treatment (specimens will kept
at a fixed temperature before performing the test) and cyclic loads will be considered.

3. CONCLUSIONS
The lack of a standard experimental procedure for bond test both for concrete and tuff specimens implicates different
results depending on the loading pattern and the specimen geometry, especially for masonry. Beam tests on concrete
specimens give higher experimental bond strength respect to the tensile tests due to the shear failure mode of concrete
blocks (Yao et al., 2005). Based on these previous experiences and a literature review, a new experimental program has
been planned adopting the same set-up for concrete and masonry specimens (single shear test set-up with strengthened

19
element in compression) that should give more stable results with failure modes related to bonding behavior varying
parameters. Parallel tests on concrete and tuff specimens will be performed to check the difference in bond behavior
considering that the failure loads of realized test resulted quite comparable, even if the two materials have a very
dissimilar strength. This will be useful also to check the effectiveness of the formulations predicting the delamination
loads. The available data evidence that the theoretical formulations are safe in most cases for concrete, even if the role
of the safety factors is very important; the Italian code gives good provisions for concrete specimens in tension and too
much safe predictions for masonry elements.

4. REFERENCES
Chen J. F., Teng J. G. Anchorage strength models for FRP and Steel Plates bonded to concrete. Journal of structural
Engineering, ASCE, Vol. 127, No. 7, July 2001.
CNR DT 200/ 2004. Istruzioni per la Progettazione, l’Esecuzione ed il Controllo di Interventi di Consolidamento Statico
mediante l’utilizzo di Compositi Fibrorinforzati, 2005 (in italian).
Fib Bullettin 14. FRP as Externally Bonded Reinforcement of R.C. Structures: Basis of design and safety concept, 2001,
TG9.3.
Yao J., Teng, J.G., Chen J.F. Experimental study on FRP-to-concrete bonded joints. Composites: Part B: Engineering, 36
(2005), Elsevier, pp. 99–113.
Lu, X.Z., Teng, J.G., Ye, L.P., Jiang, J.J. Bond–slip models for FRP sheets/plates bonded to concrete. Engineering Structures,
27 (2005), Elsevier, pp. 920–937.
Pecce M., Ceroni F., Manfredi G., Prota A. Tests and modelling on bond of FRP laminates. JCI International Symposium on
Latest Achievement in Technology and Research on Retrofitting Concrete Structures, Interface Mechanics and Structural
Performance, July 2003 – Kyoto, Japan, pp. 95-99.

Table 3. Experimental tests in progress on concrete and masonry specimens

B H L Bf Lb
Specimen Material n Fiber Anchorage Treatment
[mm] [mm] [mm] [mm] [mm]
2 specimens CR 150 150 400 100 150 1 C No No
+ three tests using anchorage systems at the end (FRP sheet, FRP bar, FRP fan)
2 specimens CR 150 150 400 100 100 1 C No No
+ two test using anchorage systems at the end (FRP sheet)
2 specimens CR 150 150 400 50 150 2 C No No
2 specimens CR 150 150 400 50 150 2 C FRP sheet No
2 specimens CR 150 150 400 100 150 1 C No 80°
2 specimens CR 150 150 400 100 150 1 C No 60°
3 specimens CR 150 150 400 100 150 1 C No Cyclic load
+ three tests using anchorage systems at the end (FRP sheet, FRP bar, FRP fan)
2 specimens CR 150 150 400 100 150 1 G No No FRP
+ three tests using anchorage systems at the end (FRP sheet, FRP bar, FRP fan)

block
Masonry
Concrete or
2 specimens CR 150 150 400 100 100 1 G No No
2 specimens CR 150 150 400 100 100 1 G FRP sheet No
2 specimens CR 150 150 400 50 150 2 G No No
2 specimens CR 150 150 400 50 150 2 G FRP sheet No
2 specimens CR 150 150 400 100 150 1 G No 80°
2 specimens CR 150 150 400 100 150 1 G No 60°
3 specimens CR 150 150 400 100 150 1 G No Cyclic load
+ three tests using anchorage systems at the end (FRP sheet, FRP bar, FRP fan)
2 specimens T 250 110 370 100 150 1 C No No
+ three tests using anchorage systems at the end (FRP sheet, FRP bar, FRP fan) Figure 1.
2 specimens T 250 110 370 100 150 1 C No 80° Experimental set-up for
2 specimens T 250 110 370 100 150 1 C No 60° asymmetric pull-out test
3 specimens T 250 110 370 100 150 1 C No Cyclic load
+ three tests using anchorage systems at the end (FRP sheet, FRP bar, FRP fan)
2 specimens T 250 110 370 100 150 1 G No no
+ three tests using anchorage systems at the end (FRP sheet, FRP bar, FRP fan)
2 specimens T 250 110 370 100 150 1 G No 80°
2 specimens T 250 110 370 100 150 1 G No 60°
3 specimens T 250 110 370 100 150 1 G No Cyclic load
CR = Concrete, T = Tuff, C =
+ three tests using anchorage systems at the end (FRP sheet, FRP bar, FRP fan) Carbon; G = Glass

20
Thir d Int er nati ona l Co nfer ence o n FRP Comp os ites i n Ci vil Eng ine er ing (CICE 200 6)
Decemb er 13 -15 200 6, Mi ami , Flori da, USA

COHESIVE –BRID GING ZONE M ODE L OF THE FRP -CONC RETE

INTERFA CE D EB ONDIN G

Jialai Wa ng
(Assistant Professor, The University of Alabama, Tusc aloosa, Ala bama 33487-0205, USA )

ABSTRA CT
Debon ding along the FPR -concrete interface can lead to premature failure of the structures. In this stud y, a
combined cohesive/bri dging zone model is presented to simulate the debondin g procedure between the FRP and
concrete interface . The crack processi ng zone of the interfa ce is modeled by a cohesive zone model and the particle
interlocking zone of the interface is modeled by a bridging zone model. Closed -form solutions of interfacial stress,
FRP stress and ultimate load of plate d beam are obtain ed for a typical single -lap specime n. Excellent agreements
with experime ntal results have been achieved by this model, which verifies the validation of the proposed novel
bond-slip law.

KEYW ORDS
Cohesive zone model; Fiber reinforced polymer; Strengthening; Concrete ; Debondin g

1. INTRODU CTION

Experimental studies have shown that the stre ss deformation relationship of the FRP -concrete interfa ce is
nonlinear (Chajes et al. 1995; Bizindav yi and Neale 1999 ; Dai et al. 2005;). By considerin g a nonline ar bond -slip
law, it is possible to model the whole debondi ng process of FRP -concrete interface as demonstrated recentl y by
Yuan et al. (2004). In existing cohesive zone debondin g models (Yuan et al. 2004, Wang 2006), a simple bi-linear
bond stress -slip law is used freque ntly. As obs erved in experime ntal studies (Leung and Tung 2006) the FRP-
concrete interface debondin g has two different failure stages, i.e., a crack processi ng (damaging) stage of the bond
followed by a particle bridging (interlockin g) stage until the surfaces of the FRP plate and concrete substrate are
fully separate. Due to the different failure mechanisms at crack processin g and bridging stage, the stre ss-slip law
should be differ ent. In this stud y, a novel nonlinear bond stress -lip law is proposed , in whic h two different linearl y
softening laws ar e employed in to simulate these two stages separatel y.

2. COHESIVE /BRIDGING ZONE MODEL OF INTERFACE DEBONDING

Consider a simply-supported reinforced concrete beam (RC beam) reinforced by an FRP plate subjected to
pointed loads and/or uniform distributed load, as shown in Fig. 1(a) . To simplif y the analysis, onl y a flexural crack
existing at the mid -span of the concrete beam is considere d in this stud y. Since a symmetric load is applied, onl y
half of the structure needs to be analyzed (Fig. 1(b). The geometr y of the cross -section of the plated beam is shown
in Fig. 1(a). Both the concrete beam and FRP plate are modeled as linear elastic simple beams (beam 1 and 2 in Fig.
1, respectivel y). Therefore, t he constitutive laws for these two beams r ead:
N i = C i u i' , M i = - Di wi (1)
'' i = 1, 2
where N i and M i are axial forces and bending mome nts of beam i (i = 1, 2), respectivel y; ui and wi are axial and
vertical displac ements of beam i (i = 1, 2), respectivel y; Ci and Di are axial and bending stiffnesses of beam i (i = 1,

21
2), respectivel y; and Ci = E ibihi, Di = E iIi; E i is the Young’s modulus of beam i (i = 1, 2); bi and hi are width and
height of b eam i (i = 1, 2); Ii is t he mome nt of inertia of the be am i (i = 1, 2).
b1
x h1
z d a2 a1 h2
b2
(a) t f1 k2
N1 N1+DN1
M1 t f2
M h1 N1
N2
t N2 k1
N2+DN2 h2
(c)
M2
Gf t f3
t
(b) k3
t d
II I
I x
x d1 d2 df
a1 (e)
(d)
t
t
III II
I IV III II I
x x
a2 a1u d a2 a1u
(f) (g)

Fig. 1. Interfacial stress of FRP-plated con crete beam . Fig. 2. Cohesive/bri dging bond-slip mode l

Beam 2 is bonded to Beam 1 through the FRP -concrete interface layer whic h is modeled as a large fracture
processing zone with a nonlinear bond -slip law as demonstrate d in many experimental studies (Bizind avyi and
Neale 199 9). Based on the differe nt failure stages of the interfa ce, the whole FRP -concrete interface can be divided
into four different zones as shown in Fig. 1 after full debondi ng occurs. Two different line arly softening laws are
emplo yed in this study to simulate the cohe sive zone (Zone II) and bridgi ng zone (Zone III) separatel y.
Correspondingl y, the proposed bond stres s-slip law has four segments as show in Fig. 2. This non -line ar relationship
can be describe d by the following equation:
ì t f1
ï d d < d1
ï d1
ï t f 1 -t f 2
ït f 1 - (d - d1 ) d1 £ d < d 2
t=í d 2 - d1
ï t f3
ït f 3 - (d - d 2 ) d 2 £ d < d f
ï d f
-d 2
ïî 0 df £d
(2)
The slip along the interf ace d (relative axial displace ment of the top of the FRP plate and the bottom of the concrete
beam) and given by:
(
d = u1 - Y1 w1' - u 2 - Y2 w2' ) (3)
Y1 and Y2 are the dista nces from the bottom of beam 1 and the top of beam 2 to their respective neutral axis. From
the point of view of CZ M, such a nonlin ear relationship is a m aterial propert y of the FRP -concrete interface .

Under exter nal load, interfacial shear stress is develope d along the FRP -concrete interface. Four debonding
stages s hown in Fig. 1(d) to 1(g) are analyzed as follows.

Stage I: Linear ly elastic stage

By assuming the FRP plate a nd concrete beam have the same curvature (Smith and Teng 200 1; Rashe ed and Pervaiz
2002) , the governing equation of shear stress along th e interfac e of FRP -concrete can be obtain ed as:
1 (Y1 + Y2 ) ö
2
t æ 1 t Y +Y
t '' = f ç + + ÷b2t + f 1 2 M ' (4)
d1 è C1 C2 (D1 + D2 )ø
ç ÷ d 1 D1 + D2
Considering consta nt bending Moment M in con ventional sin gle -lag specimen, the shear stress and axial force of the
FRP plate in this stag e can be expressed as:

22
 x b2 A
t = Ae - l x ,
1
N 2 = N 20 + ò b (Dt + t )dx = N
2 C 20 + e -l1x (5)
0
l1
t f1 Y1 + Y2
, C = b ç 1 + 1 + (Y1 + Y2 )
2
æ ö
Where l1 = Cl ÷ , Ct = , N 20 = b2 Ct M .
d1 l 2ç
è C1 C 2 D1 + D2 ÷
ø
(
D1 + D 2 C l2 )
Stage II: Elastic-softening stage

If the load is increase d after reaching the elastic limit, part of the interface turns to soften with the slip and two
regions along th e interfac e are forme d (Fig. 1 ( e)):

Region I: Linearly Elastic Region (d = d1 ):

t = t f 1 e - l1 (x - a1 ) , N 2 = N 20 +
b2t f1
e -l1 (x - a1 ) (6)
l1
where a1 is the softenin g zone size .

Region II: Linearly Soften ing Region (d 1 < d = d 2) :

l2
t = t f 1 cos(l 2 (x - a1 ))+ t f 1 sin (l 2 (x - a1 )), N 2 S = N 20 - b2 æçç sin(l 2 (x - a1 ))- l 2 cos(l 2 (x - a1 ))ö÷÷t f 1 (7)
l1 l2 è l1 ø
1 (Y1 + Y2 ) ö k 2 2
2
t f 1 -t f 2 æ 1
where l22 = ç + + ÷= l1
d 2 - d1 çè C1 C2 D1 + D2 ÷ø k1

Stage III: Elastic -softening-bridging Stage

In this stag e, three differ ent zones exist on th e FRP -concrete interfac e, as shown in Fig. 1(f). T he interfa ce shear
stress an d the axial force of the FRP plate i n Zone I and II can be obtai ned by shifting t he abscissas of corresponding
solut ion in Stage II by a bridging zone siz e a2.
Zone I: Elastic zone:

t = t f 1 e - l (x - a
1 1u -a2 )
, N 2e = N 20 +
b2t f 1
e - l (x - a
1 1u - a2 )
(8)
l1
Zone II: Softening zone:
æ l ö b æ l ö
t = t f 1çç cos (l 2 (x - a1u - a2 ))+ 2 sin (l2 (x - a1u - a2 ))÷÷ , N 2 S = N 20 - 2 çç sin (l2 (x - a1u - a2 ))- 2 cos(l2 (x - a1u + a2 ))÷÷ (9)
è l 1 ø l 2 è l 1 ø
Zone III: Bridging zone:
l3 æ l ö
t = t f 3 cos(l3 (x - a2 )) + t f 1 çç sin(l2 a1u ) + 2 cos(l2 a1u )÷÷ sin (l3 (x - a2 )),
l2 è l1 ø
æ æ ö ö
N 2 = N 20 +
b2 ç - sin (l3 (x - a 2 ))t f 3 + l3 t f 1 ç sin (l2 a1u )+ l2 cos(l2 a1u )÷ cos(l3 (x - a 2 ))÷ (10)
l3 ç l2 è ç l1 ÷ ÷
è ø ø

where l32 =
tf3 æ 1 1 (Y1 + Y2 )2 ö k 3 2
ç + + ÷ = l2
d f - d 2 çè C1 C2 D1 + D2 ÷ø k 2
.
Stage IV: Elastic -softening -bridging -debon ding stage

If the load is increas ed after reaching the debondin g limit, full debonding occurs along the interface (Fig. 1 (g))
and propag ates a dista nce d from the location of the pull end. In this zone, the interfac e shear stres s is zero.
Therefore N 1 and N 2 are constants. The stre ss distribution within zones I, II, and III can be obtain ed by simpl y
shifting d in abscissas in that of elastic -softenin g-bridging stage. Following the same procedure

23
3 EXPERIMENTAL VERIFICATION

As verification, the FRP stress calculated by the present analytical soluti ons for a single-lap shear test in Fig.
1(a) are compare d with experimental result (Pan and Leung 2006) and presented in Fig.3. The parameter in bond
stress-slip law shown in Fig. 2 are chosen as k 1 = 153 MPa/mm, tf1 = 5.2 MPa, k 2 = 4 MPa/mm, tf2 = 4.62 MPa, tf3 =
1.25 MPa, k3 = 0.56 MPa/mm. The total fracture energy Gf = 2.2 N/mm. Fig. 3 shows the present analytical model
has achieved good agreem ents with experimental r esults , which validate s the novel bon d-slip model of this stud y.

4. CONCLUSIONS
In this stud y, a novel cohesive/bri dging zone model is proposed to simulate the FRP -concrete interface
debodin g. This new model uses two different softening laws to describe the nonlinear bond stress -slip law of the
FRP-concrete interface at cohesive zone and bridgi ng zon e, respectivel y. Closed -form solutions of the interfacial
stress, the axial force of FRP plate for differe nt deboding stages. The valid ation of this model is v erified b y excelle nt
agreem ents with experimental data achieved by the new bond stress -slip law. The cohesiv e/bridg eing zone mod el
in this stud y provides a n efficie nt and effective wa y to anal yze more general FRP -concrete interface debondin g.

10000 7000
9.5KN, Ex p. 12KN, Exp.
(a) 9.5KN, Ana.
6000 (b) 12KN, Ana.
8000 15KN, Exp.
11KN, Exp.

strain in FRP (10-6 )

Strain in FRP (10-6 )

5000
11KN, Ana. 15KN, Ana.

13.5KN, Exp. 17.9KN, Ex p.

6000 4000
13.5KN, A na. 17.9KN, A na.

14.51KN, Exp. 3000 19.74KN, Exp.

4000 19.74KN, Ana.
14.51Kn, Ana.
2000
2000
1000

0 0
0 100 200 300 400 500 0 100 200 300 400 500
distance from the pull end (mm) distance from the pull end (mm)

Fig. 3. Comparison between the present sol ution and ex perimental result (Pan and Leung 200 6 ): (a) CFRP
thickness = 0.22 mm ; (b) CFRP thickness = 0. 44 mm

REFER ENCES:

Bizindav yi, B.L. and Neale, K.W., (1999). “Transfer Lengths and bond stren gths for composites bonde d to
concrete”. Journal of composites for constructi on 3, 15 3 -160.
Chajes, M.J., Januszka , T.F., Mertz , D.R. , Thomson Jr., T.A., Finc h Jr., W.W., (1995 ). “Shear stre ngthening of
reinforced concrete beams using extern all y applied composite fabrics ” . ACI Structural Journal 92, 295-303.
Dai, J., Ueda, T., Sato, Y., (2005). “Development of the Nonlinear Bond Stress –Slip Model of Fiber Reinforced
Plastics Sheet–Concrete Interfaces with a Simple Method” . Journal of Composites for Constructi on 9, 52 -62.
Leung, C.K.Y ., and Tung, W.K. (200 6). “Three-parameter model for debonding of FRP plate from concrete
substrate”. Journal of Engineerin g Mechanics 32 , 509 -518.
Pan, J. and Leung, C.K.Y , (2006). “Debonding along the FRP –concrete interface under combine d pulling/peeling
effects”. Engineering Fracture Mechanics. in press.
Rashee d, H.A., and Pervaiz , S.,(2002 ). “Bond slip analysis of fiber -reinforce d pol ymer-stre ngthened beams ”.
Journal of Engineering Mechanics 128, 7 8 -86.
Smith, J .G., and Teng, J.,(2001). “Interfacial stre sses in plate d beams ”. Engin eerin g Structure 23, 8 57-871.
Taljsten, B., (1996 ). :Strengthenin g of concrete prisms using the plate bonding technique ”. International Journal of
Fractures 82, 253 -266.
Wang, J. (200 6) . “Cohesive zone mod el of intermediate crack-induce d debondin g of FRP -plated concrete beam”.
International Journal of Solids and Structures . (in pres s).
Yuan H., Teng, J.G., Seracino, R., Wu, Z.S., and Yao, J ., (2004). “Full -range behavior of FRP -to -concrete bonded
joints ”. Engineering Structures 26, 553-565.

24
Third international Conference on FRP Composites in Civil Engineering (CICE 2006)
December 13-15 2006, Miami, Florida, USA

CONSTITUTIVE MODEL FOR TIME-DEPENDENT BONDING AND

DEBONDING ALONG FRP-CONCRETE INTERFACE
Hesham Diab
PhD Candidate, Ibaraki University, Hitachi, Ibaraki, Japan

Zhishen Wu
Professor, Ibaraki University, Hitachi, Ibaraki, Japan

ABSTRACT
The time-dependent behavior of FRP-concrete interface is big common for concrete structures strengthened with
externally bonded fiber reinforced polymer (FRP) sheets or plates. This paper is devoted in developing a new
nonlinear viscoelastic model for studying the long–term behavior of FRP-concrete interface. The model has the
ability to describe creep behavior of FRP-concrete adhesive layer and creep fracture propagation along the FRP-
concrete interface. The creep of FRP-concrete interface is taken into account by using Maxwell’s generalized
rheological model through a step-by-step time increment model. The nonlinear time-dependent behavior of adhesive
layer depends on the time-dependent degradation of the bond strength and fracture energy. The proposed model has
been calibrated by using the results of the double lab shear creep test specimens to verify its validity. The results
demonstrate the reliability of the proposed models and their capability to predict the time-dependent propagation of
debonding. The results show also the ability of this model to predict a wide range of the creep fracture not only
under low sustained loading but also under high sustained loading.

KEYWORDS: FRP-concrete interface, Creep fracture Propagation, Nonlinear viscoelastic model

1. INTRODUCTION
For realistic predictions of the long-term behavior of structures externally strengthened by FRP sheets not only the
instantaneous response of FRP-concrete interface under static loadings, but also the influence of long-term
deterioration processes is of important. In this case it has been shown, that basic creep strongly affects the
serviceability of concrete structures and contributes to a reduction of lifetime and carrying capacity. Although in the
last two decades great achievements have been directed to understand the bonding and debonding behavior of FRP-
concrete interface and develop different evaluation methods based on bond strength (Wu et al. 2002), it is difficult
till now to understand and model the creep fracture at FRP-concrete interface. The present study extends the
capability of linear viscoelastic model presented by the authors to simulate the interfacial time-dependent debonding
along FRP-concrete interface. This new model has the ability to predict and to evaluate the initiation of micro-cracks
and formation of a macro-crack i.e predicts the creep Fracture propagation at FRP-concrete interface. A model able
to reproduce these nonlinear effects is described in the following part of this contribution in combination with the
previously linear viscoelastic model (Wu, and Diab, 2005).

2. OBJECTIVES AND FORMULATION OF CONSTITUTIVE RELATION

The previous rheological model describes satisfactorily the creep of FRP-concrete interface for low sustained loads
i.e. under the debonding limit. This previous model describes only a single feature of the much complex behavior of
FRP-concrete interface. As time increases the shear stress along FRP-concrete interface decreases due to stress
relaxation function that completely define an ideal viscoelastic material which means the debonding will not occur
after the short-term loading based on the bond strength criteria. Recent creep experimental tests show that the FRP-
concrete interface behaves in a much complex manner than viscoelastic behavior. The creep of the adhesive leads to
initiate and increase the debonding along FRP-concrete interface. This result restricts the applicability of linear
viscoelastic model to fully represent the real behavior of FRP-concrete interface especially under high sustained
loads. Therefore, modifications that allow for creep fracture in the linear viscoelastic model are necessary. Clearly

25
Studies of the fracture test data of concrete structures strengthened with FRP sheet showed that the debonding
models with softening shear stress-slip relation are inevitable for describing the FRP-concrete interface. Although
the interfacial fracture energy is affected by the mechanical property of adhesive and by the concrete strength, it is
difficult to say that it affected by time. Based on this discussion, a modified constitutive relation must satisfy the
following requirements:
1. The constitutive relation must be reduced to that for linearly viscoelastic aging creep and it should predict when
the micro crack is initiated, If the initial load is under the debonding limit.
2. In the absence of creep (e.g., for very fast deformation), the constitutive relation must reduce to an algebraic
bond-slip relation which describes softening behavior.
3. The fracture energy of FRP-concrete interface remains constant and does not depend on the time.
Requirement 1 is crucial. It makes it difficult to use debonding failure models which depend only on the bond
strength criteria without any external equation because the shear stress descends with time due to the relaxation
function. Therefore, a new factor should be considered for the bond failure model coupling with linear relaxation to
attain the requirement 1. The rheological model for describing the nonlinear behavior and for attaining the previous
requirements can be effectively be represented through a series of analogical models shown in Figure 1(a). A system
with N Maxwell’s chains that models the increment in creep and elastic slip is placed in series with an element that
schematizes the softening model. The softening element describes the softening diagram after attaining the micro
debonding, and their respective deformations, S e , S c and S p must be added. So we may write (see Figure 1(a))
S  Se  Sc  S p (1)
e c
where S , S , S and S p
= total , elastic, creep and softening slip deformation at FRP-concrete interface element
respectively.
The increment in elastic and creep slip deformation has been presented in our previous research (Wu, and Diab,
2005) for linear viscoelastic behavior of FRP-concrete interface.

 Local Bond Stress

f
1

 ft /  f
0.8
S 0.6
K1 K2 Kn
1
K0
2  n Sc S K 0.4

0.2
1 Gf Slip S
p


S 0
Sd Sft0 0 20000 40000 60000 80000
Time (Min.)

(a) (b) (c)

Figure 1. A nonlinear viscoelastic model, Linear Model and time-dependent behavior of maximum shear stress

Most of debonding models for representing short-term behavior of FRP-concrete interface depend on the bond stress
criteria, which is difficult to use solely for time-dependent models as mentioned before. Therefore, a new factor
should be added to these types of debonding models.

2.1 Fictitious Bond Stress Criteria Model (FBCM)

For the present investigation, a new interfacial debonding model was developed that attempts to combine bi-linear
debonding model shown in Figure 1(b) with time-dependent bond strength. The new fictitious bond stress criteria
model (FBCM) depends on the assumption that the bond strength is decreased with time and the fracture energy is
being constant with time where the fracture energy is represented by the area under the   s curve. The form of
bond strength time-dependence equation is taken to be,
    t



 i  i 1  e 
n
ft
 i

   t  
(2)
f i 0

where n is the number of terms of the series,  i are the relaxation times, t is the time increment,  i is a
coefficient define the percent of reduction of the bond stress within each relaxation time, and  f , ft are bond
strength and time-dependent bond strength respectively,
One important point in choosing the parameters of equation 2 is that this equation must be able to realistically model
the time-dependent maximum bond stress where the FRP-concrete interface will be initiated to debonding if the
time-depend shear stress increased than this value at a generic time. These parameters depend on the time-

26
dependent parameters of concrete, adhesive and FRP sheets. From experience, The best way to find these parameters
( n ,  i ,  i ) is by solving the linear viscoelastic problem i.e. neglecting debonding and then drawing the time-
dependent maximum shear stress for FRP-concrete interface as shown in Figure 1(c), from this relation, the
relaxation times and reduction coefficient can be obtained as discussed in the previous work.
Base on the requirement 3, the fracture energy which is represented by the area under the curve is being constant and
the maximum time-dependent slip is calculated as follows:

s ft  f
 s ft 0

(3)
ft

where s ft0 and sft are maximum slip at instantaneous load (t0) and maximum time-dependent slip at time (t).
It should be pointed out that the value s ft 0 is very important to be obtained based on the fracture energy and
maximum bond strength. The slope of the linear softening part depends on this value, so the small value of
s ft 0 leads to high slope which allows the debonding to evolve quickly and causes difficulties due to the rapid
debonding. However, the different values of s ft 0 based on the same fracture energy and different bond strengths
lead to the same result. It is clear that the exponential form of equation 2 is physically motivated. At short-term
loading, t should be equal zero and in turn the requirement 2 will be attained and due to the linear viscoelastic
model the interfacial bond stress descends with time and the bond strength is also reduced with time which means
requirement 1 will be attained.

3. CALIBARATION OF THE PROPOSED MODEL

The proposed model (FBCM) is calibrated by using the experimental results obtained from double lab shear test
specimens (S23, S16.8, S14 and S9.8, where S23 means that the sustained load is 23 kN) carried out at Ibaraki
University (Takeshi, 2006). A two-dimensional FE analysis is carried out to simulate the experimental specimens
using a commercially available FE code DIANA. The details of the prism specimens are schematically shown in
Figure 2. The FRP-concrete interface is represented by an interface element (L8IF) with zero thickness. Using
supplied subroutine mechanism offered by DIANA, The long-term behavior of adhesive is represented by the
proposed nonlinear viscoelastic model and the linear viscoelastic behavior of the material is represented by a
Maxwell chain formed by five rheological elements in parallel. The relaxation times (  ), given in minutes, are
selected according to (Wu and Diab, 2006), based on actual linear experimental results. The values adopted
are  0   , 1 =14.4,  2 =1440, 3 =14400 and 4 =144000, and the values of stiffness are  0 =0.12,  1 =0.4,
 2 =0.24,  3 =0.2 and  4 = 0.04 (where  n  K n / K ) and K is the total interfacial.
100
Anchoring with FRP sheets
(Circumferential direction) Concrete Block
Clip strain gage 1
5 10 30 30 20 50 40 mm

50 100

5
Clip strain gage 2
mBonded length 200 mm
250 mm

Figure 2 Detail of FRP sheet-concrete bonding joints (Double-lap shear test)

3.1 Comparison between Experimental and Analytical Results

The time-dependent maximum bond strength equation of all specimens is represented by equation (4) which is
obtained based on the previous illustration
 ft
 0 . 4  0 . 2 e  t / 14 . 4  0 . 2 e  t / 1440  0 . 12 e  t / 14400  0 . 08 e  t / 144000 (4)
 f
The fracture energy of FRP-concrete interface is obtained by changing the value of the maximum slip s ft 0 until
obtaining a good agreement between numerical and experimental results while the initial maximum bond strength
remains constant then a comparison between experimental and numerical fracture energy is done. A comparison

27
between experimental fracture energy obtained based on (JSCE, 2001) and that obtained from the proposed model is
shown in Figure 3(a). The proposed model gives a high accuracy at low and high sustained loads as shown in Figure
3(b) which represents time-dependent displacement with semi logarithmic scale for S23 and S16.8 (82 and 60% of
their static capacity, where the static capacity was 28 kN). Figure 3(c) and 3(d) show the time-dependent FRP strain
distribution along bonded FRP sheet. The results demonstrate the ability of the proposed model to simulate the creep
fracture propagation along FRP-concrete interface. It should be mentioned that the divergence between numerical
and experimental strain at time zero for S14 is a result of a macro crack occurred at concrete block due to static load
which it is not considered in this study, where concrete behaves in elasticity. Although it seems that the creep of
FRP-concrete interface stopped after 22930 minutes at the specimen S9.8, the numerical model gives a small
increasing at FRP strain as shown in Figure 3(d), more experimental study is needed to confirm this observation.

1 2
Experimental Experimental
Proposed Model (FBCM) S23
Fracure Energy (N / mm)

0.8 1.6
Proposed model

Displ. (mm)
1.2 (FBCM)
0.6

0.4 0.8
S16.8
0.2 0.4

0 0
S9.8 S14 S16.8 S23 1.0 10.0 100.0 1000.0
Specimens Time (Min.)

(a) Fracture energy (b) Semi-log scale for time-dependent

displacement ( S23 and S16.8)
6000 4000
0 Min.
0 Min.
60 3500
5000 2010
1000 FR P Sheet Strain 3000 22930
FRP Strain (µå)

3000
4000 60000
6200 2500
FBCM (Gf=0.2)
FBCM (Gf=0.324)
3000 2000
.
1500
2000
1000
1000
500
0 0
0 50 100 150 200 0 50 100 150 200
Distancefrom the loaded end (mm) Distancefrom the loaded end (mm)

(c) FRP-shear stress distribution (S14) (d) FRP-shear stress distribution (S9.8)

Figure 3. Comparisons between numerical results for Model (FBCM) with experimental results from double-lap
shear test
4. CONCLUSION
From the analysis presented above, the following conclusions may be drawn:
 The proposed model seems to be able to predict the debonding and creep fracture propagation along FRP-
concrete interface.
 By using the fictitious bond stress criterion model (FBCM), the propagation of the fracture process zone
along FRP-concrete interface can be numerically simulated for high and low sustained loads.

5. REFERENCES
Japan society of Civil engineers (JSCE) (2001).”Recommendations for Upgrading of Concrete Structures with Use
of Continous Fiber sheet.” Concrete Engineering Series 101.
Takeshi Sasaki (2006).”Study on creep behavior of FRP sheets and FRP-concrete interface.” Graduated paper
(supervised by Prof. Wu, Z. S.), Ibaraki University.
Wu, Z. S., and Diab, H. M. (2005).”Interfacial constitutive model for long term behavior of FRP-concrete adhesive.”
Proceedings of International Symposium on Innovation & Sustainability of Structures in Civil Engineering,
Southeast University, China, pp. 1757-1769.
Wu, Z. S., and Diab, H. M.” Constitutive model for time-dependent behavior of FRP-concrete interface”. Journal of
Composites for Construction, ASCE. (Primarily accepted)
Wu, Z. S., Yuan, H., and Niu H. (2002). “Stress transfer and fracture propagation in different kinds of adhesive
joints.” Journal of Engineering Mechanics, ASCE, Vol. 128, 562-573.

28
Third International Conference on FRP Composites in Civil Engineering (CICE 2006)
December 13-15 2006, Miami, Florida, USA

DEBOND CHARACTERISTICS OF CARBON FIBER LAMINATES FOR

BRIDGE REHABILITATION
M. Arockiasamy
(Professor and Director, Center for Infrastructure and Constructed Facilities, Florida Atlantic University, Boca
Raton, Florida, USA)

M. Sivakumar
(Senior Design Engineer, PTE Strand Co Inc., Hialeah, Florida, USA)

M. Shahawy
(President, SDR Engineering, Tallahassee, Florida, USA)

ABSTRACT
Recent bridge inspection statistics nation wide show that more than a third of the United States’ half million
highway bridges are either “structurally deficient” or functionally obsolete”. Bridge components are exposed to
deterioration effects because they are directly subjected to environmental conditions, corrosive action, accidental
vehicular impacts, and cyclic load variations. This may cause defects in concrete bridges including cracking, scaling,
spalling, leaching, chloride contamination, delamination and partial or full depth damage, and lead to posting load
restrictions on the use of bridges. Retrofit and rehabilitation is essential in order to restore the bridge to its original
design load carrying capacity.

Fiber reinforced polymer (FRP) composites are increasingly being used for the repair and strengthening of
deteriorated concrete structural components through adhesive bonding of prefabricated strips/plates and the wet lay-
up of fabric. Concrete structural elements strengthened with FRP reinforcement can fail in different ways. Bond
failure modes have attracted the attention of the designers besides the classical failure modes, such as reinforcing
steel rebar fracture, concrete crushing, or shear failure. The objective of the present study is to evaluate the
interfacial bond strength between the FRP composite strip, adhesive and the concrete under harsh exposure
conditions-namely simulated tidal conditions and freeze-thaw exposure. The bond strength is experimentally
determined by the peel test and the results compared with those based on finite element stress analysis. The fracture
toughness for debonding is evaluated and expressed as the critical strain energy release rate.

KEYWORDS
Debond,Carbon Fiber Laminates, Bridge Rehabilitation

1. INTRODUCTION
The deterioration and critical need for renewal of civil infrastructure has recently been the focus of considerable
discussion among the researchers in North America, Europe and Japan. The repair of concrete beams by the external
bonding of a steel/ composite plate is an effective method to protect concrete and restore a part of the stiffness of the
structure (Meier, et al. 1991; Arockiasamy, et al. 1995). Whether the plate repair is of steel or advanced composite
material, the plate is typically bonded to the concrete using an adhesive. The corrosion of the steel plates continues
to be a problem. The effect of adverse environmental conditions on the externally bonded steel plates can be over
come by using FRP materials which have the distinct advantages of strength, non-corrosive properties and ease of
application with limited interruption to traffic.

29
The fiber-reinforced polymer (FRP) composite laminates have been used to strengthen an aging single-span
reinforced- concrete T-beam bridge in the state of New York (Hag-Elsafi, et al, 2001). Load tests were conducted to
evaluate effectiveness of the strengthening system and investigate its effect on structural behavior, and tests results
were compared with those obtained using classical analysis. The load test was repeated to monitor in-service
performance of the system after two years of service (Hag-Elsafi, et al, 2004). Further inspection using an infrared
thermography camera did not show any significant delamination in the system.

Beside the classical failure modes, such as steel fracture, concrete crushing or shear failure which can appear in
normal RC and PC members, bond failure modes have to be taken into account in a concrete member strengthened
with externally bonded reinforcement (Blaschko M. et al., 1998). These bond failures can occur in the interface
between the externally bonded reinforcement and the concrete body. The following are the possible bond failure
modes:• i) FRP peeling-off at the outermost flexural crack in the uncracked anchorage zone, • ii) FRP peeling-off at
flexural cracks in the maximum bending area, • iii) FRP peeling-off at flexural cracks between the outermost crack
and the maximum bending area, • iv) FRP peeling-off caused by shear cracks and • v) FRP peeling-off caused by the
unevenness of the concrete surface.

The susceptibility of open sandwich beams (concrete/adhesive/steel) was investigated to static fracture by crack
propagation along the adhesive layer, using linear elastic fracture mechanics (Anandarajah, A. and Vardy A. E.,
1985). This study was focused on a simply supported beam with four point loadings. It was found that beams
constructed with a suitable adhesive should not be at risk from such failure, whether they were conventionally
reinforced beams strengthened by an externally bonded plate or plain concrete beams reinforced solely by an
external plate. In addition, it was reported that the beams were relatively insensitive to the thickness of the adhesive
bond layer.

The present study evaluates the interfacial bond strength between the FRP composite strip, adhesive and the
concrete under harsh environmental exposure. Interface crack propagation and peeling of FRP laminate from
concrete was investigated in the experimental work (Baker, W. A., 1999). Multiple specimens were tested using a
specially designed test fixture for characterizing the concrete/FRP laminate debond fracture. The compliance and
critical load for crack propagation were obtained for each specimen. The experimental results are compared with
those based on finite element stress analysis. The fracture toughness for debonding is expressed as the critical strain
energy release rate.

2. FRACTURE ENERGY
The energy balance approach can be used to find the adhesive fracture energy, Gc. The energy criterion for fracture
describes quasi-static crack propagation as the conversion of the work done, Wd, by the external force and the
available elastic energy stored in the bulk of the specimen, U, into surface free energy, γm. The relationship may be
written as
d d
Wd U γ m. A
da da (1)
where dA is the increase in surface area associated with an increment of crack growth, da. For a crack propagating
in a lamina of width, b, Eq. (1) becomes
1. d
Wd U γ m
b da (2)
Bonded structures exhibiting bulk linear-elastic behavior, i.e., away from the crack tip regions, obey Hooke's Law.
Gc may be expressed as:
2
Fc d
Gc . C
2. b d a (3)

30
where Fc is the load at the onset of crack propagation and C is the compliance of the structure and is given by
displacement/load, i.e., v/Fc. The compliance function, C, will be determined by experimental data obtained by the
debond tests and regression analysis.

3. EXPERIMENTAL SETUP
Twenty seven concrete blocks were made from a commercially available 5,000 psi ready-mix concrete. After an
initial curing period of seven days, CFRP composite plates were then bonded to the top of the concrete blocks. An
initial crack was then introduced to these specimens i) at the concrete-epoxy interface, ii) through mid-thickness of
epoxy layer, and iii) at the epoxy-CFRP plate interface. The test specimens were grouped into three categories of
nine specimens each based on exposure conditions. One group consisted of only control specimens and the other
two were subjected to freeze-thaw and simulated tidal conditions. The tidal conditioning was simulated exposing the
specimens to alternative wet/dry cycles of 24hrs for 100 days. A similar cycle was adopted to expose the specimens
to freeze-thaw conditioning by placing the specimen in the freezer (– 10°F to 0°F) for 24 hours and then removed
and allowed to thaw for another 24 hours at room temperature.

The bond strength of the CFRP plate-epoxy-concrete interface is evaluated using the peel test. The test method was
adapted from the peel tests that have been used extensively for composite materials. The specimen was fixed to a
tilted base (tilt angle ψ) and vertical load applied at the end of the plate as shown in Figure 1.

Fig. 1 – Specimen configuration

The specimen was loaded at a constant displacement rate of about 1.5 mm/min until the crack propagates and arrests
at a new crack length. The specimen was then fully or partially unloaded. This procedure was repeated until the top
plate becomes totally separated from the concrete. By this process, several compliance and critical loads for crack
propagation data points were obtained for each specimen, and from this data the fracture energy, Gc was then
calculated from Eq. 3 above.

4. RESULTS, DISCUSSIONS AND CONCLUSIONS

The load vs. displacements and corresponding crack lengths were recorded and graphs generated for critical load,
compliance and fracture energy for each crack length. Typical graphs shown in Fig. 2 represent the data for
specimens with initial crack locations at the CFRP/epoxy interface. Similar graphs were generated for the specimens
with initial cracks through the epoxy and concrete/epoxy interface (Baker, W. A., 1999). The overall test results
show that the control specimen exhibited a slightly higher average fracture energy compared with the specimens
exposed to tidal simulation and the freeze/thaw specimens. From the peel test data, no significant difference between
fracture energies of the control and tidal specimens can be deduced. However, the freeze/thaw action had an adverse
effect on the fracture energy of the specimens.

A scanning electron microscope (SEM) was used to examine the surface of the epoxy-CFRP interface after
debonding to further evaluate the effect of the harsh environmental conditions on the bond. The micrographs from
the SEM at 25,000volts and 1000X showed no discernable difference between the control and conditioned

31
specimens. Based on the limited data, it can be stated that the composite structural system was not adversely
affected by harsh environmental conditions.

A finite element model (FEM) was developed using four node plane strain elements and analyzed to obtain
compliance vs. crack length relationship. Fig. 4 shows the comparison of the experimental results and good
correlation with the values from the FEM model and the beam theory. The compliance calculation using beam
theory was made for a fixed ended beam with load applied perpendicular to the longitudinal beam axis, whereas the
compliance based on the FEM model and the experiment was for a force applied at a 20° tilt angle. The significant
difference between the compliance values at higher crack lengths can be attributed to the effect of the tilt angle.

C1T Critical Load F1T Critical Load

T1T Critical Load

F1T Compliance
C1T Compliance
T1T Compliance

F1T Fracture Energy

C1T Fracture Energy T1T Fracture Energy

Control specimen Tidal specimen Freeze-Thaw specimen

Fig. 2 – Critical load, compliance and fracture energy for each crack length

Control specimen Tidal specimen Freeze-Thaw specimen

Fig. 3 – The micrographs from the SEM at 25,000volts and 1000X

32
 Fig. 4 – Comparison of experimental and analytical results

REFERENCE
Anandarajah, A. and Vardy A. E., (1985). “A theoretical investigation of the failure of open sandwich beams due to
interfacial shear fracture”. The Structural Engineer, Vol. 63, No. 4, pp 85-92.
Arockiasamy, M., Sowrirajan, R,. and Zhuang, M.,(1995). “Behavior of beams prestressed or strengthened with
fiber reinforced plastic composites”, Proceedings of the IABSE Symposium, San Francisco, CA, pp 997-1002.
Baker, W. A., (1999). “Debond of CFRP laminate bonded with concrete exposed to harsh environmental
conditions”, MS thesis, Florida Atlantic University, Boca Raton, FL.
Blaschko M., Niedermeier, R. and Zilch, K.,(1998). “Bond Failure Modes of Flexural Members Strengthened with
FRP”. Second International Conference on Composites in Infrastructure, (ICCI ’98) Vol 1. pp 315-327.
Hag-Elsafi,O., Alampalli, S., Kunin, J.,(2001). “Application of FRP laminates for strengthening of a reinforced-
concrete T-beam bridge structure”. Composite Structures 52. pp 453-466.
Hag-Elsafi,O., Alampalli, S., Kunin, J.,(2004). “In-service evaluation of a reinforced-concrete T-beam bridge FRP
strengthening system”. Composite Structures 64. pp 179-188.
Meier, U., and Kaiser, H., (1991). “Strengthening of structures with CFRP laminates”, Proceedings of the Specialty
Conference on Advance Composite Materials in Civil Engineering Structures, ASCE, Las Vegas, NV, pp 224-
232.

33
34
Third International Conference on FRP Composites in Civil Engineering (CICE 2006)
December 13-15 2006, Miami, Florida, USA

DEBONDING OF CFRP LAMINATES EXTERNALLY BONDED TO

CONCRETE SPECIMENS AT LOW AND HIGH TEMPERATURES
Ernst L. Klamer
(Ph.D. Researcher, Eindhoven University of Technology, Eindhoven, The Netherlands)

Dick A. Hordijk
(Professor, Eindhoven University of Technology, Eindhoven, The Netherlands and
Director, Adviesbureau ir. J.G. Hageman B.V., Rijswijk, The Netherlands)

Cees S. Kleinman
(Professor, Eindhoven University of Technology, Eindhoven, The Netherlands)

ABSTRACT
This paper presents the results of an experimental investigation into the influence of temperature on small scale
concrete specimens, strengthened with externally bonded Carbon Fiber Reinforced Polymers (CFRP). Debonding of
the CFRP, due to high shear stresses in the concrete at the interface with the adhesive, governs the failure of these
specimens at room temperature. Temperature changes however will affect the bond properties of the CFRP-
adhesive-concrete joint, both due to the significant difference in the Coefficient of Thermal Expansion (CTE)
between concrete, adhesive and CFRP and due to the change of the material properties with increasing temperatures.
Both effects can affect the load level at which debonding occurs. Especially the adhesive shows a significant
decrease in strength and stiffness when the temperature reaches the glass transition temperature (Tg). To investigate
the influence of temperature on the debonding of externally bonded CFRP, two different test setups were used; a
double-lap shear test setup and a three point bending test setup. Test results have shown that a change in temperature
affected both the failure load and the type of failure, especially when the Tg of the adhesive was reached.

KEYWORDS
Carbon Fiber Reinforced Polymers, Temperature, Thermal stresses, Debonding, Bond

1. INTRODUCTION
One of the strengthening techniques that has become increasingly popular in the construction industry lately, is the
strengthening with externally bonded reinforcement based on Carbon Fiber Reinforced Polymers (CFRP). This
strengthening technique has proven to be a cost-effective and durable solution to strengthen existing (reinforced)
concrete structures. By now, guidelines have become available and numerous applications have been realised world
wide. Despite these developments, some questions have still remained unanswered. One of these questions that has
received very little attention so far is the effect of the ambient temperature on the debonding behavior of externally
bonded CFRP. Temperature changes will cause thermal stresses due to the significant difference in the coefficient of
thermal expansion (CTE) between concrete and CFRP and will change the properties of the applied materials, like
the strength and stiffness of the adhesive. Both effects can affect the load level at which debonding occurs.

So far, only limited research has been carried out into the influence of temperature on the debonding of externally
bonded CFRP. Experimental results from literature (Tadeu and Branco, 2000; Di Tommaso et al., 2001; Blontrock,
2003; Wu et al., 2005) have shown that the influence of temperature can be significant. However, based on the
reported failure loads as function of the temperature no distinctive conclusions can be drawn. It is also not known to
what extent the applied test setup had influenced the results and to what extent thermal stresses had developed
during the heating of the specimens. The aim of this project was therefore to further investigate the influence of
temperature on the debonding behavior of CFRP, externally bonded to concrete specimens.

35
2. EXPERIMENTS
Two different types of experiments, double-lap shear tests and a three point bending tests, were carried out. Each
test set-up was carried out with two different concrete grades (fcm,cube = 40.1 N/mm2 and 70.8 N/mm2) to investigate
the influence of the concrete strength. The specimens were tested in the temperature range from -20°C till 100°C.

2.1 Double-lap Shear Tests

The double-lap shear test specimens were produced in two series, one for each concrete grade, of twelve specimens
each. To be able to connect the specimen to the tensile testing machine, a steel threaded rod (M20 for the low and
M24 for the high strength concrete specimens) of about 1 m length was placed into the center of the specimens
(Figure 1). The specimens, including the rod, were cut in half after curing of the concrete. Two CFRP laminates
(Sika CarboDur S512) of 50×1.2×650 mm3 were then bonded to two opposite sandblasted surfaces with a two-
component epoxy adhesive (SikaDur-30). 25 mm at each side of the saw cut remained unbonded to prevent local
stress concentrations near the saw cut (Figure 1). Before testing, the specimens were heated in the oven for 12 hours.

Figure 1: Double-lap Shear Test Setup

The double-lap shear test specimens were tested in a 250 kN tensile testing machine, by pulling the two parts away
from each other. Steel clamps were applied at one side of the specimen, to make sure that debonding of the CFRP
was initiated at the other side. In this way, strain gauges, which measured the strain distribution over the length of
the CFRP, had to be applied on only one side. All specimens were packed with isolation. The temperature of the
concrete surface and the adhesive were measured during the experiment and varied with a maximum of 3°C.

2.2 Three Point Bending Tests

The three point bending test specimens were also produced in two series, one for each concrete grade, of twelve
specimens each. The concrete specimens were cut at midspan till half the height of the beam to create an initial
bending “crack” (Figure 2). The specimens spanned 750 mm, were supported at one fixed hinge support and one
roller support and were loaded at midspan. No internal reinforcement was applied. One CFRP laminate (25×1.2×650
mm3) was applied to the sandblasted soffit of the specimens. 25 mm at each side of the saw cut remained unbonded
to avoid local stress concentrations. The three point bending test specimens were tested in a 100 kN testing machine.
The specimens were not isolated during the tests, in order to have a clear sight on the specimens. The temperature of
the concrete surface and adhesive were measured during the tests and varied with a maximum of 5°C. Steel clamps
were applied to make sure debonding occurred at one side, which reduced the number of needed strain gauges.

Figure 2: Three Point Bending Test Setup

36
3. TEST RESULTS
3.1 Failure of the specimens

Both the double-lap shear tests and the three point bending tests failed by debonding of the CFRP laminate due to
high shear stresses in the concrete-adhesive interface. This was followed by a bending crack at midspan in case of
the three point bending tests. Debonding was initiated at the loaded end near the saw cut and propagated towards the
end of the laminate. The specimens which were tested at temperatures from -20°C till 50°C failed in an explosive
way by failure of the concrete at the interface with the adhesive, leaving a small layer of concrete (1–3 mm) attached
to the adhesive (Figure 3). The specimens that were tested at temperatures of 50°C and higher also failed at the
interface between the adhesive and concrete, but no concrete remained attached to the adhesive (Figure 4).

Figure 3: Failure in the Concrete at 20°C Figure 4: Failure in the Adhesive at 70°C

3.1 Failure Load – Temperature Relation

Figure 5 and Figure 6 show the measured failure loads as a function of the applied temperature for both concrete
grades, respectively for the double-lap shear tests and for the three point bending tests. Two linear regression lines
were plotted in each figure to give an impression of the tendencies. Basically, two different temperature zones can
be distinguished, first a zone in which the failure load increased with increasing temperature (zone 1), followed by a
zone in which the failure load decreased with increasing temperature (zone 2).

Figure 5: Failure Load Double-lap Shear Tests Figure 6: Failure Load Three Point Bending Tests

3.3 Influence of Temperature on the Failure Load

It was expected that debonding was initiated at the loaded end of the CFRP laminate due to the high shear stresses in
the concrete at the interface with the adhesive. Increasing the temperature was expected to affect the shear stress
distribution in the concrete due to three important effects (Figure 7): (1) The development of thermal stresses due to
the difference in CTE between concrete and CFRP; (2) the reduced adhesive stiffness; and (3) the reduced bond
strength of the concrete-adhesive interface (Klamer, 2006). These temperature related properties were used in a
Finite Element Analysis to simulate the shear stress distribution of the double-lap shear tests at 20°C and 70°C.

The difference in CTE between concrete (αc = ± 10 × 10-6 /°C) and CFRP (αf = ± 0 × 10-6 /°C) caused an initial shear
stress distribution in the concrete at the interface with the adhesive, when increasing the temperature ((1) in Figure
7). By subsequently loading the specimen, first the initial shear stresses had to be compensated at the loaded end
(τ0kN,70°C). The difference in CTE between concrete and CFRP therefore had a positive influence on the failure load.

37
Figure 7: Shear Stress in the Concrete-Adhesive Interface. Figure 8: Measured Strain in the CFRP at 30 kN

The second effect was the reduction of the Young’s modulus of the adhesive when increasing the temperature
(Klamer, 2005). As a result, the strain in the CFRP laminate, at the same external load, became more equally
distributed over the length of the CFRP laminate (Figure 8). A more equal strain distribution implied that the normal
stress in the CFRP was transferred to the concrete more equally. The increase in shear stress at the loaded end, when
applying an external load, will therefore be smaller at 70°C (∆τ70°C) compared to the increase at 20°C (∆τ20°C) ((2) in
Figure 7). The reduced adhesive stiffness therefore had a positive influence on the debonding load.

The third effect however was the bond strength of the concrete-adhesive interface. At temperatures near or above Tg,
the concrete surface bond strength was significantly reduced ((3) in Figure 7) and governed over the positive
influence of the reduced adhesive stiffness and the difference in CTE. The reduced bond strength therefore explains
the decreasing failure load with increasing temperatures in zone 2 after the initial increasing failure load in zone 1.

4. CONCLUSIONS
The experimental results have shown that temperature affected the load at which debonding was initiated, but also
affected the type of failure. Basically three effects of temperature were expected to have affected the debonding
load: (1) The difference in CTE between concrete and CFRP; (2) the reduced Young’s modulus of the adhesive at
elevated temperatures; and (3) the reduced bond strength of the concrete-adhesive interface at elevated temperatures.

5. ACKNOWLEDGEMENTS
The authors are indebted to the Dutch Ministry of Transport, Public Works and Water Management that supported
this investigation and SIKA Nederland BV for supplying the CFRP laminates and the adhesive free of charge.

6. REFERENCES
Blontrock, H. (2003). “Analyse en modellering van de brandweerstand van betonelementen uitwendig versterkt met
opgelijmde composietlaminaten”. Ph.D. thesis, Ghent University, Ghent, Belguim. (In Dutch).
Di Tommaso, A., Neubauer, U., Pantuso, A., and Rostásy, F.S. (2001). “Behavior of adhesively bonded concrete-
CFRP joints at low and high temperatures”. Mechanics of Composite Materials, Vol. 37, No. 4, pp. 327-338.
Klamer, E.L., Hordijk, D.A., and Janssen, H.J.M. (2005). “The influence of temperature on the debonding of
externally bonded CFRP”. Proceedings of 7th symposium on Fiber Reinforced Polymer Reinforcement for Concrete
Structures, Editors: C.K. Shield, J.P. Busel, S.L. Walkup and D.D. Gremel, Kansas City, pp. 1551-1570.
Klamer, E.L. (2006). “The influence of temperature on concrete structures strengthened with externally bonded
CFRP, Experimental Research”. O-2006.04, Eindhoven University of Technology, Eindhoven, The Netherlands.
Tadeu, A.J.B., and Branco, F.J.F.G. (2000). “Shear tests of steel plates epoxy-bonded to concrete under
temperature”. Journal of Materials in Civil Engineering, Vol. 12, No. 1, pp. 74-80.
Wu, Z.S., Iwashita, K., Yagashiro, S., Ishikawa, T., and Hamaguchi, Y. (2005). “Temperature effect on bonding and
debonding behavior between FRP sheets and concrete”. J. Society of Material Science, Vol. 54, No. 5, pp. 474-480.

38
Third International Conference on FRP Composites in Civil Engineering (CICE 2006)
December 13-15 2006, Miami, Florida, USA

EFFECTS OF BOND CONFIGURATIONS ON FLEXURAL RESPONSE

OF RC BEAMS EXTERNALLY STRENGTHENED WITH CFRP SHEETS
Jianguo Dai
(Researcher, LCM Research Center, Port and Airport Research Insititute,Yokosuka, Japan)

Keisuke Sumiyoshi
(Graduate student, Division of Built Environment, Hokkaido University,Sapporo, Japan)

Tamon Ueda
(Professor, Division of Built Environment,Hokkaido University, Sapporo, Japan)

Hiroshi Yokota
(Director General, LCM Research Center, Port and Airport Research Insititute,Yokosuka, Japan)

Yasuhiko Sato
(Associate Professor, Division of Built Environment,Hokkaido University, Sapporo, Japan)

ABSTRACT
This paper investigates experimentally how various bond configurations influence the flexural response of FRP-
strengthened RC beams. A total of seven beams(200×200×2200mm) were tested. One was the reference without
strengthening. Two were strengthened with CFRP sheets using hard and soft adhesive bonding, respectively, along
the whole test span. The remaining four were applied a combination of hard and soft adhesives but in two different
ways. One way is to use the soft adhesive at the flexure-dominating zone only and another is to use it for the shear-
dominating zone only. Pair tests were conducted for these four specimens under both normal and low temperature (-
10°C). It is revealed that a sole use of soft adhesive improves the ultimate member strength and ductility but leads to
a loss of stiffness. Using the soft adhesive at the flexure-dominating zone can not improve the member strength but
improve the member ductility only. However, use of hard adhesive at the flexure-dominating zone and the soft one
for the remaining parts can improve the member strength and ductility while avoiding loss of member stiffness. This
test program provides an in-depth understanding of the debonding mechanisms in the FRP-strengthened RC beams.
In addition, it is verified that that a low temperature of -10°C does not influence the global flexural response of FRP
strengthened beams using the combined bonding system.

KEYWORDS
FRP, adhesive bonding, flexural strengthening, bond configuration

1. INTRODUCTION
Flexural strengthening of functionally deficient RC members is a main application of FRP strengthening technology.
Since stress transfer between the external FRP and existing RC members is usually achieved through adhesive
bonding, appropriate selection of adhesives plays a critical role in optimizing the mechanical and durability
performance of the strengthened system. Unfortunately, the issue of how to achieve optimized macro-bond
performance remains unsolved up to date. As has been widely known in the FRP flexural strengthening cases,
premature interface debonding causes a sudden drop of loads and loss of ductility of the whole composite system
except when additional mechanical anchorages are available at the end of FRP. This debonding occurs often at a low
stress level in the FRP when an epoxy adhesive with high elastic modulus is chosen to be the bonding material. A
fewer researchers tried to use a flexible adhesive with low elastic modulus for the FRP flexural strengthening
application [Maeda et al, 2001, Sato et al. 2002]. They found that the flexible bonding can improve greatly the

39
flexural strength, ductility and the strain level in FRP at the ultimate stage. However, it is noticed that the flexible
bonding system hardly contribute the strength and stiffness enhancement before steel reinforcement yields. This is
not favored when the serviceability strengthening is also required. With the aim to consider simultaneously the
factors of strength, stiffness, ductility, and also, the utilizing efficiency of FRP.materials, this study tries to find an
optimized bond configuration for the flexural strengthening while providing an in-depth understanding of the
debonding mechanism.

2. TEST PROGRAM

A total of 7 RC beams were designed to fail in flexure in this test program. Each beam has the section area of 200×
200mm and was longitudinally and transversely reinforced with 2D13mm and D6@75mm, respectively (see Fig.1).
The yielding strengths of the longitudinal and transverse reinforcement are 370 and 356MPa, respectively. As
shown in Table 1, among the seven beams SP-1(0-0-0) was the reference beam and the remaining six were
strengthened with 0.334mm thick CFRP sheets. The elastic modulus and tensile strength of the CFRP sheets are
230GPa and 3430MPa, respectively. Two types of bonding adhesives EE2000 and HP-430 were used. EE2000 is a
hard adhesive that has an elastic modulus of 1.95GPa and HP-430 is a soft adhesive with an elastic modulus less
than 50MPa. Four types of bond configurations were applied (SP-2 to SP-5 in Table 1) that correspond to specimens
H-H-H, S-S-S, H-S-H, and S-H-S, respectively. Description of these different bond is shown in Fig.1, where ‘‘H’’
and ‘‘S’’ mean the hard and soft adhesives, respectively. Concept for this design is based on the presumption that
that soft bond around the flexure-dominating zone can increase the beams’ rotation and delay the mid-span
debonding while soft bond at the shear-dominating zone can increase the interface slip and bond force transfer. The
thickness of adhesive layer was kept 0.5mm for all strengthened beams. Specimens SP-6 and SP-7 had the same test
variables with SP-4 and SP-5 but they were put in a low-temperature chamber for two weeks before test but after the
curing to investigate the bond performance under low temperature, which is a popular environmental concern in
Hokkaido. Specimens SP-1 and SP-2 were from a previous test (Dai et al. 2005). Their concrete strength was 33MPa.
Concrete strength of the remaining beams was 30MPa.

D6@75 2D13

165

200
200
200 750 300 750 200
2200

Adhesive H-H-H
bond of S-S-S
H-S-H Unit: mm
FRP and H: Hard adhesive
S-H-S
concrete 620 500 620 S: Soft adhesive

Shear-dominating zone Flexure-dominating zone

Figure 1: Beam dimensions, bonding configurations and loading arrangement

Table 1 Arrangement of specimens and test results

No Test Py Pu δy δu εfrp,max Tempe- Failure description
Code (kN) (kN) (mm) (mm) (μs) rature
SP-1 0-0-0 40.2 44.1 4.08 25.7 - 20°C: Concrete failure after steel yields
SP-2 H-H-H 61.6 109.0 5.79 16.4 7,170 20°C: FRP peeling (failure at concrete side)
SP-3 S-S-S 48.6 114.5 4.95 25.6 12,004 20°C: FRP peeling (failure in the adhesive layer)
: FRP peeling (failure at concrete side)
SP-4 H-S-H 49.2 104.4 4.8 18.2 8,810 20°C
SP-5 S-H-S 54.1 120.3 5.26 25.1 11,216 20°C: FRP peeling (failure in the adhesive layer )
: FRP peeling (failure at concrete side)
SP-6 H-S-H LT N/A 103.5 N/A 18.0 N/A -10°C
SP-7 S-H-S LT N/A 117.0 N/A 27.1 N/A -10°C: FRP peeling (failure in the adhesive layer )
Note: Py: load at the yielding of steel reinforcement; Pu: the maximum load capacity; δy: deflection at the yielding of steel
reinforcement; δu: the ultimate deflection; εfrp,max: the maximum strain observed in FRP; and fc’: concrete strength. N/A: unclear
yielding load and deflection since gages on the FRP sheets and steel reinforcement were incorrectly recorded in the chamber.

40
3. TEST RESULTS AND DISCUSSION

3.1. Load-Deflection Response

Figures 2 and 3 shows load-deflection curves of all tested beams. The yield and ultimate load and deflection are
presented in Table 1 as well. It is clear that all strengthened beams have greater member stiffness and ultimate
member strength than the reference one (see Fig.2). Test beams H-H-H and H-S-H have similar ultimate strength.
On the other hand, test beams S-S-S and S-H-S have similar ultimate strength. So the bond properties of adhesives at
the shear-dominating zone determine the efficiency in strength enhancement. As shown in Table 1, the maximum
strains in the FRP sheets in the cases of H-H-H, H-S-H, S-S-S and S-H-S are 7170, 8810, 12004, and 11216μ,
respectively, meaning that the efficiency of utilizing FRP’s strength is much improved by using soft adhesives at
the shear-dominating zone. Interestingly, test beams H-H-H and S-H-S have similar member stiffness before steel
yields. So have the test beams H-S-H and S-S-S. This comparison indicates that the bond at the flexure-dominating
zone determines the strengthening efficiency in terms of member stiffness. The hard interface bond seems favorable
for the flexure-dominating area. The ductility performance sequence is: S-H-S ≈ S-S-S > H-S-H >H-H-H, indicating
that the interface slip capacity at both flexure and shear zones influence the ductility but the latter’s influence is
much greater. Therefore, it is a good choice to combine the hard bond between FRP and concrete at mid-span area
for the stiffness enhancing purpose and the soft bond at anchorage areas for the strength and ductility enhancing
purpose. Figure 3 shows the effects of low temperature. The global flexural responses of FRP strengthened beams
under 20°C: and -10°C: do not show noticeable differences at the whole loading stage.
140
Load(kN)
140
Load(kN)

120
120

100 100
0 -0 -0
H -H -H 80 H -S -H L T
80
S -S -S S -H -S L T
H -S -H 60 H -S -H
60
S -H -S S -H -S
40 40

20 20

0 0
0 5 10 15 20 25 30 0 5 10 15 20 25 30
D e f le c t io n ( m m ) D e fle c tio n ( m m )

Figure 2: P-δ curves: effects of bond configuration Figure 3: P-δ curves: effect of temperature
Progressive debonding within the shear zone
140
Load(kN)

Mid-Span 25000 H-S-H FRP

H-H-H FRP
120 debonding
Micro strain in FRP/Steel

H-S-H Steel
H-H-H Steel
20000 S-H-S FRP
100 S-S-S FRP
S-H-S Steel
S-S-S Steel
80 15000
60
10000
40 S-S-S Steel S-H-S Steel
H-S-H Steel H-H-H Steel
20 S-S-S FRP S-H-S FRP 5000
H-S-H FRP H-H-H FRP
0
0 5000 10000 15000 20000 25000 0
-1000 -500 0 0 500 1000
Maximum microstrain in steel/FRP
Distance from the mid-span (mm)
Figure 4 Strain development in FRP/steel Figure 5 Strain profile in FRP/steel at failure

3.2 Strain Development and Distribution in the FRP Sheets and Steel Reinforcement

Figure 4 indicates development of the maximum strains observed in the FRP and steel at the flexure zone. Strains in
both steel and FRP increase with the load and strains in the FRP are larger before the steel yields. In the contrast,
strains in the steel are much greater than that in FRP after the steel yields. This can be understood as the occurrence
of mid-span debonding between the FRP and concrete that is triggered by the yielding of steel. Clearly, this mid-
span debonding is not the critical factor to cause the ultimate failure of FRP strengthened RC beams although it
occurs first. It does not stop the continuing increase of strain in FRP (see Fig.4). After this mid-span debonding
period, stress concentration in the mid-span can be released. Rapid increase of the maximum strain in the steel stops
and but strains in the FRP increase further. As a result, the debonding between FRP and concrete propagated stably

41
in the shear-dominating zone, where the steel has not yet yielded. A substantial increase of member strength is
observed during this period (see Fig.4). Soft adhesive in the shear zone allows more increase of strain in the FRP
since it can accommodate larger interface slip and opening ability until a critical debonding. Figure 5 presents the
strain distributions in the FRP and steel for the peeled side of each beam at the ultimate stage. It is shown that
different bonding adhesives hardly influence the strain distribution pattern in the flexural zone at the ultimate stage.
However, use of the soft adhesive in shear zone [see S-S-S and S-H-S in Fig. 5] corresponds to a wider-range steel
yielding area that brings more member ductility. Moreover, the average strain gradients of FRP within the shear
zone are comparatively large. Hence using soft adhesives brings higher average bond stresses along the shear span,
which may be related to the increased number of cracks within the shear span.

3.3 Failure Mode and Debonding Mechanisms

All beams failed due to the peeling of CFRP sheets at one side as shown in Fig.6. However, the FRP/concrete
interfaces failed at concrete side and in adhesive layer, respectively, when the hard and soft adhesives were used in
the shear-dominating zone. In all the cases, the peeled FRP sheets were attached with some cover concrete at the
mid-span zone. In the case that the amount of cover concrete is small, a horizontal crack at the height of steel
reinforcement is yet visible (see S-H-S and H-S-H in Fig.6). As mentioned in the last section the FRP/concrete
interface at the mid-span debonds right after the steel yields to keep the deformation compatibility. These horizontal
debonding cracks were consequently attributed to the further yielding of steel reinforcement. The impact from
energy release of FRP induced spalling of concrete cover. In the shear-dominating zone, the peeling of FRP from
concrete in all tested beams are related to a major flexure-shear crack hence a critical factor for the ultimate failure
is the interface deformation ability at the tip of a major diagonal flexural-shear crack. Hence a comprehensive mix-
mode bond modeling for FRP/concrete interface and the discrete modeling of concrete fracture are needed to
simulate the above-two different failure mechanisms. Fig.6 also shows that soft bond in the shear zone allows
development of more cracks (see S-S-S and S-H-S) compared to hard adhesive case at the ultimate stage. This
should be considered when numerical simulation based on crack spacing is performed for bond strength prediction.

Figure 6 Failure modes of tested beams

4. CONCLUSIONS
1. A further insight into the whole-range debonding mechanisms of FRP strengthened RC beams in flexure is
provided. It is found that concrete cover failure at the mid-span area is not closely related to the bond of
FRP/concrete interface but the yielding of steel. The ultimate failure of strengthened beams occurs far beyond the
mid-span debonding between FRP sheets and concrete. The interface deformation capacity at the tip of a major
flexural-shear crack is confirmed to be a critical factor dominating the ultimate strength and ductility.
2. Combined use of soft and hard adhesives at the shear and flexure-dominating zones, respectively, seems to give
an optimized flexural response. More studies are needed to analyze the effective bond length of FRP to concrete
with multi-cracks, so that an optimum proportion of hard to soft adhesive bonding lengths can be formulated.
3. The strengthened beams show no difference in their flexural response under low temperature of 20°C and -10°C.

5. REFERENCES
Maeda, T.; Komaki, H.; Tsubonai, K. and Murauei, K.(2001), Strengthening Effects of CFRP Sheet Bonding using
Soft Resins, Proceedings of the Japan Concrete Institute, Vol.23, No.1, pp.817-822(in Japanese)
Sato, Y., Ito, T., Komaki, H. and Maeda, T. (2002), Flexural Behaviors of Reinforced Concrete Beams Strengthened
by CFS with Soft Layer, Proceedings of the Japan Concrete Institute, Vol.24, No.2, pp.1375-1380(in Japanese)
Dai, J.G., Ueda, T., Sato, Y. and Ito, T. (2005), Flexural Strengthening of RC Beams Using Externally Bonded FRP
Sheets through Flexible Adhesive Bonding, Proceedings of the International Symposium on Bond Behavior of FRP
in Structures (BBFS 2005), Edited by J.F. Chen and J.G. Teng, Hong Kong, China, pp.213-221.

42
Third International Conference on FRP Composites in Civil Engineering (CICE 2006)
December 13-15 2006, Miami, Florida, USA

ENVIRONMENTAL SENSITIVITY AND DEFECT CRITICALITY IN FRP

BOND TO CONCRETE THROUGH A FRACTURE MECHANICS
APPROACH
Rajiv Navada
(University of California San Diego, La Jolla, CA 92093-0085, USA)

Vistasp M. Karbhari
(University of California San Diego, La Jolla, CA 92093-0085, USA)

ABSTRACT
Although fiber reinforced polymer (FRP) composites are extensively used for the rehabilitation of concrete
infrastructure there is a critical lack of information regarding the effect and criticality of defects, both from a short-
term perspective and as related to long term durability making assessment of these difficult. This paper reports on
the use of a fracture based methodology for determination of criticality and discusses some results based on defect
type and exposure.

KEYWORDS
Bond, Fracture energy release rate, defect, effect, criticality..

1. INTRODUCTION
Although fiber reinforced polymer (FRP) composites are increasingly being used in the form of externally bonded
reinforcement for the rehabilitation of deteriorating and/or under-strength concrete structures, there are still a
number of issues that have not been resolved and are hence being addressed either through use of excessive caution
or through inordinately high factors of safety. Predominant on this list are the issues of durability and the criticality
of defects. While a body of research is being conducted on the former topic, the latter has still to be addressed in an
integrated fashion. Although a series of recent publications has provided lists of potential defects and an assessment
of techniques to identify them (Kaiser and Karbhari, 2003; Kaiser and Karbhari, 2004; Kaiser et al, 2004), and
guidelines for repair of a range of defects were recently published (Mirmiran et al, 2004) there is a singular lack of
understanding related to the criticality of defects and the effect of these defects on the integrity of the rehabilitation
scheme.

Since most rehabilitations are expected to provide significant periods of service it is important that the effect of
defects be clearly determined so as to provide the engineer with guidance as to whether a defect found during
inspection has to be immediately treated, or can be monitored for a period of time, or merely neglected as being non-
critical and aesthetic in nature. In addition, it is important that the effect of environmental exposure on the system
be assessed since this could have an accelerative effect on the progression of growth of the defect, and hence on its
criticality. In this paper a fracture based approach to the assessment of criticality of defects is outlined and some
results are discussed providing initial guidance both to effects of defects and to effects of environmental exposure.

2. APPROACH AND METHODOLOGY

In an attempt to simulate the growth of separation between the externally bonded FRP and the concrete substrate
(which could occur within the concrete, within the adhesive layer, between layers of FRP, or in combinations
thereof) a method initially proposed by Yamaguchi et al (1999) and modified by Kaiser (2002) is used in the current
investigation. The test setup essentially consists of a concrete beam with a central notch over which the FRP layer is
bonded. When loaded as shown in Figures 1(a) and (b) a mixed mode condition is imparted at the two crack tips
thereby closely simulating actual field conditions for crack progression. An additional advantage of the method is

43
that two interfacial regions, one on either side of the central notch can be studied simultaneously enabling multiple
tests under similar conditions, as well as providing a means for direct comparison of effects of defects occurring
along the bonded length of the FRP composite. Details on the test method, including the use of a video monitoring
system capable of accurately recording crack growth along the four edges (Figure 2) are provided by Kaiser (2002)
and hence will not be repeated herein.

Figure1(a): Schematic of Test Set-up Figure 1(b): Test Setup Showing Progression of
Interfacial Crack

Figure 2: Typical Images of Crack Growth Obtained By the Cameras

In the current set of experiments strips of unidirectional carbon fabric were laid-up on the concrete beams and
impregnated with an epoxy resin having a viscosity of 700-900 cps, leaving a central unbonded region of 374.7 mm
on either side of the central notch. This gap both ensured that crack progression was stable and that the effect of the
notch (used for placement of the rod to cause peel as in Figure 1) on propagation was minimized. Two layers of the
fabric were placed and the specimens were allowed to cure under ambient conditions for 4 days prior to testing.

In order to assess the effect of defects arising from conditions related to bonding specimens were tested after
application of both inadequate primer and excessive primer (with the latter representing the case where the excessive
application of primer resulted in a discernable resin rich layer at the interface), in the presence of a moisture
saturated substrate (equivalent to 4% moisture measured at the surface), and with a variety of debonds and
delaminations at the interface. Specimens were also tested after exposure for periods of 30 and 60 days to (a) heat at
65oC, (b) freeze at -10oC, (c) immersion in water at 23oC and (d) ponding (representative of conditions where the top
of the concrete was exposed to water which was allowed to seep in towards the FRP-concrete interface. For the
purpose of the current investigation a uniform and constant level of 2 mm of water was maintained on the top
surface of the specimen, i.e in contact with the concrete on the face opposite to the FRP).

44
3. ANALYTICAL DEVELOPMENT
As shown in Figure 1, the CFRP is loaded by means of a loading rod which is inserted in the notch in the concrete
beam. From a force balance and use of trigonometric relations the potential energy stored can be determined by
integrating the forces with respect to Δ under the condition that peel length, a, remains constant (wherein E11 is the
Young’s Modulus of Elasticity of the composite membrane, ε is the strain in each of the arms, F is the vertical
Force in each arm and can be expressed as S sin θ, where θ is the peel angle, a is the peeled distance on either side,
Δ is the vertical displacement of loading arm x = Δ / a , α = s / a and S = E11 Bt as shown in Figure 3) and
using the initial condition that at Δ=0, U s =0 as

⎡ Δ2 + s 2 + 3a 2
U s = E11 Bt ⎢ − (a − s ) 2 + Δ2 − (a + s ) 2 + Δ2 +
{(a + s) 2
}{ }
+ Δ2 (a − s ) 2 + Δ2 ⎤
⎥
⎢⎣ 2a 2a ⎥
⎦

Figure 3: Geometry and force balance

Finally, fracture toughness can be calculated by differentiating the equation for potential energy with respect to the
average peel distance, a, under the condition that membrane displacement, Δ, remains constant, resulting in
dU
G=−
2 Bda Δ =const
⎧ ⎧
(1 + α )⎨2 − (1 − α ) 2 + x 2 ⎫⎬ (1 − α )⎨2 − (1 + α ) 2 + x 2 ⎫⎬
1 ⎩ ⎭ ⎩ ⎭
G = E11t ⎡ x 2 + α 2 − 3 + (1 + α ) 2 + x 2 (1 − α ) 2 + x 2 + +
4 ⎢⎣
(1 + α ) 2 + x 2 (1 − α ) 2 + x 2
This represents the “slip” case as defined by Kaiser ( ). Under constraints of the test fixture if the slip were
restricted, the strain in the two arms of the composite membrane would be different. This would cause the strain
energy stored in the two arms and hence the fracture energy release rates (FERR’s) for the two arms to be different
such that S1 = ε 1 E11 Bt and S 2 = ε 2 E11 Bt leading to the generalized form of
⎡ ⎧⎪⎛ ⎞ ⎫⎪ ⎧⎪⎛ ⎫⎤
a2 ⎟ / a ⎬ − ⎨⎜ − a a 2 + Δ2 + Δ ⎞⎟ / a 2 ⎪⎬⎥
2
G = E11t ⎢− 1 + ⎨⎜ − a 2 + Δ2 − ⎜
⎢ ⎪⎩⎜⎝ ⎟
a 2 + Δ2 ⎠ ⎪⎭ ⎪⎩⎝ 2 ⎟⎠ ⎪⎭⎥
⎣ ⎦
Using this equation it is possible to get 2 values of FERR per concrete specimen, one from each side of the notch. It
should be noted that for this case the result of the differentiation for G involves the division by B instead of 2B as in
the earlier case.

4. RESULTS AND DISCUSSION

Due to space restrictions only representative results are presented herein. For ease of comparison results in Figures
4(a)-(d) results are shown as normalized values with the base-line being the defect free specimen set. Thus lower
values represent greater levels of deterioration. As can be seen there is a substantial effect of exposure on FERR. It
is interesting to note that there is a pronounced effect of freeze primarily in cases where the substrate has moisture
prior to application of the FRP. Further it can be seen that gradients, as compared to saturation, of moisture are
more deteriorative, as could be expected from a thermo-elastic perspective. It is interesting to note that approximate
relationships can also be derived through such experiments to assess effects of defects. The effect of length of a
delamination was seen to affect FERR in linear fashion till a limiting size when the length equaled the width of the
fabric strip. This trend can be simply expressed as
Normalized FERR = 1 – 0.64(Delamination width)

45
 1 .2 1 .2
B a s e - lin e B a s e - lin e
1 m o n th 1 m o n th
3 m o n th s 3 m o n th s
1 1

0 .8 0 .8

Normalized FERR
Normalized FERR

0 .6 0 .6

0 .4 0 .4

0 .2 0 .2

0 0
I n a d e q u a te E x e c s s iv e M o is t S atu rated In ad eq u a te E x e c s s iv e M o is t S a tu ra ted
P r im e r P r im e r S u b stra te S u b stra te P r im e r P r im e r S u b stra te S u b stra te
C atego ry C a teg o ry

Figure 4(a): FERR After Immersion in Water Figure 4(b): FERR After Ponding
1 .2 1 .2
B a s e - lin e B a s e - l in e
1 m o n th 1 m o n th
3 m o n th s 3 m o n th s
1 1

0 .8 0 .8

Normalized FERR
Normalized FERR

0 .6 0 .6

0 .4 0 .4

0 .2 0 .2

0 0
In a d eq u a te E x ec ssiv e M o is t S a tu r a te d In a d eq u a te E x ec ss iv e M o is t S a tu ra ted
P r im e r P r im e r S u b str a te S u b str a te P r im e r P r im e r S u b stra te S u b str a te
C ateg o ry C a teg o ry

Figure 4(c): FERR After Exposure to Heat Figure 4(d): FERR After Exposure to Freeze

5. SUMMARY
This study provides a methodology for the comprehensive assessment of the effect and criticality of defects and
environmental exposure. The use of the fracture mechanics approach ensures that it is amenable to scaling to allow
results obtained on small specimens to be later appropriately scaled to field structures providing a means for
development of inspection and defect criticality guidelines.

6. ACKNOWLEDGEMENTS
The authors gratefully acknowledge the support of SCCI and the Oregon Department of Transportation.

6. REFERENCES
Kaiser, H. and Karbhari V.M. (2003), “Identification of Potential Defects in the Rehabilitation of Concrete
Structures With FRP Composites,” International Journal of Materials and Product Technology, 19[6], pp. 498-520.
Kaiser, H. and Karbhari V.M. (2004), “Non-Destructive Testing Techniques for FRP Rehabilitated Concrete: I – A
Critical Review,” International Journal of Materials and Product Technology, 21[5], pp. 349-384.
Kaiser, H., Karbhari V.M. and Sikorsky, C. (2004),“Non-Destructive Testing Techniques for FRP Rehabilitated
Concrete: II – Assessment,” International Journal of Materials and Product Technology, 21[5], pp. 385-401.
Mirmiran, A., Shahawy, M., Nanni, A. and Karbhari, V.M., (2004) “Bonded Repair and Retrofit of Concrete
Structures Using FRP Composites: Recommended Construction Specifications and Process Control Manual,”
NCHRP 514, Transportation Research Board of the National Academies,
Yamaguchi, K., I. Kimpara, and K. Kageyama (1999), Evaluation of Debonding Energy Release Rate of Externally
Bonded FRP Sheets for Rehabilitation of Infrastructures, in Proceedings of ICCM-12, Paris, France.
Kaiser, H.(2002), Assessment of Defect Criticality and Non-destructive Monitoring in CFRP-rehabilitated Civil
Structures, Masters Thesis Department of Structural Engineering, UCSD: La Jolla. p. 264.

46
Third International Conference on FRP Composites in civil Engineering (CICE 2006)
December 13-15 2006, Miami, Florida, USA

IMPROVING THE BONDING AND DUCTILE BEHAVIOR OF FRP

STRENGTHENED RC BEAMS
H. Fatemi,
Graduate student, the University of Tabriz, Tabriz, E. Azerbaijan, IRAN

A. Davaran,
Assistant Prof., the University of Tabriz, Tabriz, E. Azerbaijan, IRAN

J. Esmaeili
Assistant Prof., the University of Tabriz, Tabriz, E. Azerbaijan, IRAN

ABSTRACT
Fiber reinforced polymer (FRP) is one of the most commonly used materials for strengthening and repairing of
the concrete structures. Debonding between fiber and concrete is a common failure mode, which is brittle and
abrupt. In this paper, it has been shown that adding the U-shaped FRP layers in the orthogonal direction to
longitudinal FRP, extended upward to the beam’s web, can improve the ductile behavior of the RC beam. A new
and effective detailing for providing end-anchorage of FRP is proposed which easily can be constructed.
Flexural behavior of several RC beams strengthened with the FRP by different arrangement of the U-shaped
orthogonal reinforcing FRP were studied experimentally and the effectiveness of each arrangement has been
verified.

KEYWORDS:
FRP, debonding, RC beams, ductility

1. INTRODUCTION

Fiber reinforced polymers (FRP) are progressively used for repairing and retrofit of reinforced concrete (RC)
structures. The flexural weakness or defects of RC beams are tackled normally with gluing FRP plates to the
tension face of the beam, Figure 1. Among the versatile types of the fibers, carbon textiles with higher tensile

2 8
A R C be a m

6@ 7c m
A dhesive layer 2 10

C ross section A
A
C FR P layer

Figure 1. RC Beam Strengthened with FRP

strength is commonly preferred, because of easy handling and performance. Many researchers have studied the
behavior of simply supported beams strengthened with carbon FRP (CFRP) attached to the bottom face of beam,
(Smith and Teng 2002) and (Buyukozturk et al, 2004). Several important failure modes, observed
experimentally are shown in Figure 2. These well-known modes are:
(a) Rupture of FRP in tension, (b) Shear failure near the supports or concentric force, (c) Crushing of concrete in
compression and, (d) Debonding of the FRP or splitting of concrete cover.
The debonding mode can be separated to three groups, as indicated in (Smith and Teng 2002). These are:
(d1) Cover splitting of concrete cover attached to FRP, (d2) debonding of the FRP and glue from the concrete
skin and, (d3) debonding of the FRP because of concrete cracks in the intermediate part of beam.
An extensive description of debonding modes has been verified by (Oehlers and Seracino 2004). Many analytic-
empiric formulas have been proposed based on different theories to explain the debonding failure mode by the
aforementioned references. Recent test results have shown that debonding has an abrupt nature, so a rapid

47 1
decrease of the strength is usually observed immediately after the ultimate load is attained. Promotion of
bonding strength and formation of some ductile behavior can be viewed as important goals in this regard.
In this paper, an innovative method for anchoring of CFRP for strengthening of beams is presented. In addition,
the effects of adding U-shaped FRP, which are orthogonal to the longitudinal fibers and extended upward to the
beam's web are investigated.

2. MATERIALS AND GEOMETRY AND METHODS

In this research, seven lightly reinforced concrete simply supported beams are verified experimentally. The first
four specimen are aimed to observe the ductile behavior by adding some U-shaped, specially detailed FRP
orthogonal to main FRP. Moreover, the last three specimens are tested to verify an innovative anchorage
detailing.

2.1. Geometrical Property of Specimens

All the seven beams have a dimensions of 150x200x1200 millimeters (mm.). See Figure 1.
The top and bottom reinforcing steels are two No. 8 and two No. 10 REBARS (having diameters of 8 and 10
mm.), respectively. Loop stirrups with 6 mm. in diameter at 70.mm spacing are used for shear reinforcing. This
reinforcing complies with minimum demands of ACI 318-99 .The difference between flexural and shear
nominal load bearing capacity of the strengthened beam, provides a wide variation rang for verifying the
debonding force. So getting a detailing to cause the debonding strength moves up toward the upper limit of
shear failure load, was thought to be a good plan. It can be mentioned that the U-shaped FRP could also increase
the shear strength in some cases, which must be taken into account. The concrete cover of tension and
compression reinforcement are 35 mm. and 30 mm., respectively. All the other size and spacing parameters are
depicted in the Figures 1 and 3.

2.2. Physical properties of the material

The compressive strength of the cylindrical samples of the concrete was about 20.6 Mpa. The yield strength of
the longitudinal and transverse rebar was 344 and 275 Mpa, respectively. The carbon fiber textiles used was
SIKAWRAP-Hex230 C and the epoxy for gluing the fibers had a grade of SIKADUR 330, both manufactured
by the SIKA-co of the Switzerland. The tensile strength, modules of elasticity, and one layer thickness of carbon
fibers were 3650 Mpa, 231000 Mpa and 0.12 mm, respectively.

2.3. Methods of Strengthening of the Specimens

The four specimens of the first group have been made as follows:
A0: a specimen with no CFRP, A1: having two layers of CFRP with dimension of 150 mm. x 900 mm.,
A2U: having two layers longitudinal CFRP and one U-shaped transverse (150mm width) attached to each end of
the longitudinal CFRP and A3X: having two layers of longitudinal CFRP and one crossed shaped transverse
(150mm x 500mm) CFRP at each end. These specimens have been shown in Figure 2. The other three
specimens in the second group are made as follows: B1: specimen with two longitudinal and one U-shaped
transverse CFRP at each end, B2AN: specimen with two layers of CFRP and two anchoring CFRP at a distance
of 240 mm. from each end and B3AN: specimen similar to B2AN but the distance of the anchoring CFRP from
each end was 140 mm. The anchor detail is made of following steps. First, a slit with a width a little less than
concrete cover is prepared. Then the end of longitudinal FRP is wrapped around a coupled of 10 mm rebar
which is smeared with glue. Finally, they are placed into the slit and then the finished face is smoothed by extra
sikadure31/41. The slit is covered by two layered U-shaped CFRP, as shown in Detail 1 of Figure 2.
X-Shaped CFRP Layer
A RC beam

Adhesive
layer

A
Specimen A3X
Specimen A1 CFRP layer

U-Shaped CFRP Layer U-Shaped CFRP Layer

Specimen A2U Specimen B1

Figure 2. The Different Arrangement of Specimens

48 2
 U -S h ap ed CFRP L ayer U -S h ap ed CF RP Layer

D et ail 1 D eta il 1

S pec im e n B 2 A N Spe cim en B 3A N

S ik ad u re3 1 /4 1
8

Figure 2(Continued). The Different Arrangement of Specimens

2.4. Test Setup

Every seven specimen was loaded by a universal testing machine. The loading method is shown in Figure 3. The
points A and C have 100mm distance from beam’s end. The two moving support A and C, compressed the
specimen against the fixed support B. It was not seen any local concrete crushing near the supports in this tests.
100 500mm

A C

Figure 3. The Loading Setup

3. RESULTS
The load-displacement graphs of the first group have been traced in Figure 4. The load setting view and
observed failure mechanisms have been shown in Figure 6. The load-displacement behavior of the second group
has been shown in Figure 5. A new observed failure mode for specimen B2AN and B3AN of the second group
are indicated in Figure 6. In this mode the lateral cover splitting followed by bottom face splitting of the
concrete cover was observed. The initiation of lateral cover separation can be attributed to the considerable
tensile force, which is produced by the U-shaped CFRP at the anchoring zones.
The broader area of the separation zone is thought to be the main reason for the increased load capacity of the
specimens B2AN and B3AN in second group. It is evident that the proposed anchoring detail can drastically
raise the load capacity in comparison to traditional bonding method such as one used in specimen B1.

12000
12000
A3X
10000
10000
A1 8000
8000
Load(kg)
Load(kg)

A2U A1 A0
6000 B3AN
6000 A2U B1 B1
B2AN
A3X 4000
B2AN
4000 A0 A0
A0 B3AN
2000
2000

0
0 0 400 800 1200 1600 2000
0 400 800 1200 1600 2000

Mid-span deflection(1/100 mm) Mid-span deflection(1/100 mm)

Figure 4. Comparison of 1st Group Figure 5. Comparison of 2nd Group

49 3
 (a) Load Setting View (b) End Cover Splitting, A1 (c) Intermediate Debonding, A2X

d) Side cover Splitting, B2AN (e) Side cover Splitting, B3AN (f) Shear Failure Between, B3AN

Figure 6. Observed Failure modes

4. CONCLUSION
Eight simply supported RC beams were investigated experimentally. Some observed results could be mentioned
as follow:
a) Using the U-shaped transverse CFRP can partly remove the abrupt and brittle nature of the debonding failure
mechanism, but further researches are needed to confirm the results.
b) An innovative and simple anchoring detailing for multi layer CFRP is proposed. This method can postpone
the end debonding mechanism and considerably increase the load bearing capacity of simple RC beams.
Extensive experimental and analytical works are required to characterize the proposed method.
c) Using the U-shaped end anchorage causes the intermediate debonding precedes the end debonding
mechanism.
d) The presence of anchorage leads to the shear failure to occur near that zone.

5. REFERENCES
Smith S. T. and Teng J. G. (2002).”FRP-strengthened RC beams. I: review of debonding strength models”.
Engineering Structures, Vol. 24, pp 385-395.
Buyukozturk O., Gunes O. and Karaca E. (2004). “Progress on understanding debonding problems in reinforced
concrete and steel members strengthened using FRP composites”. Construction and Building Materials, Vol. 18,
pp 9-19.
Oehlers D. J. and Seracino R. (2004). Design of FRP and Steel Plated RC Structures, 1st edition, Elsevier Ltd.

50 4
Third International Conference on FRP Composites in Civil Engineering (CICE 2006)
December 13-15 2006, Miami, Florida, USA

INTERMEDIATE CRACK DEBONDING ANALYSIS OF REINFORCED

CONCRETE BEAMS STRENGTHENED WITH EXTERNALLY BONDED
FRP PLATES
Luciano Ombres
Associate Professor, Department of Structural Engineering. University of Calabria, Rende (CS), Italy

ABSTRACT
The present paper concerns with intermediate crack debonding failure modes in FRP strengthened reinforced
concrete beams; a theoretical non linear model derived from a cracking analysis based on slip and bond stress is
adopted to predict the stresses and strains distributions at the failure. The analysis refers to beams with multiple
cracks; an analytical form of the local bond-slip law is, also, considered. In order to validate the effectiveness of
the model, a comparison between theoretical predictions and experimental results available in the literature, is
carried out. Results of a numerical analysis, varying geometrical and mechanical parameters involved in the
physical problem are, also, presented and discussed.

KEYWORDS
Crack Debonding, Reinforced Concrete, FRP

1. INTRODUCTION

The behaviour of reinforced concrete structures strengthened with externally bonded Fiber Reinforced Polymers
(FRP) plates is governed from debonding failure modes that, as well known, can drastically reduce their flexural
capacity. Debonding failures can take place at the ends of the FRP plates in presence of high stresses at the
interface between the FRP and the concrete or away from the ends of the bonded FRP plates when they are
induced by flexural or flexural-shear cracks. The first failure modes have been extensively studied and many
models of analysis have been proposed; on the contrary, the second type of failure, commonly named
Intermediate Crack-induced debonding failure (IC debonding failure), has received less attention especially from
a theoretical point of view.
The mechanism of IC debonding is related to the formation of cracks at the tensile side of the concrete elements.
When a crack is formed in the concrete, the tensile stresses released by the cracked concrete are transferred to
the FRP plate; consequently high local interfacial stresses between the FRP plate and the concrete are induced
near the crack. Further increase of the applied loading produce an increase both in the tensile stress in the plate
and in the interfacial stress between the FRP plate and the concrete near the crack. When the interfacial stress
achieves the critical values, debonding initiates and then propagates away from the crack. The IC debonding can
occurs in presence both of a single flexural crack and of multiple cracks; in the last situation a succession of
plate debonding between adjacent cracks can occurs simultaneously with a sudden failure of the beams.
The IC debonding is governed by the FRP-to-concrete bond-slip behaviour, which, generally, is affected by
elastic deformations and microcracking; the interfacial debonding is likely to occur near the crack intersection
immediately after flexural cracking.
Even if the IC debonding represents the most common failure mode in reinforced concrete structures
strengthened with externally bonded thin FRP plates, there is a lack of research on this topic and a limited
number of models are available in the technical literature; among those the models proposed by Niu and Wu,
Chen and Teng and Liu, Oehlers and Seracino.
Niu and Wu (Niu and Wu, 2001) developed a methodology for predicting the debonding failure load due to
flexural cracks in FRP-strengthened reinforced concrete beams based on the fracture energy criterion; the
debonding failure load is established by combining the reinforced concrete beam theory with the solution from
the simple shear test.

51
 The model proposed by Chen and Teng (Teng et al. 2003) is based on the similarity between the IC debonding
failures and those in simple shear tests; a bond strength model is combined with a simple section analysis for
predicting the strength of beams and slabs, which fail by IC debonding. The maximum force that the bonded
FRP can take is expressed as.

bf 1 if L ≥ Le 
Ef t f 2−  
bc
P =α β f β L f c b f Le , where Le = ; βf = ; β L =  π L   (1)
fc bf sin   if L ≤ Le 
1+ 2 Le 
bc

and bc=concrete width, b f,tf and E f = FRP width, thickness, and elastic modulus, respectively, L= bond length;
€ fc= concrete compressive strength, α= calibration factor to account for any difference between the behaviour of a
beam failed by IC debond € and a shear-lap specimen failed by interfacial debond. The α-value proposed by Chen
and Teng on the basis of experimental results is 0.4. Recently, an experimental study carried out by Pham and
Al-Mahaidi (Phan and Al-Mahaidi, 2006) on CFRP retrofitted reinforced concrete beams, indicates that α can
be taken as 1.04.
A discrete crack model based on partial interaction theory has been developed by Liu, Ohlers and Seracino (Liu
et al., 2005) to analyse the IC debonding behaviour of plated members; the model takes into account slip at both
the plate/concrete and bar/concrete interfaces. Through the model the maximum plate strain reached and the
strain at debonding failure can be determined.
In the present paper, a theoretical non linear model derived from a cracking analysis, founded on slip and bond
stresses, is adopted for the analysis of the debonding induced from intermediate flexural cracks in FRP
reinforced concrete beams. The analysis refers to beams in which a multiple cracking takes place. An analytical
form of the bond-slip laws at the interfaces FRP-to- concrete and steel rebars –to-concrete are used. Through the
model the strains and stresses in the concrete element between two contiguous cracks for any loading level can
be evaluated. The IC debonding occurs when the strains and stresses in the cracked element reach the values that
correspond to the failure condition at the interface FRP-to- concrete. To validate the model, a comparison
between theoretical predictions, experimental results available in the literature, and predictions of others
theoretical models, is made. Obtained results put in evidence as the predictions of the proposed model are in
good agreement with those experimental.

2. THE NON-LINEAR MODEL

The model, derived from a cracking analysis based on slip and bond stress, refers to a beam element between
two consecutive cracks (block) subjected to a constant bending moment higher than the first cracking bending
moment. The analysis refers to the stabilized crack formation phase and considers two limit cracking
configurations corresponding to the maximum and minimum crack spacing that bound all possible cracking
configurations. The following equations are used to solve the structural problems:
i) Equilibrium conditions on the cross-sections (translational and rotational)

∫Ω σ c dΩ c +i∑
c
=1, n
ω ri σ ri =0
(2)
∫ σ c y dΩ c + ∑ σ ri y i ω ri =M
i =1, n
Ωc

where Ωc is the concrete area, ω ri is the area of the ith reinforcement (FRP sheet, steel rebars); yi is the distance
between the neutral axis of the cross section and the centroid of the ith reinforcement.

ii) Strain compatibility between two points, initially fully bonded, belonging to the steel rebar and the concrete
du
u s' ( z )= s =ε s ( z )−ε cts ( z ) (3)
dz
and to the FRP sheet and the concrete
du
u r' ( z )= r =ε r ( z )−ε ct ( z ) (4)
dz

52
where us (z) is the slip between the concrete and the steel rebar; εs (z) and εcts (z) are strains of the steel rebar and
the concrete in tension at the level of the steel rebars, respectively; ur (z) is the slip between the concrete and the
FRP sheet; εr (z) and εct (z) are strains of the FRP sheet and the concrete at the tensile side of the cross-section,
respectively;

iii) Longitudinal stress equilibrium of the concrete reinforcements:

dσ s ( z ) 4 dσ r ( z ) 1
σ s' = = τ s ( z) σ r' = = τ r ( z) (5)
dz db dz tf

where db is the steel rebar diameter; tf is the thickness of the FRP sheet; σs (z) and σr (z) are tensile stresses in the
steel rebar and the FRP sheet respectively; τ s (z) and τr (z) are bond stresses between steel rebar and concrete and
FRP sheet and concrete, respectively.
The equations 2-5 furnish a system of differential equations that cannot be solved in closed form because the in-
homogeneity of the boundary conditions; as a consequence a numerical solution is required being impossible to
solve the mathematical model in a closed form (Aiello and Ombres, 2004). The utilized numerical procedure is
founded on the finite differences method, by dividing the space between two cracks in n-1 discrete elements
with small length Δ z. An iterative procedure that transforms the problem of limit conditions in the iterative
solution of an initial value problem is adopted (Aiello and Ombres, 2004).

2.1. Limit conditions for the IC debonding

The following limit conditions have to be imposed to evaluate the occurrence of the IC debonding:
• At the halfway, for symmetry, the slip between FRP/concrete and steel rebar/concrete is equal to zero;
• At the distance lb from the midway, immediately before the crack formation, that is when the tensile
strain in the concrete is equal to zero, at the FRP/concrete interface the slip is maximum and the bond
stress is zero. These conditions are the most favourable to the IC debonding failure.

2.2. The load at IC debonding failure

The numerical procedure developed considering initial values allows evaluating strains, stresses and loading
value corresponding to the IC debonding. For an assigned load, the procedure starts evaluating strain and stress
distribution at the halfway by equilibrium conditions (2) and imposing the limit value ur=us=0. Therefore it is
possible to evaluate, by means of equations (2)-(5), the stress and strain distribution at the edge cross section of
the first discrete element contiguous to the halfway; subsequently, the procedure is extended to every contiguous
discrete elements. The limit values at the distance lb from the midway are attained iteratively; the iteration is
made varying the external load. The procedure halts when the IC debonding failure conditions are reached.

3. COMPARISON WITH THEORETICAL AND EXPERIMENTAL RESULTS

Theoretical predictions of the proposed model are compared with experimental results and with predictions of
other theoretical models available in the literature. In particular the database of experimental results reported in
(Teng et al., 2003) are considered together with predictions of the Niu and Wu (Niu and Wu, 2001) and the
Chen and Teng (Teng et al., 2003) models. These models, as previously mentioned, are based on the similarity
between IC debonding failures and those in shear tests on FRP-to-concrete bonded joints; analytical
relationships are obtained combining a simple section analysis of concrete beam with the solution from the shear
test. Both models do not consider the local bond-slip FRP-to-concrete. The proposed model requires an explicit
bond-slip relationship that adequately describes the effective interaction between the reinforcements (FRP sheets
and steel rebars) and the concrete. In the analysis the bond-slip law between the FRP and the concrete is
expressed by the Ueda et al. model (Ueda et al., 2005), while the well-known Bertero-Popov-Eligheausen model
is adopted for the bond-slip law between the steel rebars and the concrete. Results of the comparison in terms of
shear force at the IC debonding failure are reported in the Table 1. The analysis of results evidences, as
predictions of the proposed model are in very good agreement with experimental results; the mean value, the
standard deviation and the coefficient of variation of the ratio Vpred/Vexp are equal to 0.9588, 0.09678 and 0.0982,
respectively.

53
 Table 1. Shear force at the IC debonding failure: theoretical - experimental comparison

Beam designation Experimental Chen and Teng Niu and Wu Proposed

(Teng et al, 2003) (kN) (kN) (kN) (kN)
A1 72,80 65,52 65,77 80,59
A2 84,90 67,96 68,31 85,68
A7 86,10 65,25 64,00 77,52
C1 77,20 65,52 65,77 79,49
B2 148,00 96,07 94,79 137,75
4 14,80 11,37 12,99 15,15
5 15,30 13,51 16,29 16,07
6 14,00 14,45 13,84 13,12
7 12,80 12,53 9,47 13,43
8 18,70 17,20 12,03 20,41
B3u 17,00 11,85 13,73 20,65
B4u 17,25 12,22 12,45 17,90
B5u 17,30 10,61 14,17 17,48
B1u 30,00 31,93 24,24 28,27
B3 27,60 25,71 23,37 25,40
B4 26,30 25,57 24,73 25,79
B5 34,90 33,63 31,97 29,53
B6 34,80 33,69 29,17 29,59
B7 29,60 29,31 28,20 27,44
B8 30,80 29,03 28,12 28,17

4. CONCLUDING REMARKS

The results of the analysis, allow to drawn the following concluding remarks:
• The predictions of the model are in very good agreement with experimental results and in some cases
more accurate than models usually adopted for the analysis of the IC debonding failure;
• The local bond slip law is fundamental for the IC debonding mechanism; consequently, a reliable
structural analysis imposes the use of prediction models founded on the local bond-slip law like the
proposed model.
Further comparisons with experimental and numerical results are needed for a better validation of the model that
seems to be very useful for the accurate prediction of the IC debonding failure of FRP-strengthened reinforced
concrete beams.

5. REFERENCES
Aiello, M.A. and Ombres, L. (2004). “Cracking and deformability analysis of reinforced concrete beams
strengthened with externally bonded carbon fiber reinforced polymer sheets”. Journal of Materials in Civil
Engineering, ASCE, Vol. 16, No 5, pp. 392-400.
Liu, I., Oehlers, D.J., and Seracino, R. (2005). “FRP plated reinforced concrete hinges: partial interaction
numerical simulation”. Proceedings of the Third International Conference Composites in Construction-CCC05
Editors: P. Hamelin, D. Bigaud, E. Ferrier and E. Jacquelin, Lyon, France, pp. 69-76.
Niu, H. and Wu, Z. (2001). “Interfacial debonding mechanism influenced by flexural cracks in FRP-
strengthened beams”. Journal of Structural Engineering, Vol. 47A, pp. 1277-1288
Pham, H.B., and Al-Mahaidi, R. (2006). “Prediction models for debonding failure loads of carbon fiber
reinforced polymer retrofitted reinforced concrete beams”. Journal of Composites for Constructions, ASCE,
Vol. 10, No 1, pp. 48-59.
Teng, J.G., Smith, S.T., Yao, J. and Chen, J.F. (2003). “Intermediate crack-induced debonding in RC beams and
slabs”. Construction Building and Materials, Vol. 17, pp. 447-462.
Ueda T., Dai J.G., and Sato Y. (2005). “Development of nonlinear bond stress slip model of fiber reinforced
plastics sheet-concrete interfaces with a simple method”, Journal of Composites for Constructions, ASCE, Vol.
9, No 1, pp. 52-62.

54
Third International Conference on FRP Composites in Civil Engineering (CICE 2006)
December 13-15 2006, Miami, Florida, USA

EVALUATION OF FRP BOND SPECIFICATIONS FOR DEVELOPMENT

LENGTHS
Nathan Newman
(Graduate Research Assistant, University of Missouri-Rolla, Rolla, Missouri, United States)

Ashraf Ayoub
(Assistant Professor, University of Missouri-Rolla, Rolla, Missouri, United States)

Abdeldjelil Belarbi
(Distinguished Professor, University of Missouri-Rolla, Rolla, Missouri, United States)

ABSTRACT
This paper presents an analytical study performed on concrete girders reinforced with FRP bars with the goal of
evaluating current ACI-440 design equations for bond development length. The analytical model is based on finite
element formulation of anchored bar problems with inelastic constitutive laws for the corresponding materials. The
bond stress-slip behavior between FRP bars and concrete was calibrated from experimental data obtained from a
large database set of over 50 test specimens available in the literature. A large number of analytical simulations were
conducted in order to investigate the effect of different parameters on the development length. These parameters
include: bar type (GFRP, CFRP), bar strength, bar modulus of elasticity, bar diameter, surface coatings, concrete
strength, and confinement. The analytical study concluded by proposing a new equation for development length that
accounts for the effect of these parameters.

KEYWORDS
Bond, FRP, Development Length

1. INTRODUCTION
In more recent years, the use of FRP bars as reinforcement in concrete has grown due to its corrosion resistance,
fatigue resistance, magnetic transparency, and lightweight properties (Okelo and Yuan, 2005). However, the
knowledge of the properties of FRP and its behavior with respect to concrete are limited. This presents a problem
when deriving design equations, specifically when searching out a reliable equation for the bond development length
of FRP bars embedded in concrete. The lack of understanding of this material is reflected in that essential design
equations and the ability to accurately design structures using FRP are inadequate. The equation for the
development length of steel bars in concrete can not be directly applied to FRP bars due to several obvious
differences in material properties and surface variations between the two products, but more importantly the
differences in their respective bond failure mechanisms prevent this. This is due mainly to their material properties,
specifically their stress-strain relationships. Steel generally displays an elastic region with a large plastic region in
its stress-strain curve. This plastic region contributes ductility to the bond system on behalf of the reinforcement
during loading. However, the stress-strain relation for FRP displays a large elastic region immediately followed by
ultimate failure with no preceding warning. There is no plasticity in the FRP reinforcement; therefore, all of the
ductility of the system comes from the bond behavior due to the interaction of the FRP and the concrete. It is
important to ensure that the FRP is protected and does not yield, but rather the bond. Doing this causes the bond
system failure to better resemble the failure mode of the bond and not that of the FRP. In order to derive a length of
embedment which satisfies this ideal scenario for a set condition, numerical modeling was used. Using one-
dimensional finite element analysis, the behavior of the FRP bond with concrete was analyzed, and used to derive a
more accurate design equation for the development length of a straight FRP bar embedded in concrete.

55
2. APPROACH
For the duration of the study, no new physical testing (such as direct pullout testing) was to be conducted. However,
data needed to be obtained through other means. This was done by contacting a plethora of researchers whom had
done similar testing and by gathering test data from various journal articles and papers. A sizable database of
specimens was thereby accumulated. Each specimen’s stress-slip behavior of the FRP to concrete bond was
documented and their respective parameters were obtained and recorded. In order to find the required development
length for an FRP bar, the ACI 440.1R-03 equations for development length were first used to find an embedment
length which would expectantly fall close to the actual development length. This calculated length was used in the
numerical model as the embedment length for the respective specimen. The analysis was run using a computer
model, and the fFRP stress versus slip relationship for the setup was plotted. From this graph, the maximum achieved
stress in the FRP bar was found. The embedment length of each specimen for a set diameter was varied until the
defined development length criterion was met. The embedment length used to achieve the criterion was then taken
as the development length of the specimen for that exact diameter. In this manner, the development length of all of
the specimens and their variations were found. The results were then analyzed to develop a design equation.

3. CRITERION FOR DEVELOPMENT LENGTH

In order to establish a criterion for determining an acceptable value for development length, the original criterion for
steel was examined. It was found that the criterion for steel is based on achieving a slightly magnified value of the
yield stress for steel in the bar. By doing this, the bar is pushed beyond its elastic range and into plastic deformation
and thereby ensuring ductility in the system. Since FRP does not present a plastic deformation range, an alternate
criteria was needed. It was decided upon to use a reduction factor to lower the FRP ultimate tensile stress to a level
which would allow the bond to fail before the bar failed. Since FRP fails in an abrupt fashion without warning, it is
desirable to have bond failure first. The criterion used to obtain an acceptable development length for steel was a
target stress in the steel bar of 1.25fy (Jirsa and Lutz, 1979). The reciprocal of this factor was taken in order to apply
it to FRP as a reduction factor. It was decided to use a reduction factor of 0.8fpu for the FRP development length
criterion. This value was then compared with the model results. When the maximum axial stress achieved in the
bar matches 0.8fpu, the coinciding embedment length for that particular case is taken as the required development
length for the respective specimen. An ideal bond relationship between FRP embedded bars and concrete is shown
in Figure 1. From this diagram it can be seen that the FRP is protected and that the ductility is provided by the bond.
The ductility comes from the bond hardening slope just prior to bond yielding. The criterion defined for this testing
would be met when the maximum stress in the FRP achieved is equal to 0.8fpu.

Figure 1: Proposed FRP Stress-Slip Relationship Controlled by Bond Failure

4. MODEL
The analytical model used in this project is based on a one-dimensional finite element formulation of an anchored
bar with inelastic constitutive laws for the materials involved. A total of six elements with seven nodes are taken for
each bar specimen. The program is displacement controlled, so either one end was pulled a set distance for a pullout
test, or one end was pulled while the other end was pushed a set distance for a push-pull test. This displacement was

56
a fixed preset variable. The program was arranged around the following variables: modulus of elasticity of the bar,
ultimate tensile strength of the bar, bond behavior for the corresponding specimen, diameter of the bar, and the
embedment length.

In order to depict the behavior of bond, the following constitutive model was used. The bond stress versus
displacement plot for a given specimen was described by three key points. These points are illustrated in Figure 2.
Using these points, the essential components concerning the bond properties are addressed. These include
FRP/bond stressing and yielding, and bond failure.

Figure 2: Idealized Bond Stress-Slip Relationship

5. NUMERICAL RESULTS
The ACI 440.1R-03 equations were used to calculate suggested development lengths using the following equations:

d b2 ffu
lbf = K 2 Eq (1)
f c'
d b ffu
lbf = Eq (2)
2700
The corresponding development length for a set condition were found to be; 30 inches using the ACI 440.1R-03 Eq.
2, and 58 inches using the ACI 440.1R-03 Eq. 1, taking an average value for the suggested K2 factors (ACI, 2003).
The numerical simulation was run using both of the acquired development lengths, the results are shown below in
Figure 3 along with the development length which was found using the criterion previously defined. The
development length found using the criteria was 17 inches for this test specimen.

Figure 3: Stress-Slip Relationships for Various Embedment Lengths

57
From Figure 3 it can be seen that the development length of 17 inches is the more viable length which exhibits a
bond hardening region. From this region comes the desired ductility. The plots of specimens with 30 and 58 inches
both fail due to rupture of the FRP bar.

Figure 4 displays several stress-slip curves for three different embedment lengths. Each of these specimens is
pullout failure controlled due to the bond hardening region found at the peak of each curve, however, only one meets
the criterion. The fpu of each of these specimens is the same at 80ksi, which gives a target ffrp of 64ksi. The
specimen having an embedment length of 17 inches peaks at approximately 64 ksi and therefore satisfies our
criterion and is taken to be the development length for the set condition of its parameters.

Figure 4: Comparison of Stress-Slip Relationship in terms of Embedment Length and Stress Level Criteria

6. CONCLUDING REMARKS
This study attempts to approach the evaluation of the FRP bond from a different perspective than steel. The failure
mechanism and criteria are based around preserving the ductility of the bond when using FRP. Using a wide spread
of data, the effect of differing parameters were investigated via the use of finite element analysis. The work is
currently in progress, and will result in a set of proposed design equations for development length that considers the
effect of the different parameters investigated.

7. REFERENCES
ACI (2003). “Guide for the Design and Construction of Concrete Reinforced with FRP Bars,” American Concrete
Institute.
Jirsa, James O. and Lutz, LeRoy A. (1979). “Rationale for Suggested Development, Splice, and Standard Hook
Provisions for Deformed Bars in Tension”. Concrete International, July, pp 47-61.
Okelo, Roman and Yuan, Robert L. (2005). “Bond Strength of Fiber Reinforced Polymer Rebars in Normal Strength
Concrete”. Journal of Composites for Construction, May/June, pp 203-213.

58
Third International Conference on FRP Composites in Civil Engineering (CICE 2006)
December 13-15 2006, Miami, Florida, USA

PRACTICAL USE OF PULL-OFF STRENGTH TESTING (ASTM D4541)

FOR ASSESSING QUALITY OF FRP-TO-CONCRETE BOND

Kent A. Harries
(Assistant Professor, Department of Civil and Environmental Engineering, University of Pittsburgh)

ABSTRACT
Many researchers report, and many practitioners specify, pull-off strength testing (ASTM D4541 or similar) as a
quality assessment measure for bonded FRP retrofit procedures. Such testing techniques were not originally
developed for bonded FRP applications typical of civil infrastructure and require careful application and
interpretation of results. Nonetheless, pull-off strength testing has a variety of applications including substrate
assessment, FRP installation quality assessment and assessment of bond degradation due to environmental exposure.
A variety of experimental data are presented representing laboratory and field applications of pull-off testing of FRP
applications. Guidance for the use of pull-off testing is provided.

KEYWORDS
Adhesive, Debonding, Pull-off testing, Surface preparation, QC/QA

1. INTRODUCTION
The quality, integrity and overall performance of bonded FRP retrofit systems is largely dependent on adhesion of
the FRP system to the concrete substrate. Although a great deal of research has been conducted focusing on bond
behavior, and a number of methods for assessing this have been proposed, few may be practically adapted for in situ
quality control/assurance (QC/QA) or acceptance testing. The essence of a sound QC/QA method includes a)
simplicity of application; b) the ability to be rapidly deployed with minimal preparatory work; c) easily understood
acceptance criteria; and d) excellent repeatability ensuring integrity of data. The existing ASTM Standard D4541
Standard Test Method for Pull-Off Strength of Coating Using Portable Adhesion Testers (2002) provides such a
method which is well suited to acceptance testing of bonded FRP systems.

2. ASTM D4541 DIRECT TENSION TESTING

The ASTM D4541 test method was developed to assess the adhesion of overlays for concrete repair. Guidance for
this application is found in ICRI (2004). Overlay repairs, however, differ from bonded FRP applications in a number
of crucial ways. The ASTM D4541 test method is, nonetheless, appropriate for the qualitative and quantitative
assessment of bond quality of a bonded FRP system.

The ASTM D4541 test method involves adhering a rigid disk (or dolly) to the surface of the FRP to be tested (Figs.
1(a) and (b)). The test sample is isolated from the surrounding FRP by a circular hole produced using a core drill.
Using a pull-off test apparatus (an example is shown in Fig. 1(c)), the disk is subject to direct tension. Failure occurs
through the weakest plane within the depth of the test sample. Seven failure (Fig. 2) are possible in a bonded FRP
system. The implications of each failure mode are discussed below. If bond is sound, the method will result in a
cohesive failure in the concrete substrate (Failure Mode G) and thereby provide a lower bound value for the
adhesive bond strength. If the bond capacity is deteriorated the failure mode, failure surface and pull-off strength can
all provide insight into the bond behavior and deterioration mechanism. Classification of the failure allows
qualitative assessment of the degradation and may help to identify the nature of the degradation. Pull-off strength,
assuming it is no longer governed by the substrate concrete, may be a surrogate quantitative measure of bond
degradation.

59
2.2 ASTM D4541 Test Apparatus and Procedure

Five test apparatus types are recognized in ASTM D4541. Comparison of results from one apparatus to the next is
inappropriate (ASTM 2002). Only a Type I Fixed Alignment Adhesion Tester, as defined by ASTM D4541, is
appropriate for assessing FRP bond. Such an apparatus uses 50 mm (or larger) disks. All other apparatus types
permitted use smaller disks (usually 20 mm) and are suited to assessing the adhesion of coatings (such as paint
systems) where the substrate is not expected to be the weak plane. Because the weak plane is expected to be in the
concrete substrate, a test area having a diameter greater than twice the maximum aggregate size is suggested.
Vaysbund and McDonald (1999) assessed the field performance of three commercially available testers. A DYNA
Z15 tester (Fig. 1(c)) was used in the present work. With this unit, the entire test process is easily executed by one
person and testing may be carried out in any orientation (downhand, horizontal (Fig 1(c)), or overhead).

deck soffit GFRP application

deck drain

610 mm
girder soffit
(a) cored circle (b) (c) (d)

Figure 1: Pull-off test application: a) view of application on web of reinforced concrete interstate girder;
b) 50 mm rigid disk; c) manually operated pull-off test apparatus; d) Failure Mode G. (Harries 2001)

In the ASTM D4541 test procedure, providing a cored circle around the perimeter of the test sample is considered
optional. This cored circle must be provided for useful results to be obtained for FRP system. Without the core, load
spreading from the rigid disk is likely and a greater region of bonded material is engaged. The extent of this region
is a function of the FRP system thickness, flexural properties and through-thickness shear properties (Harries et al.
2004). In drilling the core it is important not to exert stresses (dominantly torsional) on the test sample. This is best
accomplished using a smooth (no teeth) diamond core drill. For a concrete repair application, Vaysbund and
McDonald (1999) recommend that the core extend a minimum of 25 mm or twice the disk diameter into the
substrate material. Although no parametric study has been conducted on bonded FRP applications, it is known that
the typical debonding plane is located within the concrete immediately adjacent the adhesive. For this reason, the
author proposes that only a minimum penetration into the concrete substrate, equal to the greater of 6 mm or the sum
of the FRP and adhesive thickness, is required. Bonding the rigid disk to the FRP is easily accomplished using any
manner of high-strength adhesive. The author typically uses a two-part “5-minute” epoxy with excellent success.

3. INTERPRETATION OF PULL-OFF TEST FAILURE MODES

Seven failure modes possible in a pull-off test conducted on a bond FRP system are shown in Figure 2.
Combinations of these modes (with the exception of Mode F) are generally unlikely.

rigid disk
FRP adhesive

concrete substrate

Failure Mode A Failure Mode B Failure Mode C Failure Mode D Failure Mode E Failure Mode F Failure Mode G
adhesive failure cohesive failure adhesive failure cohesive failure adhesive failure mixed Mode E cohesive failure in
at rigid disk in FRP at FRP/adhesive adhesive at FRP/concrete and Mode G concrete substrate
interface interface

Figure 2: Pull-off failure modes for FRP applications.

Failure Mode A: adhesive failure at rigid disk provides little information beyond a lower-bound pull-off capacity for
the system tested. Mode A is an indication of the use of an inappropriate adhesive system for fixing the test disk.
Failure Mode B: cohesive failure in FRP is an indication of poor through-thickness properties of the FRP. Such
failures are observed largely due to incomplete wet-out of the fibers or plies comprising the FRP. Figure 4 shows a
Mode B failure which occurred between plies of a multi-ply wet layed-up GFRP installation (Harries 2001). Mode B

60
failures have also been observed in preformed CFRP strip materials having a too high fiber volume ratio (Reeve
2005). Such failures may also result from environmental degradation of the FRP material itself.
Failure Modes C and E: adhesive failure at either adhesive interface is an indication of poor adhesion properties
which may result from a) improper selection (matching) of adhesive for adherand materials; b) contamination of
adhesive; c) improper or incomplete cure of adhesive; d) contamination or improper preparation or cleaning of
adherand surfaces; or e) environmental degradation. In the author’s experience, Mode E is more common than Mode
C; this is believed to reflect the greater variability in concrete surface preparation and the greater likelihood of
contamination of this interface during installation.
Failure Mode D: cohesive failure in adhesive is unlikely to be observed in conventional FRP applications. If
observed, Mode B is an indication of very poor adhesive properties likely resulting from contamination, incomplete
cure or environmental degradation. In these cases, however, Modes C or E are more likely.
Failure Mode F: mixed cohesive failure in substrate and adhesive failure at the adhesive/substrate interface is
commonly observed in FRP pull-off tests (Fig. 3). It is generally believed that this failure initiates in the concrete
substrate. The failure plane in concrete is circuitous; if it reaches the adhesive/concrete interface, the failure plane is
likely to follow the interface plane resulting in a mixed mode failure. It is typically observed that such a failure
propagates through the concrete on one “side” of the disk and through the interface on the other (Fig. 4), supporting
the described behavior. It is rare to have a mixed mode failure with the concrete failure interspersed with the
interface failure across the area of the disk. Mode F failures are reported with their proportions of each failure mode.
Failure Mode G: cohesive failure in concrete substrate is the desired failure mode, representing a sound FRP-
adhesive system. This is not to say that a Mode G failure is will meet acceptance criteria, but rather that the concrete
substrate – as the least controllable constituent – should be the “weak link” in the bonded FRP system. An initially
degraded substrate or poor surface preparation will lead to low Mode G pull-off strength values. The extent of
aggregate failure (as opposed to failure through only the cement paste) should be noted.

3.1 Lower-bound test results and acceptance criteria

If the pull-off test is being used as a QC/QA method having a minimum specified pull-off strength as an acceptance
criteria, it would seem reasonable to accept any failure mode whose capacity exceeds the acceptance criteria. Such
an approach must be used with great caution. For example, a Mode A failure that exceeds the acceptance criteria
may be considered to be a valid test although unseen damage in the remainder of the bonded system may exist
making such an inclusion ill-advised. Modes B and D, although exceeding an acceptance criteria suggest flaws or
poor preparation of the constituent materials of the bonded system. Modes C and E, similarly, suggest poor surface
preparation or contaminated interfacial surfaces. Pull-off strength acceptance criteria is limited by the tensile
capacity of the substrate concrete. In most applications, this substrate will consist of the paste-rich cover concrete
located at a formed surface. Estimates of the tensile strength of concrete vary. It is recommended that pull-off
strength acceptance criteria not exceed 0.33 f c ' (MPa).

4. APPLICATIONS OF PULL-OFF STRENGTH TESTING

Pull-off strength testing has been used by the author in a number of studies largely to establish QA/QC of bonded
FRP systems. In such cases, samples are selected at random from areas of the FRP application not affecting the
performance of the retrofit. In all studies a protocol ensuring that the concrete age exceeds 56 days and the FRP
system has cured for at least 7 days is followed. All samples are prepared with a smooth (no teeth) diamond core
barrel. The left side of Fig. 3 shows one such QA/QC program (for the beams reported in Harries et al. 2006). This
program assessed a CFRP application using two different adhesive types. As seen in Fig. 3, generally excellent pull-
off strength values were found for the high-modulus adhesive (average pull-off strength = 0.58 f c ' ) while acceptable
results (average = 0.52 f c ' ), although with significantly greater variability, were found for the low-modulus
adhesive. As reported in Fig. 3, most failures were Mode G, giving a good degree of confidence to this testing.
Pull-off strength testing has been used to study the effect of environmental exposure (Malvar et al. 2003) and initial
surface condition (Wan et al. 2006) on bonded FRP applications. One such study is reported by Harries et al. (2004)
and shown in the middle section of Fig. 3. In this study, pull-off testing of an epoxy-based concrete pipe lining
system was conducted for cases where the substrate was surface dry and where it was saturated. It can be seen in
Fig. 3 that the epoxy applied to the saturated surface had essentially the same pull-off strength (average = 0.43 f c ' )
although the failure mode changed from F and G in the dry application to E in the saturated application. This result
indicates some retarding effect on the adhesion due to the presence of water on the substrate for the material used.

61
Finally, Harries (2001) reports a small field study using pull-off testing to assess the in situ quality of a five-year old
GFRP application on the exterior girder of an Interstate overpass. This application is shown in Fig. 1. Not only was
the in situ pull-off capacity of interest, but also the effect of a road drain was questioned. As shown in Fig. 1, three
test samples were taken beneath the road drain and the remaining samples in regions unaffected by the drain. The
results are shown on the right side of Fig. 3. In this study, the pull-off strength results were relatively low (average
= 0.38 f c ' ) and no effect was seen due to the presence of the road drain. The results from this study were
inconclusive and would suggest that a low “guaranteed” pull-off strength be adopted. The highest test results were
Mode A failures – indicating that the adhesive used to secure the disks was barely adequate. Additionally a Mode B
failure (Fig. 4) was observed between laminates of the multi-ply hand layed-up GFRP application. Investigation of
all disks indicated excessively thick resin layers between often dry GFRP fibers. This observation clearly indicates
poor quality GFRP installation where wet-out was not ensured and may indicate inadequate rolling or working of the
GFRP sheets or sagging of the sheets during installation due to excessive resin.
1.0
normalized pull-off strength (1/f c' ) (MPa)

low-mod high-mod Harries et al. 2004 Harries 2001

G
0.8
1/2

A dry wet drain

G G
GG
A G
0.6 G G E
F CFG F
E A
G F
G GG F A
F G E F
0.4 F E GGA
E BF
G F

0.2

0.0 B
Figure 3: Pull-off strength results. Figure 4: Representative
failure modes.
5. CONCLUSIONS
ASTM D4541-type direct tension testing is easily adopted for QC/QA and other investigative applications of bonded
FRP systems. For applications to concrete, a Type 1 (ASTM 2002) pull-off test apparatus is used and the sample
should be core-drilled for reliable results to be obtained. Careful interpretation of the failure modes and resulting
disks can provide significantly more qualitative information than simply the pull-off strength. Representative
examples of QC/QA and surface preparation and environment exposure studies are presented, illustrating the
potentially wide applicability of this simple test technique.

6. REFERENCES
American Society for Testing and Materials (ASTM) (2002) D4541-02 Standard Test Method for Pull-Off Strength
of Coatings Using Portable Adhesion Testers, ASTM, West Conshohocken, PA.
International Concrete Repair Institute (ICRI) (2004) Guideline No. 03739 Guide to Using In-Situ Tensile Pull-Off
Tests to Evaluate Bond of Concrete Surface Materials, ICRI, Des Plaines, IL.
Harries, K.A. (2001) In Situ Tests on FRP Retrofit Application: I585 Northbound Bridge over SC85 Southbound in
Spartanburg SC, report to South Carolina Department of Transportation, October 2001. 5 pp.
Harries, K.A., Reeve, B. and Zorn, A. (2006) “Effect of Adhesive Modulus on the Monotonic and Fatigue Behavior
of Externally Bonded CFRP Strips” Proceedings of the 3rd International Conference on FRP Composites in Civil
Engineering (CICE 2006) Miami, December 2006.
Harries, K.A., Young, S., McNeice, D. and Warren, D. (2004) “Sprayed Epoxy Composite Materials for Structural
Rehabilitation” Proceedings of the 10th Underground Construction Technology Conference, Houston, Jan. 2004.
Malvar, L.J., Josji, N.R., Beran, J.A. and Novinson, T. (2003) “Environmental Effects on the Short-Term Bond of
Carbon Fiber-Reinforced Polymer (CFRP) Composites” ASCE Journal of Composites for Construction, Vol. 7,
No. 1, pp 58-63.
Reeve, B. (2005) “Effect of Adhesive Stiffness and CFRP Geometry on the Behavior of Externally Bonded CFRP
Retrofit Measures Subject to Monotonic Loads” M.Sc. Thesis, University of Pittsburgh. December 2005.
Vaysburd, A.M. and McDonald, J.E. (1999) “An Evaluation of Equipment and Procedures for Tensile Bond Testing
of Concrete Repairs” Technical Report REMR-CS-61, US Army Corp of Engineers, Washington, DC.
Wan, B., Petrou, M.F. and Harries, K.A. (2006) “Effect of the Presence of Water on the Durability of Bond Between
CFRP and Concrete” Journal of Reinforced Plastics and Composites, Vol. 25, No. 8, pp 875-890.

62
Third International Conference on FRP Composites in Civil Engineering (CICE 2006)
December 13-15 2006, Miami, Florida, USA

PREDICTION OF BOND FAILURE OF CONCRETE PRISMS BONDED

WITH FRP COMPOSITES
H. Toutanji
(Professor, University of Alabama in Huntsville, Huntsville, AL, USA)

P. Saxena
(Graduate Research Assistant, University of Alabama in Huntsville, Huntsville, AL, USA)

L. Zhao
(Graduate Research Assistant, University of Alabama in Huntsville, Huntsville, AL, USA)

ABSTRACT
The use of fiber reinforced polymer (FRP) for strengthening concrete structures has grown rapidly during the past
few years. In spite of exhibiting superior properties, the safety of usage is questionable as FRP undergoes brittle
debonding failures. Amongst the reported failure modes of FRP strengthened concrete beams, there has been limited
research in terms of intermediate crack induced interfacial debonding and fewer strength models are developed for
predicting such failures. It is suggested that conducting a simple shear test on the FRP bonded to concrete substrate
can simulate this type of failure mode. This paper is a study of the existing experimental and analytical work done to
understand the failure mode of FRP bonded concrete. Twelve specimens were tested to study the influence of
concrete strength and the amount of FRP on the ultimate load capacity of FRP-concrete bond under direct shear.
Existing experimental work was collected from literature and consists of an extensive database of 351 concrete
prisms bonded to FRP and tested in direct shear. The analytical models from various sources are applied to this
database and the results are presented.

KEYWORDS
FRP, Bond, Inorganic Epoxy, Bond length, Anchorage strength models.

1. INTRODUCTION
The concern with the use of FRP is the premature debonding failure. The aim of this study is to review and compare
the existing studies and predictive models on the FRP-to-concrete bond. Of the various premature failure modes
exhibited by FRP strengthened RC beams, intermediate crack induced debonding is a less commonly reported
failure mode. It however has a significant control over the strength of a major portion of the FRP strengthened beam
(Smith and Teng 2002). Few strength models focus on predicting midspan debonding. It is suggested in many
studies (Teng et al. 2003, Chen and Teng 2001, Seracino 2001) that conducting a simple shear test on the FRP
bonded to concrete substrate can simulate this type of failure mode. Hence, a model based on a simple shear test
may predict intermediate crack induced debonding accurately. This paper presents an experimental study of 12
specimens of FRP-to-concrete bonded prisms. An inorganic epoxy was used for bonding FRP sheets to the concrete
surface. This gave an excellent opportunity to compare the models even with specimens having a different epoxy
system. Further, anchorage strength models from various sources are applied to an extensive database of 351
specimens collected from literature and tested in single or double shear tests.

2. EXPERIMENTAL PROGRAM AND RESULTS

The pull out test specimens consisted of a concrete prism bonded with FRP strips by an inorganic epoxy. The
concrete prism had a dimension of 200 mm × 200 mm with a thickness of 130 mm. Concrete cylinders made from

63
the same batches were cast and tested according to ACI provisions. Also, the average modulus of elasticity of
concrete Ec was obtained as 33.5 GPa. Commercially available carbon fiber sheets with unidirectional plain weave
were used. The thickness of each layer of the carbon fiber is 0.165 mm measured by a dial caliper. Yarn fiber type
T300C, 3K was used. Its properties obtained by conducting a FRP coupon test are listed in Table 1.

Table 1. Properties of Inorganic Epoxy and CFRP

Ultimate Strain Tensile Strength Compressive Modulus of
(%) (MPa) Strength (MPa) Elasticity (GPa)
Inorganic Epoxy 0.6 - 23.6 4.1
CFRP 0.006 660 - 110

The surface preparations for concrete included roughening and sandblasting, followed by rinsing with a water jet.
The prepared surface was primed with inorganic epoxy. The carbon fiber sheet was also impregnated with the epoxy
and was placed on the marked bonding area of 50 mm × 100 mm. They were left at room temperature for at least 3
days for the matrix to fully dry. Each layer of fiber sheet had the same length. The thickness of bonded FRP sheet
was also measured after curing with a dial caliper. Five strain gages were installed on the FRP before testing at an
interval of 20 mm to monitor the strain distribution along the bond length, as shown in Figure 1. Loading and strain
information was obtained by a data acquisition system. The load cell and strain gauges were scanned at a time
interval of 1 second, recording the load-strain relation for each specimen.
P
1
FRP

Fracture Energy (N/mm)

0.8

Strain Gauges 0.6

Concrete Block
0.4
Experiment
Clamp Average Value
0.2
Proposed Model
Testing Pad
0
0 10 20 30 40 50 60 70
Compressive Strength (MPa)
Figure 1: Shear Test Set-up Figure 2: Variation of Fracture Energy

The concrete strength, experimental loads, and failure modes for all the specimens are listed in Table 2. The modes
of failure primarily observed are concrete shearing (CS) and fiber delamination (DL). Since the failure was initiated
due to either shearing of concrete (CS) or the fracture of concrete edge (initiation of DL mode of failure), it is
reasonable to assume the dependency of fracture energy on concrete strength. Zhao (2005) in his study compared the
experimental load to the bond failure load proposed by Taljsten (1996) and plotted the interfacial fracture energy per
unit area of the joint Gf (N/mm) with respect to the compressive strength of the concrete (MPa), as shown in Figure 2.
Zhao (2005) proposed a simple bilinear relationship between the variation of fracture energy and concrete strength, as
⎧⎪0.014 f c' 0 ≤ f c' ≤ 46.2 MPa
Gf = ⎨ (1)
⎪⎩ 0.65 f c' ≥ 46.2 MPa
and Pu = b f 2G f t f E f (2)

Equations (1) and (2) can be used together to find the ultimate debonding load Pu of the FRP-to-concrete bond.
Specimens were divided into three groups, namely, Group I, II and III according to the concrete compressive
strengths of 17.0 MPa, 46.2 MPa and 61.5 MPa, respectively. Further specimens in each group were numbered
based on the layers of carbon fiber sheet. Numbers 1, 2, 3, and 4 corresponded to three, four, five, and six number of
carbon fiber layers, respectively. A significant increase in the ultimate load was noted with an increase in the
number of carbon fiber sheets for group I. However, for groups II and III there was a slight increase in the ultimate
load between 5 and 6-layered fiber bonded concrete substrate. Also, an increase in the ultimate load was noted with
an increase in the concrete strength.

64
 Table 2. Experimental Results
Specimen I-1 I-2 I-3 I-4 II-1 II-2 II-3 II-4 III-1 III-2 III-3 III-4
Concrete Strength 17.0 17.0 17.0 17.0 46.2 46.2 46.2 46.2 61.5 61.5 61.5 61.5
(MPa)
No. of Layers 3 4 5 6 3 4 5 6 3 4 5 6
Experimental Load 7.56 9.29 11.64 12.86 12.55 14.25 17.72 18.86 13.24 15.17 18.86 19.03
(kN)
Failure Mode CS CS DL DL DL DL DL DL DL DL DL DL

3. ANALYTICAL REVIEW
Various models readily available in different studies were collected to predict the bond strength of FRP-to-concrete
bonded specimens. The models were collected from the studies of Chen and Teng (2001), Lu et al. (2005), Adhikary
and Mutsuyoshi (2001), and Dai et al. (2005) and were categorized according to the consideration of effective bond
length. The effective bond length is defined as the length beyond which any increase in the bond length of the fiber
cannot increase the anchorage load. To ascertain the extent of usefulness of shear test for predicting midspan
debonding in flexurally strengthened RC beams, an extensive database consisting of 351 rectangular concrete prisms
bonded with FRP plates tested in single and double shear tests was developed. These specimens were collected from
experimental studies conducted or reported by Zhao (2005), Adhikary and Mutsuyoshi (2001), Sharma et al. (2006),
Seracino (2001), Yao et al. (2005), Fu-quan et al. (2001), Chen and Teng (2001), Lu et al. (2005), and Nakaba et al.
(2001). The following conditions were applied to the prisms in order to include them in the database:

1. It was not preloaded or precracked,

2. It did not have any form of external anchorage between the concrete and FRP sheet or plate, and
3. It was not devised or subjected to any physical condition in order to induce a particular failure mode.

Different models were applied to the database to study the predictability. The experimental versus predicted load
was plotted for each of the models and statistical analysis was carried out. Figure 3 shows the scatter of the modified
bond strength model by Chen and Teng (2001). This model predicts the best results in terms of average
experimental-to-predicted bond strength ratio (1.58) and percentage of unsafe design (2.5%). Most of the specimens
lie in the safe zone, for this model providing conservative yet efficient predicted debonding loads. Zhao’s (2005)
model based on the interfacial fracture energy gave an average bond strength ratio of 1.314 and 21% unsafe design
as shown in Figure 4. While most of the specimens lie in the safe zone, they are highly scattered.

The model by Dai et al. (2005) highly overestimates the bond strength for most of the test data as seen in Figure 5,
with a large number of specimens lying in the unsafe zone. The average experimental-to-predicted bond strength
ratio is found to be 0.673 and the percentage of unsafe design is 94%.

Based on the tests conducted on the FRP bonded concrete prisms and the comparative study of all the models, a
pattern is observed in the results. It implies that some of the better predicting models could be used as a basis for
developing a model for intermediate crack induced debonding. This could be achieved by applying the model to
beams tested in bending and checking the validity through modifications.

40 40
35 35 Zhao
Chen & Teng Modified
Predicted Load (kN)

X=Y Line
Predicted Load (kN)

30 X=Y Line 30
25 25
Unsafe Unsafe
20 20
15 15
Safe Safe
10 10
5 5
0 0
0 10 20 30 40 0 10 20 30 40
Measured Load (kN) Measured Load (kN)

Figure 3: Chen and Teng’s Modified Model Figure 4: Zhao’s Model

65
 60

50

Predicted Load (kN)

40
Unsafe
Dai et al.
30 X=Y Line
Safe
20

10

0
0 10 20 30 40 50 60
Measured Load (kN)

Figure 5: Dai et al.’s Model

4. CONCLUSIONS
With a consideration of balance between the average ratio of predicted to experimental bond strength and percentage
of safe design, models by Chen and Teng, Khalifa et al., Yang et al., and Lu et al. are effective in predicting the
bond strength. Lu et al.’s model is dependent on several parameters that require lengthy calculations. Interfacial
fracture energy based models by Zhao, and Yuan and Wu are also recommended with emphasis on Yuan and Wu’s
model. Since none of the models considered, had any parameter dependent on the properties of the epoxy, it was not
possible to infer the effects of epoxy on the bond strength in the current study. Empirical models underestimate the
bond strength and have a scattered load prediction. It was also found that the underestimation of bond strength is
mainly the reason for attaining a low percentage unsafe design (<10%) for some of these models. The models that
overestimate the bond strength (average experimental-to-predicted bond strength ratio < 1) were usually found to
have a very high percentage of unsafe design (65~95%) as well.

5. REFERENCES
Adhikary, B. B., and Mutsuyoshi, H. (2001). “Study on the bond between concrete and externally bonded CFRP
sheet.” Proc., 6th Int. Symp. on Fiber Reinforced Polymer Reinforcement for Concrete Structures (FRPRCS-5), Vol.
1, pp 371-378.
Chen, J. F., and Teng, J. G. (2001). “Anchorage strength models for FRP and steel plates bonded to concrete.”
Journal of Structural Engineering, Vol. 127, No. 7, pp 784–791.
Dai, J., Ueda, T., and Sato, Y. (2005). “Development of the nonlinear bond stress-slip model of fiber reinforced
plastics sheet-concrete interfaces with a simple method.” Journal of Composites for Construction, Vol. 9, pp. 52–62.
Fu-quan, X. U., Jian-guang, G., and Yu, C. (2001). “Bond strength between CFRP sheets and concrete.” Proc., FRP
Composites in Civil Eng., Vol. 1, pp. 357–364.
Khalifa, A., Gold, W. J., Nanni, A., and Aziz, A. (1998). “Contribution of externally bonded FRP to shear capacity
of RC flexural members.” Journal of Composites for Construction, Vol. 2, No. 4, pp. 195–203.
Lu, X. Z., Teng, J. G., Ye, L. P., and Jiang, J. J. (2005). “Bond–slip models for FRP sheets/plates bonded to
concrete.” Engineering Structures, Vol. 27, No. 6, pp. 920–937.
Nakaba, K., Toshiyuki, K., Tomoki, F., and Hiroyuki, Y. (2001). “Bond behavior between fiber-reinforced polymer
laminates and concrete.” ACI Structure Journal, Vol. 98, No. 3, pp. 359–367.
Seracino, R. (2001). “Axial intermediate crack debonding of plates glued to concrete surfaces.” Proc., FRP
Composites in Civil Eng., Vol. 1, pp. 365–372.
Sharma, S. K., Ali, M. S. M., Goldar D., and Sikdar, P. K. (2006). “Plate-concrete interfacial bond strength of FRP
and metallic plated concrete specimens.” Composites Part B, Vol. 37, No. 1, pp. 1–10.
Smith, S. T., and Teng, J. G. (2002). “FRP-strengthened RC beams I: review of debonding strength models.”
Engineering Structures, Vol. 24, No. 4, pp. 385 –395.
Taljsten, B. (1996). “Strengthening of concrete prisms using the plate bonding technique.” International Journal of
Fracture, Vol. 82, pp. 253–266.
Teng, J. G., Smith, S. T., Yao, J., and Chen, J. F. (2003). “Intermediate crack–induced debonding in RC beams and
slabs.” Construction Building and Materials, Vol. 17, No. 6-7, pp. 447–462.
Yao, J., Teng, J. G., and Chen J. F. (2005). “Experimental study on FRP–to–concrete bonded joints.” Composites
Part B, Vol. 36, No. 2, pp. 99–113.
Zhao, L. (2005). “Characterizations of RC beams strengthened with carbon fiber sheets.” Ph.D. thesis, University of
Alabama in Huntsville, Alabama, USA.

66
Third International Conference on FRP Composites in Civil Engineering (CICE 2006)
December 13-15 2006, Miami, Florida, USA

STRENGTH MODEL FOR INTERMEDIATE CRACK DEBONDING IN

FRP-STRENGTHENED CONCRETE MEMBERS CONSIDERING
ADJACENT CRACK INTERACTION
J.F. Chen
(Lecturer, Institute for Infrastructure and Environment, Edinburgh University, Edinburgh, UK)

J.G. Teng
(Chair Professor of Structural Engineering, Department of Civil & Structural Engineering,
The Hong Kong Polytechnic University, Hong Kong, China)

J. Yao
(Professor, Department of Civil Engineering, Zhejiang University, Hangzhou, China)

ABSTRACT
Intermediate crack (IC) debonding is a common failure mode in RC beams strengthened by bonding an FRP plate to
the tension face. Such a debonding failure initiates in the high moment region and propagates towards a plate end.
This paper presents a new strength model for IC debonding failures. This strength model is based on a recent
analytical solution for interfacial debonding between concrete and externally bonded FRP and considers the
interaction between two adjacent cracks in an explicit manner. This enables factors such as the effect of load
distribution to be automatically considered. Other factors such as the internal reinforcement on IC debonding
behaviour can also be accounted for if their effect on cracking spacing is properly captured through an appropriate
crack spacing equation.

KEYWORDS
FRP, concrete beam, strengthening, interface, IC debonding.

1. INTRODUCTION
The performance of the FRP-to-concrete interface in providing an effective stress transfer is of crucial importance in
reinforced concrete (RC) structures strengthened with externally bonded fibre reinforced polymer (FRP) plates (or
sheets). A number of failure modes in FRP-strengthened RC members are directly caused by interfacial debonding
between the FRP and the concrete. One of these failure modes, commonly referred to as intermediate crack
debonding (IC debonding) (e.g. Teng et al. 2003; Lu et al. 2006), involves debonding of the FRP plate which
initiates at a major crack where the plate is under high tension and propagates along the FRP-to-concrete interface
towards a stress-free end of the plate.

The stress state of the FRP-to-concrete interface in an IC debonding failure may be compared with that in a simple
pull-off test specimen in which a plate is bonded to a concrete prism and is subjected to tension. However, a key
difference exists: in a normal RC beam with a tension-face FRP plate, the FRP plate spans over a series of cracks,
and as a result, the propagation of debonding from the initiation crack to the adjacent crack depends not only on the
tensile force in the FRP plate at the initiation crack but also that at the adjacent crack. The effect of this adjacent-
crack interaction has been considered in a recent model by Lu et al. (2006) for RC beams bonded with a tension face
FRP plate developed based on finite element results. This paper presents an alternative IC debonding strength
model, which accounts for adjacent-crack interaction in an explicit manner.

67
2. ANALYTICAL MODEL FOR ADJACENT-CRACK INERACTION
The behaviour of the FRP-to-concrete interface between two adjacent cracks (Figure 1) may be approximated by a
simple model as shown in Figure 2. The model is similar in geometry to a simple pull-off test. Their chief difference
lies in that both ends (i.e. at both cracks) of the FRP plate are subjected to tension in the present model. Several
recent studies have been conducted to investigate the behaviour of such a bonded joint model (Schilde and Seim,
2004, Teng et al. 2006, Chen et al. 2005, 2006). Teng et al. (2006) presented an analytical solution for this bonded
joint (Figure 2) based on a bilinear local bond-slip model between the FRP and the concrete but the solution is
involved and does not provide an explicit expression for the ultimate load. Chen et al. (2006) subsequently
developed a simpler analytical solution for the same FRP-to-concrete bonded joint model (Figure 2) based on a
linearly softening bond-slip model (i.e. neglecting the ascending branch in the bilinear bond-slip model). They
produced a simple yet fairly accurate solution in comparison with that given in Teng et al. (2006).

Load

Plate
x tp
P2 P1
tc Concrete prism
Adhesive P P3
FRP 4

(a) Elevation

Concrete prism

Cracks bc bp Plate
L
Concrete
Force in FRP (b) Plan
Force in FRP
Adhesive P1
P2
FRP

Figure 1: Intermediate crack debonding in FRP- Figure 2: FRP-to-concrete bonded joint model
plated RC beams between two adjacent cracks

The FRP-to-concrete bonded joint model (Figure 2) investigated by Teng et al. (2006) and Chen et al. (2006) is
subjected to the following loading: the plate is subjected to two tensile forces, P1 at the right end and P2 at the left
end with P1 ≥ P2 ≥ 0 ; the concrete prism is subjected to two forces P3 and P4 which can be either tensile or
compressive; and all these forces remain proportional throughout the loading process. The width and thickness of
the plate are bp and tp respectively, and those of the concrete prism are bc and tc respectively, which are all constant
along the length. The length of the bonded part of the plate (i.e. bond length) is L. The elastic moduli of the plate and
the concrete are Ep and Ec respectively.

If the tensile deformation of the concrete is neglected, which is usually conservative (Teng et al. 2006), Chen et al.’s
(2006) solution leads to the following expression for the ultimate load at the right end of the plate for the FRP-to-
concrete bonded joint:
b p 2G f E p t p ( 1 − β 2 ) −1 / 2 if L ≥ au (1)
P1,u =
b p 2G f E p t p sin( λL )[1 − β cos( λL )]
−1
if L < au
where
P2 (2a)
β=
P1
1 (2b)
au = arccos β
λ
τ 2 1 τf 1 (2c)
λ2 = f =
2G f E p t p δ f E p t p
in which τf is the peak shear stress of the linearly softening FRP-concrete bond-slip model, δf is the slip value when
the interfacial shear stress reduces to zero which signifies the shear fracture (or debonding or macro-cracking) of a
local bond element and Gf = τfδf/2 is the fracture energy of the interface.

68
3. NEW IC DEBONDING STRENGTH MODEL

It is significant to note that when P2 = 0 (i.e. β = 0, the plate is subjected to a tensile force at the right end only), and
Eq. (1) reduces to the solution given by Yuan et al. (2001) for a simple pull-off test of an FRP-to-concrete bonded
joint. It is also noted that Yuan et al.’s (2001) solution was the basis for Chen and Teng’s (2001) bond strength
model for the simple pull-off test. Following the same procedure as employed in Chen and Teng (2001), a new IC
debonding strength model considering the interaction between two adjacent cracks is thus proposed as follows based
on Eq. (1). The model is expressed in terms of the maximum (i.e. ultimate) stress in the plate at its right end for
convenience of design application. The maximum tensile force in the plate at the right end can be easily obtained by
multiplying the maximum plate stress by the cross sectional area of the plate bptp.

The ultimate stress in the plate at the critical crack where debonding initiates is then given by
Ep f c' (3)
σ IC = αβ σ β w β L
tp
where f c' is the cylinder compressive strength of concrete. The coefficient βσ reflects the effect of the ratio between
the plate stresses at the adjacent crack and the critical crack ασ; the coefficient βw reflects the effect of the difference
between the width of the plate bp and the width of the concrete prism (the width of the beam) bc; and the coefficient
β L reflects the effect of the bond length. They are found from
1
if L ≥ Le
1 − α σ2 (4)
βσ = 1
if L < Le
πL
1 − α σ cos
2 Le
b p + 2δ
if b p + 2δ < bc
bp (5)
βw =
bc
if b p + 2δ ≥ bc
bp
1 if L ≥ Le
βL = πL (6)
sin if L < Le
2 Le

The effective bond length in Eq. 4 is given by

2 E pt p (7)
Le = arccos α σ
π f c'
which reduces to the effective bond length equation in Chen and Teng’s (2001) model when the stress ratio ασ
equals zero. If both the internal reinforcement and the external reinforcement do not vary along the beam length, the
plate stress at the critical crack where IC debonding initiates σIC and that at the adjacent crack in the direction of IC
debonding propagation σadj may be approximately related to the bending moments ΜIC and Μadj at these locations.
That is
σ M (8)
α σ = adj ≈ adj ≤ 1
σ IC M IC

It may be noted that in Chen and Teng’s (2001) bond strength model, β w = ( 2 − b p / bc ) /(1 + b p / bc ) . Eq. (5) is used here
instead to better reflect the width ratio effect. Further details will be published separately. Based on a test database
collected from the literature, δ =10 mm may be adopted. Both bp and bc should be in mm in Eq. (5). Because of this
new expression for β w , the coefficient α in Eq. (3) needs to be reduced from 0.427 to 0.315 as the best-fit value, and
from 0.315 to 0.25 as the 95 percentile lower bound value for design use. It should also be noted that the bond length
L in this model should be taken as the distance between the crack where IC debonding initiates and the adjacent
crack in the direction of debonding propagation.

69
4. COMPARISON WITH TEST DATA
The above model has been compared with the test results of 18 cantilever slab specimens reported in Yao et al.
(2005) which failed due to IC debonding in the concrete with a thin layer of concrete attached to the FRP plate after
debonding. The remaining four specimens reported in Yao et al. (2005), which failed by either debonding at the
adhesive-concrete interface with little concrete attached to the FRP plate or concrete crushing, were excluded in this
comparison. Using the crack spacings and crack positions observed in the tests, the predictions from the present
model are in close agreement with the test results. The ratio of the test value to the prediction has an average value
of 1.02 with a CoV of 26.6% for the bending moment. The corresponding values for the axial strain of the FRP are
1.01 and 26.1% respectively. More details of these comparisons and further comparisons with other test data can be
found in Chen et al. (2006).

5. CONCLUSIONS
This paper has presented a new strength model for IC debonding failure in RC beams bonded with a tension-face
FRP plate based on an analytical solution for a simple FRP-to-concrete bonded joint model between two adjacent
cracks. The new strength model was found to show close agreement with a set of test data. An important feature of
the new model is that it considers the interaction between two adjacent cracks in an explicit and rigorous manner.
This feature enables factors such as the effect of load distribution to be automatically considered. The effects of
other factors such as the type and amount of internal tensile reinforcement can also be reflected if their effect on
crack spacing is properly captured by an appropriate crack spacing equation.

6. ACKNOWLEDGMENTS
The authors gratefully acknowledge the financial support provided by the Research Grants Council of the Hong
Kong Special Administrative Region, China (Project No: PolyU 5173/04E), the Natural Science Foundation of
China (Project No: 50378084), and Edinburgh University, UK.

7. REFERENCES
Chen, J.F., Yuan, H. and Teng, J.G. (2005). “Analysis of debonding failure along a softening FRP-to-concrete interface
between two adjacent cracks”, Proc., International Symposium on Bond Behaviour of FRP in Structures
(BBFS2005), Chen & Teng (eds), 7-9 Dec., Hong Kong, pp103-111.
Chen, J.F., Yuan, H. and Teng, J.G. (2006). “Debonding failure along a softening FRP-to-concrete interface between
two adjacent cracks in concrete members”, Engineering Structures, accepted for publication.
Chen, J.F. and Teng, J.G. (2001). “Anchorage strength models for FRP and steel plates bonded to concrete”, Journal of
Structural Engineering, ASCE, 127(7), 784-791.
Chen, J.F., Teng, J.G. and Yao, J. (2006). "Strength model for IC debonding in FRP-plated concrete members", In
preparation.
Lu X.Z., Teng J.G., Ye L.P. and Jiang J.J. (2006). “Intermediate crack debonding in FRP-strengthened RC beams: FE
analysis and strength model”, Journal of Composites for Construction, ASCE, accepted for publication.
Schidle, K. and Seim, W. (2004). “Experimental and numerical investigations of bond between CFRP and concrete”,
Proc., 2nd Int. Conference on FRP Composites in Civil Engineering, 8-10 Dec., Adelaide, Australia, 381-387.
Teng, J.G., Smith, S.T., Yao, J. and Chen, J.F. (2003). “Intermediate crack-induced debonding in RC beams and
slabs”, Construction and Building Materials, 17(6-7), 447-462.
Teng, J.G, Yuan, H. and Chen, J.F. (2006). “FRP-to-concrete interfaces between two adjacent cracks: theoretical
model for debonding failure”, International Journal of Solids and Structures, 42, in press.
Yao, J., Teng, J.G. and Lam, L. (2005). “Experimental study on intermediate crack debonding in FRP-strengthened RC
flexural members”, Advances in Structural Engineering, 8(4), 365-396.
Yuan, H., Wu, Z.S. and Yoshizawa, H. (2001). “Theoretical solutions on interfacial stress transfer of externally
bonded steel/composite laminates”, Journal of Structural Mechanics and Earthquake Engineering, JSCE, 18(1),
27-39.

70
Third International Conference on FRP Composites in Civil Engineering (CICE 2006)
December 13-15 2006, Miami, Florida, USA

STUDY ON BOND CHARACTERISTICS OF CFRP/STEEL DOUBLE-

LAP SHEAR JOINTS AT SUBZERO TEMPERATURE EXPOSURE
Ahmed Al-Shawaf, Riadh Al-Mahaidi and Xiao-Ling Zhao
(Department of Civil Engineering, Monash University, Victoria 3800, Australia)

ABSTRACT
This paper addresses the effect of subzero environmental exposures on the bond characteristics between CFRP
and steel plates. A series of tensile tests were conducted on CFRP/steel plates specimens joined together in
double-lap shear joints and subjected to 20ºC, 0ºC, -20ºC, and -40ºC environmental exposures. Normal modulus
(240 GPa) unidirectional carbon fibre plies were used in strengthening the epoxide matrix. The aim of the
current experimental work is to determine the debonding strength, elongation development and lap-shear stress
and slip variation under subzero exposures. This will help in establishing the feasibility of using CFRP in
retrofitting steel structures at extremely cold environments.

KEYWORDS
CFRP (Carbon Fibre Reinforced Polymer), Debonding Failure, Double-Lap Shear, Effective Bond Length,
Subzero temperature.

1. INTRODUCTION
The adoption of FRP composite laminates in strengthening aged steel structures has acquired some acceptance
recently as an alternative to traditional strengthening methods by virtue of their advantages, amongst which are
high stiffness-to-weight-ratio, enhanced fatigue life, corrosion resistance, controllable thermal properties and
faster field application. However, the utilization of this methodology depends on the FRP/steel durability, which
usually varies according to the nature of environment and the overall properties of the reinforcing fibres, resin
matrix, adhesive interface and the fabrication and conditioning techniques.
In terms of subzero environments, considerable work has been conducted on the behaviour of concrete
structures strengthened with FRP sheets subjected to cold climate conditions (Baumert, Green et al.,1996),
(Rivera, Karbhari et al.,1999),(Neale, Labossiere et al.,2001), (Karbhari,2002).Throughout all previous studies,
FRP/concrete bond-related issues have been experimentally investigated, concluded and recommendations were
proposed for adopting this retrofitting methodology under such climates. Nevertheless, when it comes to steel
adherends, the majority of the concrete practice design guidelines can not be superimposed due to
micro/macrostructural differences between both materials (Fawzia, Al-Mahaidi et al.,2004a). Literally, more
than 90% of past FRP literature under subzero conditions has concerned itself with the area of environmental
degradation and development of databases of test results for specific composite materials, especially those
utilized in aeronautical applications (Springer,1981). On the other hand, a very limited number of experimental
investigations were addressing mainly the bond durability of the CFRP/steel interface at subzero exposures
intended for civil applications (Karbhari and Shulley,1995),(Colombi, Fanesi et al.,2005). However, these works
adopted different matrix resins, fibres, and fabrication, conditioning and experimentation procedures than the
present work. This paper reports the results of a series of double strap shear tests conducted at Monash
University in order to characterise the debonding phenomena between wet lay-upped CFRP and steel plate
specimens loaded statically in axial tension under ambient and subzero temperatures. The prime objective of the
current study is to investigate and provide some feedback data on the short-term effect of subzero conditions,
encountered in civil infrastructure applications (i.e. retrofitting bridges, petroleum industry structures...etc), on
bond strength. Brief discussions are presented on failure modes, joint capacity, strain distribution along joint
effective bond length and effect of temperature on specimens’ lap-shear stress and slip.

2. EXPERIMENTAL PROGRAM
2.1. Materials

71
For the fabrication of the wet lay-up CFRP laminates, normal modulus unidirectional carbon fibre sheets were
used with a nominal modulus of elasticity of 240 GPa, tensile strength of 3800 MPa, and thickness of 0.176 mm.
Araldite 420 A/B (epoxy adhesive) was chosen as the resin matrix and adhesive interface at the same time since
wet lay-up method was adopted. Mild steel plates with a 350 MPa yield strength, and nominal thickness of 4.85
mm were used in the current test program.

2.2. Specimens Geometry, Test Setup and Thermal Conditioning

The proposed overall dimensions for the double-doubler joint-specimens (refer to Figure 1) were decided
according to the load-frame gripping system and the internal space of the environmental chamber used for
composite-specimens’ conditioning. A steel rig for producing the whole eight specimens to be tested in the

steel plate (L=150 mm) steel plate (L=150 mm)

L2 = 100 mm L1 = 75 mm
2 mm gap between plates

7 8
50 mm
1 2 3 4 5 6 control
S.G.

3-ply unidirectional
CFRP laminate
5 strain gauges 10 mm
@ 15 mm c/c
(Back Face) (Front Face)

Figure 1: Geometry of Double-Doubler CFRP/Steel Specimen Showing Strain Gauges Positioning

program in a single sample was devised. The aim was to attain the maximum possible consistency and
alignment for the specimens and also to have a flat composite outer surface for a reliable strain gauging.
Traditional mechanical surface-pretreatment of the main steel plates were followed according to the epoxy-
manufacturer’s data sheet before applying the first epoxy layer on both faces. These included cleaning,
degreasing, abrading and final degreasing. Three unidirectional carbon fibre plies were applied consecutively on
both faces of the steel plates. The sample was then cured for 14 days at room temperature and post-cured for 4
hours at 50°C. Cutting into the predetermined specimens’ geometry was achieved by means of a water-jet
cutting machine. Before testing, foil strain gauges were attached to CFRP outer surfaces as depicted in Figure 1.
All specimens were thermally aged inside an environmental chamber and tested individually in direct tension
with a screw driven load frame under displacement control (2 mm/min). The current test method has a good
conformance with both ASTM standard (D3528-96) and (D5868-01). However, main dissimilarity is due to the
adoption of wet lay-up FRP composite. For each target temperature (viz. 20°C, 0°C,-20°C and -40°C), two
specimens were tested. Once the target temperature, measured by a thermocouple attached to the CFRP surface,
was attained; an extra ten minute period was allowed to ensure specimen’s thermal stabilisation before the onset
of the test. The soaking time at each temperature was later validated using an experimental calibration procedure
similar to the one described in ASTM: D3528.

3. RESULTS AND DISCUSSION

3.1. Failure Patterns

Figure 2: Typical Failure Modes for Double-Doubler CFRP/Steel Specimens (Edge Views)

The prevailing failure mode for the CFRP/steel specimens tested at all present temperatures’ range (i.e. from 20
→ -40 °C) was found to be, as expected, complete debonding failure at the interface intermixed with minor
patches of CFRP fibre-tear failure which could be interpreted as an indication of good steel surface preparation.
As illustrated in Figure 2, the failure occurred at the shorter bondline (i.e. effective bond length) which is equal

72
to 75 mm. This failure was found similar to those observed from previous tests on CFRP/steel adherends
,(Colombi, Fanesi et al.,2005) at ambient temperatures.

3.2. Ultimate Load Capacity

The plot of the applied load with the joint strain readings for specimens tested at (20°, 0°,-20°,-40°C) are shown
in Figure 3 . This figure reveals explicitly the irrelevancy of the short-term subzero exposure within the current
temperature range on the lap-shear joint capacity, as all the CFRP specimens failed between (70-75) KN with no
specific trend. It also depicts a general tendency of higher composite strains with temperature increase.

80.0
70.0
60.0 CFRP Specimens
Axial Load (KN)

NS 20.1
50.0 NS 20.2
40.0 NS 0.1
NS 0.2
30.0 NS -20.1
20.0 NS -20.2
NS -40.1
10.0 NS -40.2
0.0
0 2000 4000 6000 8000
Strain @ the joint (με)

Figure 3: Axial Load vs. Strain at the Joint for Subzero Exposures (i.e. strain gauge no. 1, ref, Figure 1 )

3.3. Strain Distribution, Lap-Shear Stress and Slip relations

The strain distribution along the effective bond length was captured by the strain gauge readings at the surface
of the CFRP laminate. These readings are plotted with the distance from the joint under two exposures (i.e. 20°C
and – 40°C) at different load levels as shown in Figure 4. Close examination of Figure 4 (a) and (b) reveals that
for the same load ratio, the slope of the strain curves for the (-40°C) exposure is steeper than that of the (20°C)
ones. This means that the shear (bond) stress values are higher in the subzero temperature exposure. This could
be attributed to the effect of increased embrittlement and stiffness of the subzero-aged adhesive.

5000 5000
Specimen NS 20.2 Specimen NS -40.2

4000 4000
A xial S train (με)

Axial Strain (με)

3000 3000
Load Level increment Load Level Increment
from 0.1Pult. → 0.9 Pult. From 0.1 Pult.→ 0.9 Pult.
2000 2000

1000 1000

0 0
0 10 20 30 40 50 60 70 0 10 20 30 40 50 60 70
Distance from the joint (mm) Distance from the joint (mm)

a : Exposure Temperature= 20°C b : Exposure Temperature = - 40 ° C

Figure 4 : Strain Distribution along Effective Bondline for Increasing Load Level

The effect of adhesive increased stiffness at subzero temperature-exposure is further emphasised in Figure 5. It
shows plots of slip values vs. lap-shear stress for all CFRP/steel specimens (two at each temperature). The
maximum lap-shear stress occurs close to the joint (i.e. at 7.5 mm) which is believed to be due to changes in the
rheological and mechanical properties of the adhesive at subzero temperatures and the relatively short effective
bond length. A systematic and considerable increase in the maximum lap-shear stress is evident when the
exposure temperature dropped from (20°C) to (-40°C). All curves plateaued and lap-shear stresses declined after
attaining their maximum values. The exception to this is the (-40°C)-exposure specimens, where it is conceived
that the effect of growing embrittlement due to the extreme subzero exposure has prevented the lap-shear stress

73
from exhibiting this plateau with the current, relatively short effective bond length, and thus, any shifting in the
maximum lap-shear stress away from the joint.

50

Specimens
40

Lap-Shear Stress (MPa)

NS 20.1
NS 20.2
30
NS 0.1
NS 0.2
20
NS -20.1
NS -20.2
10 NS -40.1
NS -40.2
0
0 0.05 0.1 0.15
Slip (mm)

Figure 5 : Slip (at 7.5 mm from joint) vs. Lap-Shear Stress at Experimental Subzero Temperatures

4. CONCLUSIONS
Tensile experiments on subzero preconditioned CFRP/steel plates double-lap joints were conducted to
investigate the short-term interfacial bond characteristics. The following conclusions can be drawn based on the
limited test results:
• The devised rig for fabricating wet lay-up CFRP/steel plate main sample and the water-jet cutting method
have proved to yield high specimen’s reproducibility and thus good experimental consistency. As such, it is
recommended for such applications.
• The predominant failure mode encountered at all current short-term thermal exposure was an adhesive/steel
interface debonding failure.
• The short-term subzero exposure seemed to have insignificant effect on the lap-shear strength of the
CFRP/steel composite.
• The well-known embrittlement and ductility reduction phenomena of FRP composites at low temperatures
seemed to be responsible for decreasing strain, and consequently increasing lap-shear stress values of the
CFRP/steel specimens with temperature decrements from 20°C to -40°C.

5. REFERNCES
Baumert, M.E., Green, M.F. and Erki, M.A. (1996). "Low temperature behaviour of concrete beams
strengthened with FRP sheets", CSCE Annual Conference, Alberta, Canada, pp 179-190.
Colombi, P., Fanesi, E., Fava, G. and Poggi, C. (2005). "Durability of steel elements strengthened by FRP plates
subject to mechanical and environmental loads", CCC2005: Third international conference composites in
construction, Lyon, France.
Fawzia, S., Al-Mahaidi, R., Zhao, X.L. and Rizkalla, S. (2004a), Comparative study of failure mechanisms in
steel and concrete members strengthened with CFRP composites, in Developments in Mechanics of Structures
and Materials, A. J. Deeks and H. Hao, London, Balkemea Publishers: pp 71-76.
Karbhari, V.M. (2002). "Response of fiber reinforced polymer confined concrete exposed to freeze and freeze-
thaw regimes". Journal of Composites for Construction, Vol. 6, No. 1, pp 35-40.
Karbhari, V.M. and Shulley, S.B. (1995). "Use of composites for rehabilitation of steel structures -
determination of bond durability". Journal of Materials in Civil Engineering, Vol. 7, No. 4, pp 239-245.
Neale, K.W., Labossiere, P. and Theriault, M. (2001). "FRPs for strengthening and rehabilitation: Durability
issues", Composites in Construction: A reality, American Society of Civil Engineers, Capri, Italy, pp 102-109.
Rivera, J., Karbhari, V.M. and Dutta, P.K. (1999). "Effects of extended freeze-thaw exposure on composite
wrapped concrete cylinders". International SAMPE Symposium and Exhibition (Proceedings), Vol. 44, No. pt 2,
p 2231.
Springer, G.S. (1981). Environmental effects on composite materials, Technomic Pub. Co., Westport, CT /
USA.

74
Third International Conference on FRP Composites in Civil Engineering (CICE 2006)
December 13-15 2006, Miami, Florida, USA

THE MECHANISM OF EFFECT OF STRUCTURAL SPAN ON

INTERMEDIATE CRACK-INDUCED DEBONDING
Hemdan SAID
(Ph. D. Candidate, Ibaraki University, Hitachi, Ibaraki, Japan)

Zhishen WU
(Proffessor, Ibaraki University, Hitachi, Ibaraki, Japan)

ABSTRACT
In the current paper, a finite element analysis based on the smeared crack model and interfacial fracture mechanics
has been carried out to focus upon the effect of structural span on the intermediate crack-induced debonding.
Considering that for structures with multi distributed cracks, local debonding at some points does not mean a
complete debonding failure, the mechanism of debonding has been explained from the macroscopic point of view.
Based on this consideration, it has been shown that at the same levels of midspan bending moments of FRP-
strengthened girders with same cross section, as the span length increases, the gradient of FRP stress distribution
decreases. This implies that interfacial shear stress along the FRP-concrete interface decreases and consequently the
possibility of debonding decreases.

KEYWORDS
FRP sheets, Debonding; Smeared crack; Fracture mechanics; Span effect.

1. INTRODUCTION
Premature failure due to intermediate crack-induced debonding (IC debonding), which limits the full utilization of the
material potential, has been observed to be the most dominant failure mode in FRP-strengthened structures. Till date,
Extensive research has been carried out for investigating the IC debonding and studying different parameters which
affect it (e.g. Triantafillou, 1992; Täljstin, 1997; Wu et al. 1998; Wu and Niu, 2000; and Wu & Yin, 2002). Through
these researches, it has been recognized that debonding in flexural structures is affected by many parameters such as
flexural cracks spacing, amount and properties of FRP sheets, and mechanical properties of the adhesive layer. Another
important parameter which is observed to affect the debonding behavior is the span of FRP-strengthened structures
with same cross section. Limited research has dealt with this issue (Wu and Said, 2005 and Wu et al. 2006). In previous
study, Wu and Said, (2005) have shown that the effective bonding length increases with the increase of structural span.
This implies relieving shear stresses by transferring them over a larger length. Wu et al., (2006) observed
experimentally and confirmed numerically that for large spans, shear stress along the FRP-concrete interface decreases
and consequently the failure mode may shift from intermediate crack-induced debonding to the rupture of FRP sheets.
Based on the previous discussion, it is believed that the span length of an FRP-strengthened structure has a significant
effect on the interfacial shear stress distribution at the FRP-concrete interface and consequently on the IC debonding.
Till now, the mechanism of such effect is still ambiguous and need to be clarified. In the current study, a finite element
analysis based on the smeared crack model and interfacial fracture mechanics is performed to explain the mechanism of
span effect on the IC debonding. First, details of the finite element modeling have been discussed briefly. Second, the
mechanism of span effect has been explained by carrying out numerical simulation of a general kind of structures.
Finally, span effect in different code provisions has been discussed.

2. PROBLEM STATEMENT
It is observed from the available experiments of FRP-strengthened girders that girders with large spans may fail due
to FRP rupture before debonding. On the other hand, because most of the available guidliens and code provisions for
predicting debonding failure are based on experiments and analyses of small-size structures, they may not be

75
appropriate for application to large-scale ones and need to be modified. Understanding deeply the failure mechanism
associated with the IC debonding is of great importance for introducing powerful guidelines lead to devoloping safe
and economic designe methods. Therefore, the current study focuses upon clarifying the mechanism of the effect of
structural span on the IC debonding, which may be useful in future for updating the current design guidelines.

3. FINITE ELEMENT MODELLING

As a structural model, the full-scale girder shown in Fig. 1 is considered. The selected girder is a general kind of
structures with an I-shape section and it is widely used in the field applications for bridge construction. During
analyses, clear spans ranging from 10m to 30m are considered while cross sectional dimensions and reinforcement
are kept unvaried. For all different spans, girders are considered to be strengthened with 3 layers of FRP sheets.
Thickness and width of each layer is 0.128mm and 300mm respectively. The FRP amount has been selected to give
different failure modes (IC debonding or FRP rupture) for different spans. Although this type of girders is usually
prestressed in the practical applications, no prestressing of strands or FRP sheets has been considered in the current
simulation for comparing accurately among different spans. For accurate simulation, it is necessary to establish
appropriate models for considering propagation of cracks in concrete, bond-slip behavior between reinforcing bars
and concrete, and bonding behavior along the adhesive layer. In the following sections, the adopted concrete
cracking, material, interfacial and structural models have been presented.

3.1. Concrete Cracking Model

The rotating smeared crack model is adopted to simulate the initiation and propagation of cracks where the tensile
behavior is modeled by a simplified linear softening curve.

3.2. Material Models

The response of concrete in compression is modeled by the pre-defined parabolic curve given by the software. The
parabolic curve is based on fracture energy in compression by the definition of the crack band width of the element.
Concrete compressive strength and modulus of elasticity are considered as 58 MPa and 33.9 GPa respectively.
Steel reinforcing bars are considered as a linear elastic-perfectly plastic material. The elastic-perfectly plastic behavior
is modeled by Von Mises yield criterion with a yielding stress of 300 MPa and Young’s modulus of 200 Gpa.
Strands are considered as a linear elastic-perfectly plastic material with strain hardening. The elastic-perfectly plastic
behavior is modeled by Von Mises yield criterion with a yielding stress of 1730 MPa.
FRP sheets in general behave in a linear elastic manner up to rupture. The modulus of elasticity and the tensile
strength are considered as 240 GPa and 4.0 GPa, respectively.

3.3. Interfacial Models

The linear softening model shown in Fig. 2 is adopted to simulate the mode-II fracture of the adhesive. In this model,
when the local bond stress attains the local bond strength τf, micro-debonding is initiated and followed by a decrease
in the local bond stress until it becomes zero where macro-debonding is formed. The slope of the ascending branch
represents the interfacial stiffness ks, while the area under the curve represents the mode II fracture energy Gfb. The
values of τf, ks, and Gfb are chosen to be 4.0 MPa, 80 MPa/mm, and 0.4 MPa.mm respectively. On the other hand, the
interfacial behavior between concrete and reinforcing bars is modeled according to (CEB-FIP model code 1990).

3.4. Structural Model

Due to symmetry, only half of each girder has been analyzed with appropriate boundaries. Girders have been solved
under 3-point loading using the load control in combination with the arc-length control. Concrete is modeled by 4-
node plane stress elements, steel bars and FRP sheets are modeled by 2-node linear truss elements connected to
concrete by zero-thickness line interface elements. In order to accurately compare the behavior of different-span
girders, initial stresses due to own weight of girders are neglected.

4. VALIDATION OF THE MODEL

The validity and rationality of the finite element model has been checked by investigating the suitable mesh and by
comparing with some experimental results. Due to the limited space, it is not available to present these comparisons
here. Detailed validation and comparisons can be referred to in Wu et al. (2006).

76
 Local bond stress, τ
750mm Dia. 13mm Micro debonding initiation
*Unit: mm τf
P
Steel bars
Steel bars Unloading/reloading
Ks

530mm
Dia. 13mm

1000mm
Steel bar and strands PFRP sheets
Stirrups
Dia. 10mm
Anchorage
1 Macro debonding initiation
L(Variable) Steel bars
Dia. 6mm Gfb
Strands
Dia. 12.4mm
Slip δf
FRP sheets 350mm
Fig. 2: Linear softening model
Fig. 1: Details of the Simulated Girders. for FRP-concrete interface.

5. NUMERICAL SIMULATION AND DISCUSSION

As stated previously, the main purpose of the current research is to clearly show the mechanism of the effect of span on
the IC debonding. Therefore, three girders having spans within the field application range of 10m, 20m, and 30m are
simulated. All the simulated girders have the same cross sectional dimensions and reinforcement as shown in Fig. 1.
Therefore, results of the different spans are compared at the same levels of midspan bending moments. Results are
discussed for the case of multi distributed cracks from both microscopic and macroscopic points of view.
From the microscopic point of view, Fig. 3 shows a comparison among the shear stress distributions of the simulated
girders at the same level of midspan bending moment before the initiation of micro debonding. It is clear form this
figure that as the span length decreases, the local bond stress tends to be higher; this implies an increase in the
possibility of micro debonding and consequently an increase in the possibility of macro debonding. This result can
also be confirmed by drawing the shear stress distributions for the three girders at a bending moment corresponding to
the initiation of macro debonding of the 10m-girder as shown in Fig. 4. This figure illustrates that macro debonding
initiated for the 10m-girder around mid-span, while no debonding in 20m and 30m girders occurred. The shear stress
distribution of the first element attained macro debonding of the 10m-girder is shown in Fig. 5. It should be noted that
for structures with multi distributed cracks, local debonding at some points does not mean a complete debonding
failure. Therefore, the mechanism of debonding will be further explained from the macroscopic point of view. Herein,
the FRP stress distributions for different spans have been drawn in Fig. 6. This figure illustrates that at the same level
of midspan bending moments, FRP stresses at midspan are approximately equal for different spans. It is clear that as
the span length increases, the gradient of the FRP stress distribution decreases which means a reduction in the
interfacial shear stress and hence the possibility of debonding decreases.
Let us assume that the gradient of FRP stress distribution of the 20m-girder or the 30m-girder increases to the same
level of the gradient of the 10m-girder. This requires a higher level of midspan bending moment which causes a
5 4 1.2
4 3 1
Shear stress (MPa).

3 10m
Shear stress (MPa)

0.8 30m
Shear stress (MPa).

2 2 20m
0.6
1 1 0.4
0
0 0.2
-1 0 500 1000 1500 2000
0 500 1000 1500 2000 0
-2 -1
-0.2 0 500 1000 1500 2000
-3 -2
-4 -0.4
-5 -3 -0.6
Distance from midspan (mm) Distance from midspan (mm) Distance from midspan (mm)

Fig. 3: Shear Stress Distributions for Different Spans at the Same Level of
Midspan Bending Moment before Micro Debonding.

4 5 5
3 4 4
Shear stress (MPa).

3
Shear stress (MPa).

20m 3
Shear stress (MPa).

2 10m 30m
2 2
1 1 1
0 0
0
-1 0 1000
4
2000 3000 4000 5000 -1 0 2000 4000 6000 8000
-1 0 2000 4000 6000 8000 10000 12000
3

2
-2
-2 1
Initiation of -2
0 -3
-3 -1
0 20 40 60 80 1
Macro-Debonding -4 -3
-2

-4 -3
-5 -4
-4

Distance from midspan (mm) Distance from midspan (mm) Distance from midspan (mm)

Fig. 4: Shear Stress Distributions for Different Spans at a Level of Midspan Bending
Moment Corresponding to the Initiation of Macro Debonding of the 10m-Girder.

77
 0.5

FRP axial stress (MPa)

0 2000
Shear stress (MPa). -0.5 0 200 400 600 800 1000 10m
-1 1500 20m
-1.5 30m
-2 1000
-2.5
500
-3
-3.5 10m 0
-4
0 5000 10000 15000
-4.5
Bending moment at midspan (KN.m) Distance from mid-span (mm)

Fig. 5: Shear Stress Distribution of the First Element Fig. 6: FRP Stress Distributions for Different Spans at
Attained Macro Debonding of 10m-Girder the Same Level of Midspan Bending Moment

higher level of midspan FRP stress. In other words, for same levels of gradient of FRP stress distributions which
means same levels of interfacial shear stresses for girders with same cross section, the midspan bending moments of
girders having longer spans should be higher than those of girders having shorter ones. This implies that values of the
maximum FRP stress of long-span girders are higher than those of short-span ones under same conditions of
interfacial shear stresses. Therefore, long-span girders may have a debonding failure at higher levels of maximum
FRP stress than those of short-span ones and in some cases they may fail due to FRP rupture before attaining the
critical values of interfacial shear stresses. It is worth mentioning, for the case of a single localized crack, the situation
is analogous to the simple shear test and hence the structural span has no effect on debonding in this case.

6. SPAN EFFECT IN DIFFERENT CODE PROVISIONS

By reviewing the most famous code provisions for predicting debonding, it is noticed that the ACI 440.2R-02 code,
which adopts a reduction factor in the FRP ultimate strain, does not take the span effect into account. On the other
hand, although both of the JSCE and the FIB codes do not consider the span effect directly, they take it into account
indirectly. This is because both of the JSCE and the FIB codes are based on the difference in FRP stress between two
successive flexural cracks. Therefore, more quantitative investigations are required for updating these codes.

7. CONCLUSION
In the current research, a numerical simulation has been carried out for studying the effect of span of FRP-strengthened
structures on the intermediate crack-induced debonding and for explaining clearly the mechanism. Based on the
rotating smeared crack model and interfacial fracture mechanics, girders with different spans and same cross section
have been analyzed. FRP stress distributions as well as shear stress distributions for the different spans are compared at
the same levels of bending moments. It is found that for FRP-strengthened structures with multi distributed cracks, as
the span increases, the gradient of FRP stress distribution decreases and consequently local shear stresses along the
FRP-concrete interface decreases. Therefore, long-span girders may have a debonding failure at higher levels of
maximum FRP stress than those of short-span ones and in some cases they may fail due to rupture before attaining the
critical values of interfacial shear stresses. Based on this conclusion; more quantitative investigations are required for
updating the current code provisions for accurate prediction of debonding failure of large scale structures.

REFERNCES
CEP-FIP Model Code (1990).
Täljsten, B. (1997). "Strengthening of beams by plate bonding", ASCE Journal of Materials in Civil Engineering,
9(4), pp. 206-212.
Triantafillou, T. C., and Plevris, N., (1992). "Strengthening of RC Beams with Epoxy-Bonded FRP Composite
materials", Materials and Structures, V. 25, pp. 201-211.
Wu, Z. S., Matsuzaki, T., and Tanabe, K., (1998). "Experimental study on fracture mechanism of FRP-reinforced concrete
beams", Proceeding of JCI Symposium on Non-metallic FRP Reinforcement for Concrete Structures, pp. 119-126.
Wu, Z. S. and Niu, H. D. (2000). "Study on Debonding Failure Load of RC Beams Strengthened with FRP Sheets",
Journal of Structural Engineering, Vol. 46A, pp.1431-1441.
Wu, Z. S. and Said, H. (2005). "Debonding in FRP-strengthened flexural members with different shear-span ratios",
7th International Symposium on FRP Reinforcement for Concrete Structures (FRPRCS-7), USA, pp. 411-426.
Wu, Z. S., Said, H., and Iwashita, K. (2006). "Performance evaluation of strengthened structures with prestressed PBO
fiber reinforced polymer sheets", Annual Meetings and Symposium (IABSE 2006), Budapest, Hungary, (accepted).
Wu, Z. S. and Yin, J. (2002). "Numerical analysis on interfacial fracture mechanism of externally FRP-strengthened
structural member", JSCE Journal of Material, Concrete Structures and Pavements, 55(704), pp. 257-270.

78
Third International Conference on FRP Composites in Civil Engineering (CICE 2006)
December 13-15 2006, Miami, Florida, USA

STRAIN RATE EFFECTS ON STRENGTH OF UNIDIRECTIONAL FRP

FABRICS AND BOND TO CONCRETE
M. Saiid Saiidi
Professor, University of Nevada, Reno, Nevada, USA

Rita Johnson
Staff Engineer, Stantec, Reno, Nevada, USA

E. Manos Maragakis
Professor, University of Nevada, Reno, Nevada, USA

ABSTRACT
As part of a study towards the development of fiber reinforced polymer (FRP) seismic restrainers, the effect of
dynamic loading on the tensile strength and bond properties of unidirectional glass and carbon fiber fabrics was
studied. The bond studies concentrated on the FRP bond to concrete. Nearly 120 samples were tested at strain rates
of up to 100,000 µε/sec. The data showed that the tensile strength of FRP fabrics is insensitive to strain rate. In
contrast, bond strength is increased with strain rate as a logarithmic function. An equation developed based on the
regression analysis of the bond data is presented.

KEYWORDS
Dynamic tests, Fiber reinforced polymer, FRP/concrete bond, Strain rate

1. INTRODUCTION

Seismic restrainers are used to prevent excessive relative movements of bridge superstructures. They are mostly
made of steel cables, rods, or occasionally, plates. The potential use of fiber reinforced polymer (FRP) fabrics as
seismic restrainers was investigated in an exploratory study at the University of Nevada, Reno. The study consisted
of material testing, development of FRP restrainers, and shake table testing of the restrainers. Because seismic
restrainers are subjected to dynamic loading, the effects of strain rate on the tensile and bond strength of FRP was
investigated. This article presents a brief summary of the study of the materials. More details are provided in
(Johnson et al., 2005).

2. STRAIN RATE STUDIES

Past research has shown that material strength generally increases at higher strain rates. For example at 50,000
µε/sec, concrete compressive strength increases by approximately 25% (Kulkarni and Shah 1998). In steel, loading
at a high rate results in the yield stress increasing by 10 to 20 percent (Paulay and Priestley 1992). The reason for
the increase of strength at higher strain rates is believed to be from the extremely localized yielding due to enhanced
bond among the molecules at high rates. Shake table testing of steel restrainers has shown that strain rate in the
restrainers may reach 100,000 µε/sec (Camargo-Sanchez et al). Past studies on FRPs and bond to concrete have
been under static loads. The effect of strain rates on FRP fabrics and FRP bond to concrete has not been studied.

2.1 Tensile Strength of FRP Specimens

A test plan was developed to assess the strain rate effect on the strength of unidirectional FRP fabrics tested in
tension in the fiber direction. Eighty, 25.4 mm wide, 305 mm long glass and carbon fabric strips were tested in an

79
MTS load frame testing machine at various strain rates. The FRP material properties are listed in Table 1. The
maximum strain rate that could be applied was 100,000 µε/sec. All the material tests were conducted at a static rate
and five constant dynamic strain rates of 1,000, 5,000, 10,000, 50,000, and 100,000 µε/sec. A minimum of three
specimens were tested at each rate. One-half of the samples were tested without any coating but the remaining
samples were coated with a silicon elastomer, Sylgard 184. This coating was used to serve as a matrix for the
fabrics while maintaining their flexibility. Because the application of FRP in this study was for seismic restrainers,
it was essential that the fabrics maintained their flexibility and buckled out of plane when subjected to compression.
This would avoid overstressing the bonded area. Past research has shown Sylgard coated FRPs can undergo large
deformations while maintaining their strength and durability (Luo and Mitra 1999). The end segments of the strips
were coated with epoxy to facilitate clamping by the grips in the testing machine. Figure 1 shows a typical test
setup.

Table 1 – FRP Material Properties

Fiberglass Composite
Design Specified
Ultimate Tensile Strength 460 MPa 575 MPa
in Primary Direction 0.58 kN/mm width 0.75 kN/mm width
Elongation at Break 2.2% 2.2%
Tensile Modulus 20.9 Gpa 26.1 Gpa
Laminate Thickness 1.3 mm 1.3 mm
Primary Fiber Glass

Carbon Fiber Composite

Design Specified
Ultimate Tensile Strength 745 MPa 876 MPa
in Primary Direction 0.75 kN/mm width 0.89 kN/mm width
Elongation at Break 1.2% 1.2%
Tensile Modulus 61.5 Gpa 72.4 Gpa
Laminate Thickness 1.0 mm 1.0 mm
Primary Fiber Carbon

Figure 1: FRP Strip Test Setup Figure 2: FRP-Concrete Bond Test Setup

The measured tensile strengths for different strain rates and fibers were evaluated to determine the influence of
strain rate. Table 2 shows the results for CFRP samples. The table shows data for both the uncoated and coated
CFRP specimens. It can be seen that the FRP strength showed considerable scatter and that no consistent trend
could be observed in the data with respect to the effect of strain rate. A similar pattern of response was observed in
GFRP specimens. The tensile strength of the coated samples was always greater than that of the uncoated
specimens, indicating the binding action of the coating material that better mobilized the fibers. The rupture of the
uncoated strips was always at the edge fibers and was controlled only by a few fibers. In coated fibers a larger
number of fibers were ruptured. Nonetheless, increasing strain rate did not appear to affect the strength of the FRP
samples regardless of coating.

80
The insensitivity of FRP laminate strength to strain rate may be attributed to the mechanism by which the strength is
provided. It is well known that FRP fabric strength is always lower than the strength of the constitutive fibers
because of the presence of the matrix. Because of the relatively low modulus of elasticity of the matrix, high strain
rates are unlikely to affect the matrix strength. In materials such as steel and concrete, the bond among the
molecules is sensitive to strain rate and hence dynamic loading increased the capacity.

Table 2 – Measured Data for CFRP Samples

25.4 mm Fabric Strips 25.4 mm Elastomer Coated Fabric Strips Carbon Bond Tests (25.4 mm strips)
Rate Strip # Max Strength Ave Strip # Max Strength Ave Strip # Max Max Ave Ave
Force Stress Force Stress Force Bond Stress Bond
(Strength) Strength
(µε/sec) (kN ) (MPa) (MPa) (kN ) (MPa) (MPa) (kN ) (MPa) (MPa)
Static 1 11.73 469 1 20.92 824
2 7.38 295 2 19.32 761 3 8.72 4.50
3 15.12 610 3 18.59 732 4 6.67 3.45
4 10.91 441 454 4 18.99 748 766 5 8.18 4.23 4.06
5000 8 14.66 586 8 22.59 889 9 9.34 4.83
9 13.74 607 9 20.21 796 10 10.85 5.61
10 12.43 496 563 10 17.84 702 796 11 7.65 3.95 4.80
10000 11 14.27 572 11 18.61 733 0.00
12 13.91 531 12 19.52 769 12 12.01 6.21
13 13.31 496 13 17.50 689 13 8.81 4.55
533 14 16.49 649 710 14 9.88 5.10 5.29
50000 14 11.82 448 15 18.82 741 15 8.54 4.41
15 14.27 510 16 17.26 679 16 7.25 3.75
16 11.26 496 485 17 20.24 797 739 17 13.34 6.89 5.02
100000 17 13.28 517 18 17.05 671 18 9.61 4.96
18 14.11 600 19 17.09 673 19 8.85 4.57
19 16.12 586 567 20 16.24 639 661 20 10.28 5.31 4.95

2.2 Bond Strength

FRP restrainers that were developed as a part of the larger study were to be attached to concrete by epoxy resin.
Bond strength to concrete is a function of the strength and stiffness of the epoxy resin, the bond area, and strength of
concrete (Teng et al 2001). Typical static bond strength tests have shown that failure occurs in the form of near
surface shear in concrete. The effect of dynamic loading on bond strength was studied by testing 40 single-lap,
FRP-concrete bond specimens (Fig. 2). Twenty specimens consisted of GFRP strips and the other 20 of CFRP.
The specimens were loaded such that the eccentricity of the load was near zero. The middle 203 mm portion of the
FRP strips that were not attached to either the grips or bonded to the beams were coated with the flexible elastomer.
A 279 mm x 102 mm x 12.7 mm steel plate was anchored to the concrete beams to attach the beams to the grips in
the testing machine. A minimum of three specimens per strain rate (static; 1,000; 5,000; 10,000; 50,000; 100,000
µε/sec) for each of the glass and carbon FRP/concrete bond specimens were tested until failure.

An effective bond length of 69 mm and 76 mm was used for the GFRP and CFRP strips, respectively. The bond
lengths were purposely made shorter than what was necessary to develop the full strength of the composite strips to
force bond failure (Teng et al 2001). The surface area of concrete was prepared using standard procedures for bond
preparation, the details of which are presented in (Johnson et al. 2005).

Bond failure occurred typically in concrete near the interface with FRP strips regardless of the strain rate (Fig. 3).
The right three columns in Table 2 list the bond test results for CFRP specimens. The average measured static bond
strength of the samples was 3.7 MPa for the GFRP specimens and 4.06 MPa for the CFRP specimens. The fact that
these strengths are close confirms that concrete strength controls FRP/concrete bond strength. The bond strength
data in Table 2 show that bond strength is improved as strain rate increases. This is attributed to the sensitivity of
concrete shear strength to strain rate. The shear strength of concrete is a function of its compressive strength and the
compressive strength is known to increase with strain rate (Kulkarni and Shah 1998). Because the bond strength is
controlled by concrete properties and not the fiber type, the entire bond test data were combined (Fig. 4) and a
regression analysis was conducted. The amplification factor for bond strength as a function of strain rate was found
to be: 0.052 Ln (strain rate) + 0.736. Based on this equation at a strain rate of 50,000 µε/sec the bond strength

81
increases by approximately 25%. Because bond failure is undesirable and for added conservatism, the increase in
bond strength under dynamic loading was neglected in the design of the seismic restrainers.

1.8

Strength Ratio (Dynamic/Static)

1.7 y = 0.052Ln(x) + 0.736
1.6
1.5
1.4
1.3
1.2
1.1
1
Figure 3: Typical Bond Failure
100 1000 10000 100000
Strain Rate (Microstrains/Second)

Figure 4: Effect of Strain Rate on Bond

3 CONCLUSIONS
The data discussed in this article demonstrated that the tensile strength of unidirectional GFRP and CFRP fabrics is
independent of the strain rate of up to 100,000 microstrains per second. The effect of higher rates was not studied.
In contrast, the bond strength of FRP fabrics to concrete was found to depend on the strain rate. An equation
representing the trend in the data was developed based on regression analysis of the combined data for GFRP and
CFRP specimens. Even though the bond strength improves under dynamic loading, it is recommended that the bond
design for seismic restrainers be based on the static bond strength for a conservative design.

4 ACKNOWLEDGEMENTS

This study was funded by the National Cooperative Highway Research Program, Innovations Deserving Exploratory
Analysis (NCHRP-IDEA) with supplemental funds from the Nevada Department of Transportation (NDOT). The
authors are indebted to S. Luo, P. LaPlace, P. Lucas, and R. Nelson of UNR for their advice and assistance. E. Fyfe
and S. Arnold of the Fyfe Co. are thanked for donating the FRP fabrics and the epoxy resins.

5 REFERENCES

Camargo-Sanchez, F., Maragakis, E. M, Saiidi, M., Elfass, S., (2004) “Seismic Performance of Bridge Restrainers at
In-Span Hinges”, Civil Engineering Department, Report N. CCEER 04-04, University of Nevada Reno.
Johnson, R., Saiidi, M. and Maragakis, E., (2005) "A Study of Fiber Reinforced Plastics for Seismic Bridge
Restrainers," Center for Civil Engineering Earthquake Research, Department of Civil Engineering, University of
Nevada, Reno, Nevada, Report No. CCEER-05-2.
Kulkarni, S. M. and Shah, S. P., (1985) “Response of Reinforced Concrete Beams at High Strain Rates,” ACI
Structural Journal, 95 (6), 705-715.
Luo, S. Y., and Mitra, A., (1999) “Finite Elastic Behavior of Flexible Fabric Composite under Biaxial Loading,”
Journal of Applied Mechanics, 66, 631-638.
Nakaba, K., Kanakubo, T., Furuta, T., Yoshizawa, H., (2001) “Bond Behavior between Fiber-Reinforced Polymer
Laminates and Concrete,” ACI Structural Journal.
Paulay, T., and Priestley, M.J.N., (1992) Seismic Design of Reinforced Concrete and Masonry Buildings, John
Wiley and Sons, USA.

82
Third International Conference on FRP Composites in Civil Engineering (CICE 2006)
December 13-15 2006, Miami, Florida, USA

BOND BEHAVIOR OF CFRP STRENGTHENED FULL-SCALE

PRESTRESSED CONCRETE BRIDGE GIRDERS
Owen Rosenboom
(PhD Candidate, North Carolina State University, Raleigh, NC, USA)

Sami Rizkalla
(Distinguished Professor, North Carolina State University, Raleigh, NC, USA)

ABSTRACT
This paper presents the bond characteristics of carbon fiber reinforced polymer (CFRP) flexural strengthening
systems for prestressed concrete. Two large-scale prestressed concrete bridge girders were strengthened with
externally bonded CFRP precured strips and tested monotonically to failure. Both girders failed due to FRP
debonding propagating from intermediate flexural cracks. The main test variable considered in the experimental
program was the amount of FRP debonding mitigation provided. Test results show that using transverse anchorage
wraps along the length of the girder can increase the debonding strains of the longitudinal CFRP strips by over 20
percent. The full program with testing of five girders is described along with preliminary analytical results.

KEYWORDS
Reinforced concrete, prestressed concrete, FRP, debonding.

1. INTRODUCTION

Strengthening concrete structures with CFRP materials is becoming attractive solution to strengthen existing
bridges. While the flexural failure modes due to concrete crushing and FRP rupture are well known (Rosenboom et
al., 2006), the bond behavior of large-scale plated structures can still not be predicted with confidence. For long-
span FRP plated reinforced concrete, with the longitudinal strengthening extending to near the supports, the
common failure mode is interface debonding propagating from flexural cracks around midspan towards the supports.
Various analytical models have been proposed to predict the strain in the FRP at debonding including fracture based
models (Teng et al., 2000) and models based on mechanics (Sebastian 2002).

As part of an extensive research project funded by the North Carolina Department of Transportation examining the
cost-effectiveness and value engineering of CFRP repair and strengthening for prestressed concrete, a study has
been initiated to determine the bond characteristics of CFRP strengthening systems for prestressed concrete. Two
girders were strengthened with CFRP precured strips and tested to failure, which occurred due to FRP debonding
propagating from the midspan region. The main test variable examined was the amount of debonding mitigation
provided. Test results show that debonding strain could be increased with a properly designed anchorage system.

2. EXPERIMENTAL PROGRAM

2.1 Test Girders and Strengthening Configuration

Two 45 year-old 9140 mm long prestressed concrete test girders were extracted from a bridge and delivered in good
condition prior to strengthening with no visible cracks. Both girders were prestressed with ten 1723 MPa stress-
relieved strands with an average prestressing force of 71.7 kN per strand measured experimentally. The strand
configuration consisted of both straight and harped prestressing strands as shown in Figure 1. Concrete core
samples were taken after testing and the average concrete compressive strength for both girders was 61.4 MPa. For
both girders the main longitudinal CFRP strengthening was two CFRP precured strips (AFRP=119 mm2), having a

83
modulus of elasticity of 164.8 GPa and a tensile strength of 2800 MPa. The strips were bonded to the concrete with
a high-modulus, high-strength 2-part structural adhesive with a shear modulus of 1.03 GPa provided by the
manufacturer. Transverse wet lay-up CFRP sheet “U-wraps” 152 mm wide were provided on each web at a spacing
of 914 mm along the length of girder EB1S. Similar U-wraps were provided for girder EB1SB on one side only, to
study the bond characteristics without U-wraps. The 1.3 mm thick U-wrap material was also tested, and the
modulus of elasticity was found to be 51.7 GPa with an ultimate tensile strength of 710 MPa. The CFRP was
installed according to NCHRP Report 514 (Mirmiran et al. 2004) under simulated field conditions by experienced
contractors.

Figure 1: Girder cross-section and strengthening configuration (in mm)

2.2 Test Setup and Instrumentation

The girders were supported by concrete blocks and tested using a 490 kN actuator in 3-point bending with 216 mm
elastomeric bearing pads provided on each side. Displacement was measured through the use of potentiometers at
midspan and at the support sections. Strain gauges were used to measure concrete compressive strain and were also
applied to the longitudinal CFRP along the length of the girder to measure FRP tensile strains for girder EB1SB.
Each girder was loaded up to the cracking load then unloaded to determine the effective prestressing force, and then
the girder was loaded at a rate of 2.5 mm per minute up to failure.

3. TEST RESULTS
The cracking load of girder EB1S was 57 kN. The effective prestress force per strand, determined from crack re-
opening load, was found to be 72.1 kN. At yielding of the prestressing strands (at a load of 102.3 kN), numerous
flexural cracks had formed around midspan at a stabilized spacing of approximately 100 mm. Near failure the
flexural cracks bifurcated at the bottom of the web due to the plating constraint. Failure occurred due to
intermediate crack debonding which propagated from the midspan towards the supports at a load level of 176 kN.
The experimentally measured debonding strain at failure was 0.0122 mm/mm, 72 percent of the ultimate strain
measured from tensile tests. The interface along which the debonding propagated was mixed: in certain locations
around midspan the failure surface was the concrete paste layer just above the longitudinal CFRP, whereas closer to
the supports the failure surface was either at the CFRP-adhesive interface or within the strip (interlaminate failure).

The cracking load of girder EB1SB, with debonding mitigation on one side only, was 57.8 kN. From the crack re-
opening load, the effective prestress force was determined to be 71.2 kN. The behavior of this girder was similar to

84
girder EB1S before and after cracking and yielding of the prestressing strands, which indicates that the presence of
the U-wraps did not influence flexural cracking. The girder failed due to intermediate crack debonding at a load
level of 161.9 kN as shown in Figure 2b, with an experimentally measured debonding strain of 0.0100 mm/mm,
which is 59 percent of the tensile strength of the CFRP. The debonding initiated near midspan on the side without
U-wraps and propagated towards the support. The failure interface initially was the concrete paste layer at the
bottom of the web and then shifted to the CFRP-adhesive interface towards the supports. The increase in FRP strain
at debonding due to the presence of the U-wraps at 914 mm spacing was 22 percent higher. The load versus
deflection curves for girders EB1S and EB1SB along with a control girder described in Rosenboom et al. (2005) are
shown in Figure 2a. The increase in ultimate load compared to the control girder was 19.3 percent for girder EB1SB
and 9.6 percent for girder EB1S. The measured axial tensile strain in the externally bonded CFRP strip versus the
distance along the beam is shown in Figure 3a.

Figure 2a: Load-deflection behavior of tested girders Figure 2b: debonding failure of girder EB1SB

Figure 3a: Length along beam v. tensile strain in FRP Figure 3b: Length along beam v. interface shear stress

4. ANALYTICAL MODELING
There are several behaviors of prestressed concrete that make strengthening with externally bonded CFRP is more
effective than regular reinforced concrete. The effective prestress force in the beam imposes an initial compressive
strain in the beam soffit that is present regardless of whether the beam is propped during strengthening. This initial
strain in the beam soffit is directly translated into extra CFRP force that can be developed prior to debonding failure
compared to what is traditionally developed for a reinforced concrete beam. Most of the developed models assume
that the member is already cracked at the time of strengthening, which is rare for typical prestressed concrete

85
members. Even if the prestressed concrete member has been subjected to overloading conditions during its service
life and is cracked at the time of strengthening, the prestressing force will promote crack closure increasing the
debonding strains of the CFRP at failure.

From equilibrium of the axial forces along the length of a plated beam the interface shear stress (τi) can be
determined by the following equation:
dε p
τi = tp Ep
dx
where tp and Ep are the thickness and modulus of elasticity of the plate and εp is the axial strain in the plate along the
length of the beam, x. Application of this equation to the measured strain values for girder EB1SB are shown in
Figure 3b. The predicted strain gradient in the externally bonded FRP material along the beam can be found by
conducting a moment-curvature analysis of the section assuming flexural failure. Since the axial strain gradient is
an important factor affecting the bond behavior, the stress-strain characteristics of the longitudinal steel
reinforcement becomes very important. Stress relieved prestressing strands, with a less abrupt transition from elastic
to plastic behavior than regular reinforcing steel, place less of a demand on the interface shear stress in this
transitional zone along the beam. At the plate termination point a different state of stress exists, with shear and
normal stresses acting together due to many factors including the abrupt termination of the plate. At a large distance
away from the termination point, peeling stresses become smaller in magnitude, and are influenced mainly by
aggregate interlock once debonding propagates. Therefore, an assumption can be made that intermediate crack
debonding is mainly due to mode II interface shearing stresses, not an interaction between shear and normal stresses.

Future tests will include three additional prestressed concrete C-Channel girders to examine the effect of the number
of CFRP layers and the type of CFRP material (precured or wet lay-up). In each case the tensile strains along the
length of the girder will be measured and used to identify failure criteria to an analytical model which will be
presented at the conference.

5. CONCLUSIONS
As part of a comprehensive research project sponsored by the North Carolina Department of Transportation, five
prestressed concrete bridge girders are considered to evaluate the bond behavior and performance of CFRP
strengthening systems. Test variables include the amount of CFRP material, type of CFRP system (precured or wet
lay-up), and presence of transverse U-wrap anchorage. The test results of two girders are presented in this paper
showing that increases in debonding strain of 22 percent can be achieved through the use of transverse wet lay-up
CFRP U-wrap anchorage.

6. ACKNOWLEDGEMENTS
The authors would like to thank the North Carolina Department of Transportation for funding this research, and for
numerous industry partners for supplying materials and labor for FRP installation. The authors would also like to
thank the dedicated staff at the Constructed Facilities Laboratory.

7. REFERENCES
ACI Committee 440, (2002) “Design and Construction of Externally Bonded FRP Systems for Strengthening
Concrete Structures (ACI 440.2R-02)”, ACI Manual of Concrete Practice, American Concrete Institute.
Mirmiran, A., Shahawy, M., Nanni, A., Karbhari, V., “Bonded Repair and Retrofit of Concrete Structures Using
FRP Composites,” NCHRP Report 514, 2004.
Rosenboom, O.A., Rizkalla, S., (2006) "Flexural Behavior of Aged Prestressed Concrete Girders Strengthened with
Various FRP Systems", accepted for publication in Construction and Building Materials.
Sebastian, W.M. (2002a) “Sensitivities of strength and ductility of plated reinforced concrete sections to preexisting
strains,” Journal of Structural Engineering, V. 128, No. 5, pp. 624-636.
Teng, J.G., Chen, J.F., Smith, S.T., Lam, L. (2002) FRP Strengthened RC Structures, John Wiley and Sons, Ltd,
England.

86
Third International Conference on FRP Composites in Civil Engineering (CICE 2006)
December 13-15 2006, Miami, Florida, USA

BOND CHARACTERISTICS OF GFRP POST-INSTALLED ANCHORS

Ehab A. Ahmed
(Ph. D. Candidate, University of Sherbrooke, Sherbrooke, Quebec, Canada)

Ehab F. El-Salakawy
(Canada Research Chair Professor, University of Manitoba, Winnipeg, Manitoba, Canada)

Brahim Benmokrane
(NSERC Research Chair Professor, University of Sherbrooke, Sherbrooke, Quebec, Canada)

ABSTRACT
This paper addresses the bond behavior of GFRP post-installed adhesive anchors embedded in plain concrete. A
total of 15 sand-coated GFRP V-ROD bars of 25.4 mm diameter (#8) were tested at different embedment lengths
ranging from 5db (127 mm) to 15db (381 mm). The anchors were installed in wet holes using epoxy based adhesive
(HIT RE 500). The experimental results indicated the adequate performance of this type of post-installed GFRP
adhesive anchors. The capacity of these anchors is achieved at a considerably smaller embedment length.

KEYWORDS
Bond, GFRP, Anchor, Adhesive.

1. INTRODUCTION
The resistance to corrosion and chemical attack, high strength-to-weight ratio, and ease of handling of FRP rods
make them a better alternative to steel reinforcement in concrete members subjected to severe environmental
conditions. The adhesive anchor generally consists of a reinforcing bar inserted into a drilled hole in hardened
concrete with a structural adhesive acting as a bonding agent between the concrete and the reinforcing bar (Cook et
al. 1992). The post-installed anchors can be driven in almost any desired position in hardened concrete by installing
in holes drilled into concrete. These anchors are usually needed for strengthening and rehabilitation of deteriorated
concrete structures or even attaching a structural concrete element to an existing concrete structure. The load-
transfer mechanism of adhesive anchor is different than that of cast-in-place one. For adhesive anchor, the load is
transferred through the adhesive to the concrete along the entire embedded portion of the anchor. This load-transfer
depends on the strength of the adhesive-bar bond, the adhesive-concrete bond, and also on the extent to which the
adhesive impregnates the concrete surrounding the drilled hole (Cook et al. 1992). Although the use of these post-
installed adhesive anchors provides greater flexibility in design and strengthening of concrete members, their
behavior is less understood than that of the cast-in-place anchors.

The bond strength of post-installed FRP anchors depends on the mechanical properties of the FRP bars, the
adhesive, and the concrete. The high strength and low modulus of elasticity as well as the differences in the
properties of the fiber material and the matrix may lead to different bond characteristic from those of steel bars
(Wang et al. 1999). Additionally, the behavior of post-installed FRP anchors seems to be more complex because it is
dependent on the adhesive-bar interface, adhesive-concrete interface, as well as the surface and material properties.

To experimentally investigate the behavior of GFRP adhesive anchors, an extended experimental program was
planned. The program included two different diameters, 15.9 mm and 25.4 mm as well as two different types of
adhesive; epoxy based and cement based. The anchors were tested with different embedment lengths ranging from
5db to 15db; where db is the bar diameter. The details of the test program are shown in Table 1. This paper presents
the results of the completed portion of that program; series 1, group I. The FRP bars and the adhesives used in this
study were manufactured by Pultrall Inc. (2005) and Hilti Inc. (2005), respectively.

87
2. EXPERIMENTAL STUDY
2.1. Test Specimens

This research includes tension testing of 15 sand-coated GFRP V-ROD adhesive anchors of 25.4 mm diameter. The
used adhesive, Type HIT RE 500, is a high strength epoxy based adhesive specially designed for fastening into solid
base materials in a wide range of material temperatures ranging from 49oC down to -5oC. It may be also used in
underwater fastening for oversize holes up to two times the bar diameter but with a maximum of 76 mm hole
diameter (Hilti Inc. 2005). The specifications of this adhesive are shown in Table 2.

Table 1: Details of Test Specimens

Series Adhesive Type Diameter Embedded Length (mm) No. of

(mm) 5 db 10 db 15 db Samples
Group I Epoxy based 25.4 127 254 382 3×5=15
Series 1
Group II Cement based 25.4 127 254 382 3×5=15
Group I Epoxy based 15.9 79.5 159 238.5 3×5=15
Series 2
Group II Cement based 15.9 79.5 159 238.5 3×5=15

The test specimens included three different embedment lengths 5db, 10db, and 15db with five replicate samples each.
Table 2 gives the details of the test specimens. The FRP bars were cut to the desired length then steel tubes were
installed on the FRP bars (using epoxy grout) at one end keeping the other end free to be driven into the concrete.
Each bar is designated by a set of symbols and numbers to be uniquely identified. As an examples for G#25-5D-E4;
the first letter G denotes glass, #25 denotes the bar diameter in mm, 5D denoted the embedded length in bar
diameter multiplications, E4 denotes the fourth sample out of the five replicates with epoxy adhesive.

Table 2: Material Specifications (Hilti 2005)

Adhesive Compressive Tensile Modulus of Bond Strength Absorption Resistance

strength (MPa) strength (MPa) elasticity (MPa) (MPa) (%) (Ω/m)
HIT RE 500 82.7 43.5 7032 12.4 0.06 % 6.6×1013

Concrete slabs with dimensions of 3450×1750×400 were cast using ready-mixed normal weight concrete (type V,
MTQ). After casting, the concrete slabs were cured with water for 14 days and stored out the laboratory for 6 weeks
before installing the FRP anchors. Layout of the GFRP anchors was made in accordance with ASTEM E 488-96. To
install the FRP anchors, holes were drilled using rotary hammer pits. The holes were cleaned by wire brushes and
compressed water as this type of epoxy performs adequately in moisture. After cleaning the holes on the inside, the
two-component adhesive package was installed in the dispenser then injected into the holes. Consequently, the bars
were pushed into the holes in a screwing fashion. The installed specimens were stored out the laboratory for another
five months. During this period, the specimens were subjected to real environmental conditions as wet-dry cycles,
freeze-thaw cycles and temperature variation. The concrete cylinders were kept in the same conditions and its
compressive strength at the day of testing was 45 MPa.

2.2. Test Set-up and Procedure

A test set-up, in accordance with the requirements of the ASTM E 488-96, was used (Figure 1). The test bars were
pulled using a hydraulic jack connected to a manual pump. Each anchor was instrumented with one LVDT to
measure the bar displacement at different loading stages. Not to break it, the LVDT was removed at approximately
80% of the expected failure load. The loading pump and the LVDT were connected to a data acquisition system to
continuously record data up to the anchor failure.

2.3. Test Results and Discussion

The test results are presented in Table 3 while the failure modes are shown in Figure 2. For bars with small
embedment length (5db), the mode of failure was concrete failure (concrete cone or concrete damage followed by
bar pull-out). Additionally, in case of concrete failure, severe damage was observed in the concrete around the
anchors. However, for bars with greater embedment lengths (10db and 15db), the mode of failure was bar rupture.

88
From the test results, there was no significant difference between anchors with 10db and 15db embedment lengths in
terms of maximum load and mode of failure. As a result, 10db seems to be enough development length. Gesoglu et
al. (2005) conducted an experimental study on the post-installed steel adhesive anchors of 12 and 16 mm diameter
with embedment length ranged from 3.3db to 10db in normal-strength, high-strength and steel fiber-reinforced
concretes. In this study, the steel bar failure of the 16 mm diameter adhesive anchors was achieved at an embedment
length of 10db in normal-strength, high-strength, and steel-fiber reinforced concretes.

The Steel Tube

Hydraulic Jack

345 mm
Steel Plate
A

410 mm
2 I-Beams

310 mm
LVDT
400

The Concrete Slab

1750 mm A
Section A-A

Figure 1: Test Set-up

At the maximum load of each test, the corresponding bond stress was calculated using the following equation:

Fu
τ= (1)
π do le

where τ is the bond stress (MPa); Fu is the maximum load (N); do is the bar diameter (mm) and le is the embedded
length (mm). Comparing the obtained values listed in Table 3 to the listed one in Table 2, it can be noticed that all of
these values are lower than the epoxy adhesive bond capacity reported by the manufacturer. Gesoglu et al. (2005)
reported that the maximum bond stress for the 10db steel anchors was 11.9 MPa for anchors in normal-strength
concrete, 12.15 MPa in normal-strength steel fiber reinforced concrete, 12.75 MPa in high-strength plain concrete,
and 12.85 MPa in high-strength steel-fiber reinforced concrete.

Table 3: Test Results of GFRP Anchors Series 1, Group I

Load Le Stress Bond Stress Av. Bond SD COV Mode of

Designation
(kN) (mm) (MPa) (MPa) Stress (MPa) Failure
G#25-5D-E1 110 217.2 10.61 Concrete
G#25-5D-E2 107 127 211.3 10.32 Concrete
10.59 0.504 0.190
G#25-5D-E3 105 (5 db) 207.3 10.13 Concrete
G#25-5D-E4 117 231.0 11.28 Concrete
G#25-5D-E5 80 Affected by the nearby ones Concrete
G#25-10D-E1 250 493.6 12.34 Rupture
G#25-10D-E2 211 416.6 10.42 Rupture
254
G#25-10D-E3 227 448.2 11.21 10.85 0.941 0.708 Rupture
(10 db)
G#25-10D-E4 207 408.7 10.22 Tube
G#25-10D-E5 204 402.8 10.07 Rupture
G#25-15D-E1 206 406.8 6.62 Rupture
G#25-15D-E2 220 434.4 7.07 Rupture
382
G#25-15D-E3 231 456.1 7.43 7.07 0.483 0.187 Rupture
(15 db)
G#25-15D-E4 239 471.9 7.68 Rupture
G#25-15D-E5 205 404.8 6.59 Rupture

Although all GFRP anchors with embedded depth of 5db failed by concrete damage, a one piece cone was not
obtained. This may be due to the high brittleness of the concrete with relatively high compressive strength.

89
 a) b)

c) d)

Figure 2: Failure Modes the Test Specimens

3. CONCLUDING REMARKS
An experimental study on the behavior of GFRP post-installed adhesive anchors was conducted using sand-coated
GFRP V-ROD bars driven into plain concrete slabs using epoxy based adhesive with different embedment lengths
(5, 10, and 15 times the bar diameter). The installed specimens were left out the laboratory for five months. During
this period, the specimens were subjected to real environmental conditions as wet-dry cycles, freeze-thaw cycles and
temperature variation. Based on the test results the following conclusions can be drawn:

1. Although the anchor installation was performed according to the recommendations of ASTM E-488-96, the
corresponding ASTM spacing was not sufficient to prevent the overlapping of the concrete failure at smaller
embedded depths.
2. There is no significant difference between the anchors with 10db and 15db embedment length in terms of
capacity and mode of failure for all the tested specimens.
3. The 10db (254 mm) embedded length for the sand-coated GFRP V-ROD bars of 25.4 mm diameter seems to be
enough to provide ample strength of the GFRP bars.
4. The used epoxy adhesive functioned properly in wet and partially submerged conditions provided that the holes
are clean and free of loose sand or concrete particles.

4. REFERENCES
ASTM E 488-96. (1996). “Standard Test Methods for Strength of Anchors and Masonry Elements.” ASTM
International, West Conshohocken, PA, USA, pp. 65-72.
Cook, R.A., Collins, D.M., Klingner R.E., and Polyzois D. (1992). “Load-Deflection Behavior of Cast-in-Place and
Retrofit Concrete Anchors.” ACI Structural Journal, Vol. 89 No. 6, pp. 639-649.
Gesoglu, M., Özturan, T., Özel, M., and Güneyisi E.(2005).“Tensile Behavior of Post-Installed Anchors in Plain and
Steel Fiber-Reinforced Normal-and High-Strength Concretes.” ACI Structural Journal, Vol. 102 No. 2, pp 224-231.
Hilti Inc. (2005). Product Technical Guide, http://www.ca.hilti.com.
Pultrall Inc. (2005). Product Technical Specifications, http://www.Pultrall.com.
Wang, Z., Goto, Y., and Joh, O. (1999). “Bond Strength of Various Types of Fiber Reinforced Plastic Rods.” Fourth
International Symposium on Fiber Reinforced Polymer Reinforcement for Reinforced Concrete Structures, Editors:
C.W. Dolan, S.H. Rizkalla, and A. Nanni, ACI SP-188-93, pp. 1117-1130.

90
Third International Conference on FRP Composites in Civil Engineering (CICE 2006)
December 13-15 2006, Miami, Florida, USA

EFFECT OF ADHESIVE MODULUS ON THE MONOTONIC AND

FATIGUE BEHAVIOR OF EXTERNALLY BONDED CFRP STRIPS
Kent A. Harries
(Assistant Professor, Department of Civil and Environmental Engineering, University of Pittsburgh, USA)

Andrew Zorn
(Engineer, DMJM Harrie, Pittsburgh, PA, USA)

Benjamin Reeve
(Field Engineer, American Bridge International, Coraoplolis, PA, USA)

ABSTRACT
This study investigates the effect of adhesive modulus on the bond behavior of externally bonded fiber reinforced
polymer (FRP) composite materials used for repair of reinforced concrete elements under both monotonic and cyclic
fatigue loading conditions. Eighteen moderate-scale beam specimens having externally bonded carbon FRP (CFRP)
retrofits are tested. Different commercially available adhesive systems are used to bond the CFRP in otherwise
identical applications. The deleterious effects of fatigue loading on FRP-to-concrete bond are also identified.

KEYWORDS
Adhesive, Debonding, Fatigue, Flexure

1. INTRODUCTION
Oehlers (2005) presents a summary of five “generic” debonding mechanisms observed in FRP-concrete systems:
Plate End (PE); Critical Diagonal Crack (CDC); Flexural Intermediate Crack (FIC); Shear Intermediate Crack (SIC);
and, Axial Intermediate Crack (AIC). In reinforced concrete beams having relatively long shear spans and those
where the PE mode has been effectively mitigated, debonding of the FRP initiates at flexural (FIC) and/or
flexural/shear (SIC) cracks near the region of maximum moment. Under loading, these cracks open and induce high
interfacial shear stress accompanied by a small amount of peeling causing delamination, which propagates across
the shear span in the direction of decreasing moment. The following parameters have been identified as affecting
bond behavior. In many cases, opinions vary on the specific effect and considerably more study is required to
understand and quantify these parameters: a) concrete tensile strength; b) adhesive mechanical properties; c)
effective bond length of FRP; d) member scale and size; e) concrete section geometry; f) member loading geometry,
duration and rate; g) FRP retrofit geometry; and h) environmental and mechanical exposure. The present work
focuses on the effect of adhesive material properties and the nature of the applied loading: monotonic or fatigue.

1.1 Adhesive Properties

For the desired performance of a member to be realized, the FRP retrofit must be effectively bonded to the concrete
member’s tension face. Stiffer adhesives are viewed as being superior since they transmit stress across the FRP-
concrete interface in a more efficient manner. Stiffer adhesive systems however are associated with debonding
failure of the FRP at lower stress/strain levels and thus result in a less efficient use of the FRP material itself. A
flexible adhesive improves the performance of the retrofit by relieving stress/strain concentrations in the FRP while
adequately transmitting stresses between the FRP and concrete substrate. In general, reducing the shear stiffness of
the adhesive layer results in greater interfacial ductility and higher fracture energy. This study considers two
commercially available adhesive systems having shear moduli differing by about a factor of two.

2. EXPERIMENTAL PROGRAM
Eighteen reinforced concrete beams having externally bonded carbon fiber reinforced polymer (CFRP) strip flexural
retrofits are reported. The beams are 254 mm deep, 152 mm wide and are supported over a simple span of 4540 mm.

91
All beams have 3 - #4 primary reinforcing bars. The internal steel had yield and ultimate strengths of 429 MPa and
667 MPa, respectively. Twenty-eight day concrete compressive strength determined from cylinders was 23.3 MPa.

A commercially available 100 mm wide by 1.4 mm thick CFRP strip was used to retrofit sixteen of the beams (the
remaining two beams were unretrofit control (C) specimens). In all cases, the CFRP was extended to within 100 mm
of the beam supports. The CFRP has a reported tensile strength of 2792 MPa, modulus of 155 GPa and rupture
strain of 0.018. Eight beams were retrofit using commercially available low modulus “Adhesive L; the remaining
eight were retrofit using commercially available high modulus “Adhesive H”. Material properties of the adhesive
systems are given in Table 1. In all cases the adhesive bondline was approximately 1.6 mm thick.

Table 1: Manufacturer reported material properties of adhesive.

tensile strength tensile modulus elongation at shear strength bond strength

(ASTM D638) (ASTM D638) rupture (D638) (ASTM D732) ASTM (C882)
Adhesive L 14 MPa 2.2 GPa 0.063 21 MPa 18 MPa
Adhesive H 25 MPa 4.5 GPa 0.010 25 MPa 22 MPa

As shown in Figure 1, the CFRP strips were applied in widths of 25 mm, 51 mm and 102 mm. Four beams were
provided with each CFRP geometry: two each applied with the L and H adhesive system, respectively. Specimen
designation (given in inch units) is shown in Figure 1.

152 mm
2 - #3 (9.5 mm dia.)
L1 L2 L2x1 L4 L = low modulus Adhesive L
254 mm

C L1F L2F L2x1F L4F H = high modulus Adhesive H

CF H1 H2 H2x1 H4 F = fatigue loading
H1F H2F H2x1F H4F
4 - #4 (12.7 mm dia)

25 51 25 51 25 102 25 mm cover all around

Figure 1: Test specimen and CFRP retrofit geometry.

All beams were tested in three point flexure over a 4540 mm simple span. Applied load, midspan displacement and
coincident reinforcing bar and CFRP strains were recorded. One series of nine specimens was tested under
increasing monotonic load to failure (L and H specimens). The second series (designated LF and HF) was tested
under cyclic fatigue loading conditions. The applied load at midspan was cycled between 4.45 kN to 22.24 kN at
rate between 1.2 and 1.7 Hz (depending on the test). This fatigue stress level was selected to result in fatigue life of
the unretrofit control specimen CF of approximately N = 500,000 cycles. This fatigue regime resulted in fatigue-
induced reinforcing bar rupture in five of the nine fatigue specimens (CF, L1F, H1F, L2x1F and H2F). The
remaining specimens (L2F, H2x1F, L4F and H4F) were cycled to N = 2,000,000 and subsequently tested under
increasing monotonic load to failure. These latter specimens are considered to be “runout” specimens having
undergone 2 million cycles of fatigue conditioning. A complete description of the test programs reported here are
found in Reeve (2005) and Zorn (2006).

3. EXPERIMENTAL RESULTS
A summary of experimental results for all specimens is provided in Table 2. The ultimate deflection is defined as
that corresponding to a post-peak load of 80% of the peak load attained. Displacement ductility is defined as the
ratio of the ultimate displacement to that at general yield. For the specimens undergoing fatigue loads, data is given
for the first post-cracking cycle (N = 2) and the final recorded cycle (N = Nf). The reinforcing bar stress range is
determined from the recorded strain range and assumes Es = 200 GPa. The secant stiffness is the slope of the line
between the load and displacement at the lower (4.45 kN) and upper (22.24 kN) limits of the applied fatigue loading.

In all cases, the CFRP strain corresponding to debonding is determined by careful analysis of the CFRP and
reinforcing bar data as described by Reeve (2005). Debonding can only be identified as it propagates past at the
discrete locations of the strain gauges. For this reason, the strains reported should be interpreted as the greatest
CFRP strain observed while still bonded to the concrete, and thus are believed to represent a lower-bound on the
strain to initiate debonding.

92
 Table 2: Summary of Key Results.

increasing monotonic load to failure

Specimen H1 C L2L1 H2 L2x1 H2x1 L4 H4
maximum load (kN) 31.0
37.7 39.8
44.3 43.4 45.5 45.2 51.8 49.2
ultimate deflection (mm) 78.2
61.2 84.9
56.9 55.3 64.7 56.1 72.2 47.4
displacement ductility 2.61
1.83 2.48
1.70 1.60 1.89 1.70 1.91 1.33
CFRP strain at debond (µε) n.a.
2900 5300
6688 3550 7878 3200 4540 2850
fatigue loading (cycling from 4.45 to 22.24 kN)
Specimen CF1 L1F H1F L2F H2F L2x1F H2x1F L4F H4F
rebar stress range (MPa) 241 247 239 207 199 207 220 174 175
N=2
secant stiffness (kN/mm) 1.43 1.43 1.42 1.71 1.53 1.56 1.63 1.89 1.71
final cycle, Nf 329324 400892 424422 2M 1128006 447695 2M 2M 2M
rebar stress range (MPa) n.a. 231 250 240 217 230 251 201 196
N=Nf
secant stiffness (kN/mm) n.a. 1.34 1.33 1.35 1.37 1.42 1.38 1.56 1.48
increasing monotonic load to failure following N = 2000000
maximum load (kN) 45.5 45.3 51.3 53.8
fatigue-induced
ultimate deflection (mm) fatigue-induced 67.3 51.3 60.5 52.6
reinforcing bar
displacement ductility reinforcing bar rupture 2.05 1.45 1.67 1.3
rupture
CFRP strain at debond (µε) 4300 3909 3854 3259
1
due to a power failure, Specimen CF was loaded to failure at N = 329,324 prior to fatigue-induced reinforcing bar rupture.

Figure 3 summarizes the load-displacement relationship for the nine monotonic tests conducted and Figure 4
presents a representative example of the load-deflection response of a fatigue specimen (L4F) showing the
2,000,000 cycles of fatigue conditioning and the subsequent monotonic test to failure.
60 60

L4
50 50
H4 L4
L2x1 L4F
L2
40
applied load, kN

40
applied load, kN

H2x1 H2
H1 L1
30 30
C C
20 20
2 million fatigue cycles
shown: N = 48
N = 138,314
10 10 N = 502,791
L Specimens
N = 2,000,000
H Specimens
0 0
0 20 40 60 80 100 0 20 40 60 80 100
midspan deflection, mm midspan deflection, mm

Figure 3: Load deflection behavior of Figure 4: Representative load deflection

monotonically loaded specimens. behavior of Specimens L4 and L4F.

3.1 Performance of Monotonically Loaded Specimens

As is typically observed for externally bonded FRP retrofit applications, the application of the CFRP increases the
capacity of the beam although the incremental increase in strength decreases with increasing CFRP area. Similarly
the displacement ductility decreases with increasing CFRP area as the section reinforcing ratio increases.

As can be seen in Table 2, the load carrying capacity of the specimens tested is not affected by either the adhesive
type (L or H) or the 2,000,000 cycle fatigue conditioning. The ultimate displacement and thus displacement ductility
is improved with the softer adhesive system (L). Fatigue conditioning generally results in a reduction in the ultimate
displacement and thus ductility of the monotonically loaded beam.

3.2 Performance of Fatigue Loaded Specimens

In all cases, the stress range in the reinforcing steel increases with continued fatigue cycling. This increase is
proportional to the observed fatigue life (Nf) and is not affected by the amount of CFRP applied. Similarly, the
degradation of secant stiffness is affected by fatigue life. In most cases, the softer adhesive (L) appears to result in a

93
greater stress range and/or stiffness degradation. This is believed to result from the marginally less efficient stress
transfer affected by the softer adhesive requiring the internal reinforcing steel to carry proportionally greater stress.

4. DEBONDING OF EXTERNALLY BONDED CFRP

FIC debonding was the primary mode of failure for all monotonically loaded specimens. At the very least, local
debonding was evident in all fatigue specimens except L2F. In the specimens that experienced fatigue-induced
reinforcing bar rupture (L1F, H1F, H2F and L2x1), this debonding was caused by the energy release of the
reinforcing bar and remained localized near the rupture location. Minor debonding was visually identified in runout
specimens H2x1F, L4F and H4F. The debonding in these cases did not propagate beyond a strain gauge location and
thus no specific debonding strain may be identified. Nonetheless, the CFRP strains in the region immediate adjacent
the observed debonding ranged from only 1545 to 2285 µε.

4.1 Effect of Adhesive Stiffness

In FRP retrofit design, the mitigation of debonding is typically addressed through stress and/or strain limits applied
to the FRP material. Although, recommendations from various international standards documents vary considerably,
all are based on a function of the FRP axial modulus (Ef) and all neglect adhesive line properties. The results of this
study show the dramatic effect of adhesive modulus on the CFRP debonding strain. In all cases, the softer adhesive
(L) has a higher strain at initiation of debonding and may thus be interpreted as resulting in a more efficient use of
the bonded CFRP. This beneficial effect is most pronounced for thinner CFRP strips (L1 and L2x1, each having 25
mm strips).

4.2 Effect of Fatigue Conditioning

In all cases reported, the monotonic tests conducted following 2,000,000 cycles of fatigue conditioning experienced
lower strains at debonding than did their monotonically tested counterparts. The superior behavior of softer adhesive
(LF) is significantly less pronounced following fatigue conditioning.

5. CONCLUSIONS
The following primary observations and conclusions were drawn from this study:
1. Due to the long shear span, the primary failure mode of all monotonically loaded specimens was FIC
debonding. Some FIC debonding was evident during fatigue loading of all specimens except L2F.
2. As observed in most similar studies, the addition of CFRP material increased the load carrying capacity and
reduced the displacement capacity (ductility) of the beams. The incremental effect, in each case, is reduced with
increasing CFRP area.
3. The stress range in the internal reinforcing steel in the fatigue loaded specimens was observed to increase
proportionally with the number of cycles of fatigue loading. Similarly the secant stiffness of the fatigue load
cycle degrades with fatigue life. Both effects are marginally worse in the specimens having the lower modulus
adhesive (LF specimens).
4. CFRP strains corresponding to the initiation of debonding are significantly greater in the specimens having the
low modulus adhesive (L and LF specimens). The improved behavior is more pronounced in those specimens
having thinner CFRP strips (L1 and L2x1).
5. The effect of 2,000,000 cycles of fatigue conditioning appears to significantly reduce the previously described
beneficial effect on debonding strain of using a lower modulus adhesive.

The results of the present study indicate the need to address: a) the effect of the mechanical properties of the
adhesive system used; and b) the development of reduction factors for adhesively bonded FRP systems subject to
fatigue loading in developing and updating design guidelines for externally bonded FRP systems.

6. REFERENCES
Oehlers, D.J. (2005) “Generic Debonding Mechanisms in FRP Plated Beams and Slabs” Proceedings of the
International Symposium on Bond Behavior of FRP in Structures, December 7-9, 2005, Hong Kong. pp 353-44.
Reeve, B. (2005) “Effect of Adhesive Stiffness and CFRP Geometry on the Behavior of Externally Bonded CFRP
Retrofit Measures Subject to Monotonic Loads” M.Sc. Thesis, University of Pittsburgh. December 2005.
Zorn, A. (2006) “Effect of Adhesive Stiffness and CFRP Geometry on the Behavior of Externally Bonded CFRP
Retrofit Measures Subject to Fatigue Loads” M.Sc. Thesis, University of Pittsburgh. April 2006.

94
Third International Conference on FRP Composites in Civil Engineering (CICE 2006)
December 13-15 2006, Miami, Florida, USA

EXPERIMENTAL STUDY ON THE LOCAL BOND BEHAVIOR OF NSM-

FRP BARS TO CONCRETE
Donatella Galati
(PhD Candidate, Department of Innovation Engineering University of Lecce, Lecce, Italy)

Laura De Lorenzis
(Assistant Professor, Department of Innovation Engineering University of Lecce, Lecce, Italy)

ABSTRACT
Near-surface mounted (NSM) fiber-reinforced polymer (FRP) reinforcement has proven effective for strengthening
reinforced concrete (RC) structures and is gaining increasing attention. The NSM reinforcement is installed by
grooving the surface of the member to be strengthened and embedding FRP bars or strips in the grooves with an
appropriate binder. In this paper bond tests conducted on short NSM-bar anchorages are presented and results are
evaluated to determine the effect of the test variables on local bond-slip behavior and development capacity.
Parameters of the present study are the groove depth and width to depth ratio and the mechanical properties of the
groove-filling epoxy. During the tests, conducted with a direct shear setup using a specifically suited pulling
machine, the loaded-end and free-end slips and the strain distributions in the transverse plane were monitored. The
local bond-slip curves obtained from the tests are modelled analytically and used to predict the capacity of a
previous series of long bond length specimens characterized by the same other test variables.

KEYWORDS
Bond, fiber-reinforced polymers, near surface mounted reinforcement, pull-out testing, slip

1. INTRODUCTION
This work follows up to a previous experimental study on long bond length NSM bar anchorages (De Lorenzis and
Galati 2006), and reports test results on short NSM bar anchorages focusing on the same test variables, i.e. groove
depth and width to depth ratio and mechanical properties of the groove-filling epoxy. Beside gaining more test data
on the influence of the mentioned variables, the objectives of this investigation were to verify the consistency in
failure modes between short and long bond length specimens and to calibrate a local bond-slip relationship based on
average bond stress and slip measurements on the short bond length specimens, evaluating whether the use of such
local bond-slip relationship would allow an accurate prediction of the ultimate load of long bond length specimens.
These results would validate the use of local bond-slip laws from short bond length specimens for prediction of
quantities of design interest for strengthening with NSM reinforcement.

2. TEST PROGRAM AND MATERIALS

This paper presents results from a total of 24 NSM joints, see Table 1. The test variables were the groove size and
width to depth (w/d) ratio and the mechanical properties of the groove-filling epoxy. Two commercially available
epoxies, labeled as “a” and “b”, were selected for their significantly different values of tensile strength and modulus
of elasticity. Square grooves of two different sizes were evaluated, corresponding to ratios of groove depth d to
actual bar diameter db,a (indicated as k in Table 1) equal to 1.5 and 2.0. For a given groove depth, three different w/d
ratios were evaluated, equal to 0.75, 1.00 and 1.25. All joints had a bond length lb of 37.5 mm, corresponding to 5
times the nominal diameter of the bar db, and sufficiently short to consider bond stresses and slips uniformly
distributed along the bond length and hence to allow the evaluation of the local bond-slip behavior of the joints. The
specimen codes in Table 1 are structured as follows: two initial digits, corresponding to the groove width in mm, are
followed by two digits corresponding to the groove depth in mm; then the type of epoxy is reported (a or b), and
finally the letters A, B, C indicate different repetitions tested to control the experimental scatter.

95
The grooves were saw-cut onto the surface of the concrete blocks and then the bar was installed therein. The joints
were instrumented with two LVDTs, to monitor slip of the NSM bar with respect to the concrete at the loaded and at
the free end of the bond length. Strain gages were also applied to the outer surface of the epoxy, in the direction
perpendicular to the bar axis. Testing was conducted in displacement-control mode with a direct pull-out setup,
using a portable pulling machine suitable for in-situ testing.
The NSM joints tested in this program were constructed on two concrete blocks, having an average compressive
strength of 41.2 and 46.3 MPa and an average splitting tensile strength of 2.3 and 2.7 MPa, respectively. Type-a
epoxy had a direct tensile strength of 18.6 MPa (ASTM D 638M), a compressive strength of 46.3 MPa, an elastic
modulus of 4150 MPa (ASTM D 695M). Type-b epoxy had a direct tensile strength of 22.8 MPa, a compressive
strength of 83.4 MPa, an elastic modulus of 12870 MPa.
The round carbon FRP (CFRP) bars used in this study are spirally wound with a carbon fiber tow and sand coated,
having a nominal diameter db of 7.5 mm and an actual diameter db,a of 8.0 mm. The experimental tensile strength is
2214 MPa and the modulus of elasticity is 145.7 GPa. For more details see De Lorenzis and Galati (2006).

3. TEST RESULTS
The main test results are reported in Table 1 and Figures 1 and 2. The Table reports the ultimate load of the joints,
Pmax, the average bond strength at the bar-epoxy interface τavgp=Pmax/(πdblb), the loaded-end and free-end slips at
peak load, slep and sfep, and the failure mode.
For all specimens with type-a epoxy, the primary bond failure mechanism was slip at the bar-epoxy interface. Post-
failure observations revealed that a very thin layer of epoxy was still visible on the surface of the bar, indicating that
failure had resulted from cohesive shear failure in the epoxy. For specimens 12-12/a longitudinal splitting cracks in
the epoxy paste and inclined cracks in the concrete adjacent to the groove became visible after the peak load, and
widened during the descending branch of the load - slip curves. For joints 12-16/a, bond failure occurred by a
combination of slip at the bar-epoxy interface and inclined cracking of the concrete adjacent to the groove. Joints
16-16/a and 20-16/a failed at the bar-epoxy interface with no visible cracks in the concrete or in the epoxy. Type-b
epoxy has a tensile strength 23% larger and a modulus of elasticity 210% larger than type-a epoxy. This difference
in the epoxy properties caused the average bond strength to increase significantly (Table 1) and in some cases
(specimens 12-12 and 16-16) led to a different failure mode. Joints 12-12/b failed abruptly by formation of inclined
cracks in the concrete adjacent to the groove before the shear strength of the epoxy was reached. Slip at the bar-
epoxy interface occurred only after the peak load under decreasing bond stresses. For joints 12-16/b and 16-16/b,
failure occurred by a combination of slip at the bar-epoxy interface and inclined cracking of the concrete. Only
joints 20-16/b failed at the bar-epoxy interface with no visible cracks in the concrete.
Figure 1 illustrates the curves giving the average bond stress at the bar-epoxy interface versus loaded- and free-end
slips. Only one representative repetition per type of specimen was selected for the sake of clarity.
For specimens 12-12/b and 12-16/b the gain in strength compared with the same specimens with type-a epoxy was
accompanied by a more abrupt decrease of the bond stress after failure. For all the other joints the average bond-slip
behaviour is characterized by a stiff initial ascending branch, followed by a rather smooth softening branch that
indicates a pseudo-ductile nature of bond failure at the bar-epoxy interface. Besides, for all specimens the average
bond stresses in the post-peak phase increase as the groove dimensions (i.e. both k and w/d) increase, due to the
higher degree of confinement of the bar-epoxy interface. Hence increasing the groove dimensions, while not
significantly improving the bond strength, yields a significant increase of the fracture energy of the joint, and this
implies that the bond failure load of long bond length specimens will also increase.
Figure 2 shows the transverse strains measured on the epoxy cover at two different locations versus the applied load
for selected specimens (strain 1 and 2 are located at a distance of 12.5 mm and 25 mm from the loaded end,
respectively). Tensile strains are taken as positive and the curves are limited to the pre-failure phase. All curves
display the same general trend. As illustrated in Figure 2, the use of a larger w/d ratio appears to delay the phase
where tensile strains are developed in the cover and to reduce the maximum transverse strains at the peak load,
indicating that the larger groove dimensions increase the transverse stiffness of the cover of the NSM bar. It is also
evident that using type-b epoxy causes a marked reduction in the slope of the strain-load curves, and a substantial
reduction of the transverse tensile stresses at peak load. Similar conclusions can be drawn on the effect of k ratio.

4. ANALYTICAL MODELING
As mentioned earlier, the local bond stress–slip relationship is here approximated by the average bond stress-slip
relationship of the short bond length specimens. For all specimens the ascending branch of the bond–slip law is well
interpreted by the Bertero–Popov–Eligehausen (BPE) relationship, that is τ=τm(s/sm)α valid for 0≤s≤sm, where τ is
the local bond stress, s the local slip, τm and sm are bond stress and slip at the peak point, and α is a parameter that
varies between 0 and 1. The descending branch of the bond-slip law can be described by an equation formally

96
 identical with the exception that a parameter α’ varying between -1 and 0 is used in place of α. Parameters τm, sm, α
and α’, calibrated by best fitting of the experimental results, are reported in Table 2, where τm is taken equal to the
average bond strength and sm is evaluated as the average between loaded-end and free-end slips at the peak point. As
previously noted, the local bond strength is weakly affected by the groove dimensions, whereas a dramatic increase
in the joint fracture energy Gf (here evaluated at a slip of 3 mm) is observed.

Table 1: Test program and results

Specimen w/d k = d/ Measured Measured Pmax τavgp slep sfep

Failure mode
db,a w/d k (kN) (MPa) (mm) (mm)
12-12/a/A 1.00 1.50 1.05 1.47 11.5 13.02 0.31 0.13 BE-C+LC +SP
12-12/a /B 1.00 1.50 1.06 1.50 10.9 12.34 0.23 0.10 BE-C+LC + SP
12-12/a /C 1.00 1.50 1.03 1.45 11.4 12.90 0.20 0.05 BE-C+LC +SP
12-12/a/AVERAGE 11.3 12.75 0.25 0.09 BE-C+LC +SP
16-16/a/A 1.00 2.00 0.93 2.00 11.7 13.24 0.35 0.23 BE-C
16-16/a/B 1.00 2.00 1.07 1.95 12.2 13.81 N/A 0.32 BE-C
16-16/a/C 1.00 2.00 0.97 2.04 14.0 15.84 0.30 0.16 BE-C
16-16/a/AVERAGE 12.6 14.30 0.32* 0.24 BE-C
12-16/a/A 0.75 2.00 0.78 1.98 11.1 12.56 N/A 0.37 BE-C+ SP
12-16/a/B 0.75 2.00 0.70 2.04 14.1 15.96 0.52 0.34 BE-C+ SP
12-16/a/C 0.75 2.00 0.78 1.92 11.1 12.56 N/A 0.29 BE-C+ SP
12-16/a/AVERAGE 12.1 13.69 0.52* 0.33 BE-C+ SP
20-16/a/A 1.25 2.00 1.20 2.00 12.5 14.15 0.53 0.29 BE-C
20-16/a/B 1.25 2.00 1.21 2.02 12.3 13.92 0.42 0.35 BE-C
20-16/a/C 1.25 2.00 1.34 1.96 11.1 12.56 N/A 0.40 BE-C
20-16/a/AVERAGE 12.0 13.54 0.47* 0.35 BE-C
12-12/b/A 1.00 1.50 1.01 1.52 21.5 24.33 N/A 0.23 SP
12-12/b/B 1.00 1.50 0.96 1.47 20.4 23.09 N/A 0.17 SP
12-12/b/C 1.00 1.50 1.00 1.45 17.3 19.58 0.29 0.17 SP
12-12/b/AVERAGE 19.7 22.33 0.29* 0.19 SP
16-16/b/A 1.00 2.00 1.01 1.97 24.3 27.50 N/A 0.18 BE-C+ SP
16-16/b/B 1.00 2.00 1.03 1.96 24.8 28.07 0.16 0.08 BE-C+ SP
16-16/b/C 1.00 2.00 1.08 1.94 23.0 26.03 0.20 0.15 BE-C+ SP
16-16/b/AVERAGE 24.0 27.20 0.18* 0.14 BE-C+ SP
12-16/b/A 0.75 2.00 0.82 1.91 17.3 19.58 0.33 0.22 BE-C+ SP
12-16/b/B 0.75 2.00 0.74 1.95 22.1 25.01 0.16 0.14 BE-C+ SP
12-16/b/C 0.75 2.00 0.77 1.96 18.0 20.37 0.24 0.19 BE-C+ SP
12-16/b/AVERAGE 19.1 21.65 0.24 0.19 BE-C+ SP
20-16/b/A 1.25 2.00 1.10 2.14 23.5 26.60 0.41 0.30 BE-C
20-16/b/B 1.25 2.00 1.18 2.07 21.3 24.11 N/A 0.22 BE-C
20-16/b/C 1.25 2.00 1.21 2.08 23.7 26.82 0.54 0.31 BE-C
20-16/b/AVERAGE 22.8 25.84 0.47* 0.28 BE-C
Note: BE-C = failure at the bar-epoxy interface - cohesive shear failure in the epoxy; LC = longitudinal cracking of the epoxy;
SP = splitting failure by fracture in the concrete along inclined planes; N/A = not available; *average of available data.

30 30
Average bond stress (MPa)

12-12/a/A loaded end 12-12/b/A loaded end

Average bond stress (MPa)

25 12-12/a/A free end 25 12-12/b/C free end

12-16/a/C loaded end 12-16/b/C loaded end
12-16/a/C free end 20 12-16/b/C free end
20 16-16/a/C loaded end 16-16/b/C loaded end
16-16/a/C free end 16-16/b/C free end
15 20-16/a/B loaded end 15 20-16/b/B loaded end
20-16/a/B free end 20-16/b/B free end
10
10
5
5
0
0
0 1 2 3 4 5
0 1 2 3 4 5 Slip (mm)
Slip (mm)

Figure 1: Average bond stress-slip curves

97
From the calibrated local bond–slip relationship, Figure 3 shows the predicted bond failure load as a function of the
bond length. For the tested NSM systems, anchorage and economic efficiency suggest the use of 16x16 mm
grooves, i.e. square grooves with a k ratio of 2. This yields a development length of about 55 db and 30 db when
using type-a and type-b epoxy, respectively. Figure 3 also reports the experimental data obtained by De Lorenzis
and Galati (2006) for specimens with 24-diameter bond length, which appear in reasonably good agreement with
theoretical predictions. Note that curves of specimens 12-12 and 12-16 and those of specimens 16-16 and 20-16 are
rather close to each other, in agreement with the values of parameters α and α’ reported in Table 2.
5000 5000
12-16/a/A strain 1 12-16/b/A strain 1
4000 12-16/a/A strain 2 4000 12-16/b/A strain 2

Transverse strain (με)

16-16/a/A strain 1
Transverse strain (με)

16-16/b/A strain 1
16-16/a/A strain 2 3000 16-16/b/A strain 2
3000
20-16/a/A strain 1 20-16/b/A strain 1
20-16/a/A strain 2
2000 2000 20-16/b/A strain 2

1000 1000

0 0
0 10 20 30 0 10 20 30

-1000 -1000
Load (kN) Load (kN)

Figure 2: Transverse strain-load curves

Table 2: Calibrated parameters in the local bond– slip relationship

Type-a epoxy Type-b epoxy

Specimen τm sm α α’ Gf τm sm α α’ Gf
(MPa) (mm) (N/mm) (MPa) (mm) (N/mm)
12-12 12.8 0.17 0.45 -0.54 19.56 22.3 0.21 0.19 -0.82 20.35
12-16 13.7 0.43 0.18 -0.63 22.69 21.7 0.22 0.45 -0.65 23.28
16-16 14.3 0.28 0.21 -0.18 30.67 27.2 0.16 0.34 -0.26 51.67
20-16 13.5 0.41 0.21 -0.18 32.27 25.8 0.33 0.38 -0.16 60.30

5. CONCLUSIONS
The following main conclusions can be drawn: when using type-a epoxy, failure is controlled by slip at the bar-
epoxy interface, due to cohesive shear failure in the epoxy; if type-b epoxy is used, failure may occur by concrete
cracking for small groove sizes; the failure modes of short bond length specimens are consistent with those of long
bond length specimens (De Lorenzis and Galati 2006) which validates the analytical approach used in this study,
consisting in calibrating the local bond-slip law based on short bond length data; increasing the groove dimensions
does not heavily affect the local bond strength, but significantly enhances the fracture energy of the joint; as a
consequence, the bond failure load of long bond length joints increases; for the tested NSM systems, anchorage and
economic efficiency suggest the use of 16x16 mm grooves, i.e. square grooves with a k ratio of 2.
The approach used in this study is suggested as a protocol for characterization of different NSM systems. While a
general formulation of a local bond-slip law appears unfeasible, calibration of a local bond-slip law based on short
bond length testing appears a simple yet reliable procedure to predict the anchorage length of a given NSM system.
100 100

80 80
Theor. 12-12/a
Ultimate Load (kN)

Theor. 12-12/b
Ultimate load (kN)

Theor. 12-16/a Theor. 12-16/b

Theor. 16-16/a Theor. 20-16/b
60 Theor. 20-16/a 60
Exp. 12-12/a Theor. 16-16/b
Exp. 12-16/a Exp. 12-12/b
Exp. 16-16/a Exp. 12-16/b
40 40 Exp. 16-16/b
Exp. 20-16/a Exp. 20-16/b

20 20

0 0
0 20 40 60 80 100 120 140 160 180 200 220 0 20 40 60 80 100 120 140 160 180 200 220
Bond length (N. of db) Bond length (N. of db)

Figure 3: Theoretical and experimental bond failure load as a function of the bond length

REFERENCES
De Lorenzis, L., and Galati D. (2006), “Effect of Construction Details on the Bond Performance of NSM FRP Bars
in Concrete”, Proceedings FIB Congress 2006, Naples, Italy, CD-ROM.

98
Part II. Bridge Applications
Third International Conference on FRP Composites in Civil Engineering (CICE 2006)
December 13-15 2006, Miami, Florida, USA

A Case Study of Life Cycle Cost based on a Real FRP Bridge

Itaru Nishizaki
(Team Leader, Public Works Research Institute, Tsukuba, Ibaraki, Japan)

Nobufumi Takeda
(Group Leader, Obayashi Corporation, Kiyose, Tokyo, Japan)

Yoshio Ishizuka
(Group Leader, Shimizu Corporation, Koto-ku, Tokyo, Japan)

Takumi Shimomura
(Associate Professor, Nagaoka University of Technology, Nagaoka, Niigata, Japan)

ABSTRACT
In this paper, the authors considered and calculated Life Cycle Cost (LCC) of FRP and Pre-Stressed Concrete (PC)
footbridges based on an actual FRP footbridge constructed in Japan. Three types of PC bridges and two types of
FRP bridges were set as model cases. The result suggests that FRP bridges have the competitive edge in spite of
their initial cost and are more efficient when durability is required in severely corrosive environments.

KEYWORDS
FRP, PC, footbridge, life cycle cost

1. INTRODUCTION

FRP has some excellent properties as a structural material. Its application to bridges offers a possibility to solve
problems that bridges made of conventional materials are facing today such as corrosion and damages incurred early
in the life-cycle of a structure. Presently, FRP’s unit price is usually rather more expensive than that of other
conventional materials. This may increase the initial cost of the FRP superstructure and is one of the obstacles
deterring widespread use of the material in FRP bridges.
In order to evaluate the benefit of using FRP in bridges, it is important to consider FRP’s life cycle cost (LCC)
including the cost for maintenance. There has been some research1)-3) on the cost benefit of FRP structures; however,
because some of those studies begin with the design of the structures and include many suppositions, the LCC
estimates of FRP structures are not so reliable.
With this in mind, the authors tried to evaluate the LCC of an actual FRP footbridge, remaining as faithful to actual
conditions as possible. The case study is based on an FRP footbridge constructed in Okinawa, Japan, in 2000. It is
called “Okinawa Road Park Bridge” and is pictured in Figure 1.

Figure 1: View of the Okinawa Road Park Bridge

99
2. THE STRUCTURES FOR THE CASE STUDY

A FRP footbridge and PC footbridge crossing a 4-lane road were considered as the case models. The bridges are
located close to the seashore and severely affected by sea salt. The main girders of the FRP footbridge are made of
hand lay-up FRP; pultruded FRP is used for the stiffeners, decks, and floor systems. Both types of FRP were made
of glass fiber and vinylester resin. Parts of the FRP footbridge were made in several factories within the Tokyo area,
assembled in a factory in Tokyo Bay, and then shipped to Okinawa. Wall type piers and steel pipe pile foundations
were used in the substructure for both bridges.

Table 1: Model cases of FRP and PC bridges

FRP bridges PC bridges

Two span girder bridge Single span deck girder bridge with hollow
Concept
with GFRP C-girders post-tension concrete beams
Length 37.8m 36.0m
Span 19.7m+17.2m 35.0m
Width 4.3m 4.3m
350kgf/m2 for main girders
Live load
500kgf/m2 for decks

3. CALCULATION METHOD OF LCC

Direct construction costs of the initial cost and the maintenance cost for both FRP and PC bridges were calculated
based on the design reports for both bridges. LCC was obtained by the equations:

LFRP bri. = IFRP bri. + MFRP bri.

LPC bri. = IPC bri.. + MPC bri.
L: Life-cycle cost
I: Initial cost
M: Maintenance cost

We did not calculate the cost for disuse neither did we consider the discount rate to discount future costs to the base
year. Initial costs were calculated for both the superstructure and substructure. Maintenance costs were calculated
only for the superstructure. The authors tried to set realistic suppositions in situations where no data existed. In this
study, the authors made some assumptions for unknown conditions and simplified the calculation. Hence the values
of the costs in this study do not indicate the real values of the Okinawa Road Park Bridge itself.

4. RESULTS

4.1 Initial costs

4.1.1 PC footbridges

Five types of superstructure were roughly designed for the PC footbridges. A deck girder footbridge with hollow
post-tension concrete beams was selected after considering multiple viewpoints, including economy, workability,
structure, view, and maintenance. Table 2 shows the model cases of the PC footbridge. CASE-1 is the base case with
two types of corrosive protected cases added. CASE-2 adopts epoxy resin coated reinforcing bar and PC tendon.
CASE-3 also adopts coated bar and tendon, with the addition of a paint coating on the concrete surface. The
calculated the initial cost of each superstructure is: 48,240,000JPY, 50,620,000JPY and 54,370,000JPY respectively.
As regards the substructure, two piers (Pier 1 and Pier 2) were roughly designed for each of three alternatives. The
best results are shown in Table 2. The total cost of the substructure was 10,130,000JPY.

4.1.2 FRP footbridges

100
The initial cost of FRP bridges is roughly divided into three categories: (1) materials, (2) assembly, and (3) mold for
hand lay-up. Table 3 shows the initial cost of FRP bridges. The initial cost of the FRP superstructure was
73,600,000JPY. The base model case (CASE-4) of the FRP footbridge has some special points, for example, it is the
first real FRP footbridge in Japan and it is located on the seashore, suggesting that it may be possible to reduce its
initial cost. The authors considered a modified case (CASE-5) for FRP bridges so as to reduce its initial cost. These
modifications were: (1) change of handrail to aluminum, (2) change of design in the joint part of main girders, and
(3) sharing of mold by two bridges. The result of the modified initial cost became 62,350,000JPY.

Table 2: Model cases of PC bridges and initial costs

（Unit: 1000JPY）
CASE-1 CASE-2 CASE-3
Coated reinforcing bars
Corrosion protection for the Coated reinforcing bars
None Coated PC tendon
superstructure Coated PC tendon
Surface coating
Initial cost for the superstructure 48,240 50,620 54,370
Pier 1: 6 Steel pipe piles (φ600mm-9mm, L=17.5m)
Substructure system
Pier 2: 4 Steel pipe piles (φ600mm-12mm, L=20.0m)
Initial cost for the substructure 10,130
Total Initial costs 58,370 60,750 64,500

Table 3: Model cases of FRP bridges and initial costs

（Unit: 1000JPY）
CASE-4 CASE-5
(1) Aluminum handrail
Modified points for the Standard FRP bridge based on
(2) Change of joint parts of the main girders
superstructure the real bridge
(3) Sharing the mold by 2 bridges
Initial cost for the superstructure 73,600 62,350
Pier 1: 2 Steel pipe piles (φ500mm-9mm, L=15.0m)
Substructure system Pier 2: 4 Steel pipe piles (φ500mm-9mm, L=18.0m)
Pier 3: 2 Steel pipe piles (φ500mm-9mm, L=12.0m)
Initial cost for the substructure 6,910
Total Initial costs 80,510 69,260

There are three piers (Pier 1, Pier 2, and Pier 3) for the substructure of the FRP footbridge. When comparing the two
pile systems, driven steel pipe piles and PHC (Pretensioned Spun High Strength Concrete) piles with installation by
inner excavation, the steel pipe piles substructure showed better results in this case.
Comparing the total costs including both the superstructure and substructure, the difference of the initial cost of the
modified FRP footbridge (69,260,000JPY) was only 10% higher than the initial cost of the corrosion protected PC
footbridge. This result suggests FRP bridges have significant competitive power even when considering the initial
cost.

4.2 Maintenance costs

4.2.1 PC footbridge

Inspection and repair are the main maintenance considerations for bridges. Only the costs for repair were considered
in this study. The costs for inspection were omitted because it seems there are not large differences in the inspection
of PC and FRP bridges.
For the PC bridges, the authors estimated the penetration of chloride ion into the concrete after the construction, and
the repair was set when the concentration of chloride ion at steel reinforcing bars reached 1.2 kg/m3. Replacement of
covering concrete and surface coating was selected as the repair method for the PC bridges. The life of the surface
coating which protects against chloride ion penetration was set at 15 years and 30 years, and repair of the surface
coating was calculated in these intervals. Table 4 shows the results of the repair costs.

4.2.2 FRP footbridge

Since the Okinawa Road Park Bridge is relatively new, there is not enough information on its repair and
maintenance requirements. However, five years after its construction, stainless steel bolts were replaced because of
corrosion due to the severely corrosive environment. This amounted to 1,000,000JPY. We therefore considered the
same scale of repair may be required at the same interval within a severely corrosive environment and set the repair

101
cost for an FRP footbridge at 1,000,000JPY at 5-year intervals. In the modified case of FRP footbridges, the repair
cost was also modified by adopting highly durable bolts. The cost is 3,500,000JPY and the repair interval was set at
40 to 50 years.
Repainting is the major repair concern for FRP footbridges. There will be no corrosion for FRP structures caused by
weak points of painting such as edges or bolt parts like a painted steel structure because FRP does not corrode.
Hence, we set the repainting interval based on the decrease of thickness caused by the deterioration of the painting
material. The repainting interval was set at about 120 years based on the thickness (75 μm) and the material
(fluorine resin paint) of the paint. The repainting cost was calculated and the result was 5,600,000JPY including the
scaffolding for repainting.

4.3 LCC

Table 4 shows the results of initial cost, maintenance cost and LCC for both PC and FRP footbridges. At 50 years,
LCC of the FRP footbridge was 90,510,000JPY; this is lower than the 50-year LCC of the PC footbridge without
corrosion protection. The lowest 50-year LCC was that of the PC footbridge with epoxy resin coated reinforcing bar
and PC tendon (CASE-2). However, the modified FRP footbridge (CASE-5) showed the lowest 100-year LCC
among our five cases. These results suggest that FRP footbridges are more efficient when longer life is required in
severely corrosive environments.

Table 4: LCC results of both PC and FRP footbridges

（Unit: 1000JPY）

CASE-3
CASE-1 CASE-2 CASE-4 CASE-5
Repair interval: Repair interval:
15 years 30 years
Initial cost for superstructures 48,240 50,620 54,370 73,600 62,350
Initial cost for substructures 10,130 10,130 10,130 6,910 6,910
Total the initial costs 58,370 60,750 64,500 80,510 69,260
Maintenance cost for 30 years 24,500 0 18,000 9,000 6,000 3,500
Maintenance cost for 50 years 42,500 0 27,000 9,000 10,000 3,500
Maintenance cost for 100 years 69,500 24,500 54,000 27,000 20,000 7,000
50 years LCC 100,870 60,750 91,500 73,500 90,510 72,760
100 years LCC 127,870 85,250 118,500 91,500 100,510 76,260

5. CONCLUSION

In this paper, the authors considered and calculated the LCC of FRP and PC footbridges, faithfully considering
actual conditions based on a real FRP footbridge constructed in Japan. The result suggests that FRP bridges has a
competitive edge over other types of construction in spite of its initial cost and that FRP footbridges are more
efficient when longer life is required in severely corrosive environments.

6. ACKNOWLEDGEMENTS

This study was carried out as a part of the activity of “The committee for applications of advanced structural
materials” of the Japan Society of Civil Engineering (JSCE) from 2004 to 2006. The authors also thank the
government of Okinawa prefecture for their cooperation in providing the information on the plans and designs of the
FRP and PC footbridges.

7. REFERENCES

1) Meiarashi, S., Nishizaki, I. and Kishima, T. (2002). “Life-Cycle Cost of All-Composite Suspension Bridge.”
Journal of Composites for Construction, 6 (4), 206-214.
2) Ehlen, M. A. (1999). “Life-cycle costs of fiber-reinforced-polymer bridge decks.” J. Mater. Civ. Eng., 11(4), 224-
230.
3) Nystrom, H. E., Watkins, S. E., Nanni, A., and Murray, S. (2003). “Financial viability of fiber reinforced polymer
(FRP) bridges.” Journal of Management in Engineering, 19(1), 2-8.

102
Third International Conference on FRP Composites in Civil Engineering (CICE 2006)
December 13-15 2006, Miami, Florida, USA

STRUCTURAL REHABILITATION OF AN OFF-SYSTEM BRIDGE

USING EXTERNALLY BONDEND CFRP LAMINATES
Nestore Galati
(Research Engineer, University of Missouri - Rolla, Rolla, Missouri, USA)

Andrea Rizzo
(PhD Candidate, University of Lecce, Lecce, Italy)

Antonio Nanni
(Professor and Chair, University of Miami, FL, USA)

ABSTRACT
This paper presents the use of Fiber Reinforced Polymers (FRP) laminates for the flexural strengthening of a
concrete bridge superstructure. The system consists of Carbon Fiber Reinforced Polymers (FRP) laminates bonded
onto the concrete surface in order to provide the necessary flexural reinforcement for the deck. The bridge
superstructure selected for this project is a 4-span reinforced concrete (RC) slab. The bridge is located on Route
0039 in Washington County, Missouri - USA. The bridge analysis was performed for maximum loads determined in
accordance to AASHTO Design specification, 17th edition. The strengthening scheme was designed in compliance
with the ACI 440.2R-02 design guide.

KEYWORDS
Mechanically Fastened Fiber Reinforced Polymers (MF-FRP) Laminates, Load Rating, Finite Elements Modeling

1. INTRODUCTION

Fiber-reinforced polymer (FRP) materials have emerged as a practical alternative for construction, renovation and
strengthening of bridges with significant cost and time savings over conventional methods. Advantages of FRP
materials are that they resist corrosion, long outlive conventional materials, and have high strength-to-weight ratio.
It has been shown that in this technical area nowadays the engineer has different tools available in order to find the
optimal solution to each problem: manual lay-up FRP and SRP laminates, adhered pre-cured FRP laminates, near
surface mounted (NSM) FRP bars and, finally, MF-FRP laminates (Lopez et al. 2004).
This paper reports the use of Fiber Reinforced Polymers (FRP) laminates for the flexural strengthening of a concrete
bridge superstructure. The system consists of Carbon FRP laminates produced by MAPEI Corporation bonded onto
the concrete surface in order to provide the necessary flexural reinforcement to the deck. The bridge superstructure
selected for this project is a 4-span reinforced concrete (RC) slab. The bridge is located on Route 0039 in
Washington County, MO. The bridge analysis was performed for maximum loads determined in accordance to
AASHTO Design specification, 17th edition. The strengthening scheme was designed in compliance with the ACI
440.2R-02 design guide.

2. DESCRIPTION OF THE BRIDGE

Bridge No. 0390006 (see Figure 1), located in Washington County (Route 0039), MO. The bridge is actually load
posted to a maximum weight of 4.5 tonSI (5 ton). The total length of the bridge is 17.8 m (58 ft) and the total width
of the deck is about 4.3 m (14 ft). The structure is a 4-span continuous reinforced concrete (RC) slab having 178 mm
(7 in) thickness. From visual observations, the guardrail along the longitudinal North edge was found damaged with
concrete spalling and exposed steel reinforcement. No visible traces of steel rebar corrosion were observed. Even
though the concrete substrate was found sound, as a consequence of the insufficient amount of longitudinal

103
reinforcement, the deck was visibly cracked at mid-span and corresponding to the piers. Piers and abutments
appeared in good conditions.
The geometry of the bridge is summarized in Table 1. Figure 2 and Figure 3 show the longitudinal and plan view of
the bridge. Figure 3 also shows the position from where the concrete cores where extracted, and the longitudinal and
transverse steel reinforcement of the deck.

Figure 1: Bridge No. 0390006

The details of the bridge reinforcement and material properties were unknown at the time of strengthening due to the
unavailability of plans. As a consequence, at the onset of the project, these properties were determinate in-situ, based
on visual and Non Destructive Testing (NDT) evaluation. Three concrete cores were drilled and tested in
compliance with ASTM C39 and ASTM C42. The average concrete compressive strength f`c was found to be 20.0
MPa (2900 psi). The location of the steel reinforcement was accurately detected with a rebar locator. The steel
mechanical properties were determinate by testing three specimens cut from exposed bars according to ASTM A615
and ASTM A955. The average yield steel strength, fy, was found to be 512 MPa (74.2 ksi). More details can be
found in Rizzo et al. (2005).

Table 1: Geometry of the Bridge

Span D1 D2 D3 D4
Clear Span, lc 4147 mm 4178 mm 4451 mm 4204 mm
(in the Traffic Direction) (13 ft 7 1 4 in ) (13 ft 8 1 2 in ) (14 ft 7 1 4 in ) (13 ft 9 1 2 in )
Design Length, ld 4350 mm 4382 mm 4655 mm 4407 mm
(in the Traffic Direction) (14 ft 3 1 4 in ) (14 ft 4 1 2 in ) (15 ft 3 1 4 in ) (14 ft 5 1 2 in )
Deck Height, H d 190 mm 184 mm 178 mm 190 mm
(Average Value) ( 7 1 2 in ) ( 7 1 4 in ) ( 7 in ) ( 7 1 2 in )
Skew, α 48o (between Support and Traffic Direction)
Roadway Width, Wr 4267 mm (14 ft )
Curb-to-Curb Roadway
3988 mm (13 ft 1 in )
Width, Wrc
Overlay Height, H o 12.7 mm ( 0.5 in )

721" Deck D1 4' 721" Deck D2 614" Deck D3 7" Deck D4 621" 71"
EAST 2 WEST

13'-714" 13'-812" 14'-714" 13'-912"

8" 8" 8" 8" 8"
S1 S2 S3 S4 S5
1 in = 1” = 25.4 mm
Supports 1 ft = 1’ = 304.8 mm

Figure 2: Longitudinal View of the Bridge

104
 Guardrail Steel Bars Layout
CC #1 8" 8" on the Deck 8" 585"
CC #2
Deck D1 Deck D2 Deck D3 Deck D4
14'-234" 14'-478" 15'-278" 14'-558"

EAST 3'-11" 5'-7" 4'-7" 1'-218" 1'-621" WEST 13'-1" 14'-01"

4
Guardrail

47°
S1 S2 CC #3 S3 S4 S5
Visible Transverse Exposed Bars with 558"
Supports
Cracks Highly Spalled Concrete
Average Values: #5 @ 12 '' (Damaged Guardrail Length)
CC = Concrete Core
Cover 1 5 8 '' (Positive Reinforcement)
Cover 4 1 2 '' (Negative Reinforcement)

Figure 3: Plan View of the Bridge

3. BRIDGE ANALYSIS AND FRP STRENGTHENING DESIGN

The analysis of the bridge was performed according to the MoDOT Bridge Manual, 1996: the assumed load
configurations were consistent with the AASHTO Specifications (AASHTO 2002). The structural analysis of the
bridges was performed using design truck and lane loads having geometrical characteristics and weight properties as
suggested in AASHTO, 2002 Article 3.7.4. An H15-44 truck load was considered as design load since it generated
the maximum load configuration compatible with the shear capacity of the structure: in fact, for “slab” bridges the
FRP strengthening can only be used to increase the flexural capacity.
Since the FRP strengthening system does not allow increasing the flexural capacity of the deck in the negative
moment region, the analysis was conservatively conducted by neglecting the flexural continuity of the deck over the
supports. This led to model the deck as a simply-supported slab between two consecutive supports.
FRP laminate design was carried out according to the principles of ACI 440.2R-02 (ACI 440 in the following). The
properties of concrete, steel and FRP laminates used in the design are summarized in Table 2. The concrete and steel
properties were obtained by testing of samples. The guaranteed design material properties of the FRP laminates were
determined experimentally as part of a companion experimental program (Rocca et al., 2006).
The φ factors used to convert nominal values to design capacities are obtained as specified in AASHTO (2002) for
the as-built and from ACI 440 for the strengthened members. Figure 4 details the longitudinal flexural
strengthening, while Figure 5 details the moment capacity of the unit slab along its length. More details regarding
bridge analysis and design can be found in Rizzo et al. (2005). The CFRP laminates were applied by a certified
contractor in accordance to manufacturer’s specification. The strengthening of the bridge was completed in 48 hours
without traffic disruption.
The theoretical load rating (Cai and Shahawy, 2003) was calculated for four different truck types, mandatory by
Missouri Department of Transportation according to the Manual for Condition of Bridges (ASSHTO, 1996). A
detailed description of these trucks can be found in Rizzo et al. (2005). By comparing the load rating calculations
before and after strengthening it was showed that the FRP strengthening allowed increasing the posting load of the
bridge from 4.5 tonSI (5 ton) to 14.5 tonSI (16 ton).

Table 2: Bending Moments and Shear Forces per Unit of Bridge Deck

Concrete Steel FRP - LAMINATES

Compressive Yield Modulus of Tensile Modulus of Thickness
Strength Strength Elasticity Strength Elasticity
[ MPa ] ([ psi ]) [ MPa ] ([ psi ]) [GPa ] ([ ksi ]) [ MPa ] ([ ksi ]) [GPa ] ([ ksi ]) [ mm] ([in])
41.4 (6000) 455.0 (66) 200.0 (29000) 2700 (390) 230 (33000) 0.167 (0.007)

3. CONCLUSIONS

Conclusions based on the retrofitting of the bridge utilizing FRP materials can be summarized as follows:

105
 • The FRP externally bonded system showed to be a feasible solution for the strengthening of the bridge;
• As a result of FRP strengthening, the load posting of the bridge was increased from 4.5 tonSI (5 ton) to 14.5
tonSI (16 ton), corresponding to a 320% enhancement.

6 in 6 in
Deck D1 Deck D2 Deck D3 Deck D4

6 in

East
West

6 in S1 S2 S3 S4 S5

CFRP Laminates

Figure 4: Strengthening of the Deck: Plan View

Position Along the Slab, z (in)

0 5 10 15 20 25 30 35 40 45 50 55 60 65 70 75 80 85 90 95
0

1 in = 1” = 25.4 mm
1 kip = 4.45 kN
Moment (kip-ft / ft)

10

Design Moment Envelope

15

Moment Capacity
20

25

Figure 5: Diagram of the Capacity of the Deck at the Ultimate Load Conditions

4. REFERENCES

AASHTO, 1996, LRFD Bridge Design Specifications, Second Edition, American Association of State Highway and
Transportation Officials, Washington, DC.
AASHTO, 2002, Standard Specifications for Highway Bridges, 17th Edition, American Association of State
Highway and Transportation Officials, Washington, DC.
ACI Committee 440, 2002, Guide for the Design and Construction of Externally Bonded FRP Systems for
Strengthening Concrete Structures (440.2R-02), American Concrete Institute, Farmington Hills, MI.
Cai, C. and Shahawy, M. (2003). “Understanding Capacity Rating of Bridges from Load Tests” Practice Periodical
on Structural Design and Construction. ASCE. Vol. 8. No 4. pp. 209-216
Lopez, A. and Nanni, A., (2004). “Validation at a Large Scale of FRP Technology for Bridges Strengthening”,
Submitted to Concrete International – ACI.
MoDOT Bridge Load Rating Manual. Missouri Department of Transportation, Jefferson City, Missouri, 1996, pp
4.1-4.28
Rizzo, A.; Galati, N. and Nanni, A., (2005). “Strengthening of Bridge No. 390006 Route 0039 - Washington
County, MO” CIES Report 04-055-d, University of Missouri-Rolla, Rolla, MO, USA.
Rocca, S., N. Galati, and A. Nanni, (2006). “Experimental and Analytical Evaluation of Large-Size Reinforced
Concrete Columns Strengthened with FRP”, Report UTC-142, University of Missouri-Rolla, Rolla, MO, USA.

106
Third International Conference on FRP Composites in Civil Engineering (CICE 2006)
December 13-15 2006, Miami, Florida, USA

ENHANCEMENT OF SHALLOW DEPTH PATCHES FOR CONCRETE

BRIDGES USING FRP OVERLAYS
Golrokh Nossoni
(Graduate Research Assistant, Michigan State University, East Lansing, MI, USA)

Ronald S. Harichandran, Ph.D., P.E

(Professor and Chairperson, Michigan State University, East Lansing, MI, USA)

Rigoberto Burgueño, Ph.D.

(Associate Professor, Michigan State University, East Lansing, MI, USA)

ABSTRACT
The use of fiber reinforced polymer (FRP) fabrics applied as overlays on top of traditional patching materials was
investigated to provide more durable patches on concrete structures. FRP overlays provide the benefits of keeping
the patch in place for a longer period, restraining the diffusion of moisture and chloride ions, especially when
patches are cracked, and reducing further damage to the concrete element. Finite element (FE) models were devel-
oped to assess the damage on the patching material due to shrinkage, corrosion, and mechanical load. The demands
from the numerical studies were used to evaluate suitable FRP fabrics and their configurations. It was found that a
layer of bidirectional glass FRP is adequate for shallow depth patching applications. Experiments are being con-
ducted to validate this result.

KEYWORDS
Shallow depth patch, Shrinkage, Cracking, Bond Strength, Debonding, FRP, Delamination

1. INTRODUCTION

Shallow depth surface patches, used to repair damaged concrete bridge elements typically crack extensively soon
after application. These patches can be classified into two primary groups: cementitious mortars (CM) and polymer
modified cementitious mortars (PM). Cracking and full or partial delamination of the patching material from the
concrete substrate is generally unavoidable. The cracking and delamination is related to many factors and the pri-
mary factor is the shrinkage of the repair material at an early age. The tensile strength, modulus of elasticity of the
repair material at an early age, and structural characteristics govern the effect of restrained shrinkage (Yuan et al.,
2003). Restrained shrinkage results in the development of various stresses which can lead to premature failure of the
patch. Failure modes include vertical cracking due to direct tension and delamination due to interfacial stresses. The
problem is further exacerbated if the repaired structure is subjected to an aggressive environment, where cracks pro-
vide free access for intrusion of chloride ions and diffusion of carbon dioxide (Baluch et al., 2002). The choice of an
optimum repair material should be based on the best compromise of required properties, and may also be influenced
by the availability of materials and technical experience, and other constraints, such as the application technique and
construction environment. But even with patching materials that are highly compatible with concrete, the repair
typically lasts only for a few years. Most of the research done in this area has focused on improving the compatibil-
ity between the concrete substrate and the patching material to prevent cracking, but finding reliable criteria for the
required parameters is difficult. To minimize this problem and improve the performance and durability of shallow
depth surface patches on concrete substrates, the use of an FRP overlay as a secondary reinforcement is investigated
in this study. The additional reinforcement can hold the patching material in place and prevent its premature failure.
Finite element simulations of the proposed patching system under shrinkage, corrosion and mechanical loads are
reported in this paper. Experimental studies to corroborate numerical predictions are currently underway.

107
2. NUMERICAL EVALUATION OF PATCH REPAIRS WITH AND WITHOUT FRP
Two- and three-dimensional finite element analyses (FE) were conducted using the general-purpose ABAQUS pro-
gram (version 6.3). 2-D models were used to study the effect of restrained shrinkage, mechanical load, and corrosion
on the repair material and concrete substrate when the FRP overlay was not used. 3-D models were used to select
optimally configured layers of FRP on damaged patches. The analyses were conducted on models with the geome-
try, boundary conditions, and loading to be used for the experimental specimens. Simply supported beams with di-
mension of 16" × 4" × 3" are being used in mechanical load and shrinkage tests, and beams with dimension of 12" ×
6" × 6" are being used in the corrosion test. Both beam geometries have a cavity of dimension 4" × 3" × 0.625" and
4" × 5" × 0.625", respectively, on the bottom side. Full bonding was assumed between all interfaces, but some inter-
faces were allowed to separate using a node-by-node approach to simulate crack propagation and debonding. Rein-
forced concrete was modeled as an elastic-plastic material, and the patch materials were assumed to be elastic. The
physical and mechanical properties of the patch and FRP materials used in the FE models are given in Table 1.

Table 1: Patch and FRP Materials used in Finite Element Analyses

Patching Materials FRP

Elastic Shrinkage Mate- Fiber Thickness
Material Type Type E11 (psi)
Modulus (psi) Strain (µε) rial Orientation (in.)
Material 1 PM 2.0E+6 350 FRP 1 Glass Unidirectional 3.47E+6 0.05
Material 2 PM 2.0 E+6 1410 FRP 2 Glass Chopped 1.17E+6 0.04
Material 3 PM 3.6 E+6 800 FRP 3 Glass Bidirectional 2.47E+6 0.013
Material 4 PM 2.5 E+6 760 FRP 4 Carbon Unidirectional 1.02E+7 0.02
FRP 5 Carbon Bidirectional 6.60E+4 0.01

First 2-D analyses were performed to determine the most likely location of cracks in the patching material. Subse-
quently, exhaustive 3-D analyses considering different crack patterns were performed. In all the analyses only one
layer of the FRP overlay was considered. For uni-directional FRP, the fibers would normally be oriented in the lon-
gitudinal direction for mechanical load and in the transverse direction for the corrosion load. However, since only
one layer of FRP overlay is considered and the load carried by the FRP is expected to be low, all analyses for uni-
directional FRP were performed twice with fibers oriented in the transverse and longitudinal directions, respectively,
to determine whether the load normal to the fibers could be carried by the matrix alone.

2.1 Failure Modes

Failure modes in the patching material include vertical cracking and bond failure, while failure modes in the FRP
overlay include rupture of the fibers or matrix, and bond failure between the patch and the FRP overlay. The poten-
tial for bond failure was assessed through a debonding index (DI):
Fb = (σ ii / σ n )2 + (τ ij / τ n )2 (1)
where τn = shear bond strength, σn = tensile or compression strength of the patch material, and τij and σii are the shear
and normal stresses acting on the bond interface, respectively. The influence of normal stresses along the FRP bond
was neglected.

The potential for tension cracking was evaluated using the magnitude of the principal tension stress. A tension
cracking failure index (TCI) was defined as:
Ft = σ p max / f t (2)
where σp max = maximum tension principal stress, and ft = tensile strength of the patch material. Rupture in the fibers
or matrix of the FRP can be evaluated using the same concept. A fiber/matrix rupture index (F/MRI) was defined as:
Fr = σ ii / f ultimate (3)
where σii = stress in the direction of fibers or matrix, and fultimate= ultimate tensile strength of the fibers or matrix
depending on the FRP orientation. Failure is assumed to occur when any of the above functions exceed unity.

108
2.2 Simulation of Load Effects

Since the research focus is on the post-shrinkage behavior of the patch material, a complex diffusion-based tech-
nique (Baluch et al., 2002) is not considered necessary. Shrinkage strains were applied as a uniform initial strain in
the patch material and the time-dependency of shrinkage was neglected. The mechanical load was limited to that
which produced a strain demand of 0.2εy in the steel reinforcement. Figure 1(a) shows the TCI-values due to shrink-
age at critical locations along the depth of the patch calculated using 2-D analysis. The figure indicates that all
patching materials other than Material 1 are expected to crack. After the first crack opens, additional tensile cracks
are likely to open at random locations and prevent debonding on the bond surfaces. However, under the mechanical
loads debonding failure is most likely.

Path Crack (mm) Crack Path (mm)

0.0 5.0 10.0 15.0 0.0 5.0 10.0 15.0 20.0 25.0 30.0 35.0
5.0 2.5
Material1 Material1- FRP4
4.5 Material2- FRP4
Material2
Material2- FRP5
4.0 Material3 2.0
Material1- FRP2
Tension Crack Index

Fiber Rupture Index

Material4 Material2- FRP2
3.5
Material1- FRP1
3.0 1.5 Material2- FRP1
Material1- FRP3
2.5 Material2- FRP3

2.0 1.0

1.5

1.0 0.5

0.5

0.0 0.0
0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4

Crack Path (in.) Crack Path (in.)

(a) (b)

Figure 1: (a) TCI along the most critical path; and (b) FRI along the crack path in the FRP

The two patch materials that had the best and worst performance in the 2-D analyses (Materials 1 and 2) were used
to evaluate the behavior of the shallow repairs with FRP overlays using 3-D analyses. The calculated values of the
FRI in the critical model are shown in Figure 1(b). In this model a single crack was introduced in the patch material
along the transverse direction at the most likely location, and the model was analyzed under combined shrinkage and
mechanical loads. As Figure 1(b) shows, only one of the selected FRP overlays (chopped glass FRP2) fails due to
the applied load when used with patching material 2. Uni-directional FRPs had fibers oriented in the longitudinal
direction. Analyses of uni-directional FRP with fibers oriented in the transverse direction showed that the matrix is
not strong enough to carry the applied load in the longitudinal direction and would therefore crack.

When reinforcement bars corrode, the corroded steel swells to about 4 to 6 times its initial volume. However, some
of the corrosion products are likely to fill pore voids within the concrete. The expansion due to corrosion was esti-
mated by calibration with available experimental data for FRP wrapped concrete cylinders subjected to accelerated
corrosion (Baiyasi, 2000). Based on this calibration, an expansion strain of 17,500 µε was applied to the entire cross
section of the #4 bar. Figure 2(a) shows the TCI along different paths due to corrosion for the worst patching materi-
als, calculated using a 2-D model. It clearly indicates that the first crack is likely to open along Path 1.

Based on the results from 2-D analyses, a vertical pre-opened crack was introduced in the 3-D model radiating
downward from the bar. The model was studied under combined shrinkage and corrosion loads with fibers oriented
in the transverse direction. Only the patch material having the highest shrinkage (Material 2) was considered. Figure
2(b) shows the FRI along the crack path and indicates that the chopped glass FRP overlay (FRP 2) is expected to
fail. The other FRP overlays do not fail under this combined load. Analyses with unidirectional fibers oriented in the
longitudinal direction showed that the matrix could not sustain the stresses in the transverse direction and that fibers
oriented in the transverse direction would be necessary.

Figure 3 shows the DI-values calculated to evaluate the state of the bond between the FRP overlay and the patching
material. It was assumed that bond failure initiates in the patching material. The calculated DI-values transverse to
the crack path for critical models indicate that there will be debonding near the crack location over a 0.2-0.3 in.
length for some FRP overlays (left side of Figs. 3(a) and (b)), and debonding near the patch/concrete interface at the
ends of the patch over a 0.1-0.2 in. length for two overlays (right side of Figs. 3(a) and (b)).

109
 Path (mm) Crack Path (mm)
0.0 2.0 4.0 6.0 8.0 10.0 12.0 14.0 0.0 5.0 10.0 15.0 20.0 25.0 30.0 35.0 40.0 45.0 50.0
25.0 3.0
Material2-Path1 Material2 - FRP1
22.5
Material2-Path2 Material2 - FRP2
2.5
20.0 Material2-Path3 Material2 - FRP3
Tension Crack Index

Fiber Rupture Index

Material2-Path4 Material2 - FRP4
17.5
2.0 Material2 - FRP5
15.0

12.5 1.5

10.0
1.0
7.5 Path 4-3-2-1
5.0
0.5
2.5

0.0 0.0
0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.6 1.8 2.0

Path (in.) Crack Path (in.)

(a) (b)

Figure 2: (a) TCI along different paths; and (b) FRI along the crack path in the FRP overlay

Debonding Path (mm) Debonding Path (mm)

0.0 5.0 10.0 15.0 20.0 25.0 30.0 35.0 40.0 45.0 0.0 5.0 10.0 15.0 20.0 25.0
3.5
2.4
Material1- FRP4 Material2 - FRP1
2.2 Material2- FRP4 3.0 Material2 - FRP2
2.0 Material2- FRP5
Material2 - FRP3
Material1- FRP2
Debonding Index

Debonding Index
1.8
Material2- FRP2
2.5 Material2 - FRP4
1.6 Material1- FRP1 Material2 - FRP5
1.4 Material2- FRP1 2.0

1.2
Material1- FRP3
Material2- FRP3 1.5
1.0
0.8
1.0
0.6
0.4 0.5
0.2
0.0 0.0
0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.6 1.8 0.0 0.2 0.4 0.6 0.8

Debonding Path (in.) Debonding Path (in.)

(a) (b)

Figure 3: DI along the transvese crack path for: (a) model shown in Figure 1(b);
and (b) model shown in Figure 2(b)

3. CONCLUSIONS

Analytical results for mechanical, shrinkage, corrosion, and combined loading effects on patch repairs with an FRP
overlay show that one layer of uni-directional fabrics (FRP 1 and FRP 4), with fibers oriented either along the trans-
verse or longitudinal direction, are insufficient to withstand the combined load. One layer of the chopped glass fab-
ric (FRP 2) also is inadequate. However, one layer of the bi-directional fabrics (FRP 3 and FRP 5) is sufficient to
resist all loading effects with either of the patching materials considered. Since glass FRP is more economical than
carbon FRP, one layer of FRP 3, or any other bi-directional glass reinforced FRP with similar mechanical properties,
should be an effective solution for use as an overlay for improving the performance and durability of shallow depth
patches in concrete structures.

4. REFERENCES
Baiyasi, M. I. (2000). “Repair of corrosion-damaged columns using FRP wraps.” Ph.D. Thesis, Michigan State Uni-
versity, East Lansing, Michigan.
Baluch, M. H., Rahman, M. K., and Al-Gadhib, A. H. (2002). “Risks of cracking and delamination in patch repair.”
Journal of Materials in Civil Engineering, July/August, pp 294-302.
Yuan, Y., Li, G., and Cai, Y. (2003). “Modeling for prediction of restrained shrinkage effect in concrete repair.”
Cement and Concrete Research, Vol. 33, No. 3, pp 347-352.

110
Third International Conference on FRP Composites in Civil Engineering (CICE 2006)
December 13-15 2006, Miami, Florida, USA

EXPERIMENTAL STUDY ON PRESTRESSED CONCRETE GIRDERS

STRENGTHENED WITH CFRP TAPES
AT DIFFERENT BRIDGE REPAIR STAGES
Arkadiusz Mordak
(M.Sc., Ph.D. Student, Cracow Technical University, Cracow, Poland)

Zbigniew Manko
(Professor of Civil & Structural Engineering, Wroclaw University of Technology, Wroclaw, Poland)

ABSTRACT
This paper presents the results of the research on a five-span prestressed concrete road bridge under static load
performed at different stages of its repair. All bridge spans become reinforced by CFRP tapes glued on bottom
flanges of main beams, and external flat steel stirrups, as well as overlaid a new concrete deck layer. The main aim
of the conducted bridge repair was to increase its load capacity to 300 kN (class C) according to the Polish Loads
Standard (PN-85/S-10030). The results obtained during the research at different stages of repairs were conducive to
determining the behavior of the analyzed span structure under static loads, which allowed for an assessment of the
efficiency of the strengthening, as well as establishment of guidelines for future reference concerning this type of
strengthening in the engineering practice.

KEYWORDS
Bridges, Prestressed concrete beam, CFRP tapes, Field tests, Static loads, Repair stages

1. INTRODUCTION
This paper presents research on the concrete spans with the prestressed main beams of the road bridge on the Nysa
Klodzka River situated in Paczkow (Opole Province, at 72+536 km along the road no. 328 Stanowice – Paczkow).
The research was conducted at different stages of the repair works, that is, before performing the main research on
static and dynamic field load tests (acceptance inspections), which was aimed to determine the efficiency of the
applied repairing methods. Before the repair, the bridge load capacity determined in expertise was classified as class
E, that is 150 kN according to the Polish Loads Standard (PN-85/S-10030), mainly due to a very poor technical
condition of the load-carrying structure of all bridge spans, resulting mostly from transverse cracks in the main
beams (Figure 1). The main aim of the repair was to increase the object load capacity to class C (300 kN).
The aim of the conducted research was to determine the behavior of the chosen spans structures subjected to
considerable static loads at various indirect stages (phases) of the performed bridge repair. The research allowed to
find out on which elements of the load-carrying structure of span the biggest forces were exerted during the progress
of repair works on one half of the bridge. It is quite an important problem not only in terms of construction works
safety but also of vehicles and people using the non-repaired part of the bridge is concerned. Furthermore, there is
also a possibility of overloading some structure elements, like e.g. main beams.
It should be added that it is a relatively rare activity to perform research on a bridge or its elements during the repair
works. The inspection of spans and analysis of the obtained results performed each time after the accomplishment of
repairs allow determining the influence of the same load on the quality and durability of this object in the process of
strengthening as well as the efficiency and purposefulness of this process.
The final results of the bridge acceptance inspection (PN-S-10040, 1999), conducted after the complete repair under
the trial static and dynamic load (Manko and Mordak, 2002), allowed a comprehensive evaluation of the efficiency
of the main beams strengthening by applying CFRP tapes. Moreover, it enabled a comprehensive evaluation of the
change of the spans structures behavior under the same load during different stages of repair works, simultaneously
with the standard traffic running on the second, then non-repaired half of the bridge which included I, II and V spans.

111
 (a) SIDE VIEW LONGITUDINAL SECTION (b)
Kamieniec Ząbkowicki Paczków
18.000 18.000 18.000 18.000 18.000
1-6 - main beams deflection gages
Material parameters:
T1-T14, T10', T11' and T13' - strain gages
- steel:
cables - Freyssinet type 12 5,
T10', T11' 1-6 Span I T5', T8' 1-6
Span III
Span IV T2, T5, T8 1-6
T2, T5, T8 T1-T14 Span II Nysa Kłodzka River T10, T11, T13 Span V tension strength: 1650 MPa,
T13'
T10, T11, T13
stirrups: 240 MPa,
18.375 18.750 18.750 18.750 18.375
(c)
- concrete - B30,
TOP VIEW
Measurements: V, VIII and XI Measurements: VII and X Measurements IX compression strength: 22,5 MPa
- CFRP tapes -

2.500
1.850
1.980
1.275 2.840 1.320 1.695
tension strength: 2400 MPa
7.130

Measurements: I, II, III and IV Measurements VI i - numbers of main beams deflection (or strain gages)

Figure 1: Post-tensioned concrete road bridge: (a) side view from headwater and (b) longitudinal section,
(c) top view – technical parameters of truck type KAMAZ 5511 and load distribution on bridge spans during
particular stage of field tests, and extensometers localization on strengthening strips during field research

2. RESULTS OF RESEARCH AND DISCUSSIONS

The strengthening of the particular bridge spans was accomplished by gluing the tapes made of carbon fibers CFRP
(three for each beam) to the bottom flanges of the main beams and constructing an additional reinforced concrete
deck plate of variable thickness 0.12–0.185 m and adding the outer stirrups in the form of steel flats of 5×50 mm
section, with axial base by every 0.35 m (Manko and Mordak, 2002; Schoenradt 1998).
Stage I 1.62
Handrail prestressed by steel cable Stage X
(a) 1.71 Stage VII Stage VI 1.68 1.25

Span I (j)
(f)
Right side + (10 m)
3
-0.52
0.42

5.47
3.35

Sika CarboDur M1214 Span II CFRP

6.73
3.82

Span II
1.11
0.89

1.50
1.88

Left side 3
1.93
2.41

Right side 3 (10 m)

65.5 2.45
45.0 1.78

2.87
2.06

+ (10 m) +

0.60
-0.11
-7.8 -0.15

0.94
0.50
7.5 0.57

0.49
0.81

1.17
1.27
1.18
1.18
4.03
2.61

1.85
1.62
53.1
42.5

2.41
2.00
2.82
2.33
65.0
90.4

+ 6
(10 )
13.7
7.5

34.1
25.0

6
+ (10 )
-18.0
7.5

CFRP Strains

-7.0
12.5
14.8

7.5
32.5
34.4
22.1

62.5
72.0

52.5
56.9
20.0

65.4
70.0

(g) 6
75.0

1.27 53.1
124.4

60.0

90.0
89.2
65.0
47.5

(10 )
55.0
77.2

+
+ 6
104.5

87.5
93.7

(10 )
60.0

Stage II Span II Stage VII

1.24
1.71
2.05

1.58 Left side 3

2.78
2.06

+ (10 m)
3.40

CFRP Strains
-0.19
0.51

(b)
0.49
0.87

60.0
93.8
2.37

+ 6
(10 )
25.0
32.9

40.0
-5.2
55.0
58.5

60.0
48.9
Span I

55.0
13.8
70.0
31.3
Sika CarboDur M1214
50.0
69.5

6
75.0
77.0

+ (10 )
5.0
-7.3
99.0
97.5

10.0
12.1

Right side + (10 m)

3
1.25
CFRP Strains Stage XI
-0.51

5.22
0.53

3.35
0.83
1.12

6
6.46
3.99

+ (10 )
29.6
20.0
1.97
2.29

(k)
53.3
12.2

36.7
-7.2

Stage VIII
3.84
2.67

1.25
Span I CFRP
Left side 3
1.71
1.87
1.90
2.43

6 (h)
2.84
2.17
119.5

(10 m)
99.2
82.5

+ (10 ) +
95.0

0.50
-0.12
17.5
-17.5

0.85
0.50
20.6
10.0

1.17
1.29
45.0
61.5

73.1
55.0

1.24 Span I Sika CarboDur M1214

Stage III
Left side 3
3.67
2.07

+ (10 m)
0.38
-0.57
0.86
0.67

60.0
92.3

52.5
67.4

40.0
53.1

15.0
13.2
25.0
33.1

7.5

(c)
-6.8

Steel
2.07
1.41
5.28
2.62

stirrups + 6
Span I Sika CarboDur M1214 5 x 50 (10 )
6.85
3.22

7.5
-19.7

Right side
1.73
2.17

3
2.80
2.12

+ (10 m)
15.0
16.8

CFRP Strains
3.24
2.50

125.0
67.4
0.78
0.57
0.48
-0.17

30.0 45.0

6
46.1 53.1

+
110.0
1.37

91.1

(10 )
1.32

55.0
73.2
77.5
104.0

6
97.5
130.5

+ (10 )
20.0
-4.9
50.0
13.3
10.0
-19.7

65.0
30.3

6
80.0

+
97.9

(10 )
16.8

80.0
47.4
10.0
-6.2

22.5
14.0
20.0

30.0
38.6

CFRP Strains
54.8
40.0

71.4
60.0

6
(10 )
1.71 Stage V Stage IV 1.76 Stage IX
1.25 - measured values
- calculated values
(i)
(d)
Span I Sika CarboDur M1214 Span V CFRP
1.68
2.05

Left side 3
1.46
1.82

(10 m)
Figure 2: Diagrams of total
2.37

Right side
1.78

3
2.77

3.62
2.02

2.37

+ (10 m) +
0.46
-0.55
5.21
3.10
-7.1 -0.16

0.67
1.05
0.79
0.47
0.0 0.41

1.14
1.14

6.74
3.79

6 deflections δt and strains εt received

87.3
77.5

+ (10 )
2.5
14.1

7.5
-19.2

from research (values from the left)

20.0
34.3

(e)
37.5
50.6

52.4
45.0

16.7
30.0
70.4
50.0
64.2
55.0

Span I
6
and from calculation δc and εc
92.5

97.5
124.8

75.0

+ (10 )
1.98

Left side 3
1.72
2.68
2.07

+ (10 m)
3.28

-0.19
0.47
0.83
0.47

(values from the right) of main

85.0 2.39

1.23
1.20

CFRP Strains
6
122.8

80.0
97.8

beams in I or II and V span cross-

85.0

6 + (10 )
72.5
74.3

+ (10 )
95.4

5.0
-7.0
12.6
10.0

15.0
-19.3
16.8
55.0

55.0
45.6
27.5
31.7

60.0
69.3

section in stages I–XI

56.5
52.5

112
The research was conducted at 11 different stages during the bridge repair. Figure 1 shows the load scheme on the
tested spans at different stages with the measurements points localization. The following sizes were made:
– six main beams deflections,
– vertical and horizontal displacements of the chosen expansion and fixed bearings, made by dial indicators with
1×10–5 m accuracy,
– strains (indirectly – normal stresses) in the main beams, which were performed by strain gages (extensometers)
and mechanical indicators,
– strains in the strengthening tapes in half and 1/4 of the effective span of the main beams, and
– strains in the outside stirrups in bearing zones and in midspan, which were performed by strain gages.
The measured permanent deflections of the particular main beams differ inconsiderably among each other, and are
not proportional to elastic deflections and do not exceed permissible values (δperm = lt / 800 = 18.00 / 800 =
22.50×10–3 m). The relations of the main beams permanent deflections to their total deflections were established on
the basis of diagrams shown in Figure 1and in Table 1.
The maximum total main beams deflections; measured during the research under static load at different stages of
bridge repairs, oscillate between 2.02×10–3 m and 3.99×10–3 m.
The average deflection values of spans cross-sections were smaller than calculated ones in all cases. For the possible
load schemes to execute during the repair works, average measured correspond to calculated elastic deflection
relations for particular stages were as follow: I – 0.696, II – 0.752, III – 0.895, IV – 0.904, V – 0.925, VI – 0.906,
VII – 0.900, VIII – 0.588, IX – 0.702, X – 0.991, XI – 0.969. The measured main beams deflection distributions in
span cross-sections were different than calculated ones. It is proved that a real torsional stiffness of the bridge was
higher than assumed in static calculations.
The differences between the initial and the final readings found in measurements were approximately the same in all
spans and examined main beams cross-sections. It may prove that they were rather a result of new bearings
settlement or under bearing joints and possible insignificant reading errors of the measurement instruments, and only
to a minimal rate of permanent strains of the prestressed main beams. Prestressed concrete spans structure should not
indicate bigger permanent strains, however old bearings became replace by new ones, previously not loaded with
such significant loads. This might have resulted in the occurrence of some insignificant settlements and differences
in readings under proportionally heavy static load.

Table 1: Values of total (upper row) and permanent (bottom row) deflections and strains of the main beams

Deflections δt and δp in (10–3 m) Strains εt and εp in (10–6)

Stage of Date of
Main beam number
research measurement
B1 B2 B3 B4 B5 B6 B1 B2 B3 B4 B5 B6
30.08.2001 3.82 3.35 2.61 1.90 1.11 0.56 37.50 57.5 47.5 52.5 35.0 25.0
I
(span I) –0.05 –0.05 –0.07 –0.19 –0.17 –0.32 0.0 0.0 –20.0 5.0 –7.5 0.0
10.09.2001 3.99 3.35 2.67 1.97 1.12 0.53 132.5 82.5 95.0 85.0 100.0 80.0
II
(span I) 0.14 0.16 0.15 0.11 0.02 –0.06 80.0 25.0 45.0 25.0 82.5 47.5
20.10.2001 2.50 2.12 1.73 1.33 0.78 0.48 150. 0 135.0 95.0 85.0 47.5 0.0
II
(span I) 0.13 0.11 0.06 0.02 –0.05 –0.01 97.5 25.0 40.0 0.0 2.5 0.0
10.11.2001 2.02 1.78 1.46 1.14 0.79 0.41 57.5 30.0 80.0 107.5 57.5 20.0
IV
(span I) –0.03 –0.05 –0.08 –0.04 –0.07 –0.15 20.0 27.5 37.5 –5.0 22.5 7.5
10.11.2001 0.47 0.83 1.23 1.72 2.07 2.39 5.0 7.5 32.5 135.0 117.5 87.5
V
(span I) 0.04 0.02 0.01 –0.01 –0.05 –0.03 0.0 –2.5 22.5 –5.0 –7.5 2.5
17.11.2001 2.06 1.78 1.50 1.18 0.81 0.57 120.0 45.0 72.5 17.5 0.0 32.5
VI
(span II) 0.09 0.02 0.03 0.02 –0.16 –0.16 55.0 2.5 45.0 2.5 –7.5 25.0
17.11.2001 0.51 0.87 1.27 1.71 2.07 2.37 20.0 15.0 25.0 75.0 30.0 97.5
VII
(span II) 0.08 0.04 0.00 0.00 –0.05 –0.01 2.5 –2.5 0.0 2.5 –5.0 2.5
16.02.2002 0.38 0.86 1.41 2.07 2.62 3.22 17.5 55.0 45.0 55.0 107.5 72.5
VIII
(span I) –0.08 –0.21 –0.19 –0.28 –0.38 –0.53 –5.0 37.5 –2.5 10.0 30.0 15.0
08.03.2002 0.46 1.05 1.68 2.37 3.10 3.79 7.5 40.0 45.0 50.0 75.0 92.5
IX
(span V) –0.02 –0.09 –0.12 –0.16 –0.14 –0.14 7.50 –2.5 5.0 0.0 25.0 15.0
18.05.2002 0.60 0.94 1.27 1.62 2.00 2.33 27.5 52.5 77.5 103.5 182.5 87.5
X
(span II) –0.02 –0.03 –0.01 –0.02 –0.01 0.00 12.5 37.5 22.5 69.0 182.5 52.5
25.05.2002 0.50 0.85 1.29 1.71 1.90 2.17 –32.5 75.0 25.0 60.0 82.5 60.0
XI
(span I) –0.08 0.02 0.06 0.13 0.00 0.06 –32.5 75.0 2.5 50.0 2.5 2.5

113
3. CONCLUSIONS
The practical experiments of the performed bridge spans research under static load at different stages of its repairs
and also conducted observations of their structure interaction together with the comprehensive analysis of the results
obtained from measurements and their comparison to calculated ones allow to formulate the following conclusions
of the general character (Manko and Mordak, 2002; Meier and Deuring, 1991; Täljsten, 2004):
1. The span structures with the post-tensioned prestressed concrete main beams strengthened by CFRP tapes and
outer stirrups in the form of steel flat did not arouse any reservations as far as the sizes of section forces, displace-
ments and strains obtained from the research and calculations are concerned. Deflection and strain values of the main
beams obtained from the measurements did not raise any doubts, as far as the strength and carrying capacity in tested
spans, in behavior and interaction of the bridge at 11 repair stages are concerned. As the most loaded span structure
elements, which need a detailed study and analysis, one should consider the elements of the bridge deck plate where
stresses due to their direct load with stresses due to their interaction with main beams and crossbeams sum up.
2. A satisfactory agreement of main beam deflections in all of the tested spans was obtained in comparison with
calculated ones in all cases for each repair stages (stiffness of elements, layers thickness, etc.). This shows
correctness of assumptions taken for calculation and static-strength analysis of these spans or also the correctness of
assumed analytical structure model with their real behavior at particular repair stages.
3. In most cases of tested main beams sections and elements of superstructure (deck plate) in the analyzed spans at
the different repair stages of the bridge, the deflections and strains (and normal stresses which were calculated on
their basis) during field load tests have had an elastic nature. Even without consideration of bearing settlements and
displacements, they have appeared mostly smaller than the calculated ones. The noticed some minimal values of
permanent displacements were partially the permanent deflections of the main beams and to some extent reading
errors and errors of measuring devices.
4. The grid model of variable load-capacity structure which was assumed for the first calculation at different repair
phases depending on layers and strengthening tapes seems to be sufficient tool to determine the deflections and
strains in tested structures on an engineering level (what it shows also in the presented juxtaposition of the obtained
results – Table 1). To complete the detailed analysis of interaction between particular pavement layers and structure
components and the assumed strengthening manner it is necessary to use more complex model which should reflect
better a real interaction in a such type of span structures at the considered repair stages of the bridge, especially on
the contact section of concrete and CFRP tapes.
5. Normal stresses in tapes (which were calculated on the basis of strain values) show unambiguously that their load-
capacity used was only in about 5–10%. However, no strains were noticed in outer stirrups, but the small strain
gages movements are contained in a range of reading accuracy or measuring device errors. A job that can be
ascribed them could be connected with the protection against to the delamination of tapes (together with lag of
reinforcement bars), what was often happened in the strengthening solutions (Manko and Mordak, 2002).
6. The differences in the expected deflections and strains of main beams in the relation to those obtained from meas-
urements after the repair of one half of the bridge span proves that there was a good interaction between a new
bridge deck layer and pavement layers of the roadway. However, the application of strengthening of main beams
with CFRP tapes and outer stirrups did not bring about the significant changes in deflections and strains values of
main beams. The efficiency of the strengthening work was assessed as poor because of the lower level of stresses.
In the fact, above summary and main conclusions refer to the structures of the tested span elements of preset
geometric characteristics, particular element stiffness, and determined effective spans. However, it may be stated
that spans strengthening constructed by lamels and steel outer stirrups is not the best solution as far as this type of
structures are concerned, mostly from the economical point of view. In order to use expensive CFRP tapes to a
higher extent, one should install on the beams already known prestressing devices for CFRP tapes (Täljsten, 2004).

4. REFERENCES
Manko, Z., and Mordak, A. (2002). “Submitting Project and Report on Static and Dynamic Trial Load of Prestressed
Road Bridge over Nysa Klodzka River along National Road no. 382 Stanowice – Paczkow in km 72.536 in Paczkow
after its Repair.” Scientific-Research Center for the Development of Bridge Industry MOSTAR, Wroclaw, Poland.
Schoenradt, Z. (1998). “Technical Design of Bridge Repair over Nysa Kłodzka River along National Road no. 382
Stanowice – Paczków in Paczków in km 72.536.” Technical-Engineering Office KARO, Poznan, Poland.
Täljsten, B. (2004). FRD Strengthening of Existing Concrete Structures Design Guidelines. Division of Structural
Engineering, Luleå, Sweden.
Meier, U., and Deuring, M. (1991). The Application of Fiber Composites in Bridge Repair. Strasse und Verkehr, H.
9, S. 7–11.

114
Third International Conference on FRP Composites in Civil Engineering (CICE 2006)
December 13-15 2006, Miami, Florida, USA

FEASIBILITY STUDY OF THERMOPLASTIC WRAP FOR BRIDGE

PROTECTION
Nasim Uddin
(Associate Professor, Department of Civil Engineering at University of Alabama, Birmingham, Alabama,USA)

John D Purdue
(Graduate Student,Department of Civil Engineering at University of Alabama, Birmingham, Alabama,USA)

Uday Vaidya
(Associate Professor, Department of Materials Engineering at University of Alabama, Birmingham, Alabama,USA)

ABSTRACT
The research focused on the effects of low velocity impact loading on high strength concrete confined by a
prefabricated polypropylene jacket and comparing the results with similar specimens confined by carbon fiber. In
order to accomplish this, both static and dynamic load tests were performed. Concrete cylinders were used for static
loading. Twelve concrete cylinders were prepared for static load testing: three were plain concrete and used as
control specimens, three were wrapped with one layer of unidirectional carbon fiber, and six were confined by the
polypropylene pipe. The thickness of the polypropylene wrap was machined to different thicknesses; three 3 mm
and three 6 mm. The cylinders were standard 152 mm x 305 mm. Cylinders were loaded to failure in uniaxial
compression using a Tinius-Olsen Universal Testing Machine. Impact testing was performed using four 152 mm x
914 mm columns. The columns consisted of one control sample; one carbon fiber wrapped, and two (one of each
thickness) wrapped with polypropylene. Impact testing was conducted using an Instron drop-tower testing machine.

KEYWORDS
Thermoplastic composite, polypropylene jacket, low velocity impact.

1. INTRODUCTION
Concrete bridge piers are designed to withstand large compressive axial loads, but often fail under eccentric out-of-
plane loads, such as those created by an impact, earthquake or explosion. In the wake of recent terrorist attacks,
such loading is of increased concern. Retrofitting the piers with continuous-fiber-reinforced thermoplastic (CFRTP)
polymers could reduce vulnerability to these loads. Presently, fiber-reinforced thermoset polymers are used to add
stiffness and tensile strength of concrete bridge members, but there has been no effort made to use thermoplastic
composites for damage containment in structures such as bridges because of perceptions of high cost and expensive
manufacturing. This study explored thermoplastic composite material produced in continuous pultruded form to
produce a cost-effective split product form of directionally oriented glass fiber in polyurethane (or polypropylene)
thermoplastic matrix for a representative bridge column. Two split halves will encapsulate the column with on-site
mounting feasibility. The advantage of using pre-fabricated thermoplastic forms is they can be thicker than
conventional thermoset wraps (such as presently used in bridge structures, only from a standpoint of enhancing
stiffness/tensile strength). It is envisioned that under impact from unknown threats such as collisions from
trucks/trailer or blasts, the structure have progressive failure potential, in place of catastrophic fracture presently
witnessed. The tape and pultruded thermoplastic form has flexibility to accommodate curvatures encountered as part
of the structure, and can be used either alone (only to suppress catastrophic failure) or in conjunction with
conventional thermoset wraps if ductility improvement is also needed. An example of the concept is shown in Fig.
1. The split halves can be connected by a combination of thermoplastic tape wraps around the halves and a
secondary mechanical reinforcement. Furthermore, the rate of strain induced on the structure is severe. The
polypropylene/glass is expected to enhance failure strain of concrete structures by several orders of magnitude.

115
 Joining thermoplastic
tape rings to join the split halves. Secondary
mechanical bonding will also be provided

Glass/PP pultruded split shells with or

without thermoset warp inside to
encapsulate column structures

Fig. 1: Schematic of the Concept

Preliminary guidelines for field implementation of polypropylene jacket (PP) wrap systems are shown below: (1)
Existing RC columns can be strengthened using prefabricated PP or glass reinforced PP (glass/PP) shells. They can
be fabricated in half circles, half rectangles, circles with a slit in continuous rolls, so that they can be opened and
placed around columns (Figure 1). Two half shells of PP or glass/PP plate rolled to a radius of 0.5 to 1.0 in. (12.5 to
25 mm) larger than the column radius can be positioned over the area to be retrofitted and the vertical seams may be
site-welded (using ultrasonic welding) to provide a continuous tube with a small annular gap around the column. For
effective confinement to be achieved, a full contact between the column and the PP or glass/PP shell is essential.
This can be ensured by either bonding the shell to a thermoset wrap bonded to the pier (if ductility improvement is
needed); otherwise bonding the shell to the column using adhesives, or injecting shrinkage-compensated cement
grout or mortar into the space between the shell and the column; (2) Prefabricated PP or glass/PP shells can be used
as a stay-in-place forms for a precast modular bridge pier system for new bridge pier construction. The work will
compare the effects of dynamic loading of this type of confinement with the most common composite strengthening
technique to date, carbon fiber wrapping. Two series of tests will be performed in this research: uniaxial
compression testing of cylinders and impact loading of columns. The first letters are used to denote the type of
specimen, “Cy” for cylinder and “Co” for columns. The second letter establishes the confinement type, “N” for
plain samples, “C” for carbon FRP, and “P” for polypropylene wrapped cylinders. The next letter denotes the type
of concrete “B” for high strength. The first number in the scheme is for the confinement thickness (mm) or number
of plies. Finally, the last number represents the sample number. The Research Objectives:
• To investigate the effects of static loading on the polypropylene wrapped concrete cylinders and compare
with carbon fiber wrapped cylinders.
• To compare the energy absorption characteristics of columns confined by PP and carbon fiber reinforced
polymer (CFRP).

2. STATIC COMPRESSION LOADING

The purpose of the uniaxial compression tests of the concrete cylinders was to evaluate static loading phenomena
such as the effect on axial strength and strain capabilities. For this study, two variables were investigated:
confinement material (polypropylene and carbon fiber (for reference)) and thickness of the polypropylene
confinement. The average compressive strength of the concrete was 58.6 MPa. The cylinders were grouped as
follows: three control specimens; three 3 mm and three 6 mm thick polypropylene; and, finally, three 1-ply
unidirectional carbon fiber. Unidirectional SikaWrap Hex 103C was used for the CFRP, with Sikadur 300 used for
the bonding agent to the concrete surface. The fibers were oriented such that they provided reinforcement in the
hoop direction (perpendicular to the applied compression load). After preparing the concrete surface and the carbon
fiber, the material was rolled onto the cylinders. George Fischer beta (β)-PP was used for the polypropylene and
came in the following dimensions: 140 mm outer diameter, 13 mm wall thickness, and 5 m in length. This material
was chosen since it had many desirable characteristics, some of which are high impact strength, abrasion resistance,
low weight, and a sizable operating temperature range making it ideal for load bearing applications. Since the
polypropylene reinforcement is meant to act as passive reinforcement, the material was machined down from its
original 13 mm wall thickness to the two thicknesses previously mentioned. Table 1 below gives the material
properties for the two types of confinement. Comparison of the compression test data is presented in Fig. 1, and a
summary is presented as follows: (1) Effect of Confinement Thickness. No appreciable differences in load-bearing
and/or uniaxial concrete compressive strength were observed for the 3 mm or 6 mm polypropylene jackets. This
may be due to the high strain capacity within the PP, which masks the influence of the wall thickness (within the 3

116
 Table 1. Mechanical Data for Reinforcing Materials
Tensile strength Tensile modulus Elongation Nominal thickness
(MPa) (GPa) (%) (mm)
SikaWrap Hex 103C 958 73 1.33 1
β-PP 30 2.0 120 3&6
Note: β-PP = beta-nucleated polypropylene.

120 CyCB1
100 CyPB3
80 CyPB6
Stress (MPa)

60 CyNB0
40

20

0
0.000 0.002 0.004 0.006 0.008 0.010 0.012 0.014 0.016 0.018 0.020

Axial Strain

Fig. 1. Comparison of stress versus strain for cylinders (averaged values)

mm – 6 mm wall thickness range considered here); (2) Progressive failure potential. Both PP thicknesses illustrated
a significant improvement in deformability when compared to the unconfined concrete. This enhancement is
comparable and much higher to the improvements of the CFRP wrapped cylinders; (3) Compressive Strength. As
expected, neither series of polypropylene jacketing could produce a significant increase in compressive strength.
This was due to the ability of this material to bulge and dilate. As mentioned before, axial strengthening applications
are not the intended application of this material, however resistance to transverse loads is expected which is directly
related to sustaining higher compressive strains; (4) Stress-Strain Response. The stress-strain response of the PP
confined concrete is multi-linear in nature, and the mode of failure of PP wrapped specimens reflects a progressive
mode of specimen failure relative to a specimen, for example, CFRP jacket which had failed by sudden rupture of all
the fibers in a region of the jacket. The bulging of pp jacket, on the other hand, was concentrated along the upper
gage length, though some also occurred within the bottom regions to a much less extent. No separation or debonding
from concrete surface occurred along the PP. In term of normalized strain data given by the ratio εtu/εto (strain at
failure of the confined cylinder by strain of the unconfined cylinder), the 3 mm PP jacket produced an average ratio
of 8.4, and the 6 mm PP jacket yielded an average ratio of 9.6, which is impressive when compared with the 15.6
average ratio of the glass/PP confinement reported in earlier research (www.UTCA.org). The modes of failure
observed during these tests varied depending on the confinement. The PP confined samples exhibited a barreling
effect. The ability to dilate considerably allowed the confined concrete to crush and compact inside the PP jacket.
While this dilation was drastic, yielding of the polypropylene is evident in only a few places on the samples, and
only one sample showed signs of material failure. Failure of the CFRP wrapped cylinders occurred due to fiber
rupture near midheight.

3. LOW VELOCITY IMPACT LOADING

Low velocity impact testing was conducted using a total of four concrete columns: one control specimen, one CFRP
confined, and two confined by a PP wrap of 3 mm and 6 mm thicknesses. All columns tested were 152 mm x 914
mm. In an effort to illustrate a similar loading situation as would be seen in reality, the columns were placed
horizontally inside a testing jig and subjected to axial compression. The system used for impact testing was an
Instron Model 8250 drop-weight impact machine with an instrumented striker (tup) assembly. A flat striker was
used for this test and had an impact area of 76 mm x 102 mm. For this study, the impact weight was 246 N. The
hammer (tup) contained an internal load cell, which was used to record the contact load between the falling
assembly and the column during the impact event. The load cell was rated for a maximum load of just over 44 kN.

117
A drop height of 30 cm was used for all tests, since the combination of this height and the weight of the striker
assembly produced loading close to that of the maximum allowed by the load cell. Deflection, velocity, and energy
absorption were recorded using the DynaTup software that accompanied the Instron drop-tower. All samples were
impacted from a height of 30 cm in an effort to keep the load cell free from damage. Due to the limitations of the
drop-tower machine, the impact loading can be classified as low velocity impact or velocity less than 10 m/s (Bartus
2003). Average impact velocity for these tests was 2.4 m/s. A summary of test data is given in Table 2, and fig. 2.
Table 2. Summary of Impact Test Results
Specimen ID Peak load (kN) Maximum deflection at midspan (mm) Maximum strain Change in strain (%)
CoNB0 38 3.38 0.0024 ---
CoCB1 45 2.72 0.0047 143
CoPB3 36 4.52 0.0057 147
CoPB6 34 5.00 0.0058 148
50

6 45

40 CoNB0
5 CoNB0 CoCB1
CoCB1 35 CoPB3
CoPB3 CoPB6
CoPB6
4 30
Displacement (mm)

Force (kN)
25
3
20

2 15

10
1
5

0 0
0 2 4 6 8 10 12 0 1 2 3 4 5 6
Time (msec) Displacement (mm)

Figure 2. Comparison of load versus displacement; and displacement versus time for tested columns

The impact tests were conducted to assess the energy absorption capacity of three concrete columns strengthened by
PP confinement, a carbon/epoxy confined and one unconfined control specimen. From the test results, the following
conclusions were made: (1) Peak loading of the columns varied based on the stiffness of the confinement. Since the
PP confined columns were the least stiff, they also exhibited the desirable least peak loading from the impact
resistant design perspectives; (2) Deflection of the PP confined columns was greater than the unconfined and CFRP
wrap columns. This is very favorable given that the time this displacement occurred was nearly six times greater
than that of the plain specimen; (3) Transverse flexural strain values recorded across the middle gage length showed
that PP wrap strain demonstrates an increase in capability of approximately 145% over the unconfined concrete; (4)
Energy absorption of the polypropylene was significantly higher to that of the carbon fiber confinement. Energy
absorption for 3 mm PP wrap was higher than that of 6 mm wrap which can be attributed to the lower stiffness of the
former. Polypropylene wrap produced higher deflection than the CFRP wrap column and unconfined column. This
effect was due to the ability of the material to compress and further absorb the energy from impact; and (5) though
failure was not possible for the PP wrapped specimens, the above results demonstrated that the usage of a
thermoplastic prefabricated wrap has the potential to suppress the catatrophic failure due to the threat of impact.

4. CONCLUDING REMARKS AND ACKNOWLEDGEMENT

This study explored and demonstrated the potential of thermoplastic composite material (e.g., Polypropylene)
produced in continuous pultruded form to suppress catastrophic failure for a representative bridge column. However
much more studies are required before any definitive conclusions can be made regarding this issue. We gratefully
acknowledge the financial support of UTCA under the director Dr. Danial S. Turner.

5. REFERENCES
Vaidya U., Husman G. and Gleich K. (2004). “Design and manufacture of woven reinforced glass/polypropylene
composites for mass transit floor system”. Journal of Composite Materials, Vol. 38, No. 21, pp 1949-1972.

118
Third International Conference on FRP Composites in Civil Engineering (CICE 2006)
December 13-15 2006, Miami, Florida, USA

FIELD TESTS OF PRESTRESSED CONCRETE BEAMS

OF ROAD BRIDGE SPANS STRENGTHENED BY CFRP TAPES
Zbigniew Manko
(Professor of Civil & Structural Engineering, Wroclaw University of Technology, Wroclaw, Poland)

Arkadiusz Mordak
(M.Sc., Ph.D. Student, Cracow Technical University, Cracow, Poland)

ABSTRACT
The paper is presented the method, scope, results and their analysis of the research, and the main conclusions drawn
from static field load tests performed on a five-span prestressed concrete road bridge after its repair. The bridge with
the effective spans of 18.00 m is located over the Nysa Klodzka river in Paczkow (Poland). All bridge span beams
became reinforced by CFRP tapes glued on bottom flanges of main beams as well as overlaid a new concrete deck
layer. The main aim of the conducted bridge repair was to increase its load capacity to 300 kN (class C) according to
the Polish Loads Standard (PN-85/S-10030). The results obtained during the research of the bridge after its repair at
field load tests were conducive to determining the behavior of the analyzed span structure under static loads, which
allowed for an assessment of the efficiency of the strengthening, as well as establishment of guidelines for future
reference concerning this type of maintenance in the engineering practice.

KEYWORDS
Road bridge, Prestressed concrete beam, Field load test, CFRP strips, Static loads

1. INTRODUCTION
The paper is described the research method and the results of measurements and calculations of the static load test,
as well as the analysis of the obtained results of a five-span road bridge with the prestressed concrete main beams of
the Plonsk type reinforced by CFRP tapes and with a new concrete deck slab (Figures 1 and 2). The bridge over the
Nysa Klodzka River is situated in Paczkow (Upper Silesia Province – Poland).
Before the repair, the bridge load capacity determined in expertise was classified as class E, that is 150 kN according
to the Polish Loads Standard (PN-85/S-10030), mainly due to a very poor technical condition of the load-carrying
structure of all bridge spans, resulting mostly from transverse cracks in the main beams. The main aim of the repair
was to increase the object load capacity to class C (300 kN).
The aim of the tests was to determine the behavior of the road bridge spans and supports structure under known load
in order to prove the validity of the design assumptions made in the calculations and analyses of the spans and in the
load test program. It was to evaluate and determine the real rigidity and the load capacity of the bridge spans after
their repair, particularly the flexural rigidity of the strengthening main beams directly integrated with the deck slab
and the road surface.
The tests were directed to recognize the response behavior of the span structures of the old bridge under the
determined load in order to verify the assumptions made in the design (Schoenradt 1998) as well as to determine the
actual load capacity of the object (Manko and Mordak, 2002). It applied mainly to the assessment of the actual
stiffness of the prestressed beams, the deck plate and supports (Figure 2). The load tests were specifically intended
to investigate whether and to what extent span and deck elements interact, namely if there is any extra reserve of
load capacity in the bridge spans. It is manifested in smaller deflection and normal stress values in the main beams
and tapes, obtained from measurements carried out in the considered sections, in relation to calculated values
obtained on the same service load (PN-77/S-10040, PN-85/S-10030, PN-91/S-10042, PN-S-10040, 1999). Thus, the
assumptions concerning the statical scheme and the calculated model of the structure also were verified.
The main conclusions ascertain whether the bridge can be admitted to normal service and with what, if any, post-
construction recommendations. The conclusions from this research can be useful for the engineering practice, parti-
cularly in the area of the test load of strengthening bridges by FRP, because a number of identical spans were tested.

119
(a) (b)

Figure 1: General side view from the direction of the headwater of the road bridge: (a) localization of four
trucks on tested span II (load scheme IIS) and (b) bottom view on the beams strengthening by CFRP tapes

2. BRIEF DESCRIPTION OF BRIDGE STRUCTURE

The tests were carried out on the five-span road bridge. Each span consists of six main beams integrated with the
new reinforced concrete deck slab of B30 concrete class. The total width of the individual spans is constant along
the bridge length and it is 9.94 m (Figure 2). The total length of the bridge is 98.60 m. The effective length of each
of the spans is 18.00 m. The particular spans are simple-supported and made from Plonsk type prefabricated pre-
stressed concrete beams of length, L, of 18.50 m and are integrated with the RC deck slab over interior supports. The
bridge was designed to serve under the I class load (300 kN) in accordance with the PN-66/B-02015 (or C according
to the PN-85/S-10030). The span interior crossbeams are made as prestressed concrete. The tests covered spans I
and II (Figure 1). The bridge supports are in the form of massive concrete piers and abutments on spread foundation,
fixed in a reinforced concrete footing. The foundation rests directly on the virgin soil. The span main beams rest
always on single-roller and fixed steel bearings (Figure 1). The roadway was covered with bituminous pavement,
0.09 m thick, with incorporated insulation of an average thickness 0.01 m (Figure 2). The usable width of the bridge
amounts to 9.50 m, which includes the 7.00 m wide roadway and a 1.25 m sidewalk on each side.
The strengthening of the particular bridge spans was accomplished by gluing the CFRP tapes made of carbon fibers
(three for each beam) to the bottom flanges of the main beams. Additional by constructing a new reinforced concrete
deck plate of variable thickness 0.12–0.185 m as well as adding the outer stirrups in the form of steel flats of 5×50
mm section, with axial base by every 0.35 m (Manko and Mordak, 2002; Schoenradt 1998).
(a) CROSS SECTION (b)
BEFORE REPAIR AFTER REPAIR
9.92 9.94
0.21 1.25 3.50 3.50 1.25 0.22
Handrail Asphaltic concrete 0.05 m Abrasion layer asphaltic concrete 0.04 m 0.12
to disassemble Stone bricks 7/9 cm 0.08 m Lashing layer asphaltic concrete 0.05 m Aluminum
Gas =100 mm Sand-cement sub-crust 0.04 m Insulation 0.01 m handrail with
steel cable
(to disassemble) Protective concrete 0.04 m Reinforced concrete deck plate 0.12-0.185 m Elastic mass
Insulation 0.005 m Prefabricated beam PLONSK type to water stop
Asphalt 0,02 m Leveling concrete 0-0.07 m
Stone sidewalk curb 0.20x0.20 m Protection
1.10

Sidewalk slab Prefabricated beam PLONSK type pipe

0.82 0.14 0.25 1.00

thickness 0,28 m on water permability concrete

0.46 0.03

continuous footing = 110 mm

Concrete sidewalk curb 0.15x0.25 m Modified asphalt
2% (to disassemble) 2%
0.03 0.52

B1 B2 B3 B4 B5 B6 80 0.44
100 0.03
Gas
=100 mm
T1 1 T2 T3 T4 2 T5(T5') T6 T7 3 T8(T8') T9 4 T10 (T10') 5 T11 (T11') T12 6 T14
Water pipe Sika CarboDur M1214 T13(T13')
=100 mm Ti - strain gages i - main beams deflection and strain gages
0.46 0.35 0.10 0.60 0.90 0.60 0.90 0.60 0.90 0.60 0.90 0.60 0.90 0.60 0.10 0.35 0.47
9.92 9.94

Figure 2: Cross-section of spans and localization of measurement points: (a) before and (b) after, its repair

120
(a) Kamieniec Zabkowicki SIDE VIEW Paczkow
KAMAZ 5511 KAMAZ 5511
3. RESEARCH SCOPE AND APPROACH
Pf = 44.70 kN Pr = 2x73.40 kN Pf = 44.70 kN Pr = 2x73.40 kN

The static field load tests of the bridge included following

18.00 18.00
measurements (Figure 2):
(b) Span I Span II – six main beams deflections (load-carrying structure) at their
m 4( c 4) midspan by dial indicators with 1×10–5 m accuracy;
m5( c 5) TOP VIEW
m 6( c 6)
– vertical and horizontal displacements of the chosen
expansion and fixed bearings by dial indicators

2.500
1.850
1.980
with 1×10–5 m accuracy;
1.275 2.840 1.320 1.695
7.130
– strains (stresses) in the main beams, which were performed
m 4 ( c 4)
by strain gages (extensometers) and mechanical indicators;
m 5 ( c 5) – strains in the strengthening tapes in half and 1/4 of the
m 6 ( c 6)
SCHEME IS SCHEME IIS effective span of the main beams;
(c) 0 – strains in the outside stirrups in bearing zones and in
4.68
8.00

8.00
4.77

1
4.83
4.95

8.00

8.00
2 midspan, which were performed by strain gages;
5.08

8.10

4.88

8.10
3
– vertical displacements in selected intermediate supports and
5.14
8.10

4.88
4 8.10
5 abutments, and
4.88

8.20
8.20
5.16

6 – confirmation dimensions of each element of the span

4.90
8.20

8.20
5.16

7
8 structure taken during a general bridge inspection (the
[10-3m]
technical condition of the span and support structures was
(d) 0 SCHEME IN2 SCHEME IIN2
inspected prior to, during and after the finish of the testing.
3.65

5.20

3.53

5.20

1
The static load tests were conducted for two nonsymmetrical
4.35

4.08
6.50

6.50

2
4.69
5.04
7.70

7.70

3 and one symmetrical load schemes. All the measurements

4
specified by the test loading program were taken exclusively
5.45

5.17

5
8.90

8.90

6 under a static load (Manko and Mordak, 2002). Four trucks of

5.53
5.81

7 the KAMAZ 5511 type loaded by soil dump with the total
9.80

9.80

8
5.78
5.99

weight of 191,50 kN, were used in the tests (Figure 3).

10.70

10.70

9
10
[10-3m]
SCHEME IN1 SCHEME IIN1
(e) 0
1 4. TESTS RESULTS AND ANALYSIS
0.99

0.60

0.60
1.06

2
3
1.54

2.00

2.00
2.211.58

4
The final results of bridge acceptance inspection (PN-S-10040,
2.15

3.40

3.40

5 1999), conducted after the complete repair under the trial static
2.81

4.80

4.80
2.79

6 and dynamic load, allowed for a comprehensive evaluation of

3.33

6.20

6.20
3.35
3.81

7
3.88

7.50

7.50

[10-3m] the efficiency of the main beams strengthening by applying

Number of measurement points (gages):
measured values (right side) calculated values (right side) CFRP tapes (Figure 3). The maximal total main beams deflec-
m1 m6 c1 c6 tions measured during the research under static load, oscillate
m2 m5 c2 c5
m3 m4 c3 c4
between 1.37×10–3 m and 5.99×10–3 m (Table 1). The average
deflection values of span cross-sections were smaller than calcu-
Figure 3: Prestressed concrete road bridge and lated ones in all cases and did not exceed permissible values
load cases distribution on bridge during field (δperm = L / 800 = 18.00 / 800 = 22.50×10–3 m). For the possible
test: (a) side view from headwater, (b) top view three load schemes to execute after the repair works average
– technical parameters of truck type KAMAZ measured correspond to calculated elastic deflection relations
5511, and displacements of all tested main were as follow: IS – 0.629, IIS – 0.598, IN2 – 0.560, IIN2 –
beams of the bridge on span length obtained 0.540, IN1 – 0.517, IIN1 – 0.508. The measured main beam
from measurement and calculation for load deflection distributions in span cross-sections were different
schemes: (c) IS & IIS, (d) IN2 & IIN2 and than calculated ones. It is proved that a real torsional stiffness
(e) IN1 & IIN1 of the bridge was higher than assumed in the static calculations.
The measured permanent deflections of the particular main beams differ inconsiderably among each other and are
not proportional to elastic deflections as well as they do not exceed permissible values (less than 20% of elastic
deflections or strains). The relations of the main beams permanent deflections to their total deflections and strains
were established based on diagrams shown in report by Manko and Mordak (2002). The differences between the
initial and the final readings found in measurements were approximately the same in all spans and examined main
beams cross-sections. It may prove that they were rather a result of new bearings settlements or under bearing joints
and possible insignificant reading errors of the measurement instruments, and only to a minimal rate of the
permanent strains of the prestressed main beams. This might have resulted in the occurrence of some insignificant
settlements and differences in readings under proportionally heavy static load.

121
Table 1: Results of total δm and εm (upper row) and calculated δc and εc (bottom row) deflections and strains
of main beams of studied spans at the midspan for all load schemes by measurement and calculation

Deflections δm and δc in (10–3 m) Strains εm and εc in (10–6)

Span Load scheme
number number Main beam number
B1 B2 B3 B4 B5 B6 B1 B2 B3 B4 B5 B6
IS measured 4,77 4,95 5,08 5,16 5,14 5,16 315 350 310 340 190 185
calculated 5,04 5,47 5,74 5,77 5,55 5,18 347 334 324 326 337 351
measured 3,81 3,55 2,79 2,21 1,58 1,06 180 130 160 175 10 100
I IN1
calculated 5,33 4,54 3,52 2,26 1,00 0,17 211 163 138 115 91 77
measured 5,99 5,81 5,45 5,04 4,35 3,65 455 410 325 300 160 120
IN2
calculated 6,93 7,26 7,06 5,93 4,14 2,08 401 377 355 333 308 266
IIS measured 4,68 4,83 4,90 4,88 4,88 4,88 230 200 345 370 115 335
calculated 5,04 5,47 5,74 5,77 5,55 5,18 347 334 324 326 337 351
II IIN1 measured 3,65 3,19 2,63 2,06 1,37 0,88 180 120 160 150 20 75
calculated 5,33 4,54 3,52 2,26 1,00 0,17 211 163 138 115 91 77
IIN2 measured 5,78 5,53 5,17 4,69 4,08 3,53 315 270 355 355 100 290
calculated 6,93 7,26 7,06 5,93 4,14 2,08 401 377 355 333 308 266

5. CONCLUSIONS
The practical experience gained from the testing of the road bridge and the observations concerning the behavior of
the structure of post-tensioned prestressed concrete beams of the Plonsk type made during the tests, as well as the
analysis of the results obtained from the measurement and their comparison with the calculated values leads to
specific conclusions as to the real behavior of the structure of the individual spans of the bridge and allows one to
form the following general conclusions (Manko and Mordak, 2002):
1. In the light of the tests that were carried out, the structures of the bridge span and supports did not raise any doubts.
The deflections of the main beams are principally elastic. The minimal permanent displacements that were
discovered are in part residual deflections of the main beams and partly caused by the bearings settlement of piers.
2. The measured deflections and average strains values were considerably lower than the expected (calculated) ones
and the permissible ones for all the tested spans. This indicates higher rigidities of the span structures – probably
due to better interacting and integration with the RC deck slab and the road surface. This applies to all the main
beams in the tested span. In all, five almost identical spans made up of Plonsk type beams have been analyzed
and therefore, the conclusions as to the behavior of such typical structures can have major practical importance.
3. On the basis of the static tests that were carried out permission could be given for the bridge to be subjected to
dynamic tests during the entrance and the exit of a dynamic load and then the bridge could be admitted to normal
service. A close inspection of the span and supports of the bridge as well as the supplementary and verifying
measurement did not reveal any damage to the structural components or their joints in the considered spans.
4. Normal stresses in tapes (which were calculated on the basis of strain values) show unambiguously that their load-
capacity was used only in about 5–10%. However, no strains were noticed in outer stirrups, but the small strain
gages movements are contained in a range of reading accuracy or measuring device errors. A job that can be
ascribed them, could be connected with the protection against to the delamination of tapes (together with a lag of
reinforcement bars), what was often happened in the strengthening solutions (Manko and Mordak, 2002).
In the fact, above summary and conclusions refer to the structures of the tested span elements of preset geometric
characteristics, particular element stiffness, and determined effective spans. However, it may be stated that spans
strengthening constructed by lamels and steel outer stirrups is not the best solution as far as this type of structures
are concerned, mostly from the economical point of view. In order to use expensive CFRP tapes to a higher extent,
one should install on the beams already known prestressing devices for CFRP tapes.

4. REFERENCES
Manko, Z., and Mordak, A. (2002). “Submitting Project and Report on Static and Dynamic Trial Load of Prestressed Road
Bridge over Nysa Klodzka River along National Road No. 382 Stanowice – Paczkow in km 72.536 in Paczkow after its Repair.”
Scientific-Research Center for the Development of Bridge Industry MOSTAR, Wroclaw, Poland.
Schoenradt, Z. (1998). “Technical Design of Bridge Repair over Nysa Kłodzka river along National Road No. 382 Stanowice –
Paczkow in Paczkow in km 72.536.” Technical-Engineering Office KARO, Poznan, Poland.

122
Third International Conference on FRP Composites in Civil Engineering (CICE 2006)
December 13-15 2006, Miami, Florida, USA

Blast mitigation of a concrete arch bridge using composites

Achille Devitofranceschi1, Domenico Asprone2, Andrea Prota2, Renato Parretti2 and Antonio Nanni2,3
1
ANAS S.p.A., Direzione Centrale Progettazione, Rome, Italy
2
Department of Structural Analysis and Design, University of Naples “Federico II”, Naples, Italy
3
Department of Civil, Architectural and Environmental Engineering, University of Miami, Coral Gables, FL, USA

ABSTRACT
Blast mitigation has recently become a critical need for public infrastructure because the probability of explosions is
no longer linked to accidental reasons only. In the spring of 2005, an international project regarding blast mitigation
started under the coordination of the University of Missouri – Rolla and the participation of the University of Naples
“Federico II” and two Italian firms, ANAS, owner of the Italian highway network, and AMRA, a research center of
excellence. The objective of the project was to investigate a real structure under blast loading conditions; the team
had the opportunity to study and test to collapse a recently dismissed long-span reinforced concrete arch bridge in
southern Italy.
The first phase of the project has been focused on a detailed structural analysis of the bridge performed via a finite
element modeling of the structure anticipated by static/dynamic materials characterization. The conducted numerical
modal analysis was validated by in-situ vibrodyne tests. This allowed the calibration of the model later used for blast
loading analysis.
At this time, two mitigation techniques using composites have been proposed to protect the piers of the bridge.
These solutions will be modeled and tested on the structure under blast loads.

KEYWORDS
Blast mitigation, bridge, composites, FRP, pier, reinforced concrete, SRP.

1 INTRODUCTION
The Tenza Viaduct (Figure 1), located in southern Italy, was built in the sixties as part of the Salerno-Reggio
Calabria Highway and was open to traffic untill two years ago. The highway owner ANAS had planned to change
the geometry of the route, since it no longer respects current safety requirements. Accordingly, the bridge was closed
to traffic as it belonged to a substituted portion of the highway.
The Tenza Viaduct consists of three different structures, one main span and two approaching spans. The main span
is an open spandrel arch structure that is 120 m long and 50 m deep. The bridge deck and its wall piers are supported
by a ribbed, solid slab, and fixed-fixed arch. Each 30 m long approaching span is supported by multiple wall piers of
different heights. Each individual pier is made of two reinforced concrete (RC) columns connected over their entire
height by a RC wall.
The Viaduct was built during the economic growth after the Second World War and although no record exists on the
method used for its construction, wood formworks were probably used, as that was a very common practice at that
time. In the eighteens, the bridge was rehabilitated to meet seismic code requirements. The retrofitting consisted of
RC encasement of both piers and arch cross-sections (Nanni et al. 2006).

2 TRADITIONAL BRIDGE ASSESSMENT AND CHARACTERIZATION

The objective of the first phase of the project was to obtain a detailed knowledge of the structure and its behavior. A
traditional investigation was performed using: a) static and dynamic characterization of materials; b) a modal

123
analysis; and c) field validation through vibrodyne testing. Static analysis under gravity and seismic loads was
conducted on a FEM model of the structure.

Figure 1: The Tenza Viaduct

2.1 Material characterization

The objective of this campaign was to obtain complete quasi static and dynamic characterization of the materials of
the Tenza Viaduct.
First of all, four types of materials were identified:
• The original concrete, used in the sixties when the bridge was built;
• The concrete used for piers and arches rehabilitation;
• The original steel, lightly ribbed, used as reinforcement of the original concrete;
• The modern ribbed steel bars used as reinforcement of the rehabilitated concrete.
For each of these materials several specimens were taken from the structure. Tensile tests of steel specimens as well
as compression and ultrasonic tests of concrete specimens were conducted. The collection of the specimens intended
to be representative of the whole structure: arch, deck, piers and girders.
The original concrete had an average cylindrical strength of 34 MPa for the arch and 46 MPa for the deck; the
rehabilitated concrete had an average cylindrical strength of 31 MPa. Ultrasonic tests were performed before
conducting the concrete compressive tests and results were related to the cylindrical strength.
For the steel specimens standard tensile tests were performed and stress-strain curves collected recording yielding
strengths of 400 and 490 MPa and ultimate strengths of 593 MPa for original and rehabilitated specimens,
respectively.
Moreover, in order to obtain a mechanical characterization of concrete and steel under high dynamic loading rates,
tests were performed in the DynaMat Laboratory of the University of Applied of Southern Switzerland using three
modified Hopkinson Bars (Cadoni et al., 2000). Stress – strain relationships under different strain rate were
evaluated under tensile loads. The strain rate ranged between 10-4 and 10+1 s-1 for the concrete and between 10-4 and
10+2 s-1 for the reinforcing steel. The results show that:
• At high strain-rate, the new concrete seems to have a better behavior than the original one. An increment of tensile
strength varying from 100% to 400% was found. This implies also an increase of pseudo-ductility and energy
absorption capability with increasing strain-rate.
• The reinforcing steel is also strain-rate sensitive both in terms of stress and strain. As an example, the yielding
stress grows more than 60% from the quasi-static regime to the impact one (>500 s-1), while the failure strength
increases its value less than 20%. Ultimate strain and failure strain have an increment of 50%.
The results provide information about the influence of dynamic loads on the constitutive behavior of materials aged
in an existing structure and they will be used in structural analyses accounting for strain rate effects.

2.2 FEM model

Considering that the bridge is substantially made of three different structures (central arch and approaching spans),
three different numerical models were elaborated. The geometry of the models was obtained from the original
drawings of the structure and verified with a 3D laser scanner survey (Figure 2). The values of the elastic properties
of the structure were based on the results of the material characterization tests.

124
 Figure 2: Bridge resulting from the application of the laser scanner

The three models were built using SAP2000 and Straus7 software (Figure 3) (Computers and Structures, Inc., 2004;
G+D Computing, 1999). For each of them static and modal analyses were performed. The results show that the first
two vibration periods of the central structure are longitudinal (1.08 Hz) and transversal (1.45 Hz), while the first
vibration periods of the approaching spans correspond to local modes.

Figure 3: FEM models for arch bridge and approaching spans

The numerical analytical results were verified through a field vibration test conducted using both wind action and a
mechanical vibrodyne giving harmonic excitations. Accelerometers distributed on the structure recorded the data
obtained during the test. Frequency and LMS (Least Mean Square) analyses of the results confirmed a transversal
mode of vibration with a frequency of 1.35 Hz, within 10 per cent difference of the predicted value.

3 MITIGATION TECHNIQUES
The mitigation of blast action on the bridge has been focused on the piers since they appear, in the hierarchy of the
structure, the most critical elements. The typical pier has a “I” cross section (Figure 4) with a 300 mm web
thickness, a 500 mm flanges thickness, and a 600 to 1,000 mm variable width; the overall height is 3 m. The planned
mitigation technique of the bridge can not avoid modifying the shape of the cross section to obtain more
compactness to resist blast loading. In fact, the “I” cross section appears vulnerable to shock actions due to local
weakness. The planned mitigation technique will provide the filling of the section with light-weight shrinkage-free
concrete to obtain a rectangular shape. The filler will also have a sacrificial protection for the web.
500

Existing RC
variable from 600 to 1,000 mm

Light-weight
concrete filler

300

rounding of
corners

Figure 4: Pier cross section

125
After rectangular cross section has been realized and the corners are rounded, FRP reinforcement is applied to the
contour as externally bonded system due to the promising results displayed by such strengthening technique for blast
mitigation (Lawver et al., 2003; Lu et al., 2005).
A first technique calls for the use of Carbon Fiber Reinforced Polymer (CFRP) fabrics, which has shown good
results in the literature (Muszynski and Purcell, 2003; Hegemier et al., 2006). This solution also provides an increase
to the seismic performance of the structure.
A second technique consists in the application of a multilayered material system made by foam (metallic or
polymeric) sandwiched by a Steel Reinforced Polymer (SRP) system (Hardwire LLC, 2002) (Figure 5). In this case,
the jacket has also an energy dissipation function that has shown interesting potential in preliminary numerical
simulations. The possibility of using this technique for noise mitigation of bridge decks is of interest and currently
under consideration.
More detailed information on the mitigation strategy will be orally provided at the conference.

Foam
SRP

Figure 5: Multilayer technique

4 CONCLUSIONS
The paper presented the work conducted to study the possibility of blast mitigation with composites of a highway
bridge. Blast loading is becoming of relevant importance in design procedures of highway infrastructures and
composites show interesting characteristics in this field. A traditional analysis of the structure was performed and
two different solutions were proposed. Numerical modeling will be used for prediction and blast testing will be
conducted to obtain validation and further ideas for blast mitigation design of existing structures.

5 AKNOWLEDGMENTS
The authors are grateful to Technical Support Working Group (TSWG) for funding support to this project under
Contract Number N4175-05-R-4828.

6 REFERENCES
Cadoni E. et al. (2000) High strain-rate tensile concrete behaviour, Magazine of Concrete Research, 52, No.5, Oct.,
365-370.
Computers and Structures, Inc. “CSI Anal y sis Reference Manual For SAP2000®, ETABS®, and SAFE™”
Berkeley, Califor nia, USA September 2004
G+D Computing, “THEORETICAL MANUAL – Straus7”, Sydney, Australia, 1999
Hegemier, G.A., Seible, F., Rodriguez-Nikl, T., Arnett, K. (2006), “Blast Mitigation of Critical Infrastructure
Components and Systems” Fédération Internationale du Béton, Proceedings of the 2nd International Congress, June
5-8, 2006 – Naples, Italy
Lawver, D., Daddazio, R., Oh, G.J., Lee, C.K.B., Pifko, A.B., Stanley, M. (2003), "Composite Retrofits of
Reinforced Concrete Slabs to Resist Blast Loading," 74th Shock and Vibration Symposium, San Diego, CA.
Lu, B., Silva, P., Nanni, A., Baird, J. (2005), "Retrofit for Blast-Resistant RC Slabs with Composite Materials," 7th
International Symposium on Fiber-Reinforced Polymer Reinforcement for Concrete Structures, FRPRCS-7, Kansas
City, MO, pp. 1345-1359.
Muszynski, L.C., Purcell, M.R. (2003), "Composite Reinforcement to Strengthen Existing Concrete. Structures
against Air Blast," Journal of Composites for Construction, ASCE, Vol. 7, No. 2, pp. 93-97.
Nanni A, Asprone D, Ayoub A, Baird J, Filangieri A, Galati N, Kiger S, Marianos W, Prota A, Quintero R, Wang
M, and Wei J. “Blast Testing and Research-Bridge at the Tenza Viaduct”. Final Report Task 1 of TSWG Contract
Number N4175-05-R-4828, University of Missouri-Rolla, Rolla, MO, U.S.A., 2006.
Hardwire LLC, “What is Hardwire”, www.hardwirellc.com, Pocomoke City, Maryland, 2002.

126
Third International Conference on FRP Composites in Civil Engineering (CICE 2006)
December 13-15 2006, Miami, Florida, USA

PERFORMANCE OF FRP RETROFITTED BRIDGES

UNDER BLAST LOADING
Rui Zheng
(Ph. D Student, Florida International University, Miami, FL, USA)

Amir Mirmiran
(Professor and Chair, Florida International University, Miami, FL, USA)

ABSTRACT
The use of FRP system to harden the bridges under blast loading has been suggested by a number of studies. This
paper considers the material characteristics of concrete, steel and FRP at low and high strain rates, and proposes an
analytical technique to convert the three dimensional time-consuming blast loading analysis into a unidirectional
study, maintaining high level of fidelity and accuracy. The study shows the benefits of the use of FRP for hardening
bridge substructures.

KEYWORDS
FRP, Blast Loading, Dynamic Analysis, Repair and Retrofit, Hardening.

1. INTRODUCTION
Stay-in-place fiber reinforced polymer (FRP) forms have been used for columns, piles, and girders in recent years.
The use of FRP composite laminates for retrofitting concrete structural members has been shown to be very
successful in restoring or increasing the strength of the members. However, there is little research on the behavior of
FPP-confined columns under blast loadings.

The response of a structure subjected to blast loading is complicated and depends on the mass, geometry, and
material properties of the structure, as well as the pressure magnitude of the impact loading. The materials of
concrete, steel and FRP have different properties in strength and ductility when the loading rate changes. This paper
considered the material characteristics of concrete, steel and FRP at low and high strain rates.

At present, most analytical work on blast loading is focused on 3-D finite element analysis with very complicated
modeling schemes, accuracy of which depend significantly on the fineness of the mesh, and with significantly high
cost of computation. This paper presents the details of an ongoing study on the behavior of FRP-confined columns
under blast loading using an equivalent unidirectional analysis with high level of fidelity and accuracy and low
computing cost.

2. MATERIAL MODELLING
2.1 Material Modeling at Low Strain Rate

The steel-confined concrete core is modeled after Mander et al. (1988), with an energy-based failure criterion, which
equates the energy absorbed in concrete core with the maximum energy stored in the spiral steel reinforcement. The
FRP-confined concrete core is modeled after Samaan et al. (1998), with a bi-linear empirical equation and a concrete
failure criterion based on the FRP tube rupture in the lateral direction. Both models are cast into the modified Kent-
Park model (Scott et al. 1982), which is characterized with a tri-linear curve to fit the concrete stress-strain

127
responses of the two models. Tensile strength of concrete is neglected for simplicity. The typical hysteretic stress-
strain response of concrete core is shown in Fig. 1(a).

The steel reinforcement is simulated as an elasto-plastic material with a tri-linear hysteretic model, as shown in
Figure 1(b). Based on earlier coupon tests (Shao 2004 and Zhu 2004), the stress-strain curve for the FRP tube in the
longitudinal direction is defined as below, and shown in Figure 1(c):

(
σ = (ε / ABS (ε )) × − 21 .2 + 451 + 638,143 × ABS (ε ) ) ABS (ε ) ≤ 0.05 [in MPa] (1)

(
σ = (ε / ABS (ε ))× − 3.08 + 9.49 + 13,423 × ABS (ε ) ) ABS (ε ) ≤ 0.05 [in ksi] (2)

5 500 200
Compression Tension
Compression Tension
0 400

-5 150
300

F R P T u b e A x ia l S tr e s s ( M P a )
C o n c r e te S tr e s s ( M P a )

-10
S te e l R e b a r S tre s s (M P a )
200
100
-15
100

-20
0 50
-25
-100
-30 0
-200
-35
-300
-50
-40
-400
-45
-0.016 -0.014 -0.012 -0.010 -0.008 -0.006 -0.004 -0.002 0.000 0.002 0.004 -500 -100
-0.02 -0.01 0.00 0.01 0.02 0.03 0.04 0.05 -0.03 -0.02 -0.01 0.00 0.01 0.02 0.03 0.04 0.05 0.06
Concrete Strain (mm/mm)
Steel Rebar Strain (mm/mm) FRP Tube Axial Strain (mm/mm)

(a) (b) (c)

Figure 1: Typical Hysteretic Stress-Strain Curve of Model for
(a) Concrete, (b) Steel and (c) FRP

2.2 Material Modeling at High Strain Rate

Concrete, steel and FRP generally show higher strength and ductility at higher loading rates. For the unconfined and
steel-confined concrete, Scott et al. (1982) reported that both the peak stress and the post-peak descending slope
were affected by the strain rate. The dynamic effect is introduced to the concrete model with dynamic increase factor
Kd. The study recommended a constant value of 1.25.

There are few reports regarding the FRP-confined concrete performance under high strain rate. Zhang (2003) stated
that the strength for FRP confined concrete is a function of strain rate, as shown in the following equation.

f c'd = f c ' ( 1 + α lg ε& / ε& o ) (3)

'd '
where f c is the dynamic loaded confined concrete strength; f c is confined concrete strength under low strain rate;
ε& is the strain loading rate; ε&o is the standard strain loading rate (10-5 1/s); α is sensitivity index of concrete
strength to loading rate, which is a function of confining ratio ξ. Confining ratio is defined as:
σ f ⋅t
ξ = (4)
fc ⋅ r
where σ f is the FRP rupture stress; t is the external FRP thickness; f c is unconfined concrete strength under low
strain rate; and r is concrete cylinder radius.

The concrete model is therefore modified for the high rate of loading. For this study, the loading rate is at the scale
of 10-5 1/s. The FRP-confined concrete model under high and low strain rates are plotted in Figure 2(a).

128
Steel also shows higher strength and ductility under high strain rate. Soroushian and Choi (1987) proposed a steel
model considering various loading rates. The following equation showed that the increment of steel yielding stress is
proportional to the logarithm function of the strain rate.
f y'
fy
(
= − 0 . 451 × 10 −6
) (
f y + 1 . 46 + − 9 . 20 × 10 −7
f y + 0 . 0927 )log ε&
10
(in MPa) (5)

The ultimate capacity and ultimate strain of the steel material follow the same trend, as follows:
f p'
fp
(
= − 6 . 83 × 10 − 6 f y + 1 . 72 + − 1 . 37 × 10 ) ( −6
)
f y + 0 . 144 log ε&
10 (in MPa) (6)

ε u'
= (− 8 . 93 × 10 −6
) (
f y + 1 . 4 + − 1 . 79 × 10 −6
f y + 0 . 0827 )log ε&
(in MPa) (7)
εu 10

In this study, steel material is modeled with elastic material and linear strain hardening up to failure. The steel model
at low and high strain rates are plotted in Figure 2(b).
c o n c re te s tre s s

low strain rate

high strain rate

concrete strain

(a) FRP Confined Concrete (b) Steel Reinforcement

Figure 2: FRP Confined Concrete Stress-Strain Curve under Various Strain

Rates for (a) FRP Confined Concrete and (b) Steel Reinforcement

For FRP materials under high strain rate, no literature has been found to provide a definite statement. However,
Kuksenko and Tamuzs (1981) reported that the FRP material does not have significant strength increase under high
loading rate. Therefore, the FRP model did not include the dynamic effect in this analysis.

3. SIMPLIFIED BLAST ANALYSIS WITH HIGH FIDELITY

Both the conventional reinforced concrete (RC) column and the concrete-filled FRP tube sections are discretized
into layers. For each section, material models are adopted from the constitutive material models under high strain
rate. Each section is analyzed separately to generate its own moment-curvature response.

In order to simplify the blast analysis with high fidelity and no loss of accuracy, this study incorporates the concept
of equivalent section with pseudo materials. The equivalent section was proposed to transform the existing section
geometry into an I-shaped section, where the web thickness is considered to be negligible (Figure 3), and the entire
section is lumped into two flanges with limited thickness, thus limited stress variation along the flange thickness.
Since the material is only present at the two tips of the section, there is no strain or stress gradient in the section.
With the concept of equivalent section, the section performance is fingerprinted by the pseudo material defined at
the two flanges. In order to produce exactly the same sectional moment-curvature response, the stress-strain curve of
the pseudo material is defined based on the moment-curvature response of the actual cross section.

129
The blast analysis is then conducted on a line element along the column with the pseudo materials using ANSYS®.
The study takes significantly less time than the finite element analysis of a complete three-dimensional mesh.

Discretized Section Equivalent Section

Pseudo Material

Identical Sectional Response

Figure 3: Illustration of the Equivalent Section Concept

4. Blast Analysis
Figure 4 shows the blast loading chosen for this study. A full blast
analysis was carried out to compare the performance of pier
frames made of either RC or CFFT columns. The study generated
a number of results in terms of stresses and deflections of the two
systems, as well as the safe distance from the blast loading for the
bridge substructure. The study did not include any blast pressure
on the bridge deck as it was assumed that the blast pressure is
exerted directly on the bridge column. The study showed superior
performance of the CFFT columns over equivalent RC columns. It
also verified that the simplified method of analysis works quite
well for blast loading.
Figure 4 : ATF Vehicle Bomb Size Matrix

5. REFERENCE
Kuksenko, V.S., and Tamuzs, V.P. (1981). “Fracture Micromechanics of Polymer Materials.” Springer, Hingham,
Ma.
Mander, J., Priestley, M., and Park, R. (1988). “Theoretical Stress-Strain Model for Confined Concrete”. Journal of
Structural Engineering, ASCE, Vol. 117, No. 8, 1988, pp. 1804-1826.
Samaan, M., Mirmiran, A., and Shahawy, M. (1998). “Model of Concrete Confined by Fiber Composites”. Journal
of Structural Engineering, ASCE, Vol. 124, No.9, 1998, pp. 1025-1031.
Scott, B.D., Park, R., and Priestley, M. (1982). “Stress-Strain Behavior of Concrete Confined by Overlapping Hoops
at Low and High Strain Rates.” ACI Journal, Vol. 79, No.1, pp. 13-27.
Shao, Y., and Mirmiran, A. (2004). “Nonlinear Cyclic Response of Laminated Glass FRP Tubes Filled with
Concrete”. Composite Structures, Elsevier Science Ltd., Vol. 65, No. 1, 2004, pp. 91-101.
Soroushian, P., and Choi, K. (1987). “Steel mechanical properties at different strain rates.” Journal of Structural
Engineering, ASCE, Vol. 113, No. 4, pp.663-672.
Zhang, B., Pan, J., and Jiang, H. (2003) “Computation of strength for FRP confined concrete under fast loading.”
Journal of Harbin Institute of Technology, Vol. 35, No. 8, pp.958-961.
Zhu, Z. (2004). “Joint Construction and Seismic Performance of Concrete-Filled Fiber Reinforced Polymer Tubes”.
Ph.D. Dissertation, North Carolina State University, Raleigh, N.C., 2004.

130
Third International Conference on FRP Composites in Civil Engineering (CICE 2006)
December 13-15 2006, Miami, Florida, USA

Design and Analysis a 10-m FRP Deployable Bridge

R.G. Wight, MA. Erki, C.T. Shyu, R. Tanovic, A. Xie
Royal Military College of Canada, PO Box 17000, Station Forces
Kingston, Ontario, Canada K7K 7B4

ABSTRACT
The Canadian Forces are replacing many of their tracked vehicles with wheeled vehicles in a move towards a
“primarily wheeled Army”. However, there is a loss of cross-country mobility associated with this change, which
creates a need for lightweight, inexpensive, short span bridges. Owing to its favourable properties, FRP was the
target material for the construction of these bridges. A structural concept, adapted to the material properties of
commercially available glass FRP (GFRP) pultruded products, is presented. A 10 m span full-scale box-beam was
built from commercially available GFRP pultruded sections that were bonded throughout to form a tapered box-
beam, with a width of 1.22 m and a height at mid-span of 0.9 m, plus 0.04 m (1.575 in.) for a ribbed GFRP panel for
the wearing surface. The box-beam represents a single track of a double-track bridge, whose tracks are to be lightly
laced together. The design considered the concepts for modular assembly, methods of transporting, launching and
recovering of the structure. The design of a hinge at mid-span to accommodate these features was a major focus.
Finite element analysis of the design was used to predict deflection behaviour.

KEYWORDS
Pultruded structural shapes, short span bridge, fibreglass

1. INTRODUCTION
Both the Canadian Forces (CF) and a wide range of Canadian and international non-government organizations
(NGOs) have an intense interest in cross-country mobility in settings where infrastructure may have been badly
damaged by conflict or natural disaster. The ability to deliver humanitarian aid so as to maintain order and civil
society is hugely dependent upon timely restoration of mobility and access. Hence, there exists a need for
lightweight, inexpensive, short-span bridges. Owing to its favourable properties, such as high strength to weight
ratio, corrosion-free characteristics, high degree of free formability and good fatigue resistance, fibre reinforced
polymer (FRP), and in particular glass FRP (GFRP), is the target material for the construction of such bridges.

The concrete/FRP structural research group at the Royal Military College of Canada has developed an expertise in
the use of GFRP for short span bridges and their related design issues since 1994 (Erki et al., 1994; Erki et al., 1995;
Yantha et al, 1995; Erki et al., 1997; Tanovic et al., 1998). With a view to designing the 10-m full-scale GFRP
bridge, a 4.8 m span prototype gap crossing aid was first built using commercially available GFRP pultruded
sections that were bonded throughout to form a tapered box-beam, with a width of 1.2 m and a height at mid-span of
0.5 m, plus a 0.04 m (1.575 in.) thick ribbed GFRP panel for the wearing surface (Wight et al., 2004a; Wight et al.
2004b; Heffernan et al. 2004, Wight et al. 2005). The goal was roll-on/roll-off capability, with minimal terrain
preparation for deployment of the bridge. Experience with the design, construction and testing of the 4.8 m
prototype tapered box-beam confirmed that a gap-crossing aid consisting of built-up box beam components would
fulfill the need for transportable, lightweight, short span bridges.

2. DESCRIPTION OF THE BRIDGE

A 10-m span full-scale box-beam was designed and built using commercially available GFRP pultruded sections
that were bonded throughout to form a tapered box-beam, with a width of 1.22 m and a height at mid-span of 0.9 m,
plus 0.04 m (1.575 in.) for a ribbed GFRP panel for the wearing surface. The box-beam represents a single track of a

131
double-track bridge, whose tracks are to be lightly laced together. As for the 4.8 m prototype, the low rigidity of
GFRP was compensated for by maximizing the moment of inertia at mid-span for the 10-m gap crossing aid.
However, unlike the 4.8 m prototype, the 10-m gap crossing aid has two slopes on either side of the mid-span,
because with only one slope the large change in angle at mid-span would have been impractical for vehicles. At the
quarter-span the height of the box-beam is 0.75 m, and at mid-span it is 0.9 m. The 10-m gap crossing aid consists of
three longitudinal beams in parallel, two for the outer faces and one on the centreline of the box. The top and bottom
flanges of the beams consist of a pair of back-to-back hollow tube sections (50.8 × 50.8 × 6.4 mm), spaced by a
6.4 mm thick flat plate section for the web. The three beams are tied together at their tops and bases by two flat
plates that make up, respectively, the top and bottom plates of the box-beam. A ribbed decking is further bonded to
the top plate to stiffen it and to provide a wearing surface for the vehicle. The full structure would consist of two of
these box-beams for the two wheel tracks of the vehicle. The two box-beams would be lightly laced together with
bolted GFRP sections. The global dimensions are shown in Figure 1.
North

MM load case

Figure 1: Tapered beam single trackway of the 10-m FRP deployable bridge

The bridge was designed for the Military Load Class (MLC) 30, which in accordance with NATO specifications
represents two standardized vehicles, namely a tracked vehicle of 27.22 Tonnes and a wheeled vehicle of
30.84 Tonnes. The design of the 10-m box-beam considered concepts for modular assembly, methods of
transporting, launching and recovering of the structure. The design of a hinge at mid-span to accommodate these
features was a major focus. Several hinge prototypes, including hybrid FRP/steel hinges, were designed, constructed
and tested, before arriving at a proposed solution. Finite element analysis of the design was used to finalize the
global design details of the structure. The bridge will be tested in the laboratory, under quasi-static and dynamic
loading, and in the field using a moving vehicle. This paper presents design considerations, including the design of
the mid-span hinge, finite element analyses for the static loading, and the construction sequence used to build the
bridge. Test results will be presented at the conference.

The GFRP materials used are all pultruded sections that were purchased from Creative Pultrusions Inc. The
materials were from the 1625 series for the rectangular tubes, the 1625 series for the flat sheets and Flowgrip
Flooring for the decking/wearing surface. The hollow tubes and plates consist of alternating layers of randomly
oriented fibres and unidirectional fibre rovings in a vinyl ester resin matrix for the hollow tubes and an isophthalic
polyester resin matrix for the flat plate. The amount of glass fibre by weight for these sections is between 45 and
50%. The material properties of the GFRP components are shown in Table 1.

To allow ease of transportation and launching, the 10-m gap crossing aid was designed to be in two halves, joined
by a hinge attached to the bottom plate at mid-span. The hinge will be predominately in tension. Two main systems
were considered, namely a hinge fabricated using carbon FRP (CFRP) sheets wound around a steel hinge pin and an
all-steel hinge. The CFRP option required that the CFRP sheets be bonded to the bottom plates of the box-beam. The
all-steel hinge was also bonded to the bottom plate, but with the option of lightly bolting the steel plates to the
bottom plate of the box-beam to prevent premature failure due to prying. It is expected that prying will occur due to
unavoidable eccentricities, in spite of measures taken to reduce these. Also, in the all-steel hinge, consideration was
given to whether the center of the steel pin would be placed concentrically or eccentrically with respect to the

132
thickness of the bottom plate of the box-beam. The system with the pin positioned below the bottom plate of the
box-beam would allow less encumbered folding of the box-beam.

Table 1. Material properties of the GFRP components

Component Property (coupon values) Units Values

Modulus of elasticity GPa 22
Tube
Poisson’s ratio - 0.35
Density Mg/m3 1.90
Modulus, of elasticity (LW*) GPa 12.4
Modulus of elasticity (CW*) GPa 6.9
Plate
Poisson’s ratio (LW) - 0.32
Poisson’s ratio (CW) - 0.28
Density Mg/m3 1.90
Modulus, of elasticity (LW*) GPa 20.7
Modulus of elasticity (CW*) GPa 10.35
Deck Poisson’s ratio (LW) - 0.32
Poisson’s ratio (CW) - 0.28
Density kg/m2 14.65
*LW: lengthwise parallel to the unidirectional fibre rovings;
CW: crosswise perpendicular to the unidirectional fibre rovings

Ultimately, even though the CFRP hinge would be a lightweight option, it was not favoured, because it had lower
failure strength than the all-steel hinge. This lower strength was a result of premature debonding failure of the CFRP
sheets from the FRP substrate. Also, it is expected that the impact and dynamic loading of the hinge in service
would exacerbate the debonding problem. Uncertainty regarding this could not be adequately addressed. Also, there
was uncertainty regarding quality control during fabrication of the CFRP hinge, protection of the hinge from damage
in service, and reparability of any damages incurred by the CFRP hinge in service.

For the all-steel hinge, the steel pin was fitted into a steel tube that was welded to the steel plates of the hinge. In
tests, welding failure never governed the maximum load obtained. Bonding was used to attach the steel plates of the
hinge to the FRP substrate, representing the bottom plate of the FRP box-beam. Pliogrip adhesive with a metal
primer worked well in this application. General observations were that for the all-steel hinge, the geometric
eccentricity of the steel pin (the centre of the pin placed below the bottom plate) greatly reduced the maximum
strength obtained for the hinge, owing to the prying resulting in debonding of the steel plates from the FRP substrate
at low loads. The hinge design chosen for the 10-m gap crossing aid is the all-steel hinge with the pin placed as
nearly as possible concentrically with the mid-depth of the bottom plate. This design resulted in a hinge strength that
was at least 20% greater than provided by the box-beam design. Construction sequence of the 10-m gap crossing aid
consisted of (a) assembly of one of the “half-webs” for half of the box-beam; (b) assembly of the three half-webs
with internal diaphragms; (c) bonding of the bottom plate on the three half-webs; (d) central hinge in disassembled
position. Figure 2 shows part of the construction sequence.

Figure 2: Part of the construction sequence of the 10 m gap crossing aid; (a) bonding of the bottom plate on
the three half-webs; (b) central hinge in disassembled position.

133
3. ANALYSIS
The finite element method, using LUSAS software, was used to analyze the static response of the 10-m FRP gap-
crossing aid. Three analyses were done for the tapered beam assuming (a) no mid-span hinge and without the ribbed
decking; (b) no mid-span hinge but with the ribbed decking, and (c) with the mid-span hinge and the ribbed decking.
For analysis (c), the two halves of the tapered beam were connected at the extreme top and bottom elements only,
neglecting as a first approximation the contact problem over the two faces of the adjoining halves. The critical
loading case considered was that of a patch load (MM) of 125kN (over 90% of the design vehicle load) at the mid-
span apex of the box-beam, as shown in Figure 1. The vertical displacements for the three analyses are (a) 49.2 mm,
(b) 26.9 mm, and (c) 61.1 mm. The ribbed decking added considerable stiffness to the displacement behaviour, but
the introduction of a hinge resulted in significantly higher displacements compared to the analysis without the hinge
but with the ribbed decking (i.e. (b)).

4. SUMMARY AND CONCLUSIONS

The design of a 10-m full-scale FRP gap-crossing aid has been completed, with consideration given to the concepts
for modular assembly, methods of transporting, launching and recovering of the structure. The design of a hinge at
mid-span to accommodate ease of transportation and launching was a major focus. Several hinge prototypes,
including hybrid FRP/steel hinges, were designed, constructed and tested, before arriving at the proposed solution to
use an all-steel hinge. Finite element analysis of the design was used to finalize the global design details of the
structure. Deflections that normally govern the design of FRP structures were acceptable.

5. REFERENCES

Erki, M.A., Yantha, P.J., Green, M.F., Johansen, G.E. and Wilson, R. (1994). Dynamic Response of an FRP
Military Vehicle Bridge. Fourth International Conference on Short and Medium Span Bridges, 8-11 August,
Halifax, Nova Scotia.
Erki, M.A., Yantha, P., Green, M.F., Johansen, G.E., Wilson, R., and Mauer, D. (1995). Experimental Behaviour of
a Reinforced Plastic Vehicle Bridge. Proceedings of the Composite Institute's 50th Annual Conference & EXPO'95,
Feb 1995, pp. 11E1-5.
Erki, M.A., Tanovic, R., Penstone, S.R., Johansen, G.E., and Wilson, R. (1997). Fatigue Evaluation of FRP
Roadway Bridges, Recent Advances in Bridge Engineering, ed. U. Meier and R. Betti, Zurich, pp. 193-201.
Heffernan, P.J., Wight, R.G., Shyu, C.T., Tanovic, R., Erki, M.A. 2004. Rapidly Deployable, Portable, GFRP
Vehicle Bridge Development Project. Proceedings of the 11th European Conference on Composite Materials,
Rhodes, Greece. CD ROM.
Tanovic, R., Erki, M.A., Penstone, S. (1998). “Fatigue Behaviour of Pultruded FRP Components for Short and
Medium Span Bridges”, 5th International Conference on Short and Medium Span Bridges, SMSB-V. Calgary. (CD).
Tanovic, R., Erki, M.A., Johansen, G.E., and Wilson, R. (1997). Extreme and Fatigue Traffic Loading of a
Reinforced Plastic Vehicle Bridge, Proceedings of the Annual Conference of the CSCE, Sherbrooke, Quebec, Vol. 6,
pp. 61-70.
Wight, R.G., Shyu, C.T., Tanovic, R., Erki, M.A., and Heffernan, P.J. (2004a). Short-span deployable GFRP
Tapered Box-Beam Bridge. 4th International Conference on Advanced Composite Materials in Bridges and
Structures, Calgary, Alberta, 20-23 July, CD-ROM.
Wight, R.G., Shyu, C.T., Tanovic, R., Erki, M.A., and Heffernan, P.J. (2004b). Deployable Tapered Box-beam
Bridge. Advanced Polymer Composites for Structural Applications in Construction. ACIC 2004 (ed. Hollaway, L.C.,
Chryssanthopoulos, M.K., Moy, S.S.J.) Woodhead Publishing Limited, Cambridge, England. pp. 428-433
Yantha, P., Green, M.F., Erki, M.A., Johansen, G.E., and Wilson, R. (1995). Dynamic Testing of an FRP Vehicle
Bridge. Proceedings of the SEM/IMAC Conference, Nashville, Tennessee, Feb 1995, pp. 11E:1-5.

134
Third International Conference on FRP Composites in Civil Engineering (CICE 2006)
Dec.13-15 2006, Maimi, Florida, USA

RESPONSE OF NO-NAME CREEK FRP BRIDGE TO STATIC

LOADS, MOVING AND IMPACT TRAFFIC LOADS
Eric Zhou and Youqi Wang
Department of Mechanical and Nuclear Engineering
Kansas State University
Manhattan, Kansas

David Meggers
Kansas Department of Transportation
Topeka, Kansas

Jerry Plunkett
Kansas Structural Composites Inc.
Russell, Kansas

ABSTRACT
In this paper, the field tests are conducted to investigate the response of the first composite bridge in the US
to the static load, moving traffic load and impact traffic loads. Static results were compared to previous
results. Dynamic responses of the bridge to dynamic traffic loads were measured and are presented in this
paper. A detailed discussion is presented.

KEYWORDS:
FRP Bridge, Impact Traffic Load, Moving Traffic Load, Field Tests

1. INTRODUCTION AND GENERAL DESCRIPTION OF FIELD TESTS

No-Name Bridge is the first all composite road bridge in the United States. It was built by Kansas
Structural Composites Inc. in Russell, Kansas in 1996. Fig.1 is a schematic illustration of the bridge. It is a
single span and self-supporting glass-fiber reinforced composite bridge. Three composite panels comprise
the bridge. The panels are supported by steel I-beams which rest on the creek banks. There is a freestanding
test platform for long term monitoring beneath the bridge. Fig.2 is a schematic drawing of the three bridge
panels. Panel length is 23’3”. Bridge width is 27’9”. Interlocking joints connect the panels side by side.
Previously, two static field tests had been conducted on Nov. 19, 1996 and May 18, 1997, respectively. On
September, 2004, field tests were conducted to examine the response of No-Name Creek Composite Bridge
to static loads, moving traffic loads, and impact traffic loads. A AASHTO (American Association of State
Highway Transportation) type 3 truck was used to apply load to the bridge. Truck dimensions and load
distribution are shown in Fig.3. Total weight was 70340 lbs., shared by front wheels and rear wheels. Load
supported by front wheels was 16320 lbs.; load supported by rear wheels was 54020 lbs. Three laser
sensors were mounted on the test platform. When the truck moved though the bridge, deflection data were
transferred to a computer. Specified maximum sensor error is 50µ. Four kinds of loads were applied
during the tests:
1. Static load tests: the truck rested on the middle span location of the bridge
2. Creep load tests: the truck moved through the bridge as slow as possible
3. Dynamic traffic load tests: the truck moved through the bridge at controlled speeds, and,
4. Impact load tests: the truck moved over an obstacle at controlled speeds. When the truck passed
over the obstacle, an impact load was applied to the bridge.
Static results were compared to previous results. Dynamic responses of the bridge to dynamic traffic loads
were measured and are presented in this paper.

135
1
Third International Conference on FRP Composites in Civil Engineering (CICE 2006)
Dec.13-15 2006, Maimi, Florida, USA

North panel

Center Joints
panel 27’
Traffic
Direction
South panel

23’ 3”

Fig.1 No-name creek bridge Fig.2 Schematic drawing of bridge panels

4’

15’ 4’ 6’
8’
16320 54020
Fig.3 Dimensions of the loading truck

2. TEST DETAIL AND TEST RESULT ANALYSIS

Bridge deflection was measured in four kinds of loads: static truck load, creeping truck load, moving truck
load with controlled speed and obstacle induced impact truck load.

2.1 Static Tests

Five static load tests were conducted. In static tests, the truck was loaded on one panel. Mid-span
deflections of all panels were measured. This allowed us to evaluate
(i) Load transfer between panels and
(ii) Rigidity of bridge panels.
Refer to Table 1. The first column of the table shows truck locations. The second, fourth and sixth columns
show the mid-span deflections of each panel. The third, fifth and seventh columns show load fractions
supported by each panel.

Table 1. Results from static tests

North Panel Center Panel South Panel

Truck Deflection Load Deflection Load Deflection Load Rigidity(EI)
Location (in) Fraction (in) Fraction (in) Fraction (lb.in2)
North panel 0.2567 0.7296 0.0878 0.2496 0.0073 0.0208 5.78E+10
Center panel 0.0878 0.2736 0.1547 0.4822 0.0783 0.2442 6.33E+10
South panel 0.0041 0.0146 0.0669 0.2416 0.2060 0.7437 7.34E+10

136
2
Third International Conference on FRP Composites in Civil Engineering (CICE 2006)
Dec.13-15 2006, Maimi, Florida, USA

2.2 Creep Test

Four creep tests were conducted on the center and south panels. For the creep tests, the truck moved
through the bridge as slow as possible. Deflections of mid-span and ¼ span points were recorded as the
truck moved through the bridge. Thus, an influence curve was derived. Fig. 4 shows the two influence
curves from two creep tests. Three horizontal curves were taken from the previous static tests.

00:00.0 00:17.3 00:34.6 00:51.8 01:09.1 0 0.0002 0.0004 0.0006 0.0008

0.05
0.02
0 0
-0.02
-0.04 -0.05

Deflection
Creep
Deflection

-0.06
-0.1
-0.08
-0.1 South
-0.15
-0.12 Static Center
-0.14 North -0.2
-0.16
-0.18 -0.25

Time(mintue:second) Time (minute:second)

4-a Test 1: Creep test on center panel 4-b Test-2: Creep test on south panel

Fig.4 Influence curves in mid-span positions

2.3 Moving Traffic Dynamic Tests

In moving traffic dynamic tests, the truck moved through the bridge at five controlled speeds: 5
miles/hour, 10 miles/hour, 20 miles /hour, 30 miles/hour and 35 miles/hour. Similar to static and creep
tests, four groups of tests were conducted. In the first and third groups of tests, the truck moved through the
center panel. In the second and fourth groups of tests, the truck moved through the south panel. Fig.5
shows mid-span deflection of the center panel under a moving traffic load in the first group of tests. One
sees deflection varied insignificantly when the speed was lower than 20 miles/hour. It began to increase
significantly afterwards. Similar results were achieved in the other groups of tests. In addition, ratios
between dynamic tests and static tests are calculated and are defined as dynamic factors. Fig.6 shows the
dynamic factors from the first and third groups of tests.

0 0.0002 0.0004 0.0006 0.0008 0.001

0.05
1.5

0
1.25
Dynamic Factor

creep
-0.05
5 miles/H 1
-0.1
10 miles/H
Group 1
-0.15 20 miles/H 0.75
Group 3
-0.2 30 miles/H
35 miles/H 0.5
-0.25
static 0 10 20 30 40
T i m e ( m i n u t e :se c o n d )
Moving Speed of Truck(Miles/hour)

Fig.5 Mid-span deflection in moving traffic tests Fig.6 Dynamic factors

137
3
Third International Conference on FRP Composites in Civil Engineering (CICE 2006)
Dec.13-15 2006, Maimi, Florida, USA

2.4 Impact Tests

For impact tests, a 2”x 6” wood stud was placed onto the middle of the bridge. The truck drove over the
wood stud. Similar to dynamic tests, four groups of tests were conducted. Fig.6 shows results from the first
group of tests on the center panel. Deflection increases with increasing moving speed. Impact effect cannot
be neglected even when speed is very slow. A similar phenomenon was observed in other groups of tests.
The impact factors are shown in Fig.7. The impact induced the south panel vibration as shown in Fig.8.

00:00.0 00:17.3 00:34.6 00:51.8 01:09.1 01:26.4 1.5

0.05

Impact Factor
0 1.25
Deflection (in)

-0.05 creep
1
static
-0.1 Group 1
10 miles/H Group 3
0.75
-0.15 20 miles/H
-0.2 30 miles/H
0.5
35 miles/H 0 10 20 30 40
-0.25
Time (minute:second) Truck Moving Speed

Fig.6 Mid-span deflection under impact loading Fig.7 Dynamic factor

00:33.26 00:34.13 00:34.99 00:35.86

0.04
Deflection (in).

-0.04

-0.08

-0.12
Time (Minute:Second)

Fig.8 Impact induced south panel vibration (Truck speed: 30 miles/hour, loading: center panel)

3. CONCLUSIONS

Several conclusions are reached through these field tests:

1. Two static tests were conducted in 1996 and 1997. It were found that the rigidity of bridge panels
were 6.72 x 1010 lb. in2 and 5.93 x1010 lb. in2. In this field test, 6.48 x1010 lb. in2 of the static
rigidity was derived. This indicates that there has been no significant change of bridge rigidity
after 8 service years.
2. Bridge response to traffic load with a speed lower than 20 miles/hour is similar to static load.
Dynamic effect can be neglected. Deflection increases with increased traffic speed after traffic
speed reaches 20 miles/hour. When traffic speed reaches 35 miles per hour, bridge deflection
increases approximately 30%.
3. Bridge deflection increases with traffic speed in impact tests. Impact effect cannot be neglected
even when speed is slow. When traffic speed reaches 35miles per hour, deflection increase
approximately 45%.
4. Impact load may trigger significant free vibration of south panels. Free vibration was found in
impact tests with traffic speed of 30miles and 35 miles per hour. Frequency approximates 18.6 Hz.

138
4
Part III. Bridge Decks
Third International Conference on FRP Composites in Civil Engineering (CICE 2006)
December 13-15 2006, Miami, Florida, USA

BEHAVIORS OF NEW GENERATION OF FRP BRIDGE DECK

WITH OUTSIDE FILAMENT-WOUND REINFORCEMENT
Peng Feng
(Lecturer, Dept. of Civil Engineering, Tsinghua Univ., Beijing, China)

Lieping Ye
(Professor, Dept. of Civil Engineering, Tsinghua Univ., Beijing, China)

ABSTRACT
OFR (outside filament-wound reinforcement) is an effective configuration to improve the performance of FRP
bridge decks. It can enhance the transverse stiffness of FRP decks and restrict the swelling and dispreading trends of
the FRP components to increase the ultimate load and deformation capacity of FRP decks. The new generation of
FRP bridge deck with OFR, named HD deck, was designed. Four HD decks were manufactured and tested in static
and fatigue loads. The results show that the HD decks satisfy the demands of China Bridge Design Codes. The
efficiency of OFR was verified. Based on the results, the reinforcement mechanism was analyzed.

KEYWORDS
Corrosion resistance, fatigue, light-weight superstructure, debonding, failure simulation.

1. INTRODUCTION
Many different types of FRP bridge decks have been designed and investigated in the last decade as their favorable
characters, such as corrosion resistance, light weight and convenient installation (Karbhari et al, 2000; Brown and
Zureick, 2001; Stone et al, 2001; Luke et al, 2002; Williams et al, 2003; Keller and Schollmayer, 2004). It was
found in the experimental studies of FRP decks that the FRP longitudinal strength, which is much higher than the
strengths in shear, transverse and inter-layer, was not utilized efficiently in most cases. The observed failure
characteristics, including cracking, fiber breaking, debonding and delaminating, have the corresponding maximum
load respectively. The damage always occurs in the mode with the lowest ultimate load: the strength failure in weak
direction, such as crack in fiber direction and delaminating, or the assembly failure between components, such as
debonding, slip and pullout. To prevent these failure modes and improve the performance of FRP decks, the
configuration of OFR (outside filament-wound reinforcement) was presented (Feng et al, 2004). As the outside
filament-wound layers enhance the transverse stiffness of the surface of FRP decks and restrict the swelling and
dispreading trend of the components, the ultimate load and deformation capacity can be increased obviously with the
failure mode changing. Based on this concept, the new generation of FRP bridge deck with OFR was designed,
named HD deck. Four HD decks with different OFR layers were tested, among which three were loaded in static and
one in fatigue load. And the behavior of HD decks was simulated with finite element software. Based on the results,
the reinforcing efficiency of OFR is investigated.

2. HD DECKS
The HD deck is composed of four filament-wound square tubes, two pultruded plates and the OFR layer, as shown
in Figure 1. HD decks will be laid on girders in parallel with Z-shaped connectors to form the superstructure of
bridges as illustrated in Figure 2. The square tube made of 4400Tex E-glass fiber roving and epoxy resin is wound in
±45 . Its average thickness is 8mm and the fiber volume ratio is 60.5%. The thickness of the face plate is 12mm
which is made of E-glass 4400Tex fiber roving, fabric and mat with vinyl ester resin. They are assembled together
and glued with epoxy resin. Then the deck is wound with the fiber filaments dipped in resin in the cross-angle ±θ.

139
The OFR layer is made of the same material as the tubes. The mechanical properties of plates and tubes were
determined by tests, which are listed in Table 1.

OFR layer Filament wound tube Putruded plate

Figure 1: HD Deck Section Figure 2: HD Decks in Bridge

Table 1: Material Properties of HD Decks

Modulus / GPa Strength/ MPa

Poisson ratio
Longitude Transverse Shear Longitude Transverse Shear
Plate 31.84 17.41 5.79 0.213 >304 (tension) 57.4 48.0
Tube 23.96 23.96 9.33 0.483 191(compress) - -

3. EXPERIMENTAL STUDY OF HD DECKS

Four HD decks with difference OFR layer thickness or cross angle manufactured for the tests, including three for
static tests and one for fatigue test, as listed in Table 2. All decks were simply supported. There were two kinds of
loading conditions as shown in Figure 3: the central point of 200×200mm area to simulate the wheel load and the
four-point line bending to estimate their stiffness.

Table 2: Tested HD Decks

Specimens Length/m Span/m OFR Load cases

Case 1: four-point line bending to 350kN
HD0 3.0 2.8m None
Case 2: central point to failure
±85° Case 1: four -point line bending to 350kN
HDW3 3.0 2.8m
3mm Case 2: central point to failure
±85° Case 1: central point to 630kN
HDW5 3.0 2.8m
5mm Case 2: four -point line bending to failure
±60° Fatigue: central point to 20-100kN 2000000 times
HDW3-60 3.0 2.8m
3mm Static: central point to 150kN 9 times

900 100 800 100 900 2800

2800
200

200

3000 3000

Figure 3: Four-point Line Load and Central Point Load

3.1 Static Load Tests

For HD0 without OFR, the tubes and bottom plate debonded suddenly when the central load reached 485kN as
shown in Figure 4(a). There was an obvious gap between the bottom plate and the tubes. Before the failure, HD0

140
was linear elastic, and the ultimate deflection was 30.1mm. Under the central load of 70kN, which is the design
wheel load, the deflection was 4.11mm (span/681). HDW3 deck failed in the punching under the load position as
shown in Figure 4(b). The fibers around the loading pad were cut and broke. The maximum load was 618kN with a
deflection of 52.5mm after a short ductile process. Comparing with HD0, HDW3 had a 26.4% increase of the load
capacity. Its deflection under the design load was 3.97mm that indicates the stiffness was improved a little by OFR.
HDW5 didn’t fail when the load went to 632kN which is higher than the maximum value of HD0 and HDW3 under
the same load pattern. It had a 3.35mm deflection under the design load. After unloading of Case 1, HDW5 was
loaded to failure at 1737kN in Case 2. The deck deformed obviously before failure, and broke suddenly at 1737kN.
The fiber under the left load point was crushed and fractured as shown in Figure 4(c). In this failure mode, FRP is
utilized effectively although it failed in brittle. Comparing these tests, it is shown that the ultimate strength increase
obviously with the OFR layer thickness. The effects of OFR can enhance the ultimate load capacity and make the
failure process more ductile.

(a) HD0 (b) HDW3 (c) HDW5

Figure 4: Failure Modes of HD Decks

3.2 Fatigue Load Test

The HDW3-60 was tested in fatigue to simulate the traffic loads. The load was applied periodically from 20kN to
100kN for 2 million times. After 5k, 20k, 50k, 100k, 800k, 1.2M, 1.6M and 2M times, a static load of 150kN was
applied once to get the stiffness of the deck. During the loading process of 7 days, no obvious failure was observed,
the deck deformed and bounced periodically under the cyclic load. After loading, there was a residual deflection of
0.3mm. The stiffness deterioration of HDW3-60 in the test is illustrated in Figure 5, which are lower than 7% and
correspond linearly with the logarithm of load times.

-1140 8.0
G2 S3
-1160 7.8
Deflection (mm)
变形(mm)

-1180 7.6
应变(με )
Strain (με)

y = -14.792lgN - 1101.1 y = 0.1667x + 6.5116

-1200 7.4

-1220 7.2

-1240 7.0
lgN 0 1 22 3 4 5 6 77 lgN 0 1 2 3 4 5 6 7
3 4 5 6 2 3 4 5 6 7
N 1 10 10 10 10 10 10 10 N 1 10 10 10 10 10 10 10
加载次数 加载次数
Load times Load times

(a) Strain (b) Deflection

Figure 5: Changes of HDW3-60 with Load Times

3.3 Test Analysis

The stiffness of FRP deck is the control parameter in design. All tested HD decks satisfy the request of Heavy
Truck-20 in China Bridge Design Code: the deflection is less than span/600 (4.67mm) under a wheel load of 70kN.
The differences of their bending stiffness and shear stiffness determined by four-point line tests are no more than
10.8% to each other. It means that the OFR layer with a thickness less than 5mm doesn’t affect the stiffness of the
decks obviously. However, the ultimate load and the deformation capacity are improved considerably, which can be
seen from the deflection-load curves in Figure 6.

141
The transverse strains in OFR disclose the reinforcing mechanism. Figure 7 shows the transverse strain distribution
of deck surface of HD0 and HDW3 in midspan. When the load is 200kN, which is lower than the half of the
ultimate strength of HD0, the strains (the left parts) in these two decks are almost same. But when the load closes to
the failure load, the difference (the right parts) at the section corner and near the loading pad can be seen where the
strains in HDW3 are much higher than those in HD0. Hence, the OFR layers enhance the inner deck at the corner
mainly, which is also verified by FEA (finite element analysis).

HDW3:punching
700
618kN
HDW5: no failure
600 632kN 3572
500 HD0: debonding 1900 1112 2136
485kN 1715
Load(kN)

715 740 689 527

400 550 302
209
300 200kN 488kN 689
200kN 597kN 2136

200 -2609 598

100 -598 831

-251 -976 -836
-765 -817
-1050 -322
0
0 20 40 60 1245
Deflection(mm) 1672 815
1420

(a) HD0 (b) HDW3

Figure 6: Load-deflection Curves of Tests Figure 7: Transverse Strain Distribution of Deck Surface

4. CONCLUSIONS
New FRP bridge deck with OFR is design and studied. Four HD decks were tested in static and fatigue loads. The
conclusions can be summarized as below: (1) the HD deck with OFR has high bearing strength and low stiffness
deterioration; (2) the OFR layer is very effective to enhance the strength and the deformation capacity of FRP bridge
deck; (3) OFR layers act at the corner of section mainly.

5. ACKNOWLEDGEMENTS

The authors are very appreciated to Chinese Natural Science Fund to support the key Program of FRP application in
civil engineering in China (No. 50238030) and the National High Technology Research and Development Program
of China (863 Program) (No. 2001AA336010).

6. REFERENCES
Brown, R.T., Zureick, A.H. (2001). “Lightweight composite truss section decking”. Marine Structures, Vol.14,
pp115-132.
Feng, P., Ye, L.P., Zhang, L.W., et al. (2004). “Experimental study of outside filament winding reinforced FRP
bridge decks”. Proc. 4th International Conference on Advanced Composite Materials in Bridges and Structures,
Calgary, Alberta, Canada, (CD-ROM).
Karbhari, V. M., Seible, F., Burgueno, R., et al. (2000). “Structural characterization of fiber-reinforced composite
short- and medium-span bridge systems”. Applied composite materials, Vol.7, No.2-3, pp151-182.
Keller, T., Schollmayer, M. (2004). “Plate bending behavior of a pultruded GFRP bridge deck system”. Composite
Structures, Vol.64, pp285-295.
Luke, S., Canning, L., Collins, S., et al. (2002). “Advanced composite bridge decking system – Project ASSET”.
Structural Engineering International, No.2, pp76-79.
Stone, D, Nanni, A, Myers, J. (2001) “Field and laboratory performance of FRP bridge panels”. Proc. International
Conference Composites in Construction, Porto, Portugal, pp701-706.
Williams, B., Shehata, E., Rizkalla, S. H. (2003). “Filament-wound glass fiber reinforced polymer bridge deck
modules”. Journal of Composites for Construction, Vol.7, No.3, pp266-273.

142
Third International Conference on FRP Composites in Civil Engineering (CICE 2006)
December 13-15 2006, Miami, Florida, USA

TESTS OF A FRP DECK PANEL AT VERY COLD TEMPERATURES

Z. John Ma, Ph.D., P.E.
(Associate Professor, University of Tennessee, Knoxville, Tennessee,USA)

Usha Choppali
(Project Engineer, Tendon Systems LLC, Columbus, Georgia, USA)

ABSTRACT
The FRP deck panel with an overall geometry of 2.134m x 0.340m x 0.178m was cyclically tested under a three-
point loading setup in a cold room. During a three-month testing period, the temperature of the cold room was
changed from 260C to 00C, -200C, -310C, back up to 270C, then to -380C, and finally to -550C. In addition to the
experimental program, the Classic Beam Theory (CBT) method and the Finite Element (FE) method were used to
predict the panel behavior at room temperatures. Based on this research, the following conclusions were made: (1)
The low-temperature stiffness was about 3.8% higher than at the high temperature as long as the low temperature
was above a certain degree; (2) When the temperature was lowered from -200C to -550C, the low-temperature
stiffness decreased 11.5% comparing with the stiffness at room temperatures. This degradation in stiffness could
not be recovered by raising the panel temperature; (3) Very cold temperature exposure resulted in matrix hardening,
matrix microcracking and fiber-matrix debonding as demonstrated by the AE testing data at different temperatures;
and (4) At the room temperatures, the FRP panel showed a little stiffer performance than both the CBT method and
the FE method would predict.

KEYWORDS
FRP deck panels, very cold temperatures, cycling test, stiffness degradation, Acoustic Emission

1. INTRODUCTION
Fiber-Reinforced Polymer (FRP) composites, used in various applications and environments, are often chosen over
conventional materials for their durability. However, there remains a “durability gap” for FRP composites
(Karbhari, et al., 2003). Based on a recent study performed by Dutta et al. (2002), it was observed that the low-
temperature (-300C) stiffness in the load-deflection curves was higher than at the high temperature (500C). In cold
regions like Alaska, the temperature during the winter season may go well below -50°C. It is difficult to do cast-in-
place concrete for bridge decks in winter. As a result, the use of the FRP composites to replace existing,
deteriorated bridge deck systems offers a very attractive alternative in cold regions. The paper presented here
focused mainly on the study of the behavior of the FRP deck panel at very cold temperature (below -500C) cycling
combined with a service load cycling.

2. EXPERIMENTAL PROGRAM
The FRP deck panel specimen was obtained from Kansas Structural Composites Inc. (KSCI), Kansas. It had an
overall geometry of 2.134m in length, 0.340m in width, and 0.178m in thickness. The specimen was of sandwich
type construction with a vertical corrugated core and was manufactured using the wet lay-up method, as discussed
by Plunket (1997). The core of the panels was 152.0mm high and consisted of corrugated cells of 2.5mm thick in
the form of honeycomb structure having a standard geometry of 51.0mm and amplitude of 102.0mm wave length.
Also, the edge lengths were covered with two layers of 85.0 grams of chopped strand mat. According to the
manufacture’s data, the top and bottom plates, of thickness 13.0mm, were composed of several layers of
unidirectional, bi-directional and chopped E-glass fibers. Both polyester and vinyl ester resin will be used as matrix
in the testing program. However, polyester resin was used as matrix in the specimen reported here. Figure 1 shows

143
the specimen and the testing setup. The sensor instrumentation can also be seen in the figure. The testing was
conducted using a self-equilibrating structural steel frame in a cold room. The cold room is 4.3m by 4.3m in size
and was designed to provide a low temperature of -700C. Figure 2 shows the control panel of the cold room while
one of tests was in progress. It was decide to use the service load level of 35.6 KN to test the specimen at various
temperatures. At each temperature, the specimen was loaded up to 35.6KN at a loading rate of 2.2 to 4.4 KN per
minute. Figure 3 shows the loading and temperature cycling history of the experiment. The detailed testing program
was reported elsewhere (Ma et al., 2006).

Figure 1: Three-Point Bending Test Setup Figure 2: Control Panel of the Cold Room
60

40

20 Temperature
Test Performed Dates
Temperature (degree C)

0
5/3/2005 5/18/2005 6/2/2005 6/17/2005 7/2/2005 7/17/2005

-20

-40

-60

-80
Time (days)

Figure 3: Loading and Temperature Cycling History

3. BAVIOUR OF THE FRP PANEL AT 260C

After the first stage of room temperature test, it was decided to perform an analysis of the specimen to compare the
result from the experiment with the available material data from the manufacturer of the specimen. To achieve this
objective, two methods were used: the Classic Beam Theory (CBT) and Finite Element Modeling. In the CBT
method, assume that the flexure rigidity for all of the core sheets is the same and all three layers are firmly bonded
together. The second assumption was also made that cross-sections which are plane and perpendicular to the
longitudinal axis of the unloaded panel remain so when bending takes place. Besides the CBT method, the panel
was also modeled and analyzed by the finite element analysis program ABAQUS with a three dimensional
deformable shell model. The FE model consists of top and bottom flanges connected by a sandwich core as shown
in Figure 4. Figure 5 shows the comparison of the strain between the test result and the FE prediction. Similar to
the deflection, the test result showed stiffer performance.

144
 40.0
35.0
30.0
25.0

Load (KN)
20.0
15.0
Test
10.0
FE method
5.0
0.0
-600.0 -500.0 -400.0 -300.0 -200.0 -100.0 0.0
Strain (uE)

Figure 4: FE Model of the Panel Figure 5: Load vs Strain

4. IMPACT OF THE VERY COLD TEMPERATURE CYCLING

After the first stage room temperature testing, the temperature of the cold room was lowered from 260C to 00C as
shown in Figure 3. The panel was tested at 24 hours and 48 hours from the time of lowering the temperature. The
data points in Figure 6 shows the relationship between the applied load and the deflection of the panel at mid-span.
Figure 6 also shows that the panel stiffness, as represented as the slope of the trend line, is 10.82 KN/mm. The load
vs. deflection curve for tests at other temperatures has the same pattern as shown in Figure 6, but with different
stiffness values. Figure 7 shows a comparison of the panel stiffness at different temperatures. Please note that the
testing conditions remained the same throughout the experiment. The only variable in each data set of the
experiment was the temperature. Referring to Figure 7 for the test performed at 26 0C, a stiffness of 10.42 KN/mm
was obtained. As the temperature was lowered to 00C, a static loading test was performed at the 24th hour and the
stiffness value attained was 10.82 KN/mm. Another static testing was performed at the 48th hour at the same
temperature, and the stiffness value was decreased to 10.59 KN/mm. When the temperature was further lowered to
-200C, the stiffness was found to be 10.60 KN/mm at the 24th hour. Until this stage of testing, the panel had been
subjected to the temperature change of 460C and four cycles of loading and unloading static tests. At the end of the
4th cycle loading test, the panel stiffness increased the maximum of about 3.8% as the temperature decreased from
260C to 00C and further to -200C. However, when the panel was kept in the -200C for another 24 hours the stiffness
did not further increase. Instead, it decreased 5.0% comparing with the stiffness at room temperature. When
subjecting the panel to further colder temperature as cold as -550C, the panel stiffness showed a gradual decrease,
and stabilized around the value of 9.22 KN/mm. When comparing with the stiffness at original room temperature, it
decreased 11.5%. Most importantly, this degradation in stiffness could not be recovered by raising the panel
temperature, as shown in Figure 7. Figure 8 shows an example comparison of the AE testing data at different
temperatures. On comparing the two figures [(a) and (b)], it is seen that more activity happened at -310C. There
was also more energy at -310C. A higher amplitude range (50-60 dB) was observed at -310C when compared to a
temperature of 260C. Please note that in Figures 8 (a) and (b) the longer the time of observation the greater is the
load applied on the specimen.

5. CONCLUSIONS
The following conclusions were made: (1) The low-temperature stiffness was about 3.8% higher than at the high
temperature as long as the low temperature was above a certain degree (about -200C under the testing condition
presented in the paper); (2) When the temperature was further lowered from -200C to -550C, the low-temperature
stiffness did not further increase. Instead, it decreased 11.5% when comparing with the stiffness at the original room
temperature. Most importantly, this degradation in stiffness could not be recovered by raising the panel
temperature; (3) Very cold temperature exposure resulted in matrix hardening, matrix microcracking and fiber-
matrix debonding as demonstrated by the AE testing data at different temperatures; and (4) At the room
temperatures, the FRP panel showed a little stiffer performance than the CBT and the FE methods would predict.

145
 40.0
11.00
35.0 y(Load) = 10.82 x (Deflection)
R2 = 0.997 10.50

Stiffness (KN/mm)
30.0

10.00
25.0
Load (KN)

9.50
20.0

9.00
15.0

8.50
10.0

8.00
5.0
26 0 0 _20 _20 _31 27 27 _38 _55
(24hr) (48hr) (24hr) (48hr) (24hr) (24hr) (48hr) (24hr) (24hr)
0.0
0.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5
Deflection (mm)
Temperature ( 0 C )

Figure 6: Load vs Deflection at 00C (24 hours test) Figure 7: Stiffness vs Temperature and Time

(a) at 26 0C (b) at -31 0C

Figure 8: Overview of the AE Data at Different Temperatures

6. ACKNOLEDGEMENT
The test specimen was provided by the KSCI, and we thank Dr. Jerry Plunkett of KSCI for his technical contribution
and support. Partial financial support for this study is received from the National Science Foundation – NSF
CAREER program (CMS – 0343865) and the Department of Civil and Environmental Engineering at the University
of Alaska Fairbanks.

7. REFERENCES
Dutta, P.K., S.C.Kwon, G.Dullel, and R. Lopez-Anido (2002), “Fatigue Evaluation of Composite Bridge Decks,” In
proceedings, 10th U.S.Japan Conference on Composite Materials, September 16-18, 2002, Stanford University,
Stanford, C.A., DEStech Publications, 336-344.
Karbhari, V.M., et al. (2003), “Durability Gap Analysis for Fiber-Reinforced Polymer Composites in Civil
Infrastructure,” Journal of Composites for Construction, Vol. 7, No. 3, August 2003, pp. 238 – 247.
Ma, Z. J., Choppali, U., and Li, L., (2006), “Cycling Tests of a Fiber-Reinforced Polymer Honeycomb Sandwich
Deck Panel at Very Cold Temperatures,” International Journal of Materials and Product Technology (IJMPT), Vol.
26 (in press).
Plunket, J. D. (1997), “Fiber-reinforced Polymer Honeycomb Short Span Bridge for Rapid Installation,” IDEA
Project Report, November.

146
Third International Conference on FRP Composites in Civil Engineering (CICE 2006)
December 13-15 2006, Miami, Florida, USA

STATIC AND FATIGUE TESTING OF INNOVATIVE ALL-COMPOSITE

BRIDGE SLAB
René J. Roy
Ph. D. Candidate, Department of Civil Engineering, University of Sherbrooke, Sherbrooke, Quebec, Canada

Mathieu Robert
Ph. D. Student, Department of Civil Engineering, University of Sherbrooke, Sherbrooke, Quebec, Canada

Brahim Benmokrane
NSERC Research Chair Professor in Innovative FRP Composite Materials for Infrastructures, Department of Civil
Engineering, University of Sherbrooke, Sherbrooke, Quebec, Canada

Ahmed S. Debaiky
Fellow NSERC Industrial Postdoctoral Researcher, A.S. Composites Inc., Boucherville, Quebec, Canada

Hossein Borazghi
President, A.S. Composites Inc., Boucherville, Quebec, Canada

ABSTRACT
In the context of the state of deterioration of North America’s bridges, researchers have been working on improving
the performance of current designs. All-FRP bridge slabs have attracted increased interest since the mid 90s. These
slabs offer notable advantages over traditional reinforced concrete slabs. These advantages include lighter weight,
better durability and faster installation among others. This research focuses on the development of a thermoplastic
fiber reinforced polymer (FRP) bridge slab for the replacement of wood logs that are part of the deck structure of a
popular steel/timber bridge design. Different structural profiles have been explored through finite element analysis
and two slab prototypes have been fabricated. These full-scale prototypes have dimensions of approximately
3200mm×1200mm×250mm and cover two lateral spans. Both prototypes were tested under concentrated static
loading and one was subjected to above two million cycles of loading. The slab deflection and FRP laminate stress
levels under loading were recorded. The retention of the slab rigidity was monitored at regular intervals during
cyclic fatigue loading. No significant rigidity loss was observed during this period.

KEYWORDS
Bridge, deck, FRP, thermoplastic, fatigue

1. INTRODUCTION
The U.S. Federal Highway Administration estimated that close to 30% of the 588 288 bridges in the United States
either have a deficient structure or their capacity is obsolete (Busel, 2002). Some of these bridges are located in
regions prone to harsh weather conditions, particularly Nordic regions where de-icing salts are commonly used to
keep roads functional. As a result, these bridges tend to deteriorate faster. FRP materials are known to be immune to
corrosion that affects steel rebars present in reinforced concrete bridge slabs. Along with resistance properties in the
same range as steel, FRP composites represent a better alternative for bridge materials. The concept of all-FRP
bridge slabs in place of reinforced concrete slabs has been investigated by various researchers and companies
(Market Development Alliance, 2000). Many field projects have been realized in the United States since the mid
90s. These slabs had different section profiles composed of FRP laminates either produced by pultrusion, hand lay-
up or vacuum assisted molding. This project focuses on the development of an FRP replacement bridge slab for a
steel/timber bridge application where the current deck structure includes staggered timber logs. These timber logs

147
need replacement after about 15 years. With the actual deck design, the steel girders are also exposed to fluid
spillage from the bridge surface and this increases their rate of deterioration.

2. DETAILS OF THE EXPERIMENTAL PROGRAM

2.1 Design Parameters

The replacement slab is to be installed in similar conditions as the existing timber logs; hence the span length (L)
was kept to 1450 mm for this design. It has been observed through experimentation on this bridge type that the
wheel load of a truck axle is supported by three consecutive timber logs. To mirror this, the width of the FRP slab
was set to an equivalent of 1200 mm. The thickness of the new slab would match the timber log at 203 mm if the
actual wearing surface is kept or it may be increased if a thinner wearing surface is chosen, up to a total of 299 mm
including the wearing surface. The prescribed truck axle service load is established from the Canadian Highway
Bridge Design Code at 110.25 kN per wheel (CHBDC, 2000). The length of the slab was chosen as to cover two
spans to reproduce continuity. The truck axle load is to be applied in two concentrated areas (2×110.25 kN)
separated by 1800 mm center to center as prescribed by the code. The loads are centered from the middle support, in
this way creating the maximum negative moment over this support. The maximum deflection criterion under the
service load was set to L/400, and maximum stress in the FRP to 30% of its ultimate strength.

Figure 1 Actual Steel/Timber Bridge Deck.

2.2 Structural Design and Prototype Fabrication

Glass fiber reinforced polypropylene was chosen as a candidate material for a bridge slab. It was used in the form of
commingled glass and polypropylene fibers 0º/90º weaved fabrics sold under the commercial name TwinTex®
(Saint-Gobain Vetrotex, 2006). Its properties are listed in Table 1.

Table 1 TwinTex® 60% Mass Fiber Material Properties (Robert, 2006).

Tensile Modulus Tensile Strength Flexural Modulus Flexural Strength

12 GPa 229 MPa 11 GPa 189 MPa

Different section profiles made of shaped FRP panels were considered with regard to structural efficiency and ease
of fabrication. It was chosen to fill the hollow spaces in the section with polyurethane foam to increase rigidity and
help prevent buckling of the FRP laminates in the profile. The polyurethane foam considered had a density of about
3.5lbs/ft3 for a modulus of elasticity taken at 10 MPa. These section geometries were modeled for finite element
analysis with the SAP2000® software (Computers and Structures, 2006). The FRP laminates were approximated as
straight line where needed and the laminate junctions were idealized as fixed (no slip). The model used linear 4-node
shell elements for the FRP laminates and linear 8-node elements for the polyurethane foam. From the simulation
results, two section geometries were chosen to be fabricated as prototype slabs for testing (Figure 2). These
geometries produced satisfactory deflection results for 4-5 mm thick laminates and were judged suitable for simple

148
fabrication. These prototype slabs were made as an assembly of FRP components joined together by bolts. The FRP
components were fabricated by vacuum bag molding in an oven at 180ºC. Electrical strain gages were installed on
the FRP laminates at various locations in the slabs. Once all the FRP components were assembled (Figure 3), one
end of the slab was blocked and the liquid polyurethane foam mix was poured in the cavities by gravity for foaming.

1200 11 55

4
5

4
204

236
25
4

12.5 12.5 t= 4 m m
5

4
125
150

Prototype #1 Prototype #2

Figure 2 Section Profile Geometries Chosen for Prototype Fabrication.

Figure 3 Prototype #2 During Construction.

2.3 Test Set-up and Cyclic Fatigue Loading Regime

Both prototypes were subjected to a static loading in the laboratory using a 500 kN hydraulic jack. Strains in the
FRP laminates were recorded and linear variable differential transformers (LVDT) were used to measure the slab
deflection. The slabs rested on three I-beams separated by 1450 mm reproducing the bridge girders. The load was
applied equally on two 600 mm × 250 mm steel plates lying on rubber pads. Prototype #2 was further subjected to
cyclic loading using two 500 kN hydraulic jacks. The load cycles had a sinusoidal pattern with amplitude from 15
kN to 110.25 kN at a rate of 1 Hz for an initial two million cycles. After these two million cycles the top load limit
was doubled to 220.5 kN until failure after 1500 cycles. Figure 4 shows the fatigue test set-up and loading regime.

Figure 4 Fatigue Test Set-up and Cyclic Fatigue Loading Regime.

149
3. TEST RESULTS AND OBSERVATIONS
Both prototype supported a load of 500 kN without any major visible damage resulting. Recorded FRP strains were
low with both prototypes, with the maximum representing 25% of ultimate capacity. Prototype #1 had a deflection
of L/180 at the service load which was greater than the design criteria. Lower than initially planned polyurethane
foam rigidity with excessive local deflection around the loading areas were identified as the cause of this defect in
rigidity. The slab was then loaded to failure with two 500 kN hydraulic jacks and supported an ultimate load of 780
kN. The failure mode was punching in the loading area. Prototype #2 had a better deflection of L/382 under service
load, which is in the range of the L/400 design criteria. It was therefore chosen for fatigue testing. The test was
stopped every 100,000 cycles to perform a recorded static loading whose results are presented in Figure 5. The
results show that there was little rigidity loss due to fatigue cyclic loading. Alteration of the polypropylene texture
was slightly visible in small regions of the top FRP panel close to the loading areas. After about 300,000 cycles of
loading this alteration gradually became more apparent. It later developed into small cracks in the polypropylene
matrix. At about the 1,500,000 cycle mark the top FRP panel locally kinked in the region of one of these cracks,
close to the end of the slab. We evaluate that these cracks were promoted by the initial static loading, which went up
to 2.3 times the service load. When the cyclic load upper limit was doubled, a crack opened in the whole FRP top
panel, with fiber ruptures, and developed. The slab failed in a punching style due to this crack after 1,500 cycles.
250.00

200.00

150.00
Total load [kN]

100.00

After 100,000 cycles

After 500,000 cycles
50.00 After 1,000,000 cycles
After 1,500,000 cycles
After 2,000,000 cycles

0.00
0 1 2 3 4 5 6 7 8 9 10
Deflection measurement besides the load surface (raw data) [mm]

(a) (b)

Figure 5 Prototype #2: (a) Slab Deflection Under Static Loading During the Fatigue Test; (b) Failure Mode.

4. CONCLUSIONS
The experiments conducted in this research lead to the following conclusions:
1. Stress levels remain low in the FRP laminates for all practical loading conditions and the designed slabs are able
to withstand a load higher than the factored design load.
2. Deflection under the service load for prototype #2 is L/382 and is in the range of the design criteria.
3. No significant loss of rigidity was recorded on prototype #2 during the two million cycles of load at 2×110.25 kN.
4. The mode of failure of the slabs is punching in the loading area and this doesn’t represent a catastrophic failure.

5. REFERENCES
Busel, J.P. (2002). “Engineered FRP bridge decks solutions for the future”, SAMPE Journal, vol. 38, no 5,
September/October 2002, p. 46-48.
CHBDC (2000) Canadian Highway Bridge Design Code, Standards Council of Canada, Ottawa, Canada.
Computers and Structures (2006). SAP2000 software, www.csiberkeley.com, Berkeley, California, USA.
Market Development Alliance (2000). “Product Selection Guide: FRP Composite Products for Bridge
Applications”, American Composites Manufacturers Association (ACMA), Arlington, VA., USA.
Robert, M. (2006). “Durabilité du composite polymérique de polypropylène renforcé de fibres de verre”, Ms. thesis,
Department of Civil Engineering, Université de Sherbrooke, Sherbrooke, Quebec, Canada.
Saint-Gobain Vetrotex (2006). “Twintex Woven Products Data Sheets”, www.twintex.com, Chambery, France.

150
Third International Conference on FRP Composites in Civil Engineering (CICE 2006)
December 13-15 2006, Miami, Florida, USA

RAPID CONSTRUCTION OF CONCRETE BRIDGE DECK USING

PREFABRICATED FRP REINFORCEMENT
Fabio Matta
(Graduate Research Assistant, University of Missouri-Rolla, Rolla, MO, USA)

Antonio Nanni
(V. & M. Jones Professor, University of Missouri-Rolla, Rolla, MO, USA)

Thomas E. Ringelstetter
(Graduate Research Assistant, University of Wisconsin-Madison, Madison, WI, USA)

Lawrence C. Bank
(Professor, University of Wisconsin-Madison, Madison, WI, USA)

ABSTRACT
The development of durable structural systems for accelerated bridge construction is key to reducing the economic
and social costs associated with replacement operations on a large scale. This paper reports on the field application
of stay-in-place reinforcing panels, entirely made of glass fiber reinforced polymer components and specifically
developed for the rapid construction of concrete bridge decks. The salient features of the system are illustrated,
along with significant research and development outcomes. The five-day construction of the cast-in-place deck and
open-post rail of Bridge 14802301 in Greene County, MO, is documented, and the major outcomes outlined. The
project demonstrates how lightweight and noncorrosive FRP reinforcement is a practical alternative to steel, with the
potential of versatile structural forms that add relevant constructibility and economic advantages.

KEYWORDS
Bridge Deck, Fiber Reinforced Polymers, Accelerated Bridge Construction.

1. INTRODUCTION
During the last four years, increasing investments have been made to support the research and development of
innovative technologies for accelerated bridge construction, primarily under the sponsorship of the Federal Highway
Administration (FHWA), the American Society of State Highway and Transportation Officials (AASHTO
Technology Implementation Group), and the Transportation Research Board (TRB Task Force on Accelerating
Innovation in the Highway Industry). Emphasis has been placed on improving safety and minimizing traffic
disruption while enhancing quality and durability. The issue arises from the urgent need of upgrading and
maintaining a significant portion of the bridge inventory while facing inevitable budget restrictions. Redecking
operations are rather frequent, since corrosion of steel reinforcement is a major instrument of degradation in
reinforced concrete (RC) decks and safety appurtenances. In the case of off-system bridges, cost-benefit analysis,
contractors know-how and equipment availability typically result in the adoption of either partial or full-depth cast-
in-place (CIP) technologies. The most popular solution limits the use of prefabricated elements to standardized
partial-depth precast prestressed concrete panels as structural stay-in-place (SIP) forms between the girders, with
CIP concrete topping, as opposed to traditional removable plywood forms. SIP steel metal deck forms, with a full-
depth CIP configuration that eliminates the problem of reflective cracks, are less attractive due to three major
drawbacks: a) safety concerns due to risks of accidental damage of relatively thin metal sheets, resulting in local
buckling problems under wet concrete load; b) corrosion issues under aggressive environments; c) efficient
inspection of the underside of the deck is complicated.

In the project presented herein, an innovative prefabricated glass Fiber Reinforced Polymer (FRP) SIP reinforcement

151
has been selected to construct the replacement deck of Bridge 14802301 in Greene County, MO. Corrosion resistant
FRP reinforcement gratings and SIP form plates are integrated into very large-size modular panels. The structural
form takes advantage of FRP composites tailorability and lightweight to provide improved constructibility, resulting
in enhanced construction speed and safety.

2. PREFABRICATED STAY-IN-PLACE FRP REINFORCEMENT

2.1 Description and Detailing

The FRP SIP panels are prefabricated assemblying off-the-shelf pultruded glass/vinylester components, typically
used in floor grating applications in corrosive environments, into a three-dimensional grating made of two (top and
bottom) layers (Figure 1). The main load-carrying elements are 38 mm I-bars, spaced at 100 mm on-center, which
run continuously in the direction perpendicular to traffic (transverse). Both shape and spacing of the I-bars have
been thought to allow ease of walking over the three-dimensional assembly. Three-part cross rods, spaced at 100
mm on-center and running through pre-drilled holes in the I-bars web in the direction parallel to traffic
(longitudinal), provide shrinkage and temperature reinforcement, enhance the in-plane rigidity of each reinforcing
layer, and constrain the core concrete to ensure mechanical compatibility with the structural I-bars. Top and bottom
reinforcing layers are integrated using two-part vertical connectors that space them at 100 mm on-center. The two
components forming the connectors are shaped to be epoxy-bonded to the I-bars and then fastened together. The
formwork consist of 3.2 mm thick and 1.22 m long plates that are epoxy-bonded to the I-bars in the bottom layer.

Cross Vertical Epoxy-bonded Chair at

TRAFFIC Rods I-bars Connector SIP Plate Overlap

0.3 m 1.22 m 1.22 m

(1 ft) (4 ft) (4 ft) (a) (b)

Figure 1: Prefabricated FRP SIP Reinforcement Panels: Longitudinal Section (a) and Close-up (b)

The system concept, detailing and construction procedure have been addressed to improve constructibility by
introducing original solutions when needed, and constantly seeking input from practitioners. Each SIP panel has a
width of 7.06 m, a typical length of 2.44 m (Figure 1(a)), and a weight of about 409 kg (23.7 kg/m2). The width
corresponds to that of the bridge deck minus 127 mm per side, to allow a traditional drip edge notch to be formed
on-site. The use of large-size and lightweight panels allows easy placement of the SIP reinforcement on the bridge
girders with single picks of a crane at four anchorage points. Hence, both time-consuming and labor-intensive
setting/removing of plywood forms and tying of rebars are eliminated. Adjacent panels are connected in a non-
mechanical fashion by means of 0.30 m overlaps, formed by offsetting the top and bottom grating layers (Figure
1(a)), thereby preserving a degree of continuity in the longitudinal direction (Figure 1(a) and Figure 2(a)). 3.2 mm
thick strips are inserted to cover the SIP plate-to-plate butt joints in order to prevent concrete leaking during casting
(Figure 2(b)). When using steel girders, each SIP unit is anchored to the top flanges via stainless steel threaded bolts
at every 2.44 m, keeping the bottom reinforcing layer in place with 6.3 mm thick FRP washers (Figure 2(c)). Holes
in the SIP plate are drilled on site. When composite action is sought between girders and deck, the panels can be
supplied with pre-drilled holes with longitudinal and transverse spacing of 10 cm on-center to accommodate welded
shear studs. No cambering of the panels is required to match the roadway crown, which is formed using the finishing
machine. The length and layout of the end panels are designed to fit the actual bridge length and accommodate the
expansion joints. Since glass FRP is easy to saw-cut, adjustments can be readily made on site (Figure 2(d)).

Left Panel Chair at Overlap

Plates Butt-joint Right Panel

with Cover Strip

TRAFFIC (a) (b) (c) (d)

Figure 2: Panel-panel Connection (a, b); Anchoring to Girder (c); End Panels at Expansion Joint (d)

152
The steel-free reinforcement system is completed with the prefabricated glass FRP rebar cages of a newly designed
open post Modified Kansas Corral Rail (Matta and Nanni, 2006). Cut-out pockets in the panels overhang
reinforcement facilitate insertion of the post cages at the correct spacing. The continuous top rail reinforcement is
made of either 2.44 m or 4.88 m long cages with 1.22 m rebar splices, thought to be rapidly mounted prior to rail
forming. Again, the use of lightweight FRP cages greatly simplifies handling and mounting operations, while
eliminating on-site rebar tying that is particularly labor-intensive in this case.

2.2 Research and Development

Extensive research and development work during the last 14 years has demonstrated the structural effectiveness of
pultruded FRP gratings as internal reinforcement of concrete bridge decks. Two recent pioneer construction projects
have been completed in Wisconsin, USA (Bank et al., 2006; Berg et al., 2006). The solution presented herein
features the last-generation system, and the first with fully-integrated reinforcement and SIP forms (Ringelstetter et
al., 2006). The project Special Provisions included FRP Material Specifications, in compliance with a model
specification developed for the FHWA (Bank et al., 2003). Performance Specifications were also defined for the SIP
panels by imposing stress and deformation limitations to test panels when simulating typical construction loads, i.e.
vertical and lateral loads, in-plane racking, vertical load on overlaps, and wet concrete load (Matta et al., 2005).

The FRP RC open post rail was designed following the ACI 440 guidelines (ACI, 2006) to meet the AASHTO
LRFD (AASHTO, 1998) and Standard Specifications (AASHTO, 2002). In the case of the LRFD provisions, where
a yield-line approach is recommended to evaluate the equivalent transverse static strength, deformation
compatibility was assumed to account for the lack of moment redistribution in FRP RC structures, along with
conservative failure scenarios (Matta and Nanni, 2006). In addition, the end posts located at the expansion joints and
approach deck, where rail continuity is not provided, were designed to exceed the required crash Test Level 2
strength FT = 120 kN. The deck and rail design was validated through laboratory testing of full-scale deck slabs and
rail post/deck connections, which was performed at key steps of the optimization process, and confirmed the
significant safety margin of the layout selected for the field implementation (Matta et al., 2005).

3. FIVE-DAY BRIDGE REDECKING

The old Bridge 14802301 (Greene County, MO) slab-on-girder superstructure, built in 1933, was in need of
replacement because of severe corrosion-induced degradation of deck and safety appurtenances, and increased load
requirements. The load rating was 3.9 t (2004), versus an original design based on a 9.1 t truck load with 30%
impact factor. The new superstructure has four symmetrical spans of 11.3 m (exterior) and 10.7 m (interior) length,
for a total length of 43.9 m. The cross section comprises four W610×25 steel girders spaced at 1.8 m on-center and
acting non-compositely with a 178 mm thick deck. The out-to-out deck and clear roadway width are 7.3 m and 6.7
m, respectively. The girders are continuous over two spans, with a closed expansion joint at the central support.

Transition from research and development to field implementation was conducted in coordination with the
manufacturers of the FRP deck and rail reinforcement, and the engineer of record. The construction operations were
planned with the contractor parties to minimize the amount of time and work. Construction of the RC deck and
railing from the SIP panel installation to rail casting is documented in Figure 3. The job was completed in November
2005 in five days, instead of the typical 2-3 weeks needed for similar steel reinforced bridges built by the contractor.
Installation of the deck panels was finalized in six hours during the first day by six workers. During the second day,
the 36 rail post cages were mounted, the deck details formed (expansion joints, chamfers, drip edges), and the
finishing machine was set. Deck casting and finishing was completed in the third day. The remaining two days were
used to mount the open post concrete rail top continuous cages and the formwork, and finally casting.

4. CONCLUSIONS
The first application project of a innovative prefabricated FRP reinforcement for rapid bridge deck construction has
been presented. The use of very large-size and lightweight modular stay-in-place panels, comprising a double-layer
grating with epoxy-bonded form plates and designed for improved constructibility, eliminates the need of formwork
and on-site tying of reinforcing bars. The five-day redecking resulted in over 70% reduction in deck construction
time, with a similar reduction in labor cost. Shape and spacing of the reinforcing profiles, devised to facilitate

153
walking over the three-dimensional assembly, allowed an increase of about 50% in concrete placement productivity
while improving safety and working conditions, as confirmed by the field workers.

(a) (b) (c) (d) (e)

(f) (g) (h) (i)

Figure 3: Bridge Redecking Operations: Panels Installation (a); Mounting of Post Cages (b); Deck Casting (c)
and Finishing (d); Mounting of Top Rail Cages (e); Rail Casting (f); Finished Superstructure (g-i)

A conservative cost estimate for the deck as-built is $409/m2 ($38/ft2), of which $280/m2 ($26/ft2) from the
prototype FRP panels delivered to the site. The amount increases to $483/m2 ($44.9/ft2) including the cost of the
open post railing ($271/m, $82.6/ft). The competitive potential of the proposed system is also enhanced by the
durability of FRP reinforcement, with prospective increased service life and reduced maintenance costs.

5. ACKNOWLEDGEMENTS
The financial support of the US DOT, the UMR UTC on Advanced Materials and NDT Technologies, and the
Federal Highway Administration, thorugh the Innovative Bridge Research and Construction Program, is
acknowledged. The assistance of Strongwell and Hughes Brothers, industry members of the UMR NSF I-U/CRC
“Repair of Building and Bridges with Composites” (RB2C), is also acknowledged. Thanks are due to Greene
County, MoDOT, Great River Engineering, Hartman Construction, and Master Contractors LLC, for active support.

6. REFERENCES
American Association of State Highway and Transportation Officials (1998). AASHTO LRFD Bridge Design
Specifications, 2nd edition, AASHTO, Washington, DC.
American Association of State Highway and Transportation Officials (2002). Standard Specifications for Highway
Bridges, 17th edition, AASHTO, Washington, DC.
American Concrete Institute Committee 440 (2006). ACI 440.1R-06 - Guide for the Design and Construction of
Concrete Reinforced with FRP Bars, ACI, Farmington Hills, MI.
Bank, L.C., Gentry, T.R., Thompson, B.P., and Russell, J.S. (2003). “A Model Specification for FRP Composites
for Civil Engineering Structures”. Construction and Building Materials, Vol. 17, No. 6-7, pp. 405-437.
Bank, L.C., Oliva, M.G., Russell, J.S., Jacobson, D.A., Conachen, M.J., Nelson, B., and McMonigal, D. (2006).
“Double Layer Prefabricated FRP Grids for Rapid Bridge Deck Construction: Case Study”. Journal of Composites
for Construction, Vol. 10, No. 3, pp. 204-212.
Berg, A.C., Bank, L.C., Oliva, M.G., and Russell, J.S. (2006). “Construction and Cost Analysis of an FRP
Reinforced Concrete Bridge Deck”. Construction and Building Materials, Vol. 20, No. 8, pp. 515-526.
Matta, F., and Nanni, A. (2006). “Design of Concrete Railing Reinforced with Glass Fiber Reinforced Polymer
Bars”, Proceedings of 2006 ASCE-SEI StructuresCongress, CD-ROM, 9 pp.
Matta, F., Nanni, A., Galati, N., Ringelstetter, T.E., Bank, L.C., Oliva, M.G., Russell, J.S., Orr, B.M., and Jones,
S.N. (2005). “Prefabricated Modular GFRP Reinforcement for Accelerated Construction of Bridge Deck and Rail
System”, Proceedings of FHWA Accelerated Bridge Construction Conference, pp. 129-134.
Ringelstetter, T.E., Bank, L.C., Oliva, M.G., Russell, J.S., Matta, F., Nanni, A. (2006). “Development of a Cost-
Effective Structural FRP Stay-In-Place Formwork System for Accelerated and Durable Bridge Deck Construction”.
Proceedings of 85th Transportation Research Board Annual Meeting, CD-ROM #06-2218, 16 pp.

154
Third International Conference on FRP Composites in Civil Engineering (CICE 2006)
December 13-15 2006, Miami, Florida, USA

TESTING OF A FRP DECK SYSTEM FOR MOVEABLE BRIDGES

Lei Zhao
Assistant Professor, University of Central Florida, Orlando, Florida, USA

Jignesh S. Vyas
Graduate Research Assistant, University of Central Florida, Orlando, Florida, USA

Marcus Ansley
Chief Structures Research Engineer, Florida Department of Transportation, Tallahassee, Florida, USA

ABSTRACT
Moveable bridges in Florida typically use open steel grid decks due to the weight limitation of 1.2 kN/m2. These
decks are noisy to ride on, typically less skid resistant than a solid riding surface, and costly to maintain. A new FRP
deck system providing a solid roadway surface, satisfying the weight requirements, and having the potential of
reducing maintenance cost is being investigated. Fatigue and failure tests were performed on full-size panels in a
two-span configuration. Results from the fatigue test showed no signs of degradation after 2 million cycles. The
failure tests indicate that a significant factor of safety exists when the AASHTO truck load is considered.

KEYWORDS
FRP composite deck, safety, moveable bridge, fatigue test.

1. INTRODUCTION
It has been reported that approximately 28 percent of America’s 600,000 public bridges are either structurally
deficient or functionally obsolete (FHWA, 2004). The existing quality of the highway infrastructure has been on a
decline, due to insufficient maintenance, heavy loads, unexpected or harsh environmental conditions. This problem
has created an urgent need for effective means of structural repair, rehabilitation, and replacement. As a result, there
are tremendous opportunities for FRP bridge decks that are corrosion resistant, lightweight, have a high strength to
weight ratio and can be easily and rapidly installed. Although FRP material costs are greater than traditional
concrete and steel materials, they have shown promise in applications such as decks on moveable bridges, where the
advantages of FRP outweigh its high initial costs.

Moveable bridges are bridges across waterways that may lift up or rotate out of the way to allow ships to pass. The
weight limitation on the deck for moveable bridges is typically 1.2 kN/m2, which makes open steel grid deck the
only viable option amongst conventional deck systems. However, these decks present safety and environmental
problems, for they are typically less skid resistant than a solid riding surface, create loud noises, and allow debris to
fall through the grids. Although the initial cost of construction of a steel grid deck is low, maintenance cost is very
high. FRP decks, in comparison, are not only lightweight, but also provide a solid riding surface, which has the
potential of improving driving safety, reducing noise levels, and preventing falling debris. Also, the maintenance
cost is expected to be significantly lower.

2. DECK SYSTEM DESCRIPTION

The system consists of mainly a pultruded bottom panel and a top plate. The bottom panel, which has four I-sections
and a bottom plate, is mechanically fastened to adjacent panels (at their edges) and the top plate (at the top flange of

155
the I-sections) as shown in the Fig.1. The center-to-center distance between the webs of the I-section is 203 mm.
The total deck thickness is 127 mm, which includes a 114-mm-deep bottom panel and a 13-mm-thick top plate.

Mechanical Fasteners 13

Mechanical Fasteners
13

13 114

178 203 203 203

Unit: mm
a) Cross section of the deck system b) Picture of the deck showing all components

Fig. 1 Deck Assembly

3.TEST SETUP
The prototype deck specimen, which was made of two pultruded sections and top plates, was supported by three
steel I-beams that were 1219 mm apart. Loading was applied on top of a polymer-based wear surface that was
installed on the top plate of the deck. Fatigue loading was applied simultaneously at locations shown in Fig. 2a up to
80 kN per loading pad. The locations of failure loads in FAIL 1 and FAIL 2 are shown in Fig. 2b. Two failure tests
were performed:
• Loading in test FAIL 1 was applied in one span and above two I-beams in the deck (Fig. 2c).
• Loading in test FAIL 2 was applied in the other span and above the center of the web of one of the I-
sections in the deck (Fig. 2e).
The deck needed to demonstrate—in both FAIL 1 and FAIL 2—failure loads that are above the AASHTO-required
strength of 164 kN. Deflection and strain in the deck during the tests were measured at locations shown in Fig.2. D0
through D5 and S0 through S13 represent the locations of displacement transducers and strain gages, respectively,
mounted on the bottom of the specimen.

4. TEST RESULTS

4.1 Fatigue Test

As shown in Fig. 3a, the progression of the mid-span deflection of the deck stabilized after 1 million cycles, except
for one measurement location, D4, where there was a rising trend after 1.0 million cycles. The maximum deflection
under the fatigue load grew from 3.7 mm at zero fatigue cycle to 5.2 mm at 2 million fatigue cycles. The long term
deflection growth will be studied further and field-monitored on bridges that use this system.

The wear surface showed no signs of cracking or degradation from the fatigue test.

4.2 Failure Tests

The load-deflection and load-strain relations observed from FAIL1 (Fig. 3b) were largely linear-elastic. The peak
load and the displacement at the peak load were 370.5 kN and 23.3 mm, respectively. At the peak load, the
maximum strain in the bottom plate dropped from 4800 µε to 3500 µε (Fig. 4a), indicating a delamination in the
panel. Loading was terminated to avoid damage to the other span, which was to be tested in FAIL 2.

The load-deflection and load-strain relations observed from FAIL 2 (Fig. 3b) were largely linear-elastic up to 313.6
kN, when a load-drop of approximately 12% occurred. When loading continued, the specimen was able to regain the
load level of the first peak and eventually failed from web buckling at a load of 396.8 kN and deflection of 48.9mm.
The system was able to achieve significant deflection after the first peak. Post-test inspection revealed delamination
between the web and the bottom flange of the deck as shown in Fig. 4b. The transverse displacement and strain
profiles (Fig.5) indicate that the load is not effectively distributed in the transverse direction. The I-sections directly
under the loading pad carried majority of the load.

156
 Fig. 2 Loading and Transducer Configuration

5 FAIL 2
400 (D1)
FAIL 1
(D4)
Displacement (mm)
(mm)

4 D4
300
Displacement

(kN)
(kN)

3 D1
Load
Load

200

2
D3
D5 100
D0
1 D2

4 5 6 0
1000 10 10 10 0 10 20 30 40 50
Number of Load
Number of LoadCycles (logscale)
Cycles (log scale) Displacement (mm)
Displacement (mm)
a) Fatigue b) Load-Displacement of Failure tests

Fig. 3 Overall Displacement Response

FAIL 2
400 FAIL 1
(S2)
(S10)

300
(kN)
(kN)
Load

200
Load

100

0
0 2000 4000 6000 8000
-6 -6
Strain( (10
Strains 10 ) )
a) Strain Response b) Delamination at Joint

Fig. 4 Strain Response and Failure Mode

157
 0
0
0 % P ea k

10
Displacement (mm)(mm)

Displacement (mm) (mm)

5
2 5 % P ea k

20
Displacement

Displacement
10
5 0 % P eak

30 0 % Pea k3
15
2 5% Pe ak 3
7 5 % P eak
5 0% Pe ak 3
20 40 P ea k 1 (313 .5 k N )
P ea k 2 (355 .6 k N )
P eak (3 6 8 .5 k N )
P ea k 3 (396 .9 k N )
25 50
-6 0 0 -4 0 0 -2 0 0 0 200 400 600 -60 0 -4 0 0 -2 0 0 0 20 0 400 600
L o catio n (m m ) L o catio n (m m)
Location (mm) Location (mm)

0 0
0 % Peak
1000
1000 25 % Peak
2000

(10 -6) ) 3000

))

-6
2000
(10 -6

50 % Peak Strains(10
-6
Strains(10

4000
Strain
Strain

3000 5000
75 % Peak 0% Peak3
25% P eak3
6000 50% P eak3
4000 P eak1 (313.5 kN )
7000 P eak2 (355.6 kN )
Peak (368.5 kN) P eak3 (396.9 kN )
5000 8000
-500 -250 0 250 500 -600 -40 0 -200 0 200 400 600
Location (mm)
Location (mm) Lo catio n (mm )
Location (mm)

a) Test FAIL 1 b) Test FAIL 2

Fig. 5 Displacement and Strain Distribution Profiles

5. CONCLUSION

• The deck system displayed no noticeable signs of degradation after 2 million fatigue cycles. The progression of
the deck mid-span deflections, on a log-scale, stabilized after 1 million cycles, except for one location (D4)
where deflection showed a slight rising trend. The maximum deflection under the fatigue load was 5.2 mm.
• The peak loads measured from the FAIL 1 and FAIL 2 tests were 370.5 kN and 396.8 kN, respectively, which
were 2.3 and 2.42 times the AASHTO requirement for factored load level.
• In FAIL 2, the deck system showed significant deflection capability beyond the first load drop caused by
delamination.

6. ACKNOWLEDGEMENT
The authors would like to express their appreciation to Mr. Frank Cobb, Mr. Paul Tighe, Mr. David Allen and Mr.
Steve Eudy at the Florida DOT Structures Laboratory for their technical assistance during the testing.

7. REFERENCES
FHWA, “Status of the Nation's Highways, Bridges, and Transit: 2004 Conditions and Performance Report”.

158
Part IV. Composites Systems
Third International Conference on FRP Composites in Civil Engineering (CICE 2006)
December 13-15 2006, Miami, Florida, USA

URBAN LIGHT TRANSPORT (ULTRA) GUIDEWAY PROJECT

Thanongsak Imjai
(PhD student, Department of Civil & Structrual Engineering, Sheffield University, Sheffield, UK)

Maurizio Guadagnini
(Lecturer, Department of Civil & Structrual Engineering, Sheffield University, Sheffield, UK)

Kypros Pilakoutas
(Professor, Department of Civil & Structrual Engineering, Sheffield University, Sheffield, UK)

Peter Waldron
(Professor, Department of Civil & Structrual Engineering, Sheffield University, Sheffield, UK)

ABSTRACT
This paper presents one of the case studies that was examined during the European funded CRAFT RTD Project
CurvedNFR. A 6m long concrete plank reinforced with thermosetting FRP bars as longitudinal reinforcement and
thermoplastic FRP strips as shear reinforcement was manufactured and tested in the structural laboratory of the
University of Sheffield. The use of FRPs allowed the reduction in the required concrete cover (hence a reduction in
the overall depth and total weight) and will guarantee the construction of a highly durable structure, thereby
minimising the need of structural maintenance. This study shows that current design recommendations for FRP RC
structures can be effectively used to predict deflection and crack width of the RC plank at service load. It is also
found that serviceability limit state, such as deflection and crack width, control the design criteria.

KEYWORDS
Fibre Reinforced Composite, Reinforced Concrete, Crack width, Deflection

1. INTRODUCTION
The European funded CRAFT RTD project CurvedNFR (2003) was established in 2003 with the aim of developing
materials, methodology, and manufacturing process for low-cost, curved fibre reinforced plastics (FRPs) for
concrete structures. The ULTra guideway project has been employed as a case study to design and analyse a RC
plank using FRPs as internal reinforcements. Important aspects of the project are: 1) analysis of the environmental
impact of the infrastructure; 2) visual intrusion from the overhead portions of the structure (Figure 1-right), and 3)
evaluation of the durability of the structural elements. A full scale specimen of the RC plank was cast and tested at
the University of Sheffield under different load arrangements to simulate the load conditions that the structure will
be subjected to during its service life. GFRP thermosetting bars were used as longitudinal reinforcement whilst
GFRP thermoplastic strips were used to manufacture the shear reinforcement. The preliminary design of the beam
was carried out according to the design recommendations proposed by the ACI 440 committee (ACI 440.1R, 2003).
The ULTra project is briefly introduced in the following along with the methodology that was followed for
designing and testing the FRP RC plank.

2. ULTRA GUIDEWAY PROJECT

2.1 The ULTra system

Transport by conventional means involves energy use, resource use and emission output which cannot be regarded
as sustainable. Current transport is dominated by the car. This provides the flexible personal transport required by

159
modern forms of cities, but is widely recognised as unsustainable. Current public transport is poorly accepted.
Unfortunately, analysis also shows that current forms of public transport offer little, or even negative, benefit in
sustainability over the car. A new transport system has been devised to meet the need for transport which is both
effective and sustainable. This transport system, ULTra, offers an advanced form of personal transport system that
uses a fleet of low power, electrically driven vehicles on a dedicated guideway network of routes (Figure 1).

Figure 1: Driverless automatic ULTra vehicle (left) and overhead portion of the ULTra guideway (right)

In contrast to previous forms of public transport, there is no waiting, no stopping and no transfers within the system.
In many circumstances, it can offer better transport than available by other means. ULTra has been designed to
demanding sustainability requirements. Typically, ULTra provides a reduction by a factor of 3 in energy use and
emissions output over existing forms of transport. ULTra is also complementary to existing forms of transport. By
providing a network link to major rail or bus stations, it can improve the attraction of current transport services.
Evaluations undertaken for the Department of Transport and supported by recent questionnaire studies, suggest that
a comprehensive ULTra system could attract 25-30% of present car drivers. ULTra is now undergoing engineering
development funded by the Department of Transport, Local Government and the Regions. It is believed that the
system will offer a new approach to public transport with a real prospect of significant gains in effectiveness and
sustainability.

2.2 Design concept

The geometry of the cross section of the reinforced concrete guideway has been optimised during the design and an
attempt was made to keep the overall depth of the structural elements down to minimal values to reduce visual
intrusion (Figure 2). The design process focused on providing sufficient reinforcement to resist the applied loads
(Ultimate Limit State Design) and to control deflection and cracking under operating conditions (Service Limit State
Design). The design procedure suggested in the ACI 440.3R document was employed to determine the necessary
amount of both flexural and shear reinforcement. These include bending and shear design procedures (Imjai et al.,
2004). For design considerations, standard sectional analysis was used to determine the flexural properties of the
FRP RC section. The RC section was designed to fail due to concrete crushing in compression (over-reinforced
section) as the lack of ductility of the composite reinforcement can not provide warning of impending failure. The
nominal flexural strength of the section was determined based on strain compatibility, internal force equilibrium,
and the mode of failure. According to the design calculations, it is found that the planks under flexural loads can
meet the design requirements. The shear capacity of the RC plank provided by 4x10 mm with a spacing of 50 mm
thermoplastic composite strips is optimised according to ACI code. It is also found that serviceability limit states
such as deflection and crack widths control the design.

3. TEST PROGRAMME
3.1 Beam preparation and material properties

Glass FRP thermosetting Eurocrete bars (ffu=700 MPa, εfu=0.017 and Ef=45 GPa) were used as longitudinal
reinforcing material, and shear reinforcement was provided in the form of links manufactured from FRP
thermoplastic Plytron strips (ffu=720 MPa, εfu=0.019 and Ef=28 GPa) that were produced by Plytron GmbH Ltd.
Owing to the physical characteristics of FRP, the overall weight of the reinforcing cage was only about 13.5 Kg,
which amounts to about 2% of the total weight of the concrete. By comparison, a similar reinforcing cage made of

160
steel reinforcement would weight approximately 50 Kg (8% of the total weight of the concrete). The thermoplastic
shear links were bent in the laboratory at the University of Sheffield by heating the composite with an air gun at a
controlled temperature and shaping it around a custom made mould. The geometry of the specimen is illustrated in
Figure 2b along with a schematic view of the cross section showing the reinforcement details.

(a)

(b) (c)

Figure 2: a) Reinforcement arrangement and b) cross section of the ULTra guideway. c) Detailing of the
thermoplastic shear reinforcement at one end of the plank

Foil-type electric strain-gauges were positioned at various locations along the flexural and shear reinforcement to
monitor variations in strains. The positions where the strain gauges were to be located were accurately marked on
each bar and link and the surrounding areas were appropriately prepared to guarantee a successful installation of the
gauges. Prior to the application of the gauges on the GFRP bars, glue was used to seal the surface. Cement glue was
used to attach the strain gauges to the bars and electrical wires were soldered to the terminal of each gauge for
subsequent connection to the data logger. A ready mixed concrete obtained from a local supplier was used to
manufacture the test specimens. The specifications of the mix were: concrete C40 with 10 mm maximum aggregate
size and cement type OPC with a slump of 100 mm.

3.2 Test set-up

Figure 3 shows the loading patterns to which the beam was subjected during two successive phases of testing. Load
case 1 (Figure 3a) was applied to generate the maximum positive bending moment in the RH span, whilst the load
case 2 (Figure 3b) was applied to generate the maximum negative moment over the central support. In both cases,
the load was applied in increments of about 1 kN. At each load step, cracks were marked and the widths of selected
target cracks were measured. Overall deflections of the beam were measured at different locations using several
Linear Variable Displacement Transducers (LVDTs).
RH LH RH LH

7 7
12 12
11 11

4 10 4 10

8 8
6 6
5 5

3 3

9 1

13 13

(a) (b)

Figure 3: Test set-up and instrumentation for a) load case 1 and b) load case 2

161
4. DISCUSSION OF THE RESULTS

Three load cycles were performed at load levels corresponding to: a) the load induced by standard passenger-
carrying vehicles (service load 1, about 5 kN); b) the load induced by a road sweeper machine (service load 2, about
10 kN); and c) the design load (1.5 times the maximum service load, about 15 kN). In the case of load case 2, the
load was further increased to about 50 kN with no severe repercussions on the structural integrity of the beam. The
load-displacement behaviours for both load cases are shown in Figure 4. The overall behaviour of the beam and the
crack pattern observed during testing confirmed that adequate shear provision was achieved by using the custom
made thermoplastic links (see Table 1).

16
Load (kN)

60

Load (kN)
Design load
14
50
12
Service load 2 40
10
8 30
6 Service load 1
20
4 Design load
Service load 2
2 10
Service load 1
0 0
0 2 4 6 8 10 12 0 2 4 6 8
Displacement (mm) Displacement (mm)

Figure 4: Load-deflection response of ULTra beam: load condition 1 (left) and load condition 2 (right)

Table 1 shows a comparison of the results obtained from the test performed during the first phase of testing (load
case 1) with the values predicted according to the recommendations proposed by the American Concrete Institute for
the design of concrete structures reinforced with Fibre Reinforced Polymer Reinforcement (ACI 440.1R.03). This
table shows clearly that conservative values are generally predicted by the design recommendations and that the
tested FRP reinforced beam meets all of the serviceability requirements.

Table 1: Test results and design equation predictions: Load condition 1

Serviceability limits Predicted values Experimental values

Max crack Max Max crack Max Max crack Max
Load stage width deflection width deflection width deflection
(mm) (mm) (mm) (mm) (mm) (mm)
Service load 1 0.5 12.5 0.15 4.75 0.10 1.3
Service load 2 0.5 12.5 0.29 9.19 0.25 3.6
Design load - - 0.41 11.97 0.35 8.8

5. REFERENCES
American Concrete Institute (ACI). (2003). "Guide for the Design and Construction of Concrete Reinforced
with FRP Bars ACI 440.1R-03", ACI Committee 440, Farmington Hills, MI, USA.
CurvedNFR (2003). "Cost effective Curved Polymer Composite Rebar". CRAFT RTD European funded project,
CRAFT GIST-CT-2002-50365, http://www.curvednfr.com
Imjai T., Guadagnini M., Pilakoutas K., (2004). Case study: Urban Light Transport (ULTra) guideway Project,
Technical Report, Centre for Cement and Concrete, The University of Sheffield, Sheffield, UK, pp. 23.

162
Third International Conference on FRP Composites in Civil Engineering (CICE 2006)
December 13-15 2006, Miami, Florida, USA

COMPOSITE SANDWICH WALL PANELS FOR BUILDING

APPLICATIONS
Waleed Shawkat
Master of Sceince Candidate, Queen’s University, Kingston, Ontario, Canada

Hart Honickman
NSERC Undergraduate Researcher, Queen’s University, Kingston, Ontario, Canada

Amir Fam
Associate Professor and Canada Reseach Chair in innovative and retrofitted structures, Queen’s University,
Kingston, Ontario, Canada

ABSTRACT
This study examines the structural performance of a novel composite panel consisting of a polyurethane foam core
sandwiched between two layers of carbon fibre-reinforced polymer (CFRP) skin. Ten panels were tested as one-way
slabs, using both four-point bending and three-point bending configurations. The study was intended to examine the
consistency and repeatability of test results of similar panels, as well as investigate the effects of shear span-to-
thickness ratio, moisture absorption, and the effect of a masonry pattern coating applied to one face of the panel. The
flexural strengths and stiffnesses and various failure modes have been evaluated. The panels were also compared to
conventional reinforced concrete (RC) panels.

KEYWORDS
Composites, Sandwich Panel, FRP, Polyurethane, Cladding.

INTRODUCTION
The primary purpose of exterior cladding is to provide shelter from rain and snow, as well as providing thermal
insulation. It is also responsible for transfering wind loads to the internal skeleton of a building. Since conventional
cladding is composed of steel reinforced concrete, corrosion is a major concern especially if the cladding is placed in
a harsh environment. The new sandwich panels are polymer based, thus alleviating this concern. Also, they are
significantly lighter in weight.

The panels are constructed using a rigid polyurethane foam core sandwiched between two layers of carbon fiber
reinforced polymer (CFRP). The CFRP layers resist the tensile and compressive stresses resulting from the flexure
induced by wind loading. The polyurethane core resists shear stresses and also contributes to the moment of inertia
of the panel’s cross section by acting as a spacer that positions the CFRP skins further apart.

In order to fabricate these panels, the CFRP skins are first adhered to the upper and lower faces of a formwork. The
polyurethane foam is then injected into the void between the CFRP skins as a two-part liquid. As the two parts of
this liquid react, the chemical expands into a foam-like substance, hardens, and adheres to the CFRP skins. Because
of this expansion, the polyurethane foam applies pressure to the CFRP skins as it hardens, which helps improve the
bond.

Five large panels were trimmed and each was cut in half in order to create ten specimens for testing. Three of the
panels were approximately 68 mm thick and had a smooth white marine grade gel-coat applied to the outer surface
of the CFRP skins, while the remaining two panels were approximately 55 mm thick and had a masonary-like
patterned coating on one face. Table 1 shows partner specimens cut from the same panel.

163
This paper includes brief description of the test specimens, test parameters, test setups, instrumentation, and test
results. The paper also includes evaluation of flexural strength and stiffness of the panels and comparisons with
conventional reinforced concrete panels.

TEST SETUPS, INSTRUMENTATION AND PROCEDURES

Specimens #1, 2, 3, 5, 6, 9, and 10 were tested in four-point bending. A span of 1400 mm and width of 300 mm was
used for specimens #1, 2, 3, 5, and 6 and a span of 1360 mm was used for specimens #9 and 10 with the full width
of 320 mm being loaded using two line loads applied at one and two thirds of the span of the panel. The line loads
were applied using rollers resting on rigid steel strips. Teflon sheets were placed under the steel strip in order to
protect the panels from stress concentrations. The specimens were supported along the full width at both ends.
Longitudinal strains in the CFRP skins were monitored using four strain gages. They were configured such that
there were two strain gages at the mid span of the top face (compression) and an identical setup on the bottom face
(tension) of each panel. Mid-span deflections of each panel were monitored using two linear potentiometers.

In order to increase the chances of flexural failure (rather than shear failure) specimens #4, 7, and 8 were tested in
three-point bending. A span of 1400 mm was used for specimen #4 and a span of 1360 mm was used for specimens
#7 and 8, with loads applied at the midspan of the panels. Instrumentation was similar to the aformentioned tests.

Prior to its flexural test, specimen #6 underwent a water absorption investigation during which it was completely
submerged under fresh water for approximately 90 days. The specimen was periodically removed briefly from the
water in order to measure its weight so that the water absorption rate could be monitored. This procedure was
derived from ASTM C272-01 and ASTM D570-98.

TEST RESULTS, FAILURE MODES AND EFFECT OF VARIOUS PARAMETERS

Table 1 provides a summary of the test types and results in terms of the span (L), the ultimate load (Pu), ultimate
moment (Mu), ultimate deflection (δu), ultimate strains in the upper and lower CFRP skins (εu), and failure modes.

As shown in Fig. 1(a), the ultimate loads and deflections of specimens #2 and 3 suggest good consistency within the
panel from which they were cut. Both specimens ultimately failed by outward buckling of their upper CFRP skins.
The ultimate loads, deflections ,and failure modes of specimens #7 and 8, however, were quite dissimilar from
eachother, see Fig. 1(a). Specimen #7 exibited outward buckling of its upper skin, whereas specimen #8 exhibited a
shear failure within its polyurethane core.

Specimens #1, 2 and 5 were cut from three different panels, yet they were all from the same production batch. A
comparison of the load-deflection responses of the three specimens is provided in Fig. 1(a). The ultimate loads and
deflections of specimens #1, 2 and 5 suggest significant inconsistencies between the panels. It should be noted that
specimens #1 and 5 both failed in shear within the foam at comparable load levels, which suggests comparable
quality; whereas specimen #2 failed by CFRP skin buckling in compression at a substantially higher load.

Table 1: Summary of Test Specimens and Results

Specimen 1 2 3 4 5 6 7 8 9 10
Partner 6 3 2 5 4 1 8 7 10 9
Test Type 4-Point 4-Point 4-Point 3-Point 4-Point 4-Point 3-Point 3-Point 4-Point 4-Point
span mm 1400 1400 1400 1400 1400 1400 1360 1360 1360 1360
width mm 300 300 300 300 300 300 319 319 320 320
Pu kN 5 13.1 14 18.65 6.4 7.7 7.1 4.25 12.6 7.9
δu mm 19.4 24.6 24.8 36.3 15.8 30.9 28.9 48.1 51.6 29.75
Mu kN.m 1.17 3.06 3.27 6.53 1.49 1.80 2.41 1.45 2.86 1.79
Failure Foam Skin Foam
flexure flexure flexure flexure shear shear flexure
Mode shear buckling shear
εu top -1670 -3148 -3180 -6810 -1384 -1740 -2030 -600 -1800 -2720
εu bottom 1120 2775 2967 5945 1461 1941 3324 2904 4340 1137

164
Effect of Shear Span-to-Depth Ratio (a/t)

The effect of shear span (a)-to-depth (t) ratio (a/t) can be studied through specimens #4 and 5. As shown in Table 1,
specimen #4 resisted substaintially higher moment than specimen #5. This is attributed to their respective failure
modes. Specimen #5 failed in shear, whereas specimen #4 was the only specimen in this study to experience a pure
flexural strength failure mode by crushing of the upper CFRP skin. Therefore, the maximum strain of the upper skin
of specimen #4 represents the compressive failure strain of the CFRP skin since it was utilized to its full capacity.
This was certainly a more ideal performance than other specimens that failed prematurely in shear or skin buckling
because it demonstrated the true potential of these composite panels. It should be noted that the shear force at
ultimate flexural failure of specimen #4 was substantially higher than that of specimen #5 at its shear failure, yet
specimen #4 did not fail in shear. This can likely be attributed to poor foam quality, which may have triggered an
early shear failure in specimen #5. The performances of specimens #4 and 5, which were cut from the same panel,
also indicates a lack of quality consistency within panels.

The effect of (a/t) can also be studied through specimens #7 and 9, although these specimens were cut from different
panels. As shown in Table 1, specimen #7 and 9 had somewhat comparable moment resistances despite their
dissimilar failure modes. Specimen #7 failed by upper skin buckling, whereas specimen #9 failed in shear.
Undulations in the upper skin of specimen #7 (caused by the application of its masonry-patterend coating) may have
invited a buckling failure of the upper CFRP skin prior to reaching its compressive strength.

Effect of Moisture Absorption

Comparing specimens #1 and 6, which were cut from the same panel and both were tested under identical four-point
bending configuration. Specimen #1 was tested in dry conditions, whereas specimen #6 was submerged in water for
about three months and then tested while still wet. A comparison of the load-deflection responses of the two
specimens is provided in Fig. 1(a). It is quite interesting to notice that the ultimate load of specimen #6 was higher
than that of specimen #1, despite its moisture absorption. It should be noted, however, that failure modes were
different. Specimen #1 failed in shear as described earlier, whereas specimen #6 failed by outward buckling of the
compression CFRP skin. This failure mode is in fact quite similar to those of specimens #2 and 3, which were tested
also in four-point bending but failed at substantially higher loads. The responses in Fig. 1(a) indicate that specimen
#6 was also slightly stiffer than specimen #1. It is difficult to assess the effect of moisture on behavior in this case,
due to the potential for large variability among the specimens.
20 7
#4
18
6 RC Panel (b), Equivalent Strength
16
#3 5
14
Moment (kN.m)

#9
Specimen#4
12 #2
4
P (kN)

10
3
8 #10
#6
#5
6 #7 2 RC Panel (a), Equivalent Stiffness
#8
4 #1
1
2

0 0
0 10 20 30 40 50 60 0 0.1 0.2 0.3 0.4 0.5 0.6
δ (mm) Curvature, ψ (rad/m)

(a) Load vs. Deflection for Tested Specimens (b) Composite Panel vs. Virtual RC Panels
Figure 1: Flexural Behaviour of Wall Panels

Effect of Orientation of Masonry Pattern Coat

Specimens #9 and 10 were produced from the same large panel and were both tested using an identical four-point
bending configuration, except that the masonry pattern of specimen #9 was located on the top (in compression), to
simulate a wind pressure on a building, whereas that in specimen #10 was located on the bottom (in tension), to
simulate a wind suction condition. A comparison of the load-deflection responses of the two specimens is provided
in Fig. 1(a). The ultimate loads of specimens #9 and 10 and the deflections at ultimate suggest that having the

165
masonry pattern in compression certainly has a strengthening effect. Specimen #9 failed in shear within the
polyurethane foam as discussed earlier, however specimen #10 failed by buckling of the upper CFRP skin at a lower
load. It is clear that the presence of the masonry pattern in compression have prevented the skin buckling in
specimen #9 and forced the shear failure mode. Fig. 2 shows a summary of failure modes.

(a) Shear failure (b) Skin buckling (c) Skin crushing

Figure 2: Failure Modes

COMPARISON BETWEEN SANDWICH AND REINFORCED CONCRETE PANELS

An analytical model was used to predict the flexural responses of virtual RC panels, designed according to CSA
A23.3 to be compared to specimens #4. The RC panels were assumed to be 100 mm thick and reinforced with a
single layer of rebar at their mid-thickness. The concrete compressive strength and steel rebar yield strength are
assumed 35 MPa and 400 MPa, respectively. The width of the RC panels was assumed equal to that of the
composite panels, 300 mm. The objective of this study was to establish the steel reinforcement configuration in the
RC panels that would be required to satisfy two different criteria, namely (a) the same flexural stiffness after
cracking and (b) the same ultimate moment capacity as the composite panels, see Fig. 1(b).

For the RC panels, Response2000 (Bentz, 2000) was used to establish the moment-curvature responses. It was found
that 180 mm2 steel reinforcement is needed to satisfy the similar stiffness criterion for RC panel (a). This reinforced
ratio is equivalent to 6 No.10/m. RC panel (b), on the other hand, was designed to provide the same moment
capacity, by using 390 mm2 steel reinforcement, which is essentially 13 No. 10/m. In this case, the stiffness is higher
than that of the composite panel and the ductility of the RC panel (b) is substantially lower than that of RC panel (a).

CONCLUSIONS
This study has shown that composite sandwich panels have a great potential and are quite promising as an
alternative for reinforced concrete cladding systems. It was noticed that the structural performance is sensitive to the
quality of the panel, particularly the integrity of the polyurethane foam core. The following conclusions are drawn:

1. The composite panels tested in this study are 6 to 7 times lighter in weight than conventional reinforced concrete
(RC) panels of the same size and are 14 times lighter than a typical 100 mm thick conventional RC panel of the
same flexural strength or stiffness.
2. The masonry pattern coat provided a strengthening effect and forced a shear failure mode when positioned in
compression (i.e. simulating wind pressure loading), as the flexural strength was substantially higher than that of
a panel having the masonry pattern on the tension side (i.e. simulating a wind suction effect), which failed by
buckling of the CFRP skin in compression. There was, however, no noticeable effect on the stiffness.
3. The most common failure modes were the outward buckling of the CFRP skin in compression and shear failure
within the polyurethane foam core by diagonal tension.

AKNOWLEDGMENT
The authors wish to aknowledge financial support provided by Res-Precast Inc and Materials and Manufacturing
Ontario (MMO).

REFERENCES
American Society for Testing and Materials, ASTM standards 2005
Bentz, E. (2000) Response 2000, Computer Program for reinforced concrete sections, University of Toronto.
Code for the Design of Concrete Structures for Buildings, Canadaian Standard Association (CSA), CSA A23.3-94

166
Third International Conference on FRP Composites in Civel Engineering (CICE 2006)
December 13-15 2006, Miami, Florida, USA

BEHAVIOR OF JUTE/ EPOXY COMPOSITE

CURVED I-BEAM UNDER BENDING LOADING

Asad A. Khalid

Department of Mechanical Engineering, Faculty of Engineering,

IIUM International University Malaysia, 53100, Gombak, Malaysia

ABSTRACT
In this paper experimental work on Jute/ epoxy composite curved beam subjected to four point bending has been
carried out. Two, Four, six and eight layers of woven fabric jute/ epoxy composite curved I-beams were fabricated by
hand lay-up moulding fabrication process. The test specimen is a continuous Curved I-beam consisting of a central
circular 90o region connected by two 200 mm straight legs. The load-displacement response was obtained and the
energy absorption values were calculated for all the composite curved I-beams. Glass/ epoxy composite curved beams
were also fabricated and tested for comparison purpose. The woven fabric composite curved beams mechanical
properties have been obtained from tensile tests. Results from this investigation show that the bending load required for
jute/epoxy composite curved I-beams under four point bending was less 21% than glass/ epoxy. The first crushing
loads difference between jute and glass/ epoxy fell in the range of 19%-24%. It has been found that the failure
mechanism of jute/ epoxy specimens is similar to that of glass/ epoxy. Specimens failed after the initial crushing load
localized at the lower part of the beam center region. Then, it followed by the matrix and fiber break. Delamination was
also occurred on both materials tested.

KEYWORDS: Bending loading; Jute fiber; Glass fiber; Epoxy; Composite curved beam.

1. INTRODUCTION
Composite materials have a wide range of applications in aerospace, automotive and marine structures, because of their
high stiffness and strength with respect to their weight. In addition, composite materials have high corrosion resistance,
thermal resistively and considered as non- conductive materials. Since the 1990s, natural fiber composites are emerging
as realistic alternatives to glass-reinforced composites in many applications [1]. Natural fibers possess many advantages
over glass fibers, such as lower density, lower cost and recycle ability, they are not totally free of problems On the
other hand more carbon dioxide neutrality of natural fibers is particularly attractive. So far good number of automotive
components previously made with glass fiber composites are now being manufactured using environmentally friendly
composites [2]. Natural fibers come from renewable resources and are relatively inexpensive. These fibers are now well
recognized to impart good reinforcing capability to composites. While their tensile strengths and moduli are generally
inferior to those of polymeric fibers, they often exhibit significantly larger elongation giving them better damage
tolerance [3]. Jute Natural fiber is one of the most widely used natural fibers and is very easy cultivated. Jute is a
promising reinforcement for use in composite on account of its low cost, low density, high specific strength and
modulus, no health risk, easy availability and renews ability. Among all the natural fiber reinforcing materials Jute is

167
Third International Conference on FRP Composites in Civel Engineering (CICE 2006)
December 13-15 2006, Miami, Florida, USA

relatively inexpensive and commercially available in the required form [4]. The concept of curved beam construction
typically involves stiff and strong face sheets bonded to a thick low-density core. I-beam is widely used in weight
reduction. These shapes very often contain curved segments. The simplest case of such a structure, i.e., a curved
sandwich beam loaded in pure bending [5].
The main objective of this present investigation is to study the behavior of Jute/ epoxy composite curved I-beam under
four point bending. Also to compare the results obtained with glass/ epoxy Curved beam. The comparison will include
the effect of bending load on displacement, flexural stress-strain relation, strains at the mid-span of the beam. failure
mechanism of the tested composite curved I-beams have been studied also.

2. EXPERIMENTAL WORK
2.1 Materials and Specimens Fabrication
The test specimen was designed based on four-point flexure principle as outlined in ASTM Standard D-6415-99 [6].
Curved I-beams were fabricated by using foam mould covered with sheets of plastics. Hand lay-up molding process has
been used to fabricate the specimens. Woven roving jute and glass fibers were used as a reinforced materials and matrix
of epoxy resin (Leco 811-563-103) with hardener (Leco 811-563-104) as matrix material with 8:1 ratio, respectively
Curved Composite I-beam was of 50 mm height and 50 mm flange width.. The test specimen is a continuous Curved I-
beam consisting of a central circular 90o region connected by two 200 mm straight legs tensile test specimens were
fabricated also to obtain the required mechanical properties. The mechanical properties for jute/epoxy specimens were
E11=E22= 17.68 GN/m2, E33= 8.97 GN/m2, G13=G23= 2.86 GN/m2, G12= 3.02 GN/m2, ν13=ν23= 0.24 and ν12= 0.29.
While the properties for the glass/epoxy specimens were E 11=E22= 52.25 GN/m2, E33= 12.33 GN/m2, G13=G23= 4.2
GN/m2, G12= 3.58 GN/m2, ν13=ν23= 0.25 and ν12= 0.35. The densities of jute and glass fibers were 1.6 and 2.12 kg/m3
respectively. Samples of the fabricated moulds and composite curved I-beams are shown in Figures 1 and 2
respectively.

(a) glass/ epoxy (b) Jute/ epoxy

Fig. 1. Samples of the fabricated moulds Fig. 2. Samples of the fabricated composite curved I-beams

2.2 Test Set-up

A computer controlled servo-hydraulic instron machine type 5582 with a load capacity of 100 kN has been used to
perform the quasi static bending load. The crosshead speed was adjusted at 5 mm/min, Figure 3 show the test set-up of
the composite I-curved beam under four-point bending. On the other hand, two strain gauges, each of 3 mm gauge
length was fixed at the middle of the beam on the bottom and upper surfaces to obtain the longitudinal strain variation.
The strains gauges were bonded to the surface of the tube using CN adhesive.

168
Third International Conference on FRP Composites in Civel Engineering (CICE 2006)
December 13-15 2006, Miami, Florida, USA

Fig. 3. Test set-up for the curved I-beam

3. RESULTS
Results from this investigation include the load-displacement response for woven roving Jute/ epoxy and glass/ epoxy
composite curved I-beams under four point bending. The effect of materials used on initial failure load was
investigated. Strain-load relations at the middle surface of the beams were also drawn. Photographs of the tested curved
I- beams during loading have been taken to composite tube at the first failure load. Due to the page limitation, results
are summarized below.
3.1 Load-displacement
The load displacement relation has been obtained directly from the testing machine. A universal testing machine
( Digital Instron Model 8500) with 250 kN load capacity has been used. These tests were all carried out at a speed of 5
mm/min. The effect of number of layers on the initial crushing load is shown in Figure 4. Two, four, six and eight
layers of composite curved I -beams were considered for the study. Figure 5 show the strain-load relation up to initial
failure of the tested composite curved beams.

Bottom (Glass/ epoxy)

Bottom (Jute/ epoxy)
30 Jute/ epoxy 15 Top (Glass/ epoxy)
First crushing load (kN)

25 Glass/ epoxy Top (Jute/ epoxy)

10
20
5
Strain (%)

15
0 Load (kN)
10
0 2 4 6 8 10 12 14 16 18 20
5 -5

0 -10
0 2 4 6 8 10
Num ber of layers -15

Fig. 4. Effect of number of layers on Figure 5. Strain-load relation up to first crushing

the first crushing load is shown in of the tested composite curved beams
3.2 Failure mechanism
Failure mechanism of the fractured Jute and glass/ epoxy composite curved I-beams under bending have been taken and
discussed. A photograph sample of the fractured Jute and glass/ epoxy specimens is shown in Figure 7. As shown from
this Figure, Specimens failed after the initial with highly localized stresses at the lower and upper parts of the beam

169
Third International Conference on FRP Composites in Civel Engineering (CICE 2006)
December 13-15 2006, Miami, Florida, USA
center region. Then, it followed by the matrix and fiber break. A distinguish delamination was also noticed on both
materials tested.

(a) glass/epoxy (b) Jute/ epoxy

Fig. 6. Photographs of fractured composite curved I-beams under four point bending loading
5. DISCUSSION AND CONCLUSIONS
Composite curved beams were fabricated carefully using hand lay-up method. Three tests were done for each type in
order to get better results consistency. Load-displacement graphs and the effect of composite material used have been
examined. Longitudinal strains on top and bottom surfaces were examined also and drawn for both materials tested.
The main conclusions that could be drawn from this investigation include:
1. It has been found that for both glass/ epoxy and jute/ epoxy composite curved I-beams, the load increase linearly
with displacement until it reaches initial crushing load. The material strength and load decrease after initial crushing
load until the final failure of the specimen.
2. Result shows that the first crushing load for the tested glass/ epoxy was higher 20%-31.6% than jute/epoxy beams.
3. Based on the results obtained, it can be concluded that the specific absorption energy (SEA) is slightly higher for
glass/ epoxy composite curved I-beam than the jute/ epoxy beams.
4. Failure mechanism of Jute/ epoxy curved beam is similar to the glass/ epoxy curved beams. Fiber break and matrix
cracking were found besides fiber fragmentation in the top surface at the mid-span of the beam. Delamination was
occurred in the lower surfaces.
5. Jute/ epoxy can be considered for composite structures, as glass/epoxy. While the glass/epoxy beam support loads
higher 25% than Jute/ epoxy beams. Further analysis on moisture absorption is needed for both materials.
6. The average percentage of strains for Jute/ epoxy beams were found higher 18% than glass/epoxy beams at the
bottom surfaces of the beam. While the difference was 22% at the top surface of the beam.
REFERENCES
1. Joshi S. V., Drzal L. T., Mohanty A. K. and Arora S., Are natural fiber composites evironmentally superior to glass
fiber reinforced composites, Composites A, vol. 35, 2004, 371-376.
2. Paul Wambua, Jan Ivens and Ignaas Verpoest. Natural fibers: can they replace glass in fiber reinforced plastics.
Composites Science and technology, 63, (2003) 1259-1264.
3. Krishnan Jayaraman and Debes Bhattacharya. Mechanical Performance of wood fiber- waste plastic composite
materials.
4. T. Munikenche Gowda, A.C.B. Naidu and Rajput Chaya. 1999, Some mechanical properties of untreated jute fabric-
reinforced polyester composites. Composites A, 30, 1999, 277-284.
5. Andrew M. Layne and Leif A. Carlson. Test method for measuring strength of a curved sandwich beam.
Experimental mechanics, vol. 42, No. 2, 2002
6. ASTM standard D 6415-99. Standard test method for measuring the curved beam strength of a fiber-reinforced
polymer matrix composite. American society for testing and materials, West Conshohocken (2001).

170
Third International Conference on FRP Composites in Civil Engineering (CICE 2006)
December 13-15 2006, Miami, Florida, USA

LOAD CARRYING CAPACITIES

OF PULTRUDED FRP SHEET PILING
Yixin Shao
Associate Professor, McGill University, Montreal, Quebec, Canada

Fabien Darchis
Graduate Research Assistant , McGill University, Montreal, Quebec, Canada

ABSTRACT
Buckling initiation and post buckling behaviour of pultruded FRP sheet pilings subject to uniform earth pressure
were studied using a finite element method in an attempt to predict the load-carrying capacities of the pilings. The
FE predictions were compared with full scale sheet pile tests. It was found that, shell connection model was
sufficient to assess the performance of single panel in a connected sheet pile wall. The bifurcation, buckling
initiation and post-buckling failure occurred in a close range of loading, suggesting that there existed no post-
buckling capacity for the typical installation. With 10% additional materials added either to increase the thickness of
the compressive flanges or to create the junction stiffeners between flanges and webs, the buckling resistance of the
same profile design could be doubled.

KEYWORDS
Pultruded composites, fiberglass sheet piling, load capacities, buckling.

1. INTRODUCTION

Connected sheet piles are often used as waterfront retaining structures to prevent the erosion of the land. Corrosion
resistant FRP sheet piling is finding its way into this unique application. Although pultruded composites have shown
excellent tensile capacity, their relatively low elastic moduli and thin walled section design make the FRP sheet
piling vulnerable to the local and global buckling. To design a sheet pile section with desired load capacity,
understanding the buckling process of FRP sheet pile subject to an earth pressure is essential.

Analytical approaches have been developed to evaluate the buckling of a composite section using the compressive
plates restrained by boundary conditions (Kollar, 2003). These approaches require knowledge of the restraints
between different plates and the governing equations. They are suited to the estimation of the local buckling in a
plate, but difficult to be used to predict the post-buckling behaviour, especially at the presence of nonlinear
geometric deformation. Finite element approach appears to be promising in buckling analysis of a complex
composite section for pre-buckling, buckling initiation and post-buckling process. FEA calculates directly the load
capacity of the structure with exact section shape and takes into account the geometric non-linear deformation. The
approach has been successfully employed in assessing the post-buckling behaviour of the pultruded I-beams (Bank
and Yin, 1999) and the progressive failure of the pultruded box-beams (Palmer, et al, 1998). This paper is to present
a study on the finite element analysis of a pultruded FRP sheet piling subject to a uniform earth pressure using the
finite element code of ABAQUS/STANDARD. The main objectives are to evaluate the structural behaviour of the
sheet pile panels, including the pre-buckling, buckling initiation and post-buckling performances, and to propose
design modifications for enhanced load bearing capacity and serviceability performance of the panel.

2. FINITE ELEMENT MODELING

The earth pressure carried by the sheet pile wall is simulated by a uniform pressure so that the numerical results can
be compared with experiments. Fig. 1 shows a simply-supported sheet pile panel loaded by a uniform pressure in a

171
span of 2.13 m. In experiments, the air pressure was transferred to the panel through the foam inserts. Detailed
description of the test setup is given elsewhere (Shao and Shanmugan, 2006). In numerical simulation, the pressure
was applied directly to the nodes in vertical direction. The pultruded fibreglass reinforced polyester (FRP) sheet
piles are designed and manufactured by IBP Corporation of Canada. The profile of a single unit has a symmetric
double-Z cross section, 127 mm deep, 425 mm wide from pin-to-eye connections, and with a wall thickness of 4.7
mm in the top (main) flange and 3.2 mm elsewhere. The mechanical properties are summarized in Table 1. E11, T11
and C11 are the longitudinal modulus, tensile strength and compressive strength, and E22, T22 and C22 the transverse
modulus, tensile strength and compressive strength, respectively.

Sheet pile

Air bag

Fig. 1: Sheet pile panel and loading configuration

Nonlinear finite element ABAQUS/STANDARD package was employed with 92000 shell elements to evaluate pre-
buckling, buckling initiation and post-buckling of FRP sheet pile under pressure. The nonlinear static analysis was
used to establish the probable buckling load, the eigenvalue extraction method to determine the bifurcation load and
buckling modes, and the nonlinear static calculation with imperfection and dynamic perturbations to perform post
buckling analysis.

Table 1: Material properties

E11(GPa) E22(GPa) υ12 G12(GPa) T11(MPa) T22(MPa) C11(MPa) C22(MPa)

Top flange 31.4 7.3 0.18 3.1 422 70 314 79
Web 16.1 12.3 0.2 2.7 180 119 178 99
Bottom flange 30.2 7.9 0.18 3.0 298 84 250 66

3. LOAD CAPACITIES BASED ON FEA RESULTS

3.1. Nonlinear Static Analysis and Comparison with Experiments

The sheet pile model with shell connection was progressively loaded by a pressure up to 50 KPa in an increment of
2 KPa. During the monotonically loading, computation stopped at a bifurcation load of 42 KPa after which the
solutions to the nonlinear elastic equations were no longer unique. In order to validate the simulation, the pressure-
deflection curves and pressure – strain curves of the numerical pre-buckling calculations were compared to the
experimental data of single panel tests. They are presented in Figs. 2 and 3. The FEA results agreed well with
experimental data. Therefore, the shell-connection model appeared to be sufficient to predict the behaviour of the
single panel in a sheet pile wall.

60 50
Pressure (kN/m )

Eye flange
2

Pin flange Main flange

45
50
Shell connection model bifurcation loading
40

40
Experimental failure 35
Pressure [kPa]

30

30 25
Experimental deflection B4
Shell-connection model (main flange)
20
20 Shell-connection model (pin flange)
Shell-connection model deflection
15
Shell-connection model (eye flange)

10
B4 exp (main flange) 10
B4 exp (pin flange)
5
B4 exp (eye flange)
0 0
0 10 20 30 40 50 60 70 -10000 -8000 -6000 -4000 -2000 0 2000 4000
Deflection [mm]
Strain (x10-6)

Fig. 2: Comparison of pressure-deflection curves Fig. 3: Comparison of pressure - strain curves

172
3.2. Buckling Load and the Corresponding Buckling Mode

To determine the buckling loads and their corresponding buckling modes, a buckling extraction calculation was
performed on the deformed base state of the panel at 40 KPa. The first mode consisted of combined buckling waves
on both pin flange and pin web, whereas the second mode appeared also in combined waves on eye flange and eye
web. It was found that the two buckling loads corresponding to the first two buckling modes were so close that it
was impossible to predict with certainty which side would buckle first. Experimental results showed that buckling
could happen on either side (Shao and Shanmugan, 2006). Fig. 4 compares the FEA first buckling mode with
experimental observation. The buckling loads predicted by FEA were about 43 kPa. In experiments, the buckling
pressure of the single panels was found in a range of 42 - 48 KPa, with an average of 45 KPa. In the buckling
process, eye and pin connections acted as restrained end conditions for bottom flanges: because they were a lot
thicker than the bottom flanges, the connections did not participate in the buckling wave. The junction ends (bottom
web or flange) played a similar role: the junctions between webs and flanges were restrained end conditions for the
adjacent plates.

Fig. 4: Comparison of FEA first buckling mode (left) with experiment (right)

3.3. Post-Buckling Capacity

To investigate the post-buckling capacity of sheet pile after buckling at 43 kPa, post-buckling analysis was
performed by introducing geometrical imperfection and by introducing a loading rate. For the first method, the two
first buckling modes found by the buckling calculation shown in Fig. 4 were superimposed in the initial geometry.
Different levels of perturbation were investigated at 1.0 mm, 0.5 mm and 0.3 mm; the value of which represented
the upper bound of the geometric imperfection. As regard to the second method, several loading rates were applied
to the panel: 5 kPa/s, 2 kPa/s and 0.3 kPa/s. The last loading rate represented the loading rate in the single panel tests
and could be used to simulate the quasi static post buckling curves. The pressure applied ranged from 0 kPa to 60
kPa. Six different calculations, three with imperfection and three with dynamic loading, gave very similar results; in
each case, the same post buckling wave pattern appeared at the two junctions (web / pin flange and web / eye
flange). Fig. 5 demonstrates the deformed shape and the typical stress distribution in transverse direction (σ22) in
web at pin flange and web junction where the compressive stress (σ22) at web first reached to its ultimate
compressive strength.
60

Dymanic: 0.3 kPa/s 2.0 kPa/s 5.0 kPa/s Pressure [kPa]

50

40

Imperfection: 0.3 mm
0.5 mm
C22 =100 MPa in the Web

30
1.0 mm

Compression, C22
20

10

0
-4.0E+08 -3.0E+08 -2.0E+08 -1.0E+08 0.0E+00
Stress [Pa]

Fig. 5: Mid-section transverse stress distribution (σ22) at junction of pin flange and web

173
High stress concentrations over small areas appeared after buckling which would lead to localized failure of the
panel. At buckling of pin flange, the six curves started to separate and showed distinct buckling behaviours. The
more the imperfection was introduced in the panel, the sooner it would buckle. In contrary, the higher the loading
rate, the more the resistance to buckling. The perfect static post-buckling curve was assumed to lie between the 0.3
kPa/s curves and the 0.3 mm imperfect curves. After buckling, the sensitivity to imperfection or dynamic loading of
the panel disappeared, which was typical of post-buckling of plates, and the curves converged. The post-buckling
load capacity was about 39 KPa using imperfection perturbation and 41 KPa using dynamic loading rate and was
dominated by transverse compressive strength in web as shown in Fig. 5. The post buckling load was even lower
than the buckling load based on finite element analysis, indicating that there was no post-buckling capacity of the
sheet pile panel in the installed position.

4. DESIGN MODIFICATION FOR BETTER BUCKLING RESISTANCE

Modifications were proposed for sheet pile design based on two approaches: (1) with added compressive stiffeners
(diameter = 15 mm) at the compressive junctions to enhance the restraints on the compressive plates (Fig. 6a) and
(2) with the increased thickness (thickness = 4 mm) of two webs and two compressive flanges for better buckling
resistance (Fig. 6b). In either case, same amount of 10% additional material was added to the original profile. The
buckling load was increased to 93 KPa with added compressive stiffeners and to 66 KPa with increased thickness in
compressive members. In comparison with the capacity of 43 KPa in original profile design, the increase was
between 100% and 50%.

(a) With added compressive stiffeners (b) With increased thickness in compressive members

Fig. 6: Modification in sheet pile profile design

5. CONCLUSIONS
A finite element analysis was performed on a sheet pile wall subject to uniform pressure to assess its load capacity.
It was found that buckling initiation and ultimate failure happened almost simultaneously at buckling load,
suggesting there existed no post-buckling capacity for the sheet pile in the installed position. The pressure-deflection
curves were found to exhibit a softening behaviour due to the reduction of the section stiffness. Nonlinear finite
element analysis with perturbation function provides an efficient tool for determining the pre-buckling, bifurcation
and post-buckling loads of pultruded sheet pilings. The modified profile design with compressive stiffeners or
increased thickness in compressive members could double the load capacity.

6. REFERENCES
Bank, L. C. and Yin, J. (1999), “Failure of Web-Flange Junction in Postbuckled pultruded I-Beams”. Journal of
composites for construction, 3(4), pp 177-184.

Kollar, L. P. (2003), “Local Buckling of Fiber Reinforced Plastic Composite Structural Members with Open and
Closed Cross Sections”, ASCE Journal of Structural Engineering, 129(11), pp 1503-1513.

Palmer, D. W., Bank, L. C. and gentry, T. R. (1998), “Progressive Tearing failure of Pultruded Composite Box
Beams: Experiment and Simulation”, Composite Science and Technology, 58, pp 1353-1359.

Shao, Y. and Shanmugan, J. (2006), “Moment Capacities and Deflection Limits of Composite Sheet Piles”, Journal
of Composites for Construction, in press.

174
Third International Conference on FRP Composites in Civil Engineering (CICE 2006)
December 13-15 2006, Miami, Florida, USA

A case study on life-cycle assessment of environmental aspect

of FRP structures
Hirokazu Tanaka
(Research Engineer, SHIMIZU CORPORATION, Koto-ku, Tokyo,Japan)

Hitoshi Tazawa
(General Manager, ASAHI GLASS MATEX CO.,LTD., Sagamihara,Kanagawa, Japan)

Morio Kurita
(Group Leader, SHIMIZU CORPORATION, Koto-ku, Tokyo,Japan)

Takumi Shimomura
(Associate Professor, Nagaoka University of Technology, Nagaoka, Niigata,Japan)

ABSTRACT
In this paper, the authors evaluated the amount of carbon dioxide emissions from FRP footbridge and PC (Pre-
stressed concrete) footbridge with some reasonable scenarios based on the first FRP footbridge constructed in Japan
at 2000. As a result, the total amount of carbon dioxide emissions from FRP footbridge decreased by about 26
percent than that from PC footbridge. This is because the substructure of FRP footbridge was able to be cut down by
the superstructure of FRP footbridge which has much lighter weight than that of PC footbridge.

KEYWORDS
FRP structures, Environmental aspect, Amount of carbon dioxide emissions, Case study

1. INTRODUCTION

FRP has some excellent properties as a structural material and has a possibility to solve some problems that the
bridges made of conventional materials face today; corrosion in early stage, for example. For spreading use of FRP
bridge, it is very important to evaluate an environmental impact properly. There are some studies1),2),3) about the life-
cycle assessment of the structures such as buildings and bridges made from concrete and steel, but few studies of
FRP bridges. The unit amount of carbon dioxide emissions from FRP is much higher than that from other
conventional materials such as concrete and steel. However, FRP bridges have much lighter weight than other
conventional materials. This will reduce the total amount of carbon dioxide emissions. The authors tried to evaluate
the amount of carbon dioxide emissions from FRP footbridge. We selected the case of “Okinawa Road Park Bridge”
constructed at Okinawa in 2000 which is the first FRP footbridge in Japan. We compared this FRP footbridge case
with the PC footbridge case which was considered as an alternative plan. The amount of carbon dioxide emissions
from the materials and under construction was calculated with some reasonable scenarios. In this study, the authors
added some suppositions for unknown condition and made some simplifications for the evaluation, so the amount of
carbon dioxide emissions in this study don’t indicate the really amount of carbon dioxide emissions of Okinawa
Road Park Bridge itself.

2. TARGET STRUCTURE

The target structures to evaluate an environmental impact are FRP footbridge and PC footbridge. FRP footbridge is
about 38m in length and about 4.5m in width. The superstructure type of FRP footbridge is two span continuous
girder bridges. FRP is made of glass fiber and vinyl-ester resin. PC footbridge is about 36m in length and about

175
3.5m in width. The superstructure type of PC footbridge is one span post-tension hollow slab girder bridge. The
substructure type of FRP footbridge and PC footbridge is wall type pier. The foundation type is steel pipe pile.
Figure 1 shows the view of the FRP footbridge.

Figure 1: View of Okinawa Road Park Bridge (FRP footbridge)

3. EVALUATION RANGE

The evaluation range of an environmental impact of FRP footbridge and PC footbridge is shown in Figure 2. In this
paper, the materials stage and the construction stage were evaluated. Currently, the only options for disposal and
recycling are full recycling of the concrete vs. landfill or incineration of the FRP. But accurate numbers do not exist.
So the using stage, the dispose and recycle stage were not evaluated. As for FRP footbridge, the main beam made of
hand-lay up FRP and the bridge deck made of pultruded FRP were manufactured in separate factories and
transported to Tokyo by land. After the main beam and the bridge deck were assembled in the factory of Tokyo,
FRP footbridge was transported to Okinawa by sea.

Materials stage
Reinforcement Steel Cement, Aggregate
Main beam manufacture Bridge deck manufacture
Land transportation Land transportation
Land transportation 20km 20km
Land transportation
300km Assembling(Tokyo) 50km Concrete mixing
Land transportation
Construction stage 20km
Marine transportation Tokyo→Okinawa Reinforcement steel & form assebling
2,000km
Land toransportation Casting in place By concrete pump &
2km
vibrator
Installation By truck crane Curing General curing

Using stage Using Out of sbject Using Out of sbject

Dispose & Recycle Stage Dispose & Recycle Out of sbject Dispose & Recycle Out of sbject

FRP footbridge PC footbridge

Figure 2: Evaluation Range

4. UNIT OF CARBON DIOXIDE EMISSIONS

The unit amount of carbon dioxide emissions used in this paper is shown in Table 1. The unit amount of carbon
dioxide emissions of FRP was referred to the reference4). The unit amount of carbon dioxide emissions of concrete,
prestressing steel wire, steel pipe pile, and construction of concrete was referred to the committee report5). The unit
amount of carbon dioxide emissions of transportation was referred to the values6) showed on the website of
Ministry of Land Infrastructure and Transport in Japan.

176
 Table 1: Unit amount of carbon dioxide emissions

Heading Unit Unit amount of carbon dioxide emissions

FRP(Hand-lay up ） kg 4.97 kgCO2 /kg
FRP(Pultruded） kg 3.09 kgCO2 /kg
kg 0.0918 kgCO2 /kg
Concrete（ Fc27N/mm 2 ）
Materials m3 211.1 kgCO2 /m 3
Reinfoecement steel kg 0.755 kgCO2 /kg
Prestressing steel wire kg 1.31 kgCO2 /kg
Steel pipe pile kg 1.25 kgCO2 /kg
Marine transportation t*km 0.039 kgCO2 /t*km
T ransportation
Land transportation t*km 0.154 kgCO2 /t*km
Construction Concrete m3 39.0kgCO2 /m 3

5. EVALUATION RESULT

5.1 AMOUNT OF CARBON DIOXIDE EMISSIONS ON MATERIALS AND CONSTRUCTION STAGE

The amount of carbon dioxide emissions on the materials stage and on the construction stage is shown in Figure 4.
On the materials stage, for superstructure, the amount of carbon dioxide emissions from FRP footbridge increased
by about 8 percent than that from PC footbridge. This is because the unit amount of carbon dioxide emissions from
FRP is much higher than that from concrete, though the weight of superstructure of FRP footbridge is much lighter
than that of PC footbridge. For substructure, the amount of carbon dioxide emissions from FRP footbridge decreased
by about 50 percent than that from PC footbridge. This is because the substructure of FRP footbridge was able to be
cut down, especially the weight of steel pipe pile reduced. On the materials stage, the amount of carbon dioxide
emissions from FRP footbridge decreased by about 18 percent than that from PC footbridge.
On the construction stage, for superstructure, the amount of carbon dioxide emissions from FRP footbridge
decreased by about 80 percent than that from PC footbridge. This is because the work of FRP footbridge under
construction was almost the transportation by sea and the amount of carbon dioxide emissions on the transportation
by sea is very small. For substructure, the amount of carbon dioxide emissions from FRP footbridge decreased by
about 45 percent than that from PC footbridge. This is because the amount of concrete of the substructure of FRP
footbridge was much smaller than that of PC footbridge by the weight of superstructure of FRP footbridge being
much lighter than that of PC footbridge. On the construction stage, the amount of carbon dioxide emissions from
FRP footbridge decreased by about 70 percent than that from PC footbridge.

5 4
2 10
Amount of carbon dioxide emissions(kg)

2 10
Amount of carbon dioxide emissons(kg)

Superstructure Superstructure
5 4
Substructure
1.5 10 Substructure 1.5 10
(Construction stage)
(Materials stage)

5 4
1 10 1 10

4
5 10 5000

0 0
FPR footbridge PC footbridge FPR footbridge PC footbridge

Figure 4: Amount of carbon dioxide emissions on the materials stage and on the construction stage

177
5.2 TOTAL AMOUNT OF CARBON DIOXIDE EMISSIONS

The total amount of carbon dioxide emissions on the materials stage and the construction stage is shown in Figure 5.
The amount of carbon dioxide emissions from FRP footbridge decreased by about 26 percent than that from PC
footbridge. Though the unit amount of carbon dioxide emissions from FRP is much higher than that from concrete,
the total amount of carbon dioxide emissions on the materials stage and the construction stage of FRP footbridge
was smaller than that of PC bridge. This is because the substructure of FRP footbridge was able to be cut down by
the weight of superstructure of FRP footbridge which is much lighter than that of PC footbridge.

5
2 10
Amount of arbon dioxide emissions(kg)
(Materials & Construction stage) Materials stage
5 Construction stage
1.5 10

5
1 10

4
5 10

0
FRP footbridge PC footbridge

Figure 6: Carbon dioxide emissions on materials and construction stage

6. SUMMARY

1) On the materials stage, the amount of carbon dioxide emissions from FRP footbridge decreased by about 18
percent than that from PC footbridge.
2) On the construction stage, the amount of carbon dioxide emissions from FRP footbridge decreased by about 70
percent than that from PC footbridge.
3) The total amount of carbon dioxide emissions from FRP footbridge on the material and construction stage
decreased by about 26 percent than that from PC footbridge. This is because the substructure of FRP footbridge was
able to be cut down by the weight of superstructure of FRP footbridge which is much lighter than that of PC
footbridge.

7. REFERENCES

1) Koji Sakai (2005). “Environmental Design for Concrete Structures”. Journal of Advanced Concrete Technology,
Vol.3, No.1, pp.17-28.
2) Kenji Kawai, Takafumi Sugiyama, Koichi Kobayashi and Susumu Sano (2005). “A Proposal of Concrete
Structure Design Methods Considering Environmental Performance”. Journal of Advanced Concrete Technology,
Vol.3, No.1, pp.41-52.
3) Katz A. (2004). “The environmental impact of steel and FRP reinforced pavements”. ASCE Journal of
Composites for Construction, Vol.8, No.6, pp.481-488.
4) Japan Reinforcement Plastic Society (2003). “FRP Whoever Use -FRP handbook-” (In Japanese)
5) Japan Society of Civil Engineers (2004). “Environmental impact evaluation of concrete (Part2)” (In Japanese)
6) Ministry of Land Infrastructure and Transport (Japan). Data of the unit of carbon dioxide emissions of transporta
tion for 2003, http://www.mlit.go.jp/sogoseisaku/kankyou/ondanka1.htm, 03/21/06. (date accessed)

178
Part V. Confinement Issues
Third International Conference on FRP Composites in Civil Engineering (CICE 2006)
December 13-15 2006, Miami, Florida, USA

AXIAL BEHAVIOR OF CIRCULAR COLUMNS CONFINED WITH

FIBER-REINFORCED POLYMER JACKETS
Azadeh Parvin
Associate Professor of Civil Engineering, The University of Toledo, OH 43606, USA

Aditya S. Jamwal
Former Graduate Student, The University of Toledo, OH 43606, USA

ABSTRACT
This paper presents a finite-element analysis of the behavior of axially-loaded, small-scale fiber-reinforced polymer
(FRP) wrapped concrete columns. Parameters considered were the FRP wrap thickness, and the ply angle
orientation with respect to the lateral axis of the column. The finite element analysis results showed considerable
increase in the axial compressive strength and ductility of the FRP-confined concrete columns as compared to the
unconfined ones. The increase in the wrap thickness resulted in further axial strength and ductility enhancements of
the confined columns. Columns confined with the 0° angle ply wrap exhibited the highest axial stress capacity as
compared to the columns enhanced with wraps consisted of ±15° and 0°/±15°/0° ply angles.

KEYWORDS
Column, Confined concrete, Axial load, Fiber-reinforced polymer

1. INTRODUCTION

Fiber reinforced polymer (FRP) wrapping of concrete columns is an ideal technique to increase the strength and
ductility of these structural elements. Among the significant parameters that influence the confinement effectiveness
of FRP-wrapped concrete columns, wrap thickness or number of FRP layers, and ply angle orientation are
noteworthy (Rochette and Labossiere, 2000; Parvin and Jamwal, 2005, 2006). In the present study, effects of the
wrap thickness and the ply angle orientation on the performance of small-scale FRP-wrapped circular columns have
been investigated using a nonlinear finite element analysis software program (MSC.MARC™ 2001), to gain insight
into the behavior of such columns.

2. FINITE ELEMENT ANALYSIS: MODELS, AND SIMULATION RESULTS

Prior to parametric studies, the proposed finite element analysis (FEA) models were validated and calibrated through
an experimental study on FRP-wrapped circular columns reported in the literature (Rochette and Labossiere, 2000).
The finite element analysis and experimental results were in good agreement and are reported elsewhere (Parvin and
Jamwal, 2006).

Using the validated FEA models, the behavior of small-scale FRP-wrapped concrete columns, with various
wrap thicknesses, and ply angles under axial loading were investigated. The columns had the dimension of 152.4
mm × 304.8 mm (6 in × 12 in). The bottom of the concrete columns was fixed. The axial load was increased
incrementally and applied on the top cross-section of the column concrete core and not the FRP jacket.
Compression strength of 20.69 MPa (3 ksi) with Poisson’s ratio of 0.17 was assigned for the concrete, which was
modeled as an isotropic material. Along with the Mohr-Coulomb failure criterion, which has been used for the
concrete, isotropic work hardening rule was employed to define the concrete material’s plastic behavior. The failure

179
criterion for the unconfined concrete was the crushing strain of 0.003 and a small strain option unlike the confined
concrete, where a large strain option was used for analysis. For all cases, unidirectional E-glass FRP material was
used with a modulus of elasticity of 41,370 MPa (6×106 psi), a Poisson ratio of 0.24, and a rupture strain of 0.019
along the fiber direction. Since the FRP material was unidirectional, the stiffness in the other direction was
negligible. The rupture of the FRP material controlled the failure of the confined column. The unidirectional FRP
composite was modeled as an orthotropic material.

Circular column models utilized in the numerical study are presented in Table 1. The template used for labeling
the numerical column models leveraged the column shape, the wrap thickness, and the ply angle as follows:
a) The letter ‘C’ specifies that the column is circular.
b) The wrap thickness is categorized as 1, 2 and 3 for thicknesses of 1.3 mm (0.051 in.), 2.1 mm (0.083
in. and 3 mm (0.118 in.), respectively which is represented by the first subscript. The thicknesses of
1.3 mm (0.051 in.), 2.1 mm (0.083 in.) and 3 mm (0.118 in.) correspond to 6, 10 and 14 plies of FRP,
respectively.
c) The ply angles are categorized through the second subscript as 1, 2, and 3 for 0°, ±15°, and 0°/±15°/0°,
respectively.
As an example, the label or column identifier C11 denotes that the column is a circular with the wrap thickness of
1.3 mm (0.051 in.) and ply angle configuration of 0°.

Table 1: Parametric Case Studies for Circular Columns

Column ID Wrap Thickness Ply Angle Configuration

mm (in.)
C11, C12, C13 1.3 (0.051) 0°6 ±15°6 0°/±15°4/0°
C21, C22, C23 2.1(0.083) 0°10 ±15°10 0°/±15°8/0°
C31, C32, C33 3.0 (0.118) 0°14 ±15°14 0°/±15°12/0°

2.1 FEA Results

The strength and ductility enhancements of wrapped-column models C11 through C33 as compared to the control
model C00 are presented in Table 2. Furthermore, the axial stress versus axial and lateral strains are plotted in
Figures 1-3 to demonstrate the effects of various wrap thicknesses for each ply angle configuration. The stresses
and strains were measured at mid-height of the columns. The circular column with the ply angle of 0° provided the
highest axial strength. This increase in axial strength is more visible when the wrap thickness was increased from
1.3 mm (0.051 in.) to 3 mm (0.118 in.). However, the axial strain capacities for the columns with ±15° and
0°/±15°/0° ply angles are higher than that of columns with 0° ply angle for all the wrap thicknesses. For the
columns with 0°and 0°/±15°/0° ply angles, the increase in the axial stress was fairly consistent with the increase in
the wrap thickness. On the other hand, the same was not true for the column with the ply angle of ±15°. There was

Table 2: Maximum Axial Stress and Axial Strain for control and FRP wrapped Circular Columns

Axial Stress Axial Strain % Increase in % Increase in

Case ID
Mpa (ksi) Axial Stress Axial Strain
C00 20.85(3.024) 0.0033 -- --
C11 40.76 (5.905) 0.0221 95 569
C12 40.28 (5.840) 0.0257 93 679
C13 41.31 (5.990) 0.0246 98 645
C21 48.52 (7.035) 0.0220 133 567
C22 47.58 (6.904) 0.0263 128 697
C23 46.62 (6.765) 0.0272 124 723
C31 58.34 (8.462) 0.0219 180 564
C32 51.48 (7.464) 0.0250 147 658
C33 52.82 (7.659) 0.0246 153 645

180
 Stress vs. Strain Response for Ply Angle of 0°
9

6
(1 ksi = 6.895 MPa)
Axial Stress (ksi)

2 Ply Thickness = 0.051 in.

Ply Thickness = 0.083 in.
1
Ply Thickness = 0.118 in.
0
-0.02 -0.01 0 0.01 0.02 0.03
Transverse Strain Axial Strain

Figure 1: Stress vs. Strain for Ply Angle of 0°

Stress vs. Strain Response for Ply Angle of ±15°

8

6
(1 ksi = 6.895 MPa)
Axial Stress (ksi)

3
Ply Thickness = 0.118 in.
2
Ply Thickness = 0.083 in.
1
Ply Thickness = 0.051 in.

0
-0.03 -0.02 -0.01 0 0.01 0.02 0.03
Transverse Strain Axial Strain

Figure 2: Stress vs. Strain for Ply Angle of ±15°

181
 Stress vs. Strain Response for Ply Angle of 0°/±15°/0°

(1 ksi = 6.895 MPa)

6
Axial Stress (ksi)
5

3 Ply Thickness = 0.118 in.

2 Ply Thickness = 0.083 in.
1 Ply thickness = 0.051 in.
0
-0.03 -0.02 -0.01 0 0.01 0.02 0.03
Transverse Strain Axial Strain

Figure 3: Stress vs. Strain for Ply Angle of 0°/±15°/0°

an 18% increase in the axial strength when the wrap thickness was increased from 1.3 mm (0.051 in.) to 2.1 mm
(0.083 in.) (see axial stresses for model C12 and C22 in Table 2). On the other hand, only 8% gain materialized
when the wrap thickness was increased from 2.1 mm (0.083 in.) to 3 mm (0.118 in.) (see axial stresses for model
C22 and C32 in Table 2). Regardless of the ply angle configuration, the increase in the wrap thickness resulted in an
enhancement in the stiffness as apparent in the slope of second portion of the stress versus strain curves.

3. CONCLUSIONS

In the present study, the effects of various upgrade schemes on axially-loaded FRP-wrapped circular columns were
investigated. The following conclusions are drawn:
1. It was observed that the axial stress and strain capacities of the FRP-wrapped concrete column increased by
more than 2 to 3 times, and by 5 to 6 times, respectively as compared to the unconfined column.
2. The ply angle of 0° provided the highest axial strength for the FRP wrapped circular columns as compared
to ply angles of ±15° and 0°/±15°/0°. However, the ductility provided by the column with the ply angle of
0° was less than that of columns with the ply angles of ±15° and 0°/±15°/0°.
3. The increase in the axial stress carrying capacity is uniform with the increase in the wrap thickness for the
column with the ply angle of 0°. The FRP-wrapped circular columns with the ply angles of ±15° and
0°/±15°/0° provided a non-uniform increase in the axial stress carrying capacity as the wrap thickness was
increased. The columns with the ply angle of 0°/±15°/0° showed a softening response before failure for the
FRP-wrapped columns with higher wrap thicknesses of 2.1 mm (0.083 in.) and 3 mm (0.118 in.).
4. In general, for all the columns, the axial stress carrying capacity increased with the increase in the wrap
thickness.

4. REFERENCES

1. MSC.MARC™ 2001. MSC Software Corporation, Palo Alto, CA.

2. Parvin A, Jamwal A S. (2006). “Performance of externally FRP reinforced columns for changes in angle
and thickness of the wrap and concrete strength.” Composite Structures, 73(4), 451-457.
3. Parvin A, Jamwal A S. (2005). “Effects of wrap thickness and ply configuration on composite-confined
concrete cylinders.” Composite Structures, 67 (4), 437-442.
4. Rochette, P., and Labossiere, P. (2000). “Axial testing of rectangular column models confined with
composites.” Journal of Composites for Construction, 4(3), 129-136.

182
Third International Conference on FRP Composites in Civil Engineering (CICE 2006)
December 13-15 2006, Miami, Florida, USA

CFRP CONFINED ULTRA HIGH STRENGTH CONCRETE COLUMNS

Adnan R. Malik
(PhD Candidate, The University of New South Wales, Sydney, NSW, Australia)

Stephen J. Foster
(A/Prof., The University of New South Wales, Sydney, NSW, Australia)

ABSTRACT
An experimental investigation was conducted to examine the behaviour of RPC columns confined by CFRP and
subjected to concentric and eccentric loadings. All the 17 columns were cast with the concrete mix containing either
no fibres or 2% (by volume) of straight steel fibres with a concrete strength of 165 MPa. Experimental data for
strength, lateral and axial deformation and the failure mode were obtained for each test. The CFRP confined RPC
columns showed improvements in both ultimate compressive strength and ductility compared with unconfined RPC
columns. Because of the ultra high strength concrete used in this study, the final failure was sudden and explosive at
the point of rupture of the CFRP wrapping.

KEYWORDS
Caron fibre reinforced polymer, reactive powder concrete, columns, confinement.

1. INTRODUCTION
In recent years, the use of carbon fibre reinforced polymer (CFRP) composites for a variety of structural applications
(columns, bridge piers, piles etc.) has been rapidly increasing, mainly because of their superior strength to weight
ratio and durability in corrosive environments. The orthotropic behaviour of CFRP makes it a most suitable product
for encasing concrete columns and, in particular, in combination with an ultra high strength reactive powder
concrete (RPC). The objective of this research is to examine the behaviour of CFRP confined fibre and non-fibre
reinforced RPC circular columns without conventional reinforcement tested under concentric and eccentric loading.

2. EXPERIMENTAL PROGRAM
2.1 Test specimens

In this study, 17 RPC columns were tested consisting of 10 CFRP confined steel fibre reinforced RPC (FR-RPC)
columns, 6 CFRP confined RPC columns with no steel fibres and one FR-RPC column with no wrapping. The
columns were circular; 152 mm in diameter and 1060 mm high. No longitudinal reinforcement was used in any of
the columns. The columns tested are identified by load eccentricity, wrap type, existence of steel fibres in the
concrete and an identification number for the specimen if the test was repeated. For example, specimen FC35-1 was
cast with fibre reinforced RPC, was tested at an initial eccentricity of 35 mm and was wrapped with Type 1 layout
carbon fibre polymer sheet. Specimen PC0-2-2 was cast with RPC with no fibres, was tested under concentric
loading (0 mm eccentricity), was wrapped with Type 2 carbon fibre sheet and was a repeat test. Column FC0 was
control column with no wrap. Details for the columns tested are presented in Table 1.

2.2 RPC mix design and Material properties

The RPC was mixed using locally available materials: 920 kg/m3 of General Portland cement, 920 kg/m3 of Sydney
sand, 221 kg/m3 of undensified silica fume, 39 kg/m3 of superplaticizer and 157 kg/m3 of steel fibres for FR-RPC
mix. The steel fibres used were high strength straight steel fibres, 13 mm long, 0.2 mm in diameter and had an
ultimate tensile strength of 1800 MPa. The water-binder ratio was 0.17. The control specimen properties are

183
presented in Table 2, where ρf is the volumetric ratio of fibres, Eo is the modulus of elasticity, v is the Poison’s ratio,
fcm is the mean compressive cylinder strength, fdp is the double punch tensile strength, fsp is the split cylinder tensile
strength, fcf is the flexural tension strength and Gf is the fracture energy.

Table 1: Column Details

Load eccentricity Wrap Load eccentricity

Column Mix Wrap Type Column Mix
(mm) Type (mm)
FC60-1 1 1 60 FC10-2 4 2 10
FC35-1 2 1 35 PC60-1 8 1 60
FC20-1 1 1 20 PC35-1 8 1 35
FC10-1 2 1 10 PC20-1 7 1 20
FC0-1-1 3 1 0 PC10-1 7 1 10
FC0-1-2 3 1 0 PC0-1-1 9 1 0
FC60-2 5 2 60 PC0-1-2 9 1 0
FC35-2 5 2 35 FC0 6 - 0
FC20-2 6 2 20

2.3 FRP material properties and Wrap Details

The CFRP used in this project was MBrace® CF120 and CF530. The resin used for the bonding of CFRP was a
two-part epoxy adhesive (MBrace® Staturant). The CFRP was tested in accordance with the ASTM standard D3039
(2005). The mean ultimate tensile strength of three CF120 specimens was 3374 MPa with the strain corresponding
to failure load being 0.0145. The elastic modulus was 232 GPa. The columns were wrapped with 4 layers of CFRP.
The mechanical properties of the CFRP and epoxy as given by the manufacturer and the wrap details are presented
in Table 3.
Table 2: RPC material properties

ρf fcm Eo fdp fsp fcf Gf

Mix v
(%) MPa GPa MPa MPa MPa N/mm
1 2 163 0.13 41.7 8.2 - 26.2 20.7
2 2 165 0.1 42.0 7.7 - 35.4 32.7
3 2 168 0.1 42.6 7.9 - 17.7 14.1
4 2 172 0.13 44.2 7.6 - 26.2 18.4
5 2 168 0.15 42.6 8.7 - 24.7 21.1
6 2 165 0.15 44.2 7.7 - 33.3 28.8
7 0 139 0.13 40.5 3.1 8.0 6.1 0.03
8 0 143 - - 3.3 8.6 6.4 0.02
9 0 145 0.13 41.0 3.0 8.6 6.0 0.03

Table 3: Machanical properties of FRP and adhesive and details of CFRP wrapping

FRP and Tensile Elastic Laminate Wrap Type 1 Wrap Type 2

adhesive strength modulus Structure fibre Wrap fibre Wrap
(MPa) (MPa) Sheet direction Sheet direction
CF120 3800 240000 Layer1 &
CF120 Longitudinal CF530 Longitudinal
CF530 2650 640000 Layer 2
Layer3 &
Saturant >50 >3000 CF120 Circumferential CF120 Circumferential
Layer4

2.4 Test Setup and Instrumentation

The columns testing arrangements and gauging instrumentation are shown in figure 2. The loading was applied via
specially designed steel caps clamped to each end of the specimen. To apply the eccentric loading, knife edge end

184
assemblages were attached to the top and bottom platens of the testing machine. Testing was undertaken using a
5000 kN capacity closed loop servo controlled actuator with lateral displacement used to control the test for
eccentrically loaded columns and ram displacement control for the concentrically loaded specimens.

Ram
LVDT4

150 mm
152 mm
LVDT2

LVDT5

LVDT1 LVDT2
LVDT4
LVDT1 Base Plate Knife edge
150 mm

LVDT3 Zone A Zone B

Support
Load eccentricity
LVDT3 Steel Caps
LVDT6 (Lateral)
Column axis
(Axial)
60 deg End Plates
Circumferential SG

LVDT 1 LVDT5
(Axial)
B B Longitudinal SG (Lateral)
LVDT4
(Axial) LVDT2
LVDT locations for (Lateral) Optional Steel
Zone A & Zone B
Strain gauge locations concentric loading
plate to prevent
LVDT5 (Axial) rotation
LVDTs 1,2,3
(Lateral)

LVDT locations for

Section B-B eccentric loading LVDT4 (Axial)

(a) (b)
Figure 2: (a) Instrumentation details and (b) Test setup used for column test

3. TEST RESULTS AND OBSERVATIONS

The peak loads, moments at peak load, corresponding lateral displacements at mid height of the specimens, and
location of failure zone for the columns tested are given in Table 4.

Table 4: Peak Loads and Corresponding moments and lateral displacements

Column Failure Peak Moment ∆mid Column Failure Peak Moment ∆mid
location# Load Pu at Pu at Pu location# Load at Pu at Pu
(mm) (kN) (kNm) (mm) (mm) Pu (kN) (kNm) (mm)
FC60-1 +330 403 35.5 28.2 FC10-2 +230 1912 39.4 10.6
FC35-1 +50 714 38.9 19.4 PC60-1 +328 334 29.8 29.2
FC20-1 +200 1357 46.4 14.2 PC35-1 +238 773 43.4 21.2
FC10-1 +170 2221 42.0 8.9 PC20-1 * 1287 45.2 15.1
FC0-1-1 +150 2971 - - PC10-1 +20 1756 38.6 12.0
FC0-1-2 +50 2993 - - PC0-1-1 +150 2571 - -
FC60-2 +330 317 23.3 13.5 PC0-1-2 +160 2495 - -
FC35-2 -50 833 43.0 16.6 FC0 +200 2510 - -
FC20-2 -5 1367 47.0 14.4
Notes: # + is above and – is below column mid height, * test stopped before fibre rupture

For each CFRP confined column tested, snapping sounds were heard near the peak load as the fibres began to
rupture; however, there were no visible signs of impending failure on the surface of the wrapping. For columns
tested under load eccentricities of 10, 20 and 35 mm, failure occurred when the longitudinal FRP wrapping ruptured
in tension and the circumferential wrapping split vertically on the tensile side. Figure 3 shows that the compressive
concrete was under a considerable confining pressure towards the end of the test, particularly for the columns with
the smaller 10 and 20 mm initial loading eccentricities. In the eccentrically loaded specimens the final failure
occurred well beyond the peak loading, as shown in Figure 4. The peak axial loads and corresponding moments for
the CFRP confined RPC columns are plotted in Figure 5, together with the axial force bending moment interaction

185
diagram. The interaction diagram was obtained using an elastic-plastic stress-strain model with a yield stress of
0.85fcm, elastic modulus of 42.5 GPa and compressive failure strain of 0.005. The figure shows that the peak load is
greater than the model calculation. The concentrically loaded columns failed in a sudden and explosive manner and
the descending branch could not be captured using the ram displacement control. For CFRP confined specimens
tested the failure was characterized by CFRP failing in hoop tension followed immediately by explosive rupture of
the RPC. Table 4 shows that the confinement provided to specimen FC0-1-2 increased the failure load by 19% over
that of the unconfined specimen FC0.

FC10-1 2500
2500

Axial Comp. Strain (µε)

2000 PC60-1 PC35-1
2000
Axial Load (kN)

FC20-1
1500
1500 PC10-1

FC35-1 1000
1000

500 PC20-1
500
FC60-1 0
0
0 500 1000 1500 2000 2500 3000 3500
0 500 1000 1500
Circumferential Tens. Strain (µε) Circumferential Tens. Strain (µε)

(a) (b)
Figure 3. (a) Axial load versus circumferential tensile strain and (b) Axial compressive strain versus
circumferential tensile strain.

2400 FC10-1
3500 FCxx-1 series
2000 k3=0.85 FCxx-1 FCxx-2 PCxx-1
3000
Axial Load (kN)
Axial Load (kN

1600 FC20-1 2500

1200
2000
FC35-1 1500
800 PCxx-1 series .
1000
400
k3=0.85
500
FC60-1
0 0
0 10 20 30 40 50 0 15 30 45 60
Lateral displacement (mm)
Moment (kNm)
Figure 4. Axial load versus mid height lateral Figure 5. Peak axial loads and corresponding
displacement. moments for the columns tested.

4. CONCLUSIONS
In this study, 17 RPC columns were tested with 16 confined using CFRP. For the concentrically loaded specimens,
failure occurred at or close to the peak loading with little or no residual capacity. The CFRP confinement increased
the axial load capacity for concentrically loaded columns by 19% compared with unconfined column. For the
eccentrically loaded specimens, the CFRP was shown to be effective in controlling the failure of the specimens with
considerable straining occurring beyond the peak loading. There was no evidence, however, that the use of CFRP in
the hoop direction significantly increased the strength of the columns. As hoop strains increased, beyond the peak
loading, the stresses in the wrapping induced failure of the hoop CFRP and explosive collapse of specimen resulting
in a total loss of residual strength. Further investigations are needed to assess the effect of different types and
orientations of CFRP sheets to establish general deign guidelines for CFRP confined RPC columns.

5. REFERENCES
ASTM D3039, (2005). “Standard Test Method for Tensile Properties of Polymer Matrix Composite Materials”,
ASTM International.
Fam, A., Flisak, B., and Rizkalla, S. (2003). “Experimental and analytical modelling of concrete-filled fiber-
reinforced polymer tubes subjected to combined bending and axial loads”, ACI Structural Journal, Vol.100, No.4,
pp 499-509.
Richard, P. and Cheyrezy, M. (1995). “Composition of reactive powder concretes”, Cement and Concrete Research,
Vol. 25, No.7, pp 1501-1511.

186
Third International Conference on FRP Composites in Civil Engineering (CICE 2006)
December 13-15 2006, Miami, Florida, USA

STRENGTHENING OF SHORT CIRCULAR RC COLUMNS

WITH FRP JACKETS: A DESIGN PROPOSAL
T. Jiang
(PhD Student, The Hong Kong Polytechnic University ,Hong Kong, China)

J.G. Teng
(Chair Professor of Structural Engineering, The Hong Kong Polytechnic University ,Hong Kong, China)

ABSTRACT
This paper presents a design procedure for FRP jackets for the strengthening of short circular RC columns. The
proposed procedure is limited to FRP jackets which are continuous over a strengthened region of the column and
possess fibers oriented solely or predominantly in the hoop direction. The design procedure gives due attention to a
number of significant issues: (a) the ultimate hoop strain of FRP jacket; (b) the stress-strain model; (c) the
slenderness limit for short columns; and (d) the definition of an equivalent stress block for approximate section
analysis. The design proposal has been developed within the framework of the Chinese code for the design of
concrete structures (GB-50010 2002) for inclusion in the draft Chinese code for the application of FRP composites
in construction.

KEYWORDS
FRP, Jackets, Confinement, RC Columns, Design

1. INTRODUCTION
Strengthening of reinforced concrete (RC) columns using fiber-reinforced polymer (FRP) jackets or wraps is now a
widely accepted technique. This technique relies on the well-known fact that the compressive strength and ductility
of concrete can be substantially enhanced by the confinement from the FRP jacket. This paper presents a design
procedure for such FRP jackets which are continuous over a strengthened region of the column and possess fibers
oriented solely or predominantly in the hoop direction. The proposed procedure gives due attention to a number of
significant issues: (a) the ultimate hoop strain of FRP jacket; (b) the stress-strain model; (c) the slenderness limit for
short columns; and (d) the definition of an equivalent stress block for approximate section analysis. The design
proposal has been developed within the framework of the Chinese code for the design of concrete structures (GB-
50010 2002) for inclusion in the draft Chinese code for the application of FRP composites in construction.

2. ULTIMATE HOOP STRAIN OF FRP JACKET

It is well-known that the hoop rupture strain of FRP jackets confining concrete is significantly lower than that from
tensile tests of FRP coupons (Lam and Teng 2004). In the design of FRP jackets, the ultimate hoop strain should be
used in assessing the ultimate limit state of the jacket. This ultimate strain should be determined from one of two
approaches. For a given FRP system, particularly a new system, a sufficiently large number of FRP-confined
concrete cylinder tests should be conducted to find the characteristic value of the ultimate hoop strain. If such tests
are not conducted for whatever reason, the characteristic value of the ultimate hoop strain ε h, rup should be found from
ε h ,rup = ηε frp ,rup (1)
where ε frp,rup is the characteristic value of the material ultimate strain of FRP from flat coupon tests conducted
according to ASTM D3039/D3039M-00 (2000) or similar specifications, and η is a reduction factor. Based on

187
recent test results (Xiao and Wu 2000, 2003; Lam and Teng 2003, 2004; Jiang and Teng 2006a), it is suggested that
η =0.5 for CFRP and 0.7 for GFRP.

3. STRESS-STRAIN MODEL FOR FRP-CONFINED CONCRETE

In conventional section analysis of RC columns subjected to eccentric compression, the axial stress-strain curve of
concrete is assumed to be the same as that from axial compression tests. The same assumption is also commonly
used in the analysis of FRP-confined RC columns (e.g. Teng et al. 2002; Yuan and Mirmiran 2001). Although many
stress-strain models have been developed (Teng and Lam 2004), the design-oriented model of Lam and Teng (2003)
is believed to be particularly suitable for use in design due to its simplicity and accuracy. The model adopts a simple
form (a parabolic first portion which connects smoothly to a linear second portion) which automatically reduces to
that for unconfined concrete when no FRP is provided. It is proposed here that this model, incorporating the more
accurate expressions for the compressive strength and the ultimate axial strain recently proposed by Teng et al.
(2006) be adopted for design use. This refined version of Lam and Teng’s model is described by the following
expressions:
( Ec − E2 )
2

σ c = Ec ε c − ε c2 for 0 ≤ ε c < ε t (2a)

4 f co'
σ c = f co' + E2ε c for ε t ≤ ε c ≤ ε cu (2b)
where σ c and ε c are the axial stress and the axial strain respectively, Ec is the elastic modulus of unconfined
concrete, E2 is the slope of the linear second portion, f co' is the compressive strength of unconfined concrete. The
transition strain ε t between the parabolic portion and the linear portion and the slope of the linear second portion
E2 are respectively given by
2 f co' f cc' − f co'
εt = ; E2 = (3a); (3b)
( Ec − E2 ) ε cu
where f cc' and ε cu are respectively the compressive strength and the ultimate axial strain of confined concrete and
are given by
f cc' ⎪⎧1 if ρ K < 0.01
=⎨ (4)
f co ⎪⎩1 + 3.5 ( ρ K − 0.01) ρε
'
if ρ K ≥ 0.01
ε cu
= 1.65 + 6.5ρ K0.8 ρε1.45 (5)
ε co
where ρ K = E frp t /( Esec o R ) is the confinement stiffness ratio and ρε = ε h , rup / ε co is the strain ratio. It should be noted
that the confinement ratio f l / f co' can be interpreted as a product of ρ K and ρε ( f l / f co' = ρ K ρε ). E frp is the elastic
modulus of FRP in the hoop direction, t is the thickness of the FRP jacket, R is the radius of the confined concrete
core, Esec o and ε co are the secant modulus and the axial strain at the compressive strength of unconfined concrete,
with Esec o = f co' / ε co . When this model is used in a design specification, the model may need small adjustments so
that the curve reduces to that for unconfined concrete in a specific national code. In GB-50010 (2002), normal
strength concrete is assumed to have ε co = 0.002 and an ultimate strain of 0.0033. As a result, Ec = 1000 f co' and the
original value of 1.75 for the first item on the right hand side of Eq. (5) in the original model (Lam and Teng 2003;
Teng et al. 2006) is replaced by 1.65, so that the stress-strain model for FRP-confined concrete reduces to that for
unconfined normal strength concrete adopted by GB-50010 (2002) when no FRP is provided.

4. SLENDERNESS LIMIT FOR SHORT COLUMNS

A comprehensive numerical study (Jiang and Teng 2006b) has recently been conducted to develop a reliable
definition of the slenderness limit for short FRP-confined RC columns, in which a computer analysis of slender
FRP-confined RC columns was developed using the stress-strain model for FRP-confined concrete in compression
described in the preceding section. The analysis is for columns pinned at both ends, and is based on the following

188
assumptions: (a) the lateral deflection of the column is small in comparison with its length; (b) plane sections remain
plane; (c) the concrete does not resist any tension; (d) the steel reinforcement has an elastic-perfectly plastic stress-
strain curve; and (e) any confinement from transverse steel reinforcement is negligible. For relatively short columns
where material failure occurs before stability failure, the ultimate limit state of the column is reached when the strain
at the extreme concrete compression fiber of the critical section reaches the ultimate axial strain of FRP-confined
concrete, which is defined by Equation (5). By contrast, for longer columns where stability failure occurs before
material failure, the ultimate limit state of the column is characterized by the attainment of the maximum axial load.

The slenderness limit for short RC columns may be defined to ensure that the second order effect leads to only a
small amplification of the moment at the critical section or a small reduction (commonly 5% or 10%) of the axial
load capacity. In Jiang and Teng’s (2006b) study, the latter option was adopted. The slenderness ratio for RC
columns is commonly defined as λ = l / r , where l is the effective length of a column and r is the radius of gyration
and = R / 2 for circular columns. To develop a slenderness limit for FRP-confined RC circular columns, a large
parametric study was carried out. Based on the results from this parametric study, the slenderness limit for short
FRP-confined RC columns was proposed to be (Jiang and Teng 2006b):
e e
60 2 (1 − 1 ) + 20
D e2
λcrit = ' (6)
f cc ε h , rup
(1 + 0.06 )
f co' ε co
where D is the column diameter, and e1 and e2 (e2 ≥ 0; e2 ≥ e1 ) are the load eccentricities at the two column ends.

This expression has a clear physical meaning: the numerator defines the slenderness limit for short RC columns
without FRP confinement, while the denominator accounts for the effect of FRP confinement. When no FRP
confinement is provided and e2 D = 0.2 , Equation (6) reduces to λcrit = 32 − 12 e1 e2 , which is similar to but
slightly more conservative than the expression adopted by ACI-318 (2005). Besides, when e1 e2 = 1 , Equation (6)
reduces to λcrit = 20 , which is the slenderness limit for short RC columns without FRP confinement given in GB-
50010 (2002). Numerical results for the slenderness limit corresponding to a 5% axial load reduction obtained by
Jiang and Teng (2006b) are shown in Figure 1. These results are for columns with a longitudinal steel reinforcement
ratio ρ sc of 1% and with the circle connecting the centroids of longitudinal reinforcing bars having a radius Rs equal
to 70% of the column radius R. Columns experience a larger second order effect when the longitudinal steel
reinforcement ratio is smaller and concrete cover is thicker, although the effects of these two parameters are limited.
It can be seen from Figure 1 that Equation (6) is conservative for all cases except when the slenderness limit is very
small. However, if a 10% loss of axial load capacity is acceptable, numerical results not given here indicated that
Equation (6) provides a lower bound prediction for all cases. A 10% reduction in the axial load capacity has been
adopted as the criterion for permitted second effects in the existing literature [e.g. CEB-FIP (1993)].
300 1
Slenderness Limit, λ crit - Equation 6

250
Mean Stress Factor, α1

0.95

200
0.9

150

0.85
100 xn = 0.1R
xn = R
0.8
50
xn = 2R
Equation 7
0 0.75
0 50 100 150 200 250 300 1 1.2 1.4 1.6 1.8 2
Slenderness Limit, λ crit - analysis Strength Enhancement Ratio, f′cc / f′co
Figure 1: Slenderness Limit for Short Columns Figure 2: Mean Stress Factor α1

5. SIMPLIFIED DESIGN EQUATIONS

The strength of a short circular RC column can be found by section analysis. While such section analysis can be
conducted using the fiber element approach (e.g. Teng et al. 2002), a simplified approach using an equivalent stress

189
block is given in GB-50010 (2002). This section presents a similar simplified approach for FRP-confined circular
RC columns.

5.1 Stress Bock Factors for FRP-confined Concrete

GB-50010 (2002) defines two stress block factors for the compressive concrete in a circular RC section: α1 which is
the mean stress factor and β1 which is the block depth factor. It recommends that α1 = 1 and β1 = 0.8 . However, if
β1 = 0.8 is adopted as the block depth factor for FRP-confined RC columns, a unified and accurate expression for α1
for different neural axis depths xn cannot be obtained. From a trial-and-error process, it was found that
when β1 = 0.9 , the influence of xn on α1 can be ignored (Figure 2). The following simple expression can then be
proposed for α1
α1 = 1.17 − 0.2 ( f cc' f co' ) (7)

cu fcc'
fy
xn x
1 n
0 y
1
'2

y
2

R
Rs

fy

Figure 3: Stresses and Strains over a Circular Section

5.2 Simplified Section Analysis for Design Use

Using the stress block factors proposed above, a simplified section analysis is presented herein, which has been
modified from the procedure given in GB-50010 (2002) for the design of circular RC columns. This method is only
applicable to columns which have 6 or more evenly distributed longitudinal steel reinforcing bars. As shown in
Figure 3, the steel reinforcing bars are smeared into an equivalent steel cylinder of the same total cross sectional area
and with longitudinal strength only. 2πθ 0 , 2πθ , 2πθ1 and 2πθ 2' are respectively the central angles corresponding to
the depths of the neutral axis, the equivalent stress block, the yielded compressive steel reinforcement and the
yielded tensile steel reinforcement. θ 0 , θ , θ1 and θ 2' can be calculated from Equation (8) and it is obvious
that θ 2 = 1 − θ 2' .
arcos ⎡⎣1-ξ n (1 + Rs R ) ⎤⎦ arcos ⎡⎣1-β1ξ n (1 + Rs R ) ⎤⎦
θ0 = ; θ= (8a); (8b)
π π
arcos ⎡⎣ R Rs - (1 − β ) ξ n (1 + R Rs ) ⎤⎦ arcos ⎡⎣ R Rs - (1 + β ) ξ n (1 + R Rs ) ⎤⎦
θ1 = ; θ 2' = (8c); (8d)
π π
where ξ n = xn h0 is the ratio between the neutral axis depth and the effective height of the section h0
( h0 = R + Rs ). β is the ratio between the yield strain of the steel reinforcement to the strain of the ultimate
compressive strain of FRP-confined concrete and are given by
f
β= y (9)
Es ε cu
The axial load and bending moment carried by concrete can then be calculated from
⎛ sin 2πθ ⎞ 2 sin 3 πθ
N c = α1 f cc' θ A ⎜ 1 − ⎟ ; M c = α1 f cc' AR (10a); (10b)
⎝ 2πθ ⎠ 3 π
while those carried by the steel reinforcement can be calculated from

190
 N s = f y' As (θ1 + kc ) − f y As (θ 2 + kt ) (11a)
sin πθ1 + mc sin πθ 2 + mt
M s = f y' As Rs + f y As Rs (11b)
π π
where
⎡ξ n (1 + R Rs ) − R Rs ⎤⎦ π (θ 0 − θ1 ) + sin πθ 0 − sin πθ1
kc = ⎣ (12a)
πβξ n (1 + R Rs )

kt = −
(
⎡⎣ξ n (1 + R Rs ) − R Rs ⎤⎦ π θ 2' − θ 0 + sin πθ 2' − sin πθ 0 ) (12b)
πβξ n (1 + R Rs )
π (θ 0 − θ1 ) sin 2πθ 0 − sin 2πθ1
⎡⎣ξ n (1 + R Rs ) − R Rs ⎤⎦ ( sin πθ 0 − sin πθ1 ) + +
mc = 2 4 (12c)
πβξ n (1 + R Rs )
π (θ 2' − θ 0 ) sin 2πθ 2' − sin 2πθ 0
(
⎡⎣ξ n (1 + R Rs ) − R Rs ⎤⎦ sin πθ 2' − sin πθ 0 + ) 2
+
4
mt = (12d)
πβξ n (1 + R Rs )
Obviously, Equations (11) and (12) are too complex for design use. As a result, the following approximate
expressions are proposed:
0 ≤ θ c = θ1 + kc = 1.25θ − 0.125 ≤ 1 ; θt = θ 2 + kt = max (1.125 − 1.5θ , 0 ) (13a); (13b)
Using Equation (13), the expressions sin πθ1 + mc and sin πθ 2 + mt in Equation (11b) can be numerically approximated
by sin πθ c and sin πθ t respectively. Equation (13) was derived with the assumption of R Rs = 1.16 , as was used in
developing similar expressions in GB-50010 (2002). This assumption implies unrealistic concrete covers for very
large and very small columns, and may render the section analysis slightly un-conservative for small columns.
Nevertheless, Equation (13) is still sufficiently accurate for small columns as shown below.

Based on the above simplifications, the design equations can now be written as
⎛ sin 2πθ ⎞
N u = α1 f cc' Aθ ⎜1 − + (θ c − θ t ) f y As
2πθ ⎟⎠
(14a)
⎝
2 sin 3 πθ sin πθ c + sin πθt
M u = α1 f cc' AR + f y As R (14b)
3 π π
where N u and M u are the axial load capacity and the moment capacity, respectively.

1.2 1.2
Normalized Load Capacity, Nu/Nuo

Normalized Load Capacity, Nu/Nuo

Efrp=230GPa Efrp=80GPa
1 ε h,rup=0.0075 1 ε h,rup=0.015
ρsc=2% ρsc=2%
0.8 f′cc/f′co=1.5 0.8 f′cc/f′co=1.5

0.6 0.6

0.4 0.4

0.2 0.2
Accurate Analysis Accurate Analysis
Approximate Analysis Approximate Analysis
0 0
0 0.5 1 1.5 2 0 0.5 1 1.5 2
Normalized Moment Capacity, Mu/Muo Normalized Moment Capacity, Mu/Muo
(a) CFRP Jacket (b) GFRP Jacket
Figure 4: Comparisons between Accurate and Approximate Analyses

5.3 Accuracy of Approximate Design Equations

The predictions of Equation (14) are compared with the results of accurate section analysis in Figure 4 for a 300 mm
diameter column wrapped with either CFRP ( E frp = 230Pa, ε h , rup = 0.0075) or GFRP ( E frp = 80Pa, ε h , rup = 0.015).

191
The other material properties are: unconfined concrete compressive strength f co' = 20.1 MPa, steel yield strength
f y = 335 MPa, elastic modulus of steel Es = 200 GPa. The interaction curves are normalized by the axial load
capacity N uo (concentric compression) and moment capacity M uo (pure bending) of the column when no FRP
confinement is provided. The approximate design equations are seen to provide accurate predictions. Their accuracy
is even better for larger columns.

6. CONCLUSIONS
This paper has presented a design procedure for the strengthening of short circular RC columns with FRP jackets
which are continuous over a strengthened region of the column and possess fibers oriented solely or predominantly
in the hoop direction. The design procedure covers the following aspects: (a) the ultimate hoop strain of FRP jacket;
(b) the stress-strain model; (c) the slenderness limit for short columns; and (d) the definition of an equivalent stress
block for approximate section analysis. The design proposal was developed within the framework of the Chinese
code for the design of concrete structures (GB-50010 2002) for inclusion in the draft Chinese code for the
application of FRP composites in construction. The slenderness limit decreases with the confinement ratio, which
indicates that FRP-confined RC columns are more prone to the slenderness effect than conventional RC columns.
The slenderness limit equation presented in this paper was developed from columns with equal eccentricities at the
two ends and are thus unnecessarily restrictive for other cases. Work is currently being conducted by the authors to
develop a more accurate expression for such cases. The approximate design equations have been shown to provide
close predictions of results from accurate section analysis.

7. ACKNOWLEDGEMENTS
The authors are grateful for the financial support received from the Research Grants Council of the Hong Kong SAR
(Project No: PolyU 5059/02E) and The Hong Kong Polytechnic University (Project Codes: RG88 and BBZH).

8. REFERENCES
ACI-318 (2005). Building Code Requirements for Structural Concrete and Commentary, ACI Committee 318.
ASTM D3039/D3039M-00 (2000). Standard Test Method for Tensile Properties of Polymer Matrix Composite
Materials, American Society for Testing Materials.
CEB-FIP (1993). Model Code 1990, CEB-Bulletin No. 213/214, Comité Euro-International du Beton.
GB-50010 (2002). Code for Design of Concrete Structures, China Architecture and Building Press, China.
Jiang, T., and Teng, J.G. (2006a). “Analysis-oriented stress-strain models for FRP-confined concrete”, in
preparation.
Jiang, T., and Teng, J.G. (2006b). “Slenderness limits for short FRP-confined RC columns”, in preparation.
Lam, L., and Teng, J.G. (2003). “Design-oriented stress-strain model for FRP-confined concrete”, Construction and
Building Materials, Vol. 17, No. 6-7, pp. 471-489.
Lam, L., and Teng, J.G., (2004). “Ultimate condition of fiber reinforced polymer-confined concrete”, Journal of
Composites for Construction, ASCE, Vol. 8, No. 6, pp. 539-548.
Teng, J.G., Chen, J.F., Smith, S.T., and Lam, L. (2002). FRP-Strengthened RC Structures, John Wiley and Sons, Inc.
Teng, J.G., and Lam, L. (2004). “Behavior and modeling of fiber reinforced polymer-confined concrete”, Journal of
Structural Engineering, ASCE, Vol. 130, No. 11, pp. 1713-1723.
Teng, J.G., Jiang, T., Lam, L., and Luo, Y.Z. (2006). “Refinement of a design-oriented stress-strain model for FRP-
confined concrete”, in preparation.
Yuan, W., and Mirmiran, A. (2001). “Buckling analysis of concrete-filled FRP tubes”, International Journal of
Structural Stability and Dynamics, Vol. 1, No. 3, pp. 367-383.
Xiao, Y., and Wu, H. (2000). “Compressive behavior of concrete confined by carbon fiber composite jackets”,
Journal of Materials in Civil Engineering, ASCE, Vol. 12, No. 2, pp. 139-146.
Xiao, Y., and Wu, H. (2003). “Compressive behavior of concrete confined by various types of fiber composite
jackets”, Journal of Reinforced Plastics and Composites, Vol. 22, No. 13, pp. 1187-1201.

192
Third International Conference on FRP Composites in Civil Engineering (CICE 2006)
December 13-15 2006, Miami, Florida, USA

EXPERIMENTAL STUDY ON CONCRETE COLUMNS STRENGTHENED

WITH CFRP SHEETS CONFINEMENT UNDER ECCENTRIC LOADING
Cao Shuangyin
(Professor, Southeast University, Nanjing, Jiangsu, China)

Jing Denghu
(Asst. lecturer, Southeast University, Nanjing, Jiangsu, China)

Sun Ning
(MSC research student, Southeast University, Nanjing, Jiangsu, China)

ABSTRACT
An experimental study of five eccentrically loaded concrete columns confined by unidirectional CFRP sheets is
presented. The failure modes, strain distribution at sectional concrete, load capacities, lateral deflections, and strain
distribution of CFRP sheets along the circumference were investigated. The results indicated that, via the
reinforcement of CFRP sheets, ultimate load capacities of concrete columns were enhanced and ductility of columns
was improved. The reinforcing efficiency decreased with the eccentricity increasing. For the column with a
comparatively large eccentricity, though the reinforcing efficiency was limited, the failure mode of columns changed
from compression to tension. It is also shown that, the plane strain remains, that is, the strain in sectional concrete is
linearly proportional to the distance from the natural axis. The strain gradient in CFRP sheet, which is attributes to
the non-uniform expansion of concrete, was also observed.

KEYWORDS
Carbon fiber reinforced plastic (CFRP), concrete columns, eccentric loading, strain gradient.

1. INTRODUCTION
The wrap with carbon fiber reinforced plastic (CFRP) sheets can improve compressive performances of concrete
columns as a result of concrete core confinement. During the past decades efforts have been increasingly
concentrated on studying concrete columns, especially concrete cylinders, under concentric loading (Teng and Chen
2002; Lorenzis and Tepfers 2003; Berthet et al. 2005). However, in field applications, there is no column that is
under perfect concentric loading. Several studies on wrapping concrete columns under eccentric loading were
conducted with bidirectional CFRP (Chaallal and Shahawy, 2000) or unidirectional CFRP (Parvin and Wang 2001;
Li and Hadi 2003). Since the confinement of CFRP sheets to core concrete is passive, nonuniform-confining stress
due to uneven dilation of compressive concrete must be considered. Parvin and Wang (2001) have examined the
effect of strain gradient on FRP-jacketed concrete prisms, and concluded that the prediction of the ultimate strength
depended on the ability of the model to contain the strain value of the FRP jacket. The purpose of this paper is to
investigate the performances of rectangular reinforced concrete columns that are subjected to eccentric loading,
wrapped with unidirectional CFRP sheets.

2. TEST PROGRAMME

Five test specimens (Fig.1) designed for the study were rectangular cross section of 250mm×250mm and had an
overall length of 1350mm. The length between the corbels was 750mm. Steel reinforcement details of the specimens
were presented in Fig.1. The actual 28-day concrete cube strength was 24.4 MPa. The columns were wrapped with
one layer of unidirectional CFRP sheet with nominal thickness of 0.167mm; the ultimate tensile strength and

193
Young’s modulus of CFRP were 3399 MPa and 235 GPa respectively. The columns were treated with 20mm radius
corners. The columns were tested under monotonically increasing axial loads (Fig.2). The load was increased until
the significant strength decay was recorded. This decay indicated failure of columns. The specimens of Z1, Z2, Z3,
Z4, and Z5 had initial eccentricity of 0, 35, 55, 75, and 115mm respectively. The theoretical balanced condition
requires that the initial eccentricity of unconfined column is 130mm according to actual strength of materials.
During the test, strains distribution at sectional concrete, lateral deflections, circumferential strains in CFRP sheets
and the load were monitored.

250 150 Steel plate with 250 2 16

thickness of 10mm
6@250

150 150

250
2 16
3 3

1-1 2-2
1 1
1350

750 2 2
2 16 4 16

2 16
150 150

6@75
3 3
400 Steel plate with 3-3
thickness of 10mm

Figure 1: Details of Specimens Figure 2: The Test Setup

3. EXPERIMENTAL RESULTS AND DISCUSSIONS

For all specimens, failures began with the bulge of CFRP sheets due to the dilation of core concrete (Fig.3a). The
fiber broke off in certain corner first, and then the specimen failed in the sides with larger compression by crushing
of the concrete near the middle section (Fig.3b, 3c, 3e). At the opposite side, for specimen Z2 with a small initial
eccentricity, there were no cracks (Fig.3d), but for specimen Z3, Z4 and Z5, distinct horizontal cracks existed (fig.
3f), the crack width became wider with the increase of initial eccentricity.

(a) Bulge (b) Specimen Z1 (c) Failure face of Z2

(d) Opposite face of Z2 (e) Failure side of Z3 (f) Opposite face of Z3

Figure 3: Typical Specimens after Failure

Testing results of five specimens were presented in Table 1, where the capacity of unconfined column was calculated
in accordance with concrete theory. The test results indicated that, when the initial eccentricity was smaller than the
balance condition, the CFRP confinement was also effective in strengthening concrete columns under eccentric
loading. But when the initial eccentricity increased near to the balanced condition (Z5), the enhancement descended
drastically. The maximum compressive strain of concrete was up to about 0.005, which was higher than that of
unconfined concrete and had no large change as a result of different initial eccentricity (Except column Z2 due to its
deviant failure). The maximum tensile strain was up to 35% of ultimate tensile strain of CFRP, which was lower than
that of CFRP confining cylindrical concrete due to corner effect and cross section. The strain distributions across the

194
depth at middle section were displayed in Fig.4. The results showed the strain in the section was linearly
proportional to the distance from the natural axis, and this was the same as that in unconfined concrete.

Table 1: Summary of Testing Results

Compressive Tensile Failure Enhancement Failure

Number Initial
Deflection Capacity strain strain position compared to position
of eccentricity
（mm） （kN） （ με ） （ με ） of unconfined of
columns （mm）
CFRP column columns
Z1 0 — 1975 — 5735 Corner 33.1% Middle
Z2 35 2.26 1400 -2232 4063 Corner 33.8% Upside
Z3 55 4.03 1175 -4891 5984 Corner 35.4% Middle
Z4 75 5.58 1000 -4495 5240 Corner 36.2% Upside
Z5 115 9.43 650 -5011 5394 Corner 19.2% Middle
250 250
0 kN
0 kN
Depth from tension face(mm)

150 kN
300 kN

Depth from tension face(mm)

200 200 300 kN
600 kN
900 kN 450 kN
1200 kN 600 kN
150
1400 kN 150 650 kN

100 100

50 50

eo=35mm eo=115mm
0 0
-6000 -4000 -2000 0 2000 4000 6000 -6000 -4000 -2000 0 2000 4000 6000
-6 -6
Strain (10 ) Strain( 10 )
(a) Specimen Z2 (b) Specimen Z5

Figure 4: Strain Distribution across the Middle Section

0 950 750

In order to measure the strain distribution of externally confining fiber

along the circumference, and so to understand the effectiveness of CFRP
confinement in strengthening concrete columns under eccentric loads,
strain gauges were bonded in CFRP out layer showed in Fig.5. From the
test results showed in Fig.6, it could be seen that the higher the Loading area
compression in the inner concrete, the greater the strain of external CFRP, 250 500
and when the inner concrete were in tensile, the strains of CFRP were near
to zero, which was similar to the results by Chaallal and Shahawy (2000). Figure 5: Points of Strain Gauges

The tested load–middle deflection curves were displayed in Fig.7, which indicated that the response consisted of
three distinct regions. In the first region, the relation was linear. With the increase of load, a transition zone was
entered. Finally, the slope of curve stabilized. Fig.8 showed the comparison of the theoretical M − N curve of the
specimens without CFRP confinement and that of the specimens with CFRP confinement. Here second order
moments were considered. It could be seen that, for the specimen with an initial eccentricity near to balanced
7000 8000
0 kN 1500 kN 0 kN 800 kN
400 kN 1800 kN 7000 200 kN 1000 kN
6000 800 kN 1975 kN 400 kN 1175 kN
1200 kN 600 kN
6000
Strain( 10 )
Strain( 10 )

5000
-6
-6

5000

4000 4000

3000 3000

2000
2000
1000
1000
0

0 -1000
0 100 200 300 400 500 600 700 800 900 0 100 200 300 400 500 600 700 800 900
Perimeter( mm ) Perimeter ( mm )
(a) Specimen Z1 (b) Specimen Z3

195
 8000 8000
0 kN 600 kN 0 kN 450 kN
7000 200 kN 800 kN 7000 150 kN 600 kN
400 kN 1000 kN 300 kN 650 kN

6000 6000

Strain ( 10 )
Strain ( 10 )

-6
-6
5000 5000

4000 4000

3000 3000

2000 2000

1000 1000

0 0

-1000 -1000
0 100 200 300 400 500 600 700 800 900 0 100 200 300 400 500 600 700 800 900
Perimeter ( mm ) Perimeter ( mm )
(c) Specimen Z4 (d) Specimen Z5

Figure 6: Outspread Sketch of Strain of Externally Confining CFRP

1600 2200
Z2 eo=35 mm Z1 Experimental result
2000
1400 Z3 eo=55 mm Theoretical calculation
Z4 eo=75 mm 1800
1200 Z5 eo=115mm 1600 Compression failure
Z1
Vertical load(kN)

Z2
1000 1400
Z3

N (kN)
1200 Z2
800 Z4
1000 Z3
600 800 Z4
600 Z5 Z5
400
400 Tensile failure
200
200
Balanced condition
0 0
0 2 4 6 8 10 0 20 40 60 80 100
Lateral deflection(mm) M (kN⋅ m)

Figure 7: Curve of Load-deflection Figure 8: Curve of M − N

condition (specimen Z5), though the enhancement was not significant compared to specimens with small initial
eccentricity (Z1 to Z4), the failure mode changed from compression to tension after wrapped by CFRP.

4. CONCLUSIONS
The experiment study of five concrete columns wrapped by one layer of unidirectional CFRP sheets under eccentric
loading showed that, the CFRP confinement was also effective in strengthening eccentrically loaded concrete
columns when the initial eccentricity was comparatively small. With the increase of initial eccentricity near to
balanced condition, the reinforcement was not in much effect, but the failure mode of columns changed from
brittleness to ductility. Strain gradient in CFRP sheets existed due to the uneven dilation of core concrete, and the
assumption that the strain in sectional concrete is proportional to the distance from the natural plan remains.

REFERRENCES
1. Berthet J. F., Ferrier E., Hamelin P. (2005) “Compressive behavior of concrete externally confined by composite
jackets, Part A: experimental study”, Construction and building Materials, Vol.19, pp 223-232
2. Chaallal O., Shahawy M. (2000). “Performance of Fiber-reinforced Polymer-wrapped Reinforced Concrete
Column under Combined Axial-flexural Loading”, ACI Structural Journal，Vol. 97, No. 4, pp 659-668.
3. Li J., Hadi M. N. S. (2003) “Behavior of Externally Confined High-strength Concrete Columns under Eccentric
Loading”, Composites Structures，Vol. 62, pp 145-153
4. Lorenzis L. D., Tepfers R. (2003) “Comparative study of models on confinement of concrete cylinders with
fiber-reinforced polymer composites”, Journal of Composites for Construction, Vol.7, No.3, pp 219-237
5. Parvin A., Wang W. (2001). “Behavior of FRP Jacketed Concrete Columns under Eccentric Loading”, Journal of
Composites for Construction，Vol. 5, No. 3, pp 146-152.
6. Teng J. G., Chen J. F., and Smith S. T. (2002). “FRP Strengthened RC Structures”, England: John Willey & Sons,
LTD

196
Third International Conference on FRP Composites in Civil Engineering (CICE 2006)
December 13-15 2006, Miami, Florida, USA

INFLUENCE OF RADIUS OF CORNERS IN CONCRETE COLUMNS

CONFINED WITH FRP SHEETS
Giuseppe Campione
(Associate Professor, University of Palermo, Palermo, Italy)

Nunzio Miraglia
(Assistant Professor, University of Palermo, Palermo, Italy)

Maurizio Papia
(Full Professor, University of Palermo, Palermo, Italy)

ABSTRACT
In the present paper the confinement effects induced in short compressed concrete members externally wrapped with
fiber reinforced plastic (FRP) is investigated. Effect of radius of corners and local reinforcements constitute by
single strips applied at corners before the continuous wrapping in square cross-sections are considered. The model
allows one to evaluate the equivalent uniform confining pressure at ultimate conditions referred to the effective
stresses in FRP along the sides of section; consequently, it makes it possible to determine the equivalent
confinement pressures exercised by the FRP wraps.

KEYWORDS
FRP; confinement; square cross-section; stress-concentration; sharp corners, smoothed corners; local reinforcement.

1. PROPOSED MODEL
In order to reduce the stress-concentration at the corners of a concrete cross-section wrapped with FRP sheets, two
different techniques are available: - the first one is the well know technique (Mirmiran et al. 1998) consisting in
smoothing sharp corners with adequate radius of concrete cross-section (see Fig. 1a); - the second one is the
reinforcement technique proposed by Campione et al. (2004), consisting in local strengthening at the corner
constituted by single strips applied before the continuous wrapping (see Fig. 1b). It has to be observed that the first
technique reduces the risk of local failure of fiber at the corners and also increases the equivalent confinement
pressures, while the second technique lightly increases the confinement pressures and produces over strength of
fibers at the corners, reducing the risk of local failure.
L
L
L2
FRP
r L2

L t1 L

t t2
H H
L2
F RP
L2 r

t1 B B
t
t2
S ING LE STRIP S
FRP

FRP Single strips

L1
a) b)
Figure 1: Reinforcing technique at the corners

197
To explain in a simplified way the confinement effects in the concrete core due to the presence by FRP wraps, which
can be pointed out numerically by a finite element approach, a recent model proposed by the authors (Campione et
al. 2004) is briefly summarized. It refers to the geometrical model shown in Fig. 2, representing a three dimensional
prismatic concrete member having square cross-section of side L with smoothed corners with radius r and eternally
wrapped with FRP wraps. The continuous FRP wrap is assumed to be externally applied to the concrete surface with
an adequate overlap length and it is not directly axially loaded. If the concrete member is loaded axially and
maintains its prismatic shape, it tends to be subjected to axial and lateral strains partially reduced by the presence of
the FRP sheet. Therefore, the increase in the bearing capacity of the member is due to the share of normal stress
depending on the confinement lateral stresses and a three dimensional model is necessary to consider these effects. It
is possible to further simplify the three dimensional model by assuming a plane model (transverse cross-section) and
considering, due to the symmetry of the system, only one quarter of the transverse cross-section, as shown in Fig. 2.

σ Three dimensional model Plane model Uniaxial model

εv
δ L/2
ε εe w A y F F
FRP
νcεv A H CA r
kc
x H HH kr
A δ L L/2
A
H
H
A A
F
L F
Deformed shape

Figure 2: Geometrical model for FRP wrapped prismatic specimens

When lateral expansion occurs in the concrete member, the displacement along the diagonal direction
corresponds to the lateral elongations along the two sides of the cross-section. These elongations can be assumed to
be produced by two elastic beams (FRP sheets) on uniaxial springs (concrete) and connected at the corners of the
cross-section. The elastic beams on elastic springs are stiffened in flexure by including also the effect of axial forces
in FRP; nevertheless, because in wrapped cross-section the stresses in FRP are very low with respect to tensile
strength of material and very low thickness of reinforcing layers are utilized, the afore mentioned stiffness effect is
negligible with respect to the confinement pressures distribution. The membrane effect can be taken into account by
imposing that the displacement at the end of the beam (corner of cross-section) is equal to the elongation of the FRP
package in the perpendicular direction. To determine the confinement pressures q (interaction between concrete core
and FRP wrap acting perpendicular to the lateral concrete surface) it is assumed that the beam behaves as semi-
infinite elastic member on elastic springs, subjected to the tensile force F shown in Fig.2. More in details, the two
elastic beams having length L/2 - r simulating the FRP package have flexural stiffness proportional to the quantity Ef
t3/12, Ef being the modulus of elasticity and t the thickness of the FRP sheet. The elastic springs are considered
acting each separately from another and their stiffness takes the presence of the concrete core shell into account.
This stiffness can be assumed to be proportional to the stiffness of the confined concrete core. Its value for unit
Ec
length is k v = , E being the modulus of elasticity of concrete, L the side of the cross-section
(L 2 − r ) ⋅ (1 − 2 ⋅ ν c ) c
considered and r the radius of corner of fillet.
The equilibrium equation of the elastic beam of inertia If , on elastic springs (see Fig. 2), in term of lateral
displacements w, is governed by the following differential equation:

d 4w
+
kv
(δ − w) = 0 (1)
dx 4
Ef ⋅ I f

Where δ is the lateral displacement of the unconfined concrete specimen and w the actual displacement. If we
consider that the elastic beam of length B/2 - r is connected to the perpendicular legs by means of a quarter of

198
 Ec
cylinder having axial stiffness kr = by imposing the boundary conditions it is possible to obtain the
r ⋅ (1 − 2 ⋅ ν c )
w(x) referred to the beam of length L/2 – r.
The boundary conditions are: i) the displacement at the tip of the beam is equal to the elongation of FRP along the
round fillets and the flat portion plus the radial shortening of concrete cylinder having radius r; ii) in the same
section (x = 0 tip of the beam) the rotation is zero. Therefore the following relationships are obtained:

⎡ 1 ⎛L π ⋅ r ⎞ (1 − 2 ⋅ ν c ) ⎤
w ( x = 0) = F ⋅ ⎢ ⋅⎜ − r + ⎟+ ⎥ (2)
⎣ Ef ⋅ t ⎝ 2 2 ⎠ Ec ⎦

dw
( x = 0) = 0
dx

3kv
By setting β = 4 and by integrating the equilibrium equation, Eq. (1), Eq. (2) into account gives the
t ⋅ Ef
3

following expressions of the displacement w and of the confinement pressure q:

⎧⎪ ⎡ 1 ⎛L π ⋅ r ⎞ ⎛ 1 − 2 ⋅ νc ⎞⎤ ⎫⎪ − βx
w ( x ) = ⎨F ⋅ ⎢ ⋅⎜ − r + ⎟+⎜ ⎟⎥ − δ⎬ ⋅ e
⎟ ⋅ cos(βx ) + δ (3)
⎪⎩ ⎣⎢ E f ⋅ t ⎝ 2 2 ⎠ ⎜⎝ E c ⎠⎦⎥ ⎪⎭

q ( x ) = k v ⋅ (δ − w ) (4)

Consequently the resultant of confining pressures along the flat portion proves to be:

L / 2− r ⎧⎪ ⎡ 1 ⎛L π ⋅ r ⎞ 1 − 2 ⋅ ν c ⎤ ⎫⎪ 1
R= ∫ q ( x ) ⋅ dx = k v ⋅ ⎨δ − F ⋅ ⎢ ⋅⎜ − r − ⎟+ ⎥⎬ ⋅ ⋅
0 ⎪⎩ ⎣ Ef ⋅ t ⎝ 2 2 ⎠ E c ⎦ ⎪⎭ 2 ⋅ β
⎡ ⎛L ⎞ ⎤ (5)
⎢ − β⎜ − r ⎟
⎝ 2 ⎠ ⎛ ⎛L ⎞ ⎛L ⎞ ⎞⎥
⋅ ⎢1 − e ⋅ ⎜⎜ cos β ⋅ ⎜ − r ⎟ − sin β ⋅ ⎜ − r ⎟ ⎟⎟⎥
⎢ ⎝ ⎝2 ⎠ ⎝2 ⎠ ⎠⎥
⎢⎣ ⎥⎦

To determine the axial forces in FRP, the equilibrium between the resultant of confinement pressures (in the
beam of length B/2 - r plus the contribution due to the circular cylinder) and the force F was imposed, resulting:

⎧⎡ ⎛L ⎞ ⎤⎫
−β⋅⎜ − r ⎟
k v ⎪⎪⎢ ⎛ ⎛ L ⎞ ⎛ L ⎞ ⎞ ⎥ ⎪⎪
⋅ ⎨⎢1 − e ⎝ 2 ⎠ ⋅ ⎜⎜ cos β ⋅ ⎜ − r ⎟ − sin β ⋅ ⎜ − r ⎟ ⎟⎟⎥ ⎬ ⋅ δ
2 ⋅ β ⎪⎢ ⎝ ⎝2 ⎠ ⎝2 ⎠ ⎠⎥ ⎪
⎪⎢⎣ ⎥⎦ ⎭⎪ (6)
F= ⎩
⎧ ⎡ ⎛L ⎞ ⎤⎫
−β⋅⎜ −r ⎟
⎡ 1 ⎛L π ⋅ r ⎞ 1 − 2 ⋅ νc ⎤ ⎪⎪ k v ⎢ ⎛ ⎛ ⎞ ⎛ ⎞ ⎞ ⎥⎪
⎠ ⋅ ⎜ cos β ⋅ ⎜ − r ⎟ − sin β ⋅ ⎜ − r ⎟ ⎟⎥ ⎪⎬
L L
1+ ⎢ ⋅⎜ − r + ⎟+ ⎥⋅⎨ ⋅ ⎢1 − e ⎝ 2
⎜ ⎟
⎣ Ef ⋅ t ⎝ 2 2 ⎠ Ec ⎦ ⎪ 2 ⋅ β ⎢ ⎝ ⎝2 ⎠ ⎝2 ⎠ ⎠⎥ ⎪
⎪⎩ ⎢⎣ ⎥⎦ ⎪⎭

2⋅F
Therefore the equivalent confinement pressure is f le = . In the case of square section (r =0) and FRP layers
L
having t2 thickness and L2 length at the corners and t1 thickness in the other part with length L1 = L - 2 L2, (see Fig
1b), the expression of F force given in Campione et al. (2004).

199
2. EXPERIMENTAL AND ANALYTICAL VALIDATION
To validate the proposed model two different recent investigations are considered. The first one (Yang et al.
2001) is of experimental nature and it analyses the effect of corner radius on the confinement properties of FRP
wraps. In this case unique re-usable aluminum devices were utilised for support of plies of FRP externally bonded.
Aluminum device had side of length L and interchangeable corners insert allowing corners of different radius r.
Direct tensile test on the devices allows one to force FRP wraps in tension, transmitting stresses around the
perimeter of the aluminum devices. Strain measurement along the corners and along the flat portion of the FRP was
made allowing one to measure the FRP stress with the variation in the radius of corners. The case of one ply of high
strength and low modulus of carbon fiber (ultimate stress 4275 MPa and tensile modulus of 228 GPa) externally
bonded to aluminum devices was considered. The variation of maximum stress reached at rupture of FRP with
variation in the radius of corners, for L = 150 mm was measured. Fig. 3 shows the experimental values (maximum
stress is made dimensionless with respect to the ultimate stress measured in circular cross-section, and radius of
corner is made dimensionless with respect to the side B). Analytical prediction according to the proposed model (Eq.
6) is also given in the same figure. The analytical results are generated considering for the aluminum device the
modulus Ec = 70 GPa and the Poisson ratio v = 0.35. The comparison shows acceptable agreement between
analytical and experimental predictions for low values of radius of corners, while for the case of circular cross-
section (r/b = 0.5) the analytical model provides higher values than the experimental value, which is almost equal to
0.66.
The second case examined refers to the analytical investigation of Karam and Tabbara (2005), in which a refined
element model by using the ADINA program was developed, using a nonlinear concrete constitutive law in order to
analyze stresses in columns passively confined with FRP wraps. Rectangular and square cross-sections of variable
corner radii were investigated. Case here considered is that of concrete member with square cross-section of side L =
150 mm externally wrapped with FRP having 1 mm thickness per ply, ultimate strain of 1.5% and modulus of
elasticity 105 GPa. Concrete had initial modulus of elasticity of 26 GPa, Poisson ratio 0.18 and strength of
unconfined concrete of 30 MPa.

Exp. Data of Yang et al. (2001) L =150 mm

Ec = 26 GPa; Ef = 105 GPa 1.0
εu =1.5 %

Mander et al. (1988)

75

Eq. (6) Current model

Karam and Gabbana (2005)

N.L.F.E. (Adina)

r/L
Figure 3: (a, b) Experimental and analytical validation

Results obtained by Karam and Tabbara (2005) by using finite element analysis showed that it is possible to
simplify the problem by assuming no friction relations between confining stress and stress in FRP and by assuming
FRP as a cable around the corner similar to a pulley. The variation of the stress in FRP with variation in the radius r
is almost linear. Fig. 3b shows the variation of ultimate stress reached in FRP at concrete rupture, normalized with
respect to the ultimate stress allowable in FRP, with the increase in the radius of corner. In the same graph
prediction according to the proposed model, and to the Mander et al. (1988) model and to nonlinear finite element
analysis carried out by the Karam and Tabbara (2005) are given. In the Mander et al. (1988) model the actual stress
in FRP was taken into account (see Karam and Tabbara 2005) by means of a confinement effectiveness coefficient.
The proposed model is utilised by assuming in Eq. (6) for Ec a value equal to 2/3 of the initial value, and the
Poisson ratio equal to 0.49. It is interesting to observe that the model of Mander et al (1988) overestimates the
maximum allowable stress compared with the two other models. The proposed model is the most conservative for
high value of radius of corners; moreover, it gives a value of stress in members with sharp corner which is not zero,
but essentially depending on the confinement stiffness factor, as observed by Karam and Tabbara (2005).

200
3. CONCLUSIONS
In the present paper the confinement effects induced in short compressed concrete members by external fiber
reinforced plastic (FRP) wrap is investigated. Square cross-sections with round corners are considered.The effect of
the increases in the radius of corner in the increase in the allowable stress in FRP at concrete failure was analysed.
According with the proposed model it was demonstrated the relation between allowable force in FRP and the radius
of corner is not linear and it depends on several factors such as: - fiber type and its thickness and modulus of
elasticity; - concrete characteristic (though the modulus of elasticity, and the Poisson ratio of concrete); - dimension
of cross-section. Moreover, some other parameters such as the length of the specimens (Campione, 2006) and the
premature failure of FRP at the corners due to cutting effects in FRP can reduce drastically the effectiveness of
devices and furthers experimental and analytical work is necessary to give more general conclusions.

4. REFERENCES
Mirmiran, A., Shahawy, M., Samaan, M., EL Echary, H., Mastrapa, J.C. and Pico, O. (1998), “Effect of column
parameters on FRP-confined concrete”. J. of Comp. for Constr., ASCE, Vol. 2, n.4, pp.175-185.
Mander, J.B., Priestley, M.J.N., and Park, R. (1988), “Theoretical stress-strain model for confined concrete”.
Journal of Struct. Engng. ASCE, Vol. 114 n.8, pp. 1804-1826.
Yang, X., Nanni, A. and Chen, G. (2001), “Effect of corner radius on the performance of externally bonded FRP
reinforcement”. Proc. of FRPRCS5, Non-metallic reinforcement for concrete structures, Cambridge, UK, July,
pp. 16-18.
Campione G , Miraglia N. and Papia M, (2004), “Strength and strain enhancements of concrete columns confined
with FRP sheets”. Structural Engineering and Mechanics, Vol. 18, n. 6, pp. 769-790.
Karam G. and Tabbara M. (2005), “Confinement effectiveness in rectangular concrete columns with Fiber
reinforced polymer wraps”. J. of Comp. for Constr., ASCE,, Vol. 9, n. 5, pp. 388-396.
Campione G. (2006),” Influence of FRP wrapping techniques on the compressive behavior of concrete prisms”,
Cement & Composite Composites, Vol.28, n.5, pp. 497-505.

201
202
Part VI. Creep and Sustained Loads
Third International Conference on FRP Composites in Civil Engineering (CICE 2006)
December 13-15 2006, Miami, Florida, USA

SERVICEABILITY OF HIGH STRENGTH CONCRETE BEAMS WITH

INTERNAL FRP REINFORCEMENT UNDER SUSTAINED LOAD
Shawn P. Gross
(Associate Professor, Villanova University, Villanova, PA, USA)

Joseph Robert Yost

(Associate Professor, Villanova University, Villanova, PA, USA)

Jonathan V. Crawford
(Staff Engineer, French and Parrello Associates, Wall, NJ, USA)

ABSTRACT
Over the past several years, extensive research has studied the serviceability of concrete members reinforced with
internal FRP bars. Numerous experimental tests have been conducted and there exists an extensive body of data in
the literature on deflections and crack widths observed during such tests. This work has resulted in the continuous
development of design guidelines published by the American Concrete Institute (ACI) Committee 440 and the
Canadian Standards Association (CSA). However, there is a relative lack of data on the behavior of flexural
members subjected to sustained service-level loading. The ACI and CSA design guidelines clearly indicate the need
for additional research in this area, particularly since the design of FRP-reinforced concrete is often governed by
serviceability. This paper reports the results of an experimental test program is which six beams (two steel, two
GFRP, and two CFRP) were maintained under a constant load for a period of 90 days. Measured flexural crack
widths and midspan deflections are reported over the duration of the test period. Measured data is compared to the
limited results available from other published studies and to the current ACI and CSA provisions.

KEYWORDS
Serviceability, deflection, crack width, sustained load, high strength concrete

1. INTRODUCTION
A large body of research has demonstrated that the instantaneous service load performance of FRP-reinforced
members can be predicted using techniques similar to those used for steel-reinforced concrete, with modifications to
account for the lower stiffness of the FRP reinforcement as compared to steel. For example, instantaneous
deflections are often computed using a modified form of the well-known Branson effective moment of inertia
equation, and crack width estimates are often computed using a modified form of the well-known Gergely-Lutz
equation. The same approach – that of using methods developed for steel-reinforced concrete with modification as
necessary – has been proposed for time-dependent serviceability behavior, but these proposals have not yet been
compared to a large body of research data. The purpose of this study is to complement the small existing database
of results on the performance of FRP-reinforced concrete under sustained service level loads.

A few published studies (Brown, 1996; Vijay and GangaRao; 1998; Arockisamy et al., 1998; Gross et al. 2003) have
examined the time-dependent deflection of FRP-reinforced beams. These results have generally shown that the
deflection increase over time, as a percentage of the initial deflection, is significantly less than the deflection
increase suggested by ACI 318 (ACI 318, 2005). For reference, ACI 318 projects increases (ratio of time-dependent
to initial deflection) of 100% after three months of sustained loading, 120% after six months, 140% after one year,
and 200% for five years or more of sustained load. Brown also compared the results to companion steel-reinforced
beams and found the percentage deflection increase to be less for FRP than for steel. On the basis of the limited data
available, ACI 440 (ACI 440, 2006) suggests using 60% of the stated ACI 318 values for FRP-reinforced concrete,
while the CSA S806 (CSA, 2002) standard for FRP adopts the ACI 318 values without modification.

203
Additional experimental research (Vijay and GangaRao, 1998; Bakis and Boothby 2004) has suggested that flexural
crack widths in FRP-reinforced concrete members increase over time. For example, Bakis and Boothby found that
crack widths increased 40% and 60% in indoor and outdoor environments, respectively, for GFRP-reinforced
concrete beams under sustained load for three years. However, this time-dependent increase in crack widths is not
directly considered by ACI 440 and CSA S806, each of which provides an expression to estimate the probable
maximum crack width under full service load. The expressions for each standard are essentially expressions derived
for steel-reinforced concrete and are used with simple modification for FRP. The CSA S806 standard implicitly
identifies maximum acceptable crack widths of 0.020 in. for exterior exposure and 0.028 in. for interior exposure
conditions, and ACI 440 endorses the use of the CSA S806 limits.

2. EXPERIMENTAL TEST PROGRAM

The experimental test program conducted at Villanova University consisted of three pairs of identical beam
specimens with details as indicated in Table 1. Each beam measured 4.5 in. wide by 7.25 in. deep in cross-section
and all specimens were constructed 74 in. long to allow for a 72 in. testing span with a 4 inch constant moment
region. These dimensions were selected after consideration of existing ACI 440 and CSA S806 design provisions,
and result in a span-to-depth ratio of approximately 10 such that serviceability criteria do not excessively limit the
design efficiency. Reinforcement was chosen such that all beams had similar nominal moment capacities with the
same cross-section dimensions. One inch of clear cover was provided for all flexural reinforcement. Beams did not
contain transverse shear reinforcement.

All specimens were fabricated in a single pour using high strength concrete with a design compressive strength f’c of
9000 psi. The concrete mix included 8% silica fume replacement of Portland cement (by weight) and a high-range
water reducing admixture. The tested compressive strength of companion cylinders sampled from the beam pour
was 8970 psi at 28 days and 9350 psi at 152 days, when the beams were first placed under load. The modulus of
elasticity of the concrete at 152 days was 6.77 x 106 psi. Creep tests were also conducted on a single 28 in. long
specimen using the procedure outlined in ASTM C512. The creep coefficient for the concrete loaded at 152 days
and held under a sustained load for 90 days was found to be 0.51.

Table 1: Test Specimen Details

Specimen S1 & S2 G1 & G2 C1 & C2

GFRP – Hughes Bros. CFRP – Hughes Bros.
Reinforcement Type Steel
Aslan 100 Aslan 200
fy (ksi) 60 * N/A N/A
fu (ksi) N/A 104 356
E (psi) 29.0 x 106 * 5.85 x 106 20.2 x 106
Ar (sq. in.) 0.62 (2 #5) 0.62 (2 #5) 0.20 (2 #3)
ρ/ ρ b 0.46 3.08 3.64
Calculated Mcr (kip-in) 28 28 28
Calculated Mn (kip-in) 201 186 195
Applied Ma (kip-in) 70 65 68
Applied Msus (kip-in) 60 56 59
* Assumed value

Beams were oriented vertically and loaded in pairs in a self-reacting frame as shown in Figure 1. Prior to
application of the sustained load, each beam was loaded with a short-term “pre-cracking” load to simulate the effects
of live load. Pre-cracking loads (Ma) correspond to 35% of each beam’s respective nominal moment capacity, while
sustained loads (Msus) correspond to 30% of the nominal capacity. For each beam pair, the beams were loaded in
unison using a hydraulic cylinder for application of the pre-cracking load or a precompressed spring for the
sustained load, such that the left beam of each pair experiences tension along the left extreme fiber and the right
beam of each pair experiences tension along the right extreme fiber. The spring mechanisms used for sustained
loading utilized a set of nuts that could be adjusted to recompress the spring and maintain a constant load as the
beams continued to deflect over time. Load for each beam pair was monitored using a load cell connected to a data
acquisition system, and deflections were monitored using linear variable differential transducers (LVDTs) mounted
to the test fixture. The tests were conducted indoors, but not in a room with constant temperature and relative
humidity conditions.

204
 S1 S2 G1 G2 C1 C2

Support Points LVDT

Spring
Loading Points

Load Cell

Figure 1: Sustained Load Test Setup

0.35 0.025

C2; 0.318
0.30
C1; 0.286 C2; 0.021
C2; 0.279 0.020
0.25 C1; 0.243 G2; 0.017 G2; 0.018
G1; 0.225 C2; 0.017 G1; 0.017
Crack Width (in.)
Deflection (in.)

C1; 0.016 C1; 0.017

0.20 0.015
G2; 0.184 G1; 0.014
G1; 0.179
0.15
G2; 0.131
S1; 0.106 S1; 0.128 0.010
S2; 0.113
0.10
S2; 0.089 S2; 0.007
S1, 0.006
0.05 0.005 S1; 0.005
S2, 0.004

0.00
0 10 20 30 40 50 60 70 80 90 0.000
-20 -10 0 10 20 30 40 50 60 70 80 90
Time, days
Time (Days)

Figure 2: Deflection Test Results Figure 3: Crack Width Test Results

Table 2: Comparison of Long-Term Deflection Data with Published Results

Reinf Time δlong-term/ δinitial Ratio – Meas./Pred.

Source / Specimen Type Under Meas. ACI 440 CSA S806 ACI 440 CSA S806
Load (a) (b) (c) (a)/(b) (a)/(c)
#10 GFRP 6 mo 0.61 0.72 1.20 0.85 0.51
Brown (1996)
#15 GFRP 6 mo 0.92 0.72 1.20 1.28 0.77
Vijay and T2CR GFRP 21.5 mo 0.86 0.96 1.60 0.90 0.54
Gangarao (1998) C1CR GFRP 12 mo 1.08 0.84 1.40 1.29 0.77
B2 CFRP 15.5 mo 1.15 0.90 1.50 1.28 0.77
Arockisamy et
B3 CFRP 15.5 mo 0.65 0.90 1.50 0.72 0.43
al. (1998)
B4 CFRP 15.5 mo 0.71 0.90 1.50 0.79 0.47
G1 GFRP 3 mo 0.26 0.60 1.00 0.43 0.26
G2 GFRP 3 mo 0.40 0.60 1.00 0.67 0.40
Current Study
C1 CFRP 3 mo 0.18 0.60 1.00 0.30 0.18
C2 GFRP 3 mo 0.14 0.60 1.00 0.23 0.14
AVERAGE 0.87 0.52

205
3. RESULTS: DEFLECTIONS
Measured midspan deflections for all beams over the 90 day test period may be seen in Figure 2. All beams
behaved similarly, with gradual deflection increases over time. The average deflection increase (ratio of time-
dependent to initial) over the test period was found to be 24%, 33%, and 16% for the steel-, GFRP-, and CFRP-
reinforced specimens, respectively. These deflection increases for the FRP-reinforced specimens are significantly
less than the 100% increase over 90 days predicted by CSA S806 or the 60% increase predicted by ACI 440. The
results are compared to selected results from published studies in Table 2. The last two columns in the table indicate
that the ACI 440 approach is the more accurate method. However, there is clearly a high degree of variability in the
results. The relatively low deflection increases for the beams in the current study, as compared to the other
published results, can be explained by low creep coefficient measured for the high strength concrete used to
fabricate the beams. However, the larger percentage increase for the GFRP-reinforced specimens (33%) as
compared to the steel-reinforced specimens (24%) contradicts the previous findings of Brown (1996) and the
underlying principle of the ACI 440 approach.

4. RESULTS: CRACK WIDTHS

Measured maximum crack widths are shown in Figure 3. As expected, the GFRP- and CFRP-reinforced beams
exhibited significantly larger crack widths than the steel-reinforced beams. All four FRP-reinforced specimens
exhibited maximum crack widths close to the CSA S806 and ACI 440 limit of 0.020 in. for exterior exposure at the
end of the 90 day test period. The average crack width increase (ratio of time-dependent to initial) over the test
period was found to be 20%, 13%, and 15% for the steel-, GFRP-, and CFRP-reinforced specimens, respectively.
These time-dependent increases are consistent with the findings of previous researchers mentioned previously,
however it should be noted that almost all of the increase occurred during the first two weeks under sustained load.

5. CONCLUSIONS
This study of FRP-reinforced high strength concrete beams under a 90 day period of sustained loading showed that:
1. Deflections increased by 33% and 16% for the GFRP- and CFRP-reinforced beams, respectively, over the
duration of the test period.
2. The existing ACI 440 approach for estimation of time-dependent deflections compared more favorably to
the results than the CSA S806 approach, but both significantly overpredicted the time-dependent increase.
3. Crack widths increased by 13% and 15% for the GFRP- and CFRP-reinforced beams, respectively, over the
duration of the test period.

6. REFERENCES
ACI Committee 318 (2005). Building Code Requirements for Structural Concrete (ACI 318-05), American Institute,
Farmington Hills, MI.
ACI Committee 440 (2006). Guide for the Design and Construction of Concrete Reinforced with FRP Bars (ACI
440.1R-06), American Concrete Institute, Farmington Hills, MI.
Arockiasamy, M., Amer, A., and Shahawy, M. (1998). “Environmental and Long-Term Studies on CFRP Cables
and CFRP Reinforced Concrete Beams,” Proceedings of the First International Conference on Durability of
Composites for Construction, B. Benmokrane, and H. Rahman, eds., Sherbrooke, Quebec, pp. 599-610.
Bakis, C. E.; and Boothby, T. E. (2004). “Evaluation of Crack Width and Bond Strength in GFRP Reinforced Beams
Subjected to Sustained Loads,” Proceedings of the Fourth International Conference on Advanced Composite
Materials in Bridges and Structures - ACMBS-IV, Calgary, Canada, CD-ROM
Brown, V. and Bartholomew, C. (1996). “Long-term Deflections of GFRP-Reinforced Concrete Beams,”
Proceedings of the First International Conference on Composites in Infrastructure (ICCI-96), Tucson, Ariz., pp.
389-400.
Canadian Standards Association (2002). Design and Construction of Building Components with Fibre-
Reinforced Polymers (CSA S806-02), Canadian Standards Association, Toronto, Ontario, Canada.
Gross, S.; Yost, J.; and Kevgas, G. (2003). "Time-Dependent Behavior of Normal and High Strength Concrete
Beams Reinforced With GFRP Bars Under Sustained Loads," High Performance Materials in Bridges, American
Society of Civil Engineers, pp. 451-462.
Vijay, P. V.; and GangaRao, H. V. S. (1998). “Creep Behavior of Concrete Beams Reinforced with GFRP Bars,”
Proceedings of the First International Conference (CDCC 1998), Durability of Fiber Reinforced Polymer (FRP)
Composites for Construction, pp. 661-667.

206
Third International Conference on FRP composites in Civil Engineering (CICE 2006)
December 13-15 2006 , Miami, Florida, USA.

DEFORMATIONAL BEHAVIOUR OF FRP CONFINED CONCRETE

UNDER SUSTAINED COMPRESSION
Rajeev Kaul
(Ph.D Candidate, Centre for Built Infrastructure Research, University of Technology Sydney, Australia)

R. Sri Ravindrarajah
(Senior Lecturer, Centre for Built Infrastructure Research, University of Technology Sydney, Australia)

Scott T. Smith
(Senior Lecturer, Centre for Built Infrastructure Research, University of Technology Sydney, Australia)

ABSTRACT
Confining concrete is an effective method to enhance the strength and ductility of reinforced concrete columns.
Fibre reinforced polymer (FRP) composites are emerging as a suitable confining material to replace conventional
materials such as steel and fibre-reinforced cement composites. Past research on the behaviour of FRP confined
concrete in compression is considerable; however, limited research has been reported on the behaviour of confined
concrete under sustained compressive loading. This paper reports the preliminary results of an experimental
investigation on the deformational behaviour of carbon FRP (CFRP) confined concrete columns under sustained
compressive stress levels, corresponding to 40% and 60% of the unconfined concrete compressive strength for up to
150 days. The results show that the creep of confined concrete columns is marginally influenced under moderate
sustained stress/strength ratios.

KEYWORDS
Carbon fibre reinforced polymer, CFRP, concrete, confinement, time dependent behaviour, creep

1. INTRODUCTION

Fibre reinforced polymer (FRP) composites have significantly high tensile strength, stiffness and strain capacity
compared to concrete. These qualities contribute to the acceptance of FRP as a viable material for strengthening and
rehabilitating of concrete structures. In addition, FRP is a durable material in normal exposure conditions and is
capable of wrapping any shaped concrete sections. Based on considerable research on the performance of FRP
strengthened concrete members under short-term loadings, design recommendations (fib 2001, ACI 2002, Teng et
al. 2002) have been published. However, a lack of research on the time-dependent effectiveness of FRP deters a
wide-spread usage of FRP (Karbhari et al. 2003). Creep of FRP confined concrete under sustained loading is an
important durability parameter in the design of FRP strengthened concrete columns (Smith et. al. 2005), and in some
early applications, creep rupture of FRP in strengthened concrete columns has been reported (Naguib and Mirmiran
2002). This paper presents preliminary results of an experimental investigation on the creep behaviour of carbon
FRP (CFRP) confined concrete columns, under moderate sustained stress/strength levels.

2. FRP CONFINEMENT TO CONCRETE COLUMNS

Fibre reinforced polymer (FRP) confinement improves the strength and ductility of concrete columns in
compression. The strength Improvement depends upon the fibre type, thickness and wrapping orientation (Lam and
Teng 2003, Matthys et. al. 2005). The FRP, which confines the lateral expansion of concrete, is in tension in the
hoop direction and fails in a brittle explosive manner when the concrete is excessively compressed. Research has
shown that the tensile strength of FRP in the hoop direction is 10 to 20% lower than that in direct tension (Xiao and
Wu 2000), where the ratio of failure hoop strain to that in direct tension is termed as the reduction factor. In FRP
confined concrete under sustained compression, concrete experiences compressive creep whereas the FRP is under

207
sustained tension. As the concrete creeps with time, time-dependent increase in confinement stress in FRP will occur
due to load transfer from the low modulus concrete to the relatively high modulus FRP. The interfacial bond stress
between FRP and concrete will also experience time-dependent changes. Creep of FRP composite is a combined
response of fibres, resin and interfacial behaviour of fibre and the matrix under sustained tension. In addition, the
temperature will have a significant influence on the instantaneous and time-dependent deformation of FRP. Naguib
and Mirmiran (2002) experimentally investigated the behaviour of glass FRP (GFRP) confined concrete under
sustained compression and showed that the effect of confinement on creep of the concrete core is not significant.
They also found the ACI 209R-92 (1992) model overestimates the creep of FRP confined concrete and the creep
coefficients for the FRP tubes. The difference in the case of FRP wrapped confinement was not however significant.
Analytical studies indicated that the creep of FRP wrapped columns is similar to that of sealed concrete.

Load Cell
Hydraulic Jack

Strain gauge points

150 X 300 mm Concrete

Dummy Blocks

Figure 1: FRP confined concrete cylinders under sustained compression.

3. EXPERIMENTAL SET-UP AND INSTRUMENTATION

Normal weight medium strength concrete (28-day cylinder strength of 37 MPa) was used in this study. A number of
150 x 300 mm concrete cylinders were cast in standard steel moulds and cured in water at 20oC for 28 days prior to
testing. The cylinders were then either confined by wrapping with two layers of unidirectional CFRP formed in a
wet lay-up procedure in the hoop direction or unconfined but sealed by coating with epoxy resin used in the
formation of the CFRP. The confined and unconfined cylinders were subjected to either short-term uniaxial
compressive loading until failure at specified ages or sustained uniaxial compression (up to 150 days) in self-
reacting creep rigs, as shown in Figure 1. Each creep rig had a pair of either CFRP confined or unconfined cylinders.
The sustained stress intensity corresponded to 40 and 60% of the cylinder strength of the unconfined concrete at the
ages of loading (Table 1). The confined compressive strength of the cylinders at the ages of loading is also given in
Table 1 (last column in brackets) as well as the sustained stress intensity expressed as a percentage of the cylinder
strength of confined concrete. Loads were applied via hand-operated for 40% stress level and electrically-operated
hydraulic jacks for 60% stress level and the loads were monitored through calibrated load cells. Periodic load
adjustments were made to ensure that appropriate sustained stresses are maintained on the creep cylinders. The
confined cylinders loaded were air stored at the uncontrolled laboratory environment for 14 and 32 days respectively
after the removal of the cylinders from water curing prior to application of the CFRP.

Table 1: Details of concrete specimens under sustained compressive loading

Age of concrete Duration of Applied stress level with respect to strength of

Creep Rig Specimen # at loading load Unconfined concrete Confined concrete
No. 3 P3, P4 60 days 150 days
40% (of 42.5 MPa%) 26% (of 65.0 MPa%%)
No. 4 F3, F4 60 days 150 days
No. 5 P5, P6 100 days 150 days
60% (of 47.5 MPa%) 43% (of 65.1 MPa%%)
No. 6 F5, F6 100 days 150 days
#
P = Plain unconfined cylinder coated with epoxy resin: F = CFRP confined cylinder
%
= unconfined compressive strength at age of loading: %% = confined compressive strength at age of loading

The axial (longitudinal) deformations on both the concrete and CFRP surfaces were measured using a demountable
mechanical strain gauge (DEMEC) over a 200mm gauge length. The DEMEC points were mounted on diametrically
opposite sides of each cylinder on (i) the concrete surface in unconfined cylinders, or in confined specimens after
small holes were drilled through the CFRP, and (ii) on the surface of the CFRP in confined cylinders at a quarter of

208
the circumferential distance from the concrete DEMEC points. The hoop strain on the CFRP was measured using
electrical resistance strain gauges over 67 mm gauge length. Drying shrinkage of companion non-loaded, non-
wrapped and non-epoxy coated cylinders, stored in the same room and environment as the creep rigs, was measured
throughout the whole duration of the experiment. The creep test was conducted in uncontrolled laboratory
environment having the mean temperature and relative humidity of 23oC and 60% respectively. Tension tests were
conducted on eight identical two layered, 15 mm wide, 252 mm long and 0.234 mm thick (2 layers of 0.117 mm
CFRP nominal thickness sheet) coupons to determine the tensile strength, rupture strain and modulus of elasticity of
the CFRP.

4. RESULTS AND DISCUSSION

The mean tensile strength, rupture strain, and modulus of elasticity of the CFRP coupons was 3030 MPa, 1.10 % and
246 GPa with standard deviations of 91 MPa, 0.05 % and 11 GPa, respectively. The mean confined compressive
strength of concrete at 28 days was 65.0 MPa compared to 37.0 MPa for the unconfined concrete at the same age.
Thus, CFRP confinement increased the cylinder strength by approximately 75%. The sustained stress/strength ratios
of the confined cylinders were 26% (at 60 days of age) and 43% (at 100 days of age) which corresponded to 40%
and 60% of the unconfined cylinders as given in Table 1. The stress-strain relationships for confined and unconfined
concrete in compression indicate that the modulus of elasticity of concrete was not significantly affected by the
confinement. The values for the modulus of elasticity, calculated by dividing the applied stress by the instantaneous
strain on loading (AS 1012.16 1996) of the creep rigs for the unconfined and confined cylinders were 26.3 GPa and
27.8 GPa at the 40% stress level and 28.9 GPa and 29.4 GPa at the 60% stress level, respectively. Even though the
strength of the concrete changes with time (fifth column in Table 1), the strength of the confined concrete cylinder is
virtually identical (sixth column in Table 1) as the failure is primarily dependent on the confinement of the CFRP.

0.8 0.8
Unconfined Cylinder
0.6 Confined Cylinder
Creep Coefficient

0.6
Creep Coefficient

0.4 0.4
Unconfined Cylinder

0.2 Confined Cylinder 0.2

0.0 0.0
0 20 40 60 80 100 120 140 160 0 20 40 60 80 100 120 140 160
Time - days Time- days

(a) 40% stress level (b) 60% stress level

Figure 2: Creep of confined and unconfined concrete: up to 150 days of loading

The time dependent behaviour of loaded concrete is evaluated via its creep behaviour with time. Creep intensity is
expressed in terms of creep coefficient which ratio of creep strain to the instantaneous strain on loading. The creep
strain is equal to the total strain on the loaded cylinders minus the drying shrinkage strain from non-loaded
companion cylinder. Figure 2 shows the development of creep coefficient with time for both confined and unconfined
concrete up to 150 days, under both sustained stress levels.

Table 2: Creep coefficient for FRP confined and unconfined concrete

Time Creep Coefficient @ 40% Stress Level Creep Coefficient @ 60% Stress Level
(Days) Unconfined (Φ1) Confined (Φ2) Φ2/ Φ1 Unconfined (Φ1) Confined (Φ2) Φ2/Φ1
28 0.47 0.49 1.04 0.22 0.24 1.09
90 0.62 0.70 1.12 0.41 0.34 0.84
150 0.67 0.70 1.04 0.54 0.36 0.67

The creep coefficient at 150 days for unconfined concrete at the 40% stress level was 0.67 compared to 0.54 for that
under 60% stress level. This is partly due to the increase in the age at loading from 60 to 100 days.. With the
confined concrete more significant drop from 0.70 to 0.36 was observed after 150 days. Also evident in Figures 2a
and 2b is the creep in confined cylinders stabilised earlier than the unconfined cylinders. This is a consequence of
the latter having lower concrete stress levels. The rate of creep for the unconfined cylinders increased as the applied
stress is increased. This is particularly evident in the unconfined cylinder at 60% loading in which a high stress level

209
has led to more significant microcracking which may have increased the rate of creep in concrete (Sri Ravindrarajah
and Swamy 1989). At both stress levels, the minimally activated FRP is mainly acting as a moisture barrier to the
concrete. This was confirmed by small FRP hoop strain measurements. However, when the sustained stress level is
greater than the unconfined strength of concrete, the creep response of confined concrete may be influenced by the
creep properties of the FRP. Research is in progress to study the creep of confined concrete under high sustained
stress levels.

Table 2 summarises the creep coefficients after 28, 90 and 150 days of loading. The ratio of creep coefficient for
confined to unconfined concrete was 1.04, 1.12 and 1.04 after 28, 90 and 150 days, respectively at the 40% stress
level compared to a more decreasing trend with the ratios at 1.09, 0.84, and 0.67 at 60% stress levels. The increasing
creep coefficient of the unconfined cylinder as opposed to the stabilised creep coefficient for the confined cylinders
from 90 to 150 days results in a reduction of the ratios in Table 1.

5. CONCLUSION
Compressive strength, modulus of elasticity and creep of concrete confined with 2 layers of CFRP were studied.
Concrete cylinders were subjected to constant sustained stress levels, corresponding to 40% and 60% of the
unconfined concrete strength at the age of loading of 60 and 100 days, for up to 150 days. The confinement
increased the concrete 28-day cylinder strength by 75%. At the stress to strength level of 40%, no significant effect
of CFRP on the creep of concrete was noted, however the rate of creep slowed quicker in the confined than the
unconfined specimens. Current research is undertaken to investigate the creep of confined concrete under high
sustained stress levels. Creep tests on CFRP coupons subjected to different conditions are also being conducted in
order to give a better understanding of variations in moisture and temperature in the uncontrolled laboratory
conditions, and stress distribution and creep effects in a CFRP confined system. Time-dependent tests will also be
conducted on GFRP confined cylinders in the future, where GFRP has been traditionally observed to creep more
than CFRP.

6. REFERENCES

ACI 209R-92 (1992). Prediction of Creep, Shrinkage and Temperature Effects in Concrete Structures. American
Concrete Institute (ACI), Committee 209, USA.
ACI 440.2R-02 (2002). Guide for the Design and Construction of Externally Bonded FRP Systems for Strengthening
Concrete Structures. American Concrete Institute (ACI), Committee 440, USA.
AS 1012.16 (1996). Methods of Testing Concrete - Determination of Creep of Concrete Cylinders in Compression.
Standards Australia, Australia.
fib (2001). Externally Bonded FRP Reinforcement for RC Structures. The International Federation for Structural
Concrete (fib), Technical Report, Task Group 9.3, Bulletin No. 14, Lausanne, Switzerland.
Karbhari, V.M., Chin, J.W., Hunston, D., Benmokrane, B., Juska, T., Morgan, R., Lesko, J.J., Sorathia, U. and
Reynaud, D. (2003). “Durability gap analysis for fibre-reinforced polymer composites in civil engineering”. Journal
of Composite Construction, Vol. 7, No. 3. pp. 238-247.
Lam, L. and Teng, J.G. (2003). “Design oriented stress strain model for FRP confined concrete”. Construction and
Building Materials, Vol. 17, pp. 471-489.
Matthys, S., Toutanji, H., Audenaert, K. and Taerwe, L. (2005). “Axial load behaviour of large- scale columns
confined with fibre reinforced polymer composites”. ACI Structural Journal. Vol. 102, No. 2, pp. 258-267.
Naguib, W. and Mirmiran, A. (2002). “Time dependent behaviour of fibre reinforced polymer confined concrete
columns under axial loads”. ACI Structural Journal, Vol. 99, No. 2, pp. 142-148.
Smith, S.T., Kaul R., Sri Ravindrarajah, R. and Otoom, O.M.A. (2005). “Durability considerations for FRP
strengthened RC structures in the Australian environment”. Proceedings (CD-Rom), Australasian Structural
Engineering Conference, ASEC, 11-14 December, Newcastle, Australia.
Sri Ravindrarajah, R. and Swamy, R.N. (1989), “Load effects on fracture of concrete”. Materials and Structures,
Vol. 22, No. 127, pp. 15-22.
Teng, J.G., Chen, J.F., Smith, S.T. and Lam, L. (2002). FRP-Strengthened RC Structures. John Wiley & Sons, UK.
Xiao, Y. and Wu, H. (2000). ‘‘Compressive behaviour of concrete confined by carbon fibre composite jackets’’.
Journal of Materials in Civil Engineering, Vol. 12, No. 2, pp. 139-146.

210
Third International Conference on FRP Composites in Civil Engineering (CICE 2006)
December 13-15 2006, Miami, Florida, USA

LONG -TERM BEHAVIOUR OF CFRP LAMINATES:

AN EXPERIMENTAL STUDY
Francesco Ascione
(PhD Student, Department of Civil Engineering, University of Rome “Tor Vergata”, Rome, Italy)

Geminiano Mancusi
(Assistant Professor, Department of Civil Engineering, University of Salerno, Fisciano, Italy)

ABSTRACT
In this paper the authors present many experimental results on the creep behaviour of CFRP pultruded laminates
commonly used for civil structural applications. The study considers both a low stress-state in the laminate,
representative of service conditions, and a higher one which is about 70% of CFRP tensile strength. The results here
discussed, which only concern the first phase of the experimental study (low stress state), show that creep effects,
when high longitudinal normal modulus is considered, can be assumed small.

KEYWORDS
Composite structures, FRP, Pultruded shapes, Long-term behaviour, Creep.

1. INTRODUCTION
Over the past few years, the strengthening of concrete structures with externally epoxy-bonded fiber reinforced
polymer (FRP) composites has become increasingly user-friendly among engineers and researchers worldwide
(Barbero E., 1998; Barbero E. and Harris J.S., 1998; Holloway L.C. and Leeming M.B., 1999; Ascione L. et.al.,
2004). With regard to this technique, one of the most relevant topics is represented by the durability and the
reliability, closely related to the time-depending behaviour of fibre-reinforced composite (FRP) materials (Ferry
J.D., 1980; Ma C.C.M et.al, 1997; Maksimov R.D. and Plume E., 2001; Triantafillou T.C. and Plevris N, 1994;
Petermann J. and Schulte K., 2002). In fact, FRPs are made up of a matrix, highly sensitive to creep, and of purely
elastic fibres. When dealing with aged concrete structures, the rate of the viscous deformations in the concrete can
be considered equal to zero. Consequently, only FRP creep deformations have to be taken into consideration. A
similar problem occurs in the case of the steel-concrete composite structures. In this case, in fact, creep phenomena
only concern the concrete component, while the steel component does not flow (CEB-FIP, 1984).
A wrong evaluation of the viscous effects could introduce relevant errors in the design of the structural plating. At
the moment, current literature does not provide any final conclusions on this important topic. Furthermore, studies
developed in the aeronautic or naval field dealing with the creep properties of composite materials cannot be directly
used for civil engineering purposes, mainly due to the different technological processing of these materials. From a
more practical point of view, guide-lines and technical codes only introduce stress limitations in order to verify the
safety of composite materials when there are viscous effects (JSCE 1997; ACI Committee 440, 2000; CEB-FIP,
2001; CNR DT-200, 2004).
The aim of the paper is to show some experimental results on the creep behaviour of a CFRP laminate investigated
by a specific approach, as described below, confirming that such effect can be assumed small if a high value of
longitudinal normal modulus of the laminate is considered. The results here discussed belong to the initial phase of a
two years long experimental programme currently underway at the University of Salerno.

2. DESIGN OF THE EXPERIMENT

As mentioned above, in order to study the creep behaviour of a CFRP laminate a long term (two years) experimental
programme is being carried out.

211
It concerns a thin CFRP plate bonded to the top flange of a simply supported titanium beam (Ti-6Al-4V). More
details are shown in Figg.1a-b. Within this experiment the actual geometrical conditions occurring in the retrofitting
of a concrete structure have been simulated. In fact, the behaviour of the beam, made of a high performance titanium
alloy, is not time-dependent as is the case of aged concrete structures. Moreover, like FRP, the titanium alloy also
exhibits linear elastic response up to very large normal strains. As a consequence, it is possible to evaluate the
stresses inside the materials only by measuring corresponding strains at any given time.

F: dead load
a
F: dead load Fb Fb
auxiliary steel beam CFRP plate

titanium alloy beam

b a b
a

Figure 1.a: Global experimental scheme. Figure 1.b: Experimental scheme (central bay).

The static scheme adopted by the authors is represented in Fig. 1.a. The total length of the scheme is equal to
7300mm = a + 2 · b, where a is the length of the titanium alloy beam (3500mm), while b (1900mm) denotes the
length of each lateral rigid auxiliary beam. Last ones are connected to previous by bolted joints (Fig. 2.a). The bond
of CFRP plate to titanium alloy beam has been made by using Loctite Multibond 330 adhesive. Finally, the length of
the laminate, a', is equal to 2900 mm.
Dead loads have been applied in symmetry at the free ends of the two rigid auxiliary beams, according to the scheme
in Fig.1.a. In particular, a defined number of 1000mm×200mm×10mm steel sheets, about 0,20 kN each one
(Fig.2.b), have been used. It is easy to verify: i) constant bending moment ( M = - F · b ) occurs within the supports
(Fig. 1.b); ii) no shear stresses arise.

Figure 2.a: Bolted joints. Figure 2.b: Dead loads.

CFRP plate

Ti-6Al-4V beam

Figure 3: Cross-section of the beam.

The cross-section of the beam, after the CFRP laminate has been bonded, is shown in Fig. 3 (length unit: mm). The
most important mechanical properties of the titanium alloy, CFRP and adhesive are described below (Tab.1).

212
 Table 1: Mechanical properties

Multibond
Ti-6Al-4V CFRP
330
Longitudinal elastic modulus [ N / mm2 ] 110000 300000 -
Yield normal stress [ N / mm2 ] 790 (20°C) - -
Ultimate normal stress [ N / mm2 ] 895 (20°C) 1450 -
Ultimate shear stress [ N / mm2 ] - - ≥16,5
Ultimate normal strain 0,1000 ≥0,0045 -

The experimental programme has been divided into two phases, each one performed with a fine control of
temperature (20°C ± 1°C) and humidity (50% ± 5%):
-1st phase: generation of a low stress state (20% of CFRP tensile strength) in the laminate over a 6-month
period. This phase has been completed, with the results obtained by the authors being discussed here for the first
time.
-2nd phase: generation of a higher stress level (70% of CFRP tensile strength) over a 18-month period. This
phase is still underway.
The strain state evolution is monitored by a continuous data acquisition hardware/software system using 35 strain-
gauges applied both to the titanium beam as well as the CFRP plate (Fig. 4.a) at defined positions along the
longitudinal axis. Due to the linear elastic response of both materials, stresses can be easily related to strains. An
optometric system, OptoNCDT 1401 by µε Micro Epsilon, is used in order to measure the mid-span deflection over
time.

a b c d e

f
Figure 4.a: Some strain gauges. Figure 4.b: Strain-gauges position (mid-span).

3. EXPERIMENTAL RESULTS
In this section the results obtained during the first phase of the experimental programme are discussed.
As indicated above, this phase concerns a continuous monitoring of the stress-strain state over a six-month period.
Dead loads have been applied so that the initial stress state in the CFRP laminate is about 20% of its ultimate tensile
strength.
The following Fig.5.a shows the values returned by strain-gauges at defined times (every ten minutes) over six
months. They refer to the mid-span cross-section (Fig. 4.b). In particular, the cyan, yellow and magenta lines refer to
strain-gauges at the top of the CFRP laminate (b, c and d positions), while the blue one refers to the bottom of the
CFRP plate. This last value has been established by measuring strains in the titanium alloy, close to the laminate (a
and e positions).
Fig.5.b highlights the evolution of vertical deflection at mid-span cross-section of the beam, evaluated by an
optometric sensor applied at the bottom flange (f position in Fig. 4.b).
As it can be noted, strains inside the CFRP follow a typical secondary-creep evolution law. The percentage
difference is equal to 2,3 % over 180 days (magenta line). On the other hand, strains in the titanium alloy increase
exhibiting a percentage difference equal to 1,0 %. Finally, the deflection at mid-span cross-section increases

213
following a quasi-linear evolution law, characterized by a maximum variation equal to 6,7 % over the whole
monitored time.

890 30
normal strain [ 106 ]

deflection [mm]
870 20
d
b
850 10
a,e

830 0
30 60 90 120 150 180 30 60 90 120 150 180
elapsed time [days] elapsed time [days]

Figure 5.a: Normal strains over time (mid-span). Figure 5.b: Deflection over time (mid-span).

4. CONCLUSIONS
As it is easy to realize, the experimental results show that the investigated CFRP pultruded laminate, characterized
by a high longitudinal normal modulus, exhibits small creep strains. In particular, the data obtained have been
compared with results of classical creep tests on the same manufactured laminate (Berardi V.P. et al., 2006).
Comparisons highlight a less sensitive to creep behaviour and confirm the durability and the reliability of the use of
CFRP laminates in structural plating purposes.

5. REFERENCES

ACI Committee 440 (2000). “Guide for the design and construction of externally bonded FRP systems for
strengthening concrete structures”.
Ascione L., Berardi V.P., Mancusi G. (2004). “Time-depending behaviour under sustained loads of Rc beams
externally plated with FRP laminates”, Proceedings of IMTCR International Conference. University of Lecce, Italy.
Barbero E. (1998). “Introduction to composite material design”, Taylor & Francis.
Barbero E. & Harris J.S.(1998). “Prediction of creep properties from matrix creep data”. Journal of Reinforced
Plastics and Composites 17(4).
Berardi V.P., Feo L., Giordano A., Ascione F. (2006). “On the creep behaviour of CFRP pultruded laminates: an
experimental study”, Proceedings of II FIB International Conference. Naples, Italy
CEB-FIP (1984). “Manual on Structural Effects of Time-Dependent Behaviour of Concrete”. Bullettin
d’Information n. 142-142bis.
CEB-FIP (2001). “Externally bonded FRP reinforcement for RC structures”.
CNR DT-200/2004. “Istruzioni per la Progettazione, l'Esecuzione ed il Controllo di Interventi di Consolidamento
Statico mediante l'utilizzo di Compositi Fibrorinforzati. Materiali, strutture di c.a. e di c.a.p., strutture murarie”.
Ferry J.D. (1980). “Viscoelastic properties of polymers”. New York: J. Wiley & Sons.
Holloway L.C. & Leeming M.B. (1999). “Strengthening of reinforced concrete structures”. CRC Press.
JSCE 1997. “Recommendation for design and construction of concrete structures using continuous fiber reinforcing
materials”.
Ma C.C.M., Tai N.H., Wu S.H., Lin S.H., Wu J.F., Lin J.M. (1997). “Creep behaviour of carbon-fiber-reinforced
PEEK [+/-45] laminated composites”. Composites: Part B, 28: 407-417.
Malvern L.E. (1969). “Introduction to the mechanics of a continuous medium”. Prentice-Hall.
Maksimov R.D. & Plume E. (2001). “Long-Term creep of hybrid aramid/glass fiber-reinforced plastics”, Mechanics
of Composite Materials 37 (4).
Petermann J. & Schulte K. (2002). “The effects of creep and fatigue stress ratio on the long-term behavior of angle-
ply CFRP”. Composite Structures 57: 205-210.
Triantafillou T.C. & Plevris N. (1994). “Time-dependent behavior of RC members strengthened with FRP
laminates”. Journal of Structural Engineering 120 (3): 1016-1042.

214
Third International Conference on FRP Composites in Civil Engineering (CICE 2006)
December 13-15 2006, Miami, Florida, USA

CREEP TESTS ON GFRP PULTRUDED SHAPES STIFFENED WITH

CARBON FIBER SHEETS
Marina Bottoni
(PhD Student, University of Bologna, Bologna, Italy)

Claudio Mazzotti
(Associate Professor, University of Bologna, Bologna, Italy)

Marco Savoia
(Full Professor, University of Bologna, Bologna, Italy)

ABSTRACT
Preliminary results of an experimental campaign on creep deformation of pultruded GFRP-shapes subject to long-
term loadings are presented. Both tension and shear tests have been performed, for the duration of about three
months. Some specimens have been stiffened by FRP carbon sheets, in order to reduce creep deformation.

KEYWORDS
pultruded shapes, creep, carbon fibers, composites

1. INTRODUCTION
As well known, verification against deformability is the most important design requirement for composite pultruded
shapes, due to their very high strength-to-modulus ratio. Moreover, in the case of glass fibers, creep deformation due
to long-term loading can be significant when compared with elastic deformation, and must then be taken into
account.
The subject of the research is the creep strain evolution of GFRP-pultruded shapes subject to long-term loadings
(Figure 1), for which few contributions can be found in the literature (Mottram, 1993, McClure and Mohammadi,
1995, Shao and Shanmugam, 2004, Berardi et al. 2006).
Since deformation of structural elements is due to both longitudinal and shear strains, two different creep functions
have been experimentally evaluated. With this aim, uniaxial tension and pure shear tests have been performed on
specimens extracted from flanges of pultruded GFRP beams. Long-term tests have been conducted in a climate
room at 20°C temperature and RH of 60 percent.
Some specimens have been also stiffened by gluing unidirectional carbon fibers with 0° fiber orientation with
respect to longitudinal direction for specimens in tension, and two-directional ±45° CFRP sheets on specimens under
shear (Figure 2). The effect of CFRP stiffening in reducing long-term deformation of pultruded elements has been
investigated.

2. GFRP PULTRUDED SPECIMENS FOR LONG -TERM TESTS

Specimens used for long-term tests have been extracted from pultruded plate elements or from flanges and web of
wide-flange pultruded beams. Plates and beams have different mechanical properties, whereas thickness of all
pultruded specimens is 9.53 mm. They are made of E-glass longitudinal fibers embedded in a polyester matrix. From
failure tests on pultruded plate elements, tensile strength of 79.5 MPa and elastic modulus E = 9.69 GPa have been
obtained. Specimens extracted from plates and flanges have been subject to long-term tension test, whereas
specimens obtained from beam webs have been used for tests on long-term shear behavior. Some specimens for both
tension and shear tests have been also strengthened by using externally glued carbon fiber sheets. With more details,

215
 Strengthened specimen Strengthened beam

CFRP unidirectional sheets

n°2 ply

9,53mm

coupon
y
extraction area
9,53mm
x x
152,4mm
Web
CFRP unidirectional sheet
n°1 ply y

152,4mm

Figure 1: Extraction of specimens from beam flanges.….. Figure 2: Two frames for creep tests on specimen.
four unstiffened specimens extracted from plates have been used for long-term tension tests. Two different widths
have been adopted, in order to test the effect of stress level on creep strain evolution with time: two specimens were
46 mm wide and 245 mm long, while the remaining two were 65 mm wide and 335 mm long. Four specimens have
been stiffened by externally gluing FRP carbon sheets. Two specimens obtained from plates are 31 mm wide and
245 mm long while other two (obtained from beam flanges) are 44 mm wide and 335 long. They have different
dimensions with respect to unstiffened specimens in order to prescribe the same stress level to pultruded elements,
calculated by homogenization criteria based on linear elastic behavior.
As for long-term shear tests, two specimens 120 mm × 220 mm (extracted from beam web) have been prepared.
Two additional specimens (from beam web), with same dimensions, have been stiffened with CFRP sheets.
Strengthening of pultruded elements subject to tension tests (Figure 1) has been done with three unidirectional plies
of carbon fiber sheets, glued to pultruded specimens (two on one side and one on the other side) extracted from
beam flanges and plates. Carbon sheets with fibers content 320 g/m2 and longitudinal elastic modulus equal to 240
GPa (according to producer data) have been used. Moreover, specimens for shear tests have been strengthened by
means of one ply of two-directional sheets with ±45° fiber orientation on both sides (carbon fiber sheets with fiber
content 200 g/m2 in each direction and longitudinal elastic modulus of 240 GPa).
As for the application of the reinforcement, an epoxy resin has been applied to pultruded specimen, followed by a
carbon fiber sheet; then, epoxy in excess has been taken away by using a roller. This procedure has been repeated for
each layer. Specimens have been then wrapped in a cotton fabric and a thicker plastic tissue, in order to adsorb
exceeding resin; finally, they were watertight sealed with a nylon sheet and void was created by means of an
autoclave. Curing of specimens, at environmental temperature, was five days long.

F LVDT F
stiffening bar

A 4 strain gages: 45

stiffening bar

B
specimen

2F

Figure 3: Instrumentation and load transmission for Figure 4: Instrumentation and load transmission for
tension tests. shear tests.

3. EXPERIMENTAL SET-UP FOR LONG -TERM TESTS

Experimental setups for both kinds of long-term tests (tension and shear) are based on the leverage system of Figure
2, giving a constant load with time. Load is applied by means of a number of steel weights suspended from the
longer lever arm and is transmitted to specimen chains attached at the shorter lever arm. Amplification factor equal
to 8.12 was determined with a load cell before testing. Each frame system has two specimen chains, carrying 50
percent of applied load each, through another leverage system.

216
Specimens used for tension test have their extremities glued to two steel plates, which are then bolted to other steel
plates connecting different specimens (Figure 3); each specimen belonging to one of two chains of the frame is
subject to 8893 N axial loading. Specimens for shear test have steel plates glued to their extremities and to their
middle portion, where other steel elements are bolted, so that both panels of specimens are subject to uniform shear
stresses. Each chain has two specimens, which are subject to 9393 N shear force (Figure 4). The adopted mechanical
system is similar to that recommended in standard ASTM D4255 for shear tests on composites.
In tension tests, deformations are measured by means of two longitudinal strain gauges placed on two faces of
specimens (Figure 3). For shear-loaded specimens (Figure 4), two strain gauges are placed on both sides with ±45°
orientation with respect to longitudinal direction. From strain gauge measures, assuming a state of pure shear in
panels A and B, shear strain is |γ|=2|ε|, where ε is the average of experimental strains from a couple of strain gauges.

4. EXPERIMENTAL RESULTS
Figures 5-6 show experimental results obtained from tension test on specimens obtained from pultruded plates; each
series of data in Figure 5 represents the mean behavior obtained from two identical specimens during the loading
phase. In order to compare results of unstiffened and stiffened (with CFRP sheets) specimens, load per unit width is
reported (instead of tensile stress) versus axial strain. Linear regression analysis of experimental data gives, as
expected, an high degree of correlation in all cases (coefficient of least-square procedure R2≅0.996). Slope of each
line then gives the equivalent rigidity (E⋅s)eq for unstiffened and stiffened specimens, being E the longitudinal Young
modulus and s the specimen thickness. As expected, strengthening by using carbon sheets reduces remarkably axial
deformability. With reference to specimens with small width, stiffness increase due to CFRP sheets is about 2.5
times.
With reference to long-term tests, ratio between longitudinal creep strain and elastic strain (creep coefficient) is
reported in Figure 6 (mean value from two identical specimens). Applied long-term load corresponds to 25 percent
and 8 percent of composite strength, respectively, for small and large unstiffened specimens. Unstiffened specimens
show approximately the same creep behavior (difference less than 5 percent): creep strain is more than 25 percent of
elastic strains after 100 days of loading. Moreover, slope of curves at that time is still appreciable and a significant
growth of strains with time can be expected. Stiffening of pultruded elements by means of carbon fiber sheets
provides for a strong reduction of delayed deformation (more than 50 percent for small specimens - see Figure 6).

small specimens with CFRP small specimens without CFRP large specimens without CFRP Small specimens with CFRP Small specimens without CFRP Large specimens without CFRP
350,0 0,3

300,0 y = 0,2276x
0,25
R2 = 0,9952
load per unit width (N/mm)

250,0
0,2

200,0
εcr/εel

y = 0,0905x 0,15
y = 0,0871x
150,0 R2 = 0,9957 2
R = 0,9957
0,1
100,0

0,05
50,0

0,0 0
0 10 20 30 40 50 60 70 80 90 100
0 500 1000 1500 2000 2500
6
strains ×10 time (days)

Figure 5: Tension tests – strain measures during Figure 6: Tension test – creep coefficient of specimens
loading phase of specimens extracted from plates. extracted from plates.
large specimens with CFRP Large specimens with CFRP
250,0 0,3

0,25
200,0
load per unit width (N/mm)

y = 0,4264x
2
R = 0,9947 0,2
150,0
εcr/εel

0,15

100,0
0,1

50,0 0,05

0
0,0
0 10 20 30 40 50 60 70 80 90 100
0 500 1000 1500 2000 2500
6
strains ×10 time (days)
Figure 7: Tension tests – strain measures during Figure 8: Tension test – creep coefficient of specimens
loading phase of specimens extracted from beam extracted from beam flanges.
flanges.

217
Moreover, slope of curve after 100 days is small, suggesting a smaller strain increase with time with respect to
unstiffened specimens.
Figures 7-8 show instantaneous and long-term behavior of specimens obtained from beam flanges and stiffened with
CFRP; equivalent stiffness is much higher than that of specimens obtained from plates; analogously, creep
coefficient is less than half of that provided by those specimens, probably due to higher fiber content. Moreover,
slope of curve after 100 days is almost horizontal.
Finally, figures 9-10 show experimental results obtained from specimens extracted from beam webs under long-term
shear tests. In Figure 9, shear force acting on each composite panel (half of force on single chain) is reported vs.
shear strains γ; experimental data obtained during the loading phase have been linearly interpolated. Each
experimental data is the average value of four strain gauge measures obtained from a specimen. Data scattering is
very limited. Results for unstiffened and stiffened specimens are compared. Slope of curves represents the
equivalent stiffness (GA’)eq, with G shear modulus and A’ cross-section area (including shear factor). Shear modulus
obtained from unreinforced specimens is about 3.18 MPa, close to value 3.4 MPa provided by the manufacturer.
Strengthening by CFRP sheets provided for a 37% increase of equivalent shear modulus. Long-term evolution of
creep coefficient under shear (Figure 10) shows an increase with time similar to the case of longitudinal creep strain,
for both unstiffened and stiffened specimens. CFRP sheets are very effective also in reducing long-term shear strain
(about 50%). After 100 days of loading, shear strains still show a remarkable rate of increase, whereas in the
presence of CFRP sheets the slope of curves is significantly reduced.

5. CONCLUSIONS
Creep experimental tests have been performed on pultruded specimens subject to axial and shear forces. Specimens
strengthened with carbon fibers sheets were also tested. A significant reduction in longitudinal and shear strains in
strengthened specimens was observed, for both instantaneous and long-term behaviour. Creep coefficients in tension
were about 0.25-0.3 after three months for unstiffened specimens, and reduced by one half in the presence of CFRP;
long term shear deformability was higher with respect to axial one.

ACKNOWLEDGEMENTS
The financial support of C.N.R. - National Council of Research, PAAS Grant 2001, is gratefully acknowledged. The
authors would like to thank ECT of Ancona (Italy) for providing specimens for testing.

REFERENCES
Berardi V., Feo L., Giordano A., Ascione L. (2006), On the Creep Behaviour of CFRP Pultruded Laminates: an
Experimental Study, Proc. 2nd Int. fib Congress, June 5-8, 2006 – Naples, Italy (on CD), 1-9.
ASTM D 4255/D 4255M (2001), Standard Test Method for In-Plane Shear Properties of Polymer Matrix Composite
Materials by the Rail Shear Method.
Mottram J. T. (1993), Short-and long-term structural properties of pultruded beam assemblies fabricated using
adhesive bonding, Composite Structures, 25, 387-395.
McClure G., Mohammadi Y. (1995), Compression Creep of Pultruded E-Glass-Reinforced-Plastic Angles, Journal
of Materials in Civil Engineering, 7(4), 269-276.
Shao Y., Shanmugam J. (2004), Deflection Creep of Pultruded Composite Sheet Piling, Journal of Composites for
Construction ASCE, 8(5), 471-479.
Spec. C without CFRP Spec. D without CFRP Spec. A with CFRP Spec. B with CFRP Spec. A with CFRP Spec. B with CFRP Spec. C without CFRP Spec. D without CFRP

5,00 0,45

4,50 y = 0,0084x y = 0,0037x 0,4

y = 0,0078x R2 = 0,9987
R2 = 0,9991 y = 0,0037x
4,00 R2 = 0,9992 R2 = 0,9989 0,35

3,50
0,3
3,00
shear (kN)

γvisc/γel

0,25
2,50
0,2
2,00
0,15
1,50
0,1
1,00

0,50 0,05

0,00 0
0 200 400 600 800 1000 1200 1400 0 10 20 30 40 50 60 70 80 90 100
strains ×106 time (days)

Figure 9: Shear test – strain measures during loading Figure 10: Shear test – creep coefficient of specimens
phase of specimens extracted from beam web. extracted from beam web.

218
Part VII. Design Guides
Third International Conference on FRP Composites in Civil Engineering (CICE 2006)
December 13-15 2006, Miami, Florida, USA

A DESIGN PROCEDURE OF FRP CONFINING SYSTEMS FOR

UPGRADE R/C COLUMNS
Angela Di Nardo
(Graduate Research Assistant, University of Salerno, Fisciano, Italy)

Ciro Faella
(Full Professor, University of Salerno, Fisciano, Italy)

Roberto Realfonzo
(Associate Professor, University of Salerno, Fisciano, Italy)

ABSTRACT
This paper presents a procedure that allows computing the axial load-bending moment (N-M) interaction curves in
case of rectangular R/C columns confined by FRP. At this purpose the constitutive law for the FRP confined
concrete in compression adopted by the recently edited guidelines of the Italian National Research Council (CNR,
2005) has been used. This law presents two different branches; the first one parabolic and the second one linear and
degenerates in the well-known “parabola-rectangle diagram” (CEN, 2001) when the confinement is absent. The
obtained interaction curves are presented in a dimensionless representation (ν−µ), which is useful in strengthening
interventions design, since it allows representing in a very simple way the dependence on several parameters.

KEYWORDS
R/C columns, FRP, confinement, interaction curves, code provisions.

1. INTRODUCTION
The action of a FRP confining system essentially produces an increment of the strength and of the ultimate strain of
compressed concrete; consequently, in case of R/C columns, it provides both an enlargement of the axial force-
bending moment (N-M) interaction curve and, mainly, an improvement of the available ductility.
Recently new guidelines have been edited by the Italian National Research Council (CNR-DT200). Such provisions
are also aimed at the evaluation of the ultimate curvature and flexural capacity of FRP confined R/C columns: at this
purpose the bilinear constitutive relationship proposed in (Faella and Realfonzo, 2002) is introduced in DT200 in
order to model the confined concrete under compression (see Figure 1). Assuming ε = ε c / ε c0 this law is given by:

fc fc
if ε <1: = (1 + γ ) ⋅ ε − ε 2 (parabolic branch); if ε >1: = 1 + ( γ − 1) ⋅ ε (linear branch) (1)
f cd f cd
where:
f cd + E t ⋅ ε c0 f c1 f ccd − f cd
γ= = Et = (2)
f cd f cd ε ccu

In Eq.s (1) and (2), fcd and εc0 are respectively the design strength and the corresponding strain (typically assumed
equal to 0.2%) of the unconfined concrete, while fccd is the design confined concrete compression strength and εccu
its ultimate strain. The Italian guidelines provide for these two parameters the following relations:

2/3 0.5
f ccd f  f 
= 1 + 2.6 ⋅  leff 
 = 1 + 2.6 ⋅ f l2, d/ 3 ε ccu = 0.0035 + 0.015 ⋅  leff 
 = 0.0035 + 0.015 ⋅ f l, d (3)
f cd  f cd   f cd 

219
 fccd

fc1 Et C
C
fcd

εc0 3.5‰ εccu

Figure 1: Constitutive law, ultimate regions and stress distribution

where the dimensionless parameter fl,d is given by the ratio of the effective lateral stress produced by the FRP system
(fl,eff) to the design compression strength of unconfined concrete (according to CNR DT200/2004 the confinement is
effective if fl,d>0.05). It is to be observed that for unconfined concrete (i.e. when fl,d =0) the constitutive relationship
degenerates in the well known “parabola-rectangle” law (CEN, 2001).
According to the DT200, the effective confinement pressure is given by:

1
f l,eff = ⋅ k eff ⋅ ρ f ⋅ E f ⋅ ε fd , rid (4)
2

where: keff is a coefficient (≤1) defined as the ratio of the volume of effectively confined concrete to the volume of
the concrete member; ρf is the geometric ratio which is a function of section shape and FRP configuration
(continuous or discontinuous wrapping); Ef is the FRP Young’s modulus in fibers direction, and εfd,rid is the FRP
conventional ultimate strain, corresponding to an unacceptable degradation of concrete (which is much lower than
the FRP characteristic failure strain under tension εfk).

2. A PROCEDURE FOR EVALUATING ν−µ INTERACTION CURVES

According to the symbols reported in Figure 1, the equilibrium equations which allow computing the ultimate axial
force and the bending moment for R/C square or rectangular cross sections are:
y2 y2
h  h  h 
N u = b ⋅ ∫ σ( y)dy +AS' σS' + ASσS ; M u = b ⋅ ∫ σ( y) ⋅  − yc + y dy + As' σ's  − d '  − Asσs  − d '  (5)
y1 y1  2   2   2 

where yc is the neutral axis depth and steel stresses are assumed positive when the reinforcement is under
compression. Equations (5) can be rewritten in a dimensionless form, as follows:

Nu σ ' σ Mu σs ' σ
ν= = ξψ + ω' s + ω s µ= = ξψ(0.5 − λξ ) + ω' (0.5 − δ') − ω s (0.5 − δ') (6)
b ⋅ h ⋅ f cd f ys f ys 2
b ⋅ h ⋅ f cd f ys f ys

being: δ’(=d/h) the dimensionless cover; ξ(=yc/h) the non-dimensional neutral axis depth; fys the steel yielding
strength; ω = ( A s ⋅ f ys ) /( b ⋅ h ⋅ f cd ) and ω' = ( A' s ⋅f ys ) /( b ⋅ h ⋅ f cd ) the steel reinforcements mechanical percentages;

y2 y2
∫ σ( y)dy ∫ σ( y) ⋅ ydy
y1 y1
ψ= and λ = 1− (7)
y c ⋅ f cd y c2 ⋅ ψ ⋅ f cd

respectively, the ratio of the concrete stress nonlinear diagram area to the rectangular area obtained for a constant
concrete stress distribution equal to fcd, and, the ratio to the distance from the extreme compressed concrete fiber of
the compressive concrete resultant “C” (=ξψ) to the depth of the neutral axis (see Figure 1).
In figure 1 the six ultimate regions for a R/C section are also shown; the ultimate state limit is attained when:
regions 1 and 2 - the steel strain reaches a limiting value in tension (according to the “Eurocode 2”, εsu=0.001);
regions 3 to 6 - the concrete compressive strain is equal to its ultimate value (εccu given by Eq.3).

220
For each of the six regions, the close form solutions of ψ and λ expressions are shown in (Faella et al., 2004), where
the Authors assumed the constitutive law reported in Eq. 1.
Figure 2 shows functions ψ(ξ) and λ(ξ) obtained by integrating Eq.s 7 for values of fl,d ranging from 0 to 60%; this
last value is rarely exceeded in practice. It is possible to observe that in regions 3 to 5 (i.e. for ξ23≤ ξ ≤ 1) ψ and λ
assume constant values (i.e.ψ and λ).
The value ξ23 - which corresponds to a threshold between regions 2 and 3 - can be evaluated as follows:

ε ccu
ξ 23 = (1 − δ' ) ⋅ (8)
0.01 + ε ccu

Therefore ξ23 also depends on the non-dimensional pressure fl,d.

For a generic value of the dimensionless pressure fl,d, the simplified trend of the functions ψ(ξ) and λ(ξ) is presented
in Figure 3a. The analytical expressions of the new relationships are reported in Figure 3b; such relationships depend
on the following three coefficients:

ψ = 1.16 ⋅ f l0,d.5 + 0.3 ⋅ f l,d + 0.8 λ = 0.437 − 0.05 ⋅ f l,d β = 0.4 ⋅ f l,d + 1.25 (9)

3.50 0,6 0.55

ψ 0,5
fl,d λ
0,4
3.00 0,3
0,2 0.50
2.50 0,1
0,05
unconfined
2.00 0.45
region2 0,6
0,5
1.50
ξ2,3 0.40 0,4
0,3
fl,d

1.00 0,2
0.35 0,1
0.50 regions region 6 0,05
3 to 5 unconfined
0.00 ξ 0.30 ξ
0.0 1.0 2.0 3.0 4.0 0.0 1.0 2.0 3.0 4.0

Figure 2: ψ and λ functions versus the neutral axis position

The value ψ - assumed by the function ψ in the regions 3 to 5 - was obtained by minimizing the scatters between a
polynomial equation (black curve in Figure 3c) - the coefficients of which were obtained imposing ψ=0.8 for fl,d=0
(unconfined concrete) - and the curve obtained from the integration of the Equation 7 (red curve in figure 3c).
The value of λ - assumed by coefficient λ in regions 3 to 5 - instead, has shown a limited variability within the
investigated range of the confinement pressure (red curve in Figure 3d).
λ,ψ
βψ
Region ψ λ
ψ
ψ
ξ
Region 2 (ξ ≤ ξ23) ψ=ψ λ=λ
ξ23
λlim
λ
λ Regions 3,4,5 (ξ23≤ ξ ≤ 1) ψ=ψ λ=λ

Region 6 (ξ >1) Ψ=
βξ − β + 0.25
⋅Ψ λ = 0.5 ⋅
(
ξ − 1− 0.5 ⋅ λ )
ξ2,3 ξ= 1 ξ ξ − 0.75 ξ − 0.75
a. b.
2.40
ψ 0.45
ψ = 1,16(fl,d)
0,5
+0,3fl,d+0,8 λ
2.00 λ = 0,437-0,05*fl,d
0.44

1.60 0.43

1.20 0.42
numerical integration
0.80 simplified model 0.41 numerical integration

fl,d simplified model

fl,d
0.40 0.40
0 0.1 0.2 0.3 0.4 0.5 0.6
c. 0 0.1 0.2 0.3 0.4 0.5 0.6
d.

Figure 3: Trend of ψ and λ coefficients – simplified relationships

221
As shown in the Figure 3d a linear law λ(fld) was assumed thus slightly overestimating the values derived from the
integration and, consequently, obtaining a conservative estimate in the ν-µ curves definition.

3. INTERACTION CURVES AND ULTIMATE CURVATURE

In Figure 4, the ν−µ interaction curves obtained from the equilibrium equations (6) and the relationships between ν
and the dimensionless ultimate curvature (φud) are shown for fl,d ranging between 0 and 60%.
For what concerning the ν−µ interaction curves, the ones reported with continuous lines were obtained by the exact
solution of integrals contained in Equations 7 (coefficients ψ and λ), while the curves indicated with dotted lines
were provided by applying the simplified procedure synthesized in Figure 3; all the curves were obtained for a
dimensionless cover (δ’) equal to 0.05 and for two symmetrical reinforcement configurations, i.e. ω=ω’=0.1 and 0.3.
Observing the Figure it is evident that the simplified relationships derived for ψ and λ - although they are much
simpler - allow obtaining ν−µ curves very similar to the exact ones. The major differences between the two models
can be observed in the “region 6”, where the simplified model is more conservative.
The ultimate curvature – shown in a dimensionless form in the lower part of the Figure 4 – were obtained by
assuming as maximum steel tensile strain a value equal to 4%, according to the seismic Italian code (Ordinanza
3274, 2003) while the ultimate strain of concrete in compression was obtained by applying Equation 3. In the figure
the improvement in terms of deformation capacity due to the FRP confinement is evident (note that the curvatures
were amplified of a factor equal to 10). The design and/or the safety verifications of FRP confining interventions
with the use of the ν−µ and ν−(φud) diagrams shown in Figure 4 is immediate. Observing the ν−µ interaction curves
it is possible to evaluate, for each value of ν, a threshold value of the non-dimensional confinement pressure; further
increments of the confining system stiffness over this value do not produce improvements of strength. In the same
way from the ν−(φud) diagrams an other threshold of fl,d can be evaluated; further increments of the confinement
pressure over this value do not produce improvements of ductility.
Finally, it has to be underlined that the contribute of steel stirrups to the confinement was not considered herein.

0.6 0.6
0,6 0,6
µu 0,5 µu 0,5
0,4 0,4
0.4 0,3 fl,d 0.4 0,3 fl,d
0,2
0,2 0,1
0,1 0,05
0.2 0,05 0.2 Unconfined
Unconfined
0.0 νu 0.0 νu
-1.0 -0.5 0.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5 4.0 -1.0 -0.5 0.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5 4.0
0.2 0.2
ω = ω' = 0,1
concrete failure
d'/h = 0,05
0.4 0.4
ω = ω' = 0,3
10(φu d) steel
d'/h = 0,05
10(φu d)
failure
0.6 0.6

Figure 4: ν−µ interaction curves and trends of the nondimensional ultimate curvatures

4. REFERENCES
CEN, European Committee for Standardization (2001). “Eurocode 2: Design of concrete structures – Part 1: General rules
and rules for buildings”, European Standard, prEN 1992-1, Brussels, Belgium.
CNR, National Research Council (2004). “Guide for the design and construction of externally bonded FRP systems for
strengthening existing structures (Materials, RC and PC structures, masonry structures)”, Technical Document CNR-DT
200/2004, Advisory Committee on Technical Recommendations for Constructions, Rome, Italy.
Faella C., and Realfonzo R. (2002). “Legami costitutivi del calcestruzzo confinato con FRP”, Proceedings of the V Italian
Workshop on Composite Structures, Salerno, Italy, November 28-29 (in italian).
Faella, C., Realfonzo, R., and Salerno N. (2004). “N-M interaction curves of concrete elements confined with FRP systems”,
Proceedings of the International Conference of Restoration, Recycling and Rejuvenation Technology for Engineering and
Architecture Application, Cesena, Italy, June 7-11.
Ordinanza 3274 (2003), “Norme tecniche per il progetto, la valutazione e l’adeguamento sismico degli edifici”. Presidenza
del Consiglio dei Ministri, (Italian Seismic Code - in italian).

222
Third International Conference on FRP Composites in Civil Engineering (CICE 2006)
December 13-15 2006, Miami, Florida, USA

DESIGN AND MANUFACTURING OF LOW COST THERMOPLASTIC

COMPOSITE BRIDGE SUPERSTRUCTRE
Nasim Uddin
(Associate Professor, Department of Civil Engineering at University of Alabama, Birmingham, Alabama,USA)

Abdul Moeed Abro

(Graduate Student,Department of Civil Engineering at University of Alabama, Birmingham, Alabama,USA)

Uday Vaidya
(Associate Professor, Department of Materials Engineering at University of Alabama, Birmingham, Alabama,USA)

ABSTRACT
An integral modular fiber thermoplastic composite bridge structural system is described. The design concept is
presented by utilizing high performance thermoplastic material (i.e. Glass/Polypropylene) along with efficient low
cost manufacturing process and fabrication techniques. The design is based on detailed finite element analyses to
investigate the stiffness and strength of the structural system. To demonstrate the design concept, two bridge deck
systems with different spans are modeled. The design concept of both decks presents a unique approach for
structurally efficient and low cost bridge deck system.

KEYWORDS
Thermoplastic composite, glass/polypropylene, finite element analyses.

1. INTRODUCTION
The United States of America is facing a major challenge to keep the nation’s infrastructure systems in usable
condition as large volumes of bridges which were built in the 1960’s (the Interstate era) need maintenance, major
rehabilitation, or replacement in the near future. A solution to this challenge could be to implement new structural
systems using more efficient materials like Fiber Reinforced Composites (Thermoplastics). Thermoplastic
composites are relatively new materials in civil engineering applications and lack the history of use in civil
infrastructure. Civil applications have often used thermoset composites. Recent progress in low cost thermoplastic
materials and fabrication technologies offer new solutions for very lightweight, cost efficient composite structures
with enhanced damage resistance and sustainable designs.

Recently, a novel hot melt impregnation technology called DRIFT (Direct ReInforcement Fabrication Technology)
(Hartness et al. 2001) has been developed that allows complete impregnation of long fibers with thermoplastic
polymers, producing high quality, low cost products that can be made as continuous rods, tapes and pultruded
shapes (Hartness et al. 2001). E-glass/PP tapes of 12 mm width and an average layer thickness of 0.6 mm were
produced using the DRIFT process. The unidirectional E-glass/PP tape material with a fiber content of 67% by
weight (42% by volume) has the tensile strength 87.6 ksi, tensile modulus of 4300 ksi and density of 99 pounds per
cubic ft. The hot melt impregnated unidirectional E-glass/PP tape can be woven into a plain weave architecture
fabric form through textile weaving operation.

The deck shape (Fig. 1) based on hat-sine rib stiffened design concept is selected by considering various issues such
as the processablilty of the E-glass/PP woven tape, and the practical issues such as tooling, and design flexibility for
the prototype studies. The face and the rib portions of the deck floor can be processed through a number of choices,
which include thermoforming, double belt press consolidation of the tape forms, reaction injection molding and or
extrusion.

223
 Figure 1: Hat-sine reinforced deck shape

2. DESIGN CRITERIA AND DESIGN PROCEDURE

2.1. Applied Loads and Allowable Deflection.

The dead loads are; self weight of the deck (15 psf) and load of wearing surface (5 psf).
The live loads include the three specified types of vehicular loading:
i) Design truck load HS20-44: three axles with loads 32 kips, 32kips and 8 kips. The spacing between the 32 kips
axles varies from 14 ft to 30 ft, and is chosen by the designer to produce the maximum effect for shear, moment
and deflection.
ii) Design tandem: a pair of 25 kips axles spaced 4 ft apart with transverse spacing of 6ft.
iii) Design lane load: a uniformly distributed load of 640 lbs/ft applied over a 10 ft wide strip.

We use the AASHTO category strength I load combination to compute the ultimate capacity of the bridge i.e.
Q = 1.25DC + 1.75(LL + IM). (1)
We use the AASHTO service I loading combination for checking the deflection of the bridge design, i.e.
Q = LL + IM (2)

We use AASHTO specifications 3.6.1.3.2 and 2.5.2.6.2 to adopt the deflection limit of L/800 (where L is the span of
the bridge). The deflection resulting from the design truck/tandem alone or that resulted from 25 percent of the
design truck/tandem taken together with the design lane load should not be greater than the maximum allowed limit.
For maximum deflection, the truck or tandem is placed such that the center of gravity of the truck or tandem is on
the center of the bridge, i.e. AASHTO arrangement I. We use arrangement II (with the rear axle of the truck or
tandem at one end of the bridge) of the truck or tandem load to check for the critical shear stress.

2.2. Design Procedure

The design procedure is based on the finite element analysis which is carried out on Ansys 8.0 software, the
composite face and the hat-sine ribs are modeled using the Shell 99 elements. Each element is defined by orthotropic
material properties (as defined in Table 1) and ply orientations. The contact region between the face panel and the
hat sine stiffened ribs is developed by merging the common nodes and key points. The loading combinations as
defined in section 2.1 are applied and based on the combination of least deflection and corresponding stresses, the
deck component dimensions (amplitude and wavelength), shape and thickness are determined.

Table 1: Material properties of E-Glass / PP woven tape composite

Property E-Glass/PP woven tape composite 40 % fiber content by volume

E X , EY , E Z 1437 ksi , 1437 ksi, 149 ksi
ν XY ,ν YZ ,ν XZ 0.11, 0.22, 0.22
G XY , GYZ , G XZ 184.16 ksi, 108.75 ksi, 108.75 ksi
E FIBER , E MATRIX 10150 ksi, 149 ksi
G FIBER , G MATRIX 4350 ksi, 108.75 ksi

224
3. DESIGN CASE SUDY, VERFICATION AND COMPARISON.
3.1. Double Lane Bridge Deck System

For the system having 60 ft span bridge deck (Fig. 2), the hat-sine rib dimensions are determined to be optimal at 36
inch depth, 48 inch wavelength and 16 inch contact width with deck component thickness as shown by Fig. 3. The
maximum deflection (Fig. 4), ultimate flexural and shear stresses are 0.9 inch, 6366 psi, and 1710 psi respectively.

Deck Panel
60 ft

Traffic Direction

24 ft
Traffic Direction

Steel Girder

Figure 2: Plan of double lane bridge deck

Flat Face Thickness = 1.1 inch Contact Width = 16 inch

24 ft

Depth = 36 inch

Sine Rib Thickness = 0.9 inch Wavelength = 48 inch

Figure 3: Double Lane Bridge Deck Parameters for Case 1 using Glass / PP

Figure 4: Deflection (inches) for Case 1 of a double lane deck model

3.2. Design Verification

For design verification and accuracy we compare the results of an experiment in which an E-glass/pp panel (Fig. 5)
(material properties same as in Table 1) was tested under point loads (500 lbs to 2000 lbs).
Sine Rib Thickness = 0.24 inch Flat Face Thickness = 0.36 inch
43 inch

Wavelength = 5.375 inch Contact Width = 1.4 inch Depth =2.5 inch

Figure 5: Panel shape and dimensional parameters used in an experiment

By comparing the results (Table 2), the analysis was found to under predict the deflection by 10 to 15 percent. This
difference between the analysis and the model could be because of slight imperfectness in contact between the hat
sine rib and flat face; it can be avoided by using strengthened bond joining methods like ultrasonic bonding etc.

225
 Table 2: Experimental and Analytical Deflection Comparison

Concentrated Load Maximum Deflection (inches) Maximum Deflection (inches) Difference

(lbs) (Experimental) (Finite Element Analysis) (%)
500 0.026 0.023 10.76
1000 0.052 0.046 11.53
1500 0.07 0.06 14.28
2000 0.1 0.085 15

3.3. Comparison with other composite bridge system

We compare the performance of our double lane bridge deck system to the bridge system proposed by Aref and
Parsons (2000). A schematic of the cross-section of the bridge system proposed by Aref (2000) consists of seven
inner cells encased in an outer shell as shown in Fig. 6.
S-Glass/Epoxy

36 inch
25 ft

Figure 6. Cross-sections of the S-glass/Epoxy

The performance comparison of both systems is summarized in table 3 based on maximum deflection, failure
indices, interface shear stresses, and the self weight of the deck system.

Table 3: Performance comparison between S-glass/epoxy (Aref, 2000) and E-glass/polypropylene (proposed
design) deck system

Material S-glass/epoxy E-glass/polypropylene

Deflection (inch) 0.9 0.9
Tsai Hill Failure Index 0.24 0.28
Interface σ yz (psi), σ xz (psi) 504, 484 234, 175
Self weight of deck (lbs), DL:LL 67000, 0.46 121500, 0.84

4. CONCLUDING REMARKS AND ACKNOWLEDGEMENT

We have presented the structural system which possesses several special features that contribute to its effectiveness,
including the use of curved panels (sine ribs) which provide the nonplanar core configurations to increase the
performance of the bridge deck system. We compared our design to one published composite bridge concept
proposed by Aref (2000). Although our design has higher self weight which results in higher dead to live load ratio
than S-glass/epoxy deck system (Aref, 2000); but our design could result in better low cost deck section based on
the manufacturing and material cost comparison However detailed cost analysis directed towards the manufacturing
process and deck material is required before any definitive conclusions can be made regarding this issue.

We gratefully acknowledge the financial support of UTCA under the director Dr. Danial S. Turner.

5. REFERENCES
Aref A.J., Parsons I.D. (2000). “Design and performance of a modular fiber reinforced plastic bridge”. Journal of
Composite, Part B 31, pp 619-628.
AASHTO (2004). AASHTO LRFD Bridge Design Specification, 3rd edition.
Hartness T., Husman G., Koeing J. and Dyksterhouse J. (2001). “The characterization of low cost fiber reinforced
thermoplastic composites produced by the DRIFT process”. Journal of Composites, Part A 32, pp 1155-1160.
Vaidya U., Samalot F., Pillay S., Janowski G., Husman G. and Gleich K. (2004). “Design and manufacture of woven
reinforced glass/polypropylene composites for mass transit floor system”. Journal of Composite Materials, Vol. 38,
No. 21, pp 1949-1972.

226
Third International Conference on FRP Composites in Civil Engineering (CICE 2006)
December 13-15 2006, Miami, Florida, USA

ITALIAN DESIGN GUIDELINES FOR THE STRENGTHENING OF

EXISTING CIVIL CONSTRUCTIONS USING EXTERNALLY BONDED
FIBRE-REINFORCED POLYMERS
Luigi Ascione, Andrea Benedetti, Antonio Borri,
Angelo Di Tommaso, Luciano Feo, Roberto Frassine,
Gaetano Manfredi, Giorgio Monti, Antonio Nanni,
Maurizio Piazza, Carlo Poggi, Elio Sacco

ABSTRACT
A series of regulatory documents were issued by the National Research Council (CNR) of Italy on the use of FRP
for strengthening existing civil constructions. These documents, described in more details in the paper, set for the
first time in Italy some standards for production, design and application of FRP for reinforced concrete, masonry,
timber and metallic constructions. They are also conceived with an informative and educational spirit, which is
crucial for the dissemination, in the professional sphere, of the mechanical and technological knowledge needed for
an aware and competent use of such materials.
The documents are the result of a remarkable joint effort of almost all professors and researchers involved in this
emerging and promising field, from 15 Italian universities, of the technical managers of major production and
application companies, and of the representatives of public and private companies that use FRP for strengthening
artifacts. Thus, the resulting FRP codes naturally incorporate the experience and knowledge gained in ten years of
studies, researches and applications of FRP in Italy.

KEYWORDS
Design guidelines, Fiber Reinforced Polymers, Existing Civil Constructions, Seismic Strengthening.

1. INTRODUCTION
The peculiar situation of Italy with regards to the preservation of existing constructions, results from the
combination of two aspects: a) seismic hazard over the whole of the national territory, recently refined by a new
seismic zonation, with medium-high intensity over a large portion of it, the highest expected PGA being 0.35g for a
475 years return period, and b) extreme variety of the built environment, perhaps with no comparison in the entire
world.
Construction typology in Italy encompasses examples reckoned as Country’s (and world’s) historical, architectural
and cultural heritage – which include buildings of various function and importance, such as palaces, temples,
churches, cloisters, theatres, memorials, city walls, castles, simple dwellings, civil engineering works such as
bridges harbours and aqueducts – dating back to more than 2000 years ago, throughout the ancient- middle- modern-
and contemporary ages, down to those built in the 20th century.
The first studies on the use of FRP for the rehabilitation of existing civil constructions started in the early 90’s by
some pioneering groups that were striving to find new solutions to increase the safety of existing constructions, that
could compete with the more developed and usual ones of concrete jacketing, steel plating, base isolation, and
dissipative bracings.
Over the last ten years the interest has spread so widely and rapidly that FRP composites have now become one of
the most active and prolific research fields throughout the country.
As a testimony of the intense activity in Italy in the field of FRP, a series of regulatory documents were issued by
the National Research Council (CNR) of Italy on the use of FRP for strengthening structures: Instructions for
Design, Execution and Control of Strengthening Interventions by Means of Fibre-Reinforced Polymers: materials,
RC structures, prestressed RC structures, masonry structures (CNR-DT200/2004), timber structures (CNR-
DT201/2005) and metallic structures (CNR-DT202/2005).

227
These documents set for the first time in Italy some standards for production, design and application of FRP for the
rehabilitation of existing civil constructions. They are the result of a remarkable joint effort of almost all the
professors and researchers involved in this emerging and promising field, from 15 universities, to the technical
managers of major production and application companies, as well as the representatives of public and private
companies that use FRP for strengthening artefacts. Thus, the resulting FRP codes naturally incorporate the
experience and knowledge gained in ten years of studies, researches and applications of FRP in Italy.
The aim of the present paper is to present the aforementioned Italian technical codes in order to contribute to their
diffusion within the international scientific community.

2. CNR-DT 200/2004
The aim of the Instructions CNR-DT 200/2004 is to draw up a document that can be used for designing, execution
and controlling strengthening interventions on structures using fibre-reinforced composites, within the current
norms. The document deals with the following issues:
- Materials
- Basic FRP strengthening concepts and particular problems;
- Strengthening concrete and prestressed concrete structures;
- Strengthening masonry.
Within the area of strengthening reinforced concrete structures and prestressed reinforced concrete structures, as
well as masonry, specific indications are given relative to constructing in seismic areas according to the most recent
norms drawn up both nationally and internationally.
It starts with a description of both the positive and negative aspects of polymeric materials from a didactic
perspective as well as including an Appendix (A) which presents several of the mechanical properties, considered to
be fundamental in order to fully understand how to use such materials in the structural field. The various
peculiarities of these composites are highlighted in comparison to traditionally used isotropic materials, with
particular attention being given to joints as well as to how to verify resistance.
The remaining issues are discussed following the usual style of technical documents published by the CNR, in
accordance with the approach set out in the Euro-codes. The proposals are distinguished into Principles and
Application Rules, with each one being identified by a number and the Principles being labelled with a (P).
The main peculiarities of the CNR-DT 200/2004 help in characterising this document as well as distinguishing it
from all the other similar ones published within the international context.
One of the first innovative aspects can be found in the chapter dedicated to the materials to which the technicians of
the main companies producing FRP strengthening systems actively contributed. It defines the format of the specific
technical product sheets that the very same producers should supply therefore allowing whoever uses them to have
the necessary information considered to be indispensable when comparing the various products available on the
market.
In both the chapter dealing with the Materials as well as that dedicated to the Basic Strengthening Design Concepts
and Particular Problems, the concept of “complete strengthening system” is opportunely introduced. This is another
particularity of the CNR-DT 200/2004, that many factors are in play from the moment of executing a FRP
strengthening to its success, ranging from the compatibility between each of the single products used to the specific
structural substrate to which they must be applied (concrete, masonry, etc.).
The latter consideration is based on the distinction introduced in the document between two types of applications:
• Type-A applications: strengthening system with certification of each component as well as the final product
to be applied to a given support;
• Type-B applications: strengthening systems certified for each component only.
For the first type of application, the producer can certify the aforementioned compatibility, with lower partial
coefficient values being attributed to the state limit method as well as less severe quality control tests being carried
out.
The CNR-DT 200/2004 also gives for the first time specific indications for the use of FRP in seismic areas, in
accordance with the most recent literature as well as the most up to date Italian Codes.
Techniques and regulations for executing FRP strengthening interventions are described in the CNR Instructions as
well as their subsequent monitoring and relative controls. This was considered a useful aid to designers, works
directors and anyone else in deciding when to use FRPs for structural plating.
Another aspect that characterises the CNR-DT 200/2004 is the fact that it organically deals with the main FRP
interventions to be carried out on masonry structures, including both simple and double curved.

228
The other four Appendices of the document include further discussion from a didactic perspective of issues
relatively conceptually advanced, ranging from delamination to the behaviour of concrete pillars, strengthened both
longitudinally as well as transversely.

3. CNR-DT 201/2005
This document contains the Instructions relative to consolidation interventions of timber structures using fibre-
reinforced composites.
The issues dealt with in this document are the following:
- Basic FRP strengthening concepts and particular problems;
- Strengthening of prestressed elements;
- Strengthening of rigid elements;
- Strengthening of thin elements;
- Delamination resistance;
- Fatigue Strengthening;
- Execution, Control and Maintenance.
There are also three Appendices, including several examples of FRP strengthening interventions on structures are
described (Appendix A), main bibliographic information (Appendix B) and the reference norms (Appendix C).
This is an informative type of document. It proposes the following objectives: to diffuse within the Technical-
Professional community the knowledge acquired on the use of fibre reinforced composite materials in static
consolidation of timber structures and to identify the interventions that are effectively appropriate and safe.
Timber is considered to be technically one of the most adapt construction materials with long term load bearing
qualities as well as a long applicative history, as highlighted by the elevated durability of the number of structures
designed and built.
Over recent years, the ever increasing need for increased resistance and rigidity of both timber and glued lamellate
timber structures, has led for number of experiments to be carried out on timber-FRP one in comparison to only
wooden ones.
Fibre-reinforced composites offer several evident advantages, being easily applicable and extremely versatile for
both restructuring existing structures and designing new ones. These characteristics, due to their attractiveness, have
favoured a rapid and widespread diffusion of in strengthening techniques of concrete structures as well as masonry
plating through the use of fibre reinforced laminas.
These techniques are now an integral part of the patrimony of many designers as well as proving to be effective
tools in rapidly resolving numerous problems.
Even though experiment have been carried out on timber-FRP structures for over the past 15 years, it is a well
known fact that the current state of art is not advanced as its concrete counterpart, with it giving a limited number of
answers to problems relating to few specific applications.
It is worth noting that up until now no international guide lines have been drawn up for the use of FRP in plating
timber structures.
The work carried out so far represents the first step towards a set of design Instructions being drawn up that in the
near future, once all the theoretical and experimental studies currently underway have been concluded, will lead to a
more complete and universal understanding of the subject area.
In this perspective, the document will be useful in identifying the problems that remain unresolved, allowing the
scientific community to concentrate upon them over the next few years.

4. CNR-DT 202/2005
This document contains Instructions relative to the consolidation interventions on metallic structures using fibre
reinforced composites.
The issues dealt with in this document are the following:
- Basic FRP strengthening concepts and particular problems;
- Strengthening of prestressed elements;
- Strengthening of rigid elements;
- Strengthening of thin elements;
- Delamination resistance;
- Fatigue Strengthening;
- Execution, Control and Maintenance.

229
There are two Appendices, one including several examples of FRP strengthening interventions on structures both
nationally and internationally, while the other deals with the main bibliographic information.
It is set out as a series of Principles and Application Rules, in accordance with the Euro-codes as previously stated in
CNR-DT 200/2004 with the aim of diffusing information within the Technical-Professional community the
knowledge acquired on the use of fibre reinforced composite materials in static consolidation of metallic structures
as well as identifying the interventions that are effectively appropriate and safe.
It is worth noting that up until now the only international guide lines on the use of FRP plating for metallic
structures have been drawn up by an English research agency (CIRIA, 2004) & (ICE, 2001).
The Italian historical architectural patrimony is rich of significant examples of metallic workmanship. They had an
important role in the growth of the industrial civilisation as well as giving an impetus to the development of
structural theory and the study of material resistance.
As for all types of structures, many metallic structures need to have some sort of restructuring interventions due to
design faults, degradation of load bearing elements and variation in use.
Current metallic structure strengthening techniques include either bolting or welding steel plates to the original
structure but these can have some negative aspects. The steel plates introduce further weight that must be sustained
as well as also being susceptible to corrosion and stress phenomena. Welding can bring about many problems and
can sometimes be impractical as in the case of cast iron or steel structures.
Many of the problems associated to traditional techniques can be overcome by using fibre reinforced composites.
FRPs have an elevated weight/resistance ratio, higher than steel, are more resistant to corrosion phenomena, if not
immune, and are extremely easy to handle.
In the current version of this document reference is made to the CNR-DT 200/2004, in particular to the basic FRP
strengthening concepts for prestressed and rigid elements as well as the problems associated to them.
Several aspects regarding delamination and FRP strengthening of metallic framework fatigue damaged are also
discussed.
Also this document is useful in identifying the problems that are still unanswered, allowing the scientific community
to concentrate upon them over the next few years, with particular attention being given to delamination, the
structural behaviour of thin elements with particular reference to eulerian stability, thermal effects (particularly
relevant to metal bridges) and prolonged exposure to UV radiation.

4. CONCLUSIONS
The peculiarity of Italy, highly seismic and endowed with a built environment unique in the world, extremely
various and rich of cultural value, renders all research in this field a continuous and challenging task.
This nationwide effort has resulted in a series of regulatory documents, that were conceived both for regulating a
rapidly growing professional and technical market, as well as for an informative and educational purpose. The
documents are deemed of great importance for the dissemination, in the professional sphere, of the physical and
technological knowledge necessary to conscious and competent use of FRP in strengthening.

5. REFERENCES
CNR-DT 200/2004: ‘Instructions for design, execution and control of strengthening interventions by means of fibre-
reinforced composites: materials, RC structures, prestressed RC structure, masonry’ (2004).
CNR-DT 201/2005: ‘Instructions relative to consolidation interventions of timber structures using fibre-reinforced
composites’ (2005).
CNR-DT 202/2005: Instructions relative to consolidation interventions of metallic structures using fibre-reinforced
composites’ (2005).

230
Third International Conference on FRP Composites in Civil Engineering (CICE 2006)
December 13-15 2006, Miami, Florida, USA

LARGE-SIZE REINFORCED CONCRETE COLUMNS STRENGTHENED

WITH CARBON FRP: VALIDATION OF EXISTING DESIGN
GUIDELINES
Silvia Rocca
(Ph D. Candidate in Civil Engineering, University of Missouri - Rolla, Rolla, Missouri, USA)

Nestore Galati
(Research Engineer, University of Missouri - Rolla, Rolla, Missouri, USA)

Antonio Nanni
(Chair & Professor of Civil Engineering, University of Miami – Coral Gables, Coral Gables, Florida, USA)

ABSTRACT
Current international design guidelines provide predictive design equations for the strengthening of Reinforced
Concrete (RC) columns of both circular and prismatic cross-sections by means of FRP confinement and subject to
pure axial loading. Extensive studies (experimental and analytical) have been conducted for the case of columns
with cross-sections of the circular type, and limited studies have been conducted on the prismatic ones. In fact, the
majority of this research has been on small-scale, plain concrete specimens. In this review paper, four international
design guidelines are referenced, and a comparative study is presented that is based on the increment of concrete
compressive strength and ductility and that, includes the experimental results from six RC columns of different
cross-section shapes. The observed outcomes are used to identify and remark upon the limitations beyond the ones
specifically stated by each of the guides and to reflect the absence of effects not considered in current models. The
purpose of this study is to present a constructive critical comparison of the state-of-the-art design methodologies
available for the case of FRP-confined concrete RC columns and to indicate a direction for future developments.

KEYWORDS
Confinement, Ductility, FRP-Strengthening, Prismatic Columns, Reinforced Concrete.

1. INTRODUCTION
The confinement of Reinforced Concrete (RC) columns by means of Fiber Reinforced Polymers (FRP) jackets is a
technique being used with growing frequency to seek the increment of load carrying capacity and/or ductility of
such compression members. The need for improved strength is the result of higher load capacity demands because
of change in the use of the structure or because of more stringent code requirements. Improving ductility stems
from the need for energy dissipation, which allows the plastic behavior of the element and, ultimately, of the
structure. Ductility enhancement is typically required in existing columns that are subjected to a combination of
axial load and bending moment because of a change in code (e.g., to account for seismic provisions) or a correction
for design or construction errors (e.g., improper splicing of the longitudinal reinforcement or lack of transverse ties).
Extensive work in both the experimental and analytical areas has been conducted on RC columns of circular cross-
sections confined with FRP and subjected to pure axial compressive loading. This work has led to the development
of several models (the majority being empirical) of two types: design-oriented, where equations are provided to
determine the axial compressive strength and the ultimate axial strain (Saaman et al. 1998, Toutanji 1999, Xiao and
Wu 2000, among others); and analysis-oriented, which allows the construction of the stress-strain response of the
RC column (Spoelstra and Monti 1999, Lam and Teng 2003, among others).
Studies focused on RC columns of non-circular cross-sections have also been conducted; however, such work is
limited because the experimental research has primarily been on small specimens of plain concrete due to high cost
and lack of high-capacity testing equipment. This situation has been the main reason for overlooking the following

231
important effects on the element performance that are not accounted for in most of the available models: (a) the size
of the cross-sectional area, (b) the dimensional aspect ratio of the cross-sectional area, (c) the presence and possible
detrimental effect of longitudinal steel reinforcement instability, (d) the concrete dilation dependant on a pseudo-
Poisson ratio, and (e) the contribution of the internal transverse steel reinforcement. In spite of these obstacles,
several models have been proposed (Lam and Teng 2003, Maalej et al. 2003, among others) and have become the
basis for design provisions. In particular, the predictive equations found in the current design guides are mostly
based on approaches created for members of circular cross-section and then modified by a “shape factor” or
“efficiency factor.” This factor is intended to account for the geometry of the section and its effect on the confining
pressure, which is no longer uniformly applied by the FRP jacket as for the case of circular cross-sections.

2. RESPONSE TO AXIAL LOAD

The purpose of this study is to use pertinent experimental evidence to identify and remark on the differences in the
design methodologies used by the existing available design guides on the FRP confinement of RC columns of
different cross-sections and subject to pure axial loading. For the purpose of this paper and for the interpretation of
experimental results, clear and unequivocal definitions of strength and ductility parameters are necessary:
• f’co and f’cc represent the peak concrete strengths corresponding to the maximum load carried by the RC column
for unconfined and confined cases, respectively.
• εcu is the ultimate strain of the unconfined RC column corresponding to 0.85f’co (Figure 1(a)). For the confined
RC column, εccu may correspond to one of the following values: a) 0.85f’cc in the case of a lightly confined member
(Figure 1(b)); b) the failure strain in the heavily confined, softening case when the failure stress is larger than
0.85f’cc (Figure 1(c)); or heavily confined, hardening case, where ultimate strength corresponds to ultimate strain
(Figure 1(d)).
The definition of εccu at 0.85f’cc (or less) is arbitrary, although consistent with modeling of conventional concrete
(Hognestad 1951), and such that the descending branch of the stress-strain curve at that level of stress (0.85f’cc or
higher) is not as sensitive to the test procedure in terms of speed of loading and stiffness of the equipment utilized.

σ
f’cc (εccu,f’cc)
f’cc
f’cc (εccu,σ > 0.85f’cc)
f’co (εccu,0.85f’cc)
0.85f’cc
0.85f’co (εccu,0.85f’co)
(a) unconfined
(b) lightly confined
(c) heavily confined(softening)
(d) heavily confined(hardening)
Failure
ε

Figure 1: Schematic Stress-Strain Behavior of Unconfined and Confined RC Columns

3. COMPARATIVE STUDY
The documents considered in this comparative study are as follows: “Guide for the Design and Construction of
Externally Bonded FRP Systems for Strengthening Concrete Structures” reported by the American Concrete
Institute (ACI Committee 440.2R-02 2002), “Design and Construction of Building Components with Fibre-
Reinforced Polymers” reported by the Canadian Standard Association (CSA S806-02 2002), “Design Guidance for
Strengthening Concrete Structures Using Fibre Composite Material” Technical Report 55 by the Concrete Society
(TR 55 2004), and “Externally Bonded FRP Reinforcement for RC Structures” Technical Report by the fédération
internationale du béton (fib Bulletin 14 2001). The latter document provides two sets of equations: “exact” and
“approximated.”
To evaluate the performance and contrast the different approaches taken by the guidelines for the determination of
the compressive strength (f’cc) and the ultimate axial compressive strain for confined concrete (εccu), a total of six RC

232
column specimens (three strengthened specimens with their corresponding control units) of different cross-section
shapes (circular, square, and rectangular) and equal gross areas (Ag) were selected, designed, constructed, and tested.
These specimens were part of a recently-conducted research study on the size-effect of FRP-confined RC columns
(Rocca et al., 2006) and in the companion paper are identified as specimens A1 and A2 (circular), C1 and C3
(square), and B1 and B3 (rectangular). This assessment is not intended to be comprehensive, but the three relevant
cases presented here indicate the trends of the guidelines under study.
Figure 2(a) shows the accuracy of the different codes with respect to the experimental results in terms of strength
enhancement ([(f’cc/f’co)theo]/[(f’cc/f’co)exp]). For the case of circular cross-sections, only the Concrete Society and the
“exact” equations by fib slightly overestimate the strength enhancement (by approximately three percent). Regarding
the prismatic cross-sections, only ACI and the Concrete Society overestimate the strength increase for both square
and rectangular sections. The “exact” formulas by fib overestimate the strength enhancement for only the square
type of cross-section.

1.2 Circular 3.0 Circular

Square Square
Rectangular Rectangular
2.5
1.1
ρf ≈ 0.3%
′ ρf ≈ 0.3% 2.0
⎛ f cc ⎞ 1.0 ⎛ ε ccu ⎞
⎜ f′ ⎟ ⎜
⎝ co ⎠ Theo ⎝ ε cu ⎟⎠Theo 1.5
⎛ f cc′ ⎞ ⎛ ε ccu ⎞
⎜ f′ ⎟ 0.9
⎜
⎝ co ⎠ Exp
⎝ ε cu ⎟⎠ Exp 1.0

0.8
0.5

0.7 0.0
ACI CSA Concrete fib_exact fib_practical ACI CSA Concrete fib_exact fib_practical
Society Society

Design Guidelines Design Guideline

(a) Ratio of Theoretical Concrete Compressive (b) Ratio of Theoretical Ultimate Axial Deformation
Strength Enhancement to Experimental Enhancement to Experimental

1.0 Circular
Square
0.9
Rectangular
0.8
0.7
0.6
PTheo 0.5
ρf ≈ 0.3%

PExp 0.4
0.3
0.2
0.1
0.0
ACI CSA Concrete fib_exact fib_practical
Society

Design Guidelines

(c) Ratio of Design Axial Load Capacity to Experimental

Figure 2: Guidelines Performance

Figure 2(b) shows the accuracy of the guidelines in predicting the ultimate axial strain enhancement εccu /εcu. A
theoretical value of εcu equal to 0.003 was used in the case of ACI, and a value of 0.0035 was used in the cases of
the Concrete Society and fib. CSA does not provide expressions for the calculation of εccu. The estimations vary
within a range of approximately ±50 percent of the experimental ratios, with the exception of the value
corresponding to the “exact” equations from fib for the case of square columns (about 250 percent). In general, the
scatter of the predictions was much larger than for the strength enhancement. This may partly be because of the
difficulty in accurately representing the effects of parameters such as size and type of aggregates; mix proportions;
water/cement ratio; and in the case of confined concrete the stiffness of the FRP jacket.

233
Figure 2(c) presents the theoretical to experimental ratios of load-carrying capacity of the FRP-strengthened RC
columns: Ptheo/Pexp. The theoretical or design values of axial resistance were computed considering the material
safety factors and/or the strength reduction factors as required by each guideline. All the predictions appear to be
conservative. The results mainly vary in a range from about 60 to 95 percent of the experimentally obtained load-
carrying capacity, with the exception of the ratios corresponding to CSA that show a minimum percentage of about
40, which can be considered too conservative.

4. CONCLUSSIONS
The limits of applicability of the equations provided by the design guidelines primarily deal with the dimensions of
the cross-sections, the side-aspect-ratio (h/b), and loading type (concentric). These limits are the result of the
reduced experimental evidence on the area of FRP-confinement of real-size RC columns, which at the same time has
not allowed the appropriate implementation of key effects in the current models. These effects have been identified
as follows: the instability of longitudinal steel reinforcement, the concrete dilation dependant on the pseudo-Poisson
ratio, the contribution of the internal transverse steel reinforcement to the confinement, and an appropriate reduction
factor to account for the premature failure of the FRP jacket.
Given the present knowledge, the research community should consider further experimental and analytical work to
confirm the basic assumptions and to provide substantial data information to feed and correctly calibrate numerical
and analytical models. Although a vast experimental campaign on real-size RC columns following the conventional
testing methodology is a choice, the current available sensing technology used in a few dimensionally-relevant
specimens represents an innovative alternative testing protocol that would help obtain accurate information and most
importantly, would allow the understanding of the physical phenomena. The measurements should be targeted to
the strain distribution along the perimeter of the FRP jacket, strain distribution of the longitudinal and transverse
steel reinforcement, lateral (outward) deformation of the longitudinal steel bars product of the concrete lateral
dilation (bar instability), concrete dilation, and crack propagation detection. A more meaningful interpretation of the
experimental data currently available in the literature would become possible once performance phenomena and
controlling parameters are fully understood.

5. REFERENCES
American Concrete Institute, ACI 440.2R-02. (2002). “Guide for the Design and Construction of Externally Bonded
FRP Systems for Strengthening of Concrete Structures,” American Concrete Institute, Farmington Hills, MI.
Canadian Standards Association, CSA-S806. (2002). “Design and Construction of Building Components with Fibre-
Reinforced Polymers,” Ontario, Canada.
fédération internationale du béton (fib). (2001). “Externally Bonded FRP Reinforcement for RC Structures,”
Bulletin 14, Technical Report, Lausanne, Switzerland.
Hognestad, E. (1951). “A Study of Combined Bending and Axial Load in Reinforced Concrete Members,” Bulletin
399, University of Illinois Engineering Experiment Station, Urbana, IL.
Lam, L., and Teng, J. (2003). “Design-oriented Stress-Strain Model for FRP-confined Concrete in Rectangular
Columns,” Journal of Reinforced Plastics and Composites, Vol. 22, No. 13, pp. 1149-1186.
Maalej, M., Tanwongsval, S., and Paramasivam, P. (2003). “Modeling of Rectangular RC Columns Strengthened
with FRP,” Cement & Concrete Composites, Vol. 25, pp. 263-276.
Rocca, S., Galati, N., and Nanni, A. (2006). “Experimental Evaluation of FRP Strengthening of Large-Size
Reinforced Concrete Columns,” CIES Report No. 06-63, University of Missouri – Rolla, Rolla, MO.
Samaan, M., Mirmiram, A., and Shahawy, M. (1998). ‘‘Model of Concrete Confined by Fiber Composites,’’ ASCE
Journal for Structural Engineering, Vol. 124, No. 9, pp. 1025–1031.
Spoelstra, M. R., and Monti, G. (1999). “FRP-Confined Concrete Model,” ASCE Journal of Composites for
Construction, Vol. 3, No. 3, pp. 143-150.
The Concrete Society. (2004). “Design Guidance for Strengthening Concrete Structures Using Fibre Composite
Material,” Technical Report 55, Crowthorne, United Kingdom.
Toutanji, H. (1999). ‘‘Stress-Strain Characteristics of Concrete Columns Externally Confined with Advanced Fiber
Composite Sheets,’’ ACI Materials Journal, Vol. 96, No. 3, pp. 397–404.
Xiao, Y., and Wu, H. (2000). ‘‘Compressive Behavior of Concrete Confined by Carbon Fiber Composite Jackets,’’
ASCE Journal of Materials in Civil Engineering, Vol. 12, No. 2, pp. 139–146.

234
Third International Conference on FRP Composites in Civil Engineering (CICE 2006)
December 13-15 2006, Miami, Florida, USA

TORSION DESIGN OF CFRP PLATED RC MEMBERS

Adrian K. Y. Hii & Riadh Al-Mahaidi
(Department of Civil Engineering, Monash University, Victoria 3800, Australia)

ABSTRACT
Since 2001, there have been several experimental investigations in strengthening reinforced concrete members in
torsion. However, available tools for torsion design of FRP plated reinforced concrete members are limited and
unproven. A database of previous experimental data available in literature was compiled and compared against fib
Bulletin 14. Modifications consistent with the space truss model were proposed to correct the poor accuracy in
predictions of CFRP contribution to strength. Subsequently, a design tool consistent with several national design
codes to analyze the full torsional capacity of CFRP plated reinforced concrete beams was validated against the
experimental database.

KEYWORDS
Aggregate interlock, CFRP, design methods, reinforced concrete, torsion.

1. INTRODUCTION
A literature survey conducted by the authors (Hii and Al-Mahaidi 2005) found no experimental data for FRP
strengthening in torsion reported prior to 2001. Since then, several investigations have been conducted. However,
most experiments were exploratory in nature, making the proposal of analytical methods difficult to validate against
a single set of tests. Previous researchers have proposed a simple limiting FRP strain value (Ghobarah et al. 2002);
(Salom et al. 2004); (Zhang et al. 2001). A more sophisticated design method by fib Bulletin 14 (2001) was taken by
Ameli et al. (2004) and Panchacharam and Belarbi (2002). However, errors were found in the analysis (Hii and Al-
Mahaidi 2005). This paper evaluates the fib Bulletin 14 method against a unified database of all available
experimental data. Limitations and inconsistencies were identified. Modifications based on current national design
codes were then proposed and compared against the same database to observe any improvements in accuracy.

2. EVALUATION OF fib BULLETIN 14 GUIDELINES

Based on the assumption of the validity of the truss mechanism, the following equations were provided by fib
Bulletin 14 to predict the FRP contribution to strength, Tn,frp:
Tn, frp = 2ε fd ,e E fu b f t f s −f 1 Ac (cot θ + cot α ) sin α (1)
ε fk ,e = kε f ,e ≤ ε max (2)
ε fd ,e = ε fk ,e / γ f (3)
Where Ac = gross area of concrete section; bf, sf, tf = width, center-to-center spacing and thickness of FRP strips
respectively; Efu = Young’s modulus for FRP; k = 0.8 is used to define the characteristic effective FRP strain, εfk,e;
α = angle between principal fiber orientation and longitudinal axis of member; εfd,e = design value of effective fiber
strain, εf,e; εmax = 0.005 (for activation of the aggregate interlock mechanism); γf = material safety factor for the
FRP (range 1.2-1.5 in fib Bulletin 14); and θ = angle of crack to longitudinal axis.

In the case of FRP strips of width bf and spacing sf, in Equation 1 tf is multiplied by bf/sf, in effect ‘smearing’ the
strips along the length of member. Although the guidelines require full wrapping around the member cross-section,
Panchacharam and Belarbi (2002) and Ameli et al. (2004) observed increases in torsional capacity for both anchored
and unanchored U-wraps that were less significant than the equivalent full wrap. Therefore, Equation 1 without the
factor 2 was used for specimens strengthened with U-wraps as it was assumed only one full couple exists between
fiber strips. Due to the similar cracking mechanisms and lack of available test data, fib Bulletin 14 recommended the

235
adaptation of Triantafillou and Antonopoulos’ (2000) shear model. The model assumes the FRP laminates develop
an effective strain in the principal material direction at ultimate limit state. Details of the effective strain
relationships are located in fib Bulletin 14. The fib Bulletin 14 guidelines do not explicitly specify the FRP
reinforcement ratio, ρf, for torsional strengthening. However, it is in the authors’ view that applying ρf from shear
strengthening is inconsistent. A member behaves like an equivalent hollow tube under torsion; thus for completely
wrapped specimens, ρf for torsion is taken to be equal to tf/tc, where tc is equal to 3Ac/(4pc). pc is defined as the
perimeter enclosing the gross area of concrete. For box-section beams, the smaller of the actual wall thickness or
3Ac/(4pc) is taken as tc.

Since then, several experimental investigations with externally-bonded CFRP and GFRP on this topic have been
carried out by Ameli et al. (2004), Ghobarah et al. (2002), Panchacharam and Belarbi (2002), Salom et al. (2004),
and Zhang et al. (2001) in addition to the authors’ investigations (Hii and Al-Mahaidi 2004; 2006). Only specimens
which failed from FRP debonding or rupture were considered. Specimens which failed from concrete crushing were
discarded as the full FRP contribution to strength could not be achieved. Also, beams strengthened with longitudinal
strips only were not included as no significant improvement in post-cracking strength was observed in experiments
by previous investigators. Closer examination of the effective strain models by Triantafillou and Antonopoulos
(2000) were derived from mainly CFRP tests and some AFRP tests (Hii and Al-Mahaidi 2005). Hence, GFRP tests
were not considered here since no effective strain model exists. This left thirty relevant CFRP strengthened
specimens in the final database. Among the strengthening layouts investigated included varying fiber orientations, as
well as full/U-wraps. Varying strip spacings and layers were also investigated. Further details of the database can be
found in (Hii and Al-Mahaidi (To be published)).

For comparison purposes, the effective strain, εf,e, was used throughout this paper. All load, material and safety
factors were taken to be unity in this paper. It can be seen from Figure 1(a) that the predicted torsional contribution
from CFRP by fib Bulletin 14 was generally unsafe by 52%. The crack angle θ was taken to be 45o as it was
generally not reported in literature. For the experiments carried out by the authors, a range of crack angles for each
specimen was measured (Hii and Al-Mahaidi 2005). The large range found highlighted the importance of the θ
parameter. However, the range of values was still generally unsafe. Other parameters should be considered as well
for further refinement of fib Bulletin 14 guidelines.

80 80
Predicted FRP contribution (kN.m)

Predicted FRP contribution (kN.m)

Hii and Al-Mahaidi

70 70
Others
60 60
50 50
40 40
(a) ε f,e
30 30 (b) ε f,e
Ave. Tf,exp/Tn,frp = 0.48
20 20 Ave. Tf,exp/Tn,frp = 1.07
Hii and Al-Mahaidi
10 Others 10
0 0
0 10 20 30 40 50 60 70 80 0 10 20 30 40 50 60 70 80
Exp. FRP contribution (kN.m) Exp. FRP contribution (kN.m)

Figure 1: CFRP contribution model verification with εf,e for (a) fib Bulletin 14; (b) modified to AS3600-2001

3. CFRP CONTRIBUTION TO TORSIONAL STRENGTH

In space truss theory, a solid reinforced concrete member is modeled by an equivalent thin-walled member having
the same external dimensions, which consists of concrete compression struts inclined at an angle θ, and transverse
and longitudinal reinforcement as tension chords that tie the member together. The shear flow, which is the
tangential component of these diagonal compressive stresses, provides the resisting torque. Tests by Hsu and Mo
(1985) showed that the shear flow location was independent of the steel bar location, provided the concrete cover
was within a certain range to prevent spalling. As expressed in Equation 1 by fib Bulletin 14 and previous proposals,
the location of the resulting shear flow from the FRP laminates was assumed to be on the concrete surface where the

236
FRP was externally-bonded. However, the shear flow location is always located at a certain distance below the
concrete surface. Hence, Equation 1 is modified to Equation 4 below:
Tn, frp = 2ε fd ,e E fu A f s −f 1 An (cot θ + cot α ) sin α (4)
Where An = area enclosed by the shear flow path.

For evaluation of the effectiveness in predicting the CFRP contribution only, Equation 4 was modified to be
consistent with the Australian Standard AS3600-2001 (SAI 2001). The area enclosed by the shear flow path, An, is
taken to be equivalent to the area of a polygon with vertices at the centre of the longitudinal bars at the corners of the
cross-section. It is immediately obvious in comparing Figure 1(b) with (a) that the improvement in conservativeness
was significant; with the average Tf,exp/Tn,frp values now 1.07 compared to 0.48. The standard deviation was 0.29. In
general, the modified method not only gave better predictions of CFRP contribution, but was now consistent with
the premise of the space truss model. The full capacity of reinforced concrete beams strengthened with CFRP, Tn,
can now be analyzed in the following section.

4. DESIGN PROPOSAL OF CFRP STRENGTHENED RC BEAMS IN TORSION

The full torsional strength of CFRP strengthened reinforced concrete beams, Tn, can be analyzed using the principle
of superposition from both the steel, Tn,s, and CFRP reinforcement, Tn,frp, as shown in Equation 5:
Tn = Tn,s + Tn. frp (5)

It has been shown through photogrammetry measurements by Hii and Al-Mahaidi (2006) that the aggregate
interlock between crack faces in concrete were found to increase proportionally with the amount of CFRP bonded.
However, the effective strain model by Triantafillou and Antonopoulos (2000) already implicitly accounts for this
beneficial effect due to the way it was derived empirically. To maintain consistency with existing codes this was not
considered any further. By expanding Equations 4 and 5 to AS3600-2001 form, the final expression is shown in
Equation 6. The crack angle θ in AS3600-2001 can either be chosen conservatively as 45o, or to vary linearly from
30o when Tn is equal to the cracking strength, Tuc, to 45o when Tn is equal to the maximum torsional capacity before
concrete crushing, Tu,max. For this analysis, Equation 7 (SAI 2001) was used to determine the design crack angle.
[
Tn = φ 2 An (cot θ + cot α ) sin α f sy. f Asw s t−1 + ε fd ,e E fu A f s −f 1 (6) ]
⎡ Tn − φTuc ⎤
θ = 30 o + 15 o ⎢ ⎥ (7)
⎣⎢ φTu ,max − φTuc ⎦⎥
Where Asw = area of single stirrup leg; fsy.f = yield strength of stirrups; st = center-to-center spacing of stirrups;
and φ = torsional capacity factor, taken as unity in this investigation.

It must be noted that Tn and θ are coupled in both Equations 6 and 7, which implies an iterative process. However, in
design Tn is replaced by applied torque required, so iteration is not required. The average Texp/Tn values for Figure
2(a) were generally conservative at 1.24, where Texp is the experimental torsional capacity of the member. The
corresponding standard deviation was 0.34.

120 120
Predicted torque capacity (kN.m)

Predicted torque capacity (kN.m)

Hii and Al-Mahaidi Hii and Al-Mahaidi

100 100 Others
Others
(b) ε f,e
80 80
Ave. Texp/Tn = 1.72
60 60

40 40
(a) ε f,e
20 20
Ave. Texp/T n = 1.24
0 0
0 20 40 60 80 100 12 0 0 20 40 60 80 100 120
Exp. torque capacity (kN.m) Exp. torque capacity (kN.m)

Figure 2: Verification of model for ultimate capacity with εf,e modified to (a) AS3600-2001; (b) ACI 318-05

237
The equation form when adapting the fib Bulletin 14 design method to ACI 318-05 (ACI Committee 318 2005) is
similar to Equation 6. However, in ACI 318-05 the area enclosed by the shear flow path, An, is taken to be
equivalent to 85% of the area enclosed by the centerline of the outermost closed hoops. For design simplicity, the
crack angle θ was chosen conservatively as 45o, although more rigorous analysis using the Softened Truss Model
algorithm developed by Hsu (1993) is allowed. The average Texp/Tn value for Figure 2(b) was 1.72. The
corresponding standard deviation was 0.49. In comparison to AS3600-2001 the predictions by ACI 318-05 were
much more conservative. This was mainly due to the fact that a design crack angle of 45o was used in the analysis,
which was a conservative assumption. Again, this highlights the importance of the crack angle parameter in design.
The empirical definitions of the area enclosed by the shear flow are similar for both design codes. However,
differences are more significant for this test series due to the larger specimen sizes, leading to smaller estimates of
An compared to AS3600-2001. The opposite is true for the smaller scale specimens, due to the concrete cover and
size of the steel bars being more significant in comparison to the overall dimensions.

5. CONCLUSIONS
The method for torsional strengthening with FRP in fib Bulletin 14 was compared against a compiled database of
available experimental data, and found to be unsafe by 52%. In fib Bulletin 14, the location of the resulting shear
flow from the FRP laminates was assumed to be on the concrete surface, which was shown to be inconsistent with
the premise of the space truss model. The predictions of FRP contribution were significantly more accurate when the
shear flow location was modified to be consistent with AS3600-2001. Through the principle of superposition, the
modified equation was then combined with two design codes to analyze the full torsional capacity of CFRP plated
reinforced concrete beams. The accuracy of the modified fib Bulletin 14 method was better when adapted to
AS3600-2001 compared to ACI 318-05, which was found to be overly conservative. This can largely be attributed to
AS3600-2001 taking into account variable crack angles in design. The combination of either design codes with the
modified fib Bulletin 14 effective strain model are suitable as a torsion design tool for CFRP plated reinforced
concrete members in the future.

6. REFERENCES
ACI Committee 318. (2005). "ACI 318-05 - Building Code Requirements for Structural Concrete." American
Concrete Institute, Michigan, USA.
Ameli, M., Ronagh, H. R., and Dux, P. F. "Experimental Investigations on FRP strengthening of beams in torsion."
FRP Composites in Civil Engineering - CICE 2004, Adelaide, Australia, 587-592.
fib Bulletin 14. (2001). "Externally bonded FRP reinforcement for RC structures." fib - International Federation for
Structural Concrete, Lausanne.
Ghobarah, A., Ghorbel, M. N., and Chidiac, S. E. (2002). "Upgrading Torsional Resistance of Reinforced Concrete
Beams Using Fiber-Reinforced Polymer." Journal of Composites for Construction, 6(4), 257-263.
Hii, A. K. Y., and Al-Mahaidi, R. "Torsional Strengthening of Reinforced Concrete Beams Using CFRP
Composites." FRP Composites in Civil Engineering - CICE 2004, Adelaide, Australia, 551-559.
Hii, A. K. Y., and Al-Mahaidi, R. "Torsional Strengthening of Solid and Box-section RC Beams Using CFRP
Composites." Composites in Construction 2005 - Third International Conference, Lyon, France, 59-68.
Hii, A. K. Y., and Al-Mahaidi, R. (2006). "An Experimental Investigation on Torsional Behaviour of Solid and Box-
section RC beams Strengthened with CFRP using Photogrammetry." Journal of Composites for Construction, 10(4).
Hii, A. K. Y., and Al-Mahaidi, R. (To be published). "Torsional Capacity of CFRP Strengthened Reinforced
Concrete Beams." Journal of Composites for Construction.
Hsu, T. T. C. (1993). Unified Theory of Reinforced Concrete, CRC Press.
Hsu, T. T. C., and Mo, Y. L. (1985). "Softening of concrete in torsional members - Design recommendations." ACI
Journal, 82(4), 443-452.
Panchacharam, S., and Belarbi, A. "Torsional Behaviour of Reinforced Concrete Beams Strengthened with FRP
Composites." FIB Congress, Osaka Japan.
SAI. (2001). AS3600-2001 - Concrete Structures, Standards Australia International Ltd.
Salom, P. R., Gergely, J., and Young, D. T. (2004). "Torsional Strengthening of Spandrel Beams with Fiber-
Reinforced Polymer Laminates." Journal of Composites for Construction, 8(2), 157-162.
Triantafillou, T. C., and Antonopoulos, C. P. (2000). "Design of Concrete Flexural Members Strengthened in Shear
with FRP." Journal of Composites for Construction, 4(4), 198-205.
Zhang, J. W., Lu, Z. T., and Zhu, H. (2001). "Experimental Study on the Behaviour of RC Torsional Members
Externally Bonded with CFRP." FRP Composites in Civil Engineering, Vol. 1, 713-722.

238
Third International Conference on FRP Composites in Civil Engineering (CICE 2006)
December 13-15 2006, Miami, Florida, USA

ANALYSIS AND DESIGN OF FRP REINFORCED CONCRETE

COLUMNS
Ching Chiaw Choo
(Research Associate, Kentucky Transportation Center, University of Kentucky, Lexington, Kentucky, USA)

Issam Harik
(Professor, Dept. of Civil Engineering, University of Kentucky, Lexington, Kentucky, USA)

ABSTRACT
Analytical investigations of rectangular concrete compression members reinforced with fiber reinforced polymer
(FRP) reinforcing bars were carried out based on the ACI ultimate strength approach. The strength interaction (P-M)
relations of reinforced concrete columns with FRP bars were examined and compared to columns with steel bars.
The results identified the possibility of premature-compression and brittle-tension failures which were related to the
compression and tension ruptures of FRP bars. Brittle-tension failure, when compared with premature-compression
failure, was more likely to occur due to low ultimate tensile strain of FRP bars. The study also concluded that the
ACI minimum reinforcement ratio limit for steel reinforced concrete columns was not adequate for use in FRP
reinforced concrete columns. In this study, design aids have been developed for FRP rectangular concrete columns
to determine the minimum required reinforcement ratio (ρf,min) for averting the brittle-tension failure to a failure
controlled by concrete crushing; a failure concept used for steel reinforced concrete columns. The proposed
approach using ρf,min enabled the analysis and design of concrete columns reinforced with FRP bars to be carried out
in a manner similar to columns reinforced with steel bars based on the ACI 318-05 provisions.

KEYWORDS
Columns, FRP, Analysis, Design, Failure

1. RESEARCH SIGNIFICANCE
In certain applications, FRP bars are favored for their non-corrosive and high-strength properties as evidenced by
their use in many concrete constructions. Considerable research efforts have contributed to the understanding of
concrete members internally reinforced with FRP composites; of particular interests are the flexural and shear
behaviors of FRP concrete members and slabs (Nanni 1993; Almusallam 1997; GangaRao and Vijay 1997;
Theriault and Benmokrane 1998; and Deitz, Harik, and Gesund 1999). At present, guidelines for the design and
analysis of FRP reinforced concrete members in flexure and shear can be found in ACI440.1R-03 (2003).

With better understanding, particularly the compression attributes, FRP bars can be employed as reinforcement in
concrete columns. This analytical study investigates FRP reinforced concrete columns with emphasis placed on the
columns’ behavior and failure mechanisms. This will ultimately lead to a rational approach – currently not available
in guidelines published by ACI (e.g. ACI440.1R-03 2003) – to the design and analysis of concrete columns
internally reinforced with FRP bars.

2. STRENGTH OF CONCRETE COLUMNS INTERNALLY REINFORCED WITH FRP

BARS
The ultimate strength approach is employed to study the axial load-bending moment strength interaction (P-M)
relations of concrete columns internally reinforced with FRP bars. Assumptions pertaining to the analysis of steel
reinforced concrete column strength interactions are used. Typical strength interaction (Pu* = Pu/bhfc’, Mu* =

239
Mu/bh2fc’) relations generated for rectangular reinforced concrete columns are presented in Figure 1. The findings
and observations related to the behavior of these concrete columns are as follows:

• Based on ACI 318 provisions, the strength interaction relations of steel reinforced concrete columns exhibiting
a well-defined balance point which signify the transition from a compression-controlled region to tension-
controlled region (Figure 1.a). A balance point is obtained when the outermost concrete fiber reaches an
ultimate strain in compression (εc = εcu = 0.003) and the outermost steel layer in tension reaches a yield strain
(εs = εy), simultaneously.

• Due to a lack of plasticity, the strength interactions of concrete columns internally reinforced with FRP bars do
not exhibit a balance point; as indicated in Figure 1.b for columns with reinforcement ratios of 5 % and 8 %,
respectively. Thus, the failure of these two strength interaction relations is compression-controlled defined by
concrete crushing.

• Brittle-tension failure could potentially occur in FRP reinforced concrete columns when the outermost concrete
fiber reaches an ultimate strain in compression (εc = εcu = 0.003) and the outermost tension layer of the FRP
bars reaches the ultimate strain in tension (εf = εfut), simultaneously. This is illustrated by a concrete column
reinforced with FRP bars with 1 % reinforcement ratio as shown in Figure 1.b. The 1 % reinforcement ratio is
the minimum allowed in ACI318-05 (2005).

P u * (M Pa) P u * (M Pa)
14 8
GFRP bars:
12 ρ=8% Eft = 45 GPa
Steel bars: 6 ρ=5% Efc = Eft
10 ρ=8% εfut = 1.4 %
Es = 200 GPa
εy = 0.21 % εfuc = 0.5εfut = 0.7 %
8 ρ=1%
4
ρ=5%
6

4 ρ=1%
Balance points 2
Brittle-tension failure:
2 εc = εcu and εft = εfut

0 M u * (M Pa) 0 M u * (MPa)
0 1 2 3 0 1 2
(a) (b)

Figure 1: Strength interaction relations of rectangular concrete columns reinforced with: (a) steel bars, and
(b) GFRP bars, with a concrete compressive strength (fc’) of 35 MPa.

2.1 Effects of Brittle-Tension or Premature-Compression Failure

Brittle-tension failure could potentially occur due to the low ultimate tensile strain of FRP bars. This can sometimes
result in an incomplete strength interaction relationship (Figure 1.b). Concrete beam-columns exhibiting brittle-
tension failure are incapable of reaching pure flexure (axial load, P = 0) prior to the tension FRP reinforcement
reaching its ultimate strain and strength. Brittle-tension failure due to tension rupture of FRP bars in concrete
columns can be potentially explosive as a large amount of strain energy is suddenly released. Conversely, concrete
crushing type failure is more desirable as evidenced by a more progressive and a less catastrophic type failure
exhibited by concrete flexural members with FRP bars (Nanni 1993; GangaRao and Vijay 1997; Theriault and
Benmokrane 1998).

While not illustrated figuratively, premature compression failure of concrete columns reinforced with FRP bars
could happen when compression rupture of FRP bars occurs prior to concrete strain reaching its ultimate (i.e.
ACI318 defines maximum usable concrete strain to be 0.003). It is worth pointing out, however, that compression
rupture of FRP bars is less likely to occur because the ultimate compression strain of FRP bars is generally greater
than the ultimate concrete strain in compression.

240
 3. PREVENTION OF BRITTLE-TENSION FAILURE

From the foregoing, concrete columns internally reinforced with FRP bars if designed to be compression-controlled
would have the following advantages: (1) The analysis and design of such columns can be performed based on the
ACI code approach; and (2) from a failure standpoint, maintaining compression-controlled failure would be less
explosive and more gradual as previously stated.

It appears, from Figure 1, concrete columns reinforced with FRP bars can be safeguarded from brittle tension failure
by providing a reinforcement ratio (ρf) larger than a minimum ratio designated as ρf,min; the reinforcement ratio
required to yield compression-controlled (εc = εcu) failure. Naturally, ρf,min of a concrete column reinforced with
specific FRP bars varies with a variety of factors such as properties of the FRP bars, concrete strength, and
reinforcement layout. To expedite the determination of ρf,min, design aids (Figure 2) were generated for rectangular
concrete columns reinforced with FRP bars independent of the FRP ultimate tensile strain (εfut). Additional aids can
be found in Choo (2005). The ordinates in Fig. 2 represent the tensile elastic moduli (Eft) of FRP bars ranging from
35 GPa to 210 GPa, covering most available FRP bars. The abscissas represent the tensile strains (εft) that will
develop at the outermost tensile reinforcing bar layer under pure flexure. As depicted, εft varies and can be
computed for any combination of Eft and ρ shown in the aids. To account for compression elastic moduli (Efc) of
FRP bars, different Eft/Efc ratios were also incorporated into these Eft-εft aids.

To illustrate the use of these aids, the reinforced concrete column internally reinforced with GFRP bar in Fig. 1.b is
revisited. Based on the concrete and GFRP properties, the required ρf,min is approximately 1.3% as shown in Fig. 2.a
to avoid brittle-tension failure.

250 250
ρ = 0.5%

ρ = 0.5%
fc' = 35 MPa fc' = 35 MPa
ρ = 8%

ρ = 8%

ρ = 1%
ρ = 2%

ρ = 1%

ρ = 2%
γ =0.9 γ =0.9
Efc/Eft = 1.0 Efc/Eft = 0.6
200 200

150 150
Eft (GPa)

Eft (GPa)

100 ρf,min ≈ 1.3% 100

50 50
45 GPa (Fig. 1.a)
0.014 (Fig. 1.b)
0 0
0 0.005 0.01 0.015 0.02 0.025 0.03 0 0.005 0.01 0.015 0.02 0.025 0.03
ε ft (mm/mm) ε ft (mm/mm)
(a) (b)

Figure 2: Tensile elastic modulus-tensile strain (Eft-εft) design aids for rectangular concrete columns
reinforced with FRP bars.

3.1 Modification Factors for Concrete and Reinforcement Layout

Note that the aids in Fig. 2 depend on two parameters: concrete strength (fc’) and the ratio of the distance between
the outer layers of rebars to the height of the column cross section in the direction of bending (γ). To account for the
increase or decrease of εft due to concrete strength, a modification factor, αc , was derived (Choo 2005):

⎛
21 − 0.2 f c ⎟ ' ⎞
αc = ⎜ ≥ 1.0 When 21 MPa ≤ fc’ < 35 MPa (1)
⎜ 14 ⎟
⎝ ⎠

⎛
25.25 − 0.15 f c ⎟ ' ⎞
αc = ⎜ ≤ 1.0 When 35 MPa ≤ fc’ ≤ 55 MPa (2)
⎜ 20 ⎟
⎝ ⎠

241
To account for γ, a modification factor, αγ , was derived (Choo 2005):

αγ = 1.5 – 0.556γ ≥ 1.0 when 0.45 ≤ γ ≤ 0.9 (3)

3.2 Modified Ultimate Tensile Strain

To use the aids for other values of the two parameters, the ultimate tensile strain (εfut) of FRP rebar shall be modified
as follows:

εfut* = εfut·αc·αγ (4)

This modified value, combined with the tensile modulus of elasticity, can then be used to determine a required ρf,min.
The introduction of a modified ultimate tensile strain, in fact, permits a quick and efficient way of incorporating
other factors in determining ρf,min.

4. CONCLUSION AND DESIGN RECOMMENDATIONS

Strength interaction studies of concrete columns reinforced internally with FRP bars identified the possibility of
these failures: (1) premature-compression, and/or (2) brittle-tension failures. These failures are associated with
failure of FRP bars prior to concrete crushing, and hence can be potentially disastrous.

Design aids that take material properties and reinforcement layout into consideration were developed for FRP
rectangular reinforced concrete columns where a minimum required reinforcement ratio (ρf,min) can be determined
and used to prevent brittle tension failure. It can be seen that the introduction of a modified ultimate tensile strain
allows a quick and efficient way to incorporate other factors that may influence ρf,min.

Based on the study, it is also observed that the minimum reinforcement ratio limit (1%) set by ACI318 for steel
reinforced concrete columns may not be adequate for FRP reinforced concrete columns. The maximum
reinforcement ratio limit (8%), however, is still applicable due to the fact that the limit is set to promote
constructability and to avoid rebar congestion.

5. REFERENCES
ACI Committee 318, 2005, “Building Code Requirements for Structural Concrete (318-05) and Commentary (318R-
05),” American Concrete Institute, Farmington Hills, MI.
ACI Committee 440, 2003, “Guide for the Design and Construction of Concrete Reinforced with FRP Bars (ACI
440.1R-03),” American Concrete Institute, Farmington Hills, MI, 42 pp.
Almusallam, T.H., 1997, “Analytical Prediction of Flexural Behavior of Concrete Beams Reinforced by FRP Bars,”
Journal of Composite Materials, Vol. 31, No. 7, pp. 640-657.
Choo, C.C., “Investigation of Rectangular Concrete Columns Reinforced or Prestressed Fiber Reinforced Polymer
(FRP) Bars or Tendons,” PhD Dissertation, University of Kentucky, Lexington, KY, 2005.
Deitz, D.H., Harik, I.E., and Gesund, H., 1999, “One-Way Slabs Reinforced with Glass Fiber-Reinforced Polymer
Reinforcing Bars,” American Concrete Institute, SP-188-25, 279-286.
GangaRao, H.V.S., and Vijay, P.V., 1997, “Design of Concrete Members Reinforced with GFRP Bars,” Proceedings
of the Third International Symposium on Non-Metallic (FRP) Reinforcement for Concrete Structures
(FRPRCS-3), Japan Concrete Institute, Sapporo, Japan, Vol. 1, pp. 143-150.
Nanni, A., 1993, “Flexural Behavior and Design of Reinforced Concrete Using FRP Rods,” Journal of Structural
Engineering, V. 119, No. 11. pp. 3344-3359.
Theriault, M., and Benmokrane, B., 1998, “Effects of FRP Reinforcement Ratio and Concrete Strength on Flexural
Behavior of Concrete Beams,” Journal of Composites for Construction, Vol. 2, No. 1, pp. 7-16.

242
Third International Conference on FRP Composites in Civil Engineering (CICE 2006)
December 13-15 2006, Miami, Florida, USA

PROBABILISTIC DESIGN OF FRP STRENGTHENING

OF CONCRETE STRUCTURES
Anders Carolin
(Dr., Luleå University of Technology, Luleå, Sweden)

ABSTRACT
Today, many design proposals exists that address different strengthening issues. Most proposals are derived on
assumptions made for designing of new structures and are also based on a deterministic approach where a safety
factor is added. The use of probabilistic methods is now extending and the reliability is sometimes calculated
afterwards for the strengthened structure. This paper presents how the reliability should be chosen on beforehand
when doing design for strengthening of an existing structure, which issues to be considered and also what safety that
one can expect from a strengthened structure. Partial coefficients on material properties and loads are used to give a
uniform treatment of risk of failure when strengthening of structures. When partial coefficients are chosen the reason
for strengthening and the strengthening method may be considered to achieve an optimal strengthening with respect
to structural safety and economy. The theoretical basis, the same as for Eurocode, has been used successfully in
Swedish design codes for a long time. The results from the analysis indicate that strengthened with CFRP plate
bonding has a very desirable effect on structural safety and reduces the risk of failure by considerable proportions.

KEYWORDS
Probabalistic, Design, Retrofitting, Concrete, FRP.

1. INTRODUCTION
The world is complicated. A concrete structure subjected to loading could be seen as a complicated system. By
strengthening such a structure, things certainly become complicated. For a long period of time, engineers have tried
to make models of the world and the structures in it. This means that simplifications are used to make it possible to
describe and estimate how things behave. Models are also quite often used to make predictions. When structures and
loads started to be modeled, the idea was to make the bearing capacity the same or higher than the highest known
load. Since models are simplifications, a safety factor was then used to make the structure a certain degree stronger
to cover for deterioration, model errors, and construction errors. This approach is called deterministic and do not
consider distributions in loads, material parameters, and so on. The deterministic approach also gives altering levels
of safety for different structures. In a probabilistic approach, uncertainties of loads and bearing capacities of
structures are considered in the model. Both bearing capacity of the structure and the load effect have statistical
distributions. Failure will occur when the load effect has one of the highest values and load bearing capacity has one
of the lowest values, at the same time. Failure is expressed by the probability of load effect becoming larger than
load bearing capacity. As an example, failure will not even occur if the load bearing capacity is one of the 1 %
lowest cases, if the load effect is just normal. Many codes, e.g. Swedish BBK94 (1994) and Eurocode EN (2002),
are based upon reliability and partial coefficients. In normal design by Swedish BBK94 (1994), the lower 5 %-
fractal is used as characteristic values, f k , of the materials. These values are then reduced by partial coefficients,
ηγ n , and, γ m , as presented in Equation 1 and the bearing capacity can then be estimated from the calculated
parameters, f d .

fk
fd = (Eq. 1)
ηγ nγ m

243
The load effect is taken as a statistical upper limit. Failure occurs when the load effect, S, are higher than the bearing
capacity, R, and the probability of failure can be described as the probability of G<0, where G is described by
Equation 2.

G= R−S (Eq. 2)

By law, the yearly probability of failure in Sweden must be 10-4 - 10-6 or higher, depending on safety class (1, 2 or 3)
prescribed by code. These values are directly connected to β–indices in range of 3.72 – 4.75. A structure causing
severe consequences if it fails must be design with the lower probability of failure. A structure causing relatively
harmless consequences if it fails may be design with a higher probability of failure. The coefficient, γ m , is
considering safety class and varies between 1.0 and 1.2. The partial coefficients ηγ n together with coefficients on
loads have been calibrated so that the demand on yearly probability of failure is met. Model uncertainties,
distribution of material properties, and difficulty of construction is considered when ηγ n is determined. In this paper
a favourable aspect for external strengthening regarding structural safety will be addressed.

2. BEARING CAPACITY OF STRENGTHENED STRUCTURE

The load effect depends on many stochastic variables, wind, dynamic load effects, snow, etc. In the case of
strengthening, in some cases it can be possible to determine the future load effect more precisely compared to the
design of a new structure. It can actually be possible to measure the load effect on a structural member when a load
is applied on the structure. Dead-load can be estimated better when the true dimensions can be measured. Loads
from support settlements can in some cases be reduced when an old structure is going to be strengthened. This can
make the scatter of the load effect narrower. Strengthening with plate bonding with CFRP does not significantly
change a structure’s dead-load. The load effect on a member in a statically undetermined system can still be altered
because of changed stiffness of the studied member. The load effect will, however, not be further discussed here.
For a reinforced concrete structure, several stochastic variables affect the load bearing capacity. Concrete properties,
steel reinforcement properties, length of internal lever arm, mode of failure and anchorage of reinforcement are all
important variables for a structure subjected to flexure. To ensure a ductile failure it is often desired to have an
under-reinforced cross-section, which means that the bearing capacity is controlled by amount of strengthening. The
flexural bearing capacity for an under-reinforced cross-section can, in its most simplified form, be described as inner
lever arm multiplied by the force in the reinforcement as described in Equation 3:

R = (d − c )As f st (Eq. 3)

where d is effective height; c is centre of gravity for the compressive force measure from top of beam; As is
reinforcement area; and fst is capacity of reinforcement in tension. If a structure is strengthened with externally
bonded CFRP, new variables such as composite properties and new failure modes may be added. On the other hand,
when a structure is going to be strengthened it is possible to undertake field measurements that can give a more
determined description of the existing materials, dimensions and possible defects of the structure. When
strengthening is applied the length of the internal lever arm for the composite might be taken as deterministic. The
load bearing capacity for a strengthened structure will be dependent on more factors compared to the load capacity
of the original structure. The mode of failure may change depending on how much the structure is strengthened.
However, to a certain amount the failure mode can be quite reliably described. A normal reinforced concrete
member that “fails” by yielding of reinforcement will for a small amount of strengthening fail by fibre rupture, on
the assumption that anchorage is sufficient. In analogue of reinforced concrete the bearing capacity of a strengthened
reinforce cross-section can be described as Equation 4, where h is the height of the cross-section; Af is area of
strengthening system; εf is ultimate strain of fibres; and Ef, is modulus of elasticity for strengthening system.

R = (d − c )As f st + (h − c )A f ε f E f (Eq. 4)

Distribution of variables for reinforced concrete has for example been studied by Jeppsson (2000) and a limited
study of strengthening systems has been undertaken by Plevris et al. (1995). Partial coefficients for strengthening
have been proposed by Monti and Santini (2002) and Täljsten (2003). For an under-reinforced beam the bearing
capacity is mainly limited by the amount of reinforcement. In case of a strengthened beam, reinforcement consists of

244
both internal steel bars and CFRP. The ultimate capacities of the two materials as well as the two different internal
lever arms are independently stochastic. For failure, it is necessary that all variables are at a critical value at the
same time. A weak steel bar can be compensated by the composite having its mean value and the structure will have
a bearing capacity larger than the load effect. In the same way, a steel bar with medium performance can
compensate for a composite that has a performance lower than its 5 % -fractal. The risk for both materials, and
therefore the load bearing capacity to be lower than the minimum acceptable level for the strengthened structure,
will decrease compared to the non-strengthened structure.

It would be possible to choose partial coefficients so that the probability for failure will be the same for the original
structure subjected to original loading and the strengthened structure with the new loading. Since different amounts
of strengthening will give different importance to the different stochastic variables, it would imply that the partial
coefficient will vary with the strengthening amount. This is not a reasonable situation and it is suggested that the
partial coefficient are determined as if the variables for the existing structure are deterministic. This approach will
give an additional reliability of the structure that should be seen as an extra security provided by the strengthening
system.

3. CALCULATION EXAMPLE
With data on concrete and steel reinforcement from literature, (JCSS PMC, 2001), an ordinary under-reinforced
structure was designed deterministically by Equation 3 and partial coefficients from Swedish code. When
considering distribution of parameters and a deterministic load, the structure got a probability of failure of
2.05 * 10-5 calculated by the second-order reliability method (Melchers 1999). Material data for strengthening
system was tested (Carolin et al. 2004) and strengthening for an increased load was designed deterministically by
Equation 4 with an assumptions on partial coefficients for the strengthening system. The tested CFRP where
pultruded laminates with an average modulus of elasticity of 150 GPa. Considerations were given to ensure an
under-reinforced cross-section even after strengthening. By setting original structure as deterministic, the increased
load as deterministic, and the real distribution of composite parameters, partial coefficients for strengthening system
were calibrated. Calibration gave a partial coefficient for fibre rupture of 1.15 and a probability of failure for the
strengthened structure of 1.35 * 10-5. Then, by having all parameters for bearing capacity as stochastic the
probability of failure were calculated to 1.23 * 10-11.

4. ANALYSIS
Two stochastic undetermined load carrying systems give a higher security against failure compared to a single
system. The increase in security is dependent on the ratio of influence of each system and the distribution of
included parameters. Seeing that the importance of a strengthening system varies, partial coefficients should be
determined based on that other reinforcement is seen as deterministic. In cases of an internal steel reinforcement the
structural safety of the strengthened structure will increase. It is quite common with material investigations in
conjunction of upgrading a structure. For concrete structures in need of strengthening it is quite common to
investigate the real compressive capacity of the concrete which then quite often allows increase of reinforcement
ratio without changing the failure mode. Even though not as common, real capacity of reinforcement is sometimes
investigated. In cases when real material properties are used, it is important that the partial coefficients of the
strengthening system will provide the desired and prescribed structural safety together with the partial coefficients
on loads. Several design codes for reinforced concrete are based on stress. Design codes on CFRP strengthening are
on the other hand often based on strains that are multiplied with a modulus of elasticity. During calibration of partial
coefficients, special attention must be given so that the two different approaches harmonize and that distribution of
strain and modulus of elasticity are treated correct.

5. DISCUSSION AND CONCLUSIONS

The found partial coefficient is reasonable for a product manufactured in an industrial process like the pultruded
laminate studied. For other CFRP products, i.e. hand lay-up composites, it is probably more suitable with higher
factors. Täljsten (2003) gives a suggestion on how partial coefficient may be compiled. Nevertheless, partial factors
must be calibrated to meet prescribed and desired levels of structural safety. Structures should be designed to fail in

245
a ductile way or at least with adequate warning signals preceding a potential collapse. One argue against CFRP is
that most fibre composites are linear-elastic material without any defined yield plateau. It is important to distinguish
between material ductility and structural ductility. A structure with a brittle failure in shear may in fact be
strengthened so that the failure mode will change to a more ductile and friendly mode, even when strengthened with
a linear elastic material (Carolin, 2003). Failures that might be seen as brittle are: shear failures and anchorage
failures. Fibre rupture is also brittle but normally proceeded by large deformations. Since different failure modes are
more preferable than others, one solution is to make those more probable than others by design, i.e. make undesired
failures less probable as suggested by Pilakoutas et.al. (2002). Different probabilities for failure modes with
different brittleness are motivated by that the risk should be constant. By studying risk as the combination of
probability and consequence, brittle failures with no warnings may give larger consequences and should therefore
have a lower probability of occurring. Different partial factors could be used for different failure modes. A special
partial factor for design of anchored could be used. When designing for shear, a special partial factor on shear loads
can be applied. It is also possible with different factors on the material depending on studied failure mode.

With deterioration in mind the strategy for strengthening becomes more complicated. Probability of failure is a
suitable unit when measuring the performance level of a structure. Since all structures are deteriorating the structural
safety will decrease. This implies that structures must be built with certain margins on structural safety and with a
lower probability of failure for a new built structure. Uncertainties in construction are normally considered in
original design, which then gives the margin for decrease of structural safety caused by deterioration. The structural
safety of a new built structure should actually be investigated before deterioration starts.

Finally, loss of strengthening effect must also be considered in design. Structures that may be subjected to vandalism
or collision should have the capacity of bearing at least the dead load even if the effect of the strengthening system is
lost. Plate bonding is normally relatively sensitive of fire loads and strengthened structures must be able to carry
prescribed loads during fire for a certain time. For all cases, structures should be able to carry loads with a
prescribed probability with respect to likeliness of certain load combinations at all time.

6. REFERENCES
BBK 94 (1995). ”Boverkets handbok om betongkonstruktioner, BBK 94, Band 1, Konstruktion”. Boverket,
Stockholm, 185 pp. ISBN 91-7147-235-5 (in Swedish).
Carolin, A., Täljsten, B., Nilsson, M., Enochsson, O. and Fahleson, C. (2004). “Tillförlitlighetsanalys för reparerade
eller förstärkta byggnadskonstruktioner” Luleå University of Technology (in Swedish).
EN (2002). “Eurocode - Basis of structural design. EN 1990:2002 (E)”. CEN, European Committee for
Standardization, Brussels, 87 pp.
Carolin, A. (2003). “Carbon Fibre Reinforced Polymers for Strengthening of Structural Elements”, Doctoral Thesis,
Luleå University of Technology, 184 pp. http://epubl.luth.se/1402-1544/2003/18
JCSS PMC (2001). Probabilistic Model Code. Joint Committee on Structural Safety, JCSS,
http://www.jcss.ethz.ch/JCSSPublications/PMC/PMC.html (2006-04-15).
Jeppsson, Joakim (2003): “Reliability-Based Assessment Procedure for Existing Concrete Structures”, Doctoral
Thesis, Lund University of Technology, 188 pp.
Melchers, R. (1999): “Strutural Reliability Analysis and Prediction”, 2nd Ed, John Wiley & Sons, Chichester, UK,
1999, 437 pp. ISBN 0 471 98771 9.
Monti, G. and Santini, S. (2002): “Reliability-based Calibration of Partial Coefficients for Fiber-Reinforced Plastic”
Journal of Composites for Construction. August. pp 162-167
Pilakoutas, K., Neocleous, K. and Guadagnini, M. (2002): “Design Philosophy Issue of Fiber Reinforced Polymer
Reinforced Concrete Structures” Journal of composites for construction. August, pp. 154-161.
Plevris, N., Triantafillou, T. and Veneziano, D. (1995): “Reliability of RC Members Strengthened with CFRP
Laminates” Journal of Structural engineering. July, pp. 1037-1044.
Täljsten, Björn (2003): “FRP Strengthening of Existing Concrete Structures - Design Guidelines - Second Edition”,
Luleå University of Technology, ISBN 91-89580-03-6, 230 pp

246
Third International Conference on FRP Composites in Civil Engineering (CICE 2006)
December 13-15 2006, Miami, Florida, USA

THRESHOLDS OF CONSTRUCTION ANOMALIES

IN FRP REPAIR SYSTEMS
Baris Yalim
(Ph.D. Candidate, Florida International University, Miami, FL, USA)

Ahmet Serhat Kalayci

(Ph.D. Student, Florida International University, Miami, FL, USA)

Amir Mirmiran
(Professor and Chair, Florida International University, Miami, FL, USA)

ABSTRACT
Fiber reinforced polymer (FRP) composites have been used extensively in repair applications over the last two
decades. In addition to their high-strength and lightweight properties, important characteristics of FRP systems for
structural repair and strengthening applications include their resistance to electro-chemical corrosion, and speed and
ease of installation. Therefore, FRP systems can be efficient and economical options to traditional repair methods.
However, relationship between the long-term performance of FRP repair systems and their construction processes
are not easy to quantify. Moreover, tolerances for most construction anomalies such as surface irregularities and
crack widths are not yet based on adequate scientific data. For instance, depending on their state (i.e. shear or
flexural) and size, cracks in the concrete may either be left untreated or epoxy injected. The thresholds separating
these approaches may depend on technical and economical considerations. This paper presents the current study to
evaluate the effect of surface flaws on the performance of wet lay-up FRP systems. Four types of flaws are
considered; surface roughness, surface flatness, voids and bug holes, and cracks/discontinuities. Discussions of test
results and recommendations are also provided.

KEYWORDS
Surface flaws, FRP repair systems, Reinforced concrete, Construction Guides

1. INTRODUCTION
Many concrete structures that were originally constructed for a specific use may later be renovated or upgraded for a
different application due to higher load-carrying capacity requirements or maintenance purposes. As a result,
existing structures may need to be retrofitted to meet higher load demands. With over two decades of field
applications and research findings, it is generally recognized that the use of FRP systems is efficient and technically
reliable for repair and retrofit of concrete structures. Since its first application in Europe and Japan in the early
1980’s, use of bonded repair and retrofit of concrete structures with FRP systems has progressively increased to the
extent that today it counts for at least 25 federally supported Innovative Bridge Research and Construction (IBRC)
projects in the United States (NCHRP 514, 2004). However, there is currently a lack of generally accepted
construction specifications and process control procedures for FRP repair systems, making users heavily dependent
on specific guidelines from FRP manufacturers. As the FRP technology matures and moves into widespread use, the
need has become more urgent than ever to equip users with the means to specify and control the constituent
materials and the adequacy of the construction process. The acceptance and use of the FRP repair systems will
ultimately depend on the availability of clear design guidelines, installation procedures, and construction
specifications (Scalzi et al. 1999). Practical guidelines for such repair techniques are not yet readily available for
practicing structural engineers (Shahawy et al. 1996). Thus, the objective of this research is to develop tolerances for
surface irregularities and crack widths for bonded repair of concrete structures using FRP composites.

247
2. TEST SETUP AND PROCEDURE
The test program involves short-term static loading of 45 reinforced concrete beams bonded with FRP sheets. All
beams were T-sections with a flexural span of 6.5 ft and target compressive strength of 5 ksi (see Figure 1). A 4-in-
wide and 67.5-in-long carbon fiber sheet (SikaWrap Hex 103C) was attached as centered to the bottom of the beam
using Sikadur 300 epoxy resin. Additional two 4-in wide fiber sheets were wrapped as U-straps in each shear span to
provide anchorage for the FRP system. All beams were tested under three point bending. A self-reaction test frame
shown in Figure 1 was designed to accommodate 7-foot-long specimen with 60-kip capacity. The beam was
supported on two hinges, with mid-span load applied using an Enerpac hydraulic jack. Each beam was instrumented
with three 1.5-in-stroke potentiometers to measure deflections at ¼ span points and a strain gauge on the FRP sheet
at mid-span of the beam. A 50-kip-capacity Futek load cell measured the applied load.

12.0"

No.3 1.5" No.3

3.0"

No.3@5" 12.0"
1.0" 1.0"

No.5 1.5" No.5

6.0"

Figure 1: Test Specimen and Test Frame

A total of 9 specimens were tested for the effect of surface roughness. Three concrete surface profiles (CSP) were
selected for each three beams, corresponded to CSP 1, CSP 2-3 and CSP 6-9 according to the International Concrete
Repair Institute recommendations (ICRI/ACI, 1999). Roughness level of the concrete surface increases from 1 to 9.
The intentional roughness levels and surface profiles were developed using grinding and pressure washing after the
beam was cured. For surface flatness, a total of 12 specimens were prepared with two different levels of out-of-
flatness (1/8 in and 1/16 in over 1 ft length at the mid-span), as peaks and valley, and two duplicates for each case.
The intentional out-of-flatness was developed using grinder after concrete was cured. Another 12 specimens were
prepared for voids and bug holes with three different void depths (1/4 in, 3/16 in and 1/8 in) and three different void
diameters (1/2 in, 3/8 in and 1/4 in). Frequency of voids on the concrete surface was selected as 10% in terms of
surface area. Surface voids were created by drilling concrete after it was cured. Finally, 12 specimens were made
with three different crack widths (1/8 in, 3/32 in and 1/16 in) and three different crack spacing (1 in, 1.5 in and 2 in).

3. TEST RESULTS & DISCUSSIONS

Figure 2 shows comparisons of load-deflection and load-strain curves, respectively, for all nine beams tested for
surface roughness. Mid-span deflections and strains correspond to the peak load. It is clear that the responses are
quite similar, regardless of the surface profiles of the beams. All beams typically failed due to FRP debonding at an
ultimate load of 30-35 kip with similar mid-span strain and deflection. There is a slight difference between the
different surface profiles, as higher roughness translates into a slightly higher capacity. However, the difference is
not statistically significant. This may be attributed to the fact that the rougher surface allows for direct bond of FRP
sheet to the aggregates. The U-straps proved helpful in anchoring the FRP sheets. The FRP sheets began to peel off
as soon as the U-straps ruptured.

In order to evaluate the effect of voids and cracks on FRP repair performance, a statistical method called “Design of
Experiment” was utilized. Fundamental purpose of this approach is to determine the relationship between factors
affecting a process (surface flaws) and the output of that process (load, displacement and strain).

248
 40 40

35 35

30 30
CSP1 Test1
CSP1 Test1 CSP1 Test2
25 25

Load (kips)

Load (kips)
CSP1 Test2 CSP1 Test3
CSP1 Test3 CSP2 Test1
CSP2 Test1 CSP2 Test2
20 20
CSP2 Test2 CSP2 Test3
CSP2 Test3 CSP3 Test1
15 CSP3 Test1 15 CSP3 Test2
CSP3 Test2
CSP3 Test3
CSP3 Test3
10 10

5 5

0 0
0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 0.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5 4.0
Mid-span Deflections (in)
Midspan Strain (mE)

Figure 2: Load-Displacement and Load-Strain Curves for Surface Roughness Specimens

The DOE analysis provides 3D response surface interpolated over test data range. Figure 3 illustrates the variation
of each response measure (load, deflection or strain) with a given void depth and diameter. It is also useful to locate
the flawless specimen on the surface for comparison. Flawless specimen is shown as a separate layer cutting the
surface on various points. In addition to the flawless specimen, several layers were added for each response at
different percent reduction of that response measure. This leads to the enhanced determination of the possible
specification thresholds for both factors based on each response. Each layer in these graphs cut the response surface
at different level such that corresponding void depth and void diameter may be obtained according to the desired
percentage of reduction or increase in the response measure. Figure 3b implies that the deflection is more correlated
with void diameter than void depth.

a) Load Variation b) Displacement Variation c) Strain Variation

Figure 3: Load-Displacement-Strain Variation with Void Depth and Void Diameter

The DOE approach was again utilized for the evaluation of cracks. Figure 4a shows the load capacity in relation to
crack width and spacing. There exists an increasing trend up to 1.5 in spacing. Same conclusion can be drawn in the
case of strain, which confirms rise after 1.5 in crack spacing. Similarly, there is substantial drop in the displacement
as the spacing becomes larger, particularly when it exceeds 1.5 in. With the available test data, it may be concluded
that 1.5 in spacing is the potential limit for existing cracks as far as load capacity, displacement and strain are
considered. As the spacing increases the maximum load also increases, as expected. Maximum increase in the
ultimate load compared to control specimen (unstrengthened) is about 19% and it occurs for the beams with 3/32 in
crack width and 1.5 in crack spacing. Graphs in Figure 4 also imply that crack width does not show a clear
correlation with any of the response measures. This may be because of the negligible difference in crack widths. It
should be noted that only one test is performed for each crack width and crack spacing. Therefore, it may also be
likely that a single test result may affect possible trends or expected correlations.

Load-displacement and load-strain curves for all beams tested for surface flatness are shown in Figure 5. Debonding
of longitudinal FRP at the mid-span for the 1/8 in valleys occurs much earlier than that of the control specimens and
the 1/16 in valleys. This behavior leads to a plateau at a load level which is about 19% less than the average capacity
of control specimens. Average peak load for the 1/16 in valleys is about 4.5% less than that for the control
specimens, which is quite reasonable. Failure mode was FRP debonding for all defective specimens as well as
control specimens regardless of their out-of-flatness levels.

249
 a) Load Variation b) Displacement Variation c) Strain Variation

Figure 4: Load-Displacement-Strain Variation with Crack Width and Spacing

Load-strain curves have similar trend until failure. Significant increase in the strain is observed immediately after
cracking for the 1/16 in valleys, whereas there is a smooth transition in the strain for the 1/8 in valleys. Average
strain values at cracking for the 1/8 in valleys are below those of the control specimens and conversely, average
strain values at cracking for the 1/16 in valleys are slightly higher than those of the control specimens. Unlike the
valleys, beams with peaks do not have load plateau and their overall behavior is quite similar to that of the control
beams which underlines the criticality of valleys in comparison to the peaks. In conclusion, 1/8 in valleys as an out-
of-flatness on the concrete surface has significant effect on the composite performance. On the other hand, behavior
of the 1/16 in valleys is very similar to that of control specimens. Thus, the 1/16 in should be the controlling
threshold for valleys and depressions, and accordingly any valleys or depressions on the concrete surface deeper
than 1/16 in measured from 12 in straight edge placed on the surface should be smoothed by grinding or filled using
epoxy resin mortar. Peaks on the concrete surface have been found as less critical as compared the valleys of the
same size and no threshold is recommended for the peaks at this stage of the research.

35 35

30 30

25 18V Test2
25 18V Test2
18P Test1 18P Test1
18V Test1
Load (kips)
Load (kips)

18V Test1
20 20 116V Test1
Debonding of bottom 116V Test1 116V Test2
FRP at the curvature 116V Test2 116P Test1
116P Test1 116P Test2
15 15
Control 1
116P Test2
Control 2
Control 1
18P Test2
10 Control 2 10
18P Test2

5 5

0
0
0.0 0.5 1.0 1.5 2.0 2.5 3.0
0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4

Deflections (in) Midspan Strain (mE)

Figure 5: Load-Displacement and Load-Strain Curves for Surface Flatness

4. REFERENCES
NCHRP 514. (2004) “Bonded Repair and Retrofit of Concrete Structures Using FRP Composites - Recommended
Construction Specifications and Process Control Manual.” Report 514, Mirmiran, A., Shahawy, M., Nanni, A., and
Karbhari, V., National Cooperative Highway Research Program (NCHRP), Washington, D.C.
Scalzi, J.B., Podolny, W., Munley, E., and Tang, B. (1999). “Guest Editorial,” Journal of Composites for
Construction, American Society of Civil Engineers, Vol. 3, No. 3, p. 107.
Shahawy, M. A., Beitelman, T., Arockiasamy, M., and Sowrirajan, R., (1996) “Experimental Investigation on
Structural Repair and Strengthening of Damaged Prestressed Concrete Slabs Utilizing Externally Bonded Carbon
Laminates,” Composites, Part B 27B, pp. 217-224

250
Third International Conference on FRP Composites in Civil Engineering (CICE 2006)
December 13-15 2006, Miami, Florida, USA

STATIC AND DYNAMIC PERFORMANCE OF FRP DECK BRIDGES

UNDER VEHICLE LOADING AND IMPLICATIONS IN DESIGN
Yin Zhang
Associate Professor, Nanyang Institute of Technology, Nanyang, China; Visiting Scholar, Department of Civil &
Environmental Engineering, Louisiana State University, Baton Rouge, Louisiana 70803, U.S.A

C.S. Cai
Associate Professor, Department of Civil & Environmental Engineering, Louisiana State University, Baton Rouge,
Louisiana 70803, U.S.A

ABSTRACT
The characteristics of bridges with FRP decks (such as mass, stiffness, and damping) are significantly different from
that of the traditional concrete and steel bridges, which may result in a much different performance of FRP bridges
from these traditional bridges. For this reason, detailed finite element analyses are used in the present study to
investigate the load distribution and the dynamic response of FRP deck bridges. The study is focused on a steel
multi-girder bridge and a concrete multi-girder bridge that are modeled using the finite element method to predict
the modal characteristics. The present study analyzes the bridge-vehicle interaction based on a three dimensional
vehicle-bridge coupled model. The dynamic response of bridges is obtained in the time domain by using an iterative
procedure employed at each time step, considering the road roughness of the deck as a vertical excitation to the
vehicles. The load distribution and the dynamic response of bridges are compared between the FRP deck and
concrete deck bridges. In addition, there are some arguments whether a composite action between the deck and
girders should be pursued or if a simple non-composite design should be used. Discussions on this aspect have been
made by modeling both the composite and non-composite systems.

KEYWORDS
Fiber Reinforced Polymers; Bridge Deck; Vehicle Bridge Interaction, Dynamic Response, Load Distribution

1. INTRODUCTION
The acceptance of FRP materials in bridge engineering is mainly due to their superior properties such as a high
strength-to-weight ratio, better durability, corrosion resistance, and fatigue resistance over steel and concrete
materials. An immediate advantage of using an FRP deck to replace a deteriorated concrete deck is the reduction of
the superstructure dead load, which results in an increase in the allowable live load capacity. Over the last decade,
some FRP bridge deck systems have been proposed, and there is a growing need to understand the behavior of FRP
deck bridges. However, very little is known about the lateral distribution of vehicle loads and the dynamic response
of bridges with a FRP deck. Therefore, the development of FRP bridge decks has been limited.

The load distribution throughout the bridge deck and the vehicle-induced dynamic impact on bridges are of primary
importance in the design of bridges. The load distribution factor and the dynamic impact (or allowance) factor have
been used worldwide in bridge design, and extensive experimental and theoretical work has been conducted to
determine these factors for bridges with conventional concrete decks. The characteristics of the FRP decks (such as
mass, stiffness and damping) are significantly different from those of the traditional concrete and steel decks, which
could result in a different performance of FRP deck bridges from the traditional bridges. In the present study, a
typical steel multi-girder bridge and a concrete multi-girder bridge with a span length of 60 ft (18.288m) were
studied. Connections between an FRP deck to the girders are more difficult than those between a concrete deck and
the girders. There are some arguments whether a composite action between the deck and girders should be pursued
or if a simple non-composite design should be used. To resolve this issue, the load distribution and the dynamic
response were compared in three conditions, namely FRP deck fully composite, FRP deck partially composite, and

251
concrete deck fully composite with the girders. The dynamic response of the bridge caused by a 3-axle truck was
obtained in the time domain by using an iterative procedure employed at each time step. The influence of vehicle
velocity as well as bridge surface roughness index on bridge performance was investigated.

2. SIMPLIFIED MODEL OF FRP DECK AND DYNAMIC EQUATION OF MOTION

The FRP bridge decks used in the present study are of a

sandwich construction. As shown in Fig. 1, FRP laminates are
attached to a closed-cell FRP, honeycomb-type, sinusoidal core,
which extends vertically between the two face laminates (or
skins). The geometry of this sandwich structure is designed to
improve stiffness and buckling response through the continuous
support of core elements with the face laminates (Plunkett
1997). Due to the geometrical complexity of this panel
configuration, a finite element modeling and analysis for an
entire bridge can be very complicated, if not impossible. In this
research work, therefore, finite element modeling techniques
were employed to develop simplified equivalent properties based
on stiffness considerations for this structure. The complex
sandwich hollow panel was therefore reduced to a solid
orthotropic plate using the equivalent properties derived
Fig.1 Sandwich Panel Configuration
(Oghumu et al. 2005).

HS20-44 truck, which is a 3-axle tractor-trailer type, is a major design vehicle in the AASHTO specifications
[AASHTO 2002, 2004]. This truck is chosen in this study and is idealized to a vehicle model consisting of two
vehicle bodies and 6 wheel bodies. The interaction force between the bridge and the vehicle is dependent on the
motion of both the bridge and the vehicle, and that the vehicle displacement is related to the bridge displacement,
road surface profile, and position of the vehicle. The equations of motion for the coupled system are written as
(Zhang et al. 2006):

M b  d&&b  C b + C bb C bv  d&b   K b + K bb K bv  d b   Fbr 

  +  +  = r G
(1)
 M v  d&&v   C vb C v  d& v   K vb 
K v  d v   Fv + Fv 

where {db}, [Mb], [Cb], and [Kb] are the displacement vector, mass matrix, damping matrix, and stiffness matrix of
the bridge, respectively; {dv}, [Mv], [Cv], and [Kv] are the displacement vector, mass matrix, damping matrix, and
stiffness matrix of the vehicle, respectively; and { FvG } = the gravity force vector of the vehicle. It is assumed that
the wheels always maintain a point contact with the bridge deck without separation. The equations of motion for the
vehicle and bridge are coupled through the interaction force and the terms Cbb, Cbv, Cvb, Kbb, Kbv, Kvb, Fbr , and
Fvr stem from the contact (interaction) force. The road surface profile was simulated in the space domain, which
serves as an input to the vehicle-bridge model.

3. NUMERICAL ANALYSIS OF TYPICAL GIRDER BRIDGE MODELS

One steel girder bridge model and one prestressed concrete girder bridge model were developed. The span length for
both bridges is 60ft (18.288m). The two bridges were designed for the HS20-44 loading, and they both consist of
five identical girders (with a spacing of 2.29m) which are simply supported. Meanwhile, in order to compare the
performance, two types of the bridge decks were used: a honeycomb-type sinusoidal core FRP sandwich panel (see
Fig. 1) and a traditional concrete deck. The FRP decks were designed for fully composite or partially composite. The
thickness of FRP decks used in these models is 8 in (203 mm), and the deck was simplified as an equivalent
orthotropic solid panel, as discussed earlier. The thickness of the concrete decks is 7.5 in (191 mm) for both the steel
and prestressed concrete girder bridges.

252
The bottom-flange stress values at the mid-span of the girders obtained from the FEA were used to calculate the
Load Distribution Factors (LDFs). By comparing the two cases of “FRP Deck Fully Comp.” and “FRP Deck
Partially Comp.”, it is observed that when the deck and the girders are partially composite, the LDF values are
larger, since in this case a smaller portion of loads are shared by the other girders. In other words, bridges with
partially composite conditions cannot distribute loads as uniformly as bridges with fully composite conditions. By
comparing the two cases of “FRP Deck Fully Comp.” and “Concrete Deck fully Comp.”, it is observed that due to
the higher stiffness of the concrete deck, the LDF values of the bridge with concrete deck are smaller than those with
the FRP deck. Meanwhile, it is observed that the stress values of concrete girder bridges are smaller than those of
steel girder bridges, because concrete girder bridges have a lager stiffness compared to steel girder bridges. LDFs of
concrete girder bridges are not as sensitive to deck stiffness as steel girder bridges are.

To investigate the effect of the deck system on bridge dynamic performance, vehicle-bridge interaction analyses of
the bridge system with different deck configurations were carried out. The multi-girder bridges were assumed to be
at rest before the vehicle entered the bridge. A parametric sensitivity study was conducted to analyze the effect of
factors such as road surface condition and vehicle velocity on the bridge dynamic response. The objective was to
compare the dynamic performance of the FRP deck bridges (fully composite or partially composite) and the
corresponding concrete deck bridges, and find some correlations between the bridge dynamic performance and these
parameters. Many investigations have shown that the roughness of a bridge surface is an important factor that affects
the dynamic response of bridge structures (Wang et al. 1992). In this study, classification of road roughness based
on the International Organization for Standardization was used, and the road surface profile was simulated in the
space domain. Two road conditions were considered as inputs to the vehicle-bridge coupled model, namely: (1) road
surface condition is good; and (2) road surface condition is poor. Based on these two road conditions and the vehicle
velocity of v =10m/s, 20m/s, or 40m/s, the dynamic responses at the bottom-flange at the mid-span of the center
girder were evaluated. Fig. 2 demonstrates the vehicle velocity, road roughness, and displacement relationships of
the steel bridge with the three different deck conditions. Due to the page limitation, other results cannot be presented
here but are summarized below.

FRP Deck steel Girders Fully Composite

5 FRP Deck steel Girders Partially Composite
Displacement d(mm)

Concrete Deck steel Girders Fully Composite

-5 Road Surface Condition--Good

V=10m/s

-10
2.5 3 3.5 4 4.5 5 5.5 6 6.5 7
Time t(s) FRP Deck steel Girders Fully Composite
FRP Deck steel Girders Partially Composite
5 Concrete Deck steel Girders Fully Composite
Displacement d(mm)

-5

-10
Road Surface Condition--Poor
-15 V=10m/s

-20
2.5 3 3.5 4 4.5 5 5.5 6 6.5 7
Time t(s)

Fig. 2 Displacement Comparison of Steel Multi-Girder Bridge

253
According to the numerical simulation, deck types have seriously affected the displacement. While the FRP Deck
Partially Composite condition results in the largest displacement, the Concrete Deck Fully Composite condition
gives the lowest displacement. This observation agrees with that observed for the static loading case as discussed
earlier since the deck stiffness is in an ascending order as FRP Deck Partially Composite, FRP Deck Fully
Composite, and Concrete Deck Fully Composite. The difference between the FRP Deck Partially Composite and
FRP Deck Fully Composite are more pronounced in bridges with steel girders than those with concrete girders. For
accelerations, the values of bridges with an FRP deck are far larger than those with a concrete deck, even under a
low vehicle velocity condition; the acceleration values of bridges with a FRP deck in the partially composite
condition are generally close to the values of those with a FRP deck in the fully composite condition. Under low
vehicle velocity (10m/s), the FRP deck in the partially composite condition generally results in a higher acceleration
than that of the FRP deck in the fully composite condition. Under high vehicle velocity (40m/s), the trend is just the
opposite. The vehicle velocity effect is more pronounced when the road surface condition is poor. Road roughness
of a bridge has seriously affected the vehicle’s vibrations, thus affecting the vehicle-bridge interaction. It can be seen
from the figures that the worse the bridge road condition, the larger the bridge dynamic displacement, and the far
larger the bridge dynamic acceleration under the truck load. This situation is more obvious in steel girder bridges. A
poor road condition not only influences the bridge’s normal operation, it moreover creates a vertical acceleration,
which can make the driver uncomfortable and may cause a higher deterioration rate of the bridge. Therefore,
maintenance of the bridge road surface in a good condition is very important in reducing the vehicle impact effect.
However, the dynamic response of bridges does not increase monotonically with the increase in vehicle velocity.
There is a peak value corresponding to a specific vehicle velocity, which is considered as being related to a vehicle
induced resonant vibration.

4. CONCLUSIONS
The present study used an equivalent orthotropic solid plate model for the FRP hollow sandwich panel. For both
load distribution and dynamic response, bridge deck types have seriously affected the results. The LDF values of
FRP deck bridges are larger than those of concrete deck bridges. The dynamic response of FRP deck bridges is also
larger than that of the concrete deck bridges. The FRP deck bridges with partially composite conditions have a larger
girder distribution and dynamic response than those of the FRP deck bridges with fully composite conditions.
Therefore, in order to obtain a better performance, it is necessary to strengthen the connection between the FRP deck
and girders through structural measures. However, this is a challenge and usually an expensive requirement for a
FRP deck system. If the non-composite condition is preferred for the FRP deck system, then it should be noted that
the load distribution and dynamic allowance developed for the full composite condition may not be conservative for
the girder design. Road roughness and vehicle velocity all significantly affect the dynamic performance of both the
analyzed FRP deck and the concrete deck bridges. The road roughness will greatly affect the vibration of the bridge,
even at low vehicle speeds when the surface condition is poor. Acceleration seems to be more sensitive to a poor
road condition than to a good road condition with the same vehicle speed, especially for steel girder bridges.

5. REFERENCES

American Association of State Highway and Transportation Officials (AASHTO) 2002, “Standard Specification for
highway bridges,” Washington, DC
American Association of State Highway and Transportation Officials (AASHTO) (2004). “LRFD
Bridge Design Specifications,” Washington, DC.
Plunkett, J. D. (1997). “Fiber-Reinforced Polymer Honeycomb Short Span Bridge for Rapid Installation,”
IDEA Project Final Report, Contract NCHRP-96-IDO30, IDEA Program, Transportation Research
Board, National Research Council.
Oghumu, S. and Cai, C. S., and Zhang, Y. (2005). “Finite Element Modeling and Performance Evaluation for the
Development of FRP Bridge Panels.” The joint /ASME/ASCE/SES Engineering Mechanics and Materials
Conference, June 2005, Baton Rouge, Louisiana.
Wang TL, Huang DZ, Shahaway M (1992). “Dynamic response of multigirder bridges.” Journal of Structural
Engineering, ASCE, Vol. 118, No.8, pp 2222-38.
Zhang, Y., Cai, C. S., Shi, X. M. and Wang, C. (2006) “Vehicle Induced Dynamic Performance Of a FRP Versus
Concrete Slab Bridge” J. of Bridge Engineering, ASCE, 11(4), 410-419.

254
Third International Conference on FRP Composites in Civil Engineering (CICE 2006)
December 13-15 2006, Miami, Florida, USA

LOAD AND RESISTANCE FACTOR DESIGN FOR FRP

STRENGTHENING OF CONCRETE STRUCTURES
Rebecca A. Atadero
(Graduate Research Assistant, University of California, San Diego, La Jolla, CA, USA)

Vistasp M. Karbhari
(Professor, University of California, San Diego, La Jolla, CA, USA)

ABSTRACT
Externally-bonded fiber reinforced polymer (FRP) composites are a maturing technology for the repair and retrofit
of existing concrete structures. Current guidelines for the design of FRP strengthening measures do not explicitly
consider the uncertainties present in the FRP properties. Load and Resistance Factor Design (LRFD) provides an
ideal framework for these considerations and is compatible with existing trends in civil engineering design codes. A
recent project has studied the application of LRFD to FRP strengthening schemes. A widely applicable design
framework has been proposed using a composite specific resistance factor to consider material variability, and a set
of application factors to consider specifics of field manufacture. This paper describes the calibration of resistance
factors for the example case of flexural strengthening of T-beam reinforced concrete bridge girders. The calibration
considers FRP degradation based on existing durability models and continued degradation of the original structure is
modeled by considering general corrosion of the reinforcing steel. Representative girders were selected from a
sample of California bridge plans. The reliability was evaluated using simulation and first-order reliability methods.

KEYWORDS
Material Variability, LRFD, Design of Strengthening, Resistance Factors

1. INTRODUCTION
Externally-bonded fiber reinforced polymer (FRP) composites are an increasingly adopted technology for the repair
and retrofit of existing concrete structures. In order to encourage the further use of these materials a design code is
needed that considers the inherent material variability of the composite, as well as the variations introduced during
field manufacture and environmental exposure while in service. Load and Resistance Factor Design (LRFD)
provides an ideal framework for these considerations and is compatible with existing trends in civil engineering
design codes.

Research into the application of LRFD to FRP strengthening is ongoing. Previous work has assessed the statistical
variation in wet layup composite properties (Atadero et al., 2005) and proposed a framework for the application of
reliability-based design to FRP strengthening (Atadero and Karbhari, 2005). This framework is based on design
values for composite properties that are meant to represent best estimates of the mean values of FRP properties in
the field. These design values are found using the mean laboratory predicted values of strength and modulus as the
characteristic values, with a set of Application Factors to account for the specifics of field manufacture. A
composite specific resistance factor is proposed to account for the variation in the composite properties. This
composite specific factor is a function of the COV of the controlling composite property for a particular limit state,
and facilitates the application of the proposed procedure to a variety of composite systems displaying different levels
of variation. The proposed design framework also makes use of an environmental factor that is specific to the
exposure environment and anticipated service life of the strengthening. This paper describes the process used to
calibrate preliminary resistance factors for the specific example case of flexural strengthening of reinforced concrete
T-beam bridge girders. Complete details may be found in Atadero (2006).

255
2. CALIBRATION PROCEDURE
The resistance factors were calibrated using a procedure that assumes the load factors have already been determined.
For this example the load factors from the AASHTO Manual for Condition Evaluation and Load and Resistance
Factor Rating (LRFR) of Highway Bridges (2003) were used because the reliabilities found using the AASHTO
LRFD (1998) factors were very high. This procedure can be described as a trial and error procedure because
designs are first created with many values of the resistance factors, the reliability of all designs is evaluated, and
then the pair of factors that results in the smallest difference between the target reliability and the as designed
reliability is selected. The procedure is described in further detail in Atadero (2006).

3. RANGE OF CALIBRATION
3.1 Composite Materials

Five different sample materials with properties chosen to represent the anticipated range of properties seen in wet
layup CFRP were used for calibration. The assumed properties of these materials are shown in Table 1.

Table 1: Assumed Properties of Sample Materials Used for Calibration

Ultimate Strength Modulus 1-Layer Thickness Ultimate Strain

Material
MPa (ksi) GPa (ksi) mm (in) mm/mm (in/in)
1 620.5 (90) 51.7 (7500) 1.27 (0.05) 0.012
2 689.5 (100) 61.4 (8900) 1.27 (0.05) 0.011
3 758.4 (110) 58.6 (8500) 1.27 (0.05) 0.013
4 827.4 (120) 59.3 (8600) 1.27 (0.05) 0.014
5 896.3 (130) 68.9 (10000) 1.27 (0.05) 0.013

Time dependent degradation of the composite was modeled using Arrhenius rate relations found in Abanilla (2005).
Three conditions were modeled, no degradation, degradation following the model, and degradation following the
model, but slowed by a factor of five to better represent field environments.

3.2 Existing Structure

Representative T-beam girders were chosen from an inventory of California bridge plans. These girders were
assumed to be deficient due to a loss in reinforcing steel. Three cases were considered, 10, 20 and 30 percent loss.
Designs were created to return the girders to the flexural capacity required by the LRFR load factors. The
possibility of continuing deterioration of the girder was considered by modeling additional steel loss through a
simple corrosion model. Continuing corrosion was considered for each of the three original percentages of loss,
resulting in a total of six conditions, three without continuing corrosion, and three with. Design lives of 10, 20, 30,
40, and 50 years were considered.

4. DESIGN OF STRENGTHENING
Trial designs were created for each of the different cases within the extensive range of calibration. The checking
equation used for design is shown in Equation 1. In this equation γi and Li are the load factors and load effects,
respectively, and R is the nominal resistance as a function of many variables including the FRP contribution, xFRP.
The general resistance factor, φ, was allowed to take on values of 0.95, 0.9, and 0.85. The composite specific
factor,ψ, ranged from 0.95-0.5 in increments of 0.05.

∑γ Q
i
i i ≤ φR(...,ψx FRP ) (1)
The design loads were calculated based on the HL-93 load model (AASHTO 1998), using a bridge load analysis
program (WSDOT, 2006). The moment resistance of the girders was calculated based on sectional analysis,
assuming a linear strain distribution through the depth of the cross-section. The ultimate strength was found when

256
either the FRP or concrete reached an ultimate strain. The strain limit in the FRP considered the possibility of
debonding or composite rupture, and was found to control due to the large area of concrete in the compressive
region provided by the T-beams.

5. EVALUATION OF RELIABILITY
The reliability of the trial designs was evaluated using a hybrid procedure whereby the variation in resistance was
estimated via Monte Carlo Simulation, and the overall reliability was calculated using first order reliability methods
as outlined in Nowak (1999). In this procedure the resistance was modeled as a Lognormal variable with the mean
and standard deviation determined through simulation based on a model of resistance (sectional analysis was used in
this example) and the individual distributions of variables contributing to resistance. In order to calibrate values of
ψ, the composite specific factor, that depend on the COV of the composite, the reliability of each design was
evaluated for composite strength COVs of 0.05, 0.10, 0.15, 0.20, 0.25 and 0.30. The components of load were
modeled as normal variables following the distributions developed in Nowak (1999). This procedure follows the
assumptions used in calibration of the LRFD specifications for design of new bridges, however some of these
assumptions (particularly modeling variation in live load) are identified as needing further study.

6. RESULTS AND DISCUSSION

Resistance factors were calibrated for target reliabilities of 2.5, 3.0, and 3.5 for each of the girders and corrosion
conditions considered. These reliability indices span the range from the value used for bridge evaluation to that used
for design of new bridges. A selection of reliability targets was used to demonstrate the range of possible factors
that would be necessary, however it is expected that a code-writing agency would specify this value for the designer.

The calibrated resistance factors were found to be very sensitive to the amount of steel remaining in the section
relative to the amount of FRP applied. Due to the range of initial conditions displayed by the girders, comparisons
between girders based on the corrosion condition were uninformative. Instead, a common reference value was
defined as the amount of steel required to just meet the LRFR load demand for each girder. Different girders at
different corrosion conditions were grouped based on the percent of steel loss relative to this reference value.
Within each grouping, a range of resistance factors was selected to meet each of the three target reliability indices.
Relative losses from 0-30% were considered, with the groups based on 5% increments. As an example, Table 2
shows the factors calibrated for two of the ranges considered. The values of ψ shown in Table 2 are the average of
the higher (corresponding to low strength COVs) and lower (corresponding to high strength COVs) values of the
range found for each girder and level of corrosion in a particular group. It is important to recognize that this was a
strict average, and therefore it can be expected that in using these factors roughly half of designs will fall below the
target reliability and roughly half will be above it. The COVs listed in the table refer to the amount of variation seen
in the value of the factors, and are provided to give some sense of the amount of variation within a set of factors for
a single group, however are not necessary for use in design.

Table 2 shows that as the percent of steel loss relative to the LRFR baseline increases the calibrated resistance
factors also increase. This implies that for a fixed value of the resistance factors, as the amount of load carried by
the FRP relative to the steel increases, the reliability of the girder increases. From a reliability standpoint, this is
attributed to the ability of the steel and FRP to offset weaknesses in each other as more of the load is shared between
the two. The change in reliability due to different amounts of steel deficiency poses a difficulty for reliability-based
design of strengthening because in real applications it is very rare that the area of remaining steel will be known
with a high degree of certainty. Thankfully, conservative designs for high amounts of steel loss can be created using
the factors for lower percentages of loss. Table 2 also shows a range of values for ψ. This demonstrates that
changes in FRP variability do have the ability to impact the reliability of the girder as a whole, thereby requiring
different resistance factors for different levels of variation. The range of COVs considered in this example was quite
large, 0.05-0.30, however, it was not impractical; values from 0.12 to 0.23 have been observed by these authors for
wet layup samples. Though Table 2 only expresses the range of values, the change in the value of ψ with changes
in the strength COV was quite uniform across the different cases considered here. The higher value of the resistance
factor generally applies to COVs from 0.05-0.15. Beyond a COV of 0.15, the resistance factor decreases linearly to
achieve the low value at a strength COV of 0.30. Tests were also conducted on the effect of changes in the COV of
the composite modulus, but they were observed to be much smaller than the effect of changes in the strength COV.

257
 Table 2 : Example of Resistance Factors for Different Target Reliabilities and
Different Amounts of Relative Steel Loss

% below steel
β=2.5 β=3.0 β=3.5
needed for LRFR
18 out of 38 cases are
too high or no design φ=0.9 φ=0.85
5< , >=10 φ=0.95 Avg ψ = 0.831-0.752 Avg ψ = 0.834-0.787
Avg ψ = 0.839-0.732 COV = 11.6% &12.7% COV = 9.3% & 9.1%
COV = 10.0% & 13.0%
17 out of 45 cases are too low
For 15 out of 45 cases or no design possible. The rest
21 out of 45 cases are
φ=0.95 or 0.90 both are evenly split.
too high or no design
work. φ=0.85
10< , >=15 Avg ψ = 0.885-0.840
φ=0.95
φ=0.9 COV = 7.5% & 7.6%
Avg ψ = 0.870-0.782
Avg ψ = 0.864-0.819 φ=0.90
COV = 9.1% & 9.6%
COV = 9.7% & 9.9% Avg ψ = 0.662-0.626
COV = 6.9% & 8.3%

Though the remaining area of steel was found to be highly significant, other trends were also observed. In general,
only small differences in reliability were present for designs created with different FRP materials. The predicted
degradation of the FRP for different design lives was included in the creation of designs, and it was found that there
was very little difference in reliability for the different degradation models. Furthermore, designs maintained
uniform reliability over different design lives for the cases with no continuing corrosion. For the cases that did
consider continuing corrosion, the reliability increased slightly over time, suggesting that the corrosion parameters
used in design were slightly more conservative than those used in analysis of reliability.

7. CONCLUSIONS
This paper has described preliminary calibration procedures allowing for the reliability-based design of FRP
strengthening with wet layup materials. Results shown here highlight the importance of knowledge regarding the
state of the existing structure when the FRP is applied and over time. They also show that the amount of variability
in composite properties can impact the reliability of the repaired structure, and should be considered in design.

8. REFERENCES

Abanilla, M.A.D. (2004). “Physico-Mechanical Characterization of T700 based Carbon/Epoxy Stystems for
Infrastructure Rehabilitation”, Master’s Thesis, University of California, San Diego.
American Association of State Highway Transportation Officials. (1998). AASHTO LRFD Bridge Design
Specifications, Customary U.S. Units, 2nd edition; AASHTO: Washington D.C.
American Association of State Highway and Transportation Officials. (2003). Manual for Condition Evaluation and
Load and Resistance Factor Rating (LRFR) of Highway Bridges; AASHTO: Washington D.C.
Atadero, R., Lee, L., and Karbhari, V.M. (2005). “Consideration of Material Variability in Reliability Analysis of
FRP Strengthened Bridge Decks”, Composite Structures, Vol. 70, No. 4, pp 430-443.
Atadero, R.A., and Karbhari, V.M. (2005). “Consideration of Time-Dependent Degradation in the Development of
Probabilistic Based Design of FRP Strengthening”, American Society for Composites, Proceedings of 20th Technical
Conference, Editors: F.K. Ko, G.R. Palmese, Y. Gogotsi and A.S.D. Wang, Philadelphia, PA.
Atadero, R.A. (2006). “Development of Load and Resistance Factor Design for FRP Strengthening of Reinforced
Concrete Structures ”, Ph.D Thesis, University of California, San Diego.
Nowak, A.S. (1999) Calibration of LRFD Bridge Design Code; National Academy Press: Washington D.C.
Washington State Department of Transportation. Bridge Engineering Software - QConBridge™ Overview.
http://www.wsdot.wa.gov/eesc/bridge/software/index.cfm?fuseaction=software_detail&software_id=48 (accessed
2006).

258
Third International Conference on FRP Composites in Civil Engineering (CICE 2006)
December 13-15 2006, Miami, Florida, USA

CZECH GUIDELINES FOR STRENGTHENING OF CONCRETE

BRIDGES BY COMPOSITES
Miroslav Černý
(Head of Centre for Composites, Klokner Institute, Czech Technical University, Prague, Czech Republic)

ABSTRACT
This paper presents the new Guidelines which have been recently elaborated in the Czech Republic. The Guidelines
are focused on strengthening of concrete road bridges by fibre composite strips, mostly based on carbon fibres and
epoxy matrices. Methods of strengthening are following: bonding (gluing) of composite strips on concrete surface or
impregnating of unidirectional or multidirectional textile by polymeric resin in-situ.

KEYWORDS
Strengthening, Concrete Bridges, Composites, Carbon Fibres, Limit States

1. INTRODUCTION
Many suppliers of carbon fibre sheets are delivering the products, which can be successfully applied for
strengthening of RC road bridges. However, in the Czech Republic doesn’t exist an unified approach for application
and design for strengthening of structures by fibre composites. Since 2001 in the Czech Republic at Czech Technical
University (leading by author) has been conducting the work on „Technical Conditions (Guidelines) for
Strengthening of Concrete Bridges by Composites“, which has been supporting by Ministry of Transportation of the
Czech Republic.
„Technical Conditions“ are oriented to preparation, design and execution of strengthening of road bridges and
footbridges from concrete and reinforced concrete by unprestressed composites and will serve as groundwork for
certification of products for strengthening.
Recently, the „Technical Conditions“ for strengthening of bridges and footbridges from concrete and reinforced
concrete on roads by prestressed composites are elaborated.
Methodology of design is issuing from the following basic sources:
EN 1991-1 Eurocode 1:Basis of design and actions on structures- Part 1: Basis of design.
EN 1991-3 Eurocode 1:Basis of design and actions on structures- Part 3: Actions on structures: Bridges.
EN 1992-1-1 Eurocode 2:Design of concrete structures- Part 1-1: General rules and rules for buildings
EN 1992-2 Eurocode 2:Design of concrete structures- Part 2: Concrete bridges
EN 206-1: Concrete: Specification, performance, production and conformity
ACI 440.2R-02 (2003) Guide for the Design and Construction of Externally Bonded FRP Systems for Strengthening
Concrete Structures
CEB-FIP Technical report (2001): Externally bonded FRP reinforcement for RC structures. Bulletin No. 14

The basic principles involved in Guidelines are presented:

a) Properties of strengthening elements
b) Structural requirements, execution and quality control
c) Recommendations for design

2. PROPERTIES OF STRENGTHENING ELEMENTS

Adhesives, sheets, textile composites and other strengthening elements should have the properties evaluated by the
standard test methods. The properties have to be given by the manufacturer or supplier, who should prove that the

259
tests have been done by the independent testing laboratory according to approved testing methods. The manufacturer
should guarantee the quality of products. The control tests of products must be done before the application. The
Guidelines include the required properties of products for strengthening of road bridges. The short-term mechanical
properties of adhesives have to be tested (tensile modulus, tensile strength, fracture strain according to EN ISO 527,
bending modulus according to EN ISO 178). Further the short-term mechanical properties of composite sheets
(tensile modulus, tensile strength, fracture strain according to EN ISO 527) should be tested. Table 1 shows
required properties of epoxy adhesive for gluing of carbon fibre sheets.

Table 1: Properties of Epoxy Adhesive (Cerny, 2004)

Properties (at 200C) Epoxy Adhesive

Density[kg.m-3] 1100- 1700
Tensile E modulus [MPa] 5000
Shear modulus [MPa] 2500
Poisson’s ratio[-] 0.2- 0.3
Tensile strength [MPa] 25
Shear strength [MPa] 10- 30
Fracture strain in tension [%] 0.3 – 1
Fracture energy [kJm-2] 0.2- 1.00
Coefficient of thermal expansion [10-6K-1] 44- 70
Water absorption:7 days 250C [%w/w] 0.04-0.1

3. STRUCTURAL REQUIREMENTS
Before diagnostic recognition the ordinary or extraordinary inspection of bridge according to standards (Czech
Standard 736221) should be done. Further recognition follows according to Technical Conditions (TP 72). It should
be recognized: an amount, geometry and type of steel reinforcement including corrosion and the quality of concrete
including corrosion. The quality of bonding should be tested according to EN 1542, and by a shear test (test
proposed by CTU KI).

4. RECOMMENDATIONS FOR DESIGN

Limit states and design situations (Cerny, 2001)
The design model should consist of a verification of both the serviceability limit state and the ultimate limit state.
The following design situations have to be considered:
• persistent situation, corresponding to the ordinary use of the structure
• accidental situation, corresponding to a loss of strengthening
• special situation (important for bridges: stresses due to differences in thermal expansion, cyclic loading, long-
term behaviour)

Verification of the ultimate limit state (ULS)

The various failure modes should be considered, assuming (1) full composite action between structural member and
strengthening (bonding) and (2) different debonding that may occur.
Load combinations and partial safety factors should be applied. In case of full bonding, for the concrete, a
parabolic-rectangular diagram, for the steel a bilinear stress-strain relationship can be considered.
For composites the following stress-strain relationship is considered:
a) linear for carbon fibre sheets (in tension along fibres)
b) bilinear for textile composites in warp and fill directions
c) nonlinear for textile composites in shear

Verification of the serviceability limit state (SLS)

It should be verified that the strengthened structural member behaves adequately in normal use.

260
The verification involves:
• stresses, these should be limited to prevent steel yielding, damage, excessive creep of concrete and composite
• deformations which may limit normal use of the structure
• cracking which may limit the durability of the structure.
Linear stress-strain diagram is considered for concrete and steel.
For composites the following stress-strain relationship is considered:
a) linear for carbon fibre sheets (in tension along fibres)
b) bilinear for textile composites in warp and fill directions
c) nonlinear for textile composites in shear

The following figures illustrate a procedure for evaluation of ULS for rectangular beam in bending, strengthened by
a carbon sheet with linear stress-strain relationship (consult CEB-FIP TR, Cerny, 2002)):

Figure 1: Initial configuration

From the initial configuration an initial strain ε0 at composite level is calculated and used in ULS analysis.

Figure 2: Ultimate limit state in bending

The Figure 2 shows a full composite action when steel yields (fyd) and concrete is crushing(αfcd).

261
Failure occurs in the critical cross-section at strain limit of steel and failed concrete, composite is not damaged. The
design bending moment is calculated on condition of initial strain loaded structure from the method of strain limit by
iteration. Some results have been confirmed experimentally (Cerny, 2006).
The methodology of design for strengthening is elaborated in Guidelines for bending and shear of a rectangular
beam and T beam, some examples are presented in Appendix.

5. CONCLUSIONS
Guidelines “Technical Conditions” allow to enforce unified approach for strengthening, evaluation of materials and
design strengthening methods for road RC bridges in the Czech Republic and to achieve more safe and effective
strengthening of RC bridges.

6. REFERENCES
Cerny, M. (2002). “Limit States Design for Structures Externally Reinforced by Composites”, Proceedings of
International Symposium on Mechanics of Composites (MC/1), Editor: M.Cerny, Czech Technical University,
Prague, pp.43-46.
Cerny, M. (2001). “Limit States Design for Composite Structures”, Proceedings of the International Conference on
Composites in Material and Structural Engineering (CMSE/1), Editor: M.Cerny, Czech Technical University,
Prague, pp.57-58.
Cerny, M. (2004). „Experimental Evaluation of Carbon Fibre Composites and Adhesives for Strengthening of
Engineering Structures“, 3rd Czech /Slovak Symposium „Theoretical and Experimental Research in Structural
Engineering“, CTU Reports, No. 3.
Cerny, M. (2006). „Testing of RC Beams Strengthened by CFRP Strips “, 4th Czech /Slovak Symposium
„Theoretical and Experimental Research in Structural Engineering“, CTU Reports, No. 3.
ACI 440.2R-02 (2003) Guide for the Design and Construction of Externally Bonded FRP Systems for Strengthening
Concrete Structures, ISBN 0-87031-0887
CEB-FIP Technical report (2001): Externally bonded FRP reinforcement for RC structures. Bulletin No. 14, ISBN
2-88394-054-1

262
Part VIII. Durability Issues
Third International Conference on FRP Composites in Civil Engineering (CICE 2006)
December 13-15 2006, Miami, Florida, USA

DETERIORATION MECHANISM OF BOND BETWEEN CFRP PLATE

AND CONCRETE IN MOIST ENVIRONMENT
Zhenyu Ouyang
(Ph.D. Student, Marquette Universit, Milwaukee, WI, USA)

Baolin Wan
(Assistant Professor, Marquette Universit, Milwaukee, WI, USA)

ABSTRACT
This study experimentally investigated the mechanism of deterioration for the bond between CFRP plate and
concrete due to moisture attack. The debonding failure in the concrete structures with externally bonded FRP has
two major modes: cohesive failure in the concrete surface layer and adhesive failure in the interface between
concrete and adhesive layer. A simple experimental approach was developed in this study to directly measure the
moisture condition in the interface between concrete and adhesive layer with function of immersion duration in
water. Modified Double Cantilever Beam (MDCB) test was used to measure the interfacial fracture energy for the
CFRP plate debonding from concrete substrate under Mode I loading. The MDCB specimens were submerged in
water for different periods before they were tested. When the value of interface region relative humidity (IRRH)
increased, fracture load and fracture energy decreased and failure mode was also changed from cohesive failure to
adhesive failure. It was found that the residual thickness of concrete (RTC) on the fracture surface was directly
related to the relative humidity in the interface region. When the IRRH was greater than 55%, the value of fracture
energy Gf began to decrease significantly with the increase of the IRRH and immersion duration.

KEYWORDS
Bond, CFRP, deterioration, interface region relative humidity (IRRH), residual thickness of concrete (RTC)

1. INTRODUCTION
When reinforced concrete are strengthened with externally bonded fiber reinforced polymer (FRP) composite
materials, bond between FRP and concrete for the strengthening system is one of the important factors affecting the
structural performance. The debonding failure in the concrete structures with externally bonded FRP has two major
modes: cohesive failure in the concrete surface layer and adhesive failure in the interface between concrete and
adhesive layer. Normally, it requires larger load and fracture energy to cause concrete cohesive failure than those to
cause interface adhesive failure. In ambient environment conditions and with good quality control during
application, the performance of the bond between FRP and concrete is good by the evidence that debonding failure
happens in concrete. However, when it is exposed to moist environment, the FRP/concrete bond may deteriorate
quickly and the failure mode changes from concrete cohesive failure to interface adhesive failure (Wan et al., 2006).
It could lose as high as 87% of the gain from CFRP when the CFRP strengthened reinforced concrete beams were
exposed to moist environment (Grace and Singh, 2005). Even for very short period when the CFRP bonded
concrete specimens were submerged in water, the interfacial energy release rate was reduced 85% after 8 weeks of
exposure (Wan et. al., 2006). Nguyen et al. (1998) found that there were few water molecule layers at the interface
when concrete with epoxy cover was exposed to water for a period. Such water molecule layers seriously decrease
the free surface energy of bond interface and create a weakest link in the system. They also change the interface
bond stiffness and gradually transfer the debonding failure modes from cohesive failure to adhesive failure.
Absorbing water can change the mechanical properties of the adhesive layer. Harmon et al. (2003) found that shear
modulus of the bond layer was one of the critical factors which affected the bond performance. Therefore, the
changes of bond strength and fracture energy are related to the quantity and distribution of water at the bond
interface region.

263
Almost all of literatures relate the bond deterioration to the moisture condition of surrounding environment and
exposure duration. However, even in same moist environment and exposure duration, the actual moist environment
in the interface region might be significant different due to different sizes and geometric properties of the specimens,
moisture distribution and transportation in concrete. In this research, a simple method was developed to measure the
adhesive/concrete interface region relative humidity (IRRH). The relationship between moisture exposure history of
the interface and interfacial energy release rate for debonding FRP from concrete was determined. The IRRH value
has the potential to be a direct and uniform criterion for the deterioration of the bond between concrete and FRP
composites in moist environment.

2. EXPERIMENTAL PROGRAMS
2.1 MDCB Test

The Modified Double Cantilever Beam (MDCB) test was successfully used by Wan et al. (2006) to investigate the
water effect on the bond between FRP and concrete. The frame used in this research is the second generation of
MDCD test frame as shown in Figure 1. Using this new frame, the loading rate and magnitude can be accurately
controlled by a programmable motor. A customized data acquisition system was designed specifically for the
MDCB test and was built in the test frame. Therefore, the data can be automatically acquired by connecting this
MDCB test frame to any computer.

A ready mixed concrete and commercially available CFRP plate and adhesive were used in this research. The
dimension of the concrete specimen was 76 x 76 x 191 mm. The surface of concrete was sand blasted and cleaned
by high pressure air before primer was applied. CFRP plate was applied on the concrete surface by standard
procedures recommended by the manufacture. The control specimens were tested after they were allowed to cure
for 7 days. All other specimens were submerged in distilled water until they were tested. All specimens were tested
under Mode I (peeling) load.

Programmable
Motor
Load
Cell
Specimen
Cable

Control Laptop
Panel

Figure 1: Second generation Figure 2: Measure the Interface Figure 3: Measure the residual
of MDCB test frame region relative humidity (IRRH) thickness of concrete (RTC)

2.2 Measure the Interface Region Relative Humidity (IRRH)

In order to minimize the disturbance to the moisture content in the concrete at the interface during the test, a plastic
tape was used to gradually and carefully cover the debonded area once the peel separation became large enough.
After CFRP plate was peeled off from concrete substrate, the fracture surface of concrete was entirely covered by a
plastic film immediately. This plastic acted as the original adhesive and CFRP plate to prevent the moisture
evaporates from concrete to air. The moisture present in the concrete would evaporate into the space between
concrete fracture surface and plastic film until it reached balance. This balanced relative humidity can be
approximately considered as the relative humidity in the FRP/Concrete interface region before test. The relative
humidity at different locations was measured by an accurate digital hygrometer which had ±3% error according to
manufacturer’s manual. A short aluminum tube was used as cap to create a small closed space to measure the
relative humidity on the fracture surface. The top of the cap was sealed with a rubber washer. A layer of 3-mm-
thick water-proof silicone was glued on the bottom wall of the cap. The fully cured silicone was very soft and it
could deform with the rough concrete surface when subjected to moderate pressure. Therefore, a closed space was
created by the aluminum tube, silicone layer and concrete surface. There was no air convection inside this closed

264
space and between the space and surrounding air. When the probe was inserted into the cap through the rubber
washer, the moisture sensor which located at the end of the probe was placed in this closed and small space as
shown in Figure 2. Once the relative humidity reached equilibrium in this small closed space, this R.H. value was
recorded as the IRRH.

2.3 Measure the Average Residual Thickness of Concrete (RTC)

Before bonding to the concrete specimens, the thickness of each CFRP plate was measured. After MDCB test
finished, the detached plates were measured through a digital coordinate measuring machine (CMM) as shown in
Figure 3. For each specimen, about 100 sample points along the entire fracture surface were measured and the
average value was calculated. This value was the summation of the thicknesses of CFRP plate, adhesive and
residual concrete layer attached on CFRP plate. Each plate was then cut into three coupons along the longitudinal
direction of the plate. A high resolution digital microscope was used to measure the adhesive thickness along each
edge of these three coupons with 5 mm interval along the longitudinal direction. The adhesive thickness was the
average value of those measurements. The difference between the total average thickness from CMM and the sum of
CFRP thickness and adhesive thickness was taken as the average residual thickness of concrete (RTC).

3. TEST RESULTS AND DISCUSSION

Energy release rate of each specimen was calculated by following the method developed by Wan et al. (2004) for
MDCB test. It was observed in the tests that debonding failure mode changed gradually from cohesive failure in
concrete to adhesive failure in interface when immersion duration increased from 2 to 7 weeks. The fracture energy
release also decreased with the immersion duration as shown in Figure 4. After the specimens were submerged in
water for 7 weeks, the average Gf value was only 22% of the control specimens.

550 2.5
500
450 2
RTC (mm)

400
Gf (J/m2)

350 1.5
300
250
1
200
150
100 0.5
50
0 0
0 10 20 30 40 50 45% 55% 65% 75% 85%
Immersion duration (Days) IRRH

Figure 4: Average fracture energy Figure 5: Residual concrete thickness (RTC)

release rate vs. immersion duration vs. interface region relative humidity (IRRH)

600 600

500 500

400
Gf (J/m2)

400
Gf (J/m2)

300 300

200
200
100
100
0
0
45% 55% 65% 75% 85%
0 0.5 1 1.5 2 2.5
RTC (mm) IRRH

Figure 6: Gf vs. RTC Figure 7: Gf vs. IRRH

265
It is shown in Figure 5 that the residual thickness of concrete (RTC) on the fracture surface decreased significantly
with the increase of the interface region relative humidity (IRRH). When the IRRH was between 45% and 55%
which was similar to normal R.H. value in room, the RTC was around 2 mm. When the IRRH was close to 75%,
the average RTC was less than 0.5 mm. It is shown in Figure 6 that fracture energy increased with the increase of
RTC. The bond fracture energy Gf increased with the increase of RTC when it was less than 2 mm. However, Gf did
not change significantly when RTC was greater than 2 mm. The relation between the fracture energy Gf and the
IRRH is shown in Figure 7. It is shown that Gf deceased with the increase of IRRH. Such decrease was relatively
slow when the IRRH was lower than 55%. When the IRRH value was higher than 55%, the Gf decreased rapidly
with the increase of IRRH value and immersion duration. Therefore, the value of IRRH of 55% was the critical
value to start the bond deterioration. For the specimens used in this research, it took three weeks to reach this
critical humidity in the interface region. If the specimen size was larger or the relative humidity in the environment
was lower, it would take longer time to let the interface region to obtain the critical relative humidity value.

When metal, glass or other non-porous materials are glued together with adhesive, the adhesive can only penetrate
into a tiny depth and such penetration depth is relatively uniform. However, concrete is a porous material and there
are numerous air cavities, capillary pores, gel pores and micro-cracks on the surface. Primer is a low viscosity
material and it is easy to penetrate into deeper location in concrete. The contact area between the primer and the
walls of the cavities, pores and micro-cracks will significantly increase the bonding area between primer and
concrete. The actual bonding area is much larger than the nominal bonding area. Therefore, the bond between
primer and the walls of the penetrating path will have significant contribution on the resistance to debond during the
peeling test. When specimens are submerged in water, water will penetrate into concrete through the part below
primer-saturated concrete much easier and faster than into primer-saturated concrete, adhesive and FRP because of
the microcracks and capillary pores in concrete. Therefore, the bottom part of the primer-saturated concrete will
reach critical R.H. value for debonding and lose its bonding ability first. After local debonding between pore wall
and penetrated primer, the concrete at that part is not gripped during peel test and it will not contribute to the energy
dissipation during the global debonding process. Water will continue to penetrate up through the primer-saturated
concrete and finally reach the nominal interface between adhesive and concrete. This might be the reason why the
RTC and Gf values gradually decrease with the increase of immersion duration as observed in experimental tests.

4. CONCLUSIONS
The following conclusions can be drawn from this research:
1. A simple method was successfully developed to measure the adhesive/concrete interface region relative
humidity (IRRH) value. It could be used as a direct and uniform criterion for the deterioration of the bond
between concrete and FRP composites in moist environment.
2. The residual thickness of concrete (RTC) on the fracture surface was directly related to the IRRH values
3. When the IRRH value was greater than 55%, the Gf value began to decrease significantly with the increase
of IRRH and immersion duration.

5. REFERENCES
Grace, N.F. and Singh, S.B. (2005). “Durability evaluation of carbon fiber-reinforced polymer strengthened concrete
beams: experimental study and design”, ACI Structural Journal, Vol. 102, No.1, pp40-51.
Harmon, T.G., Kim, Y.J., Kardos, J., Johnson, T. and Stark, A. (2003)., “Bond of Surface-Mounted Fiber-
Reinforced Polymer Reinforcement for Concrete Structures”, ACI Structural Journal, Vol. 100, No. 5, pp557-564.
Nguyen, T., Byrd, W. E., Alsheh, D., Aouadi, K., Chin, J. W. (1998). “Water at the polymer/substrate interface and
its role in the durability of polymer/glass fiber composites”, Durability of Fibre Reinforced Polymer (FRP)
Composites for Construction (CDCC'98), 1st International Conference, Canada, pp451-462.
Wan, B., Petrou, M.F. and Harries, K.A. (2006). “Effect of the presence of water on the durability of bond between
CFRP and concrete”, Journal of Reinforced Plastics and Composite, Vol. 25, No. 8, pp875-890.
Wan, B., Sutton, M., Petrou, M.F., Harries, K.A. and Li, N. (2004). “Investigation of Bond between FRP and
Concrete Undergoing Global Mode I/II Loading,” ASCE Journal of Engineering Mechanics, Vol. 130, December
2004, pp1467-1475.

266
Third International Conference on FRP Composites in Civil Engineering (CICE 2006)
December 13-15 2006, Miami, Florida, USA

Durability Approach for GFRP-Rebars

André Weber
(R &D, Schoeck Bauteile GmbH, Baden-Baden, Germany)

ABSTRACT
Most GFRP-Rebars have been developed, to solve the corrosion problem with bridges and other structures. But one
component of these rebars, the glass fibre, is itself susceptible to corrosion in the highly alkaline environment of
concrete. The other component of the GFRP-rebar - the resin system - has to provide the protection against diffusion
and deterioration.

To begin applications in the past years different accelerated durability tests have been developed and performed.
Design and test guidelines as well as first codes have been established in different countries. Now an experience of
nearly ten years shows that some rebar systems are stable under certain conditions. In spite of all the benefits the
market growth for primary applications is smaller than expected.

There is still a lack of confidence in the new reinforcement technique as a primary reinforcement for buildings and
structures. If we do not want to wait 50 years, confidence can only be established by conservative and
comprehensible test approaches with broad acceptance in science and by the authorities.

Two different approaches for durability are reviewed from the point of view of safety and the transferability of the
results. Basis of the consideration is the international semi-probabilistic safety concept with partial safety factors for
load and material. For both approaches results are presented and compared. Furthermore, a defined path from test
results to design value of the tensile strength is shown in this paper.

KEYWORDS
Durability, Composite Rebar, Safety Concept, Design Value, Accelerated Durability Test, Extrapolation

1. INTRODUCTION
New materials have acceptance problems, especially if the material is susceptible to attack from its own
environment. Long term experience is very limited. Even the significance of 10 years of long term data is low. To
get durability results in shorter time two approaches are possibly: Time temperature shifting and extrapolation.

2. SIMPLIFIED APPROACH WITH RESIDUAL STRENGTH

In this first approach the whole service life is projected on the rebar. That means load, time, temperature and
alkalinity conditions. To accelerate this test, a higher temperature is chosen for the test. Different authors have
developed time/temperature shifting equations. Normally tests are performed over several months at 60°C. The
environment can be an artificial concrete pore solution as well as wet concrete. After this test regime the specimen
are tested by tensile tests. ACI, ISO, JSCE and CSA have defined different standard test procedures.

Normally the residual strength is reported in relation to the virgin strength. More important regarding the safety
theory is the relationship between the residual strength or the load during the test and the design load in the
application. This relationship is proposed as a simplified approach for normal conditions by the fib.
But attention: if the conditions change or the testing time is too short an extrapolation is not possible. If, for
example, a specimen is loaded for 1 month and the residual strength is 95% and the specimen is loaded for two
months and the residual strength is 90% it is not possible to give a value for 3 months or 6 months. A linear

267
behaviour can not implied. As a specimen deteriorates the load in the residual section rises with every broken fibre,
that is the stress in the fibres of the remaining cross section increases. As a result, the deterioration process
accelerates. The behaviour is highly nonlinear. There is a certain probability, for example, that the above mentioned
specimen is broken after three months.

The approach described above has been selected for the certification process of the Dutch approval authority KIWA.
A less clearly specified approach is found in the non-binding section of the Canadian Standard CSA 806. In Annex
O the code recommends that the specimens should be loaded with 1.1 times the “design allowable strength” while
ageing in an artificial concrete pore solution of pH 12.7 or in concrete.

To meet the internationally accepted safety concept the Canadian approach specifies the test load to be equivalent to
a multiple of the design load, that means the specimen is stressed over the testing time with factored loads. The
testing time is specified to be 2000h at a temperature of 60°C. For the concrete a high alkaline cement (Na2O =1%)
is used with a water cement ratio higher than 0.45, leading to the highest possible ph value. During the test the
concrete is water saturated and cracked, leading to realistic additional bond stresses.

Under the assumption that these conditions are representative of a lifecycle, the conclusion is that the rebar can
withstand the design load for the entire service life. After the ageing process the rebar is tested for residual strength.
The characteristic value of this tensile strength divided by the material factor has to be higher than the design load.
As an example, the results presented in Figure 1 lead to a design value of 350 N/mm² [Weber 04].

long term test 2000h, 60°C in artificial concrete pore solution ph 13,7,
Schöck Combar d=16mm

1300
1200
1100 new
1000 mean value and
Residual strength in N/mm ²

900
5% percentile (75% probability)

800
700 2000 h aged
600
500
400
300
200
100
0
0 50 100 150 200 250 300 350
sustained stress w hile ageing in N/mm ²

Figure 1 Load influence on Residual Strength of Aged Specimen

This simple approach leads to secure and conservative results for “normal” conditions. This means normal indoor
climate or outdoor climate with mean annual temperatures around 10°C. If a more economic design is desired, or
conditions differ from these “normal” conditions in central Europe or Northern America , a new approach, which
takes the special conditions for each particular application including all the environmental influences into account,
has to be chosen.

3. NEW APPROACH BASED ON STRESS RUPTURE TESTS

In the ACI Test Guide [ACI 2004] two different tests are interesting regarding durability. The first is the so called
durability test. The second is the so called creep rupture test. It is difficult to understand why creep rupture and
durability should be tested separately. Why should creep rupture be tested outside of the concrete environment?
Creep rupture tests of GFRP rods show different limits for air, water and concrete [Renaud and Greenwood 2005].
The environment during the creep rupture test is particularly essential to a secure and conservative design value. In
addition for the design engineer no second check for creep rupture has to be performed if the design values are
secure long term values (including creep rupture) [ACI 2003].

268
 Figure -2 Shape of different stress endurance test curves of two different 6,4mm GFRP rods in cement solution in
log/log scale. Examples left for E-glass fibres in UP-resin, right for ECR-glass fibres in UP-resin. (Renaud and
Greenwood 2005)

It is clearly to be seen that the slopes of all lines are similar, while the curves are shifted to the left side due to higher
temperatures or weaker durability.

It is proposed that FRP should be designed for durability on the basis of a simple design strength equation that
multiplies the safety-factored tensile strength by a factor which is linked to various environmental parameters that
increase or decrease the factored tensile strength depending on the aggressiveness of the exposure environment and
the diameter, as follows. As no value for 1000h load capacity in creep tensile tests is known this value can be
estimated from the tensile strength:
ftd = ftk / η1000h / ηenv / γfrp (1)
simplified to:
ftd = ftk1000h / ηenv / γfrp (2)
where:
ftd = design value of tensile strength
ftk = characteristic value of tensile strength
ftk = characteristic value of load capacity for 1000h
ηenv= environmental strength reduction factor
η1000h= environmental strength reduction factor for 1000h sustained load
γfrp = material factor (1,25)

The environmental strength reduction factor ηenv, is obtained using the following equation. An exponential approach
is used, because deterioration is a described best by the kinetics of the chemical and physical processes.
ηenv,t = 1 / (100% – R10) n (3)

Where R10 is the standard reduction of tensile strength in percent per decade due to environmental influence, the
exponent n is the sum of the different influence terms: nmo is the term for moisture condition, nT is the term for
temperature and nSL is the term for desired service-life (Table 1). As a rule of thumb n=3 for “normal” conditions.
n = nmo .+ nT +. nSL + nD (4)

Table 1: examples for environmental design terms

Material fftk ff1000h R10 Moisture nmo MAT nT Service nSL Dia- nD
N/mm² N/mm² %/dec Condition °C life in a meter
CFRP 2000 5 Dry 0 0 0,7 1 -1 Half 0,6
AFRP 2000 10 Outdoor 1 10 1 10 0 Same 0
GFRP 1 1250 1000 18 Wet 2 20 1,4 50 0,7 Double (-0,6)
GFRP 2 700 300 25 30 2 100 1

269
That means for testing that stress rupture tests have to be performed in wet concrete and the basic parameter, the
R10 value and the 1000h value, have to be determined with a good probability. For the rebar from figure 1 this is
done for different temperatures (fig. 3 and 4). [Weber 2005, Byars 2006].

Creep Te nsile Endura nce Te st 23°C w e t concrete , Combar 16m m Creep Tensile Endura nce Test 60°C in W e t Concre te , Com ba r 16mm

1200 3 ,1

1200 1200 3 ,1

1200
1100 100a 1100 1100 1100
50a 1000 Extrapolation 1000
1000 1000
3 ,0
3 ,0

acc. DIN 53768

900 mean value 900 900 900
16 values

sustained load in MPa

ff1000h
sustained load in MPa

800 800 800 18%/dec 800

2 ,9

2 ,9

700 700 700 mean value 700

100a
Extrapolation
600 2 ,8

600 600 2 ,8

600
acc. DIN 53768 char. 5% 50a
10 values
500 500 500 2 ,7
500
2 ,7

o still running o still running

x failure point 5000h
x failure point char. 5%
400 2 ,6
400 400 2 ,6
400
aim: f fd = 435 /mm²
5000h

300 2 ,5

300 300 2 ,5

300
10 100 1000 10000 100000 1000000 10 100 1000 10000 100000 1000000
time to failure in h tim e to failure in h

Figure 3 and 4: Stress Rupture Performance of the same Rebar as in fig. 1 for Room Temperature and 60°C

4. DISCUSSION
For specific conditions and especially for normal conditions a durability test under load can lead to secure design
values. The assumption is that the whole service life is simulated in an accelerated long term test. The residual
strength and the load while testing yield a sufficiently safe design value. Defining the design value solely as a
percentage of the virgin strength, however, is not necessarily safe. 10 years of experience in applications under
normal conditions without reduction in residual strength will not lead to adequate indications for the following 90
years.

For a detailed environmental design another approach has to be chosen. Time, temperature, humidity and bar
diameter have an essential influence on the design value of the tensile strength. The proposed approach does not use
time temperature shifting. Based on experimental data as well as on theoretical deterioration kinetics the proposed
approach takes all the above mentioned parameters into account leading to secure design values for each and every
application. In Canada other values are required than in Florida, while indoor applications need other design values
than marine applications.

5. REFERENCES
[ACI 2003] American Concrete Institute (2003), ACI 440.1R03 “Guide for the Design and Construction of Concrete
Reinforced with FRP Bars”, Committee 440,
[ACI 2004]American Concrete Institute (2004), ACI 440.3R-04 “Guide Test Methods for Fiber-Reinforced
Polymers (FRPs) for Reinforcing or Strengthening Concrete Structures, Comitteee 440
[CSA 2002] Canadian Standards Association, S806-02 Design and Construction of Building Components with fibre-
Reinforced Polymers, Toronto 2002

[Byars et al. 2006] Byars et al. fib Task Group 9.3. Draft: Design and use of Fibre Reinforced Polymer
Reinforcement (FRP) for Reinforced Concrete Structures. FRP Reinforcement for RC Structures. Chapter 3
Durability: Performance and Design, February 2006

[Renaud and Greenwood 2005] Renaud C and Greenwood M (2005). Effect of Glass fibres and Environments on
Long-Term Duraqbility of GFRP Composites Proc. 9. EFUC Meeting Wroclaw PL 11/2005
[Weber 2004] Weber A. Bewehrungsstäbe aus GFK - Materialverhalten und Anwendungsgebiete, Proceedings.
SKZ IBK Symposium, Leipzig 2004

[Weber 2005] Weber A. GFK-Bewehrung - Bemessung und Anwendung, in Faserverbundwerkstoffe Hrsg.: Dehn,
Holschemacher, Tue, Bauwerk Verlag Berlin 2005

270
Third International Conference on FRP Composites in Civil Engineering (CICE 2006)
December 13-15 2006, Miami, Florida, USA

Durability Assessment of Adhesives and Reinforcement

at the FRP-Wood Interface
Gary M. Raftery (Gary.Raftery@nuigalway.ie),
(PhD Candidate, Department of Civil Engineering, National University of Ireland, Galway, Ireland)

Annette M. Harte,
(Senior Lecturer, Department of Civil Engineering, National University of Ireland, Galway, Ireland)

Peter D. Rodd,
(Lecturer, Department of Civil Engineering, National University of Ireland, Galway, Ireland)

ABSTRACT
Pultruded glass fibre reinforced composites can be used for reinforcing low-grade wood. In composite design, it is
imperative that a successful bond is established between the component materials. This paper describes a test
program to assess the bond quality between glass fibre reinforced polymers and spruce that contains large juvenile
wood proportions. Durability testing involved cycling specimens five times by a vacuum-pressure-soak-drying
procedure. Two pultruded reinforcing materials were selected for the study both comprising glass fibres aligned
unidirectionally, one in a vinylester resin and the other, in an engineered thermoplastic polyurethane. Bond tests
were executed using both wood laminating adhesives and two-part structural epoxies. The tests results show that the
bond depends not only on the adhesive in question but also the FRP type. In general, the durability tested specimens
failed at equal or lower shear strengths than solid control specimens, ambient tested wood-wood bond specimens or
ambient tested FRP-wood bond specimens that were taken from the same plank. Failure modes included wood
failure, reinforcement failure and adhesion failure.

KEYWORDS
Fiber Reinforced Plastics, Adhesive Bonding, Durability, Wood

1. INTRODUCTION
Glass fibre reinforced composites are ideally suited to the reinforcement of low-grade glue-laminated timber beams
due to their superior mechanical properties, low-weight, ease of handling and corrosion resistance. An adhesive, and
surface preparations that permit hygrothermal compliance, are essential to accommodate the shear and peeling
stresses that occur at the FRP-wood interface when the wood is subject to fluctuating environmental conditions. The
commercial viability of the reinforcement process would be considerably enhanced if a single adhesive system could
be used for bonding both the wood laminations and the FRP-wood interface. This would be particularly true if
commercial wood laminating adhesives could be utilised at the FRP-wood interface in contrast to more expensive
epoxy adhesives which are generally used in FRP-wood connections. Only limited results have been published on
bond tests regarding fibre reinforced plastics and low grade wood containing large proportions of juvenile wood.
Therefore, this research focuses on quantifying the durability performance of commercially available adhesives
when bonding viable reinforcing FRP materials and Irish grown Sitka spruce.

2. TEST PROGRAMME
The experimental programme undertaken involved the assessment of bond via the block shear test and comprised
four test phases. These involved the testing of solid shear specimens, ambient wood-wood bonded specimens,
ambient FRP-wood bonded specimens and finally durability tested FRP-wood bonded specimens.

271
Two pultruded fibre reinforced plastic materials were selected for the study comprising glass fibres aligned
unidirectionally in a vinylester resin (GFRP) and in an engineered thermoplastic polyurethane (FULCRUM). The
FRP selection was based on optimising cost versus mechanical properties (tensile modulus and tensile strength).
Both FRP materials were stored in the laboratory environment of 65±5% relative humidity and temperature of
20±2oC, while wrapped in protective cling-film. All the testing utilised the same batch of Irish grown Sitka Spruce
conditioned also in an environment of 65±5% relative humidity and temperature of 20±2oC. This is the service
environment in which the FRP-reinforced glulam beams are thought most likely to be utilised.

The adhesives examined comprised five conventional wood laminating adhesives and three structural epoxies. All
the adhesive manufacturers were confident that their products would form a good bond between the FRP materials
and the wood. Two phenol resorcinol formaldehydes (PRFs), a melamine urea formaldehyde (MUF), a polyurethane
(PU), and an emulsion polymer isocyanate (EPI) were the wood adhesives selected. In all instances, the instructions
stated in the adhesive technical data sheets and the recommendations of the adhesive manufacturers were followed
rigidly. For each adhesive and FRP combination under study, five number wood-wood bonded specimens, twenty
number ambient tested FRP-wood bonded and ten number durability tested FRP-wood bonded specimens were
tested from the same plank. Ten solid shear specimens, which were also cut from the same plank, were directly
compared with the results. The solid shear strength test set-up was designed to be consistent with the adhesive bond
tested specimens (Okkonen & River, 1989). The bonded assemblies were manufactured and tested with reference to
a recognised block shear test standard (ISO, 2001). A solid test specimen width of 40mm was utilized because of a
size restraint. Specimen dimensions are illustrated in Figure 1. The thickness of the FRP material, y, was 3mm for
the GFRP and 1.2mm for the FULCRUM. A wood backing piece of thickness, x, was used to give a total (x+y)
thickness of 20mm.

Figure 1: Test Specimens: (a) Solid, (b) Wood-Wood (c) FRP-Wood

Surface preparation of the GFRP material prior to bonding involved gently abrading and then cleaning with
methylated spirits. The FULCRUM material was simply wiped clean with methylated spirits because of its
inherently rough surface. Clear, defect free wood was carefully selected to promote the highest quality bonding. The
wood surface was planed parallel to the grain directly before the bonding process to ensure that the gluing face was
uniformly smooth and free from contamination.

The durability test utilised a modified EN 391 procedure (CEN, 2001). FRP-wood test specimens were initially
subjected to a vacuum pressure of 70kPa whilst submerged in water followed by pressure soaking at 600kPa. The
specimens were subsequently dried for 90 hours in an environmental chamber at 25±5% relative humidity and
35±5oC temperature. The procedure was then repeated five times. Test specimens were finally left to recondition for
a minimum of 15 days in an environment of 65±5% relative humidity and a temperature of 20±2oC prior to testing.
All specimens were sheared parallel to the grain at constant loading rates such that the test specimen had a target
failure time of 60±20 seconds (ISO, 2001). Controlled rate of load was preferred over rate of crosshead
displacement as a more progressive application of the load is achieved with the former. A quantitative assessment of
the percentage adherend failure, estimated to the nearest five percent of the shear plane area, was executed on each
sample. Shear strength of the fractured specimens was calculated based on the assumption of a uniform stress
distribution and an instantaneous failure over the theoretical shear plane.

272
3. EXPERIMENTAL RESULTS AND DISCUSSION
The shear strengths obtained and adherend percentage failure assessments from the tests executed with the GFRP
and the FULCRUM material when bonded with the wood laminating and epoxy adhesives are presented in Figures
2-9.

KEY: A=Solid Specimens, B=Ambient tested Wood-Wood Specimens, C= Ambient Tested FRP-Wood Specimens,
D=Durability Tested FRP-Wood Specimens, I = Standard Deviation

Figure 2: Shear strength vs. Adhesive Figure 3: Shear strength vs Adhesive

– GFRP with wood adhesives – FULCRUM with wood adhesives

Figure 4: Shear strength vs. Adhesive Figure 5: Shear strength vs Adhesive

– GFRP with epoxy adhesives – FULCRUM with epoxy adhesives

Figure 6: Adherend Percentage Failure vs. Adhesive Figure 7: Adherend Percentage Failure vs. Adhesive
– GFRP with wood adhesives – FULCRUM with wood adhesives

Figure 8: Adherend Percentage Failure vs. Adhesive Figure 9: Adherend Percentage Failure vs. Adhesive
– GFRP with epoxy adhesives – FULCRUM with epoxy adhesives

The mean shear strength for PRF 1 bonded GFRP-wood durability tested specimens was marginally higher than for
solid specimens and wood-wood bonded specimens from the same board. It also compared well to ambient tested

273
GFRP-wood bonded specimens. All the MUF bonded durability tested GFRP-wood specimens failed during the
moisture cycling procedures. For the remaining wood laminating adhesives, durability tested GFRP-wood bonded
specimens failed at lower shear strengths (38% lower for PRF 2, 17% lower for PU and 35% lower for EPI) than
those of the solid specimens, ambient tested wood-wood specimens and ambient tested GFRP-wood specimens.
With regard to the epoxy bonded GFRP-wood specimens, the durability tested specimens failed at lower shear
strengths than the ambient tested GFRP-wood bonded specimens. The difference was particularly significant for
Epoxy 2 bonded specimens (60% lower) and Epoxy 3 bonded specimens (55% lower). Adherend failure percentages
of the durability tested specimens were favourable (over 80%) for the PRF adhesives and EPI adhesive and
compared well with ambient tested wood-wood specimens and ambient tested GFRP-wood specimens. Large
variability existed in the results for the PU adhesive which obtained an average failure of 63%. The adherend failure
results obtained from the epoxy adhesives were also susceptible to large variations. For example, for Epoxy 3, one
specimen failed in adhesion during the moisture cycling while another failed with an adherend failure of 95%.

The PU and EPI adhesives exhibited the highest bond quality for the bond shear strength tests executed with the
FULCRUM material when bonding with the conventional wood laminating adhesives selected for the study. In
contrast to the tests executed with the GFRP, the PRFs displayed a poor bond quality between the FULCRUM and
the wood after being subjected to the hygrothermal stresses of the durability test. It was noted that all samples
bonded to the FULCRUM material with both the PRF adhesives and MUF adhesive failed in adhesion while being
subjected to the moisture cycling. The durability tested epoxy bonded specimens had lower shear strengths than
solid specimens, ambient tested wood-wood specimens and ambient tested FULCRUM-wood specimens from the
same plank. Epoxy 3 exhibited the best performance of the three epoxies in terms of strength retention after
durability testing. It is believed that Epoxy 2 would exhibit an improved performance if thicker bond lines were
used. Due to the premature failure of the PRF and MUF bonded specimens during durability testing, zero adherend
percentage failure was assigned to these test samples. The durability tested PU and EPI adhesives displayed
excellent adherend percentage failure results with the EPI bonded specimens achieving even better results than the
ambient tested FULCRUM-wood specimens. The high percentage adherend failures obtained when bonding with
Epoxy 1 and Epoxy 3 showed that strong durable bonds were formed. The average failure for Epoxy 2 bonded
FULCRUM-wood specimens were under 20% for the reason above discussed.

4. CONCLUSIONS
Significant differences in the results for the ultimate shear strengths and adherend failures percentages between the
two FRP types illustrates that bond integrity depends not only on the adhesive in question but is also substrate
specific. In general, the durability tested specimens failed at equal or lower shear strengths than solid control
specimens, ambient tested wood-wood specimens or ambient tested FRP-wood specimens taken from the same
plank. It is seen that strong durable bonds can be formed at the FRP-wood bond interface when using specific
conventional wood laminating adhesives which can exceed the performance of that obtained from bonding with
more expensive epoxy adhesives. It is noted that the durability test used in this study is an accelerated aging test and
the hygrothermal stresses induced on the tested specimens are considered extreme. The test results provide an
excellent assessment of the resistance to hygrothermal stresses at the FRP-wood interface.

5. ACKNOWLEDGEMENTS
The authors would like to sincerely thank COFORD; National Council for Forest Research and Development,
Coillte and the Irish Research Council, Science, Engineering and Technology (IRCSET) for sponsorship.

6. REFERENCES
CEN, European Committee for Standardization (2001), Glued laminated timber, Delamination test of glue
lines, EN 391.
International Organisation for Standardisation (2001), Adhesives; Wood-to-wood adhesive bonds;
Determination of shear strength by compressive loading, ISO 6238.
Okkonen, E. Arnold and River, Bryan H. (1989), Factors affecting the strength of block-shear specimens. Forest
Products Journal 39 (1):43-50.

274
 !"#!$ %& % % '()

) % & %& % *

(+ % & %& % *

+ % & %& % *

! "
# $

"
#

"% "& "& "'

% #

#
$

( )
*
& #
# +
#

# "
, '
# +
" #

275
 !"

# # - ./0 1 2 # $ .34444
( 5 6784 ( # $ 48 #
# 5% , 9 & 96 1 2
.44 .:44 ( " 5%!;' " .44<6 2
# $ 48 #
5% , 9 846 1 2 # 3.:44 ( #
5 / 80 = 6# 80 ( .. 0 ( " 5%!;' "
.44<6 2 # $ 4 80

# $% &" '% ( &) # '&

#
3

Steel plate
500 mm

y y

1000 mm
500 mm
600 mm

x x
500 mm

6 mm 50 mm 6 mm 60 mm (preconditioned)
50 mm (fatigue)
CFRP strip: 1.4 mm

Adhesive: 1.1 mm

*+ , '+-! ") ". !# /'& * '% ( &) )'+-! ") &0' $ % & * '% (1&' '"$ !2

" # , #
# # #
5 6 #
% #
$ >0 ? 5@ 6 0?#
"# 3. # .4 = 5< 6 A04 = 5< 6
# 3 *
"# &% &%B
"

-! 3 "$ # '& '0 &4 '&% & ! % & "0' $. '0 . "# $% &"+&) "+)(

# $% & # $% & . % ! ($!" ! # ( '*

(# .4 = 5< 6A 04 = 5< 6 CD. E F >0 ? GC@ EF0?
) &%
H 3 &% .4
H . &% 84
@%3 &% .# ,
@%. &% 8# ,
@%H &% .4 .# , 5 6
I) &%B
I@%. &%B 8# ,
I@%H &%B .4 .# , 5 6

276
5
) # # $ 4 44: +
% # $ #
# # "
#
# #

# 5 " .44<6" #
# # 5 .6 ! .
# * # ! "
,
# 5 " .4476
3744 3744
.
3.44 . 3.44

3444 3 3444
8
ε Cµ + E

ε Cµ + E
:44 8 :44

<44 <44 7
7 7
744 744 8
.
.44 .44

4 4
4 0 34 30 .4 4 0 34 30 .4
* *
1 2 1-2

*+ , 6 !) 0' % '& 4"&' % !7 ) ) "#!$ % & 0' 1 2 &) 1-2 ∆

! " "
# " " #
" * # #
# ' #
'
5 " .4406 ! 8" I@%. I@%H #
4000 4000

E = 200 GPa 6 E = 200 GPa

3000 3000
εxx [µm/m]
εxx [µm/m]

1 7 6
5 2 8
2000 3 9 2000
5
4 4 10
5 11 4
6 12
1000 1000
3 2,3
1 2
1
0 0
0 400 800 1200 1600 0 400 800 1200 1600
time [s] time [s]

1 2 1-2

*+ 5 , 6 !) 0' % '& 4"&' % !7 ) ) "#!$ % & 0' 1 28 &) 1-28 ∆

9
# # ( %
' #
:8 344 ( " # # , $ 47
3/,@ 7. ,@ HJ F :8 (
.4 ,@ 04 ,@ HJF 344 ( ' 7" < "
:8 ( # 7? HJ F 344 ( "
4: # ,
' # #

277
 5( " .4476 K 5' " 3>>:6"
# 7 <: 33 0/ ( HJ F :8 (
# 0 03 38 /:( HJ F 344 ( #
"# 30 ? & ,
# , "# #

110

stiffness reduction [%]

100

F T1 ( ∆σ=83 MP a)
90 F T2 ( ∆σ=83 MP a)
F T3 ( ∆σ=100 MP a)
F T4 ( ∆σ=100 MP a)

80
1 104 1 105 1 106 1 107
number of cycles

*+ 9 , )+$ '& & " 00& ""'0 '0 ) "# $% &")+ &*0 *+ ""

:
% " K
"
* " "
"
# "
! " $ # ,
" #
% $

;
-B)!% 5! @ # , % - B 6
, # , %, !

<
' '(" & 53>>:6 L'
M" ( *" - %N 0>" 33./ 338/
" -" " 5.4476 L% O
M" & & *( + + ( , # -
& % % % %- O' & ") N "! " 00 </
" -" " 5.4406 L&
M
+ %- OB ; ") .>3 .>: B " "
5.44<6" L%
M" ./ / 0+ /& /
" N .4" .. 88
( "; N(" ** & " N 5.4476L% +
M & & *( + + ( , # -
& % % % %- O' & ") N "! " 733 7.4
%!;' 5.44<6 % O++### , + + + 30+48+4<

278
Third International Conference on FRP Composites in Civil Engineering (CICE 2006)
December 13-15 2006, Miami, Florida, USA

EFFECT OF SURFACE DETERIORATION ON STRESS TRANSFER

BETWEEN CONCRETE AND FRP LAMINATE
Amnon Katz
Department of Civil and Environmental Engineering
National Building Research Institute
Technion – Israel Institute of Technology
Haifa 32000, Israel

ABSTRACT
Fiber Reinforced Laminates are widely used to strengthen concrete members suffering from deterioration. Stress
transfer between the strengthened concrete and the strengthening laminate depends on the surface quality of the
concrete. In the case of deteriorated concrete, stress transfer is complicated due to the gradient in its properties when
going from the deteriorated surface to the healthy bulk concrete. This work presents the mechanism of stress transfer
between carbon and glass FRP laminates and concrete with deteriorated surface.

Different modes of failure were identified, ranging from horizontal shear of the deteriorated concrete at the surface
to diagonal shear of the beam. It was found that a significant amount of stress is transferred by friction within the
deteriorated surface, leading to remarkable load-bearing capacities.

Keywords: FRP Laminates, Beam-Strengthening, Concrete, Deterioration

1. INTRODUCTION
Many studies have been conducted to investigate the mechanism of stress transfer between external reinforcement
and concrete, first on steel sheet reinforcement (Swamy et. al, 1986; Van-Gemert and Vanden, 1986) and more
recently on FRP (Chen et. al., 2001; Xie and Karbhari, 1998; De Lorenzis et al., 2001). These researchers realized
the importance of proper stress transfer between the reinforcing system and the concrete.
Models predicting the behavior of externally reinforced beams were developed (Rabinovitch and Frostig, 2000;
Mukhopadhyaya and Swamy, 2001), most of which deal with the overall behavior of a beam made of concrete
(either cracked or solid) that is considered uniform throughout the cross section. Other studies (Chen et al., 2001; De
Lorenzis et al., 2001; Jia et al., 2005), which specifically investigated the bond and stress transfer between external
bonded sheets and concrete bodies, also dealt with solid bodies of concrete. According to the model developed by
Bizindavyi and Neale (1999), the shear force between the FRP laminate and the concrete is transferred across a zone
about 2-3 mm thick, thereby forming a separating zone whose mechanical properties must be taken into account
when calculating the mechanism of stress transfer. Some researchers attributed the mechanical properties of this
layer solely to the resin used to bond the FRP sheet to the concrete (De Lorenzis et al., 2001), although the
properties of the concrete at the surface are generally different from those of the bulk concrete.
FRP laminates are widely used to strengthen concrete structures in various cases in which the surface of the concrete
elements is severely deteriorated but the core concrete of such elements remains undamaged. Stress transfer from the
surface to the core concrete through layers of varied strength influences the structural behaviour of the element. This
study investigated the stress transfer between carbon and glass FRP laminates and the surface of deteriorated
concrete beams. Better understanding of the stress transfer in these cases will enable better design for strengthening
of concrete elements.

2. EXPERIMENTAL METHOD
Two types of notched concrete beams, 80X150X600 mm, were prepared for this study: solid and layered. The solid
beams were made of a single type of concrete throughout the cross section, whereas the layered beams contained a

279
layer of weak concrete used to simulate the deteriorated surface. The bulk concrete of the solid and layered beams
was prepared at three strength levels: 29.7, 46.4 and 76.8 MPa (compressive strength at 28 days), denoted as I, II and
III, respectively. Compressive strength of the concrete layer that was used to simulate a deteriorated surface was
13.8 MPa and its tensile strength (determined by a pullout test according to ASTM C1583) was 1.23 MPa. These
values are somewhat lower than the minimal values recommended by ACI 440.2 (17 and 1.4 MPa, respectively). An
additional layer of concrete was cast between the surface layer and the bulk concrete in order to allow a gradual
change in properties between the layers. The deteriorated layer and the intermediate layer were each about 10 mm
thick.

After concrete curing, the surfaces to be tested were cleaned and glass or carbon fiber laminates were wet-laid.
Additional anchoring wraps were applied on one side of the beam to force failure on the other side only. Bond
breakers, 25 mm wide, were used on both sides of the notch to prevent local concentration of stress near the notch. A
series of 18 strain-gages was applied to the sides of the beam within the tested zone to monitor strain development in
the concrete at different locations and an additional 4 strain-gages were applied on the laminate itself. Figure 1
shows the test setup and arrangement of strain-gages. Most of the results were calculated as the average of three
specimens.

75
150

FRP Adhesive
50

50 25 75 150 25 150 75 50

600
Figure 1: Test setup and strain-gages arrangement (mm).

3. RESULTS
3.1. Modes of Failure and Ultimate Loads

Three modes of failure were identified:

Mode 1: Failure of the bond at the interface between the epoxy layer and the concrete.
Mode 2: Failure of the concrete surface layer.
Mode 3: Failure by shear of the beam. The diagonal crack started close to the end of the laminate.

Failure by Mode 1 was typical to the high-strength (concrete III) solid beams; Mode 3 was typical to layered beams
with relatively weak bulk concrete; Mode 2 was typical to all other beams, i.e. solid beams with relatively weak
concrete or layered beams with strong bulk concrete. Failure within the concrete layer at the interface was expected
to occur in layered beams, due to the weak concrete in this area. However, this type of failure was observed also in
solid beams made of Type 1 concrete (30.5 MPa), which is much stronger than the ACI 440.2 recommendations.

Close examination of the specimens that failed by Mode 1 showed some concrete crumbs adhering to the end of the
laminate, which indicate that local tensile stresses developed near the end of the laminate, as was analyzed by
Rabinovitch and Frostig (2000). It is possible that similar stress development led also to the initiation of a crack near
the end of the laminate in the layered beams. This crack begins to develop near the end of the laminate within the
weak layer of the surface. When the crack is large enough, the cross section of the beam is reduced and the beam

280
fails by shear of the "healthy" concrete. If the bulk concrete is strong enough to support this load, failure will occur
according to Mode 2.

Figure 2 presents the ultimate load at failure of the tested beams. A decrease in the ultimate load-bearing capacity
was seen with layered beams reinforced with carbon laminates as the strength of the bulk concrete increased. All
layered beams reinforced with glass laminates exhibited similar ultimate loads regardless of concrete strength. The
ultimate load of solid beams reinforced with glass laminate was not affected by concrete strength, and only the high-
strength concrete reinforced with carbon laminate exhibited higher ultimate loads compared with the other carbon-
reinforced solid beams. Beams reinforced with carbon laminates exhibited somewhat higher ultimate load-bearing
capacities than did those reinforced with glass laminates.

40

35
Concrete
30
type
Ultimate Load (kN)

25 I

20 II
III
15

10

0
Layered Solid Layered Solid

Carbon Glass

Figure 2: Ultimate load of strengthened beams.

3.2. Strain distribution in concrete layers

Strain development in the concrete layers was measured by a series of strain-gages as seen in Figure 1. Figure 3
presents the strain distribution in a layered beam reinforced by a carbon laminate at two load levels. At low loads,
tensile strain begins to develop at some distance from the notch and as the load increases, greater tensile strain
develops closer to the notch, until the concrete fails proximate to the notch in the weak layer. Tensile strain begins to
develop away from the notch (near SG3/line 1 in Figure 3), however, compression strain was measured closer to the
notch (SG2/line 1). Despite debonding in the area close to the notch, strain measured in the laminate indicates that
strain continues to increase towards the notch along the unbonded area. This indicates a possible stress transfer
between the laminate and the concrete surface by means of a friction mechanism that enables the transfer of a
significant amount of load.

4. CONCLUSIONS
The difference between the stiffness of the reinforcing system and that of the concrete surface plays an important
role in transferring stresses between the layers. The presence of a weak and soft concrete layer between a layer of
stiff laminate (carbon) on one side, and stiff concrete bedding on the other side, may lead to greater concentration of
stress in that layer as the stiffness of the bedding concrete increases. As a result, the load-bearing capacity of the
beam may decrease.
A significant proportion of the shear stress was transferred by a mechanism of friction at the external concrete layer
after initial failure of this layer. Measurements of the longitudinal strain in the layers of concrete indicated a

281
complicated situation of stress transfer with simultaneous action of external and internal shear forces and moments.
A model simulating this stress distribution is currently under development.

54%

13 .00 75% 18 .06

Line 3 Line 3

100-120
100-120
80-100
80-100
60-80
Line 2 60-80 Line 2
40-60
40-60
20-40
20-40
0-20
0-20
-20-0
Line 1 Line 1
Crack SG1 SG2 SG3 ….. SG 4 Crack SG1 SG2 SG3 ….. SG 4

3000 3000

2000 2000

1000 1000

0 0
1 2 3 4 5 6 1 2 3 4 5 6

Figure 3: Strain distribution in the concrete layers (area diagram) and along the laminate (bar diagram below) in a
layered beam reinforced with carbon laminate (concrete strength II). Load level is indicated at the top of each
diagram. Some of the points were calculated by interpolation.

ACKNOWLEDGMENT
This work was partially supported by the Israeli Ministry of Construction and Housing.

REFERENCES
ACI 440.2R-02, 2002, "Guide for the design and construction of externally bonded FRP systems for strengthening
concrete structures", American Concrete Institute, Farmington Hills, MI, USA.
Bizindavyi, L. and Neale, K. W., 1999, "Transfer lengths and bond strengths for composites bonded to concrete",
Journal of Composites for Construction, Vol. 3, No. 4, pp. 153-160.
Chen, J .F., Yang, Z. J., Pan, X. M. And Holt, G. D., 2001, "Effect of test methods on plate-to-concrete bond
strength", Proceedings Fibre-Reinforced Plastics for Reinforced Concrete Structures – FRPRCS-5", Burgoyne C. J.
(Editor), Thomas Telford, London, Vol. 1 pp. 429-438.
De Lorenzis L., Miller B. and Nanni A., 2001, "Bond of fiber-reinforced polymer laminates to concrete", ACI
Materials Journal, Vol. 98, No. 3, pp. 256-264.
Mukhopadhyaya, P. and Swamy N., 2001, " Interface shear stress: a new design criterion for plate debonding",
Journal of Composites for Construction, Vol. 5, No. 1, pp. 35-43.
Jia, J., Boothby, T. E., Bakis, C. E., and Brown, T. L., 2005, "Durability evaluation of glass fiber reinforced-
polymer-concrete bonded interfaces", Journal of Composites for Construction, Vol. 9, No. 4, pp. 348-359.
Rabinovitch, O. and Frostig, Y., 2000, "Closed-form high-order analysis of RC beams strengthened with FRP
strips", Journal of Composites for Construction, Vol. 4, No. 2, pp. 65-74.
Swamy, R. N., Jones, R., and Bloxham, J. W., 1987, ‘‘Structural behaviour of reinforced concrete beams
strengthened by epoxy-bonded steel plates’’ The Structural Engineer, London, 65A(2), 59–68.
Van-Gemert, D. and Vanden B. M., 1986, "Long-term performance of epoxy bonded steel-concrete joints",
Proceedings, Adhesion Between Polymers and Concrete, Chapman & Hall, London, p. 518-527.
Xie, M. and Karbhari, V. M., 1998, "Peel test for characterization of polymer composite/concrete interface", Journal
of Composite Materials, Vol. 32, No. 21, pp. 1894-1913.

282
Part IX. Fatigue Issues
Third International Conference on FRP Composites in Civil Engineering (CICE 2006)
December 13-15 2006, Miami, Florida, USA

FATIGUE BEHAVIOR OF EXTERNALLY BONDED

STEEL FIBER REINFORCED POLYMER (SFRP)
FOR RETROFIT OF REINFORCED CONCRETE

Kent A. Harries
(Assistant Professor, Department of Civil and Environmental Engineering, University of Pittsburgh)

Patrick L. Minnaugh
(Graduate Research Assistant, Department of Civil and Environmental Engineering, University of Pittsburgh)

ABSTRACT
Steel fiber reinforced polymer (SFRP) strips comprised of multiple high-strength wires have been introduced into
the repertoire of the structural engineer in recent years. The deleterious effects of fatigue loading on FRP-to-concrete
bond have been identified in previous studies by the author; therefore the effect of fatigue loading on the bond
behavior of SFRP is investigated. Four large-scale beam specimens (4.9 m long) having externally bonded SFRP
retrofits are tested. These specimens are paired with unretrofit and CFRP-retrofit companion specimens allowing a
number of direct comparisons to be made. Of the SFRP specimens, one is tested in monotonic loading to failure
while the remaining three are tested at various fatigue load levels ranging from service load level to an extreme load
level. Service load fatigue is cycled for two million cycles and the specimen is then tested monotonically to failure
to assess the effects of fatigue on the ultimate performance of the beam. Extreme loading is selected to result in
fatigue-induced failure of the internal reinforcing steel.

KEYWORDS
Bond, Fatigue, Retrofit, Steel Fiber Reinforced Polymer (SFRP)

1. INTRODUCTION
Steel fiber reinforced polymer (SFRP) composite materials have been introduced as an alternative to glass and
carbon fiber reinforced polymer (GFRP and CFRP) composite materials (Hardwire 2002). Initial research (e.g.:
Huang et al. 2005) has indicated that SFRP is comparable to FRP materials as a measure for the flexural retrofit of
reinforced concrete members. SFRP is bonded to the tension surface of concrete members and through composite
action increases the effective reinforcing in the member active in resisting post-retrofit loads. Like all bonded
applications, the effectiveness of SFRP is only realized as long as the bond remains sound. Bond behavior and
debonding failures are known to be critical for FRP materials (e.g.: Oehlers 2005). Some retrofit applications may
be subject to transient or fatigue load conditions. Previous studies by the author (Harries 2005) have clearly
demonstrated the deleterious effects of fatigue loading on FRP-to-concrete bond. Debonding failures have been
identified in cases where the transient stress level in bonded CFRP is only 10% to 14% of the material’s capacity
(Quattlebaum et al. 2005). Additionally, deterioration has been observed in cases where the stress level is as low as
4% of the FRP capacity (Harries and Aidoo 2005). For this reason, the fatigue behavior of SFRP-retrofit concrete
sections may also be a concern. This study is believed to be the first to address the fatigue-induced degradation of
SFRP performance.

1.1 Steel Reinforced Polymer (SFRP) Material

SFRP consists of high performance steel cord reinforcement embedded in a polymer matrix. The polymer matrix
provides continuity to the final composite product and serves as the adhesive to affix the SFRP to the concrete
substrate. The steel reinforcement used in this investigation is a high carbon steel cord with a micro-fine brass or
AO-brass (Adhesion Optimized) coating. The cord is made by twisting 5 individual wires together - 3 straight
filaments wrapped by 2 filaments at a high twist angle (Hardwire 2002). Properties of the SFRP composite and FX
adhesive system used are provided in Table 1 along with properties of comparable preformed CFRP strip materials
and adhesives specified for the CFRP applications.

283
 Table 1: Manufacturer reported material properties of SFRP, CFRP and adhesives.

tensile tensile elongation thickness width axial stiffness

strength modulus (Ef) at rupture (tf) used (bf) (EA = Eftfbf)
SFRP 1138 MPa 75.5 GPa 0.014 1.2 mm 121 mm 10963 kN
CFRP 2792 MPa 155.1 GPa 0.018 1.4 mm 50 mm 10857 kN
Adhesive FX (SFRP) 31 MPa not reported 0.025 ≈0.8 mm - -
Adhesive L (CFRP) 14 MPa 2.2 GPa 0.063 ≈1.6 mm - -
Adhesive H (CFRP) 25 MPa 4.5 GPa 0.010 ≈1.6 mm - -

2. EXPERIMENTAL PROGRAM
Four reinforced concrete beams having externally bonded SFRP flexural retrofits are reported. Additionally a
number of beams having equivalent CFRP retrofit measures are reported. The beams are 254 mm deep, 152 mm
wide and are supported over a simple span of 4540 mm. All beams have 3 - #4 (12.7 mm dia.) primary reinforcing
bars. The internal steel had yield and ultimate strengths of 429 MPa and 667 MPa, respectively. Twenty-eight day
concrete compressive strength determined from cylinders was 23.3 MPa. Specimen and retrofit geometries are
shown in Figure 1. The SFRP and CFRP material properties are reported in Table 1. As shown in Figure 1, the
SFRP application was 121 mm wide while the comparable CFRP application was 50 mm wide. These applications
are equivalent in terms of axial stiffness (EA) as indicated in Table 1. For the CFRP specimens, the “2” and “2x1”
specimens (Fig. 1) performed similarly (Harries et al. 2006); thus both are reported here for a more comprehensive
comparison. The SFRP was applied using “Adhesive FX” while the CFRP applications were applied using either
“Adhesive L” or “Adhesive H”. Because the SFRP is relatively flexible and “porous”, a thinner and more uniform
adhesive bondline (following subtle contours of the substrate concrete) was obtained than in the CFRP applications.
Nonetheless, all adhesive systems used were found to perform well.

All beams were tested in three point flexure over a 4540 mm simple span. Applied load, midspan displacement and
coincident reinforcing bar and SFRP/CFRP strains were recorded. Specimens reported in Table 2 were tested under
cyclic fatigue loading conditions. The applied load at midspan was cycled between the minimum and maximum
values shown at rate between 1.2 and 1.7 Hz (depending on the test). Specimens reported in Table 3 were tested
under increasing monotonic load to failure. Specimens S4F, L2F and H2x1F did not exhibit a fatigue-induced failure
prior to 2,000,000 cycles and were subsequently tested under increasing monotonic load to failure (Table 3).
A complete description of the test programs reported here is found in Minnaugh (2006).
152 mm 400
stress range (S), MPa

350 CFRP
2 #3
300 SFRP
254 mm

SC, S4F, H2F H2x1F

250
S4.75F & &
& S5.5F L2F L2x1F 200
150
3 #4
100
121 mm 50 mm 2 x 25 mm 1.E+05 1.E+06 1.E+07
SRP CFRP CFRP cycles to failure (N)
Figure 1: Test specimen and retrofit geometry. Figure 2: S-N behavior of SFRP and CFRP retrofit beams.

3. EXPERIMENTAL RESULTS
Key results from the fatigue and monotonic tests are presented in Tables 2 and 3, respectively. For the fatigue tests,
the reinforcing steel and SFRP/CFRP stress ranges (as determined from measured strains using the moduli reported
in Table 1 and assuming E = 200 GPa for the internal reinforcing steel) and the apparent beam flexural stiffness
(determined as the slope of the line between the load and displacement at the lower and upper limits of the applied
fatigue loading) at both the initial fatigue cycle (N = 2) and at the final measure fatigue cycle (N = Nf) is presented.
The initial cycle is reported as N = 2 since the first cycle, N = 1, is required to crack the beam section and thus has a
different apparent stiffness. For the monotonic tests, the ultimate deflection is defined as that corresponding to a
post-peak load of 80% of the peak load attained. Displacement ductility is defined as the ratio of the ultimate
displacement to that at general yield. The SFRP/CFRP debonding strain reported is interpreted as the greatest
SFRP/CFRP strain observed while still bonded to the concrete, and thus represents a lower-bound on the strain to
initiate debonding.

284
 Table 2: Summary of key results from fatigue tests.

Specimen Designation CF1 S4F S4.75F S5.5F L2F H2F L2x1F H2x1F
retrofit/adhesive type (Table 1) none SFRP/FX SFRP/FX SFRP/FX CFRP/L CFRP/H CFRP/L CFRP/H
minimum-maximum applied load (kN) 4.5-22.2 4.6-22.1 4.6-25.4 4.5-28.6 4.6-22.2 4.6-22.2 4.5-22.2 4.6-22.2
rebar stress range (MPa) 241 190 238 286 207 199 207 220
SFRP/CFRP stress range
N=2 n.a. 83 98 113 170 188 176 199
(MPa)
secant stiffness (kN/mm) 1.42 1.77 1.64 1.62 1.70 1.51 1.55 1.61
final cycle, Nf 329324 2M 689671 286306 2M 1128006 447695 2M
rebar stress range (MPa) n.a. 211 254 315 240 217 230 251
SFRP/CFRP stress range
N=Nf n.a. 98 85 123 190 206 187 224
(MPa)
secant stiffness (kN/mm) n.a. 1.50 1.48 1.48 1.35 1.37 1.42 1.38
1
due to a power failure, Specimen CF was loaded to failure at N = 329,324 prior to fatigue-induced reinforcing bar rupture.

Table 3: Summary of key results from monotonic tests.

Specimen Designation C SC S4F1 L2 H2 L2x1 H2x1 L2F1 H2x1F1

retrofit/adhesive type (Table 1) none SFRP/FX SFRP/FX CFRP/L CFRP/H CFRP/L CFRP/H CFRP/L CFRP/H
maximum load (kN) 31.0 47.4 46.6 44.3 43.4 45.5 45.2 45.5 45.3
ultimate deflection (mm) 78.2 55.8 53.4 56.9 55.3 64.7 56.1 67.3 51.3
displacement ductility 2.61 1.67 1.58 1.70 1.60 1.89 1.70 2.05 1.45
SFRP/CFRP strain at debond
n.a. 3000 3170 6690 3550 7880 3200 4300 3910
(µε)
1
monotonic test following 2,000,000 cycles of fatigue conditioning as described in Table 2.

Figure 2 shows the observed S-N (stress range on internal reinforcing steel – cycles to failure) behavior of the SFRP
specimens and a number of CFRP retrofit specimens (including those reported in Table 2) having the same
dimensions, reinforcing and retrofit details and loading protocol (Harries et al. 2006; Quattlebaum et al. 2005). As is
expected, the fatigue behavior is governed by the behavior of the existing internal reinforcement in all cases. The
SFRP specimens have essentially the same behavior as the equivalently retrofit CFRP specimens. As indicated in
Table 2, the reinforcing bar stress range increases coinciding with a decrease in apparent beam stiffness as the
specimens are subject to fatigue loading (increasing N). The SFRP/CFRP stress range is also seen to increase with
cycling, reflecting the continued composite behavior of the retrofit system. The SFRP is less stiff than the CFRP,
requiring greater material width to affect the same retrofit stiffness but resulting in lower shear stress being
transferred through the adhesive and into the concrete substrate.

Load-displacement relationships for the monotonic SFRP-retrofit control Specimen SC, a representative equivalent
CFRP-retrofit Specimen L2 and the unretrofit control Specimen C are shown in Figure 3(a). The SFRP-retrofit
specimen exhibits a stiffer flexural response having a marginally greater ultimate capacity. Deflection capacity and
ductility are similar to the CFRP retrofit specimen. Similar behavior is observed in the monotonic test of Specimen
S4F following two million cycles of fatigue conditioning shown in Figure 3(b). Specimen S4F exhibits essentially
identical response to representative CFRP-retrofit Specimen L2F in the initial cycle (N =1) and through the fatigue
conditioning process. The deflection offset following fatigue conditioning is approximately the same, 8 mm, in both
cases. In the final monotonic load to failure, SF4 achieves a greater ultimate capacity although exhibits less
deformation capacity than L2F. It is noted, however, that L2F exhibited greater deformation capacity than other
similar specimens reported in Table 3. Figure 3(c) shows the load deflection cycles for S4.75F and S5.5F. In each
case, the ultimate deflection achieved prior to fatigue-induced reinforcing bar rupture was approximately 24 mm.
This deflection is approximately the same deflection achieved by Specimen SC at the initiation of reinforcing bar
yield and reinforces the observation that the deflection at fatigue failure is well-estimated by the monotonic yield
deflection (Quattlebaum et al. 2005).

3.1 Debonding of SFRP/CFRP

No debonding of the SFRP was evident during any stage of fatigue loading. This observation is in contrast to
debonding observed during fatigue loading of the CFRP-retrofit specimens identified in Table 2 (Zorn 2006). Due to
the greater width of SFRP required to affect the same retrofit, the interfacial shear stress is reduced. As seen in Table

285
 2, even at the greater stress range loading (S4.75F and S5.5F), the SFRP stress remained well below that of the
comparable CFRP specimens.

In the monotonic tests reported in Table 3, SFRP debonding is observed at comparably lower strains than in the
CFRP specimens. This lower debonding strain is believed to result from the combination of a stiff adhesive and
thinner adhesive bondline (Table 1) and reflects the importance of concrete substrate properties on debonding
beavior. Nonetheless, the increased beam capacity, similar deflection capacity and no apparent debonding during
fatigue conditioning indicate that the SFRP-adhesive system may be tougher than the comparable CFRP system.
This latter effect is believed to result from the lower interfacial stresses developed.
50
SFRP SFRP
SFRP (SC)
(S4.75F) (S5.5F)
40
applied load (kN)

CFRP (L2) SFRP

(S4F)
30
CFRP
N=1 N=1 N=1
unretrofit (C) (L2F)
20

10

N = 2,000,000 N = 689,650 N = 286,306

0
0 20 40 60 80 100 0 20 40 60 80 0 10 20 30 0 10 20 30

midspan deflection (mm) midspan deflection (mm) deflection (mm) deflection (mm)
(a) monotonic tests (b) fatigue conditioned specimens (c) fatigue-induced failures
Figure 3: Load-deflection behavior of SFRP and representative companion CFRP specimens.

4. CONCLUSIONS
Steel fiber reinforced polymer (SFRP) flexural retrofit measures were compared with comparable CFRP measures
subject to fatigue loading. The following primary observations and conclusions were drawn from this study:
1. SFRP is more axially flexible than CFRP resulting in a greater application width required to affect a comparable
retrofit. This results in decreased interfacial stresses and improved fatigue behavior with no evidence of
debonding for the (relatively high) stress ranges tested in this program.
2. The SFRP-retrofit specimens are initially stiffer than the comparable CFRP specimens resulting in a marginally
increased ultimate capacity. The deformation capacity of comparable systems is, however, essentially the same.
3. A thinner and more uniform adhesive bondline was obtained with the SFRP system. This may account for both
the marginally improved global beam behavior and the apparent earlier onset of debonding.
4. The previously-observed deleterious affects of fatigue loading on bonded CFRP systems appear less significant
in bonded SFRP systems based on this initial limited study. More study is required to identify an appropriate
reduction factor to account for fatigue effects on bonded SFRP systems.

6. REFERENCES
Hardwire LLC (2002) www.hardwirellc.com, Pocomoke City, MD.
Harries, K.A. (2005) “Fatigue Behavior of Bonded FRP Used for Flexural Retrofit” International Symposium on
Bond Behavior of FRP in Structures (BBFS 2005), December 7-9, 2005, Hong Kong. pp 555-560.
Harries, K.A. and Aidoo, J. (2005) “Deterioration of FRP-to-Concrete Bond Under Fatigue Loading” International
Symposium on Bond Behavior of FRP in Structures (BBFS 2005), December 7-9, 2005, Hong Kong. pp 561-566.
Harries, K.A., Zorn, A. and Reeve, B. (2006) “Effect of Adhesive Modulus on the Monotonic and Fatigue Behavior
of Externally Bonded CFRP Strips” Proceedings of the Third International Conference on FRP Composites in
Civil Engineering (CICE 2006), December 13-15 2006, Miami, Florida, USA. paper #026.
Huang, X., Birman, V., Nanni, A. and Tunis, G. (2005) “Properties and potential for application of steel reinforced
polymer and steel reinforced grout composites” Composites Part B, Vol. 36 pp 73-82.
Minnaugh, P. (2006) “Performance Characteristics of Steel Reinforced Polymer Composite Flexural Retrofit
Measures Subject to Monotonic and Fatigue Loads” M.Sc. Thesis, University of Pittsburgh. December 2006.
Oehlers, D.J. (2005) “Generic Debonding Mechanisms in FRP Plated Beams and Slabs” Proceedings of the
International Symposium on Bond Behavior of FRP in Structures, December 7-9, 2005, Hong Kong. pp 353-44.
Quattlebaum, J., Harries, K.A. and Petrou, M.F. (2005) “Comparison of Three CFRP Flexural Retrofit Systems
Under Monotonic and Fatigue Loads” ASCE Journal of Bridge Engineering. Vol. 10, No. 6 pp 731-740.
Zorn, A. (2006) “Effect of Adhesive Stiffness and CFRP Geometry on the Behavior of Externally Bonded CFRP
Retrofit Measures Subject to Fatigue Loads” M.Sc. Thesis, University of Pittsburgh. April 2006.

286
Third International Conference on FRP Composites in Civil Engineering (CICE 2006)
December 13-15 2006, Miami, Florida, USA

Fatigue Performance of Concrete Bridge Deck Slabs Reinforced

With Glass FRP Composite Bars
Amr El-Ragaby
Ph. D. Candidate, Department of Civil Engineering, University of Sherbrooke, Sherbrooke, Quebec, Canada

Ehab El-Salakawy
Canada Research Chair Professor in Advanced Composite Materials and Monitoring of Civil Infrastructures,
Department of Civil Engineering, University of Manitoba, Winnipeg, Manitoba, Canada

Brahim Benmokrane
NSERC Research Chair Professor in Innovative FRP Composite Materials for Infrastructures, Department of Civil
Engineering, University of Sherbrooke, Sherbrooke, Quebec, Canada

ABSTRACT
Recently, glass FRP (GFRP) bars have been widely used as internal reinforcement for concrete bridge deck slabs
subjected to harsh environmental and loading conditions due to their lower cost compared to other kinds of FRPs.
There is a lack of data on the performance of FRP-reinforced concrete structural members subjected to cyclic fatigue
loading. This research is designed to investigate both the fatigue behavior and life of concrete bridge deck slabs
entirely reinforced with GFRP bars. A total of five full-scale deck slabs (3000×2500×200 mm) were constructed and
tested under concentrated cyclic loading until failure. Different scheme of variable amplitude cyclic loading as well
as different reinforcement types (steel and GFRP), ratios and configurations for top and bottom reinforcement in
both directions were investigated. Results are presented in terms of deflections, strains in concrete and FRP bars and
crack widths at different levels of cyclic load.

KEYWORDS
Concrete, bridges, deck, FRP, fatigue.

1. INTRODUCTION
Fatigue performance is an important limit state that must be considered by designers of structures subjected to cyclic
loading such as bridges. The most susceptible bridge element to fatigue loading is the deck slab since it directly
sustains repeated moving wheel loads. FRPs, as non-corrosive materials, have been investigated by researchers as a
potential substitute for steel reinforcement in concrete structures to overcome corrosion related problems. FRP
materials possess the necessary property of high tensile strength that makes them attractive as structural
reinforcement for concrete. The behavior of FRP-reinforced concrete elements is different from those reinforced
with steel bars. This is mainly because the FRP materials have relatively low modulus of elasticity, small transverse
strength, and different bond characteristics compared to those of steel (ACI 2003). As concrete bridge deck slabs
are governed by long-term fatigue endurance and durability of constituent materials, it is necessary to understand the
fatigue behavior of such structures especially when using new materials such as GFRP bars. This research is
designed to study the fatigue behavior of concrete bridge deck slabs reinforced with different configurations of
GFRP bars under fatigue loads.

2. DETAILS OF THE EXPERIMENTAL PROGRAM

2.1. Test Prototypes

The experimental program includes five full-size bridge deck prototypes (2500 mm width, 3000 mm length, and 200

287
mm thick). Four deck slab prototypes were reinforced with different reinforcement ratios and configurations of
GFRP bars and one slab was reinforced with conventional steel bars as a control. Bottom and top concrete covers of
38 mm were used for all slab prototypes. The main bottom transverse GFRP reinforcement for the four GFRP
reinforced decks, S1, S2, S3, and S4 was calculated based on the empirical design method recommended by the
updated version of Section 16 of the CHBDC, Clause 16.8.7.1 (CSA 2006). This approach results in using 1#19
GFRP @150 mm in the bottom transverse direction with reinforcement ratio of 1.2%. The longitudinal bottom
reinforcement for the four slabs consists of 1#16 GFRP @ 200 mm with a reinforcement ratio of 0.6%. For the top
reinforcement layer, different configurations and reinforcement ratios were used. For slab S1 and S2, 1#16 GFRP @
200 was used in both directions with a reinforcement ratio of 0.6% (slabs S1 and S2, had identical bottom and top
reinforcement). For S3, a minimum reinforcement ratio of 0.3% was used in both directions, which results in using
1#13 GFRP @ 300 mm in each direction. Slab S4 had no top reinforcement. The control slab S0, reinforced with
steel bars, was designed according to the empirical method of the CHBDC, Clause 8.18.4.2 (CSA 2000), which
resulted in using 1#10M @ 210 mm in all directions.

2.2. Materials Properties

All test specimens were constructed using normal weight concrete with a targeted 28-day concrete strength of 37
MPa. Following the curing period in laboratory environment (14 days), all deck slab prototypes were stored out-
doors in real environmental conditions for at least one year. This was done to simulate the environment that a real
bridge will undergo to allow for the formation of concrete cracking that arise mainly from environmental conditions.
Also, this storage period helped to stabilize the concrete properties (compressive strength and modulus of elasticity).
The average concrete compressive and tensile splitting strengths, obtained from standard tests, were approximately
41 and 3.9 MPa, respectively. In addition to CSA grade 400 steel bars, sand-coated GFRP V-RODTM (Pultrall Inc.
2004) was used. These GFRP bars are made of high-strength E-glass fibers (73% fiber by volume) with a modified
vinyl ester resin, additives, and fillers using a pultrusion process. The GFRP bars have an average modulus of
elasticity and ultimate strength of 45 GPa and 700 MPa, respectively.

2.3. Test Set-up and Repeated Fatigue Loading

All slabs were tested under a single concentrated load at the centre of a clear span of 2000 mm. This load was
applied through a 75-mm thick steel plate that measures 250 × 600 mm, which is equivalent to the footprint of a
wheel as specified by the CHBDC (CSA 2000). The edges of the slabs were partially restrained against lateral
displacement using a special arrangement of steel sections and bolts (El-Gamal et al. 2005). Moving vehicular loads
were simulated by stationary concentrated load varying cyclically in magnitude. In this research, an accelerated
fatigue loading scheme was used. It consists of variable amplitude fatigue loading where all the slabs were subjected
to sinusoidal waveform fatigue loading steps. Each step consists of 100,000 cycles with a minimum constant load
level and peak variable load levels for each step. The peak loads were equivalent to the multipliers of the load level
for fatigue limit state, Pfls (Pfls = 87.5*1.4*1.0 = 122.5 kN according to Clause 3.5.1 (CSA 2000). If the test
prototype completed the last 100,000 at the largest peak load (490 kN) without failure, the test continued at the same
peak load level until failure. This fatigue loading scheme was applied to all slabs except slabs S2 and S3. For slab
S3, an extra 300,000 cycles at lower peak loads (100,000 at each of 122.5, 153.12 and 214.37 kN peak loads) were
applied to the deck slab to assess the effect of cycling at lower peak load levels. For slab S2, 4,000,000 load cycles
between a minimum load of 15 kN and a peak load of 122.5 kN at a rate of 4 Hz was applied. This fatigue loading
scheme represents twice the fatigue life recommended by many researches for bridge deck slabs (Kumar and
GangaRao 1998; Mufti et al. 2005).

3. TEST RESULTS AND OBSERVATION

Any deterioration of the deck slab would be evident from increasing of deflections, strains in reinforcing bars and
concrete, crack widths as well as from the loss of deck stiffness with the increase of the number of load cycles or the
peak loads. This deterioration will be referred to as damage and the overall progressive deterioration due to the total
number of load cycles will be referred to as accumulated damage. These parameters were measured during the
monotonic loading tests that were carried out at the end of each fatigue loading step. Figure (1) summarizes the
number of cycles that each slab sustained under each fatigue load step until failure. It can be noticed that the steel
reinforced slab, S0, has the shortest fatigue life compared to the other tested slabs as it sustained only 300,000 cycles
at the first three fatigue loading steps and failed after only 120 cycles at the peak load of 450 kN. Slabs S1, S3 and
S4 have similar results although they have different top reinforcement ratios. This similar behavior included slab S3,

288
which was subjected to additional 300,000 cycles at lower peak loads (100,000 at each peak load of 122.5, 153 and
214 kN). This may indicate that applying repeated loads with amplitude equal or up to 1.75 times the fatigue limit
state has insignificant effect on the GFRP-reinforced concrete deck slabs. It should be also noted that after being
subjected to 4,000,000 load cycles at the peak load of 122.5 kN, slab S2 did not fail.

Figure (2) shows a comparison between the static responses of the five tested slabs after the 367 kN fatigue load
step. All GFRP-reinforced slabs had approximately the same fatigue damage although they have different
reinforcement ratios in the top layers. It is clear that the damage accumulated to slab S1 and S4 was similar (both
have the same residual deflection and stiffness) although slab S4 had no top reinforcement. Also Slab S3, which was
subjected to more load cycles than S1 and S4, had the same accumulated fatigue damage. For the steel reinforced
slab S0, it suffered more deterioration with an accumulated damage of about 2.5 times greater compared to the other
tested slabs. This was mainly due to the big difference between the modulus of elasticity of steel and concrete and
the mechanical bond mechanism, which causes much damage to concrete during cyclic loading.

Similar behavior and changes were observed for the maximum measured strains in reinforcing bars. Figure (3)
shows comparisons between the maximum measured strains in the reinforcing bars in the transverse direction
following the 183 kN fatigue loading step. Although slab S3 has completed 300,000 cycles (at lower peak load
levels) more than slab S1, the difference in the measured strains in the GFRP bars did not exceed 10%. For the
GFRP bars, the maximum recorded strain, which was approximately 2400 micro-strain, is still less than 20% of the
ultimate strain. For steel, this value was 1700 micro-strain (approximately 85% of the yield strain of the steel). Also,
the maximum measured strain in slab S1 was about three times that of slab S2. Figure (4) presents the maximum
measured concrete strain in the transverse direction following the 367 kN fatigue loading step. The largest strain of
approximately 1250 micro-strain was measured for the steel reinforced slab, S0. While the GFRP-reinforced slab
without top reinforcement, S4, recorded the lowest value of about 400 micro-strain.

250
Slab S0
Slab S1 200 S3 S0
100000 Slab S3
S1
Load (kN)

150
80000 Slab S4
No. of Cycle

S4
60000 100
Slab S4
40000 Slab S3
Slab S1 50
20000
ab
Sl

Slab S0
0 0
122.5 153.1 183.8 204.4 245 367.5 425 450 490
0 5 10 15 20
Peak Load (kN) Deflection (mm)
Figure 1. Number of cycles for all slabs to failure Figure 2. Comparison between static responses
after 367 kN fatigue loading steps

250 250

200 S0 S1 S3 S4 200
S0
Load (kN)
Load (kN)

150 150

100 S1
100

S3
50 50

0 0
0 500 1000 1500 2000 2500 -1500 -1000 -500 0
Strain (Microstrain) Strain (Microstrain)
Figure 3. Comparison between transverse bar Figure 4. Comparison between compressive strains at top
strains after 183 kN fatigue load step face of concrete after 367 kN fatigue load step

The cracks propagated and grew (became wider) with the increase of the number of load cycles and the peak load.
At early stages of fatigue loading, cracks were observed on the bottom face of the deck where some major cracks

289
propagated in the longitudinal direction parallel to the supports. A few cracks were observed at the midspan in the
transverse direction beneath the loaded area. With the increasing load cycles, more cracks were developed in the
transverse and radial direction forming a grid-like pattern. No cracks were observed on the top face around the
periphery of the loading plate until failure occurred. When a relatively stable condition was reached, approximately
after the 245 kN fatigue loading step, the crack growth reduced considerably. However, cracks widths and depths
continued to increase. All the five slabs failed in punching shear after sustaining a different number of load cycles at
the final peak load (450 or 490 kN). The top surface of the failure zone had an elliptical shape around the corners of
the loading plate while the bottom surface has approximately a circular shape with a diameter equal to the spacing
between the two supporting girders. Figure (5) shows both the top and bottom faces of slab S1 after failure.

a- Bottom face b- Top face

Figure 5. Cracks pattern of tested slabs at failure (Punching shear failure mode)

4. CONCLUSIONS
1- The punching shear is the mode of failure of concrete bridge deck slabs reinforced with steel or glass FRP
composite bars under fatigue loads;
2- The glass FRP reinforced concrete bridge deck has a better fatigue performance and longer fatigue life; about 2.5
times more than the steel reinforced ones. This may be due to the close value of the modulus of elasticity for
GFRP composite bars and concrete and the linear-elastic behavior of GFRP bars up to failure;
3- Deterioration of concrete deck slabs subjected to fatigue loads can be noticed through the cumulative damage to
the deck slab observed from the increase in both residual and elastic deflection and strains in both reinforcing
bars, and concrete;
4- Fatigue loading at a peak load level up to 1.75 times the fatigue limit state load does not cause any significant
damage to concrete bridge deck slabs with span to depth ratio less than 15;
5- The top reinforcement has a little effect on the fatigue performance of concrete bridge deck slab at lower peak
load levels (less than twice fatigue limit state load); this effect becomes more significant at higher peak loads;
6- The proposed FRP reinforcement ratio adopted by the updated version of Section 16 of CHBDC (CSA 2000) is
adequate to meet the fatigue strength and fatigue life requirements of concrete bridge decks.

5. REFERENCES
ACI 440.1R-03, (2003). Guide for the Design and Construction of Concrete Reinforced with FRP Bars, American
Concrete Institute, Farmington Hills, Michigan, 41p.
Canadian Standards Association (CSA). (2000). “Canadian Highway Bridge Design Code, CHBDC.” CAN/CSA-
S6-00, Canadian Standards Association; Rexdale, Toronto, Ontario, Canada, 734 p.
Canadian Standards Association (CSA). (2006). “Canadian highway bridge design code- Section 16, updated
version for public review.” CAN/CSA-S6-06, Toronto.
El-Gamal, S., El-Salakawy, E.F., and Benmokrane, B. (2005). “Behaviour of Restrained FRP-Reinforced Bridge
Decks under Wheel Loads." ACI Structural Journal, Vol. 102, No. 5, pp. 727-735.
Kumar, S. and GangaRao, H.V.S. (1998). “Fatigue Response of Concrete Decks Reinforced with FRP Rebars.”
Journal of Structural Engineering, ASCE, 124(1), 11-16.
Mufti, A.A., Memon, A.H., and Klowak, C. (2005). “Study of Static and Fatigue Behaviour of Second Generation
Steel Free Concrete Bridge Decks” Proceeding of the International Workshop on Innovative Bridge Deck
Technologies, Winnipeg, Manitoba, pp. 49-61.
Pultrall Inc. (2004). “V-RODTM – Technical Data Sheet.” ADS Composites Group Inc., http://www.pultrall.com,
Thetford Mines, Quebec, Canada.

290
Third International Conference on FRP Composites in Civil Engineering (CICE 2006)
December 13-15 2006, Miami, Florida, USA

FAILURE PROCESS OF ADHESIVELY BONDED JOINTS FROM PULTRUDED GFRP ADHERENDS

Ye Zhang and Thomas Keller

(Ecole Polytechnique Fédérale de Lausanne EPFL-CCLab,
CH-1015 Lausanne, Switzerland)

ABSTRACT
Adhesively bonded joints are being used increasingly in civil engineering, especially for joints comprising pultruded
glass-fiber reinforced polymer (GFRP) laminates. The layered material architecture, however, leads to a complex
delamination failure within the pultruded material, thus necessitating understanding of the progressive failure
mechanism of such joints under axial tensile loading. In this work, adhesively bonded joints composed of pultruded
GFRP laminates, including double and stepped lap joints, were experimentally investigated. The static strengths of
joints were obtained and the failure mechanism was understood. Crack propagation and back face strain gages were
successfully employed to identify crack initiation and describe crack propagation, even though the failure
mechanism was always brittle. The dominant failure mode for both types of joints was a fiber-tear-off failure that
occurred in the mat layers of the GFRP laminates. The critical strain energy release rate was calculated and the
resulting values from double lap and stepped lap joints compared very well.

KEYWORDS
Adhesive joints, failure, pultrusion, strength.

1. INTRODUCTION
Adhesively bonded joints are being used increasingly in civil engineering, especially joints composed of pultruded
GFRP laminates. Due to the complicated material architecture of pultruded composites, it is important to understand
the failure mechanism of the joints under both static and cyclic loading, as well as the progressive failure process
considering crack initiation and propagation.
Considerable experience on the mechanical behavior of adhesively bonded FRP joints under static and fatigue loads
has already been gathered for aerospace and automotive structures (Abdel-Wahab et al., 2004). However, when
these applications are compared with adhesively bonded FRP joints in civil infrastructure, there are essential
differences in the manufacturing process, the material architecture, the dimensions of the components and the
application environments (Keller and Vallée, 2005). It is therefore urgent to fill the gap of knowledge regarding the
fracture behavior of adhesively bonded joints for the civil engineering sector.
In the present work, the quasi-static behavior of adhesively bonded double and stepped lap joints from pultruded
GFRP laminates and an epoxy adhesive are investigated. These types of joints are widely used in existing civil
engineering structures, an example is shown in Fig. 1.

2. EXPERIMENTAL PROGRAM
Two types of specimens were investigated: balanced double lap joints (DLJ) and stepped lap joints (SLJ), both
composed of pultruded GFRP laminates of 50 mm width and 6 or 12 mm thickness. The pultruded GFRP laminates
(delivered by Fiberline A/S, Denmark) consisted of E-glass fibers embedded in an isophtalic polyester resin. A two-
component epoxy was used (SikaDur 330 from Sika) for the specimens. The adhesive layer thickness was 2 mm.
The dimensions and number of the specimens are shown in Fig. 2 and Table 1 respectively.
Four crack propagation gages (HBM/RSD20) were used to detect the crack initiation and propagation in three
specimens of each joint type, as shown in Fig. 3 (labeled C) and Table 1. The crack propagation gages consisted of
20 wires, spaced at 1.15 mm intervals, perpendicular to the adhesive layer, and covered almost half of the overlap
length. In each specimen, two back face strain gages (D) were placed above the locations where crack initiation was
expected to influence the strain response and two more strain gages were used to measure the axial strains (A)
outside the joint where the stresses were expected to remain uniformly distributed.

291
Figure 1: DuraSpan bridge deck with stepped lap joint Figure 2: Double lap and stepped lap joint

Figure 3: Experimental instrumentation for DLJ (above) and SLJ (below)

Furthermore, two displacement transducers (B) were employed to measure the elongation of the joint. An Instron
Universal 8800 hydraulic machine was used to apply the axial force with a displacement rate of 0.5 mm/min for the
SLJs and of 1 mm/min for the DLJs.

3. EXPERIMENTAL RESULTS
The double lap joints had an almost linear load-elongation response up to the failure, as shown in Fig. 4, which was
very brittle. In contrast, the stepped lap joints exhibited a discontinuous two stage behavior, although the final
failure was also brittle. For the SLJs, a failure in the adhesive in the two small gaps perpendicular to the longitudinal
adhesive layer occurred, which was followed by a steep decrease of the load. After this first failure, however, the
joints continued to sustain an increase in load, although the crack initiation and propagation led to a decrease in
overall stiffness. The average static ultimate loads of the double lap and stepped lap joints were 45.6 kN and 10.5 kN
respectively (see Table 1).
The typical failure modes of the DLJs and SLJs are shown in Fig. 5. For both types of joints, the dominant failure
mode was a fiber-tear-off failure in the GFRP laminates. Failure initiation occurred in the two outer mat layers of
the 12 mm laminates below the ends of the outer 6 mm laminates in the DLJs, and in the 6 mm laminates below the
(already cracked) small gaps in the SLJs. Failure propagation then occurred in the same mat layers up to final joint
failure. Failures were brittle and sudden and their initiation and propagation were normally not observable by the
naked eye.
Figure 6 shows typical measured crack gage resistances with increasing load for the DLJs and SLJs. A jump in
resistance occurred if one of the twenty wires of one gage was cut. The results from three of the gages on one of the
DLJs are shown on the left. The crack initiated simultaneously on both sides of one joint end at a load of 26.9 kN
(locations 1 and 2 in Fig. 6, left). The crack then did not grow up to approximately 41.5 kN. Subsequently, it
propagated rapidly and cut the remaining gage wires up to final failure. On the other joint edge, crack initiation also
occurred at 26.9 kN, but only at one side (3 in Fig. 6).

292
 60 12

50 10

Load [kN] 40 8

Load [kN]
30 6

20 4

Specimen 1 Specimen 1
10 Specimen 2 2 Specimen 2
Specimen 3 Specimen 3

0 0
0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.6 1.8 2.0 0.0 0.2 0.4 0.6 0.8 1.0 1.2
Elongation [mm] Elongation [mm]

Figure 4: Selected load-elongation curves of DLJ (left) and SLJ (right)

Table 1. Ultimate loads of DLJ and SLJ (* with crack propagation gages)

DLJ SLJ
Specimen Ultimate load [kN] Adhesive failure [kN] Ultimate load [kN]
1* 42.1 4.9 10.3
2* 49.0 9.0 10.3
3* 47.5 8.9 11.1
4 45.1 7.0 10.2
5 45.9 7.0 10.0
6 43.7 5.1 10.9
Average 45.6±2.0 7.0±1.9 10.5±0.5

Figure 5: Failure modes of DLJs (left) and SLJs (right) (DLJ with broken crack gage)

3.20 3.20
3.04 3.04
2.88 2.88
2.72 2.72
2.56 2.56
2.40 2.40
2.24 2.24
Resistance [Ohm]

2.08
Resistance [Ohm]

2.08
1.92 1.92
1.76 1.76
1.60 1.60
1.44 1.44
1.28 1.28
1.12 1.12
0.96 0.96 Crack gage 1
0.80 Crack gage 1 0.80
0.64 Crack gage 2
0.64
Crack gage 3 Crack gage 3
0.48 0.48
0.32 Crack gage 4 0.32 Crack gage 4
0.16 0.16
0.00 0.00
0 10 20 30 40 50 0 2 4 6 8 10 12
Load [kN] Load [kN]

Figure 6: Measured crack propagation gage resistance versus load of typical DLJ (left) and SLJ (right).

293
Subsequently, only two more points were caught. The cutting of the remaining wires and of the wires of the fourth
gage (4 in Fig. 6) could not be recorded due to the too rapid crack propagation, even at a measurement frequency of
800 Hz. For the SLJs, crack initiation in the laminates occurred on each of the four potential locations at a different
load (between 4.7 and 6.3 kN). The cracks then propagated at different rates up to the ultimate load. The results
from the back face strain gages are shown in Zhang and Keller (in review).

4. DISCUSSION
The dominant failure mode was the same fiber-tear-off failure in the outer mat layers of the laminates for both joint
types. Comparing the ultimate loads of both joint types, the average DLJ ultimate load was more than two times
higher than the double of that of the SLJ (the double has to be taken to equalize the shear surface). This much higher
performance of DLJ was mainly due to the upkeep of a symmetric joint configuration. The adhesive failure in the
small SLJ gaps led to a system change from a stepped lap to an asymmetric single lap configuration, showing much
smaller stiffness and strength.
The crack initiation in the laminate could be caught by employing both crack propagation gages and back face strain
gages. The measured average loads at which crack initiation in the laminates occurred according to crack
propagation and back face gage measurements compared very well (see Zhang and Keller, in review).
The critical strain energy release rate values were calculated at a crack length of a = 25 mm for each crack, which
corresponds to half of the overlap length (Hadavinia et al. 2003). The resulting average values were 804.9 [J/m2] for
the DLJs and 721.9 [J/m2] for the SLJs. This small difference is supported by theory, which states that the critical
strain energy release rate depends only on the material properties. That is, since the two joint types showed the same
failure mode and failed in the same material, their two Gc values must be identical.

5. CONCLUSIONS
Adhesively bonded double and stepped lap joints composed of pultruded GFRP laminates and epoxy adhesives were
experimentally investigated under quasi-static axial tensile loading. The following conclusions were drawn:
(1) Double lap joints showed a typical, almost linear behavior under axial tensile stresses, up to sudden and brittle
failure. Stepped lap joint showed two stages of failure: adhesive failure in the small joint gap perpendicular to
loading and failure of the whole joint. The gap failure changed the joint configuration from stepped lap to single lap
with a corresponding decrease in joint stiffness and strength - a major drawback of this joint type. The dominant
failure mode for both types of joints was a fiber-tear-off failure, which occurred in the outer mat layers of the GFRP
laminates.
(2) Though the dominant failure mode of both types of joints was considered as very brittle and sudden, it was still
possible to observe the crack initiation using both crack propagation gages and back face strain gages.
(3) The critical strain energy release rate, Gc, was calculated. The resulting values from double and stepped lap joints
compared very well. The experimental results provided useful results to perform a numerical study on the
progressive failure process in the next project stage.

6. ACKNOWLEDGEMENTS
The authors would like to thank the Swiss National Science Foundation (Grant No 200021-103866/1), Sika AG,
Zurich (supplier of the adhesives) and Fiberline Composites A/S, Denmark (supplier of the pultruded laminates) for
the support of this research.

7. REFERENCES
Abdel-Wahab, M.M., Ashcroft, I.A., Crocombe, A.D., and Smith, P.A (2004). “Finite element prediction of fatigue
crack propagation lifetime in composite bonded joints”. Composites Part A, 35, 213-222.
Hadavinia, H., Kinloch, A.J., Little, M.S.G., and Taylor, A.C. (2003). “The prediction of crack growth in bonded
joints under cyclic-fatigue loading. II. Analytical and finite element studies”. International Journal of Adhesion &
Adhesives, 23, 463-471.
Keller, T., and Vallée, T (2005). “Adhesively bonded lap joints from pultruded GFRP profiles, Part I: stress-strain
analysis and failure modes”. Composites: Part B, 36, 331–340.
Zhang, Y., and Keller, T. (in review). “Progressive failure process of adhesively bonded joints composed of
pultruded GFRP”. Composites Science and Technology.

294
Third International Conference on FRP Composites in Civil Engineering (CICE 2006)
December 13-15 2006, Miami, Florida, USA

PREDICTION OF EXTENED FATIGUE LIFE OF STRENGTHENED

BRIDGE DECK WITH FRPs
Hongseob Oh
(Professor, Jinju National University, Jinju, Kyoungnam, Korea)

Jongsung Sim
(Professor, Hanyang University, Ansan, Kyeonggi-do, Korea)

Minkwan Ju
(Candidate of PH.D, Hanyang University, Ansan, Kyeonggi-do, Korea)

ABSTRACT
Concrete bridge decks are damaged directly by traffic loads. However almost all fatigue studies on model concrete
decks have been performed under a concentrated pulsating load applied at a fixed point. Therefore, a fatigue analysis
under a realistically simulated moving load is needed to evaluate the performance of a concrete bridge deck more
accurately. In this study, a live load model was used to formulate an approach to calculating extreme load for
various time periods. The maximum load effects for time periods from day one to 75 years are derived by
extrapolations and simulations. In order to predict the strengthening effect of bridge decks strengthened with one
fiber sheet, fatigue tests were performed and probabilistic and reliability analyses on the fatigue of concrete deck
were carried out. According to the result of the analysis, strengthened concrete decks are found to have sufficient
resistance effect against increasing truck during service life of the bridge.

KEYWORDS
live load model, concrete bridge deck, pulsating load, probability and reliability analysis.

1. INTRODUCTION
Because many previous studies to assess the fatigue performance of FRP strengthened deck panel are limited on the
work in the laboratory and tested on the member subjected not the moving wheel loads is similar with real condition
of bridge decks but the pulsating loads, it is hard to expect the strengthening effect obtained from the test to real
bridge deck, when the verified strengthening techniques based on the test results applied on the real deteriorated
bridge decks. These differences of strengthening effect occurred between in field and in laboratory mainly caused by
either the workmanship of strengthening or the gap of boundary conditions or loading conditions. Especially,
irregularity of traffic loads induced the structural deterioration of deck, including the velocity and the moving route
as well as the weight, cannot be simulated in laboratory.
The properties of the decks were typical of bridges for secondary design truck, i.e., DB-18 given by the Korea
Highway Bridge Specification. The applied design traffic load for this category is similar to the HS-20 truck in the
AASHTO design specification. Fatigue failure of bridge deck gradually progressed not under ultimate loading state
but under service loading state, and traffic load provided in design specification is only considered ultimate state to
design the load carrying capacity of deck panel. Therefore, the model to simulate fatigue loads for either fatigue
analysis or fatigue design of bridge deck need to develop. In this study, authors try to propose the live loading model
to assume the real vehicle loads and conduct the fatigue analysis based on the proposed live traffic model.
Probabilistic approaches and reliability analysis based on the restricted test results on the strengthened concrete
bridge decks is applied to verification of strengthening effect and prediction of extended fatigue life, because fatigue
performances on the bridge decks investigated from either laboratory test or real bridge decks are limited and also
involved in inevitable uncertainties and extensive variation.

295
2. LIVE LOADS MODEL
2.1 Axle loads of truck Margins

In this study, the live load model was developed by actually measuring the passing frequency and lane load of the
heavy vehicles on the Banpo Great Bridge, Seoul, Korea. Yoon (1996) classified 10 kinds of vehicles in Korea then
measured each weight of vehicles. Among the vehicle types, the vehicle types of 5(T2), 6(TT3), 7(ST4), 8(ST5),
9(ST6), and 10(FT5) are classified as heavy vehicles as shown in Fig. 1. In the case of traffic model, TT3 and ST5
truck types were selected as because these types showed more the traffic volume than other truck types. In this study,
TT3 truck type was selected because it was investigated that TT3 type showed more traffic volume than ST5 type
relatively. To adopt fatigue load to bridge deck, normal and overloaded loading state was considered. For lane load
of normal and overloaded state, it was selected as a value of 84.0kN and 133.9kN, respectively. Actually, the upper
tail parts of the distribution of axle weights are mainly concerned in the fatigue life of reinforced concrete deck slabs.
In this study, ADTT (Average Daily Truck Traffic) provided by Korea Ministry of Construction and Transportation
was used to simulate the traffic volume of the bridge designed as 2nd grade. The coefficient of variation is 0.85 based
on ASSHTO [1998] with a one way three-lane road. But for considering critical condition of fatigue load,
coefficient of variation in this model was selected as 1.0 so that ADTT was acquired as 4,400.

SA SA ST DT SA TA TA ST DTDT DTDT
(a) Type T2 (d) Type ST5

DTDT
ST DTDT SA TA TRA ST DT
SA TA DT
(b) Type TT3 (e) Type ST6

SA SA TA ST DT DTDT SA TA SA SA ST DTDT ST ST
(c) Type ST4 (f) Type FT5

Figure 1: Truck Type [Yoon, 1996]

Table 2: ADTT [Korea Ministry of Construction and Transportation, 2000]

Truck
Passenger
Bus Overloaded
Vehicle Design Truck
Truck
One way traffic 61,477 8,755 11,703 1,600
3,901 533
ADTT 20,326 2,918
4,434 (approximately 4,400)

2.2 Live load model

It has been studied that the moving and cyclic load can affect reinforced concrete bridge deck more 4 to 7 times than
that of pulsating load. In present, however, only pulsating load has been adopted for fatigue analysis of the bridge
deck. Therefore, the live load model for fatigue analysis of the bridge deck was developed by using Type I extreme
function adopted actually measured lane load from Table 1. The cumulative distribution function (CDF) of the Type
I asymptotic form for the distribution of the largest value, in the classification of Gumbel (1958), is as follows:

FYn ( y n ) = exp[−e −α n ( yn −un ) ] (1)

where u n and αn are, respectively, the location and scale parameters, defined as follows:
u n = The characteristic largest value of the initial variate X
αn = An inverse measure of dispersion of X1
The corresponding probability density function (PDF) is
f Yn ( y n ) = α n e −α n ( yn −un ) exp[−e −α n ( yn −un ) ] , − ∞ < y n < ∞ (2)

296
 In this study, coefficient of variation and average nominal ratio for adopting axle load to extreme function was
assumed the value of 0.25 and 1.24 from Nowak (1994). Fig. 3 and 4 shows the result of live load model of axle
load for the extreme function and the service life of bridge assumed as 75 years based on Design Specification of
Highway Bridge in Korea. Fig. 2(a) and 3(a) indicates the axle load effect of design and overloaded truck as
Probability Density Function (P.D.F.) and Fig. 2(b) and 3(b) shows Cumulative Distribution Function (C.D.F.) and
this function presents similar shape with Nowak’s live load model.

(a) Probability Density Function (b) Cumulative Distribution Function

Figure 2: Live load model (normal state tandem load: 84.0kN)

(a) Probability Density Function (b) Cumulative Distribution Function

Figure 3: Live load model (overloaded state tandem load: 133.9kN)

3. TEST PROPRAM

The CFS(Carbon Fiber Sheet) and GCFRP(Grid-type Carbon FRP) materials were bonded to the prototype deck
panel in an upside-down position. Each fiber sheet was attached to the epoxy-coated surface by pressing it into the
epoxy. The GCFS material was first fixed to the concrete surface in an upside-down position using 2.5-cm length
anchor bolts spaced every 50 cm. Then the repair mortar suggested by the manufacturer was overlayed on the
concrete surface. The nine specimens listed in Table 3 were subjected to cyclic loads to investigate their fatigue
failure characteristics. The CON means the unstrengthened reference panel. Fig. 4 and 5 shows test specimens. [Oh,
2001]
Table 3: Stress level

SPECIMENS CON CFS GCFRP

Stress Level (%) 40 70 90 60 70 80 60 70 80
Loads (kN) 260 450 580 440 510 590 430 500 570

297
 φ1 6 - 5 @ 3 0

3 12 3
30 18
150 150
10 10

B o tto m
φ1 6 - 1 1 @ 1 5

10 10

B o tto m
20 20
Top 160

φ16- 11@ 20

φ16- 23@ 10

200

200
240
CL
B o tto m

: S tra in G a g e
: LV D T (a) CFS (b) GCFRP

Figure 4: Details of specimen Figure 5: Strengthening details

4. EXPECTATION OF EXTENDED LIFE

To expect the strengthening age of the 2nd grade bridge deck, the rate of increase of overloaded truck passing on the
target bridge was assumed as 12% based on ADTT data (see Table 2) in this study. To evaluate the extended life of
the bridge deck strengthened with CFS and GCFRP, the target reliability index was assumed as 4.7 and 5.0 for the
strengthening or replacement age, respectively. Fig. 6 shows reliability index with the rate of increase 12% from the
probability analysis.
12.0

10.0
G C FR P
8.0

β6.0
C FS

4.0
CO N

2.0

0.0
0 10 20 30 40 50 60 70 80
y e a r s

Figure 6: Reliability index (when axle load has the rate of increase 12%)

5. CONCLUSIONS
(1) It was evaluated that the extended fatigue life of bridge deck strengthened with CFS or GCFRP showed as more
1.2 ~ 1.5 times than that of non-strengthened bridge deck.
(2) In the case of the strengthening type, GCFRP strengthening method had better reliability index than CFS
strengthening method in the point of view for strengthening and extended fatigue life. Therefore, it is evaluated that
the GCFRP strengthening method to concrete bridge deck is more effective that CFS strengthening.

REFERENCE
Gumbel, E. J. (1958). Statistics of Extremes, Columbia Univ. Press, New York
Nowak, A. S., Yamani, A. S., and Tabsh. (1994). “Probabilistic models for resistance of concrete bridge girders”,
ACI structural journal, Vol. 91, No. 3, pp. 269-276
Oh, H. (2001). “Strengthening Design Method of Bridge Deck Reinforced by Carbon Fiber Sheet”, Ph.D. thesis,
Hanyang University, Ansan, Korea.
Yoon, S. (1996). “A Study on the Fatigue Behavior and the Fatigue Design of the Reinforced Concrete Deck Slabs
of the Composite Bridges”, Ph.D. thesis, Seoul National University, Seoul, Korea.

298
Third International Conference on FRP Composites in Civil Engineering (CICE 2006)
December 13-15 2006, Miami, Florida, USA

STATIC AND FATIGUE INVESTIGATION OF BRIDGE DECK

CANTILEVERS
Chad Klowak
Engineer-in-Training, ISIS Canada Research Network, Winnipeg, MB, Canada

Aftab Mufti
President, ISIS Canada Research Network, Winnipeg, MB, Canada

Baidar Bakht
President, JMBT Structures Research Inc.,Scarborough, ON, Canada

ABSTRACT
One solution to the problem of deteriorating bridge decks is to replace or rehabilitate them with innovative
technologies such as steel-free bridge decks or decks reinforced with fibre reinforced polymers. Cantilevers are an
integral part of bridge decks and this paper outlines test results obtained from static and fatigue destructive testing
conducted at the University of Manitoba’s W.R. McQuade Heavy Structures Laboratory. The bridge deck contained
cantilevers with glass and carbon fibre reinforced polymers (GFRP and CFRP) and conventional steel
reinforcement. Experimental results based on the testing of the GFRP and steel reinforced cantilevers (testing on the
CFRP cantilever has yet to be completed) indicate that arching-action may be present. However, further
experimental investigation is required to better determine bending and arching components in bridge deck cantilever
behaviour. The results also indicate that cumulative strain as a result of fatigue may prove to be an issue and may
have to be considered in bridge deck design. The paper also concludes that a significant reduction in negative
moment reinforcement is feasible and such a reduction would lead to a considerable cost reduction associated with
negative moment reinforcement.

KEYWORDS
Cantilever, Bridge Deck, Glass Fibre Reinforced Polymer (GFRP), Static Testing, Fatigue Testing

1. INTRODUCTION
Bridges play an important role in the highway systems around the world and have been adversely affected by age
and weathering over the past two decades. The majority of highway bridges have steel reinforced concrete decks
supported on steel or concrete girders. Over the years, the weather has taken its toll on these steel reinforced
concrete decks. Rainwater and de-icing chemicals applied to bridge deck surfaces to prevent icing of the bridge
decks during the winter months have found their way into many concrete decks and caused corrosion of the
reinforcing steel (Memon & Mufti, 2004). A solution to the problem of the existing deteriorating bridge decks is to
rehabilitate or replace them with innovative technologies such as corrosion-free bridge decks (Mufti et al., 2005).
Cantilevers are an integral part of bridge decks and this paper takes a look at the test results obtained from both
static and fatigue destructive testing. It outlines the results obtained for a cantilever reinforced with GFRP and
discusses those results with respect to deflection and strain.

2. CANTILEVER DETAILS & TESTING SCHEME

Although the bridge deck contained an internal panel that was a second generation steel-free bridge deck this paper
will only outline the reinforcement details of the cantilevers. The bridge deck measured 9000 mm in length and
5000 mm in width and was supported on two steel girders (Figure 1). The cantilever overhangs measured 1250 mm

299
from the center line of the girders to the free edge of the cantilever. The deck was divided into three different panels
that, for explanation purposes, we call the GFRP, CFRP, and steel cantilevers. The bottom reinforcement was
consistent for all three panels and was comprised of #10 GFRP bars spaced at 200 mm center-to-center in the
transverse direction (shorter distance) and #10 GFRP bars spaced at 600 mm center-to-center in the longitudinal
direction (longer distance). The top longitudinal reinforcement was also the same for all three panels and consisted
of #10 GFRP bars spaced at 600 mm center-to-center. The only difference between the three different panels was
the top transverse reinforcement provided to resist the negative moment as a result of applied load near the free edge
of the cantilever. The GFRP, CFRP, and steel panels used two #19 GFRP bars spaced at 200 mm center-to-center,
two #13 CFRP bars spaced at 200 mm center-to-center, and one 20M steel bar spaced at 200 mm center-to-center
respectively.

GFRP CFRP STEEL

Top Transverse Top Transverse Top Transverse
2 - #19 GFRP Bars 2 - #13 GFRP Bars 1 - #20M Steel Bar
@200 o/c @200 o/c @200 o/c
Top Longitudinal Top Longitudinal Top Longitudinal
1-#10 GFRP Bar @ 1-#10 GFRP Bar @ 1-#10 GFRP Bar @
600 mm o/c 600 mm o/c 600 mm o/c
Bottom Transverse: Bottom Transverse: Bottom Transverse:
1 - #10 GFRP Bar @ 1 - #10 GFRP Bar @ 1 - #10 GFRP Bar @
200 mm o/c 200 mm o/c 200 mm o/c
Bottom Longitudinal: Bottom Longitudinal: Bottom Longitudinal:
1 - #10 GFRP Bar @ 1 - #10 GFRP Bar @ 1 - #10 GFRP Bar @
600 mm o/c 600 mm o/c 600 mm o/c

Figure 1: Bridge Deck Cantilever Details

The testing scheme for the cantilevers consisted of six different destructive tests. One of the cantilevers was
subjected to three static tests, each of which was conducted on the three different cantilever panels (Figure 2). The
static tests were conducted first, in order to determine the ultimate static capacity of each of the cantilevers, before
proceeding with the fatigue tests.

Figure 2: Static & Fatigue Cantilever Testing Scheme

300
3. GFRP CANTILEVER TEST RESULTS
The experimental results for the GFRP cantilever can be grouped into four different categories. For explanation
purposes the results are grouped into those dealing with deflection behaviour and strain behaviour.

3.1 GFRP Cantilever Load Deflection Behaviour

A total of eight displacement transducers were used to monitor deflection for both static and fatigue testing. The
results outlined in this paper strictly deal with the displacement transducer placed at the center of the loading plate
(Figure 3). The cantilever failed at an ultimate load of 294 kN and a maximum ultimate deflection of 27.2 mm. The
applied load for the fatigue testing was 186 kN or approximately 63% of the ultimate load previously determined by
the static test. The applied load of 63% is the average of the maximum load measured throughout the entirety of the
fatigue test. It is important to note that the ultimate deflection from the fatigue test very closely matches the ultimate
deflection measured from the static test. The ultimate fatigue deflection was measured to be 30.1 mm and the deck
failed at 104,777 cycles.

Figure 3: Static & Fatigue Load versus Deflection for GFRP Cantilever

3.2 GFRP Cantilever Load Strain Behaviour

Figure 4: Static & Fatigue Top Transverse GFRP Bar Strain Profile (Note: Center lines of girders occur at
distances of 1250 mm and 3750 mm.)

A total of 18 electronic strain gauges were installed on the top transverse GFRP bars in order to provide strain data
for the top transverse negative moment reinforcement (Figure 4). The maximum strain at the static load of 186 kN
occurs over the girder and is 1900 με in magnitude; however, at the ultimate load of 294 kN the maximum strain has
shifted towards the loading plate and is 4282 με in magnitude. An important observation is that the strains in the top
transverse bar are nearly zero at a distance of 2500 mm from the loaded edge of the cantilever. Looking at the

301
fatigue strain data it is important to observe that the strain in the top transverse bars remains relatively constant
throughout the entire life of the cantilever, however, the strains in the bars much closer to the applied load increase
significantly with increased number of cycles. The cumulative strain in the bars over the girder and the cumulative
strain in the bars close to the applied load amount to 628 με or approximately 4% and 3142 με or approximately
20% of the ultimate strain of the GRFP bars respectively (Figure 5).

Figure 5: Cumulative Fatigue Top Transverse GFRP Bar Strain

4. CONCLUSIONS & RECOMMENDATIONS

Although this paper only outlines the results obtained from the GRFP cantilever, it is important to briefly discuss the
results from the static test conducted on the cantilever with top transverse steel reinforcement in order to make an
important point. The static ultimate load for the cantilever with steel reinforcement was recorded as 301 kN. The
nominal moment resistance of the cantilever with top transverse GRFP when calculated using conventional design
methods was 154 kN*m per meter width and the nominal moment resistance of the cantilever with top transverse
steel reinforcement was 85 kN*m per meter width. If we compare the nominal moment resistance of the two
different cantilevers, conventional design methods would suggest that the GFRP cantilever would have a higher
ultimate load. The experimental results indicate that the ultimate load of the two cantilevers was the same,
suggesting that some arching action may be present in the cantilevers. It is clear from the experimental tests
conducted on the bridge deck cantilevers that extensive further experimental investigation is required to gain a better
understanding into the bending and arching components of cantilever behaviour. Another conclusion that can be
drawn from the experimental tests is that, under fatigue loading conditions, the measured cumulative strain
amounted to 20% of the ultimate strain provided by the manufacturer of the GFRP bars. The Canadian Highway
Bridge Design Code (CHBDC 2000) limits strain in GFRP bars embedded in concrete to 25% of their ultimate strain
under service conditions. Therefore, cumulative strain in GFRP bars may have to be considered in bridge deck
design. The final observation that can be made is that the negative moment intensity in the internal panel does not
significantly strain the top transverse bars beyond the mid-span of the next adjacent internal panel, suggesting that
the bars are not required in this region. Therefore, a 33% reduction by volume is feasible, which could result in a
significant cost reduction of the top transverse reinforcement required for negative moment regions in bridge deck
cantilevers.

5. REFERENCES
Canadian Highway Bridge Design Code (CHBDC). (2000). Canadian Standards Association International, Canada.
Memon, A.H. and Mufti, A.A. (2004). Fatigue behaviour of second generation steel-free concrete bridge deck slab.
Proceedings of the Second International Conference on FRP Composites in Civil Engineering (CICE), Adelaide,
Australia.
Mufti, A.A., Memon, A.H. and Klowak, C. (2005) Study of static and fatigue behavior of second generation bridge
decks. Proceedings of the International Workshop on Innovative Bridge Deck Technologies, Winnipeg, MB,
Canada.

302
Part X. FRP Bars
Third International Conference on FRP Composites in Civil Engineering (CICE 2006)
December 13-15 2006, Miami, Florida, USA

ENVIRONMENTAL CONSIDERATIONS IN USING FRP REBARS IN

CONCRETE PAVEMENTS
Amnon Katz
Department of Civil and Environmental Engineering
National Building Research Institute
Technion – Israel Institute of Technology
Haifa 32000, Israel

Abstract
Introducing reinforcing bars made of fiber-reinforced polymers (FRP) into concrete as a substitute for ordinary steel
reinforcing bars may have economic, as well as environmental consequences. A comprehensive approach was used
to study the environmental impact of such substitution/replacement, taking into consideration the whole life cycle of
the pavement, from cradle to grave, including maintenance activities and changes in concrete technology resulting
from the change in rebar type. The deficiencies of the current tools available for the assessment of environmental
impacts were also investigated in order to introduce a more reliable model.

It was found that the environmental impact can be reduced significantly at the erection stage due to the possibility of
reducing the cement content or the thickness of the concrete layer covering the reinforcing bars, since special
protection of the steel against corrosion is no longer needed. During the service life of the pavement, additional
environmental impacts are eliminated due to a decrease in the number of required maintenance activities, each of
which has direct (materials and construction energy) and indirect (disturbance to traffic) impacts.

Keywords: life cycles assessment, FRP rebar, environmental impacts

1. INTRODUCTION
Fiber-reinforced polymer reinforcing bars (FRP rebars) have emerged in recent years as a substitute for steel bars in
concrete. This material is considered to be more durable than conventional steel (Katz et al., 2001; Bank et al., 1998;
Umoto and Ohga, 1996) and is, therefore, proposed nowadays as a substitute for steel in structures exposed to
aggressive environments, such as marine environments, and in structures exposed to deicing salt chlorides. Recently
published design recommendations (ACI 440.1, 2001) indicate that the initial period of uncertainty for this new
product is already over and it can now be used more extensively.
The utilization of new products is not always associated with better environmental protection. Sophisticated
production processes may sometimes lead to a higher environmental load of a product despite its other beneficial
properties. Things become more complicated when the product is highly complex, such as a construction project, in
which changing the properties of one material or process may have long-term effects on other materials or processes.
Edvardsen and Tollose (2001) concluded that using stainless steel reinforcing bars instead of ordinary steel rebars
may lead to an overall reduction in CO2 emission due to reduced maintenance and repair, despite the increased
emission during the production of the bar itself.

Using FRP rebars as a substitute for conventional steel changes an entire set of parameters used in the design of new
constructions. Water/cement ratio, minimum cement content, thickness of concrete cover over rebars, and periodical
maintenance are all parameters that undergo changes when replacing steel with FRP. The change in these parameters
is usually in the direction of less restricting constraints/limitations (Katz, 2004).

The current study describes the change in the environmental impact following the substitution of steel with glass
FRP as a reinforcing material for concrete pavement subjected to marine environments.

303
2. ANALYSING THE ENVIRONMENTAL IMPACTS THROUGHOUT THE WHOLE
LIFE CYCLE

2.1. Erection
Protection of steel reinforcing bars in reinforced concrete structures is achieved by lowering the water/cement ratio,
increasing the cement content and increasing the thickness of the concrete layer that covers the steel. When FRP
rebars are used to replace steel ones, the above stringent requirements can be relaxed. Thus, a higher water to cement
ratio can be used, which leads to a lower cement content per concrete unit. The thickness of the non-structural
concrete layer that protects the steel can also be reduced, leading to smaller quantities of concrete per unit area of
pavement.

In this study, the environmental impacts of steel-reinforced concrete pavement made according to AASHTO
concrete Class IV was compared with Class II concrete pavement reinforced with glass fiber-reinforced polymers
(GFRP). Table 1 presents the parameters used in this study.

Table 1: Composition and processes for the two pavements.

Steel-reinforced FRP-reinforced
pavement pavement
Concrete class Class IV Class II
Slab thickness 200 mm 165 mm
Concrete composition
Cement 390 kg/m3 335 kg/m3
Water 160 kg/m3 165 kg/m3
Gravel 1100 kg/m3 1205 kg/m3
Sand 750 kg/m3 695 kg/m3
Reinforcement 104 kg/m3 32.2 kg/m3
Processing
Mixing power 9.3 MJ/m3 9.3 MJ/m3
Average distance of 100 km 100 km
mineral transport
Average distance of 500 km 1500 km
reinforcement transport
Average distance of 30 km 30 km
concrete transport

2.2. Service-Life Stage

The environmental impact of maintenance activities is divided into two categories: (1) direct impact of construction
works, and (2) indirect impact of disturbance to traffic. It was assumed that maintenance activities required for FRP
reinforced pavements are at least 50% of those required for steel-reinforced pavements. According to Ehlen (1999)
and Ehlen and Marshall (1996), maintenance activities on bridge decks begin on the 28th year after erection and are
performed every three years. During each such maintenance activity, 2.5% of the surface is chipped away,
reinforcement is repaired and treated with protective paint, and the deck is repaired using new concrete. For steel-
reinforced pavement, these assumptions lead to approximately 8 additional maintenance activities over 75 years of
service, which are the direct result of steel corrosion (activities resulting from concrete corrosion in both types of
pavement were not counted). In such pavements, deterioration of the concrete itself enhances the deterioration of the
steel, thus the assumption that 50% of the maintenance activities are a result of steel corrosion only seems
conservative.
Direct environmental impacts of the materials and processes were calculated according to the above assumptions.
The indirect environmental impact of traffic disturbance (ETD) due to maintenance works was calculated as follows.
It was assumed that a slower traffic speed leads to a proportional increase in all environmental impacts, according to
Equation 1.

304
ETD = VE x ADT x Nc x ∆Leq (Eq. 1)

where VE = vehicle environmental impact per unit length of road, ADT = average number of cars per day, Nc =
duration of traffic disturbance in days, ∆Leq = equivalent additional length of work zone, taking into account the
actual length of road affected by the maintenance works, L, and the increased emissions due to the longer time
needed to pass this area, calculated as follows:

⎛V ⎞
∆Leq = L⎜⎜ n − 1⎟⎟
⎝ Va ⎠ (Eq. 2)

Where Vn = average normal traffic speed and Va = average traffic speed in work zone.

2.3. Disposal
Is quite common nowadays to recycle old concrete into aggregate for use in the production of new concrete as well
as to recycle all the reinforcing steel. It was, therefore, assumed that ~85% of the concrete could be re-used as
aggregate in the production of new concrete (the fine fraction, ~15%, is of minor value). Lacking knowledge on the
recycling possibilities of FRP reinforced concrete, it was assumed that all FRP reinforcement and FRP reinforced
concrete is not recycled in any way.

3. RESULTS AND DISCUSSION

Data was analyzed using the Eco-indicator 99 method. Environmental impacts were analyzed in light of their impact
on human health (years of life lost or years of disability), impact on eco-systems (loss of species over a certain area)
and depletion of resources (surplus energy required for their future production). A single score was determined after
weighting, grouping and normalization.

Table 2 presents a summary of the analysis performed on the two pavement types. The total environmental load of
FRP reinforced pavement was more than 50% lower than that of steel-reinforced pavement. A significant decrease,
of approximately 36%, was identified in the erection stage, and resulted mainly from the reduced load of the
reinforcement, the lower cement content and the smaller quantity of concrete required due to the reduced slab
thickness. Figure 1 presents the environmental load of cement, aggregate, reinforcement, transportation and
processes involved in the erection of the two kinds of pavements, together with their relative part of the erection
activity. In ordinary steel-reinforced pavement, steel is responsible for 27% of the environmental load; this value is
reduced to only 11% for FRP reinforcement. The absolute impact of FRP rebars is only 25% of that of steel rebars of
the same diameter. Similarly, the impact of cement was found to decrease by 29%. However, due to the significant
decrease in the environmental load of other components, cement still seems to be a significant contributor to the
pavement's overall environmental load.

The indirect environmental load resulting from traffic disturbance seems to be the most significant factor during the
maintenance stage and is decisively greater than that of the direct maintenance work performed. Thus, an attempt
should be made to minimize traffic disturbance during maintenance activities.

Table 2: Summary of the environmental load of the tested pavements (in Eco-indicator 99 points)

Slab type Erection Maintenance* Disposal Total

Steel-reinforced pavement 179,000 n x13,200 6,020 291,000
FRP-reinforced pavement 114,000 N/A 7,680 122,000
* n=8

The environmental impact of demolishing the pavement is quite small. Despite the larger quantities involved, the
impact resulting from the demolition of steel pavement is smaller than that of FRP pavement due to the positive
impact of recycling the old concrete (recycling of the steel has already been taken into account in the production of
new steel).

305
Sensitivity analysis was carried in order to study the influence of uncertainties involved in determining the
parameters listed above. The following parameters were investigated: effect of concrete thickness (changing
concrete class but not concrete thickness), effect of concrete class (changing from Class IV to Class II-bridge),
improving the recyclability of the FRP pavement, and the effect of different evaluation methods. All these
investigations indicated that the environmental impact of FRP pavement is lower compared with that of steel-
reinforced pavement.

Process, Process,
Steel pavement FRP pavement 1
22,100 , 18,200 ,
12% 16%
Cement, Cement,
53,300 , 38,000 ,
30% 34%

Transporta
tion,
50,100 , Aggregate
28% Transporta Aggregate
, 4,500 , , 3,800 ,
3% tion,
40,800 , 3%
36% Reinforce
Reinforce ment,
ment, 12,600 ,
49,100 , 11%
27%
Figure 1: Distribution of the environmental load during the erection stage, for steel- and FRP reinforced pavements,
expressed in Eco-points (Eco-indicator 99) and as percentages.

4. CONCLUSIONS
A comparison was made between the environmental impacts of concrete pavement reinforced with steel or FRP
bars. It was found that the environmental load might be significantly lower when considering changes in the
concrete composition or pavement thickness resulting from better chloride resistance of the FRP rebars. In addition,
the environmental impact of the FRP rebar itself is much smaller than that of steel rebar of the same diameter.
Sensitivity analysis that examined the effect of different assumptions regarding concrete composition and pavement
structure led to the same conclusion. Transportation was found to be an important environmental parameter in the
production of concrete due to the large masses of materials involved in the production of concrete pavement.

References
ACI 440.1R-01, (2001), "Guide for the design and construction of concrete reinforced with FRP bars", American
Concrete Institute (ACI), Farmington Hills, MI, USA.
Bank, L.C., Puterman, M., and Katz, A., (1998), “The Effect of Material Degradation on Bond Properties of FRP
Reinforcing Bars in Concrete”, American Concrete Institute (ACI) Materials Journal, Vol. 95, No. 3, pp. 232-243.
Edvardsen C. and Tollose K., (2001), "Environmentally "green" concrete structures", Proceedings, FIB symposium
Concrete and Environment, Berlin, Germany.
Ehlen M. A., (1999), "Life cycle cost of fiber reinforced polymer bridge decks", Journal of Materials in Civil
Engineering, Vol. 11, No. 3, pp. 224-230.
Ehlen M. A. and Marshall H. E., (1996), "The economics of new technology materials: a case study of FRP bridge
decking", NISTIR 5864, National Institute of Standard and Technology, Gaithersburg MD 20899, USA, 80p.
Katz, A., " The environmental impact of steel and FRP reinforced pavements", ASCE Journal of Composites for
Construction, Vol. 8, No. 6, 2004, pp. 481-488.
Katz A., Bank L. C. and Puterman M., (2001), "Durability of FRP Rebars After Four Years of Exposure", Fiber
Reinforced Plastics for Reinforced Concrete Structure- FRPRCS-5, Burgoyne C.J. Editor, Cambridge, Thomas
Telford press, UK, pp. 497-503.
Uomoto, T. and Ohga H, (1996), "Performance of fiber reinforced plastics for construction reinforcement", in
Proceedings: Advanced Composite Materials in Bridges and Structures, El-Brady et al. Eds., Canadian Society for
Civil Engineers, Montreal, Canada, pp. 125-132

306
Third International Conference on FRP Composites in Civil Engineering (CICE 2006)
December 13-15 2006, Miami, Florida, USA

EVALUATION OF CRACK WIDTHS IN CONCRETE FLEXURAL

MEMBERS REINFORCED WITH FRP BARS
Charles E. Bakis
(Professor, Penn State University, University Park, Pennsylvania, USA)

Carlos E. Ospina
(Senior Engineer, Berger/ABAM Engineers Inc., Federal Way, Washington, USA)

Timothy E. Bradberry
(Support Branch Manager, Bridge Division, Texas Department of Transportation, Austin, Texas, USA)

Brahim Benmokrane
(Professor, Université de Sherbrooke, Sherbrooke, Québec,Canada)

Shawn P. Gross
(Associate Professor, Villanova University, Villanova, Pennsylvania, USA)

John P. Newhook
(Associate Professor, Dalhousie University, Halifax, Nova Scotia, Canada)

Ganesh Thiagarajan
(Assistant Professor, University of Missouri, Kansas City, Missouri, USA)

ABSTRACT
The width of flexural cracks in a concrete member internally reinforced with fiber reinforced polymer (FRP) bars
depends on the characteristic of interfacial bond between the reinforcement and concrete. This paper reports the
findings of a task group formed under the aegis of ACI Sub-committee 440 H (Reinforced Concrete) to evaluate and
compare the effects of bond characteristics in previous and proposed equations for predicting crack widths. Based
on the experimental data assembled by the task group, it is shown that the bond coefficient in the proposed equation
by Frosch is 19% greater than the bond coefficient in the previous equation attributed to Gergely and Lutz.

KEYWORDS
Crack width, flexure, bond

1. INTRODUCTION
At the October 2004 meeting of Sub-committee H (Reinforced Concrete) of ACI Committee 440 (Fiber Reinforced
Polymer Reinforcement), a task group was formed to evaluate the proper value of bond parameter, kb, for use in a
new equation for crack width in flexural members reinforced with deformed FRP bars. Included in the task group’s
activity were FRP bars reinforced with glass fibers (GFRP), carbon fibers (CFRP) and aramid fibers (AFRP).

The equation for crack width in ACI 440.1R-03 (2003) is referred to as the modified Gergely-Lutz equation:

ff
w = 2.2 β kb 3 d c A (1)
Ef

307
where w (mm) is the maximum likely crack width at the tension face of a beam or one-way slab (generally taken as
the 90-th percentile crack), ff (MPa) is the stress on the bar calculated by elastic cracked section theory, Ef (MPa) is
the longitudinal modulus of elasticity of the bar, β is the ratio of the distance from the neutral axis to the depth were
the crack width is measured to the distance from the neutral axis to the centroid of the tensile reinforcement, dc is the
thickness of concrete cover measured from the tension face to the center of the closest reinforcement bar (mm), A is
the effective tension area of the concrete surrounding the main tension reinforcement and having the same centroid
as that reinforcement divided by the number of bars (mm2), and kb is a factor that accounts for the bond between a
bar and concrete. For uncoated steel bars, kb is nominally equal to 1. Some typical kb values for deformed GFRP
bars cited in ACI 440.1R-03 are between 0.71 and 1.83. ACI 440.1R-03 suggests that designers assume a value of
1.2 for deformed GFRP bars unless more specific information is available for a particular bar.

At the time this task group was initiated, ACI Committee 440 was considering the adoption of a modified version of
the crack width equation by Frosch (1999) in place of the modified Gergely-Lutz equation:

ff s
w=2 β k b d c2 +   (2)
Ef 2

where s is the bar spacing. The factor 2 in eqn. (2) is used when predicting the maximum crack width. A value of
1.5 is used for the mean crack width and 1.0 for the minimum crack width. Frosch also noted that kb=1 for
conventional deformed steel bars and kb=2 for epoxy coated deformed steel bars.

As can be seen, the form of eqn. (2) is different than that of eqn. (1). Therefore, the objective of this task group
activity is to recommend an appropriate default value of kb to use in eqn. (2) when no specific value of kb has been
determined for a particular FRP reinforcing bar.

2. APPROACH
The approach taken in this endeavor is to first collect the available crack width measurements for concrete flexural
members reinforced with FRP bars and determine the ratio of kb for eqn. (2) divided by kb for eqn. (1) for each crack
width measurement. The ratio is computed for the widest flexural crack in a member in experiments where not
many crack measurements are available. In data sets where many crack widths are available, the 90th percentile
crack width (mean plus standard deviation times 1.28) is used. The ratio resulting from this comparison is then
multiplied by the currently recommended default kb value of 1.2 to arrive at a proposed value of kb when using eqn.
(2). Therefore, the maximum crack width predicted by eqn. (2) with the proposed value of kb should be no different
than that predicted by eqn. (1) with kb=1.2.

Some of the bar strains used in this analysis came from strain gages mounted on the bars while others came from
calculations using a cracked elastic section analysis. When strains determined by both methods were provided, only
those measured directly with strain gages were used. While eqns. (1) and (2) attempt to relate crack width to bar
strain with fixed constants of proportionality, it is often observed in experiments that kb varies with bar strain.
Therefore, it is useful to investigate values of kb over recommended ranges of bar strains for field applications. For
example, according to ACI 440.1R-03 (2003), maximum bar strains in FRP bars subjected to creep and fatigue
conditions in non-aggressive environments should be limited to ranges of roughly 2400-6200 µε for GFRP, 2800-
9400 µε for CFRP, and 5700-13200 µε for AFRP. The actual maximum limits depend on the particular bars under
consideration, as described more fully in ACI 440.1R-03 (2003).

The values of kb included in the present analysis comprise AFRP, CFRP, and GFRP bars with various types of
surface treatments to enhance bond with concrete. Smooth bars and grids were not included in the present analysis.
It is expected that kb values for smooth bars and grids will be larger than those with the types of bars included here.

3. FINDINGS
Table 1 summarizes some key experimental parameters and mean values of kb for the indicated range of strains for
each type of bar. It is noteworthy that mean kb values reported were often quite scattered for different beams and for
any one beam measured at different loads. Much of this scatter within a particular type of bar likely results from
inaccuracies involved in the crack width measurements. The maximum value of kb, for example, could be 50%

308
greater than the mean value reported. The range of kb values across different types of bars and different concrete
member section properties underscores the need to be conservative when recommending a value of kb to use when
no specific value is available for a particular design application. Further careful analysis of the available data is
needed to make definitive statements on the causes for the large variation of kb found in the various investigations.

Table 1. Summary of flexural crack width experiments and bond parameters, kb.

Bar Type (source of data) Surface Bar Strain Mean Gergely Mean Ratio of
Treatment (µε) -Lutz kb Frosch kb kb’s, F/GL
Hughes Aslan 100 E-glass/ vinylester Spiral indent, 2000-2100 0.92 1.10 1.20
(Giernacky, 2002) sand coating
Marshall Ind. C-Bar E-glass/ PET- Molded ribs 1600-4500 1.39 1.72 1.24
polyester (Trejo et al. 2005)
Pultrall E-glass/ vinylester Sand coating 1200-3500 1.07 1.27 1.18
(Trejo et al. 2005)
Hughes Aslan 100 E-glass/ vinylester Spiral indent, 1100-3600 1.33 1.58 1.19
(Trejo et al. 2005) sand coating
Marshall Ind. C-Bar E-glass/ PET- Molded ribs 2300-9000 0.58 0.60 1.03
polyester (Thériault et al. 1998)
Marshall Ind. C-Bar E-glass/ PET- Molded ribs 2200-7900 1.00 1.14 1.13
polyester (Masmoudi et al. 1998)
Pultrall E-glass/ vinylester Sand coating 2400-4300 0.76 0.84 1.12
(Newhook, 2000)
Hughes Aslan 200 carbon/ vinylester Spiral indent, 2400-8800 0.92 1.09 1.19
(Theisz, 2004) scrim texture
DFI carbon/ epoxy Sand-blasted 1300-9800 1.07 1.16 1.08
(Thiagarajan, 2003)
Pultrall E-glass/ vinylester (El- Sand coating 1300-3400 0.60 0.67 1.12
Salakawy and Benmokrane, 2004)
Pultrall Carbon/ vinylester (El- Sand coating 1100-3200 0.64 0.79 1.23
Salakawy and Benmokrane, 2004;
Kassem, 2004)
Marshall Ind. C-Bar E-glass/ PET- Molded ribs 2100-3000 0.83 1.10 1.33
polyester (Kassem, 2004)
Marshall Ind. C-Bar Carbon/ PET- Molded ribs 1300-2800 0.82 1.09 1.32
polyester (Kassem, 2004)
Arapree Aramid/ epoxy Sand coating 3400-4300 0.92 1.22 1.33
(Kassem, 2004)
Steel (El-Salakawy and Benmokrane, Ribbed 800-1200 0.72 0.90 1.26
2004; Kassem, 2004)
FRP Maximum 1.39 1.72 1.33
FRP Minimum 0.58 0.60 1.03
FRP Mean 0.92 1.10 1.19
FRP Std. Dev. 0.25 0.31 0.093
FRP CV (%) 27 28 8

Some data for uncoated steel reinforcement bars is included in Table 1 for comparison with data for FRP bars. The
kb values for the steel bars fall within, but near the low end, of the range covered by the FRP bars. The ratio of kb
values for the two equations in the case of steel bars is similar to that found for FRP bars.

Maximum, minimum, mean, standard deviation, and coefficient of variation of kb values and the ratio of kb values
for only the FRP bars (i.e., excluding steel bars) are given at the bottom of Table 1. For the FRP bars, the mean ratio
of the kb’s for the two equations is 1.19, the maximum 1.33, and the minimum 1.03. Using the mean value of 1.19,
the recommended value of kb to use with FRP bars in eqn. (2) in the absence of more specific data is then

kb = 1.2 x 1.19 = 1.4 (3)

309
For comparison, the actual values of kb listed for the various bars when using eqn. (2) ranges from 0.6 to 1.72, with a
mean of 1.10. Hence, the recommended value of 1.4 has some built-in conservatism that encompasses most, but not
all of the data in Table 1. Equation (2) and the recommended value of kb shown in eqn. (3) have been incorporated
into the latest ACI design guide for concrete reinforced with FRP bars, ACI 440.1R-06 (2006).

4. CONCLUSION
When a specific value of kb is not known for a carbon or glass FRP reinforcing bar with sand coating, indents, sand-
blasted surface, and/or molded ribs to enhance bond with concrete, it is recommended to use a conservative value of
1.4 when using eqn. (2) to predict crack width. The limited amount of data for one type of aramid FRP bar included
in this analysis (sand coating) abides by this recommendation as well. Smooth bars and grids are specifically
excluded from this recommendation until further analysis on such reinforcements has been carried out.

5. REFERENCES
ACI 440.1R-03 (2003). “Guide for the design and construction of concrete reinforced with FRP bars.” American
Concrete Inst., Farmington Hills, Michigan, USA.
ACI 440.1R-06 (2006). “Guide for the design and construction of structural concrete reinforced with FRP bars.”
American Concrete Inst., Farmington Hills, Michigan, USA.
El-Salakawy, E.F., and Benmokrane, B. (2004). “Serviceability of concrete bridge deck slabs reinforced with FRP
composite bars,” ACI Struct. J., Vol. 101, No. 5, pp. 727-736.
Frosch, R.J. (1999). “Another look at cracking and crack control in reinforced concrete.” ACI Struct. J., Vol. 96, No.
3, pp. 437-442.
Giernacky, R.G. (2002). “Durability of E-Glass FRP reinforced concrete beams.” B.S. Thesis, Dept. of Engineering
Science & Mechanics, Penn State University, University Park, Pennsylvania, USA.
Kassem, C. (2004). “Cracking and load-deflection behaviour of one-way concrete elements reinforced with FRP
bars under flexure,” Ph.D. Thesis (in French), Dept. of Civil Engineering, Université de Sherbrooke, Sherbrooke,
Québec, Canada.
Masmoudi, R., Thériault, M., and Benmokrane, B., (1998). “Flexural behavior of concrete beams reinforced with
deformed fiber reinforced plastic reinforcing rods.” ACI Struct. J., Vol. 95, No. 6, pp. 665-675.
Newhook, J.P. (2000). “The use of fibre reinfroced concrete to reduce crack widths in GFRP reinforced concrete
beams.” Proc. 3rd Intl. Conf. Advanced Composite Materials in Bridges and Structures, ACMBS III, Editors: J. L.
Humar and A. G. Razaqpur, Canadian Soc. Civil Engineering, Montreal, Quebec, pp. 145-152.
Theisz, P. (2004). “Properties of high performance concrete beams reinforced with carbon fiber reinforced polymer
bars,” M.S. Thesis, Dept. of Civil Engineering, Villanova University, Villanova, Pennsylvania, USA.
Thériault, M., and Benmokrane, B. (1998). “Effects of FRP reinforcement ratio and concrete strength on flexural
behavior of concrete beams.” J. Comp. Const., Vol. 2, No. 1, pp. 7-16.
Thiagarajan, G. (2003). “Experimental and analytical behavior of carbon fiber-based rods as flexural
reinforcement,” J. Comp. Constr., Vol. 7, No. 1, pp. 64-72.
Trejo, D., Aguíñiga, F., Yuan, R.L., James, R.W., and Keating, P.B. (2005). “Characterization of design parameters
for fiber reinforced polymer composite reinforced concrete systems.” Report FHWA/TX-5/9-1520-3, Texas
Transportation Institute, Texas A&M University, College Station, Texas, USA (http://tti.tamu.edu/documents/9-
1520-3.pdf).

310
Third International Conference on FRP Composites in Civil Engineering (CICE 2006)
December 13-15 2006, Miami, Florida, USA

THE PERFORMANCE OF CURVED NON-FERROUS REINFORCEMENT

FOR CONCRETE STRUCTURES
Thanongsak Imjai
(PhD student, Department of Civil & Structrual Engineering, Sheffield University, Sheffield, UK)

Maurizio Guadagnini
(Lecturer, Department of Civil & Structrual Engineering, Sheffield University, Sheffield, UK)

Kypros Pilakoutas
(Professor, Department of Civil & Structrual Engineering, Sheffield University, Sheffield, UK)

Peter Waldron
(Professor, Department of Civil & Structrual Engineering, Sheffield University, Sheffield, UK)

ABSTRACT
Steel reinforcement in concrete has the tendency to corrode and this process can lead to structural damage. FRP
reinforcement represents a viable alternative for structures exposed to aggressive environments and has many
possible applications where superior corrosion resistance properties are required. The use of FRP rebars as internal
reinforcements for concrete, however, is limited to specific structural elements and does not yet extend to the whole
structure. The reasons for this relate to the limited availability of curved or shaped reinforcing elements on the
market and their reduced structural performance. Various studies, in fact, have shown that the mechanical
performance of bent portions of composite bars is reduced significantly under a multiaxial combination of stresses
and that the tensile strength can be as low as 40% of the maximum tensile strength that can be developed in the
straight part. In a significant number of cases, the current design recommendations for concrete structures reinforced
with FRP, however, were found to overestimate the bend capacity of FRP rebars. This paper presents an overview on
the current use of curved reinforcement and discusses the problems related to the performance of curved FRP bars
embedded in concrete.

KEYWORDS
Curved FRP bar, tensile strength, bend capacity, corrosion, concrete

1. INTRODUCTION

Fibre Reinforced Polymer (FRP) reinforcement has rapidly emerged as an efficient alternative to conventional steel
reinforcement to overcome the problem of corrosion. Owing to its superior durability characteristics, the use of FRP
reinforcement can extend the lifespan of concrete structures and reduce the need for maintenance or repair.
However, although FRPs are already quite extensively adopted in various sectors of the construction industry (e.g.
strengthening and repair of existing structures), their use as internal reinforcement for concrete is limited only to
specific structural elements and does not extend to the whole structure. The reason for the limited use of FRPs as
internal reinforcement can be partly related to the lack of commercially available curved or shaped reinforcing
elements used for shear reinforcement or complex structural connections.

Most of the shaped steel reinforcing bars currently used in concrete structures are provided pre-bent and cut in the
factory according to design specifications. These may be supplemented by a small quantity of special one-off shapes
bent directly on site. Whether bending occurs on site or at the factory, conventional steel reinforcing bars have a
major advantage since, due to their elastoplastic behaviour, they can be easily formed by cold bending, and hence,
most detailing needs can be easily met at very low cost.

311
Existing guidelines for the cold bending of steel reinforcement specify, for mild steel, a bend radius to diameter ratio
(r/d) of 2 (for example BS8666:2000), which would induce a maximum strain value of 20% in the material (Figure
1-left). When cold bending FRP bars, however, there are problems associated with the potential buckling of the
fibres located in the compression side. Moreover, the typical ultimate strain value of FRP products varies from 1%
to 2.5%, hence, the amount of strain that is induced in the fibres needs to be carefully controlled to avoid premature
failure of the reinforcing bar. As a result, cold bending of FRP bars requires larger bend radius to diameter ratios
than are currently specified for steel reinforcement.

30
Strain (%)

25

20 maximum strain values induced

by cold bending of steel bars
15

10
range of ultimate strain values
5 for typical FRP bars

0
0 5 10 15 20 25 30 35 40
Bending radius/Bar diameter

Figure 1: Strain induced in cold bent bar (left) and Longitudinal and transversal stress acting on a FRP bent
bar embedded in concrete (right)

In cases where tight radii are needed (i.e. for the manufacture of shear links and hooks), preformed curved bars of
FRP are required. The high production costs that are associated with the manufacturing of FRP curved elements,
however, have generally reduced the interest in using FRPs for these types of applications. In addition, various
studies (Ehsani et al. 1995; Ishihara et al. 1997; Maruyama et al. 1995; Morphy et al. 1997) have shown that the
tensile strength of FRP bars can be largely reduced under a combination of tensile and shear stresses. This
phenomenon can often become an issue when curved unidirectional composite elements are used as reinforcement in
concrete structures (Figure 1-right) and especially when the fibres are designed to carry high tensile stresses, since
premature failure can occur at the corner portion of the composite. In fact, tests by different authors have shown that
the tensile strength of a bent portion of composite bar can be as low as 40% of the maximum tensile strength that can
be developed in the straight part (Ehsani et al. 1995; Maruyama et al. 1995; Morphy et al. 1997).

The reduction in the strength of the composite, therefore, needs to be carefully taken into account since it has a
major influence on the maximum value of strain that can be safely sustained by the reinforcement. With all of these
issues in mind, the 2 years CRAFT RTD project (CurvedNFR 2003), funded by the European Commission, was
carried out with the aim of developing material, methodology and manufacturing process for a low-cost, curved
fibre-reinforced plastic (FRP) rebar. The project partnership, which ended in 2005, included 8 specialist SME and 3
RTD organisations across 6 European countries.

2. BEND CAPACITY
The experimental work available in the literature (Ehsani et al. 1995; Ishihara et al. 1997; Maruyama et al. 1995;
Morphy et al. 1997; Nagasaka et al. 1989; Nakamura and Higai 1995) indicates that the tensile strength of a bent
FRP bar can reduce significantly. The reduction in strength that occurs at the corners of a FRP bar has been
quantified using empirical models such as that proposed by the Japan Society of Civil Engineers, which is described
by Equation (1) (JSCE 1997). In this equation, the strength of the bent portion, ffb, is expressed solely as a function
of the uniaxial tensile strength of the composite, ffu, and the bar geometry (i.e. bar cross-section, d, and bend radius,
r).

⎛ r ⎞
f fb = ⎜ α + 0.3 ⎟ f fu ≤ f fu (1)
⎝ d ⎠

The value of α=0.05 corresponds to a 95% confidence limit, whilst α =0.092 corresponds to a 50% confidence limit.

312
Equation (1) yields generally a conservative estimate of the maximum strength that can be developed in bent bars
and it is currently adopted in the different design recommendations for FRP RC structures proposed by the
American Concrete Institute Committee 440 (2003); ISIS Canada (2001) and the Institution of Structural Engineers
(1999).

At the University of Sheffield, experimental work has been undertaken on the use of thermoplastic composites to
manufacturer bars that can be shaped easily to meet any detailing needs. Closed shear links have been produced and
used successfully in several applications (Imjai et al. 2004). An extensive series of pullout tests on bent bars
embedded in concrete was also conducted and various parameters were investigated (r/d ratio, surface finish,
concrete strength). The behaviour of thermoplastic GFRP strips (Plytron; ffu=720 MPa; Ef=28 GPa) and pre-bent
thermosetting GFRP rods (Aslan 100; ffu=760 MPa; Ef=40.8 GPa) was examined in this experimental programme.

100
fmax/ffu (%)

90
92
80 = 0. 0
E ,α
J SC
70
α=0.05
J S CE ,
60
Plytron - Type 2
50 Plytron - Type 2, S
Plytron - Type 2, H
40
Plytron - Type 2, S, H
30 Plytron - Type 3
Plytron - Type 3, H
20 Aslan 100 - Type 2
Aslan 100 - Type 2, H
10
Aslan 100 - Type 3
0
0 1 2 3 4 5 6 7
r/d
Figure 2: Comparison of test results with predictions according to current design recommendations

The variation in the strength of the bent specimens tested during this research project is shown in Figure 2 and
compared to the predictions calculated according to the equation proposed in the JSCE design recommendations (the
shaded area demarked by dashed lines). As can be observed, the current design equation does not adequately
describe the variation in bend capacity that was observed experimentally. Moreover, whilst the JSCE equation could
be used for the design of curved Aslan 100 specimens, it would appear that the same equation could overestimate the
bend capacity of the Plytron strip that was used in this study. Sheata et al. (2000) have also reported a similar
tendency for a commercial type of CFRP reinforcement. In the present study, however, acceptable predictions were
obtained for those specimens for which a better bond between the composite and the concrete was ensured either by
providing a longer embedment length (Type 2), sand coating the strips (S) or through the use of a high strength
concrete (H). Thus, it would appear that the strength of bent bars, depends not only upon longitudinal strength of the
composite and the geometry of the bent, but also on the type of fibres, the resin type and manufacturing process.

3. CONCLUSIONS
Based on the experimental work undertaken as part of the CurvedNFR Project, the following conclusions may be
drawn:

(1) Thermoplastic composites seem to offer a valid solution for the manufacturing of bends and complex
shapes. The durability of such composites in concrete, however, needs to be investigated.
(2) The capacity of the bent portion of the composite appeared to be mainly a function of the geometry of the
test specimens, namely the bending radius.
(3) The bend capacity of the test specimens varied between 25% and 85% of the ultimate strength of the
composite.

313
 (4) Values of r/d greater than 4 are required to guarantee a minimum bend capacity of 40% of the ultimate
strength of the composite.
(5) In a significant number of cases, the equation included in the current design recommendations for concrete
structures reinforced with FRP was found to overestimate the bend capacity of the Plytron strip.
(6) The capacity of the bent specimens does not seem to vary linearly with the r/d ratio, as defined in the JSCE
equation, and does not appear to be solely a function of the bend geometry. Rather, bond characteristics
appeared to be important in controlling the development of stresses along the embedded portion of the
composite and in dictating its ultimate behaviour.
(7) The equation included in the existing design recommendation for predicting the bend capacity of a curved
FRP bar does not seem to yield consistent predictions when compared to experimental data. A new
predictive model is needed and should be based on a micromechanical approach that takes the mechanical
properties of the composite into account.

FRP composite materials need to move from low volume/high technology applications to high volume/relative low
technology applications. Before the use of FRPs becomes widely accepted for concrete structures, several significant
aspects of the materials have to be examined, including the ability to produce standard reinforcement shapes and the
ability to produce large quantities of materials with a consistent quality. All of these aspects are essential if the true
potential of FRP reinforcement is to be exploited in civil engineering applications.

4. REFERENCES
American Concrete Institute (ACI). (2003). "Guide for the Design and Construction of Concrete Reinforced with
FRP Bars ACI 440.1R-03", ACI Committee 440, Farmington Hills, MI, USA.
British Standards Institution (2000), "Specification for Scheduling, Dimensioning, Bending and Cutting of Steel
Reinforcement for Concrete", BS8666:2000, BSI, London.
CurvedNFR (2003). "Cost effective Curved Polymer Composite Rebar". CRAFT RTD European funded project,
CRAFT GIST-CT-2002-50365, www.curvednfr.com.
Ehsani, M. R., Saadatmanesh, H., and Tao, S. (1995). "Bond of Hooked Glass Fiber Reinforced Plastic (GFRP)
Reinforcing Bars to Concrete." Materials Journal, 122(3), pp. 247-257.
Imjai, T., Guadagnini, M., and Pilakoutas, K. (2004). "Small-scale and Medium-scale testing for CurvedNFR
Project." ExpR-3, The University of Sheffield, UK, pp. 13.
Institution of Structural Engineers (ISE). (1999). Interim Guidance on the Design of Reinforced Concrete Structures
Using Fibre Composite Reinforcement, IStructE, SETO Ltd., London.
ISIS Canada (ISIS). (2001). Manual No. 3 - "Reinforcing Concrete Structures with Fibre Reinforced Polymers
(FRPs) ", ISIS Canada, Winnipeg, Manitoba, Canada.
Ishihara, K., Obara, T., Sato, Y., and Kakuta, Y. (1997). "Evaluation of Ultimate Strength of FRP Rods at Bent-up
Portion." Third International Symposium on Non-Metallic (FRP) Reinforcement for Concrete Structures, Sapporo,
Japan, pp. 27-34.
Japan Society of Civil Engineers (JSCE). (1997). "Recommendation for Design and Construction of Concrete
Structures using Continuous Fiber Reinforcing Materials", JSCE, Tokyo, Japan.
Maruyama, T., Honma, M., and Okamura, H. (1995). "Experimental Study on Tensile Strength of Bent Portion of
FRP Rods." Non-Metallic (FRP) Reinforcement for Concrete Structures. Proceeding of 2nd RILEM symposium
(FRPRCS-2), pp. 163-176.
Morphy, R., Sheata, E., and Rizkalla, S. (1997). "Bent Effect on Strength of CFRP Stirrups." Third International
Symposium on Non-Metallic (FRP) Reinforcement for Concrete Structures, Sapporo, Japan, pp. 19-26.
Nagasaka, T., Fukuyama, H., and Tanigaki, M. (1989). "Shear Performance of Concrete Beam Reinforced with FRP
Stirrups." Transactions of The Japanese Concrete Institute, 11, pp. 789-811.
Nakamura, H., and Higai, I. (1995). "Evaluation of Shear Strength on Concrete Beams Reinforced with FRP."
Concrete Library, JSCE, 26, pp. 111-123.
Sheata E., Morphy R. and Rizkalla S. (2000) "Fibre Reinforced Polymer Shear Reinforcement for Concrete
Members: Behaviour and Design Guidelines", Can. J. Civ. Eng./Rev. can. génie civ. 27(5), pp. 859-872.
Ueda T., Sato Y., Kakuta Y., Imamura A., and Kanematsu H. (1995). Failure Criteria for FRP rods Subjected to a
Combination of Tensile and Shear Forces. Non-Metallic (FRP) Reinforcement for Concrete Structrues. Proceeding
of 2nd RILEM symposium (FRPRCS-2), pp. 26-23.

314
Third International Conference on FRP Composites in Civil Engineering (CICE 2006)
December 13-15 2006, Miami, Florida, USA

FLEXURAL TESTING OF SPUN-CAST HOLLOW CONCRETE PILE

SECTIONS REINFORCED WITH CFRP GRID AND BARS
Antonis Petrou Michael
(Post Doctoral Associate, University of Florida, Gainesville, Florida, USA)

H. R. Hamilton III
(Associate Professor, University of Florida, Gainesville, Florida, USA)

ABSTRACT
Corrosion of prestressing steel in precast concrete is a significant problem for bridges constructed in coastal waters.
Carbon fiber-reinforced polymer (CFRP) reinforcement has been suggested as a non-corrosive replacement for steel
in highly corrosive environments. One drawback, however, is the lack of ductility that CFRP flexural reinforcement
provides when the section is taken to its full capacity under extreme events such as earthquakes or barge impacts.
This paper covers flexural testing of circular hollow concrete sections constructed using CFRP bars for flexural
reinforcement and CFRP grid for concrete confinement. In addition, the pile sections were manufactured using the
spun-cast process providing a unique combination of materials and manufacturing processes. Experimental results
available from the flexural tests indicated no significant improvement in the ductility of the grid reinforced piles
compared to the control piles but that can be attributed to damaged confinement reinforcement and/or failure modes
involving shear.

KEYWORDS
Spun-cast piles, CFRP reinforcement, concrete confinement, ductility.

1. INTRODUCTION
The spun-cast method is commonly used to produce prestressed concrete poles for power lines or stadium lighting.
They are produced by placing the concrete in a mold and then spinning the mold at high speed for a specific amount
of time depending on the size of the pole. The centrifugal force distributes the concrete across the length of the pole
while at the same time highly compacts the concrete. Spun-cast poles have higher concrete strengths and less
permeability than regular cast concrete poles.

Corrosion of prestressing steel in precast concrete is a problem for bridges in coastal waters. Carbon fiber-reinforced
polymer (CFRP) reinforcement has been suggested as a non-corrosive alternative for steel in highly corrosive
environments. One drawback, however, is the lack of ductility that CFRP flexural reinforcement provides when the
section is taken to its full capacity under dynamic loading events such as earthquakes or barge impacts. In this
proposed configuration the piles are designed to fail by crushing the concrete and an improvement in the ductility of
the concrete section is sought through concrete confinement. A CFRP grid tied into a circular shape and cast into the
concrete in a similar configuration to spiral ties is used as confinement reinforcement.

CFRP grids (with thicknesses of 3 to 4 hundredths of an inch) are used primarily for crack control in concrete
structures. Harries and Gassman (2003) conducted tests on reinforced concrete basin knockout panels that employed
a light carbon grid to control cracking. The grid reduced cracking of the panel significantly. Shao et al. (2003) used
the same light carbon grid to control plastic shrinkage cracking in concrete. They concluded that the plastic
shrinkage cracks were reduced by 50% to 65%. Michael et al. (2005) embedded the CFRP in concrete cylinders to
provide confinement. Two layers of the CFRP grid were used and it was found that the CFRP grid provided
confinement and doubled the concrete crushing strain.

315
2. PILE MANUFACTURING
Six 38-foot long piles were constructed including three with no grid in the mid-length portion of the pile (spun cast
control (SCC) piles) and three with carbon grid over their entire length (spun cast grid (SCG) piles).

14.9" 14.9"
13.4"

8.9" to 11.3" 8.9" to 11.3"

CFRP Grid
(a) (b)

Figure 1: Mid-span cross-section: (a) Control pile and (b) Grid pile

Twelve carbon reinforcing bars with a diameter of 0.375 in. were used as flexural reinforcement for the pile
specimens (See Fig.1). A concrete mixture with a specified 28-day compressive strength of 5000 psi and a
maximum aggregate size of 0.375 in. was used to manufacture the pile specimens.

The manufacturing process included several steps: (a) Placement of flexural reinforcement, (b) Placement of the
CFRP grid, (c) Placement of concrete (See Fig. 2(a)), (d) Sealing of the mold, (e) Spinning of the mold (See Fig.
2(b)), (f) Overnight curing of the pile inside the mold, and (g) Removal of the pile from the mold.

(a) (b)

Figure 2: Spun cast pile manufacturing: (a) Concrete Placement and (b) Mold spinning on platform

3. FLEXURAL TESTING
All piles were tested in a simply supported four-point bending configuration (See Fig. 3) in displacement control
mode, that is, a constant displacement rate was applied independently of the amount of load. A hydraulic actuator
was used to apply the displacement to the piles. The applied displacement was distributed to the two load points
using a steel spreader beam. Because the spun-cast pile specimens were circular in cross-section, steel saddles were
placed at the end supports to accommodate the shape of the piles and avoid lateral movements. Two more steel
saddles were used at the load points for load transfer from the spreader beam to the piles without slippage. The piles
were loaded at a rate of 0.01-in. per second. The displacement was applied monotonically until the pile could no
longer sustain any load.

316
 Actuator
Spreader Beam
D1 D2 D3 D4 D5 D6 D7 D8 D9 D10 D11 D12 D13

1'
5' 4' 3' 3' 1.5'

16' 3' 16'

North South
Support LVDT Support

Figure 3: Test set-up

4. RESULTS
The load-displacement curves for all pile specimens were linear to peak load (Fig. 4 (a)). The load-displacement
curves for all pile specimens were similar and their behavior non-ductile. Although pile SCG1 had exhibited some
post peak behavior it was not significant. This was attributed to the fact that pile SCG1 failed in shear-flexure and
therefore the effect of concrete confinement did not contribute to the overall behavior of the pile. Pile SCG3 did not
exhibit any post peak behavior and failed in the same manner as pile SCC3. This was not expected because the
presence of the CFRP grid as concrete confining reinforcement was expected to result in an improvement in the
ductility of the pile. However, it was discovered after the end of the test that the CFRP grid had been badly damaged
during construction. The damage resulted in the rupture of the hoop direction strands. The loss of the hoop direction
strands resulted in lack of concrete confinement and consequently failure to improve the ductility of the pile.

15 140

120
12
Total Load (kips)

100
Moment (k-ft)

9 80
SCC1 60 SCC1
6
SCG1 SCG1
SCC2 40 SCC2
3 SCC3 SCC3
SCG3
20
SCG3
0 0
0 4 8 12 16 20 0 0.0005 0.001 0.0015 0.002 0.0025
Displacement (in) Curvature (rads/in)
(a) (b)

Figure 4: (a) Load-displacement curves and (b) M-Φ curves (Numerical differential method)

The failure of the control piles (SCC1, SCC2 and SCC3) was highly brittle inside the constant moment region. The
control piles shattered into pieces (See Fig. 5(a)). Specimens SCG1 and SCG3 also had brittle failures although
concrete did not shatter in both cases. The lack of improvement in ductility for these two piles was attributed to the
mode of failure for SCG1 (shear-flexure) and to construction mishaps for SCG3 (damage to grid hoop strands).
During testing of pile SCG2 a number of significant observations were made. The pile failed in flexure inside the
constant moment region adjacent to the south load point. Failure was not abrupt but gradual. Pile SCG2 did not
collapse at the end but rather was capable of supporting its own weight and an additional load of approximately 2
kips (See Fig. 5(b)). Unfortunately, a data acquisition malfunction resulted in the loss of the data for this specimen,
leaving only video and photographic evidence. This was the type of behavior that was unsuccessfully attempted to
duplicate with pile SCG3.

The moment-curvature (M-Φ) curves of all specimens were developed and plotted (See Fig. 4(b)). The numerical
differential method was used to calculate the curvature values at various load levels. The numerical differential
method calculates the curvature based on the displacement values of adjacent points and therefore any localized

317
effect such as the formation of a plastic hinge can be detected. The plastic hinge is usually manifested in the form of
large increases in the curvature.

(a) (b)

Figure 5: Pile specimens after testing: (a) Control and (b) Grid

It is apparent from the M-Φ curves that none of the pile specimens formed a plastic hinge. Only specimen SCG1
may have been in the very early stages of plastic hinge formation but the impact on the overall ductility of the pile
was minor.

5. CONCLUSIONS
Based on the results presented in this chapter the following conclusions can be drawn:
ƒ The results from the piles indicate that the ductility of the grid piles was not significantly altered. However the
visual observations during testing of pile SCG2 indicated a change in the behavior using the CFRP grid as
confinement reinforcement and that improvement in ductility of CFRP reinforced pile using concrete
confinement might be feasible.
ƒ The tests revealed problems with the manufacturing practice that proved to be very significant in influencing the
behavior of the pile specimens. These problems need to be addressed to take advantage of the possible
improvement in ductility due to confinement provided by the CFRP grid.
ƒ The results presented in this paper represent a lower bound solution to the ductility problem of spun-cast
manufactured CFRP reinforced piles.

6. ACKNOWLEDGEMENTS
This study is sponsored by the Florida Department of Transportation. The authors gratefully acknowledge the
contributions of Steve Eudy who set up and ran the data acquisition system, as well as Marc Ansley, Frank Cobb,
Tony Johnston, Paul Tighe, and David Allen who helped prepare the specimens and set-up the testing apparatus.
The authors would also like to thank Accord Industries, Inc. for manufacturing the piles at cost. The authors thank
Jaber Jaber and John French of Accord Industries for their support and help in completing the manufacturing of the
piles.

7. REFERENCES
Harries, K. A. and Gassman, S. L. (2003), “Load tests of reinforced concrete catch basing knockout panels”,
Department of Civil and Environmental Engineering, University of South Carolina, Report No ST03-01, p. 21.
Shao, Y., Johnson, C. and Mirmiran, A. (2003), “Control of plastic shrinkage cracking of concrete with TechFab
carbon FRP grids”, Department of Civil, Construction, and Environmental Engineering, North Carolina State
University, Report to Tech-Fab Inc, p. 7.
Michael, A. P., Hamilton, H. R. III, and Ansley, M. H. (2005). “Concrete confinement using carbon fiber reinforced
polymer grid”, 7th International Symposium of Fiber-Reinforced Polymer (FRP) Reinforcement for Concrete
Structures, Editors: Carol K. Shield, John P. Busel, Stephanie L. Walkup, and Doug D. Gremel, American Concrete
Institute, Kansas City, MO, Vol. 2, pp. 991-1010.

318
Third International Conference on FRP Composites in Civil Engineering (CICE 2006)
December 13-15 2006, Miami, Florida, USA

Use of GFRP Bars as Reinforcement for Concrete Bridge Deck Slabs

Brahim Benmokrane 1, Ehab El-Salakawy 2, Sherif El-Gamal 3, and Amr El-Ragaby 4

1
NSERC Research Chair Professor in Innovative FRP Composite Materials for Infrastructures, Department of Civil
Engineering, University of Sherbrooke, Sherbrooke, Quebec, Canada
2
Canada Research Chair Professor in Advanced Composite Materials and Monitoring of Structure, Dept. of Civil
Engineering, University of Manitoba, Winnipeg, Manitoba, Canada
3
Post Doctoral Fellow, Department of Civil Engineering, University of Sherbrooke, Sherbrooke, Quebec, Canada
4
Ph. D. Candidate, Department of Civil Eng., University of Sherbrooke, Sherbrooke, Quebec, Canada

ABSTRACT
Six innovative concrete bridges reinforced with FRP bars were recently constructed in North America. Five bridges,
Wotton, Magog, Cookshire-Eaton, Val-Alain, and Melbourne Bridges are located in Quebec, Canada, while the
sixth one, Morristown Bridge, is located in Vermont, USA. All the bridges are of girder-type with main girders
made of either steel or prestressed concrete with spans ranging from 26.2 to 50.0 m. The deck is a 200 to 230 mm
thickness concrete slab continuous over spans of 2.30 to 3.15 m. Glass and carbon FRP reinforcing bars as well as
conventional steel were used as reinforcement for the concrete deck slab. The bridges are well instrumented at
critical locations for internal temperature and strain data collection using fiber optic sensors. Except Melbourne
Bridge, all bridges were tested for service performance using calibrated truckloads. This paper presents the
construction details and field testing results of the constructed bridges. The construction procedure and field test
results under real service conditions showed very competitive performance to concrete bridges reinforced with steel.

KEYWORDS
Concrete, deck slabs, FRP, strains, deflection, testing.

1. INTRODUCTION
The corrosion of steel reinforcing bars in concrete bridge decks, which leads to excessive cracking, spalling, reduced
strength and ultimately loss of structural integrity, constitutes a major problem when measured in terms of
rehabilitation costs and traffic disruption (Yunovich and Thompson, 2003). Attempts in recent years to address
corrosion-related problems have included the development and assessment of alternatives to conventional steel
reinforcement. One of these alternatives, fiber-reinforced-polymer (FRP) composite reinforcement has been used
successfully in many industrial applications and, more recently, has been introduced as reinforcement in concrete
bridge decks and other structural elements. The use of the non-corrodible FRP bars as reinforcement for concrete
bridge decks provides a potential for increased service life, economic, and environmental benefits (Nanni and Faza
2002; El-Salakawy and Benmokrane 2003).

This paper presents new and innovative field applications of FRP bars as reinforcement for the concrete deck slabs
of six bridges recently constructed in Canada and USA. The variables in these bridges, which were constructed
during 2001 to 2005, are the type of FRP reinforcing bars (glass or carbon), the reinforcement ratio, the area of the
bridge deck reinforced with FRP bars, the location, and category of the bridge (traffic volume and frequency of
using de-icing chemicals). This paper summarizes the construction details and some of the field testing results of
these bridges.

2. DESCRIPTION OF THE BRIDGE DECK SLABS

All the six bridges were built with normal-weight concrete. The concrete for the Morristown Bridge had an average
28-day compressive strength of 27 MPa, compared to 37 to 52 MPa for the other five Bridges. Sand-coated glass

319
and carbon FRP bars (Pultrall Inc., Thetford Mines, Quebec) were used in reinforcing the bridge deck slabs of the
Wotton, Magog, and Morristown Bridges in 2001 and 2002. By the end of 2002, in collaboration with the NSERC
Industrial Chair at the University of Sherbrooke, Pultrall Inc. was successful in developing new sand-coated glass
FRP bars with improved mechanical and durability properties (10-15% and 20-25% more modulus and tensile
strength, respectively, compared to those of the old bars). The new glass FRP bars were used in reinforcing the
bridge deck slabs of the Cookshire-Eaton, Val-Alain, and Melbourne Bridges.

The design of the concrete deck slab for Wotton and Magog Bridges (Quebec, Canada) was originally made with
steel bars. Then, the steel reinforcement was replaced with FRP reinforcement. This design approach led to heavy
FRP reinforcement. However, the other four bridges were designed based on serviceability criteria. A maximum
crack width of 0.5 mm and allowable stress limits (15 and 30% of the ultimate tensile strength of the material under
sustained and service loads, respectively) were used as a controlling design factors. This design approach led to
more economic, yet conservative, design.

The six bridges are girder type with the main girders made of either steel or prestressed concrete. The six bridges
have different spacing between girders, slab thickness, reinforcement types and ratios. Also, the roads on which the
six bridges are located have different functional categories, which mean different traffic volumes and environmental
conditions (frequency of using deicing salt). Tables 1 and 2 give some details on these bridges. More details on these
bridges can be found elsewhere (El-Salakawy et al. 2003; El-Salakawy et al 2005; Benmokrane et al. 2006).

Table 1. Six concrete bridges reinforced with FRP bars.

Total length× Deck slab Traffic
Bridge total width, m Thick, mm Span, m Transverse Reinforcement V/day Classification
Wotton 30.6 × 8.90 200 2.60 Glass FRP bars at Top and < 1000 Rural
Magog 83.7 × 14.1 220 2.85 Carbon FRP bars at Bottom 35,000 Highway
Morristown 43.9× 11.30 230 2.36 7,000 Urban
Cookshire-Eaton 52.0 × 13.6 200 2.70 Glass FRP bars at Top and 10,000 Urban
Val-Alain 50.0 × 12.6 225 3.15 Bottom 40,000 Highway
Melbourne 89.4 × 12.5 200 3.15 35,000 Highway

Table 2. Reinforcement details of the concrete deck slab of the six bridges.
Bar Transverse Direction Longitudinal Direction
Bridge
Type Top Bottom Top Bottom
No.15M@150 No.15M@150 No.15M@225 No.15M@225
Steel
Wotton (1.00 %) (0.85 %) (0.67 %) (0.57 %)
(2001) No.16@150 3 No 10 @ 90 No.16 @ 165 No.16 @ 165
FRP
(Glass-1.00 %) (Carbon-1.50 %) (Glass-0.90 %) (Glass-0.76 %)
No.15M@ 160 No.15M@ 160 No.15M@ 240 No.15M@ 240
Steel
Magog (0.82 %) (0.70 %) (0.55 %) (0.47 %)
(2002) No 16 @ 150 3 No 10 @ 90 No.16 @ 150 No.16 @ 150
FRP
(Glass-0.87 %) (Carbon-1.34 %) (Glass-0.87 %) (Glass-0.75 %)
Morristown No 19 @ 100 No 19 @ 100 No 19 @ 150 No 19 @ 150
FRP
(2002) (Glass-1.95 %) (Glass-1.65 %) (Glass-1.30 %) (Glass-1.10 %)
Cookshire-Eaton No 19 @ 75 No 19 @ 100 No 19 @ 150 No 19 @ 150
FRP
(2003) (Glass-3.25 %) (Glass-2.00 %) (Glass-1.62 %) (Glass-1.33 %)
Val-Alain No 19 @ 125 No 19 @ 125 No 19 @ 185 No 19 @ 185
FRP
(2004) (Glass-1.60 %) (Glass-1.60 %) (Glass-1.08 %) (Glass-1.08 %)
Melbourne No 16 @ 100 No 16 @ 100 No 16 @ 200 No 16 @ 200
FRP
(2005) (Glass-1.36 %) (Glass-1.36 %) (Glass-0.68 %) (Glass-0.68 %)

3. FIELD TESTING
3.1 Instrumentation of the Bridges

All bridges are similarly instrumented at critical locations for internal temperature and strain data collection using
fiber optic sensors (FOS) and thermocouples (Figure 1a). Different types of Fabry-Perot (manufactured by Roctest
Ltd., St-Lambert, Quebec) and Bragg Grating (manufactured by Avensys Ltd., Cap. De la Madeleine, Quebec) FOS

320
were installed on reinforcing bars, embedded in concrete, or glued on the surface of the concrete or steel girders. In
addition, during testing, deflections of concrete slabs and girders were measured using a system of rulers and
theodolites (Figure 1b).

3.2 Static and Dynamic Load Test

Static and dynamic field tests using calibrated trucks were conducted on the bridges after the completion of
construction to evaluate the stress level in the FRP reinforcement, the concrete deck slab, and the girders. The tests
were carried out using either a single truck or two trucks simultaneously over the different paths, which are expected
to produce the maximum deflections and strains in both the reinforcement and concrete as shown in Figure 1c.

(a) FOS and electrical gages, (b) Deflection measurements, (c) Calibrated Trucks for field test,
(Val-Alain Bridge) (Val-Alain Bridge) (Cookshire-Eaton Bridge)
Figure 1. Instrumentation and field load test of the Bridges

4. FIELD TESTING RESULTS

4.1 Strain Measurements

Similar results were obtained for all bridges in terms of maximum measured strains in FRP bars, steel reinforcement,
and in concrete. The maximum change in the strain measured in the top and bottom FRP bars (glass or carbon) as
the truck moved across the gauge did not exceed 0.5% of the ultimate strain. The maximum concrete compressive
strain at the extremes top and bottom surfaces of concrete did not exceed 10 and 30 micro-strain, respectively.
Furthermore, the concrete tensile strains were calculated from the tensile strains measured in the FRP bars. The
maximum values of tensile strains at the top and bottom surfaces of concrete slab reached 18 and 45 micro strains,
respectively. These values were well below the cracking strain of concrete, εcr = 112 to 127 micro strain (for fc' = 27
to 37 MPa and Ec = 24 to 29 GPa).

Figures 2a and 2b show the maximum tensile strains measured in top and bottom transverse FRP reinforcement of
Val-Alain Bridge against time for different paths at speeds of 5 km/h. For the bottom transverse bars, the maximum
measured tensile strain were 53 and 48 micro-strains (compared to 57 for static loading) for speeds of 5 km/h and 50
km/h, respectively. For the top reinforcement, the maximum tensile strains of 22 and 20 micro-strains were recorded
for the two simultaneous trucks case at speeds of 5 km/h and 50 km/h, respectively.

60 60

50 B2
50 Path B B2
T3 T3 Path DE T3
S train (micro-strain)

T3
S train (micro-strain )

40 B2 40 B2

30 30

20 20

10 10

0 0
20 30 40 50 60 70 55 56 57 58 59 60 61 62
-10 -10
Time (sec.) Time (sec.)

(a) Maximum strains in bottom transverse FRP bars, (b) Maximum strains in top transverse FRP bars,
Figure 2. Strains in FRP reinforcement during dynamic load test (Val-Alain Bridge)

321
4.2 Deflection Measurements

During static tests, deflection of the concrete slabs and steel girders was measured with a theodolite and a system of
rulers installed across the mid-span section of the bridge (Figure 1b). The deflection of the concrete deck slab was
calculated by subtracting the measured value at the slab position from the average of the values measured on the two
girders adjacent to this position. For Morristown Bridge, the single truck traveling over the edge girder (following
Path D over girder E, as shown in Fig. 3a) produced the peak deflection of 7.0 mm (L/6270) in that girder. The peak
deflection for the middle girders was 8.0 mm (L/5490) measured for the case of the two trucks traveling
simultaneously. The maximum measured deflections for the concrete slab were less than 2 mm (S/1180) and were
obtained with one truck traveling directly above the measuring point. For Cookshire-Eaton Bridge, as shown in Fig.
3b, the single truck following a certain path B, C, or D (wheel load directly over the girders B, C, or D, respectively)
produced a peak deflection of 3.0 mm (L/8680) in the corresponding girder. The peak deflection for the two
calibrated trucks traveling simultaneously along Path B-D was 4.0 mm (L/6510) in girder C (Fig. 3b).

10
10 5.0
Girder E - Path D
99
4.0 Path B-D (Two Trucks)
Girder B - 2 Trucks
88 Path B (One Truck)
77 3.0

(mm)
Deflection (mm)
Deflection(mm)
(mm)

66
2.0

Deflection
Deflection

55
1.0
44

33 0.0
22 0.00 6.51 13.02 19.53 26.04 32.55 39.06 45.57 52.08
-1.0
11
GFRP-Reinforced Span Steel-Reinforced Span
00 -2.0
00 7.3
7.3 14.6
14.6 21.9
21.9 29.2
29.2 36.5
36.5 43.8
43.8

TruckTruck
location alongthethe
location along bridge
bridge (m) Truck location
Distance along
along the the(m)
bridge bridge
(a) On steel girders (Morristown Bridge) (b) On concrete girders (Cookshire-Eaton Bridge)
Figure 3. Deflection of girders Vs truck paths and locations

5. CONCLUSIONS

Based on the construction details and the results of the field tests, the following conclusions can be drawn:
1. The serviceability performance of the concrete deck slabs reinforced with FRP bars in terms of strain and
deflection was very similar to that reinforced with steel bars.
2. No obstacles to construction were encountered due to the use of the GFRP bars in the two concrete bridge deck
slabs. The GFRP bars withstood normal on-site handling and placement with no problems
3. The GFRP-reinforced bridge decks are well performing under very harsh environment. No additional or
propagation of cracks, if any, were observed under these severe service conditions
4. During the entire tests, the maximum tensile strains in FRP bars were less than 0.5 % of the FRP ultimate strain.
The maximum values of tensile strains in concrete slabs were well below the cracking strain of concrete.

6. REFERENCES
Benmokrane, B., El-Salakawy, E.F. El-Ragaby, A., and Lackey, T. (2006). Designing and Testing of Concrete
Bridge Decks Reinforced with Glass FRP Bars. ASCE J. of Bridge Engineering, Vol. 11, No. 2, pp. 217-229.
El-Salakawy, E.F. and Benmokrane, B. (2003). “Design and Testing of a Highway Concrete Bridge Deck
Reinforced with Glass and Carbon FRP Bars”. ACI Special Publication, Field Applications of FRP
Reinforcement: Case Studies, Detroit, Michigan, USA, SP-215-2, pp. 37-54.
El-Salakawy, E.F., Benmokrane, B., and Desgagné, G., (2003). FRP Composite Bars for the Concrete Deck Slab of
Wotton Bridge, Canadian Journal of Civil Engineering, Vol. 30, No. 5, October, pp. 861-870.
El-Salakawy, E.F., Benmokrane, B., El-Ragaby, A., and Nadeau, D. (2005). "Field Investigation on the First Bridge
Deck Slab Reinforced with Glass FRP Bars Constructed in Canada." ASCE Journal of Composites for
Construction, Vol. 9, No. 6, pp. 470-479.
Nanni, A., and Faza, S. (2002). Designing and Constructing with FRP Bars: An Emerging Technology, ACI
International, American Concrete Institute, Vol. 24, No.11, Detroit, USA, pp. 29-34.
Yunovich, M., and Thompson, N. (2003). Corrosion of Highway Bridges: Economic Impact and Control
Methodologies, ACI International, American Concrete Institute, Vol. 25, No.1, Detroit, USA, pp. 52-57.

322
Third International Conference on FRP Composites in Civil Engineering (CICE 2006)
December 13-15 2006, Miami, Florida, USA

OPTIMIZATION OF BRAIDED REINFORCED COMPOSITE RODS

Cristiana Gonilho Pereira
(Researcher, University of Minho,Guimarães, Portugal)

Raul Fangueiro
(Professor, University of Minho,Guimarães, Portugal)

Said Jalali
(Associate Professor, University of Minho,Guimarães, Portugal)

Mário de Araújo
(Full Professor, University of Minho,Guimarães, Portugal)

ABSTRACT
This work described the development of braided reinforced composite rods for concrete reinforcement. The research
study aims to analyse the influence of braided fabrics geometry on the core reinforced braided fabrics mechanical
behaviour. Moreover, this study intends to identify the influence of different fiber types used as core reinforcement
and of testing conditions on the mechanical properties of braided fabric composite rods,

KEYWORDS
Concrete, fiber reinforced composite materials, core reinforced braided fabric, core reinforced braided composite
rod.

1. INTRODUCTION
Typically, FRP rods are produced by pultrusion, which is a well-known manufacturing method in fabricating FRP
products with a constant cross section. In the pultrusion process, the longitudinal fibers are drawn through a resin
bath and then passed through a die, which gives the rod its final shape (Kadioglu et al, 2005). Therefore, FRP rods
present smooth surface and, when used as internal reinforcement for concrete, the bond at the interface between an
FRP rod and concrete is of paramount importance. The bond behaviour will have a direct influence on both the
serviceability and ultimate limit. To improve bond behaviour FRP-concrete, a surface treatment is required to
introduce deformations on the rod surface, and two different approaches can be considered: deformation of the
surface, due to the presence of ribs or indents or providing deformations in the outer resin layer, or surface
treatments, such as sand blasting or epoxy-coated sand (Lees, 2001). Besides pultrusion, FRP rods can also be
produced using braiding techniques (Soebroto et al, 1990). Braiding is a low cost technique allowing in-plane
multiaxial orientation, conformability, excellent damage tolerance and core reinforcement. Moreover, braiding
allows the production of ribbed structures and a wide range of mechanical properties may be improved when the
core braided fabrics are reinforced with the appropriate type of fibers (Fangueiro et al, 2006).

2. EXPERIMENTAL WORK
The current work aims to understand the influence of braided fabrics geometry on the core reinforced braided fabrics
mechanical behaviour. A study on the influence of the core reinforcement fiber type on the mechanical properties of
the core reinforced braided fabrics and of composite rods has also been undertaken. Moreover, the work aims to
understand the influence of testing conditions on the braided reinforced composite rods mechanical properties. The

323
 core reinforced braided fabrics have been produced on a vertical braiding machine. Braided reinforced composite
rods have been produced by impregnating the core reinforced braided fabrics on a vinyl ester resin, in a single step.

2.1. Optimal braiding angle of core reinforced braided fabrics

Eight polyester bobbins were used to produce the braided structure and two rovings of glass fiber were used as core
reinforcement. The braiding angle varies according to the braided fabric take-up rate. Braiding angles have been
measured for each braided fabric produced and tensile tests were carried out. Fabric delivery speed and braiding
angle are inversely proportional, as shown in Figure 1.
30,0

Braiding angle [º]

y = -480,45x + 29,367

Braiding angle [º]

25,0 2
R = 0,9523
20,0
15,0
10,0
5,0
0,0
0,00 0,01 0,02 0,03 0,04 0,05
Take-up
Take-uprate
rate(m/s)
[m/s]

Figure 1: Influence of take-up rate on braiding angle.

Analysing the influence of the braiding angle on the ultimate tensile strength and on the extension at failure, it may
be concluded that there is an inflection on the curve. The ultimate tensile strength and the extension at failure
increase as the braiding angle increases up to 18.6º. For braiding angles higher than 25º, both the ultimate tensile
strength and the extension at failure decrease as the braided angle increases (Figure 2).
[N] strength [N
Ultimate tensile strength

Extension at failure [%]

Extension at failure [%]

1000 3,50
y = -0,1544x + 6,6612
y = 13,422x + 633,35 3,25 2
900 2 R = 0,3661
R = 0,9116 3,00
Ultimate tensile

800 2,75
y = 0,0211x + 2,6057
y = -59,007x + 2359,2 2,50 2
700 2 R = 0,2232
R = 0,9442 2,25
600 2,00
8 12 16 20 24 28 8 12 16 20 24 28
Braiding
Braiding angle [º]
angle [º] Braiding angle[º][º]
Braiding angle

Figure 2: Influence of braiding angle on ultimate tensile strength and on extension at failure (mean values).

Based on the above results, it may be concluded that braided fabrics produced with 8 bobbins of polyester yarn, 2 of
them with 4 yarns, and reinforced with 1800 Tex glass fiber roving as core reinforcement, lead to higher values of
ultimate tensile strength and higher extensions at failure, when the braiding angles are between 18.6º and 25º.

2. 2. Core reinforced braided fabrics

Core reinforced braided fabrics were produced with a speed of production of 0,0156m/s. Glass, carbon, polyethylene
and sisal fibers were used as core reinforcement. Tensile tests were carried out on the different core reinforced
braided fabrics for different pre-loading conditions – 25N, 50N and 100N (Figures 4 and 5).

Braided fabrics reinforced with carbon fiber present the highest ultimate tensile stress (Figure 4); this does not seem
to be significantly affected by the pre-loading conditions. Braided fabrics reinforced with polyethylene HT fibers
present the highest values of extension at failure (Figure 4). The influence of pre-loading on extension is more
significant.

As it can be seen in Figure 5, the modulus of elasticity increases when the pre-load is increased from 25 to 100 N.
The carbon reinforced fabrics present the highest modulus of elasticity.

324
 Ultimate tensile stress [MPa]
Ultimate tensile stress
1250

Extension at failure [%]

4,0

Extension at failure [%]

1000
3,0

[MPa]
750
2,0
500
250 1,0

0 0,0
0 25 50 75 100 125 0 25 50 75 100 125
Pre-load [N]
Pre-load [N] Pre-load [N]
Pre-load [N]

Glass Carbon Polyethylene Sisal Glass Carbon Polyethylene Sisal

Figure 4: Influence of initial pre-load on the ultimate tensile stress (mean values).

of elasticity [GPa]
60
Modulus of elasticity

50
40
Modulus [MPa]

30
20
10
0
0 25 50 75 100 125
Pre-load [N]
Pre-load [N]

Glass Carbon Polyethylene Sisal

Figure 5: Influence of initial pre-load on modulus of elasticity (mean values).

The effect of initial pre-loading of the core reinforcement braided fabrics presents a significant influence on their
modulus of elasticity, as can be seen by analysing the core reinforced braided fabrics tensile behaviour (Figure 6).

1000,0

900,0
Stage II
800,0

700,0
a) Stage I
Load [N]

600,0
Load [N]

500,0

400,0

300,0 Stage III b) Stage II

200,0

100,0

0,0
0,0 2,0 4,0 6,0 8,0 10,0 12,0
Stage I Elongation [mm] c) Stage III
Elongation [mm]
Figure 6: Tensile behaviour of a core reinforced braided fabric.

Three stages can be identified in the load-elongation curve for a core reinforced braided fabric: Stage I – The load is
supported by the core reinforced fibers; even though, the fibers are not yet completely straight (Figure 6 a)); Stage
II – The reinforcement fibers are now completely straight and the load is supported by the core reinforced fibers.
There is a significant increase in the load required to stretch the reinforcement fibers to the breaking point (Figure 6
b)); Stage III –The braided fabric starts to bear the load due breaking of the core reinforcement fibers. Even though
braided structures present much better tensile properties comparatively to compressive ones, elongation is much
higher than that present by fiber rovings (Figure 6 c)).

2. 3. Braided reinforced composite rods

Braided reinforced composite rods have been produced on a vertical braiding machine with an incorporated
impregnation system. Tensile and bending tests were carried out on core reinforced composite rods.

325
Braided fabric composite rods reinforced with carbon fiber present the highest ultimate tensile stress, on both pre-
loading test conditions (Table 4). Regarding to extension at failure and modulus of elasticity, the best results are
obtained when core reinforcement fibers are subjected to 25N pre-load. Composite rods reinforced by carbon fibers
present significantly higher modulus of elasticity and one of the lowest extensions at failure. Braided fabric
composite rods reinforced with carbon fiber present the highest bending stress and significant higher bending
modulus (Table 5).

Table 4: Tensile test results for composite rods (mean values).

Pre-load Reinforcement Ultimate tensile Extension at Modulus of elasticity

[N] fiber stress [MPa] failure [%] [GPa]
Glass 537,9 4,5 9,7
Carbon 793,5 3,3 25,0
25
Polyethylene 525,3 3,9 8,7
Sisal 121,8 2,7 4,4
Glass 454,5 3,5 9,0
Carbon 685,7 2,6 23,3
100
Polyethylene 473,9 3,9 7,5
Sisal 114,6 2,3 4,3

Table 5: Bending test results (mean values).

Reinforcement fiber Bending stress [MPa] Bending modulus [GPa]

Glass 161,0 5,9
Carbon 351,7 20,3
Polyethylene 115,8 4,1
Sisal 103,0 3,0

3. CONCLUSIONS
It was concluded that for a braided fabric structure there is a braiding angle that promotes the optimized mechanical
performance of the core reinforced braided structure and. Analysing the test results obtained it is possible to identify
different performances among the different types of core reinforced braided fabrics and the different types of
braided reinforced composite rods. The trends in properties of core reinforced braided fabrics are similar to that of
composite rods. The mechanical behaviour of the core reinforced braided fabrics is mainly dependent on the core
reinforcement performance. It is also concluded that it is necessary to set a pre-tension on the reinforcement fibers to
guarantee an optimized mechanical behaviour of core reinforced braided fabrics.

This work is being funded by the Foundation for Science and Technology (Portugal) within POCI PROGRAMME,
Project POCI/CTM/6086/2004, “Development of braided reinforced composite elements for concrete reinforcement
and monitoring”.

5. REFERENCES
Fangueiro, R., Sousa, G., Araújo, M., Gonilho Pereira, C., Jalali, S., (2006), “Core reinforced composite armour as a
substitute to steel in concrete reinforcement”, International Symposium Polymers in Concrete – ISPIC2006, 2 – 4
April, Universidade do Minho, Guimarães, Portugal.
Kadioglu, F., Pidaparti, R. M. (2005), “Composite rebars shape efect in reinforced structures”, Composite
Structures, No. 67, pp 19-26.
Lees, J. M. (2001), “Fibre.reinforced polymers in reinforced and prestressed concrete applications: moving
forward”, Prog. Struct. Engng. Mater., No. 3, pp 122-131.
Soebroto, H.B., Pastore, C.M., Ko, F.K. (1990), “Engineering design of braided structural fiberglass composite”,
Structural Composites: Design and Processing Technology, 6th Annual Conference, Advanced Composites, Detroit

326
Part XI. Health Assessment
Third International Conference on FRP Composites in Civil Engineering (CICE 2006)
December 13-15 2006, Miami, Florida, USA

Nondestructive Evaluation of FRP Bonding by Shearography

F. Taillade
(Researcher, Design and Physical Systems Unit - French Public Works Research Laboratory - LCPC, Paris,
France)

M. Quiertant
(Researcher, Structural Engineering Unit - French Public Works Research Laboratory - LCPC, Paris, France)

C. Tourneur
(Technical Manager, Freyssinet France, Palaiseau, France)

ABSTRACT
This paper describes the application of shearography to the evaluation of the adhesion of externally bonded
fiber-reinforced polymer (FRP) on concrete surface. The principle of the method is first reviewed. Theoretical
and experimental analysis of the disbonds is presented. To validate the feasibility of the proposed method,
different sizes of defects are simulated by replacing adhesive by TEFLON® discs between the concrete and the
carbon epoxy film at known locations of slab type concrete model specimens. To perform shearography analysis,
these samples are loaded by vacuum stressing. Presented results show that shearography enables one not only to
determine locations and areas of bond defects but also to evaluate the quality of adhesion (assessment of partial
debonding).

KEYWORDS
Nondestructive testing, shearography, disbonds, concrete, carbon epoxy, fiber-reinforced polymer, FRP.

1. INTRODUCTION
The strengthening or retrofitting of existing concrete structures must be accomplished according to design and
construction guidelines to ensure the durability and long-term performance of the FRP strengthening system
(ACI Committee 440, 2002; AFGC, 2003; fib Task Group 9.3, 2001). After being applied, these systems (fibers
and resins) have to be evaluated to check their conformance with specifications. The quality-control program
must be achieved through a set of inspections and tests. Methods such as acoustic sounding (coin or hammer
tap), ultrasonics and thermography are used to detect delaminations (disbonds). These methods are efficient to
locate bond defects but are not capable of quantifying evaluating the quality of adhesion of the FRP to the
substrate (partial delamination, damage of the resin, bad mechanical property of the resin). Consequently, a
shearography nondestructive evaluation (NDE) method for FRP bonding was previously developed by Taillade
(Taillade, 2006) and is here applied in the field of the strengthening of concrete structures.

2. PRESENTATION OF THE NONDESTRUCTIVE METHOD

Shearography is a speckle interferometric technique providing full-field and in near-real time quantitative images
of structural surface displacements. This technique can be applied to detect the disbonds in a structure composed
of a concrete substrate, one layer of adhesive, and one layer of carbon-epoxy composite. Principles of the
method are those described by (Hung, 2001) but application proposed by authors is also capable of evaluating
the bonding quality between the jacket and the concrete.

2.1 The shearography technique

The principle of an interferometer with a video split, called shearography, is to cause the interference of two
waves that had been submitted to nearly the same random fluctuations in optical path during their trajectories
between the studied object and the CCD (Charge Coupled Device) matrix. In order to produce this effect, one

327
conventional technique consists in carrying out a differential measurement of the optical phase ϕ proportional to
the product of the refraction index and the geometrical length along the respective path of each of the two beams.
The shearography based on this principle thus causes the interference of two waves coming from two points
close to the object A and B and separated by a quantity δ x (1 mm≤ δ x ≤ 10 mm). Various devices allow
determining this shift, which at times is referred to as the rate of shear; these would include: the Michelson
interferometer (see Figure 1 and Leendertz, 1973), glass corner (Hung, 1979), biprism (Hung, 1989), and
development defect evaluation (Hung, 1974). The set-up using the Michelson interferometer enables one
regulating the rate of shear by simply tilting one of the two interferometer mirrors and then implementing the
phase-shift technique (Creath, 1994), by means of translating the other mirror.

In the reference state, ϕ is equal to ϕS, which represents a random dephasing due to surface roughness and
topography of the target object. This random dephasing lies at the origin of the speckle figure that may be
observed by examining a rough surface lit by a coherent light source. By assuming that the object undergoes a
small deformation, the measured phase becomes: ϕS +∆ϕ, where ∆ϕ denotes the variation in dephasing between
the two states.

Figure 1: Principle of a shearographic interferometer set-up

In the case of plane waves, for directions of illumination and observation normal to the plate, and with a shear
distance δ x small compared with the characteristic distances over which the deformations occur, the phase
difference is expressed at the first order by:

4π  ∂w 
∆ϕ =  δ x
λ  ∂x 

in which w is the amplitude of the displacements normal to the object surface and λ is the illumination laser
wavelength.

By performing an uncertainty budget (Taillade, 2006), we found that the phase difference equivalent to noise is
roughly 2π/50 (noise including calibration procedure). Without any particular precaution (in field applications),
displacement difference can thus be mapped with a 5 nm uncertainty.

2.2 Excitation method

The partial vacuum (or depressure) ∆P is the most employed in nondestructive testing by shearography (Clarady,
1993; Deaton, 1993) to diagnose aircraft structures (Newman, 1991; Bobo, 1991) and the cryogenic tanks of
rockets (Burleigh, 1993). It makes it possible to detect very well various kinds of defects and mainly
delaminations in composite materials as well as disbonds in the metal structures. The difference in pressure
between the blade of air inside the defect and the surface subjected to the stress creates a deformation in the
shape of "bump" into defect. The depressure which should be applied to obtain a measurable deformation by
shearography can be very weak (a few Pascal). It depends primarily on the mechanical characteristics of material
analysis (elasticity coefficients) and of width/depth ratio of the defect. To envisage the depression to be applied
and to analyze the field of deformation to be measured, we use finite elements models in order to design
abacuses usable by the operators. It is then possible to consider that concrete substrate is itself cracked or other
producing a particular field deformation. The stressing can be applied by a weak depressure to surface study by
means of a suction cup.

328
3. EXPERIMENTATIONS
3.1 Description

Two concrete samples (300 x 300 mm2) have been manufactured (Figure 2). The defects are simulated in
replacing adhesive by a TEFLON® disc (width 0.5 mm) between the concrete samples and the carbon epoxy
film. Sample n°1 contains four different discs (40, 30, 20, 10 mm dia) and sample n°2 contains four discs with
identical 40 mm diameter (Ø 40) with different holes (in number and size represented by the percentage of
remaining disc mass) to vary quality of adhesive properties. In the experimental set-up (Figure 3), we show that
shearography visualizes deformation through suction cup (Plexiglas® chamber 180 mm x 180 mm x 70 mm and
with 20 mm wall thickness).

Ø 40 Ø 30 80 mm 0% 45%
80 mm

300 mm
300 mm
Ø 10 Ø 20 30% 21%

80 mm
80 mm

80 mm 300 mm 80 mm 80 mm 300 mm 80 mm
(a) Sample n°1: variable diameter (in mm) (b) Sample n°2: variable adhesive properties

Figure 2: Concrete samples with bonded carbon epoxy film containing different disbonds (made of
Teflon® discs, size and form variables)

Sample
Shearographic camera

PressureMeter

Beam expander Suction cup

Figure 3: Experimental set-up

3.2 Results

The results obtained with sample n°1 (Figure 4) show the phase difference measured by shearography through
suction cup. The applied depressure, to visualize defects by shearography, increases when the diameter of
defects decreases. The magnitude of pressure is about 100 hPa± ∆P/2 that is not detrimental to structure. With
sample n°2, we show (Figure 5) that when the percentage of hole on disc is more important (i.e. when the
disbond is less important), it is necessary to impose a more important depressure to measure a difference optical
phase of the order of 2π . All the results are confirmed by the finite elements analysis.

329
(a) Ø40 mm, ∆P=8.0 hPa (b) Ø30 mm, ∆P=10.6 hPa (c) Ø20 mm, ∆P=26.6 hPa (d) Ø10 mm, ∆P=133.0 hPa

(e) Adhesive 0% (f) Adhesive 21% (g) Adhesive 30% (h) Adhesive 45%
∆P=8.0 hPa ∆P=53.2 hPa ∆P=79.8 hPa ∆P=186.2 hPa

Figure 4 : Measured strain around disbonds of different sizes on sample 1, under different partial vacuum
levels (a, b, c, d) and disbonds (40 mm dia) on sample 2 for different adhesive percentage (d, e, f, g)

4. CONCLUSIONS
In this paper, the principle of shearography NDE method is reviewed. An application of this method to the
evaluation of the adhesion of externally bonded fiber-reinforced polymer on concrete surface is presented.
Results demonstrate that this NDE method enables one to determine not only locations and areas of defects but
also to quantify the adhesion in the case of a partial debonding. Moreover, this technology is particularly adapted
to field evaluation due to its real time assessment capability and hand portability. This method for local
evaluation of bond defects can be complementary to thermography for global and rigorous inspections of
repaired structures.

5 . REFERENCES
ACI Committee 440.2R02. (2002). Guide for the Design and Construction of Externally Bonded Systems for
Strengthening Concrete Structures, ACI, Michigan, U.S.A.
AFGC. (2003). Réparation et renforcement des structures en béton au moyen des matériaux composites –
Recommandations provisoires, Bulletin scientifique et technique de l’AFGC. (in French).
Bobo S. (1991). Shearographic inspection of a Boeing 737. Technical report, Federal Aviation Administration.
Burleigh D. D., Engel J. E. & Kuhns D. R. (1993). “Laser shearographic testing of foam insulating on cryogenic
fuel tanks”. Review of Progress in Quantitative Nondestructive Evaluation, Vol. 12, pp 411-418.
Clarady J. F. & Summers M. (1993). “Electronic holography and shearography NDE for inspection of materials
and structures”. Review of Progress in Quantitative Nondestructive Evaluation, Vol. 12, pp 381-386.
Creath K. (1994). Phase shifting holography interferometry. In Rastogi P. K. (ed), Holographic interferometric.
68, (Springer Series in Optical Science), pp 109-150.
Deaton J. B. & Rogowski R. S. (1993). “Electronic Shearography: current capabilities, potential limitations and
future possibilities for industrial nondestructive inspection”. Review of Progress in Quantitative Nondestructive
Evaluation, Vol. 12, pp 395-402.
fib Task Group 9.3. (2001) Externally bonded FRP reinforcement for RC structures, fib bulletin 14, Lausanne,
Switzerland.
Hung M. Y. Y. (2001). “Shearography and applications in nondestructive evaluation of structures”, Proceedings
of the international conference on FRP Composites in civil engineering 2001; p. 1723-1730.
Hung Y. Y. (1974). Optics Communication, Vol. 11, No. 2, pp 132-135.
Hung Y. Y. (1979). Applied Optics, Vol. 18, No. 7, pp 1046-1051.
Hung Y. Y. J. (1989). Nondestructive Evaluation, Vol. 8, No. 2, pp 55-67.
Newman J. W. (1991). “Shearographic inspection of aircraft structure”. Materials Evaluation, Vol. 49, No. 9, pp
1106-1109.
Leendertz J. & Butters J. J. (1973). Phys. E : Sc. Inst., Vol. 6, pp 1107-1110.
Taillade F. (2006). “Metrological Analysis of Shearography”. European Physical Journal – Applied Physic, to
be published.

330
Third International Conference on FRP Composites in Civil Engineering (CICE 2006)
December 13-15 2006, Miami, Florida, USA

Structural Health Monitoring in cold climate of a CFRP strengthened

concrete hollow box girder bridge
Björn Täljsten
(Professor, Denmark Technical University, Lyngby, Denmark
Luleå University of Technology, Luleå, Sweden)

Arvid Hejll
(Tech Lic., Luleå University of Technology, Luleå, Sweden)

ABSTRACT

CFRP (Carbon Fibre Reinforced Polymer)-strengthening with plates and sheets of existing concrete structures is
today a well-known method, proven with good results all over the world. However, limited research has been carried
out regarding the behaviour of CFRP strengthen structures in cold climates. This paper addresses field testing of a
CFRP strengthening pre-stressed concrete hollow box-girder bridge in Stockholm, Sweden, the Gröndals Bridge.
Also how the strengthening behaves in cold climate is investigated. Two long term monitoring systems has been
installed on a, one traditional system with LVDTs (Linear Vertical Displacement Transducers) mounted over cracks
for continuous monitoring and one system with FOS (Fiber Optical Sensors) mounted over cracks and on CFRP-
plates. Results show that difference in temperature causes most of the live load. Other results from summer and
winter measurements shows that the CFRP plates are not tensioned during winter. A conclusion from this is that the
strengthening would have been much more effective if the CFRP would be applied in the winter when the existing
structure is contracted due to the temperature effect. Furthermore, the opening of cracks could be followed over
time.

KEYWORDS
Structural Health Monitoring (SHM), CFRP Strengthening, cold-climate, strengthening effect

1. INTRODUCTION

To obtain the most out of a measurement program it is important to carry out SHM in a structured way and that a
well-planned procedure is followed c.f. (Hejll 2004). When it comes to long term monitoring there are different
strategies on how to monitor the structure, e.g. continuous, periodic, automatically triggered and manual triggered.
This may be different from one bridge to another and it can also be changed during time. For example, in the
beginning of a monitoring project continuously monitoring may be preferable but later the measurements are taken
once a week e.g. periodic monitoring or only when traffic passes the bridge e.g. automatically triggered monitoring.
One also discusses global and local monitoring. Local monitoring is defined as the monitoring undertaken on a
certain part of the structure. It is useful for laboratory tests or for structures with known damages where the defined
area is equipped with sensors. To monitor a global behaviour of a structure other parameters are needed, parameters
or methods that can provide information of the whole structure. For the Gröndals Bridge a manual triggered long
term monitoring method with local monitoring were used for the FOS (Fibre Optic Sensors).

In 2002 a monitoring and strengthening program was launched at the Gröndals Bridge, a large concrete hollow box
girder bridge with severe shear cracks in the web were found, just two years after it was opened to tram traffic. The
purpose with the monitoring is twofold, firstly to investigate if the crack opening is stable over time and do not
propagate, secondly to use and evaluate FOS for monitor crack opening on concrete. In addition to this a comparison
between the traditional monitoring technique and the FOS technique has been carried out. Furthermore, we also
wanted to use the FOS monitoring to investigate any strengthening contribution from the CFRP plates.

331
2. STRENGTHENING OF THE BRIDGE
The main span of the Gröndals Bridge is 120 meters with two adjacent spans each of 70 meters. The bridge carries
two railway tracks which are placed symmetrically about the cross-section of the bridge. Bridge inspections carried
out on the newly built, 2000, Gröndals Bridge revealed extensive cracking in the webs of its concrete hollow box-
girder section. The bridge was designed to the currently applicable Swedish codes, BRO 94 and BBK 94. On the
basis of these regulations, it was possible to erect the bridge with extraordinarily slender webs. Relatively high shear
stresses and principal stresses are generated because of the small web widths although the webs are fully compressed
considering the normal stresses caused by longitudinal pre-stressing. Furthermore, the permanent loads on the
structure are dominant. As the permanently exerted principal tensile stresses reached the value of the tensile strength
of the concrete, shear cracks were finally created. Additionally, restraining bending moments have been
superimposed in the webs due to sun radiation. Assuming a linear temperature difference of 10 to 15 K, this,
together with the other transverse bending moments, additionally causes vertically directed tensile stresses at the
inside of the web amounting to approximately, σz = 2 to 3 MPa. The positions of the cracks in longitudinal direction
of the bridge correspond to the areas of the maximum principal tensile stresses. The cracks first appeared after only
a few years of service and subsequent inspections showed that the number and size of the cracks were increasing,
(James 2004). The cracks widths were between 0.1 – 0.3 mm and for a few isolated cases between 0.4 – 0.5 mm in
the most cracked sections. Investigations as to the cause of the cracking suggested that they were due to inadequate
web shear reinforcement. The webs have a thickness of 350 mm and a total height of the box girder close to the
main span supports of approximately 7.5 m. In addition, the flanges are quite thick; the bottom flange is, at most,
about 1300 mm. To increase the safety level of the bridge strengthening was decided. Because of the progressive
nature of the cracking combined with wariness for shear cracks, the bridge was temporarily closed for traffic
towards the end of 2001 and temporarily strengthened with external steel stays. Final strengthening were carried out
with prestressed steel stays in areas with extensive and large cracks and CFRP plates were used in areas with minor
or no cracks to limit the number and size of future possible cracking. In this paper only CFRP strengthening are
further discussed. The strengthening work was carried out in spring 2003 and the average temperature on the
structure at time for strengthening was approximately 15 °C. Before the strengthening work started the concrete
surfaces were sandblasted and holes for anchoring the CFRP plates were drilled in the upper and lower flanges.
Plates were only placed on the inside of the bridge. The surfaces were thoroughly cleaned with pressurized air and
vacuum cleaners. The surfaces to be bonded were treated with a primer for the system to enhance the bond. The
CFRP plates were bonded to the surface, i.e. the webs of the structure, with a high quality epoxy adhesive, BPE®
Lim 567, specific for the strengthening system used. The Young’s modulus of the adhesive is approximately 6.5
GPa at 20 °C. Thermal coefficient for the adhesive was approximately 25-35 10-6 /°C . The average thickness of the
adhesive was 2 mm. A total of 2 500 m of CFRP plates was used for the bridge. The Young’s modulus of the plates
was 250 GPa with a failure strain of 11‰ and the thermal coefficient was approximately 6 10-6 /°C. The CFRP
plates have been bonded at an angle of 70° to the horizontal plane. This was in order for the plates to be bonded
perpendicular to the direction of the cracks. To enhance the anchorage steel plates with welded steel bars were
bonded to the face of the CFRP plates and anchored by epoxy bonding in pre-drilled holes in the top and bottom
flanges. The anchorage length was approximately 250 mm.

3. MONITORING

The monitoring systems installed can obviously not capture the effects of the dead load, and consequently only the
relative changes with regard to temperature and live load could be recorded. The FOS sensors are of type Bragg
grating and at most 7 sensors were written on one fibre, the length of each senor was approximately 20 mm.
Additionally to the FOS system a traditional long-term monitoring system was installed by the Royal Institute of
Technology, Sweden, c.f. (James 2004). This system involved the use of LVDT’s and temperature sensors in the
form of thermocouples, however, in this paper focus is placed on the FOS system. Two temperature sensors were
also added to the traditional system and are positioned on the east web of the box girder, one on the inside and one
on the outside. The data received from these temperature sensors are used to investigate the effects of temperature
on the crack widths. In total, 32 FOS have been installed on the Gröndals Bridge, they are all installed in section A
and section B, both on the concrete and on two of the CFRP plates, see Fig. 1. In table 1 the sensors are presented
systematically. For the FOS system, the previously mentioned sensors for temperature compensation shall also be
included; however, this is not recorded in table 1. The results presented in this paper are mainly from the west side
of the bridge, the side that was most warmed by the sun.

332
 Section A
Sensor Measure
GAW1-6 Crack Opening
GWR1-3 Strain Concrete
GAWC1 Strain CFRP
GAE1-4 Crack Opening
GAEC-1 Strain CFRP

Section B
Sensor Measure
GBW1-3 Crack Opening
GBWC1 Strain CFRP
GBE1-4 Crack Opening
GBEC1 Strain CFRP

Figure 1. Position of gauges on the Gröndals Bridge. Table 1 FOS sensors for monitoring

It is not possible to present all the results from the measurements and therefore only the most interesting findings are
presented. However all data from the monitoring up to March 2004 may be found in (James 2004) and (Täljsten and
Hejll 2005), where the results from the monitoring are presented for the traditional/ and FOS systems respectively. It
was not possible to carry out a continuous FOS monitoring due to the high cost of the system; therefore a periodic
monitoring was adopted. It was found that using the chosen FOS system was convenient, in particular for periodic
monitoring. The presented results from the monitoring scheme are divided into periodic summer and winter
monitoring respectively, on both occasions the monitoring was carried out for approximately 48 hours. FOS sensors
were placed over cracks and monitoring is carried out during live loading with trams. The weights of the trams are
well defined. The periodic summer monitoring was carried out from the 21st to the 23rd of May 2003. and the
periodic winter monitoring was carried out from the 10th to the 12th of March 2005. Fig. 2 and Fig. 3 show selected
displacement curves for the FOS monitoring during summer and winter, respectively. With the exclusion of the dead
load, it can clearly be seen in both of these curves that it is the temperature on the bridge that has the largest impact
on the opening of the cracks and the opening follows a daily variation. GAW3 monitor the largest crack opening,
approximately 0.065 mm. On the curve for GAW3 also minor vertical lines can be seen. These lines indicate the
passages of the trams and give approximately a crack opening of maximum 0.001 mm. It seems that the effect of the
temperature on the opening and closing of the cracks is at least tenfold that of the traffic load. This is confirmed by
the traditional monitoring system, where the effect of passing trams is investigated using a sampling rate of 200 Hz,
c.f. (James 2004, Sundquist and James 2004).
3/10/05 3/11/05 3/11/05 3/11/05 3/12/05
0.080 26
GAW1 0.080 5
24
GAW2 GAW2
0.060 GAW3 22 GAW3
GAW4 0.060 GAW4 4
20
GAW5
Displacement, [mm]

0.040 Temp GAW5

Temperature, [oC]

GAW6 18 GAW6
Displacement, [mm]

Temp 0.040 3
Temperature, [oC]

Temp Temp
16
0.020
GAW4 14
0.020 2
12
0.000 GAW4
10 0.000 1
GAW3
-0.020 8 GAW2
6 -0.020 GAW6 0
GAW5
-0.040 4
2 -0.040 -1
GAW3
-0.060 0
-0.060 -2
0 10 20 30 40 50
Time, [h] 0 10 20 30 40 50
Time, [h]
Figure 2. Results from crack-displacement with FOS Figure 3. Results from crack-displacement
monitoring system – Summer monitoring with FOS system – Winter monitoring

333
For the winter monitoring it can be seen that all cracks are compressed and the opening is negative. In addition to
this the crack displacements are fairly small. Furthermore, since the temperature variations for the winter monitoring
are considerably less compared to the summer monitoring the changes in crack opening during winter are very
small. However, also during winter the displacement curves follow the temperature curves very well. In the figures
the effect of the tram traffic can be seen very clear as vertical lines on the displacement curves. The strain values on
the CFRP plate on the east side was distinct, both for the summer and winter monitoring, see Fig. 4 to and Fig. 6.
These plot shows that the sensor on the plate follows the temperature on the east side very well. Additionally to this
it can also be noticed that the plate is under compression during the whole periodic monitoring time period with a
maximum measured strain of -210 μstr which then correspond to a compressive stress of approximately 55 MPa.
The summer monitoring in the CFRP on the west side give a tensile stress in the CFRP plates corresponding to
approximately 30 MPa.

160 26 160 26
24 -150 5
24

120 22 120 22
GAWC1 4
20 20
Temperature, [oC]

Temp

Temperature, [oC]
18 18 -175
80 80 3

Temperature, [oC]
GAEC1
Strain [ustr]

Strain [ustr]
16 16

Strain, [μstr]
Temp
14 14
40 2
40
12 12 -200
10 10 1
0
8 0
8 Temp
GAEC1
6 0
6 -225
-40 4 -40 4
2 -1
2
-80 0
-80 0 -250 -2
0 10 20 30 40 50
Time, [h] 0 10 20 30 40 50 0 10 20 30 40 50
Time, [h]
Time, [h]

Figure 4. Results from FOS Figure 5. Results from FOS Figure 5. Results from FOS
measurement on CFRP– Summer measurement on CFRP– Summer measurement on CFRP– Winter
monitoring - West monitoring East monitoring - East

4. SUMMARY AND CONCLUSIONS

This paper presents both a periodic monitoring from the summer and the winter period and the monitoring shows
that the cracks were not propagating and that the openings for the cracks were very small. The largest crack opening
measured was approximately 0.06 mm. Furthermore, the temperature effect was at least 10 times larger than the
effect from the tram traffic. Comparing the measurements from summer and winter gives the relative effect that the
cracks are compressed during the cold period. This is also valid for the measurements on the CFRP plate, which
seems to be in tension during the summer and compressed in the winter.

ACKNOWLEDGEMENTS

The research presented in this paper has been funded by several organizations. Here the
Development fund of the Swedish Construction Industry (SBUF), Stockholm Transport (SL) and Skanska Sverige
AB should be acknowledged. Also City University in London, UK, is acknowledged for their contribution FOS
measurements.

REFERENCES

Hejll, A. 2004 “Structural Health of Bridges”. Licentiate thesis 2004, Division of Structural Engineering, Luleå
University of Technology, ISSN 1402-1757/ISRN LTU-LIC—04/46—SE/NR 2004:46
James G., 2004, Long term health monitoring of the Alvik and Gröndal Bridges, TRITA-BKN Rapport 76,
Byggkonstruktion, 2004, Kungliga Tekniska Högskolan, Stockholm, Sweden
Sundquist, H. and James, G., 2004, Monitoring of shear cracks and the assessment of strengthening on two newly-
built light-rail bridges in Stockholm. In Proceedings of the Second International Conference on Bridge
Maintenance, Safety and Management, IABMAS, Eds. E. Watanabe, D. Frangopol and T. Utsunomiya, Kyoto,
Japan, pp 257-258.
Täljsten B. and Hejll A. (2005). Tvärbanebroarna – Gröndal och Alviksbron: mätning av rörelser med hjälp av
fiberoptiska sensorer (FOS), Technical Report, LTU, Luleå, Sweden, Structural Engineering, ISSN 1402-
1536/ISRN LTU-TR-05/01-SE/NR 2005:01 (In Swedish).

334
Third International Conference on FRP Composites in Civil Engineering (CICE 2006)
December 13-15 2006, Miami, Florida, USA

Structural Health Monitoring of degrading concrete beams in a

laboratory environment
Markus Bergström
(PhD student, Luleå University of Technology, Luleå, Sweden)

Björn Täljsten
(Professor, Technical University of Denmark, Lyngby, Denmark
Luleå University of Technology, Luleå, Sweden)

ABSTRACT
Much effort has been invested separately on degradation, repair and upgrading of concrete structures. However, few
holistic studies including laboratory testing have been performed on the entire cycle during a structures life.
Reinforced concrete is the most widely used building material in the world. Normally the life of concrete structures
is very long. However, concrete structures possess one drawback; at least in severe environments, the steel
reinforcement may corrode. The effects can clearly be seen when the steel reinforcement is attacked by chlorides.
The reduced steel cross-section area and loss of bond strength between steel and concrete will lead to increased
deformations, cracking and premature ultimate load, thus affecting both the serviceability limit state (SLS) and the
ultimate limit state (ULS). Procedures to repair and upgrade the damaged structure are used to increase the structural
performance. They follow the structural member of time, a SHM (Structural Health Monitoring) approach is adapted
to this project. By applying SHM to a degrading structure it assures that it will keep up to current standards by
continuous monitoring, analysing, evaluation and eventually also retrofitting.

KEYWORDS
Structural Health Monitoring (SHM), degradation, corrosion, chloride, curvature, stiffness, life cycle

1. INTRODUCTION

1.1. In general
The current paper is generated from a large research project. The project aim is to experimentally and numerically
simulate the behaviour of concrete beams enduring a simulated life cycle procedure. The test program follows the
beams from original performance of the intact beam through degradation, repair and upgrading with FRP plate
bonding to its original load carrying capacity. Several attributes make this project unique. Amongst these are
accelerated corrosion, strain measurement using fibre optic sensors and sustained loading during the entire life cycle.
In addition, mid span curvature is monitored using a special test setup.

1.2. Structural Health Monitoring (SHM)

SHM is connected to a continuous method containing monitoring, analysis, assessment and also possible retrofitting
of a structure. This way of thinking is important since concrete structures, although appearing stable and solid, are
dynamic in terms of changing material properties and internal hard-to-see damages, such as corrosion of steel
reinforcement (Hejll, 2005). This project gives the opportunity of applying this way of thinking in the controlled
manner of an advanced laboratory test program, presented below.

335
1.3. Defined life cycle
The simulated life cycle carried out may be divided into seven stages, from a to g, presented in Figure 1 and in the
text below the figure, (Horrigmoe, 1998 and Sand, 2001). This may be a normal life cycle for a concrete structure,
e.g. a bridge which is located in an environment that is aggressive due to steel corrosion.

a. e.

b. f.

c. g.

d.

Figure 1. Seven stages create the studied life cycle (Horrigmoe, 1998).

a. In the initial stage, the beam is subjected to the full service load (SLS), during which cracking in
the tension side occurs.

b. Simultaneously as the serviceability load acts on the beam specimens, accelerated corrosion
attacks the tensile flexural reinforcement in mid-span. Thereby, the cross sectional area of the
attacked bars and the bond between the bars and concrete is reduced.

c. The structure is taken out of service, which means that the variable component of the
serviceability load is removed. Dead load remains acting on the specimens.

d. Cracked and chloride contaminated concrete is removed and the tensile reinforcement is exposed
over the entire deteriorated region of the beam. The bars are cleaned by sandblasting and will have
a permanently reduced cross section

e. After the concrete is removed the cavity is refilled with a repair mortar. This is strain-free when
applied, whereas the neighbouring concrete remains strained and cracked.

f. After repairing the beam, it is strengthened in flexure using CFRP plates that are applied on the
tensioned face of the beam. The strengthening procedure increases the stiffness of the beam, but
does not add any significant weight.

g. The life cycle is closed by finally loading the beams to failure.

2. RESULTS

2.1. Degradation stage

Tensile steel reinforcement was corroded by an accelerated corrosion setup during the 70 day degradation period.
The decrease in unit weight due to corrosion was established to be 14% as shown in table 1. Stiffness was
meanwhile monitored using a curvature test setup. A 15% reduction in stiffness was observed after 70 days of
corrosion, see figure 2.

336
Table 1. Unit weight and diameter for non-corroded and corroded steel reinforcement bars.

Unit weight [kg/m] Diameter [mm]

Non-corroded steel 1,67 16
Corroded steel @ 70 days 1,48 15,1
Reduction 14% 6%

4000

3000
15%
Stiffness [kNm2]

2000

1000

0
0 20 40 60 80
Corrosion time [days]

Figure 2. Stiffness measured during the time when corrosion attacked tensile steel reinforcement.

2.2. Life cycle behaviour

FE-calculation (Sand, 2001) has shown that the life cycle behaviour, in terms of the load deflection relation follows
a load-deflection path shown by the left part of figure 3. First, the intact beam in stage a) is loaded up to the service
load. Deflection increases as corrosion reduces the beam stiffness by reduction of the steel reinforcement content.
The concrete beam has to be repaired at a certain corrosion level. The variable load is removed before the repair
procedure. This is normally carried out for an existing structure that is taken out of service before being
rehabilitated. A further increase of deflection is predicted during the repair procedure as steel reinforcement will
move towards the centre of gravity, giving a reduced effective height. This effect arises due to the dead load acting
on the beam during removal of damaged concrete. Another mechanism giving increased deflection during repair is
that the remaining bond between steel bars and concrete is fully eliminated. It is possible that the repaired beam will
need strengthening to reach the capacity of the intact beam again.

The experimental result from the test at LTU is provided in the right part of figure 3. The general appearance of the
two graphs corresponds to each other, although no FE-calculation has at this time been carried out for this particular
experimental study.

337
 100 Intact
Repaired and strengthened
Corroded
Repaired
Failure load
80 Failure load

Cover concrete removed

60 g

Load [kN]
g
Service load
b
Load

c
Dead load 40 b Service load
d e f
a
a c
20 Dead load
def

0
Deflection
0 20 40 60 80
Deflection [mm]

Figure 3. Left: General life cycle behaviour shown by FE-simulation (Sand, 2001). Right: Life cycle behaviour
given by bold line, shown by experiment at LTU.

3. CONCLUSIONS

The experimental study has provided promising results to understand the life cycle behaviour of concrete structures
during degradation, repair and upgrading. Fundamental mechanisms behind the behaviour are identified. The next
step in this study could be to work out principles to minimize the bad influence of these mechanisms, and even take
advantage of them. The adaptation of SHM made it possible to monitor the structural behaviour during the different
stages.

4. ACKNOWLEDGEMENTS

The research presented in this paper has been funded by several organisations. Here the
Development fund of the Swedish Construction Industry (SBUF), Sto Scandinavia AB and Skanska Teknik should
be acknowledged.

5. REFERENCES

Arntsen, B. 2005. “Akselererte korrosjonsforsøk, Forslag til gjennomføring”. NORUT Teknologi AS, Narvik,
Norway.
Horrigmoe, G. 1998. “Future needs in concrete repair technology”. NORUT Teknologi AS, Narvik, Norway.
Sand, B. 2001. “Nonlinear finite element analysis of deteriorated and repaired RC beams”. NORUT Teknologi AS,
Narvik, Norway.
Täljsten, B. 2004 “FRP Strengthening of Existing Concrete Structures, Design Guidelines”. 3rd edition, Luleå
University of Technology, Luleå, Sweden.
Hejll, A. 2005 “Structural Health of Bridges”. Licentiate thesis 2004:46, Division of Structural Engineering, Luleå
University of Technology, Sweden.

338
Third International Conference on FRP Composites in Civil Engineering (CICE 2006)
December 13-15 2006, Miami, Florida, USA

DEBONDING DETECTION IN CFRP STRENGTHENED RC BEAMS

USING ACTIVE SENSORS
Seung Dae Kim
(Ph.D. Candidate, Carnegie Mellon University, Pittsburgh, PA, USA)
Chi Won In
(Master Student, Carnegie Mellon University, Pittsburgh, PA, USA)
Kelly E. Cronin
(Undergrad. Student, Carnegie Mellon University, Pittsburgh, PA, USA)
Hoon Sohn
(Assistant Professor, Carnegie Mellon University, Pittsburgh, PA, USA)
Kent Harries
(Assitant Professor, University of Pittsburgh, Pittsburgh, PA, USA)

ABSTRACT
The appropriate bonding condition between substrate reinforced concrete (RC) beams and carbon fiber-reinforced
polymer (CFRP) laminates is essential to guarantee the performance of CFRP strips as retrofitting materials. In this
study, a theoretical approach and experimental results toward developing a CFRP debond monitoring system are
presented. The goal of this study is to understand suitable guided waves propagation in the CFRP-RC structure and
to develop a new theoretical framework of nondestructive testing (NDT), in which debonding can be detected
without using past baseline data. The concept of time reversal acoustics (TRA), which has been generally applied to
body waves, is extended to complex guided waves in CFRP-RC interfacial regions. Several indices sensitive to
structural damage are extracted by comparing the known input with the time-reversed signal. Active sensing devices
such as lead zirconate titanate (PZT) wafers are used to generate known input waveforms and to measure the
time-reversed responses. Two large-scale CFRP-retrofit RC beams, one tested monotonically and the other in a
fatigue regime, are used to demonstrate the potential of the proposed debonding monitoring system.

KEYWORDS
active sensing, baseline-free nondestructive testing, debonding, structural health monitoring, time-reversal acoustics.

1. INTRODUCTION
Carbon fiber reinforced polymer (CFRP) composites have become an attractive alternate material for retrofit of civil
infrastructures due to their outstanding strength, light weight and versatility. However, the improvement of strength
and stiffness in a host structure can only be guaranteed when a reliable bonding condition between the two materials
is maintained. Infrared thermography (Levar and Hamilton 2003), Electromechanical (E/M) impedance spectrum
(Giurgiutiu et al. 2003), electrochemical impedance spectroscopy methods (Hong and Harichandran 2005) and fiber
optic sensoring (Ansari 2005) have been previously applied to similar debonding problems. The goal of this study is
to ensure the safety and integrity of CFRP strengthened RC structures and to provide early warning of debonding
based on real-time nondestructive evaluation testing (NDT). A new NDT technique is developed by applying the
concept of time reversal acoustics (TRA) (Fink and Prada 2001) to guided wave propagations within
CFRP-strengthened RC beams, and it is based on the premise that “certain types of damage can be instantaneously
detected without prior baseline data.” This new concept will be extended to develop a NDT system that can be
“rapidly” deployed to laboratory specimens or in-field structures and “autonomously” perform local damage
diagnoses in the presence of operational and environmental variations that in-service structures encounter.

2. WAVE PROPAGATION AND TIME REVERSAL PROCESS IN CFRP-RC BEAMS

Elastic waves in solid media can be classified into body and guided waves, which are governed by Navier’s wave
equations (Rose 1999). The guided waves can be further divided into Lamb, Stoneley and Rayleigh waves depending
on specific boundary conditions. The various frequency components of Lamb waves travel at different speeds and

339
attenuate at different rates due to dispersion characteristics. In spite of these unique characteristics, Lamb waves are
widely used for defect detection, because they are well guided within two closely spaced boundaries and can travel a
long distance with little attenuation. As the thickness of the plate increases, the fundamental symmetric (S0) and
anti-symmetric (A0) Lamb modes converge to a Rayleigh wave, higher modes merge to a transverse bulk wave, and
additional body waves appear. Waves propagation are further complicated in a thin layered thick medium. A
CFRP-RC beam is a good example of such a structure with two distinctive thicknesses. Therefore, a conventional
Lamb wave approach may not be applicable, and a new approach, which can be used regardless of the complexity of
waves, is necessary. To address this issue, a NDT technique based on the TRA concept is proposed. According to
TRA, an input signal can be reconstructed at an excitation point (PZT A) if an output signal recorded at another point
(PZT B) is reemitted to the original source point (PZT A) after being reversed and scaled in the time domain as
shown in Figure 1. If there are certain types of defect along the wave propagation path, time reversibility breaks
down and the shape of the restored signal will depart from that of the input signal. By examining the deviation of the
restored signal from the known input signal, as shown in Figure 2, certain types of damage can be identified without
requiring any previously obtained baseline signals. In reality, the time reversibility does not work well in guided
waves due to their multimode and dispersion. To address this issue, a combination of a specific narrowband input
waveform and multi-resolution signal processing is employed so that the time reversibility is preserved within an
acceptable tolerance for more complex configurations presented in this study (Park et al. 2004).

Figure 1: Schematic concept of TRA-based damage identification Figure 2: Definition of tl, t0, and tr
Once the time reversibility of guided waves is achieved, damage classification is based on two damage indices; time
reversibility (TR) and symmetry (SYM) indices defined in Equation (1). The TR index compares the waveform of
the original input with that of the reconstructed signal, and the SYM index measures the degree of symmetry of the
reconstructed signal with respect to the main peak in the middle:
2 2
⎧⎪tr ⎫⎪ ⎧⎪tr tr
⎫⎪ ⎧⎪ tr ⎫⎪ ⎧⎪t0 tr
⎫⎪
TR = 1 − ⎨ ∫ I (t )V (t ) dt ⎬ ⎨ ∫ I (t ) dt ∫ V (t ) dt ⎬ , SYM = 1 −
2 2
⎨ ∫ L(−t ) R (t )dt ⎬ ⎨ ∫ L(t ) dt ∫ R (t ) dt ⎬
2 2 (1)
⎪⎩ tl ⎪⎭ ⎪⎩ tl tl ⎪⎭ ⎪⎩t0 ⎪⎭ ⎪⎩ tl t0 ⎪⎭
where I(t), V(t), L(t) and R(t) denote the known input signal, the main peak, left-hand and right-hand sides of the
reconstructed signal, respectively. For the experimental study presented, a 7-peak toneburst signal is used for
excitation; tl, tr and to represent the starting, ending and center time points of the toneburst signal as defined in Figure
2. The value of the TR and SYM indices become zero when the shape of the main peak in the reconstructed signal is
identical to that of the original input signal and symmetric with respect to to.

3. EXPERIMENTAL SETUP AND TEST CASES

The overall configuration and dimension of the test specimens are shown in Figure 3. Four (1 to 4) and eight (5 to
12) strain gauges located on the reinforcing steel and the CFRP strip, respectively as indicated in Figure 3(a). These
instruments are used to establish strains and the presence of debonding at the discrete gauge locations. Details of the
beam test programs may be found in Reeve (2005) and Zorn (2006). In addition, a total of 15 PZT square wafers
(2cm x 2cm x 0.0508cm) were attached on the free surface of the FRP as shown in Figure 3(b). They were used as
both sensors and actuators to form an “active” local sensing system. In this study, one PZT wafer was designated as
an actuator, exerting a predefined waveform into the structure. Then, the adjacent PZTs became sensors to measure
the dynamic strain response signals. This process of guided-wave propagation was repeated for a total of 14 different
path combinations (PZTs #1-#2, PZTs #2-#3, … , and PZTs #14-#15). Data from the active sensing devices were
collected at several steps, during which the load was held constant. Two loading cases were investigated in this study
(Cases I and II). In Case I, one of the two full-scale specimens was subjected to monotonic loading, and the data
were collected at every loading step. The monotonic load was gradually increased until the specimen fully failed. A
force-control loading was initially used up to the 5th (40.03kN) loading step and then switched to a
displacement-control from the 6th (41.59kN) to the 24th (44.54kN). The other specimen was subjected to fatigue

340
cyclic loading. In Case II, cyclic loads with a driving frequency of 1.3Hz and the load range of 4.45kN to 22.24kN
were applied. The specimen experienced a total of 2,000,000 fatigue cycles over a time period of 16 days. Data were
gathered at 24 intermittent steps of loading cycles. During the data collection, the cyclic load was held at 4.45kN.

(a) Side view (upside down) (b) Bottom view

Figure 3: Specimens with attached active sensing devices (all units are in mm)

4. EXPERIMENTAL RESULTS
The monotonic and fatigue load tests (Case I and II) were performed on two beams to introduce debonding between
the CFRP layers and the RC beams, and data were periodically collected.
Step 1:4.18kN Step 2:13.09kN Step 6:41.59kN Step 1:4.18kN Step 2:13.09kN Step 6:41.59kN
Step 7:47.15kN Step 15:52.22kN Step 24:44.54kN Step 7:47.15kN Step 15:52.22kN Step 24:44.54kN
1.0 1.0

0.8 0.8
SYM index

SYM index
0.6 0.6

0.4 0.4

0.2 0.2

0.0 0.0
1 2 3 4 5 6 7 8 9 10 11 12 13 14 1 2 3 4 5 6 7 8 9 10 11 12 13 14
Sensing zone Sensing zone

(a) TR indices at 6 loading steps (b) SYM indices at 6 loading steps

Figure 4: TR and SYM indices along the RC beam measured at selective loading steps (Case I)
Step 1: 4.181kN Step 1: 4.181kN

Step 7: 45.365kN Step 7: 45.365kN

Step 24: 45.750kN Step 24: 45.750kN

(a) Forwarding (b) Reconstructed (c) Forwarding (d) Reconstructed

Figure 5: Forwarding and reconstructed signals at sensing zones 10 and 14 (Case I)
The TR and SYM indices at the sensing zones 10 and 14 were examined for the monotonic test (Case I). The values
of indices at zone 10 became larger than 0.2 from the 7th loading step as shown in Figure 4. In Figure 4, it is
speculated that the initial increase of the indices values at the midspan is not relevant to debonding. A 4.8 mm
diameter steel rod embedded in the concrete used to connect the midspan displacement transducer may have affected
the indices values. In this study, these TR and SYM outlier values not related to debonding were disregarded. Figure
5 shows the forwarding and reconstructed signals obtained from the sensing zones 10 and 14 at selected loading
steps. As the load level increased, the amplitude of the forwarding signal gradually decreased as shown in Figure
5(a). The deviation of the reconstructed signal from the original input signal was observed for the increased load
levels, indicating the initiation of debonding near sensing zone 10 in Figure 5(b). On the other hand, no sign of
debonding was found near sensing zone 14. This is consistent with the fact that the reconstructed signal in Figure
5(d) did not change much throughout the entire loading steps. However, it should be noted that the forwarding signal
continuously changed as loading progressed in Figure 5(c). Overall, the findings from the proposed diagnosis system
agree well with the visual inspection and a coin-tapping test results performed after the monotonic loading. The

341
second specimen was subjected to a fatigue cyclic load (Case II). As shown in Figure 6, there was no sign of CFRP
strip debonding during the test nor was any evidence of debonding found from the visual inspection or the coin
tapping test after the test. The TR and SYM indices in Figure 6 mostly remained below 0.2 except sensing zone 7.
1.0 1.0
0 cycles 0 cycles
0.8 0.8
439880 cycles 439880 cycles

SYM index
TR index
0.6 1030700 cycles 0.6 1030700 cycles

0.4
1564430 cycles 1564430 cycles
0.4
2000000 cycles 2000000 cycles
0.2 0.2

0.0 0.0
1 2 3 4 5 6 7 8 9 10 11 12 13 14 1 2 3 4 5 6 7 8 9 10 11 12 13 14
Sensing zone Sensing zone
(a) TR indices at 5 loading cycles (b) SYM indices at 5 loading cycles
Figure 6: TR and SYM indices along the RC beam measured at selective loading cycles (Case II)

5. CONCLUSION
A new real-time monitoring system is developed for detecting debonding of a CFRP layer from a host RC beam. The
uniqueness of this approach lies in the fact that debonding can be detected without using previously obtained
baseline data. Surface mounted PZT wafers are used to generate and measure complex guide waves within the
CFRP-RC coupled structures. Then, two damage sensitive indices are extracted from the measured guided wave
signals based on the concept of the TRA. The potential of the proposed method is demonstrated using experimental
data obtained from monotonic and fatigue loading tests of the CFRP-RC coupled structures Using the proposed
monitoring system, the location and area of the debonding were successfully identified for the monotonic.

ACKNOWLEDGEMENT
This research was partially supported by an NSF Grant No. CMS-0529208 and Pennsylvania Infrastructure
Technology Alliance (PITA) program. The beam test was conducted by Andrew Zorn and Benjamin Reeve in the
Watkins-Haggart Structural Engineering Laboratory at the University of Pittsburgh.

REFERENCES
Ansari, F. (2005). “Fiber optic health monitoring of civil structures using long gage and acoustic sensors.” Smart
Mater. and Struct., Vol. 14, No. 3, S1-S7.
Ekenel, M., Stephen, V., Myers, J. J., and Zoughi, R. (2004). “Microwave NDE of RC Beams Strengthened with
CFRP Laminates Containing Surface Defects and Tested Under Cyclic Loading.” Proc., 16th World Conference on
Nondestructive Testing, Montreal, Canada. August 30-September 3.
Fink, M. (1999). “Time-Reversed Acoustics.” Scientific American, Vol. 281, No. 5, 91-97.
Fink, M., and Prada, C. (2001). “Acoustic Time-Reversal Mirrors.” Inverse Problems, 17, R1-R38.
Giurgiutiu, V., Harries, K.A., Petrou, M.F., Bost, J., and Quattlebaum, J. (2003). “Disbond Detection with
Piezoelectric Wafer Active Sensors in RC Structures Strengthened with FRP Composite Overlays.” Earthquake Eng.
and Eng. Vibration, Vol. 2, No. 2, 213-224.
Hong, S., and Harichandran, R. (2005). “Sensors to Monitor CFRP/Concrete Bond in Beams Using Electrochemical
Impedance Spectroscopy.” J. of Composite for Construction, Vol. 9, No. 6, 515-523.
Levar, J. and Hamilton, H. (2003). “Nondestructive Evaluation of Carbon Fiber-Reinforced Polymer-Concrete Bond
Using Infrared Thermography.” ACI Mater. J., Vol. 100, No. 1, 63-72.
Park, H.W., Sohn, H., Law, K.H., Farrar, C.R., (2004). “Time Reversal Active Sensing for Health Monitoring of a
Composite Plate.” submitted for Journal of Sound and Vibration.
Reeve, B. (2005). “Effect of Adhesive Stiffness and CFRP Geometry on the Behavior of Externally Bonded CFRP
Retrofit Measures Subject to Monotonic Loads.” MS Thesis, University of Pittsburgh Department of Civil and
Environmental Engineering.
Rose, L. J. (1999). Ultrasonic Waves in Solid Media, Cambridge University Press, New York.
Zorn, A. (2006). “Effect of Adhesive Stiffness and CFRP Geometry on the Behavior of Externally Bonded CFRP
Retrofit Measures Subject to Fatigue Loads.” MS Thesis, University of Pittsburgh Department of Civil and
Environmental Engineering.

342
Part XII. Hybrid Systems
Third International Conference on FRP Composites in Civil Engineering (CICE 2006)
December 13-15 2006, Miami, Florida, USA

EXPERIMENTAL STUDY ON FRP-CONCRETE HYBRID BEAMS

Tianhong Li
(Graduate, Department of Civil Engineering, Tsinghua University, Beijing, China)

Peng Feng
(Lecturer, Department of Civil Engineering, Tsinghua University, Beijing, China)

Lieping Ye
(Professor, Department of Civil Engineering, Tsinghua University, Beijing, China)

ABSTRACT
An innovated form of hybrid deck system is proposed, which is composed of FRP (fiber reinforced polymer) I-
shaped profiles, E-shaped profiles and concrete overlay. The E-shaped profiles act as permanent formworks，which
make the construction more convenient. Three FRP-concrete hybrid beams and one pure FRP beam were tested. The
performance influenced by the depth of concrete and CFRP (carbon fiber reinforced polymer) sheets reinforcement
at bottom is studied in comparing. In the tests, the shear cracking failure at the web of the I-shaped profiles is the
main failure mode, which was investigated.

KEYWORDS
Permanent framework, corrosion resistance, light-weight structure, shear failure, FRP-concrete structure.

1. INTRODUCTION
The structure deterioration caused by steel corrosion is a considerable problem in civil engineering. The use of FRP
(fiber reinforced polymer) is a good approach to solve this problem for its excellent chemical and environmental
endurance. Combining FRP and concrete in a reasonable way will benefit the advantages of both the materials and
save the cost. So FRP-concrte hybrid structure as shown in Figure 1 is propsoed for bridges and floors of buildings.

Figure 1: FRP-concrete hybrid structure with permanent formworks

FRP-concrete hybrid beams with different FRP profiles have been studied from 1990s. Deskovic and Triantafillou
(1995), Canning et al. (1999) and Hulatt et al. (2004) studied the different GFRP (glass fiber-reinforced polymer)
box sections with CFRP(carbon fiber-reinforced polymer) laminate on the tensioned side. Branco et al. (2003) and
Nordin and Taljsten(2004) proposed and tested the pultruded GFRP I-profile beams composed with concrete. Fam
and Skutezky(2006) proposed rectangular pultruded GFRP tubes filled with concrete combining with concrete slabs.
In this paper, the typical element in the proposed hybrid structure system, the FRP-concrete hybrid beam as shown
in Figure 1, is investigated. The FRP permanent formworks and FRP girders are included in this system, which is
the difference with the aboves. The mechanical concept of this FRP-concrete hybrid beam is similar to that of the
steel-concrete composite beams: the bending moment is provided by the concrete under pressure and the FRP profile
in tension, and the connectors tranfer the shear force between them. So the compressive strength of concrete and the
tensile strength of FRP can be utilized. Its advantages as following: (1) light-weight, which makes transportation and
installation more convenient; (2) no needs to remove forms, which may save manual labor and material resources

343
and accelerate the construction; (3) FRP permanent formworks can act as the reinforcement in transverse; (4) the
corrosion resistance of FRP can protect the concrete inside.

In this paper, the mechanical performance of this FRP-concrete hybrid beam is investigated by the tests of three
hybrid beams. The stiffness, capacity and failure modes of FRP-concrete hybrid beams are analyzed.

2. TESTS
Four specimens were fabricated and tested, including one pure I-shaped GFRP beam and three hybrid beams. The
hybrid beams were composed of I-shaped GFRP beams and E-shaped plates, connected with the combination of
steel bolts and epoxy adhesive, and then concrete was cast on the top, as shown in Figure 2. Each specimen had 36
bolts, which are designed to supply full shear transfer without yield. The specimens and the materials’ properties are
listed in Table 1: H0 is a pure I-shaped GFRP beam, HB is the standard specimen, HB-T has thicker concrete slab,
and HB-R has three extra CFRP layers of 0.167mm thick each under the bottom flange. All specimens were simply
supported and loaded in four-point bending as shown in Figure 3. The deflections and the slipages were measured
with extension indicators, and the strains of FRP and concrete were measured with guages.

Figure 2: Manufacture of FRP-concrete hybrid beams

Figure 3: Test setup of simply supported FRP-concrete hybrid beams

Table 1: List of Specimens and Materials Properties

Concrete GFRP CFRP

hc
Specimens Strength Strength Modulus Strength Modulus
/mm
/MPa /MPa /GPa /MPa /GPa
H0 0 - 650 22 - -
HB 60 34.0 650 22 - -
HB-T 110 29.6 650 22 - -
HB-R 60 28.7 650 22 3500 235

H0 was not expected to be loaded to collapse. The load was applied to 26kN when the midspan deflection reached
16mm, then unload and reload till the midspan deflection was 21mm, 1/100 span when the load was applied to
around 34kN. H0 stayed elastic in the whole progress and the deflections of both the two sides were symmetric. No
collapse and no cracking sounud occurred during the loading, and the deflection disappeared immediately with
unloading. For HB, cracking sounds were first heard at the load of 38kN, and the beam suddenly collapsed at
99.2kN with an obvious horizontal crack on the web of FRP beam, while there were no cracks on concrete and no

344
other failure either. HB-R had the very similar behavior with HB: cracks occurred at 54kN and failed with a horzital
crack on the web under the load of 94.6kN. HB-T had a different phenomenon that some cracks appeared in the
lower part of the concrete panel at the midspan at 24kN and in the gaps of the adjoining FRP plates at 34kN. Clacks
began to be heard at 50kN till the beam collapsed suddenly at 149kN with the same failure mode. Although all
hybrid beams have the same failure mode, the locations of the crack in the webs depend on the thickness of concrete
panels in the way that the crack goes close to the concrete panel with the increase of the concrete thickness. The
load-deflection curves and the failure modes of the specimens are shown in Figure 4.

160
HB-T
140
120
HB-R
100
Load/kN

80 HB
clacks
60
40
20 H0 (no failure)
0
0 5 10 15 20
Deflection/mm

Figure 4: Load-deflection curves and failure modes of test specimens

3. TEST RESULTS AND ANALYSIS

The load-deflection curves clearly show that the bending stiffness of hybrid beams is considerably higher than the
pure FRP beam, which can reach up to 4 to 6 times. Among the three hybrid beams, the bending stiffness of HB-T is
the highest, which indicates that the increase of the concrete’s thickness is effective to enhance the bending stiffness
of hybrid beams. Comparing HB and HB-R it can be seen that the CFRP layers in the lower flange have little use in
developing the bending stiffness. Strains on the top of concrete and the bottom of FRP beam at the midspan are
illustrated in Figure 5. All strains increase almost linearly with the load. The strains of HB-T are smaller than those
of the other two at the same load, wihch indicates that HB-T has the highest rigidity; HB-R and HB have close strain
values which means that their rigidity values are almost the same.

Concrete FRP 210

160
Distance to the bottom/mm

HB-T HB-T 180

140
120 150
HB HB-R
100 120
Load/kN

HB-R
80
HB 90
60
60 99.2kN
40
20 30
20kN 60kN
0 0
-3000 -1000 1000 3000 5000 7000 -2000 -1000 0 1000 2000 3000 4000 5000
Strain /με Strain/με

Figure 5: Strains on the top and the bottom of beams Figure 6: HB strains at midspan

Figure 6 shows the strains of different locations on the midspan cross section of HB. It can be seen that the plane
assumption is well satisfied, and so do the other specimens. The slippage between the ribbed plates and the I-shaped
beams are measured in the tests, the maximum slippage of HB was 0.15mm, 0.10mm for HB-R and 0.80mm for
HB-T. All these indicate that the combination connection gives an excellent shear force transfer.

In the existing experimental researches, the FRP-concrete hybrid beams loaded in bending failed in several possible
modes, such as (1) flexural failure by concrete crush (Hulatt et al., 2004; Nordin and Taljsten, 2004); (2) connectors
or adjacent concrete failure (Nordin and Taljsten, 2004); (3) concrete panel shear failure (Kavlicoglu, 2001); (4)
FRP web shear fracture (Branco et al., 2004) . All three hybrid beams in this paper failed in the fourth mode which
is resulted by the low shear strength of the pultruded FRP profiles’ webs. The shear failure happened too premature

345
to utilize the tension strength of FRP and the compressive strength of concrete. So it is important to improve the
shear strength of the FRP web to enhance the loading capacity of the whole specimen. A short pure I-shaped GFRP
beam has been tested to achieve the shear strength of GFRP profile, which was 41.7kN. Since the net section at the
seam of two adjacent ribbed plates of the hybrid beams is composed of I-shaped GFRP and a strip of concrete with
the thickness of hc' as shown in Figure 7, it is presented that the increase of the shear capacity of the hybrid beams
compared with the pure GFRP beams is provided by the concrete above the ribs. According to the test results, Figure
8 was achieved which shows that the shear strength of hybrid beams corresponds with hc' linearly.

80

70

Shear strength / kN
60

50

40

30
0 20 40 60 80
Effective depth of concrete : h c'/ mm

Figure 7: Effective section to bear shear force Figure 8: Relation of hc’ and shear strength

4. CONCLUSIONS
An innovative FRP-concrete hybrid deck system with permanent formworks is presented and tested. The following
conclusions can be drawn from this study: (1) the bending stiffness of the hybrid beam is related to the concrete
slab’s thickness, while the CFRP layers bonded on the bottom flange of FRP have little contribution to stiffness; (2)
the shear failure of FRP web is a considerable failure mode as it makes it impossible to utilize the strength of FRP
and concrete, in which the shear strength increases with concrete slab’s thickness.

ACKNOWLEDGEMENTS

The authors are grateful to the Natural Science Foundation of China for their support to the research presented here
through a national key project on the application of FRP composites in civil engineering in China (Project No.
50238030).

REFERENCES
Branco, F.A., Ferreira, J., and Correia, J.R. (2003). “The use of GRC and GFRP-concrete beams in bridge decks”.
Proceedings of FRP Composites in Bridge Design and Engineering, COBRAE Conference, Porto, Portugal
Hulatt, J.A., Hollaway, L.C., and Thorne, A.A. (2004). “A novel advanced polymer composite/concrete structural
element”. Proceeding of the Institution of Civil Engineers Structures & Building, Vol.157, pp 9-17.
Kavlicoglu, B.M., Gordaninejad, F., Saiidi, M., and Jiang, Y. (2001). “Analysis and testing of graphite/epoxy
concrete bridge girders under static loading”. Proceedings of 9th International Conference on Structural Faults and
Repairs, London, United Kingdom.
Nordin, H., and Taljsten, B. (2004). “Testing of hybrid FRP composite beams in bending”. Composites Part B:
engineering,Vol.35, pp27-33.
Canning, L., Hollaway, L., and Thorne, A.M.(1999). “An investigation of the composite action of an FRP/concrete
prismatic beam”. Construction and Building Materials, Vol. 13, pp417-426.
Deskovic, N., and Triantafillou, T.C.(1995). “Innovative design of FRP combined with concrete: short-term
behavior”. Journal of Structural Engineering, Vol.121, No.7, pp1069-1078.
Fam, A. and Skutezky, T.(2006). “Composite T-beams using reduced-scale rectangular FRP tubes and concrete
slabs”. Journal of Composites For Construction, Vol.10, No.2, pp172-181.

346
Third International Conference on FRP Composites in Civil Engineering (CICE 2006)
December 13-15 2006, Miami, Florida, USA

HYBRID BONDING OF FRP LAMINATES TO REINFORCED

CONCRETE STRUCTURES
Yu-Fei Wu
Assistant Professor, City University of Hong Kong, Hong Kong

Yue Huang
Research student, City University of Hong Kong, Hong Kong

ABSTRACT
This paper reports a new FRP bonding technology that combines the adhesive bonding technology with a new type
of mechanical fastening. The new mechanical fastening system does not rely on the bearing to transmit the
interfacial shear. Instead, it increases the frictional bond resistance by resisting the separation of the FRP laminate
from the concrete substrate. Experimental testing has demonstrated that this new technology can provide a bonding
strength that is several times that of the conventional adhesively bonded FRP system.

KEYWORDS
FRP, reinforced concrete, bond, strengthening, retrofitting, testing

1. INTRODUCTION
External bonding of fiber reinforced polymer (FRP) has emerged as one of the most popular methods for structural
rehabilitation in recent years (Nanni, 1997; Hollaway and Head, 2001). Up to date, three typical methods of bonding
FRP onto concrete structures have been developed. The first and most common method is by adhesively bonding
FRP to the external surface of reinforced concrete (RC) members, a method known as externally-bonded FRP or
EB-FRP. The second technology is the near-surface mounting (NSM) where the FRP reinforcement is embedded in
the cover layer of the concrete structures. The third recent development is the mechanically-fastened FRP (MF-FRP)
technique (Bank, 2004), where a special type of FRP strip that has a significant bearing strength is mounted onto the
concrete surface by mechanical fasteners. Apart from the above three method, it has been found that anchoring FRP
strips at their ends with large anchors, namely end anchorage, can effectively increase both the bond strength and the
ductility. Wrapping a longitudinal FRP strip with transverse U shaped FRP strips (U-jackets) has also been found to
be effective in delaying the premature debonding. Other methods such as fiber anchors (Lam and Teng, 2001) and
interlocking-anchorage with epoxy keys (Grace, 2001) have been investigated to improve the bond capacity as well.

Despite the extensive efforts around the world in recent years in studying the debonding mechanism and in
improving the bond strength, the bond remains the weakest link that fundamentally limits the efficacy of the FRP
applications. Therefore more effort in finding alternative means of bonding FRP is needed.

2. HYBRID BONDING TECHNOLOGY AND ITS MECHANISM

Careful observations of the debonding process in an EB-FRP beam have revealed that the debonding causes not only
a longitudinal slip between the FRP reinforcement and the concrete substrate but also a vertical separation between
them. It was this separation that caught the attention of the first author who proposed and developed the new
bonding technique by suppressing the separation with a special type of mechanical fastener that is shown in Fig. 1a.
This mechanical fastener is composed of a thin steel capping plate that serves to apply normal pressure on the FRP
laminate to restrain the vertical separation, and two anchors (concrete nails) that are used to fasten the steel plate
onto the concrete substrate.

347
 Mechanical FRP strip
fasterners

(a) Mechanical fastener (b) Test specimen (bottom view) after installation of the HB-FRP system
Figure 1: HB-FRP System

This new technique combines the EB-FRP system with the MF-FRP system, and therefore, is named the hybrid
bonded FRP (HB-FRP) system. However, the mechanical fasteners in the HB-FRP system work with a very
different mechanism from that in the MF-FRP system. No bearing resistance of the FRP is required. Therefore, it is
applicable to any commercially available FRP laminates. The application of the HB-FRP system consists of two
steps. The first step involves the same procedure as that for an EB-FRP system, by adhesively attaching the FRP
laminate onto the surface of the concrete. After the hardening of the adhesive for the EB-FRP, another coat of epoxy
resin is applied on top of the FRP strip and the special mechanical fasteners are then installed along the longitudinal
direction of the FRP reinforcement at a specified spacing, as shown in Fig. 1b. The installation of each mechanical
fastener involves the drilling of two small holes in the concrete and driving the concrete nails into the two pre-drilled
holes through the steel capping plate with a hammer, so that the capping plate firmly covers the FRP strip.

Due to the roughness of the debonding surface, any longitudinal slip of the FRP reinforcement against the concrete
surface will result in a vertical movement of the FRP strip, as shown in Fig. 2a. This perpendicular movement of the
FRP against the fasteners that are firmly embedded into the concrete produces significant vertical pressure on the
FRP strip. This vertical pressure in turn causes frictional shear resistance at the interface. Therefore at debonding
failure, the fasteners and the concrete around the fasteners will be pushed out by the movement of the vertical
separation. This is evident in Fig. 2b where the nail was pushed out vertically from the concrete and a piece of
concrete was pulled out by the nail at bond failure. Consequently, the mechanism of the HB-FRP system is
completely different from that of the mechanically fastened system where the interfacial shear comes from the
bearing of the attached plate/strip on the mechanical fasteners. As the nails did not deform laterally (see Fig. 2b)
after debonding failure in the HB-FRP system, it was evident that no significant bearing occurred in the HB-FRP
system. Therefore very thin anchors such as normal concrete nails are sufficient in the HB-FRP system.

Concrete
substrate

FRP longitudinal slip

FRP perpendicular movement
Pushing against capping plate
and pulling out of concrete

(a) Frictional shear due to slip (b) Local debonding failure

Figure 2: Mechanism of the HB-FRP system

3. EVALUATION OF THE HB-FRP SYSTEM

Experimental testing has been performed by the 2nd author to evaluate the HB-FRP system. Tests involved 4 one-
way slab members, one strengthened with EB-FRP and the other three by HB-FRP with different FRP thickness.

348
3.1 Test Specimens

The design details of the RC slab are given in Fig. 3. The specimen with the HB-FRP system is shown in Fig. 4a and
the details of the mechanical fastener are shown in Fig. 4b. The specimen with the EB-FRP was identical to that in
Fig. 4a except that no mechanical fastener was provided. Properties of the test specimens are given in Table 1. Each
ply of FRP had a nominal fiber sheet thickness of 0.165 mm.

2R10 2R10
150

30
250
13R10 – 200
2500 300

Figure 3: Details of RC Specimen

30 3 mm thick
Steel Plate steel plate
Anchorage Strain Gauge FRP Strip
7

60 φ 4.1
70
50 c/c 37
50 100
Concrete Slab 6

1800 φ 4.1 Dimensions in (mm)

(a) Details of the specimen with HB-FRP (b) Details of the mechanical fastener

Figure 4: The HB-FRP system for test specimens

Table 1: Material Properties of the Test Specimens

Concrete Steel Bars FRP Strip
Specimen Strengthening Cube Yield Elastic Tensile Elastic Ultimate
Method Strength Strength Modulus Strength Modulus Strain
(MPa) (MPa) (GPa) (MPa) (GPa) (%)
S1 EB-FRP (2 plies) 80 343 208
S2 HB-FRP (2 plies) 81.2 349 206
4519 257 1.76
S3 HB-FRP (4 plies) 82 332 208
S4 HB-FRP (6 plies) 82 346 208

3.2 Test Results and Discussion

The test setup and the test results are given in Fig. 5. The lowest curve in Fig. 5b gives the response of the member
strengthened by the conventional EB-FRP technique, with 2 plies of CFRP. The other three response curves are for
members strengthened with the HB-FRP system, with 2, 4 and 6 plies of CFRP, respectively.

The EB-FRP system increased the flexural strength from about 8 kN (calculated) for the un-strengthened member to
17 kN. The failure was by a debonding and complete detachment of the FRP strip from the bottom of the member,
which indicated that the bond strength was unable to fully mobilize the tensile rupture strength of the 2-ply CFRP
strip. Both the members with the HB-FRP of 2 and 4 plies of CFRP failed by the rupture of the CFRP strip, as
shown in Fig. 5c. This clearly indicated that the bond strength with the HB-FRP system was greater than that was
needed to fully utilize the material tensile strength of the 4-ply CFRP strip and hence resulted in the rupture of the
CFRP strip. The member with a 6-ply CFRP strip failed due to debonding of the strip as shown in Fig. 5d. In this

349
case, the bond strength of the HB-FRP system was exhausted before the tensile strength of the 6-ply CFRP strip was
reached.

From the test result of the conventional EB-FRP strengthened member, it can be seen that the strength increase due
to the EB-FRP system was about 9 kN (from 8 to 17 kN). In other words, the EB-FRP system contributed 9 kN of
the ultimate load. The ultimate load of the member with the 6-ply HB-FRP system was 70 kN. Taking away the 8
kN contributed by the steel bars, the ultimate load due to the HB-FRP system was 62 kN. This is about seven times
(62/9) of that contributed by the conventional EB-FRP system. The effectiveness of the HB-FRP system is clearly
demonstrated by these test results. Preliminary finite element analyses show that the EB-FRP has only one bond
stress block that moves towards the plate end as debonding propagates. However, the HB-FRP system produces
many bond stress blocks (one under each mechanical fastener) that increase the total bond resistance many times.
Detailed results of the finite element analyses are not included in this paper due to the page length limit. Effort is
made at CityU currently to simply the mechanical fastener into a staple, and the mechanical fastening process can be
as simple as operating a stapling-machine that is similar to a Powder Actuated Fastening “gun”.
80
HB-FRP (6 plies)
70
60

Applied load (kN)

50
HB-FRP
40 (4 plies)
Loading at Mid-span 30
HB-FRP
20
Concrete Slab EB-FRP (2 plies)
10
(2 plies)
0
LVDT
0 10 20 30 40 50
Mid-span displacem ent (m m )
2000

(a) Test setup (b) Response curves

(c) FRP rupture failure mode (d) Debonding failure mode

Figure 5: Flexural testing and results

ACKNOWLEDGEMENT
The work described in this paper was fully supported by a grant from the Research Grant Council of the Hong Kong
Special Administrative Region, China [Project No. CityU 122106].

4. REFERENCES
Bank, LC. Mechanically fastened FRP (MF-FRP) - a viable alternative for strengthening RC members. Keynote
Lectures, Proc. the Second International Conference on FRP Composites in Civil Engineering (CICE 2004), 8-10
December 2004, Adelaide, Australia.
Grace, N.F. Improved anchoring system for CFRP strips, Concrete International ACI, 2001; 23(10): 55-60.
Hollaway, L.C., and Head, P. Advanced Polymer Composites and Polymers in the Civil Infrastructures. Elsevier
Science, London, UK. 2001.
Lam, L., and Teng, J.G. Strengthening of RC cantilever slabs bonded with GFRP strips, Journal of Composite for
Construction ASCE, 2001; 5(4): 221-227.
Nanni, A. Carbon FRP strengthening: New Technology becomes mainstream, Concrete International, 1997; 19(6): 19-23.

350
Third International Conference on FRP Composites in Civil Engineering (CICE 2006)
December 13-15 2006, Miami, Florida, USA

HYBRID FRP-CONCRETE SANDWICH BRIDGE DECK

Thomas Keller, Erika Schaumann, Till Vallée

(Ecole Polytechnique Fédérale de Lausanne EPFL-CCLab,
CH-1015 Lausanne, Switzerland)

ABSTRACT
This paper presents a new concept for a lightweight hybrid-FRP bridge deck. The sandwich construction consists of
three layers: a fiber-reinforced polymer composite (FRP) sheet with T-upstands for the tensile skin, lightweight
concrete (LC) for the core and a thin layer of ultra high performance reinforced concrete (UHPFRC) as a
compression skin. Mechanical tests on eight hybrid beams were performed with two types of LC and two types of
FRP/LC interface: unbonded (only mechanical interlocking of LC between T-upstands) and bonded with an epoxy
adhesive. The ultimate loads of the beams increased by 104% on average due to bonding. However, the beam failure
mode changed from ductile to brittle. The beams using a LC of 44% higher density exhibited an 81% increase in the
ultimate load. The manufacturing of the beams proved to be economic in that epoxy and concrete layers were
rapidly and easily applied wet-in-wet without intermediate curing times. The experimental results demonstrated the
feasibility of the suggested hybrid bridge deck.

KEYWORDS
Bridge deck, sandwich, hybrid, FRP, lightweight concrete.

1. INTRODUCTION
Fiber-reinforced polymer (FRP) composites have found increased applications in bridge structures in recent years.
Applications for strengthening and repair are already well established. A growing number of new bridges have been
constructed as all-FRP or hybrid-FRP structures (FRP combined with traditional materials) (Keller, 2006). Several
research projects pointed out that the combination of FRP with conventional materials is a promising application of
FRP composites in bridge engineering (Deskovic et al. 1995, Hall and Mottram 1998, Canning et al. 1999, Van Erp
et al. 2002, Bank et al. 2006).
The hybrid sandwich bridge deck concept suggested in this paper consists of three layers of different materials: FRP
composites for the tension skin, lightweight concrete (LC) as a core material and ultra high performance fiber
reinforced concrete (UHPFRC) for the compression skin (Fig. 1). The FRP layer, consisting of a 5 mm thick GFRP
sheet with T-upstands, also serves as formwork. The fibers in the 30-50 mm thick UHPFRC layer are needed to
carry possible local bending moments due to concentrated wheel loads and due to the jointless application
(described below). A ductile or at least pseudo-ductile system behavior should be achieved by failure in the top
concrete layer and crushing of the lightweight aggregates.
The target total weight of the deck is less than 50% of a normal concrete deck. The FRP layer together with the LC
core is easily prefabricated in large elements (approximately 2.50 m x width of the bridge) that are transported to the
site and rapidly installed on the main girders. The joints between deck elements as well as between deck and main
girders are adhesively bonded. Subsequently, the thin UHPFRC layer is jointlessly cast on site onto the LC core. In
regions of negative bending moments, GFRP reinforcement grids are laid into the UHPFRC layer. Since the deck is
steel free and the UHPFRC layer is watertight, no waterproofing layer is required and the surfacing is directly
applied onto the UHPFRC.
To examine the feasibility of the proposed concept, flexural experiments on hybrid beams were performed. The
benefit of a bi-dimensional plate behavior was therefore neglected. The objective of the experiments was to show if
a pure mechanical connection between the FRP and LC through the T-upstands is sufficient to provide full
composite action or if an additional layer of adhesive is required. Furthermore, the shear capacity of different
densities of LC was explored, because no reliable shear properties were available for the LC concretes. To this end,
the beams were designed to obtain a shear failure in the lightweight concrete.

351
2. BEAM DESCRIPTION
The experimental program consisted of eight beams 3600 mm long, 400 mm wide and 200 mm deep (span/15).
Figure 1 shows the cross-section of the beams. For the FRP layer, the standard pultruded GFRP element Plank
40HDx500 from Fiberline was used, which was cut to a width of 400mm. Only a 30 mm normal concrete (NC) layer
was applied on the top. Lightweight concretes with average densities of 900 and 1300 kg/m3 were used.
Furthermore, two types of FRP/LC interfaces were investigated: unbonded (LC directly cast on the FRP) and
adhesively bonded (LC cast on wet epoxy adhesive applied between and on T-upstands). Table 1 gives an overview
of the four beam configurations and their labeling. For each configuration two beams were examined.

Table 1: Experimental Results at Ultimate Load

Beam Type of LC FRP-LC Ultimate Load Mid-span

interface [kN] deflection [mm]
900-1/2 LC900 unbonded 9.3 / 11.3 11.5-17.3 / 9.4
900E-1/2 LC900 epoxy bonded 31.2 / 30.4 8.7 / 8.1
1300-1/2 LC1300 unbonded 32.1 / 23.3 47.8 / 40.4
1300E-1/2 LC1300 epoxy bonded 42.8 / 50.5 12.3 / 14.1

NC

30
130

200
LC
50
5

40
FRP 400

Figure 1: Cross-Section of Hybrid Beams

3. EXPERIMENTAL SET-UP
All beams were simply supported on rollers with a span length of 3000 mm and subjected to three-point bending
using a hydraulic jack with a capacity of 200 kN. The load was applied at a constant displacement rate of 1 mm/min.
The load was measured with load cells between the jack and the steel plate. For all beams, seven linear voltage
displacement transducers (LVDT) were used to measure the deflection of the beams along the span in the centerline
and the slippage of the LC and NC from the FRP element at the beam ends. Up to 20 strain gages on the top and at
the bottom of the FRP sheet and on top of the T-upstands, and up to 30 Omega gages on the NC and LC concretes
were applied to measure the strain distributions through the cross-sections.

4. EXPERIMENTAL RESULTS
4.1 Load-Deflection Response and Failure Mode of Unbonded Beams

The load-deflection curves at mid-span, measured for the unbonded beams with low density LC (beams 900-1, 900-
2) and high density LC (1300-1, 1300-2), are shown in Fig. 2. The four beams showed an almost linear-elastic
response up to 6.8-11 kN when small vertical cracks first appeared in the tension zone of the LC below the loading
plate. The cracks always extended through the LC aggregates. Then the response of all four beams changed.
After first cracking, the load of the beams with high density could be increased significantly up to a first peak (at
31.0 kN for beam 1300-1 and 22.8 kN for 1300-2). The LC started to debond from the FRP sheet (at 20.7 kN for
1300-1 and 17 kN for 1300-2) and the stiffness decreased slightly. Increased slippage at the beam end was measured
from this load on. After the first load peak, an oscillating load phase was observed for both beams as the
displacement was increased, whereby at each intermediate peak a loud crack was audible. After the oscillating
phase, a slight yet steady load increase was observed up to ultimate load. In this last loading phase, a single crack
developed slowly through the whole depth of the beam and then horizontally in the LC/NC interface to the loading

352
plate and on the height of the T-up stands to the support. The LC was further pushed out at one beam end. Upon the
sudden drop in load, the crack, which always went through the LC aggregates, reached the support.
The load of the beams with low density LC could only be slightly increased up to the ultimate load after the first
cracks were observed (from 6.8 to 9.3 kN for 900-1 and 9.2 to 11.3 kN for 900-2). At ultimate load, debonding of
the LC from the FRP sheets started and slippage between LC and FRP could be measured at the beam end. The
maximum load of beam 900-1 could be maintained up to a mid-span deflection of approximately 17 mm, after
which the load started to decrease slightly up to 34 mm deflection. After this point a sharp drop in the load was
observed. The load of beam 900-2 started to decrease slightly immediately after ultimate load up to a mid-span
displacement of 32 mm, when also a sharp drop occurred. In this phase, the same failure mechanism developed as
described for the unbonded beams with higher density, however, at significantly smaller deflections and lower loads.
The ultimate loads of the four unbonded beams and corresponding mid-span deflections are listed in Table 1.

55

50 1300E-2

45
1300E-1
40

35
1300-1
30
900E-1
Load [kN]

25
900E-2
20 1300-2

15

10 900-1

5 900-2

0
0 5 10 15 20 25 30 35 40 45 50
Deflection at midspan [mm]

Figure 2: Load-Deflection Response of Unbonded Figure 3: Failure Pattern of Bonded Beam 1300E-2
and Bonded Beams

4.2 Load-Deflection Response and Failure Mode of Bonded Beams

The load-deflection curves at mid-span, measured for the bonded beams with low density LC (900E-1, 900E-2) and
high density LC (1300E-1, 1300E-2), are shown in Fig. 2. The responses of all four beams showed a slight yet
steady decrease in stiffness up to ultimate failure, which occurred at 31.2/30.4 kN for beams 900E-1/2 and 42.8/50.5
kN for beams 1300E-1/2. First cracks in the tension zone of the LC were observed between 20 and 26 kN for all
beams. Subsequently, a multitude of small vertical or slightly inclined cracks developed in the tension zone below
the loading plate. At ultimate failure, one of the cracks suddenly grew through the depth of the beam, along the
LC/NC interface to the loading plate in one direction and at the height of the FRP T-upstands, just above the
LC/FRP interface, to the support and on to the end of the beam in the other direction, as shown in Fig. 3. The epoxy-
bonded interfaces remained undamaged. The beams showed no debonding of the LC from the FRP and no slippage
at the beam end was measured. After brittle failure, the load dropped and the experiments were stopped. The
ultimate loads of the four bonded beams and corresponding mid-span deflections are also listed in Table 1.

5. DISCUSSION
5.1 Influence of Interface Bonding on Flexural Behavior
In the first loading phase, up to 7-11 kN, all eight beams showed almost the same stiffness. At LC concrete cracking,
however, the first differences in the behavior of unbonded and bonded beams were seen. While in the LC of the
unbonded beams only a few, wide cracks developed with subsequent debonding of the LC from the FRP in the
interface, a multitude of small cracks developed in the bonded beams, without debonding of the LC from the FRP.
Accordingly, the stiffness of the unbonded beams decreased much more than that of the bonded beams.
After debonding at the LC/FRP interface in the unbonded beams, the composite action between concrete and FRP
was partially lost and the FRP sheet participated increasingly in the load transfer with increasing load. Due to this,

353
the deflections increased considerably up to ultimate failure. The bonded beams, in contrast, showed full composite
action up to ultimate failure, which occurred at significantly higher loads and smaller deflections.
The basic failure mode of both beam types was similar, with the exception of two differences. For both beam types,
a shear failure in the LC occurred in one beam part. The failure thereby always crossed the LC aggregates. In the
unbonded beams, this failure developed slowly and stopped at the support, while the debonded LC was pushed out
at the beam end. In the bonded beams, the failure was sudden and brittle and the crack propagated above the T-
upstands (in the LC) over the supports up to the beam end. In contrast to the brittle failure of the bonded beams, the
unbonded beams showed a very ductile behavior. The oscillating response was interpreted as a slow, progressive
failure of the mechanical interlocking between LC and T-upstands, showing a high dissipation of inelastic energy.

5.2 Influence of LC Density on Flexural Behavior

The effect of low versus high density LC on the ultimate load of the unbonded beams was seen to be greater than on
the bonded beams (an increase of 169% vs. an increase of 51%, on average). The higher density LC enabled an
effective mechanical interlocking between the LC and FRP T-upstands, which could not be developed with the
lower density LC.

6. CONCLUSIONS
Experiments on eight hybrid FRP-concrete beams provided useful information about their load carrying behavior
and the influence of the two investigated parameters: the FRP/LC interface (unbonded or epoxy bonded) and the
type of lightweight concrete (low and high density). The following conclusions were drawn:
1) Varying the FRP/LC interface from unbonded to adhesively bonded increased the ultimate load by 104% on
average, but changed the failure mode from ductile to brittle.
2) Increasing the LC density by 44% increased the ultimate load by 81% on average due to improved mechanical
interlocking between LC and T-upstands. The effect was much more pronounced for the unbonded beams (169%
increase) than for the bonded beams (51% increase).
3) The manufacturing of the beams proved to be rapid and easy. Epoxy, LC and NC were applied wet-in-wet
without intermediate curing times within less than 30 minutes per beam. From this point of view the fabrication was
very economical.
4) The experimental investigation proved the feasibility of the proposed new hybrid bridge deck. Further
optimization is possible regarding LC material properties, mechanical FRP/LC interlocking, and local
reinforcements over the supports to enable an arch-tie mechanism after concrete cracking.

7. ACKNOWLEDGEMENTS
This research was funded within the project New Road Construction Concept (NR2C) of the 6th European
Framework Program. The authors wish to acknowledge the support of Fiberline Composites A/S, Denmark, supplier
of the GFRP Plank elements; Sika AG, Zurich, Switzerland, supplier of the epoxy adhesive.

8. REFERENCES
Canning, L., Hollaway, L., and Thorne, A. (1999). “An investigation of the composite action of an FRP concrete
prismatic beam”. Construction and Building Materials, Vol. 13, No. 8, pp. 417-426.
Deskovic, N., Meier, U., and Triantafillou, T.C (1995). “Innovative Design of FRP Combined with Concrete: short
term behavior”. Journal of Structural Engineering, Vol. 121, No. 7, pp. 1069-1078.
Van Erp, G.M., Heldt, T., McCormick, L., Carter, D., and Tranberg C (2002). “An Australian approach to fiber
composite bridges”. Proceedings of the International Composites Conference ACUN4, Composite Systems: Macro
Composites, Micro Composites, Nano Composites, UNSW Sydney, pp. 145-153.
Hall, J.E., and Mottram, J.T. (1998). “Combined FRP reinforcement and permanent formwork for concrete
members”. Journal of Composites for Constructions, Vol. 2, No. 2, pp. 78-86.
Keller, T. (2006). “Emerging markets – FRP composites in bridge superstructures”. 8th World Pultrusion
Conference, Emerging Markets: Globalization of the Pultrusion Industry, Budapest, Hungary, (2006).
L.C. Bank, L.C., Oliva, M.G., Russell J.S., Jacobson D.A., Conachen M., Nelson B, and McMonigal, D. (2006),
“Double Layer Prefabricated FRP Grids for Rapid Bridge Deck Construction: Case Study”. Journal of Composites
for Construction, Vol. 10, No. 3, pp. 204-121.

354
Third International Conference on FRP Composites in Civil Engineering (CICE 2006)
December 13-15 2006, Miami, Florida, USA

HYBRID SQUARE FRP/STEEL GRID TUBE CONFINED

CONCRETE CYLINDERS

Manu John
Graduate Student, Department of Mechanical Engineering, Louisiana State University, Baton Rouge, LA 70803

Guoqiang Li
Assistant Professor, Department of Mechanical Engineering, Louisiana State University, Baton Rouge, LA 70803

ABSTRACT
In this study, a lattice of steel grid tube, which is externally protected by a FRP skin, is used to confine square
concrete cylinders. Twenty seven scaled-down square cylinders, which had a height of 308mm, a length of 114.3mm
and a width of 114.3mm, were prepared. In order to evaluate the effect of grids on the structural performance of
these encased square cylinders, two types of grid tubes were used. Type 1 was prepared with ribs running in the
axial direction and Type 2 with ribs in a helical pattern. The specimens were divided into 9 groups. Groups 1 and 5
had grids on one face; groups 2 and 6 had grids on two opposite faces; groups 3 and 7 had grids on three faces and
groups 4 and 8 had grids on all the four faces. Among them, Groups 1-4 were made of Type 1 grids and groups 5-8
were made of Type 2 grids. Solid steel tubes, Group 9, which were also wrapped with the same FRP skin, were used
to encase concrete cylinders as controls. Uniaxial compression tests were conducted on all the specimens. The effect
of the steel grids on the structural behavior was evaluated based on the test results.

KEYWORDS
FRP, grids, concrete confinement, square cylinders.

1. INTRODUCTION

Over the years, steel tube and fiber reinforced polymer (FRP) tube-encased concrete beam-columns have emerged as
a novel alternative for rebuilding and new construction of engineering structures (Mirmiran and Shahawy, 1997).
This includes manufacturing plants, bridges, high-rise buildings, harbors, water-front fenders, etc. These FRP tube-
encased concrete beam-columns possess higher strength and ductility, and they are formwork-free.

In tube encased concrete cylinders, the tube is subjected to a 2-D stress condition - hoop tension and axial
compression. This shows a distinct departure from FRP repaired concrete cylinders, where only the hoop tension
dominates. Because of the 2-D stress condition, the interfacial bonding strength plays a certain role in carrying the
applied axial and transverse loads. Therefore, there is a need to increase the interfacial bonding strength. Recently, a
new type of circular steel grid tube that is externally protected by a FRP skin has been developed and tested; the
function of the steel grid tube and the FRP skin has been well defined and validated through observation (Li, 2006).
A uniaxial compressive load has been used. It is found that, due to the mechanical interlocking developed between
the grid tube and the concrete core, both the specific compressive strength and ductility have been increased as
compared to solid steel tube confined counterparts. Square steel grid tube encased concrete cylinders were also
developed and tested using a four-point bending load configuration. Again, it is found that both the specific bending
strength and ductility have been increased (Li et al, 2006). It is well known that stress concentrations exist at the
corners in square tube encased concrete cylinders. It is desired to find how the square grid tube behaves if it is
subjected to a uniaxial compression. The purpose of this study is thus to investigate the structural behavior of square
steel grid tube encased concrete cylinders when they are subjected to a uniaxial compressive load.

355
2. EXPERIMENTS
2.1 Raw Materials and Specimens Preparation

Type I Portland cement, gravel, natural sand, water, and a superplasticizer DAVA 170 were used to prepare the
concrete. The mix design followed ACI Standard 211.1 (“Standard” 1991). The mix ratio by weight was cement:
water: coarse aggregate: fine aggregate: admixture = 1: 0.56: 3.80: 2.19: 0.001. To fabricate steel grid tubes, a low
carbon steel plate, which had a thickness of 6.35mm, yield strength of 308MPa, and a modulus of elasticity of
200GPa, was obtained. The plate was cut into smaller pieces of 308 mm long by 114.3mm wide steel sheets.
Circular holes with a diameter of 25.4mm were drilled through the steel sheets. Once the steel sheets were drilled,
they were welded using seam welding to form 308mm high, 114.3mm long, and 114.3 mm wide square cylinders. A
steel grid structure is shown in Figure 1.

In order to evaluate the effect of grids on the structural performance of these encased square cylinders, two types of
tubes were prepared. Type 1 was prepared with ribs running in the axial direction and Type 2 with ribs in a helical
pattern. The encased specimens were divided into 9 groups. Groups 1 and 5 had grids on one face; groups 2 and 6
had grids on two opposite faces; groups 3 and 7 had grids on three faces and groups 4 and 8 had grids on all the four
faces. Among them, Groups 1-4 were made of Type 1 grids and groups 5-8 were made of Type 2 grids. Solid steel
tubes, Group 9, which were also wrapped with the same FRP skin, were used to encase concrete cylinders as
controls. Each group of specimens was weighed individually in order to obtain the specific load-displacement
curves. Using a balance, it was found that the weight of a group 1 tube was 5.62 kg; it was 4.96 kg for a group 2
tube, 4.25 kg for a group 3 tube, 3.55 kg for a group 4 tube. For the helical grid tubes the weights were as follows:
5.69 kg for group 5, 5.01 kg for group 6, 4.37 kg for group 7 and 3.66 kg for group 8. The weight for the solid steel
tube was 6.38 kg (group 9).

Once the steel tubes were prepared, all of them were wrapped using an ultraviolet (UV) curing E-glass 7715 fabric
reinforced vinyl ester composite to form the FRP skin. The hand lay-up technology was used. The 7715 style fiber
reinforcement was a unidirectional fabric. In this study, the fiber was aligned along the transverse (hoop) direction to
provide the maximum confinement. Each tube was wrapped using two layers of FRP with a 25.4mm overlap. After
wrapping, the specimens were moved to a UV-A light source for curing. The details of the UV-A light source can be
found elsewhere (Li et.al, 2005). The curing was completed within one hour. Before casting concrete, one end of the
tube was capped using a plastic tape. The concrete was then mixed, cast, compacted, finished, and cured for 28 days
in a standard wet curing room with 100% relative humidity.

2.2 Compressive Testing

The compression tests were conducted according to ASTM C 39 standard using a FORNEY machine. This machine
has a capacity of 2688kN. Each specimen was uniaxially compressed to about 40% of the unconfined concrete
strength, and unloaded to guarantee close contact between each component and to reduce errors in displacement
measurement. Then, the specimen was reloaded until cylinder failure. The assembled computer data acquisition
system can directly record the load–displacement curves. The loading rate was 0.23MPa/s.

3. RESULTS AND DISCUSSIONS

3.1 Axial Load-Displacement Behavior

The specific load (kN/kg) – displacement (mm) curves for each subgroup of specimens are shown in Figure 2. It is
seen that the specific load-displacement of grid tube encased concrete cylinders behave similar to the solid steel tube
encased concrete counterparts. The load-displacement curves can be generally described by a linear elastic region
followed by an almost horizontal curve, which simulates yielding, and a strain hardening region until the peak load
is achieved. Once the peak load is reached, the cylinders show a rapid decrease in load as the displacement
increases. This is an indication of the structural failure of the cylinders. The reason for this type of structural
behavior can be explained as follows. Subjected to a uniaxial compression load, a tensile stress is developed on each
face of the square tube along the hoop direction. This type of stress is the highest at the four corners of the tube. The
stress concentration at the corners results in yielding and hardening of the welds, similar to a low carbon steel
subjected to tension.

356
 Figure 1: Steel Grid Structure

Figure 2: Specific Compressive Load and Displacement of

various types of cylinders

3.2 Compressive Strength and Ductility

The specific peak load and the displacement are summarized in Table 1. Comparing the steel grid tube with the solid
steel tube, it is seen that the specific axial load of the grid tube is reduced. The reduction becomes larger and larger
as the number of grid faces increases, in particular for the axial grid tubes. The reason for this reduction may be due
to the fact that the tube and the core carry the load collaboratively. Due to the removal of steel, the stiffness of grid
tubes is lower than that of the solid steel tubes. Subject to the same axial load, the concrete core has a higher share of
load. As a result, the core expands more laterally. This larger transverse expansion results in larger hoop stress in the
tube. Consequently the cylinders with grid tubes fail at a lower axial load.

Table 1: Variation of Specific Load and Displacement with Number of Ribs

Peak Specific Load Load decrease Displacement Ductility change
Specimen
(kN/kg) (%) (mm) (%)
Group 9 237.97 - 27.37 -
Group 1 215.76 9.30 23.27 -14.90
Group 2 207.25 12.91 31.82 16.29
Group 3 204.17 14.20 26.61 -2.77
Group 4 174.68 26.60 41.74 52.50
Group 5 198.44 16.61 32.65 19.29
Group 6 202.09 15.07 23.67 -13.52
Group 7 196.13 17.58 22.83 -16.59
Group 8 184.02 22.67 29.82 8.95

Compared with the solid steel tubes, the grid tube encased concrete cylinders show an increase in ductility. This
increase becomes more pronounced for the group 4 samples. The reason for this increase in ductility can be
explained as follows. For the solid steel tubes and the solid face in steel grid tubes, it can be noticed that local

357
buckling occurs; see Figure 3. As a result failure occurs abruptly and leads to smaller displacement. On the other
hand, for the steel grid faces, no buckling can be seen due to the mechanical interlocking and the support provided
by the concrete within the bay area. The failure of the grid faces is mainly due to axial deformation of the bay.
Consequently the cylinders with grid tubes possess more ductility.

(a) Solid steel tube locally buckles (b) Grid face deforms and solid face locally buckles

Figure 3: Failure modes of solid steel face and grid face

Comparing the axial grid tube encased cylinders with helical grid tube encased cylinders, it is found that the axial
grid tube leads to a higher reduction in strength and a higher gain in ductility. Depending on the applications, axial
grid tube confined cylinders may be more applicable in the case where more ductility is required; on the other hand,
helical grid tube encased cylinders may be more proper for the cases where compressive strength is the control
factor. It is observed during the testing that the local buckling of the solid steel face or the axial deformation of the
grid face is close to the end of the cylinders. For helical steel grid face there is more solid steel at the end of the face
than that of the axial grid face. The more solid steel near the end of the face results in higher strength and lower
ductility, similar to the solid steel face.

4. CONCLUSIONS
Steel grid tube encased square concrete cylinders are prepared and tested in this study. The test results show that the
grid tube encased concrete cylinders behave similarly to the solid steel tube encased control concrete cylinders.
Compared to the control specimens, the grid tube encased cylinders have a higher ductility, due to the mechanical
interlocking, and a lower strength, due to the reduced stiffness of the steel grid faces. Compared with axial grid
tubes the helical grid tubes have a higher compressive strength and a lower ductility, possibly due to a higher
amount of solid steel near the end of the helical grid face.

5. ACKNOWLEDGMENT
This study was partially sponsored by the Louisiana Board of Regents and EDO Fiber Science under contract
number LEQSF(2004-07)-RD-B-05 and an EDA Fellowship. The encased concrete cylinders were prepared and the
tests were conducted in the Concrete Laboratory at the Louisiana Transportation Research Center (LTRC). The
support is greatly appreciated.

6. REFERENCES
Li, G., Torres, S., Alaywan, W., and Abadie, C. (2005). “Experimental Study of FRP Tube-Encased Concrete
Columns”. Journal of Composite Materials, Vol. 39, pp. 1131-1145.
Li, G. (2006). “Experimental Study of Hybrid Composite Cylinders”. Composite Structures, (in press)
doi:10.1016/j.compstruct.2005.08.028.
Li, G., John, M. and Maricherla, D. (2006). “Experimental Study of Hybrid Composite Beams”. Construction and
Building Materials, (in press) doi:10.1016/j.conbuildmat.2005.11.003.
Mirmiran, A and Shahawy, M (1997). “Behavior of concrete columns confined by fiber composites”, Journal of
Structure Engineering, Vol 123, pp 583-590.

358
Third International Conference on FRP Composites in Civil Engineering (CICE 2006)
December 13-15 2006, Miami, Florida, USA

STRUCTURAL CHARACTERIZATION OF COMPOSITE SANDWICH

PANEL WITH HYBRID FRP-STEEL CORE FOR BRIDGE DECKS
Hyo Seon Ji
(Visiting Scholar, The University of Tennessee, Knoxville, Tennessee, USA)

Byung Jik Son

(Assistant Professor, Konyang University, Nonsan-Si, Chungnam-Do, Korea)

Z.John Ma
(Associate Professor, The University of Tennessee, Knoxville, Tennessee, USA)

ABSTRACT
The use of FRP panels for bridge deck is limited, due to the relatively higher cost of the composite materials. A
hybrid concept of composite sandwich panel with hybrid FRP-steel core is proposed for highway bridges in order
not only to improve stiffness and buckling response but also to be cost-efficiency compared to all FRP decks. The
composite sandwich bridge deck system is composed of the hybrid FRP-steel core and the FRP facings. Its structural
performance under static loading was conducted, and the ANSYS finite element program was used to predict the
panel behaviour. This study is focused on the structural characterization of flexure performance of the composite
sandwich panels for bridge decks under static loading. It is found that the presented composite sandwich panel with
hybrid FRP-steel core is very efficient for use in bridges. And roughly estimated cost of this composite sandwich
deck is $250/ m2, whereas roughly estimated that of an all FRP deck is $400/m2. Therefore, the use of the hybrid
composite sandwich deck in bridge construction may be able to compete with the conventional decks.

KEYWORDS
Composite sandwich panel, hybrid FRP-steel core, Cost-efficiency, Finite Element

1. INTRODUCTION
In recent years the interest in using Fibre-reinforced polymer (FRP) in construction field application has
significantly increased worldwide. Due to its high specific stiffness and strength, FRP may offer a number of
advantages over traditional materials, including environmental durability and ease of construction. The
competitiveness of FRP in bridge structures likely falls in the potential for lower life-cycle and installation cost. The
first of all FRP short-span bridge in South Korea was installed in May 2002 (Ji et al., 2006). Although many
different types of deck systems are currently designed and constructed, the most commonly used type of deck for the
slab-on-girder system is a cast-in-place reinforced concrete deck. Recently, compared to conventional materials the
application of FRP can significantly improve the performance of bridge deck. However, the use of FRP panels for
bridge deck is limited, due to the relatively higher cost of the composite materials. The costs of the bridge decks
with different materials are compared such as an estimated cost of $400/m2 for all Glass Fibre-Reinforced Polymer
(GFRP) deck, and $200/m2 for all Concrete deck in US dollar. In terms of initial costs, FRP composites are,
however, still too expensive to compete with other conventional materials used in civil engineering applications. To
make the best use of materials, combinations of FRP and conventional materials have recently been investigated by
a number of researchers. The advantages of the combined or hybrid structural systems include the cost-effectiveness
and the ability to optimize the structural section based on material properties of each constituent material. The
objective of this paper is to present the proposed hybrid concept of composite sandwich panel system for bridge
decks. The panel system is composed of the hybrid FRP grid core and steel box tube core. This study is focused on
the flexure performance of the proposed system under static loading through an experimental program. Its structural
performance under the DB-24 live load (MOCT, 2000) is also analysed by using the ANSYS finite element program.

359
2. BEHAVIOR AND DESIGN OF HYBRID FRP-STEEL DECK PANELS
The proposed hybrid FRP-steel sandwich panel for highway bridge deck consists of FRP composite top and bottom
facings, FRP composite grid and multiple steel box cells core, as shown in Figure1.

25 6.5 UFL Grid

4
25
3
4.5
CFL + Bonding layer
Steel Tube 116

Bonding layer + LFL

4.5
3
4@125 = 500

Figure 1: Cross sections of composite sandwich panel (All Dimensions in mm)

In order to consider local effect due to tire contact at the top facing, the upper core was designed as a composite grid.
The constituents properties used for the sandwich panel are shown in Table 1.

Table 1: Fiber and matrix constituent material properties

Material E ( GPa ) G ( GPa ) ν ρ ( g / cm3 )

E-glass fiber 72.5 27.6 0.22 2.54
Polyester resin 3.38 1.38 0.38 1.24
Vinyl ester resin 3.91 1.43 0.37 1.15

By using a simple rule of mixtures approach, the layer stiffness and properties of the composites can be determined
as shown in Table 2.

Table 2: Layer stiffness and properties predicted using micro mechanics model

Ply thickness Plys per Ply E1 E2 G12

Ply name Ply type Vf ν 12
( mm ) layer orientation ( GPa ) ( GPa ) ( GPa )
CSM 3.2 4 0.194 Random 7.99 7.99 1.69 0.349
UFL
◦ ◦
Roving 2.4 2 0.234 0 /90 19.55 4.35 1.77 0.343
LFL CSM 3.2 4 0.194 Random 7.99 7.99 1.69 0.349
◦ ◦
CFL Roving 2.4 2 0.234 0 /90 19.55 4.35 1.77 0.349
◦
Grid core Roving 25.0 25 0.246 0 20.38 4.42 1.80 0.341
Steel core - 125.0 - - Isotropic 204 204 76.9 0.300
Bonding layer CSM 1.6 2 0.183 Random 8.17 8.17 1.73 0.343

Where UFL = upper facing laminate; CFL= core facing laminate; LFL = lower facing laminate; CSM =
chopped strand mat; Roving = woven roving; V f = fiber volume fraction. In general, the polyester resin is used.
However, the bonding layers consist of the vinyl ester resin due to the presence of the elongation difference between
the steel tube and facing laminate. The stiffness of each ply can be predicted from micromechanics models (Julio et
al., 2001) shown in Table 2. To predict the deflection, a preliminary analysis was conducted using the theory of
anisotropy sandwich beam. Then, the detailed finite element analysis was used. The composite sandwich panel was
modeled and analyzed by the finite element analysis program ANSYS. GFRP laminates were modeled by 8–node

360
shell elements, while the FRP grid and steel tube were modeled by solid elements. Perfect bonding between FRP
grid and steel tube was assumed in all analyses. The boundary conditions were modeled as a pin and a roller for the
end supports. Pressure load was applied in the middle of the deck. The FE analysis was conducted for a composite
sandwich panel using the material properties in Table 2. The composite sandwich panel deck was fabricated by a
hand lay-up method according to the detailed design.

3. EXPERIMENTAL STUDY
The testing setup for the deck is shown in Figure 2. The bending test was conducted in order to evaluate the
structural performance of the composite sandwich panel of 1.0m × 3.0m × 0.163m . Two composite sandwich deck
specimens were tested. The load was applied at center of the deck onto the tire contact area of 500mm × 200mm by
the 500kN-MTS actuator. Figure 3 shows the location of LVDTs (D1, D2, and D4) and strain gages (TR1, T1, TR4
= top strain; BR1, B1, BR4 = bottom strain).

D1 TR1 BR1

T 1 B1

D2 D 4 TR 4 BR 4

2.7m

Figure 2: Experimental setup Figure 3: The location of LVDTs and strain gauges

Figure 4: Load-deflection, strains response of composite sandwich panel

The load-deflection curves from testing and analysis and strains curves of the composite sandwich deck are shown
in Figure 4. The stiffness was calculated to be 16.7kN/mm. The comparison of deflection results under a service
load for the two specimens is shown in Table 3. The one rear axle load of DB24 (MOCT, 2000) is 94.08KN. The
deflection for this service load, 94.08kN was 6.00mm (mean value). This is less than the span design limitation of
L/425 (LRFD, 2004) of 6.35mm for the composite sandwich deck.

Table 3: Comparison of results for a composite sandwich deck under P=94.08 kN

Theory of anisotropy beam ANSYS Exp. Remark

Deflections 5.65 6.00
4.67 4.35 Displacement Limit (L/425; 6.35mm)
(mm) 6.35 (Mean value)

361
The maximum load applied was 491KN and lateral deflection was 35.6mm (average). The predicted failure load
was under 500kN. However, no visible damage was observed during the testing up to 491KN.
The comparison of displacement and strain results under the testing load of 400kN is shown in Table 4.

Table 4: Comparison of displacement and strain under P=400kN

Displacement (D4) Strain (T1) Strain (B1)

25.56mm 25.66mm -0.00329 -0.00333 0.00298 0.00296
Experiment
25.77mm (Mean value) -0.00337 (Mean value) 0.00294 (Mean value)
Analysis (ANSYS) 18.11mm -0.00288 0.00208

As the failure mode, the significant de-lamination between the FRP grid and steel tube was observed, but there
was no collapse of the deck. It is possible to confirm no collapse because of the flexibility failure of the steel tube by
comparing the all GFRP deck. On visual inspection, any local damage failure at the loaded part of the top flange was
not observed. The construction costs for traditional small highway bridges are estimated as shown in Table 5.

Table 5 Cost of the bridge decks

Description of cost Cost for a bridge deck with different materials (US $)
categories Concrete All GFRP Proposed Hybrid
Total direct costs $200/m2 $400/m2 $250/m2
Material cost 120 300 180
Handling 12 63 39
Shipping 4 1 1
Installation 24 7 4
Corporate overhead 40 29 26

Although roughly estimated that of the hybrid FRP are 1.25 higher than those for a concrete bridge deck, these
increases are offset by savings on other costs such as maintenance and repair costs. Therefore, the hybrid FRP decks
may be able to compete with the concrete ones.

4. CONCLUSIONS
The static testing and analysis of a hybrid FRP-steel composite sandwich panel for highway bridges are briefly
discussed in this paper. Based on the study presented in this paper, several conclusions are summarized in the
following. It is possible to confirm the efficient use of steel tube core in the core of sandwich panel to increase the
system stiffness instead of using GFRP, and to reduce the amount of GFRP composites and initial costs. It was
demonstrated that the proposed composite sandwich deck is considered to possess safety factors for strength and
serviceability much higher than the Korean Highway Code requirements. The depth of deck may be decreased 20%
when comparing with the all GFRP deck so that the same flexural rigidity can be obtained, and the strength can be
used more efficiently as well. Roughly estimated cost of this composite sandwich deck is $250/ m2 comparing with
an estimated cost of $400/m2 for all GFRP deck. Therefore, the use of the hybrid FRP-steel composite sandwich
deck in bridge construction may be very efficient and be able to compete with the conventional and all FRP decks.

5. REFERENCES
AASHTO LRFD (2004), “AASHTO LRFD Bridge Design Specifications”, American Associations of State
Highway and Transportation Officials, 3rd Ed., Washington D.C.
Ministry of Construction and Transportation (MOCT) (2000), “Standards specifications for highway bridge ” 2nd
Ed., Korea.
ANSYS. Version. 10.0.
Julio F. Davalos et al. (2001). “Modeling and characterization of fiber-reinforced plastic honeycomb sandwich
panels for highway bridge applications”. Composite Structures, Vol. 52, pp. 441-452.
Ji, H.S., Son, B.J., and Chang, S.Y. (2006) “Field Testing and Capacity-Ratings of Advanced Composite Materials
Short-span Bridge Superstructure, Composite Structures, in press

362
Part XIII. Masonry Structures
Third International Conference on FRP Composites in Civil Engineering (CICE 2006)
December 13-15 2006, Miami, Florida, USA

CHARACTERIZATION OF MASONRY WALLETTES AND

SHEAR TRIPLET SPECIMENS RETROFITTED WITH
GFRP COMPOISTES
Prakash, S.S
(Graduate Research Assistant, University of Missouri Rolla, Rolla, Missouri, USA)

Alagusundaramoorthy, P
(Associate Professor, Indian Insitute of Technology Madras,, Chennai, Tamil Nadu,India)

ABSTRACT
Masonry is one of the oldest construction materials that is in use around the world for reasons due to its
accessibility, functionality and cost. Masonry buildings are the major building stock around the world, fail in
earthquakes mainly because of their low tensile and shear resistance. These masonry structures are retrofitted with
FRP composites is found to increase its seismic performance. Experimental studies on masonry wallettes and shear
triplet specimens with and without retrofitting using GFRP composites were conducted under monotonic loading.
The increase in deformation limit and ultimate load resistance of the retrofitted wallettes and triplets were compared
to that of control specimens. Finite element (FE) models were created, and the results were verified with
experimental data. Macromodeling was adopted for the analysis of masonry wallettes. Shear triplets were analyzed
by micromodeling. Parametric studies were conducted to study the changes in the deformation limit and ultimate
load resistance for varying thickness of GFRP composites and fiber orientations. It is concluded that GFRP
composites retrofitting increases the strength and stiffness of wallettes and strength and ductility of shear triplets.

KEYWORDS
Masonry; Wallettes; Shear triplets; FRP composites; Retrofitting

1. INTRODUCTION
This study focuses on the characterization of the masonry wallettes, shear triplets, and GFRP composite materials to
study the effect of retrofitting on the strength and deformational capacity of brick masonry. The behavior of
unreinforced masonry is brittle, with little or no ductility, and it suffers form damages ranging from invisible
cracking to extensive crushing and that leads to brittle/sudden failure during earthquakes. Earthquakes like Bhuj in
India in 2001 have shown that the damage to lives and property are mainly by the failure of unreinforced brick
masonry walls. To understand the behavior of retrofitted load bearing and infill walls, the basic material properties
of masonry components are essential. In this study, the strength and elastic properties of bricks, cement mortar, brick
masonry wallettes and shear triplets, and GFRP composite materials are evaluated. The effect of GFRP retrofitting
on the behavior of masonry wallettes and shear triplets are studied by testing the specimens and finite element
analysis. Based on the results of finite element analysis and experimental study, a set of conclusions are drawn.

2. EXPERIMENTAL STUDY
Bricks and mortar were tested to determine the elastic properties as per IS: 1905, 1987. The modulus of elasticity of
the brick and mortar are found to be 808 N/mm2 and 1540 N/mm2 respectively. A composite laminate was fabricated
using the epoxy resin with hardener in 10:1 ratio and glass fiber woven roving mat with a resin to fiber ratio of 1:1.
Tests were conducted on the GFRP composite to find water absorption, hardness, volume fraction of fibers, tensile
strength, and flexural strength as per BS 2782, 1996.

2.1 Tests on Masonry Wallettes

Fifteen wallettes were cast with and without retrofitting using GFRP composites (Table 1). Wallettes were cured for
fifteen days and tested in the compression testing machine both normal and parallel to bed joints (EN 1052, 1998).
The specimens (WNB1 to WNB5) were tested under uniaxial compression normal to bed joint and they failed

363
through vertical cracking in the middle. When the principal tensile stress reaches the tensile strength of the bricks,
crack forms, and failure occurs.. The wallettes (WPB6 to WPB9) under compression parallel to bed joint failed by
the formation of vertical separation of cracks along the bed joint from differential normal deformation of bricks and
the mortar joint. The modulus of elasticity under compression parallel to bed joint is higher than the modulus of
elasticity normal to bed joint for the solid brick masonry wallettes (Table 1). Three GFRP retrofitted specimens
(RWNB1 to RWNB3) were tested under compression normal to bed joints, and the other three GFRP retrofitted
specimens (RWPB4 to RWPB6) were tested under compression parallel to bed joints. Under compression normal to
bed joints, the increase in the strength was around 20%. However, the FRP retrofitting reduced the ultimate strain
significantly (Up to 40%), which is detrimental to the seismic demand (ductility). The behavior of the retrofitted
specimen is stiffer with Eavg= 336 N/mm2, compared to the control specimen with Eavg=168 N/mm2 (Table 1). The
stiffness of the wallettes increased by around 100% by retrofitting. The retrofitted wallettes failed under
compression by crushing due to the confining effect of retrofitted laminates. Eventhough, debonding of the
laminates was noticed at the initial phase of loading, the overall bond with brick surface was intact throughout the
loading. GFRP retrofitting increased the ultimate load under compression parallel to bed joint. The increase in the
load resistance was around 3.5%. However, the ultimate strain under compression parallel to the bed joint does not
reduce due to GFRP composite retrofitting as in the case of compression normal to bed joint. The behavior of the
retrofitted specimen under compression parallel to bed joint is stiffer with Eavg=563 N/mm2 compared to that of the
control specimen Eavg=405 N/mm2. The increase in the stiffness is about 39%. After reaching the peak load, a
bursting sound was heard as a result of separation of the bed joint mortar from the bricks.

Table 1. Ultimate Strength of Control and Retrofitted Wallettes under Compression

Specimen Direction Size Density Ultimate Ultimate Young’s Modulus Axial Secant
of Loading (l x b x t) Strength Strain Stiffness at Failure
(mm) (kN/m3) (N/mm2) (N/mm2) (kN/mm)
Normal to Strain readings were Strain readings were Strain readings were not
WNB1 393 x 410 x 132 19600 3.3
Bed Joint not consistent not consistent consistent
WNB2 ,, 395 x 410 x 130 19400 2.6 ,, ,, ,,
WNB3 ,, 396 x 405 x 122 19300 3.3 0.01900 176 21.9
WNB4 ,, 398 x 408 x 126 19600 3.1 0.01920 163 21.1
WNB5 ,, 390 x 404 x 123 19500 3.0 0.01850 164 20.9
Average 395 x 407 x 126 19520 3.0 0.01890 168 21.3
Parallel to
WPB6 392 x 408 x 123 19900 3.3 0.00775 425 54.4
Bed Joint
WPB7 ,, 396 x 409 x 125 20800 3.2 0.00765 420 54.2
WPB8 ,, 398 x 406 x 127 19600 3.0 0.00770 393 50.9
WPB9 ,, 393 x 408 x 128 21200 2.9 0.00775 381 50.6
Average 396 x 407 x 127 20375 3.1 0.00771 405 52.6
Normal to
RWNB1 395 x 408 x 130 20400 3.6 0.01140 316 42.4
Bed Joint
RWNB2 ,, 396 x 406 x 131 20600 3.9 0.01120 348 46.8
RWNB3 ,, 396 x 405 x 128 20500 4.1 0.01190 345 45.2
Average 396 x 406 x 130 20500 3.9 0.01150 336 44.8
Parallel to
RWPB4 394 x 409 x 130 20300 3.5 0.00580 603 81.4
Bed Joint
RWPB5 ,, 397 x 407 x 129 20800 3.2 0.00570 561 74.2
RWPB6 ,, 396 x 406 x 128 20200 3.0 0.00570 526 69.1
Average 396 x 406 x 129 20433 3.2 0.00570 563 74.9

2.3 Tests on Shear Triplets

Shear triplet specimens without retrofitting (STS1, STS2, and STS3) and with retrofitting (RSTS1, RSTS2, and
RSTS3) were cast for testing. Specimens STS1, STS2, and STS3 were tested to determine the initial shear strength
under zero vertical compression. The shear resistance increased with relatively no deformation up to a load of 10
kN. Thereafter, the load resistance linearly increased from 10 kN to 14 kN up to a deformation of 0.01 mm. After
the deformation of 0.02 mm the crack started propagating suddenly along the bed joint, and the specimen failed.
This result shows that the unretrofitted triplet has failed through slipping mode of brick and bed joint mortar at very
low displacement levels. The average shear strength of triplets was found to be 0.29 N/mm2 (Table 2). The
specimens RSTS1, RSTS2, and RSTS3 were retrofitted with one layer of GFRP composite to study the effect of
retrofitting on increase in strength and ductility. The test results of shear triplets showed that GFRP retrofitting was

364
observed to increased the ultimate load from 13.8 kN to 39 kN. The increase in the load resistance was up to 180%.
The ultimate displacement has increased dramatically from 0.016 mm to 3.0 mm. Eventhough initial debonding
occurred, the final failure of the triplet is by cracking of the middle brick and this shows that the retrofitting by
GFRP composites increases the ductility of the masonry under shear. This failure mode indicates that with good
bonding of GFRP laminates on the brick surface, the shear failure can be totally eliminated, and the increase in
strength will be governed by the compressive strength of masonry.

Table 2. Ultimate Shear Strength of Control and Retrofitted Triplets

Specimen Size (L x B x H) Failure Load Ultimate Shear Maximum Shear

(mm) (kN) Strength (N/mm2) Deformation (mm)
STS1 233 x 110 x 230 13.5 0.28 0.020
STS2 232 x 110 x 230 14.2 0.29 0.020
STS3 233 x 110 x 230 13.9 0.29 0.014
Average 233 x 110 x 230 13.8 0.29 0.018
RSTS1 234 x 112 x 230 37.8 0.78 3.200
RSTS2 232 x 112 x 230 39.3 0.81 2.900
RSTS3 232 x 112 x 230 40.0 0.83 2.800
Average 233 x 112 x 230 39.0 0.81 2.967

3. NUMERICAL STUDY
Homogeneous macromodeling with smeared cracking concrete material model for the analysis of masonry wallettes
and heterogeneous micromodeling using elastic properties for analysis of shear triplets are adopted. Macromodeling
is adopted for the analysis of brick masonry wallettes due to its simplicity and practical significance (Giordano et al
2002). The material properties for macromodeling are obtained from the testing of brick masonry wallettes
(Table 1). The modulus of elasticity for the numerical study is obtained from the experimental results as Ex=168
N/mm2 normal to bed joint and Ey= 405 N/mm2 parallel to bed joint. The Poisson’s ratio for the masonry material is
normally around 0.15 to 0.2, and it is taken as 0.2. For micromodeling, the average modulus of elasticity of the clay
brick masonry unit is found to be 808 N/mm2. The modulus of elasticity of the mortar is found to be 505 N/mm2.
The Poisson’s ratio for the brick and mortar is taken as 0.2 (Sahlin, 1971). The friction on the interface between
brick and mortar normally varies from 0.45 to 0.80 and is taken as 0.6. The normal stiffness for the contact analysis
is given a very high value in order to avoid interpenetration of continuum. FRP material used for retrofitting is a
glass fiber woven roving mat with an area density of 360 gsm. It is modeled assuming elastic behavior through the
lamina option in ABAQUS (Hibbit et al, 2002). The tensile strength of the laminate was determined as per
procedures given in the standard (BS 2782, 1996). The average Young’s modulus in tension along the warp and weft
directions is found to be Ex= 12984 N/mm2 and Ey=11500 N/mm2. The Poisson’s ratios υxy and υyx are found to be
0.126 and 0.112, respectively. Perfect bond is assumed to exist between GFRP composite and the masonry wall
surface.

3.1 Macromodeling of Masonry Wallettes

Masonry wallettes are modeled using macromodeling concepts with the smeared crack concrete model. The average
Young’s modulus and Poisson’s ratio from the corresponding wallette results are used for the elastic zone of the
model. For the plastic zone, the parameters are taken from the nonlinear part of uniaxial compression stress strain
curve (stress and corresponding absolute plastic strain). The “failure surface” option is used to define the shape of
the failure surface. The GFRP composite is modeled as an elastic orthotropic material. In the retrofitted and
unretrofitted specimens, wallettes are modeled by eight noded linear continuum elements (C3D8R) with reduced
integration and hourglass control. The GFRP composite is modeled using 3D shell elements (S4R) with both
rotation and translations at each node. The loading is applied at the top surface of the wallette as displacement. The
bottom of the wallettes is assumed to be fixed with all the degrees of the freedom arrested. It is assumed that a
perfect bond exists between the wallette surface and GFRP composite. The behavior of the FE-model is found to be
in close agreement with the experimental data upto the elastic limit in the unretrofitted control wallettes and GFRP
composite retrofitted wallettes (Figs. 1a and 1b).

3.2 Micromodeling of Shear Triplets

In micromodeling of shear triplets, the brick as well as mortar are taken as linear elastic materials with appropriate
material properties from the experimental data. Both the brick and mortar are modeled using 3D continuum elements
(C3D8R). The GFRP composite is modeled using 3D shell elements (S4R) with orthotropic material properties. The
interface between brick and mortar was modeled using surface to surface contact algorithm with interface elements.
365
The brick surface is considered as the master surface and mortar surface is considered as the slave surface.
Tangential behavior of the interface between the brick and mortar is modeled using the penalty friction formulation
with friction coefficient as 0.6. The normal behavior of the interface is modeled using linear pressure over closure
with normal stiffness as 1000 N/mm. Initially, the increase in load resistance was stiffer before cracking of the bed
joint up to 10 kN in the case of the control triplet. After the cracking at the bed joint, the GFRP composite started to
resist the load. In FE-model, the increase in the load resistance was linear up to the ultimate displacement of 3.0 mm
(Fig. 1c). From the analysis of the FE-model, the load resistance was 70 kN compared to the experimental value of
40 kN (Fig. 1c) at the ultimate displacement of 3.0 mm. This result is due to the perfect bond assumption made in
the modeling between the triplet surface and the GFRP composite surface.

5 5 80
RWNB
70
4 4

L o a d R e s is ta n c e (k N )
FEM
RWPB 60
RWNB Experiment

S tr e s s ( N /m m 2 )
S tres s (N /m m 2 )

3 3 50
40
WNB
2 WNB
2 30
WPB RWPB
20
Experimental 1 Experimental
1 10
FEM FEM
WPB
0
0 0
0.001 0.01 0.1 1 10
0 0.005 0.01 0.015 0.02 0.025 0 0.005 0.01 0.015
Displacement (mm)
Strain Strain
(a) Wallettes-Normal to Bedjoint (b) Wallettes Parallel to Bedjoint (c) Shear Triplets
Fig. 1 Comparison of FEM Results with Experiment Data

4. DISCUSSION OF RESULTS
Both the experimental and FE analysis the results showed that the retrofitting of masonry wallettes and shear triplets
using GFRP composites increased their load resistance. However, the failure strain of wallettes reduced from 0.0189
to 0.0115 under compression normal to bed joint and under compression parallel to bed joint no reduction in the
failure strain is observed. In general, under compression, retrofitting using GFRP composites increased the strength
and stiffness of wallettes but reduced the ductility. In the case of shear triplets, retrofitting increased strength and
ductility. The increase in the ultimate shear deformation was from 0.018 mm to 3.0 mm. This result shows that the
GFRP retrofitting can significantly enhance the seismic performance of the masonry walls under shear, since the
forces generated in the earthquake are predominantly shear. The FE model results showed good correlation with the
experimental results for the unretrofitted and retrofitted specimens.

5. SUMMARY AND CONCLUSIONS

The behavior of masonry wallettes with and without retrofitting using GFRP composites under compression parallel
to bed joint, normal to the bed joint, and their failure modes were explained. The behavior of GFRP composite
retrofitted shear triplets were compared to that of control shear triplets. FE results were compared to experimental
data, and were found to be in good agreement with the test data. The following conclusions were drawn based on the
experimental and analytical studies: (i) GFRP composite retrofitting increases the strength and stiffness but reduces
the ductility in the case of masonry wallettes under compression normal and parallel to bed joint, (ii) GFRP
composite retrofitting increases the strength and ductility in the case of shear triplets, (iii) Smeared crack concrete
model is found to predict the behavior of masonry reasonably well under monotonic loading and can be further
extended to study the crack pattern, and (iv) Micromodeling is feasible for the analysis of small-scale specimens like
triplets, but suitable algorithms have to be developed to improve the convergence in large scale modeling.

6. REFERENCES
1. IS: 1905 (1987). “Indian standard code of practice for structural use of unreinforced masonry,” Bureau of Indian
Standards, New Delhi.
2. EN 1052-1(1998). “Methods of test for masonry - Part 1: Determination of compressive strength”, European
Standards.
3. BS 2782-10: Method 1006 (1996). “Methods of testing plastics, glass reinforced plastics.”
4. Sahlin, S. (1971). Structural Masonry, Prentice Hall, Englewood Cliffs, Newjersey.
5. Hibbit, D., Karlson, B., and Sorensen, P. (2002). ABAQUS/ Standard theory manual.
6. Giordano, A., Mele, E. and De Luca, A. (2002). “Modeling of historical masonry structures: comparison of
different approaches through a case study”, Engineering Structures, 24(2002), 1057-1069.

366
Third International Conference on FRP Composites in Civil Engineering (CICE 2006)
December 13-15 2006, Miami, Florida, USA

FRP COMPOSITE REINFORCEMENTS ON MASONRY VAULTS:

EFFECTIVENESS AND RELIABILITY
Alessandro Baratta
(Full professor, University of Naples Federico II, Naples, Italy)

Ottavia Corbi
(Associate professor, University of Naples Federico II, Naples, Italy)

ABSTRACT
In the paper, for treating the problem of FRP reinforcements in barrel masonry vaults, one first presents a theoretical
approach for the analysis of this vault typology. Under the basic assumptions of purely membrane state of stress and
indefinite length of the vault, the referred model is demonstrated to behave as a series of independent arches; the
case of FRP reinforcement is then considered and theoretical and experimental results are shown to be is good
agreement, also emphasizing the effectiveness of the adopted reinforcement.

KEYWORDS
Masonry vaults, theoretical setup, FRP reinforcements, experimental validation.

1. INTRODUCTION
Advanced technologies and the identification of innovative materials, such as composites, able to respond in a
satisfactory way to needs rarely met by the adoption of traditional materials and methodologies, are acquiring more
and more interest for protection of historical buildings. On the other side, the NRT (not-resisting-tension) model for
structural analysis of masonry structures proves to be an effective tool for analyzing the behavior of original
structures as well as the effectiveness of reinforcements, also with respect to seismic thrust.
In the present paper, this problem is approached by modeling the basic masonry structure by the no-tension pattern,
and the FRP reinforcement, working in uni-axial stress state. The behavior of basic masonry vault models reinforced
with FRP strips, and subject to vertical and horizontal loads, is analyzed and compared to experimental results.

2. THEORETICAL SETUP OF THE PROBLEM OF BARREL MASONRY VAULTS

WITH INDEFINITE LENGTH
As regards to barrel vaults, first of all, one should consider that, since the vault geometrically derives by the
translation along a directrix of a generating arch curve, in this case, the meridian lines coincide with the generatrix in
their shapes; if one considers a rectilinear directrix, the vault parallels are horizontal and rectilinear as well (Figure
1). The surface of the shell representing the mid-surface of the vault may be defined by the equation z = f(x).
Because of the vault geometry, one has that

∂z ∂z
θ = 0, tan ϕ = , tan θ = =0
∂x ∂y
(1)
dx dy dx dy dx
ds x = , ds y = = dy , dA = ds x ds y = = dy
cos ϕ cos θ cos ϕ cos θ cos ϕ

where dsx and dsy denote the length of the sides of the generic vault element ABCD of area dA dx and dy the length
of the corresponding sides on the element A’B’C’D’ projected in the xy-plane, and ϕ and θ denote the angles formed
by the meridian sides AB and DC of the element with the x-axis and by the parallel sides AD and BC with the y-
axis, respectively.

367
 y

Figure 1: Barrel barrel vault with horizontal directrix.

As concerns equilibrium, hypothesizing that the vault is in a membrane state of stress, a correspondence can be
established between forces acting on the element ABCD (stresses N x , N y , N xy = N yx and applied load for unit
area, p x ,p y ,p z ) and projected forces acting on the associated element A’B’C’D’ ( N x , N y , N xy = N yx and p x , p y , p z )
in the xy-plane [Baratta and Corbi 2006]. In absence of horizontal loads and if the vertical load is not dependent on
"y", as it happens when the vault is subject to only vertical loads due to the self-weight (i.e. p z = p z (x ) ≥ 0 ), and,
additionally, assuming that the vault has an indefinite length in the direction y, equilibrium may be expressed in the
form

∂ 2 ψ (y ) ∂ 2 z ∂2ψ(y )
= −p z , = Nx (2)
∂ y2 ∂x 2 ∂ y2

which reduces the problem to the determination of stress function ψ(y).

Assuming that the directrix curve of the vault is a circular arch (Figure 2) of radius R, with constant thickness "s"
and unit weight γ, and imposing suitable constraint conditions, one yields the final solution [Baratta and Corbi 2006]

R2 ⎡arcsin(t ) + 1 − t 2 + C⎤
z(t ) = − γ s ⎢⎣ ⎥⎦ (3)
H
⎛ H zo ⎞ R2 ⎡
with C = −⎜⎜1 + ⎟, H = γs 1 − t arcsin(t1 ) − 1 − t12 ⎤⎥ (4)
⎝ R 2 γ s ⎟⎠ (z1 − zo ) ⎢⎣ 1 ⎦

where zo and z1 are arbitrary ordinates, conditioned by the fact that z(t) should be contained in the interior of the
profile of the vault.
After this result, it is possible to calculate the internal forces N x ≤ 0, N y = N xy = 0 and N x ≤ 0 , N y = N xy = 0

Nx H
Nx = = (5)
cos ϕ cos ϕ

It is also possible to realize that the equilibrium solution allows the structure to behave as a sequence of identical
independent arches. From this result, in the following one refers to the portal arch model, reinforced or not with
some FRP strips, whose analytical problem implementation is shown to give theoretical results in perfect agreement
with the produced experimental data, also exhibiting very effective results in the reinforced case.

368
 z(t)
s
zo
R
x x z1 C
x1
L

z z
Figure 2: Cross section of a barrel vault with circular arch generatrix.

3. NUMERICAL/EXPERIMENTAL RESULTS FOR FRP REINFORCED OR

UNREINFORCED MASONRY MODELS
Results from analytical and experimental investigations [Baratta 1991, Baratta and Corbi 2005] developed on
masonry portal arches made of tuff brick with lime mortar, reinforced or not with carbon fibre strips and subject to
the self-weight and to an increasing horizontal force, are briefly reported in the following. Actually, as regards to the
consolidated case, one refers to the adoption of a continuous mono-directional FRP strips applied on the extrados of
the arch.
In both the reinforced and unreinforced cases, the representation of the collapse condition deriving from the
theoretical settlement of the problem leads to a situation relevant to the opening of fractures and subsequent collapse
mechanism activations which perfectly agrees with the real situation monitored during experiments: this is clear
from Figure 3, showing the deformed configurations for the two cases obtained by the codes which have been
produced for numerically implementing the problems, whence one can deduct the hinges distribution.
Figures 4 depict the numerical/experimental comparison of the absolute displacement u (mm) of the right abutment
versus the horizontal force F (N) for the two cases, showing a perfect agreement of the data and, moreover, a pretty
consistent increase in the model loading capacity when adopting the FRP reinforcement.

a) b)
Figure 3. Sketches of the deformed configuration at collapse condition for the
a) unreinforced and b) reinforced models as deducted from calculus codes.

369
 10 u (mm) 10 u (mm)

8 Experimental 8
results
Experimental
6 6 results

4 4

Trend line Trend line

2 2

0 F (N) 0 F (N)
0 20 40 60 80 100 0 200 400 600 800 1000
Figure 4. Absolute displacement u (mm) of the right abutment versus the horizontal force F (N):
a) unreinforced and b) reinforced model.

4. CONCLUSIONS
The paper focuses on the possibility of proposing a theoretical treatment for the reinforcement with FRP of barrel
masonry vaults. To this aim a theoretical approach of the problem of analysis of barrel vaults is presented, leading,
under the hypothesis of membrane state of stress, to equilibrium and admissibility conditions. The case of barrel
vault with indefinite length is specifically considered and demonstrated to behave according to the model of a series
of independent arches. Upon this result, one refers to the portal arch model, reinforced or not with some FRP strips,
whose analytical problem implementation is shown to give theoretical results in perfect agreement with the
produced experimental data, also exhibiting very effective results in the reinforced case.

5. ACKNOWLEDGEMENTS
The present research has been developed thanks to the financial support by the Dept. of "Protezione Civile" of the
Italian Government, through the RELUIS Pool (Convention n. 540 signed 07/11/2005, Research Line n. 8)

REFERENCES
Baratta, A., (1991). “Statics and reliability of masonry structures”, in “Reliability Problems: General Principles and
Applications in Mechanics of Solids and Structures”, F.Casciati & J.B.Roberts Eds, CISM, Udine, Italy.
Baratta, A., Corbi, O. (2003) “The No Tension Model for the Analysis of Masonry-Like Structures Strengthened by
Fiber Reinforced Polymers”. Intern. Journal of Masonry International, British Masonry Society. Vol.16, No.3, pp
89-98.
Baratta, A., Corbi, O. (2005) “On Variational Approaches in NRT Continua”. Intern. Journal of Solids and
Structures, Elsevier Science. Vol. 42, pp 5307-5321.
Baratta, A., Corbi, O. (2005) “Fibre Reinforced Composites in Civil Engineering. Experimental Validation of C-
Fibre Masonry Retrofit”. Intern. Journal of Masonry International, British Masonry Society. Vol.18, No. 3, pp 115-
124.
Baratta, A., Corbi, O. (2005) “Relationships of L.A. Theorems for NRT Structures by Means of Duality“. Intern.
Journal of Theoretical and Applied Fracture Mechanics, Elsevier Science. 2005, Vol. 44, pp 261-274.
Baratta, A., Corbi, O. (2006)“Analysis of Masonry Vaulted Systems: the Barrel Vaults”. SAHC 2006, the 5th
International Conference on Structural Analysis of Historical Constructions. November 6-8, 2006, New Delhi, India
Heyman, J., (1966). “The stone skeleton”. Journal of Solids and Structures. Vol.2, pp 269-279.
Heyman, J., (1977). “Equilibrium of shell structures”, Oxford University Press.
Ugural, A.C., (1999). “Stresses in plates and shells”, McGraw-Hill.

370
Third International Conference on FRP Composites in Civil Engineering (CICE 2006)
December 13-15 2006, Miami, Florida, USA

ON THE REINFORCEMENT OF MASONRY WALLS

BY MEANS OF FRP PROVISIONS
Alessandro Baratta
(Full Professor, University of Naples “Federico II”, Naples, Italy)

Ileana Corbi
(University Researcher, University of Naples “Federico II”, Naples, Italy)

ABSTRACT
In a much more extended research a computational procedure for solving two-dimensional equilibrium problems,
which are representative of the behavior of masonry walls loaded by in-plane forces, has been developed. So in
order to test the procedure some laboratory tests concerning masonry panels designed to be similar as the theoretical
model are made. The two tested panels have been built: the first panel with tufa bricks not jointed to each other, and
the second panel with tufa bricks jointed by a pozzuolana mortar, in order to test the influence in terms of resistance
that is conferred to the masonry. After the collapse some reinforcements are directly applied on the panel and a
comparison between the behavior of the masonry panel before and after the application of the refurbishments is
shown.

KEYWORDS
Structural analysis, Laboratory tests, Masonry panel, Refurbishments tecnique, FRP reinforcements.

1. INTRODUCTION
This paper represents a part of a more extended research about a procedure for solving two-dimensional equilibrium
problems, which are representative of the behavior of masonry walls loaded by in-plane forces. The elaboration of
the mechanical model of a structure, e.g. a masonry panel, requires first of all that the characteristics of the masonry
texture are formally and qualitatively defined because they strongly condition the behavior and the resistance of the
structure. In the case of a structure composed by a regular distribution of masonry square blocks, it can be modeled
by assuming that the material has not the capacity to transmit any tensile stress along the joints’ direction, but that a
low and significant tensile resistance can arise by means of a suitable relative stagger between the blocks. So the
panel can be associated to an homogeneous bi-dimensional continuum by considering a very low tensile resistance
due to the combined action of the friction and of the stagger of the bricks, in absence of the breaking of the single
block. On the other hand, in order to study a masonry having an irregular texture, the Not Resisting Tension (NRT)
model can be assumed as reliable, exhibiting a simple linear elastic behavior under compression stress states and no
resistance in tension, and, thus, resulting in an overall fragile non-linear behavior. If the loading capacity of NRT
structures can be investigated by means of the tools of the Limit Analysis (L.A.) theory, on the other hand, the study
of the intermediate crack situation can not be performed by L.A.-techniques, whilst the elastic analysis of the
masonry tissue under the assumptions of perfect integrity of the structure and of purely compressive stresses can
lead to significant results. In coclusion, some optimization (stress or strain) procedures, deriving from the
implementation of the basic variational methods extended to NRT models, can be developed [for the extended
procedure see e.g. Baratta and Toscano (1982), Baratta and Voiello (1988), Baratta (1991), Baratta and Corbi
(2003), Baratta and Corbi (2004)].
In order to test the developed computational procedure some laboratory tests concerning masonry panels designed to
be similar as the theoretical model are made. In the following a comparison between the behavior of a masonry
panel made with alone tufa bricks and a masonry panel made with tufa bricks jointed by a light mortar is shown.
After being collapsed the two panel have been reinforced by means of some usually reinforcements.

371
Fiber Reinforced Polymers (FRPs) are a type of composites characterized by a polymeric matrix reinforced with
continuous fibers (see Schwegler, 1994; Traintafillou, 1996), which exhibit desirable features, such as high
mechanical properties, lightweight, high resistance to chemical agents and corrosion, increased fatigue resistance,
reliability and durability, low thickness, adaptability and easy applicability to complex structural shapes, low
invasiveness on the construction. A wide range of amorphous and crystalline materials can be used as the fiber in
FRP materials. In the construction industry the most common fiber used are: glass fiber, carbon fiber or aramid
fiber. Carbon, glass and aramid fibers can be used separately or in conjunction as a hybrid to increase the stiffness of
a structural member or the area within a structure or some the other mechanical characteristics of the FRP fibers
(e.g. density, elastic modulus, tensile strength, ultimate elongation, etc). The embedded fiber matrix (polyester,
urethane, vinilester or epoxy) is responsible not only for keeping the fibers together but also to protect them against
environmental and localized effects. In order to help the application of the FRP reinforcements which are directly
laminated on the structure, some fire retardants are usually incorporated in the resin itself or as an applied gel-coat.
Fillers and pigments are also used in resins for a variety of purposes, the former principally to improve mechanical
properties and the latter for appearance and protective action. In the field of structural rehabilitation the most
commonly adopted forms of FRP are strips and tissues; strips, which are given by parallel continuous fibers, exhibit
a mono-directional behavior, and are then strongly non-isotropic; tissues, which are obtained by the plait of two
series of parallel fibers, are characterized by reduced mechanical characteristics of the final product, but are reduced
in their anisotropy as well.

2. LABORATORY TESTS ON THE MASONRY PANEL

Some masonry panels have been realized at the Laboratory of Materials and Structural Testing of the University of
Naples “Federico II”, which are symmetrical, with a central hole covered by a steel architrave, and having upper part
characterized by a concrete fascia lightly reinforced by steel. The panel geometry is shown in Figure 1(a).

2,3 m

0,124 m 1

0,200 m
3 2
2,23 m

4
architrave
1,322 m

FRP

0,382 m
0,775 m 0,750 m
Transducer Strain-gauge
(a) (b)

architrave architrave

FRP FRP

(c) (d)

Figure 1: (a) Masonry panel geometry, (b), (c) and (d) different applications of the FRP strips on the panel.

2.1 Masonry panel without mortar

In the first case, the panel is made of tufa bricks (type “yellow tufa of Naples”, Italy) not jointed to each other, in
order to do not confer any additional resistance to the masonry; the masonry itself is characterized by unit weight
γ=10300 N/m and Young modulus E=5.5 GPa. In correspondence of the concrete fascia on the top of the panel a

372
varying force has been applied, and some loading/unloading cycles have been made. After the collapse some FRP
strips has been directly laminated on the masonry as in Figure 1(b), at the same time with the impregnation of the
fibers by means of a special bi-component epoxy resin, and some loading/unloading cycles have been made again.
The adopted reinforcement, produced by FTS, is a BETONTEX system GV330 U-HT, made of 12 K carbon fiber,
jointed by an ultra light net of thermo-welded glass. The mechanical characteristics of the employed carbon fibers
are: tensile limit stress σfrp= 4.89 GPa, elastic modulus in traction Efrp=244 GPa, limit elongation εfrp=2%. The FRP
strip is characterized by thickness of 0.177 mm and depth of 200 mm.
The induced displacements at some selected points [1, 2, 3 and 4 in Figure 1(a)] of the panel are recorded by a
monitoring equipment consisting of: 4 transducers, placed at different locations of the panel in order to record the
absolute displacements, and 15 strain-gauges, arranged in 3 blocks of 5 strain-gauges, each block is devoted to
record the related strain situation. In details two transducers are located horizontally at two different heights on the
panel right side (transducers 1 and 2), and two in correspondence of the opening, one in horizontal position at the
top of the left side of the hole (transducer 3) and the other one under the architrave, which is devoted exclusively to
control the panel deflection (transducer 4). A sample of the displacements s(mm) versus the varying force F(N) read
by the transducer 1 during the loading/unloading cycles in the not-reinforced case and in the reinforced case with
some horizontally applied C-FRP strips is shown in Figure 2(a) and (b).

2.2 Masonry panel with mortar

In the second case, the panel is made of tufa bricks jointed by a pozzuolana mortar in order to confer a light
additional resistance to the masonry. The masonry is made with the same type of tufa bricks used in the building of
the first panel, and a varying force has been applied in the middle left part of the panel, rather than on the top, in way
to mitigate the proneness of the panel to sliding of bricks with respect to each other. The induced displacements at
the selected points 1, 2, 3 and 4 have been recorded during some loading/unloading cycles on the alone masonry
panel [Figure 2(c)] and after the application of the C-FRP reinforcement. Two applications of the FRP strips have
been made: in the first case the C-FRP strips have been laminated along the vertical direction on the panel in order
to contrast the principal tensile stresses [Figure 1(c)]; in the second case other horizontal strips have been
superimposed on the last intervention [Figure 1(d)]. A sample of the displacements read by the transducer 1 during
the loading/unloading cycles in the not-reinforced case and in the two reinforced cases is shown in Figure 2(c), (d)
and (e).

2.3 General considerations

By the diagrams in Figure 2, which report the displacements s(mm) versus the varying force F(N) read by the
transducer 1, some considerations can been made. In first instance it is evident the effect of the mortar between the
bricks in terms of global resistance, so the panel with mortar collapses in correspondence of load value about 5000
N instead of 2500 N in case of the panel without mortar.
Then, with reference to the panel’s reinforcement by means of the application of some C-FRP strips, both in the
panel with mortar that without mortar, the major effect of the C-FRP intervention is the reduction of the stress in the
masonry. In general lower displacements at the locations monitored by the transducers can be recorded in the
consolidated case with comparison to the unconsolidated case. Actually one can notice that, with reference to the
same load intensity [e.g. in correspondence of the load value 3000 N in Figure 2(c), (d) and (e)], lower
displacements can be recorded in case of FRP insertions. Obviously the effect results much more evident in the
panel with mortar with respect to the panel without mortar, and is evident still more when the intervention becomes
more invasive as in the scheme of Figure 1(d). Moreover, the increase of the overall stiffness of the panel results in a
higher loading capacity with respect to not-reinforced masonry wall. In particular the trend of each curve, shows that
it is closer to the x-axis (representing the load variable), thus indicating an increase in the stiffness which is also
related to an higher collapse value of the load.

3. CONCLUSIONS
In the paper some laboratory tests developed for evaluating the effectiveness of alternative technologies, and based
on the adoption of innovative materials for repairing ancient constructions, are reported.
Actually the modern technology of materials offers a wide variety of possibilities for the refurbishment of existing
structures: new advanced materials are often preferable to traditional materials for their enhanced characteristics in
terms of effectiveness, reliability and flexibility, which always ensure high standard of performance.

373
In details one focuses on Fibre Reinforced Polymers (FRP), which are special composites currently attracting much
attention for refurbishing and/or consolidating masonry structures with effective and low invasive interventions.
All the considerations above mentioned give the idea that researchers and technicians before to make any
intervention on a structure it should be fundamental to know very well the stresses distribution and its fracture
scenario, in order to select the really most appropriate typology of reinforcement and use correctly the innovative
materials as the composites.

a) T rd 1 b) T rd 1 c ycle 1
c ycle 1
c ycle 2 20 c ycle 2
20

18 18

16 16

14 14

12 12

s(mm)
s(mm)

10 10

8 8

6 6

4 4

2 2

0 0
0 5 00 100 0 1 500 200 0 25 00 3 00 0 0 5 00 100 0 150 0 2 000 25 00 30 00
F(N ) F(N )

c) c ycle 1 d) e) T rd 1 cyc le 1
T rd 1 T rd 1 cyc le 1
c ycle 2 cyc le 2
40 40 cyc le 2 60
c ycle 3 cyc le 3

35 35
50
30 30

40
25 25
s(mm)

s(mm)

s(mm)
20 20 30

15 15
20

10 10

10
5 5

0 0 0
0 100 0 20 00 30 00 4 000 5 00 0 600 0 700 0 0 1 00 0 20 00 30 00 4 000 5 000 600 0 700 0 0 2 000 40 00 6 000 8 000 10 000 12 000
F (N ) F(N ) F(N )

Figure 2: Displacements recorded on the masonry panel without mortar in the not-reinforced case (a) and
with FRP strips (b), and on the panel with mortar in the not-reinforced case (c) and in the case with different
application of the FRP strips (d) and (e).

ACKNOWLEDGEMENT
The present research has been performed thanks to the financial support by the Department of “Protezione Civile” of
the Italian Government, through the RELUIS Pool (Convention n. 540 signed 07/11/2005, Research Line n. 8).

REFERENCES
Baratta, A. (1991). “Statics and reliability of masonry structures”, In: F.Casciati & J.B.Roberts (Eds), “Reliability
Problems: General Principles and Applications in Mechanics of Solids and Structures”, CISM, Udine, Italy.
Baratta, A., and Corbi, I. (2003). “Investigation of FRP consolidated masonry panels”. Proceedings of Ninth
International Conference on Civil and Structural Engineering CC03, paper n. 99, The Netherlands.
Baratta, A., and Corbi, I. (2004). “Iterative Procedure in No-Tension 2D Problems: theoretical solution and
experimental applications”, In: G.C.Sih & L.Nobile (Eds.), “Restoration, Recycling and Rejuvenation Technology
for Engineering and Architecture Application”, Aracne Ed, p. 67-75, Bologna, Italy.
Baratta, A., and Toscano, R. (1982). “Stati Tensionali in Pannelli di Materiale Non Reagente a Trazione”,
Proceedings of Fourth National Congress of AIMETA, Genova, Italy (in Italian).
Baratta, A., and Voiello, G. (1988). “Teoria delle pareti in muratura a blocchi: un modello discretizzato di calcolo”,
In: Franco Jossa e la sua opera, Giannini Ed., Napoli, Italy (in Italian).
Schwegler, G. (1994). “Masonry Construction Strengthened with Fiber Composites in Seismically Endangered
Zones”, Proceedings of the Tenth European Conference of Earthquake Engineering, Vienna, Austria.
Traintafillou, T.C. (1996). “Innovative Strengthening of Masonry Monuments with Composites”, Proceedings of the
Second International Conference Advanced Composite Materials in Bridges and Structures, Montréal, Canada.

374
Third International Conference on FRP Composites in Civil Engineering (CICE 2006)
December 13-15 2006, Miami, Florida, USA

STRENGTHENING MASONRY ARCHES WITH COMPOSITES

Giulio Castori
(Ph.D. Candidate, University of Perugia, Perugia, Italy)

Antonio Borri
(Professor, University of Perugia, Perugia, Italy)

Skip Ebaugh
(Vice President, Hardwire LLC, Pocomoke City, Maryland, USA)

Paolo Casadei
(Lecturer, University of Bath, Bath, United Kingdom)

ABSTRACT
The present research project investigates the application of innovative composite materials, based on fine steel cords
embedded in either an epoxy (Steel Reinforced Polymer) or cementitious matrix (Steel Reinforced Grout), to
strengthen masonry arches. This application combines, to the traditional advantages proper of Fiber Reinforced
Polymers (FRP), the performances of this new family of composite materials, reducing installation and material
costs, and inducing an increase of ductility particularly when a cementitious matrix is used. In this way the use of
these materials could become extremely interesting in the restoration of masonry arches in historical building, and
more particularly, in road, rail, and waterway infrastructures. In the UK, for instance, there are over 40,000 masonry
arch bridges, the majority of which, being at least 100 years old, are in need of repair due to natural deterioration or
lack of maintenance, or in need of strengthening due to ever increasing traffic volume and vehicle weight. In
response to this situation a comprehensive study on the behavior of masonry arches strengthened by composite
laminates is here presented. The influence of the type of fibers (steel and carbon), matrix (epoxy and cementitious),
their location (intrados and extrados) and boundary conditions are investigated in the laboratory on scaled samples.

KEYWORDS
Arches, Masonry, Fiber reinforced polymers, Steel reinforced polymers, Reinforcement.

1. INTRODUCTION
Thanks to their adaptability to the changes of the geometric configuration, masonry arches are able to distribute the
strain along the mortar joints, avoiding the formation of significant cracks. Thus the collapse mechanism doesn’t
depend by the materials’ limit strength, but it is due to the incapability of the structure to fit the horizontal and
vertical displacements of the abutments. Consequently it is clear that similar displacements should be considered
when strengthening masonry arches, introducing only systems which are able to reinforce the arch without changing
its constructive features. Such aims have led researchers to suggest strengthening masonry shells with FRP
laminates in the form of bonded surface reinforcements. There are several advantages related to this strengthening
technique: very low weight, high tensile strength and low thermal expansion coefficient. On the contrary, their up-
to-failure linear elastic behavior, which doesn’t allow for any ductility of the system, their lack of fire resistance and
their relatively high cost, may represent an obstacle for a widespread use. A new family of composite materials
based on high strength twisted steel wires of fine diameter (0.20-0.35mm), that can be impregnated with epoxy resin
or cementitious grout is presented in this paper. SRP has the potential to address the three shortcomings mentioned
for FRP, in fact: a) steel cords have some inherent ductility (thanks to their manufacturing process) and reduced cost
when compared to CFRP; b) impregnation with cementitious paste may overcome the problems of fire endurance
and reduce installation costs. The steel cords used in SRP/SRG are obtained from the same manufacturing process
used for making the reinforcement of automotive tires, and re-manufactured, to obtain the shape of the fabric tape

375
prior to impregnation. The twisting of the cords allows some mechanical interlock between the cords and the matrix,
and may also induce a ductile behaviour upon stretching. Huang et al. (2002) investigated the mechanical properties
of SRP/SRG, testing different kinds of matrix (epoxy and cementitious). Test results showed that the material does
not experience a substantial yielding, but rather a similar behaviour to the one experienced by high-strength steel
used in prestressed concrete construction, with a slight non-linear range prior to rupture of the cords. Thus, to clarify
the behavior of brick masonry arches strengthened by composites, nine arch specimens have been tested under
monotonic vertical loads applied to ¼ of their span. Such condition is the most severe case of loading for an arch,
and can be considered to simulate particular situations of the arches (e.g. bridges, library, etc.). The aim of the study
is to compare the behavior up to collapse of arches strengthened by different types of fibers (steel and carbon) and
matrix (epoxy and cementitious), and by placing the reinforcement in different positions (intrados or extrados).

2. BEHAVIOR OF THE STRENGTHENED ARCHES

The application of composite laminates, as externally bonded strengthening materials, modifies the static behavior
of an arch because the reinforcement can bear the stresses occurring at the tensed edges. Therefore, the brittle failure
of such structures, typically caused by the formation of four hinges, can be avoided. Depending on the position of
the laminate, in fact, the formation of the forth hinge can be prevented (Foraboschi, 2004). In the case of extrados
strengthening the line of thrust can fall outside the lower edge of the arch. As a results, the arch becomes an isostatic
structure (three hinges arch) consisting of two curved beams strengthened on their upper sides. Conversely, in the
case of a structure strengthened at the intrados, although the outcoming static scheme is similar to the one adopted in
the previous case, the distribution of the stress is different: the thrust line falls outside the upper edge of the structure
and the fibers prevent the forth hinge formation close to the load point. Consequently, in both cases, the collapse is
due to other mechanisms, which are involving the limits of strength of the constituent materials (masonry and
reinforcement) and their interactions at the local level. Thus depending on the position and of the amount of the
reinforcement, the modified failure mode are: masonry crushing, sliding, debonding, reinforcement rupture.

3. EXPERIMENTAL STUDY
3.1 Characterization of the Materials and their Interaction

The experimental program comprises a series of preliminary tests for the mechanical characterization of the
constitutive materials of the arches. Concrete paviours (200x100x50mm) and an hydraulic lime mortar were used for
the construction of the arches. The bricks reached a compressive and a flexural stress equal to 43.3 and 10.9 MPa
respectively, whereas compressive tests on mortar prisms after 28 days of curing gave 0.8 MPa. The adhesion
properties between the masonry and the reinforcement were investigated for loads perpendicular (pull-off tests) to
the reinforcement, on sixteen strengthened specimens. In particular, two groups of specimens were cast for this part
of research, representing different type of reinforcement (SRP or SRG). The mean value of the tensile bond strength
was 1.29 and 1.57 MPa for specimens strengthened with SRG and SRP, respectively. As for the failure mode, while
all SRP specimens failed in the substrate, SRG specimens failed or in the overlay. Table 1 resumes some
geometrical and mechanical characteristics of the steel and the carbon fibers used as strengthening material.

Table 1: Properties of the fibers

Fibers Type
Property
Steel (3X2-4) Steel (3SX-12) Carbon (T700 SC)
Tensile strength (MPa) 2479 1657 4900
Elastic Modulus (MPa) 210000 210000 230000
Thickness (mm) 0.89 0.81 0.44
Ultimate strain (%) 1.6 1.2 2.1

3.2 Tests on the Strengthened Arches

A series of nine arch specimens (Figure 1) built by bricks arranged in a single layer (100mm of thickness) have been
tested under monotonic vertical loads applied to ¼ of their span. Different laminates arrangements and different

376
types of fibers have been used, namely: a control unreinforced specimen (UN.01), three specimens have been
strengthened with steel fibers (3SX) at the extrados by using a cementitious grout (EX.01 and EX.03 tests) and a
polymeric matrix (EX.02 test), a specimen have been strengthened at the extrados with carbon fibers (EX.04 test),
three specimens have been strengthened with steel fibers (3X2) at the intrados by using a cementitious grout (IN.01
and IN.03 tests) and a polymeric matrix (IN.02 test), whereas steel fibers (3SX and 3X2) and cementitious grout
have been used to strengthen the intrados and the extrados (IN+EX.01 test) of the last specimen. A single ply of
laminate, 150mm wide, has been applied for each arch, with the exception of the IN.03 specimen, where two plies of
laminate have been used; furthermore, in two cases (EX.03 and IN.03 tests), in addition to the reinforcement, steel
anchors were adopted: in the first case two angle plates were used to anchor the ply to the abutments, while in the
second case flat plates, screwed into the bricks, were used to secure the ply to the arch soffit.

Figure 1: Experimental set-up: dimensions of the specimens and scheme of loading.

The unreinforced arch showed a brittle failure (four hinges mechanism) and a very small load capacity (0.7 kN). The
arches strengthened at their extrados with 3SX fibers presented, with the exception of specimen EX.01, the same
failure mode (Figure 2a). In particular, while the specimen EX.01 showed, because of a set-up problem, a notable
rotation of the abutments, that not allowed any further increment in load, in the other two cases (specimens EX.02
and EX.03) the collapse occurred because of the sliding between brick and mortar in the first joint closest to the
springer and to the edge of the steel anchor, respectively; in both cases such collapse occurred without any warning.
The ultimate load was 9.2 kN for specimen EX.01, whereas it was 13.3 and 23.5 kN for specimens EX.02 and
EX.03, respectively. Conversely, the arch reinforced at its extrados by CFRP (specimen EX.04) showed the same
failure mode (sliding in the first joint closest to the springer) and a lower ultimate load capacity (11.5 kN). The
arches strengthened by 3X2 fibers at their intrados presented different patterns of collapse: specimens IN.01 and
IN.02 showed a brittle failure due to the reinforcement rupture, whereas in specimen IN.03 (where there were two
plies of laminate) collapse occurred due to local debonding of the reinforcements under the point of application of
the load (Figure 2b). In such case the failure was not brittle because the fibers contributed in holding the bricks
together during the last phase. The ultimate load was 16.2 and 14.7 kN for specimens IN.01 and IN.02 respectively,
whereas it was 23.0 kN for specimen IN.03. Finally, the arch strengthened both at its extrados and intrados with 3SX
and 3X2 laminate, respectively (specimen IN+EX.01), presented a different pattern of collapse (sliding in the joint
under the point of application of the load, Figure 2c) and a considerably higher ultimate load (32.8 kN).

a) b) c)

Figure 2: Failure mode: a) and c) sliding along a mortar joint (EX.03 and IN+EX.01), b) debonding (IN.03).

4. ANALYSIS OF THE RESULTS

The analysis of the experimental results allowed to evidence some aspects of the behavior of the strengthened arches
and propose some suggestions about the use of SRP/SRG in real cases. Figure 3 shows a load vs displacement
(measured at the location where load was applied) plot for all specimens. As for the ultimate behavior of the arches,

377
some important considerations can be drawn. For the arches strengthened at their extrados, despite different fiber
types have been used, the results showed that masonry sliding is the prevalent failure mechanism. Such kind of
failure takes place only on the arches strengthened at the extrados because the weakest point of the structure is the
hinge forming at the abutment. Thus, in the repair phase of real structures, a solution that could increase the ultimate
load capacity could be achieved, as done in specimen EX.03, by anchoring the first ply to the abutment by means of
steel plate bolted to the support and simply adhered to the reinforcement. For the arches strengthened at their
intrados, the reinforcement rupture in proximity of the loaded section has been detected to be the critical one. Thus,
in the repair phase of real structures, a solution that could increase the ultimate load capacity could be, as seen with
IN.03 test, the use of two plies of laminate, to avoid the reinforcement rupture, and steel anchors, in the form of steel
plates screwed in the masonry substrate, able to tie the strip to the arch soffit retarding premature delamination.
Finally the application of the reinforcement both at the intrados and at the extrados of the arch is clearly an ideal
application, used as well as the unreinforced arch, to evaluate the efficiency of the adopted strengthening method.
35 35

30 30
EX+IN.01

25 25
EX.03
IN.03

Load (kN)
20 20
Q (kN)

IN.01
15 15
IN.02 EX.02

EX.04
10 10
EX.01

5 5

0 UN.01
0
UN.01 EX.01 EX.02 EX.03 EX.04 IN.01 IN.02 IN.03 IN+EX.01
0 -10 -20 -30 -40 -50 -60
Q (kN) 0,7 9,2 13,3 23,5 11,5 16,2 14,7 23,0 32,8 Displacement (mm)

Figure 3: Comparison among the experimental results; load-displacement curves.

5. CONCLUSIONS
The following conclusions may be drawn from this experimental program:
• SRP/SRG composite materials have shown to be effective in increasing the ultimate load capacity of the arches.
• Mechanical anchors, not allowed in standard FRP applications, have shown to allow a substantial increase in the
ultimate load both for intrados and extrados applications.
• SRP/SRG are similar to FRP in terms of ease of installation.
• Cementitious grout well behaved in bonding the steel tape to the masonry substrate and provided an overall better
performance than the epoxy matrix allowing better redistribution of stresses between the laminate and the substrate.

6. AKNOWLEDGMENTS
The authors would like to acknowledge Hardwire LLC., Pocomoke City, MD, for providing the steel tapes and the
Department of Architecture and Civil Engineering at the University of Bath for hosting this collaborative research.

REFERENCES
Borri, A., and Castori, G. (2004). “Influence of bonding defects in masonry vaults and arches strengthened at their
intrados with FRP”, Proceeding of the 2nd National Conference, Venice, Italy, pp 7-16.
Foraboschi, P. (2004). “Strengthening of masonry arches with fiber-reinforced polymer strips”. Journal of
Composites for Constructions, ASCE 8(3), pp 96-104.
Hardwire LLC. (2002). What is Hardwire, http://www.hardwirellc.com. Pocomoke City, MD.
Heyman, J. (1982). The masonry arch, Ellis Horwood – Wiley, West Sussex, UK.
Huang, X., Birman, V., Nanni, A., and Tunis, G. (2005). “Properties and potential for application of steel reinforced
polymer and steel reinforced grout composites”. Composites, Part B, Vol. 36, pp 73-82.
Triantafillou, T. C. (1998). “Strengthening of masonry structures using epoxy – bonded FRP laminates”. Journal of
Composites for Constructions, ASCE 2(2), pp 96-104.

378
Third International Conference on FRP Composites in Civil Engineering (CICE 2006)
December 13-15 2006, Miami, Florida, USA

PRIORITIZED FRP RESEARCH FOR CONCRETE AND MASONRY

STRUCTURES
Max Porter
(Professor, Iowa State University, Ames, IA, USA)

Kent Harries
(Assistant Professor, University of Pittsburgh, Pittsburgh, PA, USA)

ABSTRACT
A workshop was held to determine the prioritized research needs of FRP use as related to concrete and masonry
structures, and hybrid structures. The workshop had seven sessions and 48 participants from different countries
representing various disciplines of academia, design, government, owner and code officials. Prioritized research
needs were established in each of the following topical session areas: internal FRP; external FRP; durability;
QC/NDE/SHM; fire and extreme loads; and hybrid structures. A final session of prioritized overall voting
determined rank-ordered research needs.

The workshop was sponsored by American Concrete Institute (ACI), ISIS of Canada, and the National Science
Foundation (NSF) was organized by ACI Committee 440-D. A final report was published in March 2005 and a
final project report was submitted to the NSF in July 2006.

The prioritized outcomes given in this paper are divided into two main categories; namely, those of Highly
Recommended Research and those of Recommended Research. The Highly Recommended Research generally fell
into the topics of Durability, Performance-Based Areas, New Materials and Systems, Need for Integrated Education,
and Research Partnerships. The paper gives a summary of these highly recommended items and references the
complete Workshop report.

KEYWORDS
FRP, Reinforcement, Durability, Reinforced Concrete, Reinforced Masonry.

1. INTRODUCTION AND CONDUCT OF THE WORKSHOP

A workshop sponsored by American Concrete Institute (ACI), ISIS of Canada, and the National Science Foundation
(NSF) was organized by ACI Committee 440-D. A final report was published in March 2005 and a final project
report was submitted to the NSF in June 2006.
The workshop goals were:
• To Identify and Prioritize Research Issues:
- By industry, practitioners and academia
• To improve our understanding of behavior of
- FRP materials
- FRP structural
- Repair systems.
The detailed objectives of the workshop were as follows:
• Develop a consensus of state-of-the-art in application FRP composites for infrastructure applications
• Identify critical research needs
• Develop a consensus on the priority of these needs.
• Identify emerging and novel applications
• Develop a coordinated plan
• Identify improved mechanisms by which research results may be disseminated

379
 • Provide a brief assessment of research facilities and capabilities in the USA

Session topics were determined based on the consensus of the Workshop steering committee and the topics
were largely based on categories identified in a survey of the profession conducted in early 2002 (Harries et al.
2002). Prioritized research needs were established in each of the following topical session areas: internal FRP;
external FRP; durability; QC/NDE/SHM; fire and extreme loads; and hybrid structures. A final session of
prioritized overall voting determined rank-ordered research needs.

The workshop participants were from various disciplines and countries as shown in Table 1 below:

Table 1: Distribution of Participants

By Discipline Number of Participants Percentage of each category

Academia 36 75
FRP Industry 6 13
Consulting/Design Engrg. 2 4
Gov’t./Owners/Code Officials 4 8
By Country
USA 34 71
Canada 10 21
UK 3 6
Belgium 1 2

2. PRIORITIZED OUTCOMES

The prioritized outcomes given in this paper are divided into two main categories; namely, those of Highly
Recommended Research and those of Recommended Research. First, the Highly Recommended Research generally
fell into the topics of Durability, Performance-Based Areas, New Materials and Systems, Need for Integrated
Education, and Research Partnerships. These are summarized below:
• Durability Topics:
- Identification of appropriate environments for durability testing - The environmental parameters
that need to be considered when using FRP materials remains a debatable question. Additionally,
the intended use, regional climates and maintenance practices will significantly impact which
parameters affect a particular application. Specific environmental exposures and/or durability
issues were identified as being critical in the application of FRP materials in concrete
infrastructure
- Development of standardized durability/environmental exposure test methods - A consensus on
accelerated environmental conditioning techniques and subsequent durability test methods is
required. Methods are required for both external FRP and internal FRP reinforcement applications.
- Durability studies of externally bonded FRP repair/strengthening measures - Identification of
time-dependent material properties and their effects on behavior, as well as the associated loading
factors (including fatigue) is needed. Clearly, the durability of the adhesive bond and/or substrate-
FRP interface is of primary concern.
- Durability studies of internal FRP reinforcement – Also, time dependent material properties and
effects on behavior and the factors affecting them (including fatigue and sustained loads) is needed
for internal reinforcement, as well. The critical issue here is the behavior of FRP embedded in
concrete; thus, research must account for the concrete environment in which the FRP is embedded,
the expected cracking behavior (which may differ from steel-reinforced concrete), the
environmental factors (which also differ somewhat from those of importance for steel-reinforced
members), and the impact on the material strength and bond of FRP in a concrete medium.
• Performance-Based Areas
- Service life prediction of FRP reinforced or strengthened structures - Models are needed to
extrapolate short term test results to long term service life models. Models for the degradation
processes are needed. Fatigue life of bonded FRP has been shown to be of particular concern and
predictive models of this behavior are required.

380
 - Fire resistance and protection of FRP reinforced or strengthened structures - The behavior of
FRP materials, whether imbedded in concrete or externally applied, subject to fire loading is
largely unknown. Modeling techniques must be developed and verified for predicting fire
performance of FRP materials and FRP reinforced or strengthened concrete structures.
- Seismic and Blast Resistance of FRP Systems - FRP systems are often used for structural retrofit
including efforts to mitigate the effects of earthquake or blast loads. Methods of assessing the
appropriateness of existing and innovative FRP systems for mitigating the effects of extreme
loading need to be developed, which also need to include the strain-rate effects.
• New materials and systems
- Innovative and hybrid materials - Research aimed at developing new and hybrid FRP materials
having properties better matched to concrete is necessary. Such systems may be as simple as
composite CFRP, GFRP and AFRP products or as innovative as polymer-free chemically
prestressed systems. More work is needed on hygrothermal behavior.
- Innovative reinforcing schemes - Development FRP materials should involve getting away from
the paradigm of “replacing steel with FRP” and toward the development of innovative reinforcing
schemes which should make both FRP reinforcement and concrete construction more cost-
effective. One role that concrete plays in reinforced concrete systems is to protect the reinforcing
system. If FRP systems can be made more robust and durable, this role for concrete becomes
obsolete and should result in a savings.
- Self-sensing FRP structural health monitoring systems - FRP materials are unique in terms of
their properties and their fabrication which lends itself well to the development of integrated
sensor systems. Such systems facilitate improved structural health monitoring and potential feed-
back to the occupants and owners over time.
• Need for integrated education – Although not specifically a research need, the education and training of
design professionals and the need for development and integration of student courseware is needed for the
following categories:
- Education in Schools of Engineering and Architecture.
- Training of design professionals
- Need for development and integration of student courseware in civil engineering, materials, and
architecture
• Research Partnerships - For seminal research to be conducted, partnerships are needed for the following:
- Close collaboration between academe and industry is essential
- Innovative industry/academe/government partnerships must be developed to facilitate seminal
research
- Interaction between industry and academe should take the highest priority and is likely to yield the
most fruitful results
• Other Recommended topics:
- Hybrid – Connections
- External FRP – Fundamental design philosophies
- Hybrid – Innovative design approaches
- Hybrid – Mechanical anchorage, bonded or not
- Durability – Extrapolate short-term test results to service life
- Internal FRP – Structural performance issues
- External FRP – Emerging systems
- Fire – Residual strength of members

The results of all of these prioritized research topics are tabulated in more detail in the workshop report (Porter and
Harries 2005). Recommendations are also given in the report for the topics to include in a plan of organized
research involving partnerships and a national laboratory.

3. SUMMARY

The NSF-sponsored Workshop had seven topical sessions and balloting was conducted to determine priority needs
for FRP research in the concrete and masonry environments. The Workshop included participants from various
engineering professions and from several countries. Those priorities presented in this paper are those deemed to be

381
of “Highly Recommended” and other “Recommended” topics. A Final Report showing details of the balloted topics
and more detailed topics was submitted to NSF (Porter and Harries 2005).

4. ACKNOWLEDGEMENTS

The contributions of all Workshop participants, the steering committee and especially the break-out session leaders
and recorders are gratefully acknowledged. The Workshop was sponsored by the National Science Foundation
(NSF) through grant number CMS 0338037. Additional support for the Workshop was provided by the American
Concrete Institute (ACI) and ISIS Canada.

5. REFERENCES

Harries, K.A., Porter, M. and Busel, J., 2003. FRP Materials and Concrete – Research Needs, Concrete
International, Vol. 25. No. 10, pp 49-54.
Porter, M.L. and Harries, K.A., 2005. Workshop on Research in FRP Composites in Concrete Construction – Final
Report. Submitted to the National Science Foundation, March 2005, followed by a complete project final report in
July 2006, 40 pp.

A summary ballot was developed and is shown below in Figure 1.

IIA IIB IIIA IIIB IVA IVB

INTERNAL FRP EXTERNAL FRP DURABILITY QC/NDE/SHM FIRE/EXTREME HYBRID
1 Material science 23 Time dependent 17 Develop standard 20 Data 7 Develop Connections 14
issues (durability, effects and factors durability test interpretation and fundamental models
ductility, creep, affecting them methods and integration for predicting fire
fatigue, fire, new interpretation resistance (materials
system and systems)
development)
2 Developing cost 20 Seismic design 9 Service life 16 Definition of 2 Performance of 17 2D and 3D grids
effective and gaps models QC/NDE/SHM seismic and blast (fabrication methods
innovative schemes objectives resistance of FRP of 3D grids)
for use of FRP reinforced and
strengthened
structural members
(concrete and
masonry)
3 Innovative 17 Fundamental 14 Extrapolate short- 12 Identify cost- 2 Residual strength 11 Innovative design 13
reinforcement design philosophies term test results to benefit of of internally and approaches
schemes (including service life QC/NDE/SHM externally reinforced
2D and 3D grids, members (fire or
etc.) impact)
4 Structural 11 Emerging sys- 11 Standardize test Self-sensing 16 Develop FRP 9 Mechanical 13
performance issues tems methods and FRP materials and/or anchorage, bonded or
(shear, bond) interpretation materials/self- systems to ensure non (flexural and
[covered in IIIA-1] powered sensors ductile failure shear anchors)
5 Fire response of FRP Material types 8 Model degradation 8 Approaches to 6 External 3 Industry 1
reinforced structures (thermoset vs. process define acceptance reinforcement – collaboration
(modeling, test thermopastics and criteria Fireproofing to meet
methods, etc.) hydrothermal ASTM E119
[covered in IVA] mechanical
mismatches)
6 Establish minimum 3 Development 5 Experimental 10
performance of new NDE data fire and extreme
requirements tools/methods loading
(FRP with
concrete)
7 Synergistic 10 Fire resistant FRP 6
effects
8 Write in topic (if you feel we have missed anything): New fireproofing 6
materials –
development and
understanding

Figure 1: Comprehensive Ballot (sample)

382
Part XIV. Modeling
Third International Conference on FRP Composites in Civil Engineering (CICE 2006)
December 13-15 2006, Miami, Florida, USA

AN ANALYTICAL AND NUMERICAL INVESTIGATION OF

DEBONDING PROBLEMS IN BEAMS STRENGTHENED WITH
COMPOSITE MATERIALS
Domenico Bruno
(Professor, University of Calabria, Cosenza, Italy)

Fabrizio Greco
(Associate professor, University of Calabria, Cosenza, Italy)

Paolo Lonetti
(Researcher, University of Calabria, Cosenza, Italy)

ABSTRACT
A refined model able to analyze edge debonding problems in beams strengthened with externally bonded composite
laminated plates, is presented. The structural system is viewed as composed by three different physical components:
the base beam, the adhesive layer and the bonded plate. Each component may be comprised by one or several
mathematical layers which adopts the first-order shear deformation laminate theory. Bonding and continuity
conditions between different layers are simulated by using the interface modelling technique. According to a
fracture mechanics approach, the analysis is carried out by evaluating the total and individual mode components of
energy release rate (ERR). Applications for typical strengthened systems, carried out by numerical integration
procedures, are proposed. The approximations introduced in the model with respect to the adopted number of
mathematical layers are analyzed and comparisons with existent FE models are given. For the simpler two-layer
model of the structure, a closed-form solution is obtained. Finally, the effect of different debonding modes on the
overall behaviour of the structural system is analyzed. These results show the capability and the accuracy of the
proposed approach to predict debonding failure behaviour in strengthened beams.

KEYWORDS
FRP strengthening, Debonding, Interface models, Computational simulation

1. INTRODUCTION
Composite materials, especially in the form of fiber-reinforced plastic (FRP) strips, are widely used for various
strengthening, upgrading or retrofitting applications of existing civil concrete or steel structures. A frequent flexural
strengthening technique, consists of externally bonding a laminated composite plates to the tension face of a
concrete or steel beam, by means of an adhesive layer. Bonding FRP plates to the external surface of structural
members leads to several improvements of the structural performance (see, for instance, Ramana et al., 2000; Lau &
Zhou, 2001). On the other hand, experimental evidences have shown that the increase in stiffness and strength
provided by the reinforcement is accompanied by a decrease of ductility leading often to debonding failure modes in
the FRP strengthened system, characterized by a brittle and unstable nature, which may compromise the
effectiveness of the reinforcement (see, for instance, Rabinovitch, 2004). In order to analyze debonding problems
several models based both on strength theory (see, for instance, Ziraba et al., 1994) and fracture or damage
mechanics (see for instance, Lau et al. 2001) have been proposed. On the other hand, despite the notable steps
forward made in this research field, additional progress should be achieved in order to reach a better understanding
of the mechanics of debonding between external strengthening system and the base structure. For instance, an in
dept characterization of the behaviour of interface fracture including the effects of mixed mode propagation is
necessary to obtain a realistic prediction of interface crack growth. Moreover, another interesting aspect is the
problem of fracture initiation which cannot be predicted by means of Fracture mechanics. The goal of the paper is to

383
analyze the capability and the accuracy of a refined mechanical model, based on a multilayer representation of the
structural system (see Bruno and Greco, 2001; Bruno et al., 2003) in coupling with interface modelling
methodology, in the analysis of debonding failure induced by cracks propagating within the adhesive layer or at the
interface between different constituents and starting at the edges of the strengthening plate.

2. MODEL OF THE LAYERED COMPOSITE STRUCTURE

The composite system consists of three physical layers: the base beam, the adhesive layer and the FRP plate, and is
assumed to contain an edge delamination at the interface between the adhesive layer and the beam as shown in
Figure 1. The delaminated structure is assumed to be described by a through the thickness sequence of mathematical
layers governed by the first order shear deformation laminate theory (FSDLT). Each of the mathematical layer may
represent several physical layers or part of a layer trough the thickness direction. The first layer is the lowest one and
the thickness of the i-th layer is denoted by ti. The loads are assumed constant in the width and the problem is
considered as a two-dimensional one in which all the generalized displacements are independent on the co-ordinate
y. Plane stress conditions are adopted in the present paper. The analysis of the composite system is carried out with
reference to a unit width.
i-th mathematical layer
Base beam
Mi(0) tb qi ti Fi j Mi(L)
Ni(0) Ni(L)
zi,wi x,ui Adhesive layer
Ti(0) ta Ti(L)
tp Composite plate a
Lu Ls
L

Figure 1. Illustration of the strengthened beam

With reference to the i-th layer the kinematics relations for membrane strain at the reference surface, curvature and
transverse shear strain, respectively, assume the following form:
ε i = ui′, κ i = ψ i′, γ i =ψ i + wi′ , (1)
where ui(x) and wi(x) are the mid-surface in-plane and transverse displacements, respectively, ψi(x) denotes rotations
of transverse normals and the prime denotes the derivative with respect to x. The constitutive relation for the i-th
mathematical layer is defined by means of the classical extensional Ai, bending-extensional coupling Bi, bending
stiffnesses Di and shear stiffnesses Hi.

2.1 Interface formulation

At the interfaces between mathematical layers displacement continuity is imposed by including appropriate zero-
thickness interface layers. Two types of interfaces are used, namely the strong interface and the collapsed interface.
The former model is implemented by means of a linear constitutive law between interlaminar tensile and shear
stresses and corresponding interface displacement jumps, ∆w, ∆u, and involves two stiffness parameters kz, kxy:
1
σ zz = 1 − d (1 + sign ( ∆w )) k z ∆w, σ zx = (1 − d ) k xy ∆u , (2)
2
where d is a damage variable, taking the value 1 for no adhesion and the value 0 for perfect adhesion, and sign is the
signum function introduced to simulate sliding contact in the delaminated region for very large values of kz. Treating
the stiffness parameters as penalty parameters leads to capture stress singularities at the interface crack tip, which in
the limit as interface stiffness approach infinity are lumped into interfacial concentrated forces, and gives the
following calculation for the ERRs:
1
im k ∆w2 ( s ) ∆w ≥ 0 1
G = GI + GII , GI = kz ,kxy →∞ 2 z , GII = im kxy ∆u 2 , (3)
kz , kxy →∞ 2
0 ∆w < 0
where interface displacement jumps are evaluated at the interface crack tip. The strong interface formulation
provides an efficient numerical model for debonding, and will be adopted for numerical applications. On the other
hand, the collapsed interface model is implemented by introducing Lagrange multipliers, representing interfacial

384
stresses, and is adopted in order to reduce the number of generalized displacement variables. The related constraint
equations imposing displacement continuity requirements between any two adjacent layers, i and i+1, are:
t t
∆ui = 0, ∆wi = 0, where ∆ui = ui − i ψ i − ui +1 − i +1 ψ i +1 , ∆wi = wi − wi +1 . (4)
2 2
In the case when a collapsed interface model is introduced, ERR and its mode decomposition can be evaluated by
applying a virtual crack closure type formula in terms of interfacial concentrated forces at the delamination tip
arising as reactions to constraint equations (4). It can be shown that results obtained for ERR and mode
decomposition are equivalent to those obtained by using the strong interface formulation. As a matter of fact, in the
limit of the strong interface procedure stress singularities become equivalent to the above mentioned concentrated
forces.

3. RESULTS
For simplicity at this stage and without loss in generality, the analysis is carried out by adopting only two
mathematical layers to model the strengthened beam and using a collapsed interface model to simulate adhesion.
The upper mathematical layer is denoted with the subscript 2 and the lower one with the subscript 1. The obtained
field equations will be used for the computation of closed-form expressions for ERR. In particular one mathematical
layer represents the element to be strengthened and the other one the composite system, comprised of the adhesive
layer and the FRP layer. Energy release rate and mode partition are evaluated by using the virtual crack closure
technique (VCCT) by using a procedure introduced by the authors (see Bruno et al., 2003) in the context of a multi-
layer model. According to the VCCT the energy released into individual modes is half the work performed by
interfacial concentrated forces at the delamination tip through displacement jumps occurring after the delamination
is extended by da. Consequently, the mode I, mode II and the total ERRs are evaluated as
1 1 t t 2
1
GI = T1 d γ 1 − γ 2 d , GII = N1 d ε1 − ε 2 − κ1 1 − κ 2 2 , G = Niε i + Miκi + Tiγ i d − Ti d ψ i (5)
2 2 2 2 d i =1 2

where the double bracket f k

=f+-f- denotes the jump across the delamination tip x=Lu+Ls-a and Ni, Mi, and Ti, are
the axial, bending and shear stress resultants, respectively. Closed-form solutions for the quantities in eqs (5) can be
obtained by using the equilibrium solution for the two-layer system.
The two-layer model suffices only to obtain an accurate evaluation of the total ERR, whereas individual energy
release rates may be computed with a reasonable accuracy only if more than one mathematical layer is used to
model each physical layer. However the use of a refined multi-layer model implies that mode partition must be
performed numerically rather than analytically, since a general closed form solution of the governing equations is
hard to be obtained. Energy release rates are obtained by using the strong interface formulation eqs (3). An iterative
collocation method is adopted which starting from an initial guess for the solution and the mesh, at each iteration the
method adapts the mesh to obtain a sufficiently accurate numerical solution. Results shown in Figure 2 with
reference to a cantilever beam subjected to an edge transverse force evidence that the total energy release rate from
the proposed approach converges to that calculated by means of an FE solution of the 2D continuum model of the
structural system as the number of mathematical layer increases. The characteristics of the bonded plate are
representative of a typical carbon-epoxy unidirectional lamina, whereas those of the beam refers to a typical
concrete beam. The numerical finite element analysis has been carried out by modeling the system with 2D plane
stress four-noded elements and computing the energy release rate by the aid of the VCCT. In addition in Figure 2 the
closed-form solution extracted from equations (5) is plotted for comparison purposes. The structure is here divided
into 7 mathematical layers: 2 for the FRP plate; 1 for the adhesive and 4 for the beam. Starting from the lowest one
their thicknesses are, respectively, 0.5 tp, 0.5 tp, ta, 0.01 tb, 0.07 tb, 0.32 tb, 0.6 tb. This layer assembly is denoted as
(2/1/4). Figure 2 shows that the total energy release rate has an excellent agreement with the FE results with relative
errors smaller than 1%. Consequently, the solution provided by the closed form expressions of eqs (5) is refined by
using an appropriate layer assembly and the accuracy is improved by increasing the number of layers. The
convergence of the total energy release rate and of its individual mode components for a fixed delamination length
(a=400mm) is also shown at the right of Figure 2, where it can be observed that individual energy release rate
components converge too, but toward asymptotic values different from FE predictions, which are mesh dependent
due to the oscillatory singularities. The (1/1/4) assembly adopts the following thickness distribution: tp, ta, 0.017 tb,
0.05 tb, 0.33 tb, 0.6 tb, whereas the (1/1/2) assembly adopts the following thickness distribution: tp, ta, 0.33 tb, 0.67
tb.

385
 25

F
GT - FEA a
2.00 a=400
Ls
20 GI - FEA L

GII - FEA
GT - Analytical
1.60
G T - Analytical
GT - Numerical (2/1/4)
15 G T - FEA
2

GI - Numerical (2/1/4)
G E 2t2/F

1.20

2
G I - FEA

G E 2 t 2 /F
GII - Numerical (2/1/4)
G II - FEA
10 0.80 G T - Numerical
G I - Numerical
0.40 G II - Numerical
5

0.00
(1/1) (1/1/1) (1/1/2) (1/1/4) (2/1/4)
0
0 200 400 600 800 1000 1200
a[mm]

Figure 2. Convergence to the FE results for a cantilever beam configuration under an edge transverse force
as a function of the delamination length and for a fixed delamination length (a=400mm).

4. CONCLUSIONS
An improved mechanical model for the analysis of debonding failure starting at the edge of beams strengthened with
externally bonded composite plates is proposed, by modelling a strengthened structural system as an assembly of
shear deformable mathematical layers. Introducing strong and collapsed interface layers and an appropriate layer
assembly based on the structure physical configuration, an accurate description of the problem is obtained. It has
been shown that for a two-layer version of the proposed model, closed-form solution for the total energy release rate
are available. Unfortunately, the energy release rate mode components predicted by the above analytical solution are
not sufficiently accurate and, as a consequence, one has to introduce more than one mathematical layer to model
each physical component to capture the local warping of the layers near the crack tip. The energy release rates
curves as a function of the crack length for different structural schemes and different debonding failure modes,
confirm the brittle and catastrophic unstable behavior which accompanies debonding growth and the observed
failure modes pointed out during experiments. Results obtained by numerical integration for typical strengthened
systems, compared with predictions from very refined 2D-continuum FE investigations, show the capability and the
accuracy of the proposed approach to predict debonding failure behavior. Contrary to the FE methodologies, which
involve considerable complexities due to the detailed mesh required to capture edge stress singularities and
differences in length scales and in mechanical properties of the single components of the system, the proposed
approach is more computationally efficient due to the use of plate variables.

5. REFERENCES
Bruno, D., Greco, F., (2001). “Mixed mode delamination in plates: a refined approach”. Int. J. Solids Structures Vol.
38/50–51, pp. 9149–9177.
Bruno, D., Greco, F., Lonetti, P., (2003). “A coupled interface-multilayer approach for mixed mode delamination
and contact analysis in laminated composites”. Int. J. Solids Structures, Vol. 40, pp. 7245–7268
Lau K.-T., Zhou L.-M. (2001). “Mechanical performance of composite-strengthened concrete structures”.
Composites Part B: Engineering, Vol. 32, pp. 21-31.
Lau KT, Dutta PK, Zhou LM, Hui D (2001). “Mechanics of bonds in an FRP bonded concrete beam”. Compos: Part
B; Vol. 32, pp. 491 –502.
Rabinovitch, O. (2004). “Fracture-mechanics failure criteria for RC beams strengthened with FRP strips-a simplified
approach”, Composite Structures, Vol. 64, pp. 479-492.
Ramana V.P.V., Kant T., Morton S.E., Dutta P.K., Mukherjee A., Desai Y.M., (2000). “Behavior of CFRPC
strengthened reinforced concrete beams with varying degrees of strengthening”, Composites Part B, Vol.31: pp.
461-470.
Ziraba YN, Baluch MH, Basunbul IA, Sharif AM, Azad AK, Al-Sulaimani GJ (1994). “Guidelines toward the
design of reinforced concrete beams with external plates”. ACI Struct ; Vol. 91(6), pp. 639 –46.

386
Third International Conference on FRP Composites in Civil Engineering (CICE 2006)
December 13-15 2006, Miami, Florida, USA

NUMERICAL ANALYSIS OF TWO-WAY CONCRETE SLABS WITH

OPENINGS STRENGTHENED WITH CFRP
M. Sc. Piotr Rusinowski
(Ph.D. Student, Technical University of Denmark, Lyngby, Denmark
Norut Technology Ltd., Narvik, Norway)

Lic. Tech. Ola Enochsson

(Ph.D. Student, Luleå University of Technology, Luleå, Sweden)

Prof. Björn Täljsten

(Professor, Technical University of Denmark, Lyngby, Denmark
Luleå University of Technology, Luleå, Sweden)

ABSTRACT
Carbon Fiber Reinforced Polymers, CFRP, offer excellent corrosion resistance to environmental agents as well as
the advantages of high stiffness-to-weight and strength-to-weight ratios when compared to conventional
construction materials. One common application for CFRP sheets is to strengthen slabs and walls when openings are
to be made. In spite of this, there have not been many studies reported on slabs with openings strengthened with
CFRP and especially, not with distributed loading. This paper presents numerical analyses of simply supported two-
way concrete slabs with openings strengthened with CFRP sheets. The finite element program Abaqus is utilized for
the analyses. The analyses are compared with full-scale laboratory tests and show a good agreement.

KEYWORDS
CFRP, FE-analysis, strengthening, slabs, openings

1. INTRODUCTION
Reinforced concrete slabs and shells are commonly used structural elements in building structures. Due to changes
in use and new functionality demands, existing constructions often need to be rebuilt and new openings in existing
slabs or shells are to be made. The effect of small openings is usually neglected due to ability of the structure to
redistribute additional stresses. The problem appears with larger openings when the static system may be altered and
strengthening must be applied. External bonded FRP sheets are becoming increasingly used in reconstruction.
Despite the growing popularity of FRP strengthening, the existing design guidelines do not normally cover
strengthening of openings. The calculation method of FRP strengthening for slabs with openings, applied also in this
paper, uses the instructions concerning required amount of steel reinforcement in slabs cast with openings given by
the building codes. This method is described in Enochsson (2005) and more generally in Täljsten (2004). The study
presented in this paper is a part of the ongoing research on two-way concrete slabs with openings at Luleå
University of Technology.

2. OBJECTIVE

The aim of the study presented in this paper was to model the behaviour of CFRP strengthened concrete slabs using
Finite Element and to compare the result with experimental tests. The FE-model can then both be used to investigate
the influence of different opening-to-slab sizes on the behaviour and how to strengthen a slab most efficient due to a
made opening. However, the idea of the entire study and the experiments was to obtain as much information as

387
possible about the effect of openings in two-way RC slabs and the efficiency of the CFRP strengthening
configuration. In order to carry out a necessary comparison the following slab configurations were tested:
− Without opening (Homogeneous)
− With opening (Weakened)
− With opening strengthened with CFRP (Strengthened)
− With an additionally steel reinforced cast opening (Reinforced)

3. EXPERIMENTAL PROGRAM AND DESIGN OF TEST SPECIMEN

All specimens were quadratic with a side length of 2600 mm and a thickness of 100 mm. The openings were also
quadratic and located in the center of the slabs. The size of the openings, 850×850 mm, was slightly larger than
allowable for the simplified design method according to the Swedish code. Figure 1a shows the types of specimens
considered in this paper.
The slabs were manufactured in four concrete batches with a designed 28 days characteristic compressive strength
of 40 MPa. All slabs were reinforced with steel bars, Nps 50 φ5, with the nominal characteristic yield strength
fyk = 510 MPa. The bars were arranged into a welded net with spacing of s = 150 mm. The strengthening system was
provided by Sto Scandinavia AB with the brand name Sto FRP Sheet S300.
The distributed load was provided by a system of airbags, embedded in a heavy steel structure, see Figure 1b. Since
the load was applied from underneath, the specimens had to be placed “upside-down”. The steel frame, made of
I-beams, provided line supports along the specimens’ edges. The four load-cells, that connected the supporting
frame with the fixture in each corner, measured the magnitude of load.
The homogeneous slab was designed using the Swedish standard method, based on the theory of elasticity and in
some extent also on the yield line theory. The method, described in Hilleborg (1990), uses a simple formula for
estimating the maximum moment, from which the required amount of reinforcement can be calculated.
The slabs with the openings were designed with the traditional method, which can be described in two phases. In the
first phase, the calculation for a homogeneous slab is to be carried out. Afterwards, the moment capacity from the
area where the opening is to be made is distributed into the bands around the opening. In principle, with an
assumption that moments in this area are constant, the method leads to spreading designed reinforcement around the
opening. The procedure is described in Enochsson (2005).

A
Test specimen Interior structure

Section A

B
q

Homogenous Weakened Support structure Opening

Exterior structure embedding airbag

Base structure
Load cell q Section B q

a) Strengthened Reinforced b)

Figure 1: a) Specimens, b) Test set-up and static system

4. NUMERICAL ANALYSIS
Numerical analysis of the slabs considered in this paper included several nonlinear considerations concerning
material properties, boundary conditions and geometries. In order to achieve the convergence application of explicit
integration had to be used. The finite element mesh was regular for all parts in the model, see Figure 2a. The
concrete part of the slab was modeled with eight-node brick elements with reduced integration. The reinforcement
grid was represented by discrete truss elements. Support steel plates, which transferred reaction forces by means of
contact formulation, were modeled by shell elements with reduced integration. Orthotropic property of CFRP sheets
was assigned by using membrane elements with no stiffness in the direction perpendicular to the fibers.
A damaged plasticity constitutive model for concrete was used in the FE-calculations. The post failure behavior in
tension was specified in terms of the stress-displacement response described by a bilinear curve, see Figure 2b.

388
The model for reinforcing steel was ideal elasto-plastic and full bond between reinforcement and concrete was
assumed. The CFRP material was considered to be elastic. The interface between the concrete slab and the CFRP
was modeled as ideal bonding, i.e. full composite action. This property was obtained by using interaction “tie”
between surfaces, which locks the distance between adjacent nodes. The FE-model assumed an ideal line support
which cannot be achieved in the field conditions due to imperfections of the test rig and uneven concrete surface.
Commonly, some elastic deformations are enabled and the supports can be then modeled as a set of discrete springs,
see Rusinowski (2005), or continually by applying simply supported deformable plates along slab edges. The latter
method was used for comparison between elastic and stiff supports. Figure 2c presents deflection curves obtained
from the experiment and the analyses of CFRP strengthened slab. The analysis with the elastic supports shows good
agreement with the experiment until plastic phase but introduces dynamic problem later on which was noticed in the
curve before smoothing. This fact cause inaccuracies especially considering concrete damaged plasticity model and
for this reason stiff line supports are used in further analyses.
60
a) b) σct c) Stiff support (FEM)
50 Experiment
fct

Load [kN/m2]
40
Elastic support (FEM)
30

20
0.3fct
0.01fct 10
0.3δ0 δ0 δ
0
0 10 20 30 40 50 60
Deflection [mm]

Figure 2: a) FE-mesh, b) Concrete tension behavior, c) Model of elastic support

5. RESULTS
The result from the numerical analysis is compared with the outcome from the experiments. The displacements were
measured in the middle of the opening’s edge, where slabs deflect most. Comparison with the load - deflection curve
is the easiest way to judge the general behavior and the load capacities of the slabs. Figure 3a presents the load -
displacement relationships for all the slabs both for the experiments and the numerical analyses. The shapes of the
curves show relatively good agreement between the experiments and the FE analyses. The CFRP-strengthened slab
hardens after the plastic phase. The possible reason for this phenomenon might be elastic behavior of carbon fibers
and the fact that the CFRP strengthening is distributed, in contrast to steel reinforcement. These two features may
delay crack initiation and propagation. The strains were measured in three locations along the openings. Strain
distribution along one of the CFRP sheets, see Figure 3b, is compared in three phases called: elastic, cracking and
plastic. The “elastic” phase is always set to load 10 kN/m2 regardless slab type and represents the state before
concrete reaches its tensile strength. The “cracking” phase corresponds the time when a slab begins to crack. The
first cracks, which are very narrow, could not be noticed during the experiments and this moment could be detected
only in numerical analyses. Finally “plastic” phase represents the moment when slab is fully cracked and begins to
deform plastically. This stage has been found individually for the experiments and the numerical analysis.

60 Strengthened FEM 10 kN/m2

50 Strengthened 300 FEM 13.9 kN/m2

Strain [μm/m]
Load [kN/m2]

FEM 35 kN/m2
Exp 10 kN/m2
40 Homogeneous
Exp 13.9 kN/m2
Weakened 200 Exp 33.5 kN/m2
30 Weakened Reinforced
Homogeneous
20 100
10
Experiments FE analyses
0 0
0 10 20 30 40 50 60 0 10 20 30 40 50 60 0 0.2 0.4 0.6 0.8 1 1.2
Deflection [mm] Deflection [mm] Distance, x [m]
a) b)

Figure 3: Comparison of results: a) maximum deflection, b) strain distribution along the opening edge

389
Figure 4 shows similarities in comparison of crack distribution between experiments and numerical analyses. The
plastic region in CFRP strengthened slab is more widely spread compared with the other analyzed and tested slabs,
which may be the effect of distributed strengthening.

Weakened Strengthened Reinforced

Experiments

FEM

Figure 4: Comparison of crack distribution obtained in experiments with final principal plastic strain distribution in
FE analyses

6. CONCLUSIONS
The FE calculations used in this study show good agreement with the experiments although relatively simple model
was applied. In opposite to the experiments, however, CFRP failure was not reached in the numerical calculations.
In FE analyses, using either smeared cracking or damaged plasticity model of concrete, cracks are distributed over a
larger region. In the reality, a few discrete cracks propagate and the stresses are concentrated in the reinforcing steel
and CFRP sheets. In these formulations mesh must be much finer to localize failure.
The model of CFRP strengthening assumes an ideal bonding, i.e. possible peeling off at crack cannot be
investigated. Furthermore, the stiffness of CFRP is assumed to be equal in compression and tension and this is
suitable for cases where CFRP sheets are only stretched. For more complex problems, as two-way concrete slabs or
structures subjected to cycle loading, different material properties of CFRP in tension and compression should be
assigned.
Apart from these observations, the explicit FE analysis gives more insight to the strengthening effect of the CFRP
sheets. It also gives more confidence to the future utilization of the non-linear FE analysis in order to evaluate larger
slabs with different opening configurations.
For further investigation of two-way concrete slabs with openings it is necessary to study the bond behavior between
the CFRP and the underlying concrete in detail. Furthermore, better models of the crack localization in the concrete
are needed to be able to accurately predict the ultimate failure.

7. ACKNOWLEDGEMENT
The research work presented in this paper has mainly been performed at Luleå University of Technology in Sweden,
and has been financed by Skanska Teknik AB Sweden and SBUF (The Development Fund of the Swedish
Construction Industry) and also partly by the European Union regional funds, Sto Scandinavia AB and the
Norwegian Research Council through the strategic institute program RECON at Norut Technology Ltd.

8. REFERENCES

Enochsson, O. (2005): CFRP Strengthening of Concrete Slabs, with and without Openings – Experiments, Analysis,
Design and Field Application. Licentiate Thesis 2005:87, Luleå University of Technology, Div. of Structural
Engineering, ISSN 1401-1757, p 168.
Hillerborg, A. (1996): Strip Method - Design Handbook. First edition. London, UK: Chapman & Hall. ISBN 0-419-
18740-5.
Rusinowski, P. (2005): Two-Way Concrete Slabs with Openings – Experiments, Finite Element Analyses and
Design. Master’s Thesis 2005:200 CIV, Luleå University of Technology, Div. of Structural Engineering, ISSN
1402-1617. p 126.
Täljsten, B. (2004): FRP Strengthening of existing concrete structures - Design Guidelines. Third edition. Luleå,
Sweden: Luleå University printing office, ISBN 91-89580-03-6, 2004, p 230.

390
Third International Conference on FRP Composites in Civil Engineering (CICE 2006)
December 13-15 2006, Miami, Florida, USA

NUMERICAL MODELING OF FRP SHEAR STRENGTHENED

RC BEAMS USING COMPRESSION FIELD THEORY
Zhe Qu
(Postgraduate, Tsinghua University, Beijing, China)

Xin-Zheng Lu
(Lecturer, Tsinghua University, Beijing, China)

Lie-Ping Ye
(Professor, Tsinghua University, Beijing ,China)

Jian-Fei Chen
(Lecturer, Edinburgh University, Edinburgh, UK)

John Michael Rotter

(Professor, Edinburgh University, Edinburgh, UK)

ABSTRACT
The modified compression field theory and an advanced bond-slip model are implemented in a general finite
element analysis package to evaluate the shear behaviour of FRP strengthened reinforced concrete beams. The
inclination angle of the critical shear crack is estimated and the debonding phenomenon is simulated. A close
agreement is achieved between the predicted average FRP strains and those in a test beam reported in the literature.
Further research is being conducted to simulate behaviour of FRP shear the interaction between the external FRP
shear reinforcement and concrete.

KEYWORDS
Shear, FRP, strengthening, concrete, modified compression field theory (MCFT)

1. INTRODUCTION

The modified compression field theory (MCFT) has been an alternative method for shear design of reinforced
concrete members since late 1980s when it was established by Vecchio and Collins (1986). It takes into account the
three basic principles of mechanics and is able to calculate the inclination of the diagonal shear crack, which is
essential in predicting the shear behaviour of reinforced concrete beams.

With the increasing interest in the technique of strengthening RC members with externally bonded FRP for shear, a
number of studies have been carried out to include the contribution of FRP in the MCFT in the last few years. Malek
and Saadatmanesh (1998) extended MCFT to include the contribution of FRP sheets with variable concrete crack
angles. However, they assumed a uniform distribution of FRP strain throughout the depth of the beam and no slip
between the FRP and concrete was taken into account. Lees et al. (2002) analysed the development of strain in FRP
sheets using MCFT based on similar assumptions. Wong et al. (2003) considered the bond-slip behaviour of FRP-
concrete interface in the MCFT model by introducing elastic or perfectly elasto-plastic link elements. They
concluded that it is necessary and viable to model the interface behaviour between FRP and concrete but a more
advanced bond-slip constitutive model must be adopted.

In this paper, the MCFT is implemented in the finite element software MSC. MARC (2003) to simulate the shear
behaviour of FRP strengthened concrete beams. An advanced bond-slip relationship is adopted to model the FRP-
concrete interface. Numerical predictions are compared with test results from the literature.

391
2. THE MODEL
The MCFT adopts a smeared crack model. The bond behaviour between the FRP and the concrete plays a crucial
role and debonding of FRP from concrete almost always happens prior to the final shear failure of a concrete beam
shear strengthened with FRP (Chen and Teng 2003). However, this interfacial bond-slip behaviour cannot be
directly incorporated into the MCFT model when the FRP starts to debond. In this study, the reinforced concrete is
modelled using a constitutive model based on the MCFT. The FRP strips are modelled separately and linked to the
concrete surface by using nonlinear springs based on Lu et al.’s (2005) bond-slip relationship (Fig. 1a).

In the MCFT, the reinforced concrete is treated as a continuous material with the reinforcements and cracks smeared
in the elements. The rotating-angle crack model is commonly used. The in-plane constitutive model in the MCFT is
established based on the uniaxial constitutive models for concrete and steel reinforcement. The widely-used uniaxial
stress-strain relationship for concrete proposed by Hognestad (1952) and that for steel reinforcements (Fig. 1b)
suggested by T. T. C. Hsu and his colleagues (Belarbi and Hsu 1995) are adopted in this study. To introduce the
compression softening effect of concrete, Hognestad’s model is modified here by including a softening coefficient ζ
which was proposed by Belarbi and Hsu (1995). The concrete compressive stress (σ) strain (ε) relationship (Fig. 1c)
is thus given as
⎡ 2ε ⎛ ε ⎞ 2 ⎤
σ = ζσ 0 ⎢ − ⎜ ⎟ ⎥ ε ≤ ε 0 (1a)
⎢⎣ ε 0 ⎝ ε 0 ⎠ ⎥⎦
⎡ ⎛ ε − ε0 ⎞ ⎤
2

σ = ζσ 0 ⎢1 − 0.15 ⎜ ⎟ ⎥ ε0 ≤ ε ≤ εu (1b)
⎢⎣ ⎝ ε u − ε 0 ⎠ ⎥⎦
0.9
where ζ = (1c)
1 + 400ε 1
where σ 0 is taken as the concrete cylinder compressive strength, ε0 and εu are the peak and ultimate compressive
strains respectively, and ε1 is the current principal tensile strain in the concrete.
σ0
τmax s
τ = τ max f
⎡ Es ⎤
Stress

s0
fs = fy ⎢(0.91− 2B) + (0.02 + 0.25B) εs ⎥ k
⎡ ⎛ ε − ε 0 ⎞⎤
Bond stress

⎣ fy ⎦ σ = ςσ 0 ⎢1 − 0.15 ⎜ ⎟⎥
⎝ ε u − ε 0 ⎠⎦
Stress

⎛ s ⎞ ⎣
−α ⎜ −1⎟ ςσ 0
τ = τ max e ⎝ s0 ⎠

⎡ 2ε ⎛ε ⎞
2
⎤
σ = ςσ 0 ⎢ −⎜ ⎟ ⎥
f s = Es ε s ⎢⎣ ε 0 ⎝ ε 0 ⎠ ⎥⎦

s ε ε0 Strain
Slip Strain
(a) FRP-concrete bond-slip (b) Steel bar (c) Concrete in compression
Figure 1: Uniaxial stress-strain relationships

To take into account the effect of the complex stress state of concrete underneath the FRP strips on the FRP-concrete
bond-slip behaviour, the biaxial strength model for concrete proposed by Kupfer (1969) is introduced into the FE
model to modify the concrete tensile strength:
⎛ σ2 ⎞
f t ' = ⎜ 1 − 0.8 ⎟ ft (2)
⎝ fc ⎠
where σ2 is the principal compressive stress in the concrete, and fc and ft are the uniaxial compressive and tensile
strengths of concrete respectively.

392
A procedure for determining the stress state from a given strain state and strain increment was derived within the
framework of MCFT, which makes use of compatibility and equilibrium conditions and the above constitutive
relationships. The procedure was implemented in MARC through the user subroutine HYPELA2.

3. NUMERICAL PREDICTIONS AND COMPARISON WITH TEST RESULTS

RC beam SCU-2-1 shear-strengthened with CFRP U-jackets and its corresponding un-strengthened control beam
S0-2-0 as reported in Tan and Ye (2003) were investigated using the aforementioned FE model. Both beams had a
shear span-to-depth ratio of 2.155 and were tested under 4-point-bending. The beams had a depth of 260mm and
material properties as listed in Table 1. The concrete cylindrical compressive strength fc was taken to be 0.8fcu.
Table 1: Material properties of specimens
Concrete cubic Web steel Longitudinal steel
Specimen FRP strips
compressive strength reinforcement reinforcement
ID
fcu (MPa) ρv fvy (MPa) ρs fsy (MPa) ρf Ef (GPa)
S0-2-0 31.8 0.19% 377 2.9% 395
S-CU-2-1 37.6 0.19% 377 2.9% 395 0.074% 235
1 2 3 4 5
FRP sheets

User defined spring elements

RC element based on MCFT

Figure 2: Finite element model for specimen S-CU-2-1

Due to symmetry, only a half span of the beams was modelled (Figure 2). The RC beams were modelled using the
user-defined 2D RC material model based on MCFT and the FRP strips were modelled as an orthotropic material.
The nodes of the FRP elements were linked to the nodes of the RC element by user-defined nonlinear spring
elements with appropriate properties of the adopted bond-slip model. The lowest row of the FRP nodes was rigidly
linked to the bottom of the beam because no slip was possible between the beam and the FRP U jackets there.

Figure 3 shows a comparison of the distribution of the predicted principal tensile strain with the test diagonal crack
pattern. It is seen that the predicted inclination angle is very close to the test crack angle. Figure 4 shows the
simulated failure process where the FRP strain value is proportional to the darkness of the colour. In the test, FRP
strip No. 4 was debonded first, followed by the debonding of strip No.3 and then No. 2. The beam eventually failed
due to the debonding of FRP strip No. 1. This failure process is closely reproduced as in Figure 4.

The predicted load-deflection curve for the un-strengthened beam specimen S0-2-0 is in good agreement with the
test results (Fig. 5), but that for the FRP strengthened specimen SCU-2-1 has some deviation from the test curve.
The main cause for this disparity may be the inability of the present FE model to simulate factors such as the
enhanced dowel action by the presence of the FRP. Figure 6 shows that the average tensile strain in the FRP strips
along the diagonal crack is close to the test value, but the predicted strain distribution is more uniform compared
with the test results which may be attributed to the adoption of the smeared crack model.

(a) Predicted principal tensile strain distribution (b) Crack pattern in the test beam
Figure 3: Predicted principal tensile strain versus test crack pattern when the middle FRP strip debonded

393
 Debonding
1 2 3 4 5 1 2 3 4 5 1 2 3 4 5 1 2 3 4 5

(a) No debonding (b) Initiation of debonding (c) F3 & F4 debonded (d) F2-5 debonded
Figure 4: Development of axial strain distribution in FRP strips
300 0.007

250 0.006
0.005
200
Load (kN)

Strain
0.004
150
S0-2-0 0.003
100 SCU-2-1 0.002
S0-2-0(FEA) Test
50 0.001
SCU-2-1(FEA) FEA
0 0
0 2 4 6 8 10 12 14 16 1 2 3 4 5
Deflection (mm) FRP sheet No.
Figure 5 Load-deflection curves Figure 6 Strain distribution along the diagonal crack

4. CONCLUSIONS
This paper has presented a study on the shear behaviour of FRP strengthened RC beams. The modified compression
field theory is built into a general-purpose finite element analysis software package. The bond-slip relationship of
FRP-concrete interface is modified by reducing the concrete tensile strength according to the biaxial stress state of
concrete underneath the FRP strips. The results show that the model is capable of predicting the inclination angle of
critical shear crack as well as the debonding procedure of FRP strips, which are both essential in predicting the shear
capacity of FRP strengthened RC beams.

5. ACKNOWLEDGEMENTS
The authors would like to acknowledge the financial support provided by the Royal Society through the Royal
Society-NSFC UK-China Joint Project (Grant No. IS 16657) and the National Natural Science Foundation of China
through a key project for FRP in construction (Project No. 50238030).

6. REFERENCES
Belarbi, A. and Hsu, T.T.C. (1995). “Constitutive laws of softened concrete in biaxial tension-compression”. ACI
Structure Journal, Vol. 92, No. 5, pp562-573.
Chen, J.F. and Teng, J.G. (2003). “Shear capacity of FRP strengthened RC beams: FRP debonding”. Construction
and Building Materials, Vol.17, No.1, pp27 – 41.
Hognestad, Eivind (1952). “Inelastic behaviour in tests of eccentrically loaded short reinforced concrete columns”.
ACI Journal, Vol. 49, No. 10, pp117-139.
Kupfer, H., Hilsdorf, H. K., Rush H. (1969). “Behavior of concrete under biaxial stresses”. ACI Journal, Vol. 66, No.
8, pp656-666
Lu, X.Z., Teng, J.G., Ye, L.P. and Jiang, J.J. (2005). “Bond-slip models for FRP sheets/plates externally bonded to
concrete”. Engineering Structures. Vol. 27, No. 6, pp938-950.
Lees, J.M., Winistorfer, A.U., Meier, U. (2002) “External prestressed carbon fiber-reinforced polymer straps for
shear enhancement of concrete.” Journal of Composites for Construction, Vol. 6, No. 4, pp249-256.
Malek, A.M., Saadatmanesh, H. (1998). “Ultimate shear capacity of reinforced concrete beams strengthened with
web-bonded fiber-reinforced plastic plates.” ACI Structure Journal, Vol. 95, No. 4, p391-399.
Tan, Z. and Ye, L.P. (2003) “Experimental research on shear capacity of RC beam strengthened with externally
bonded FRP sheets”. China Civil Engineering Journal, Vol. 36, No. 11, pp12-18
Vecchio, F.J. and Collins, M.P. (1986). “The modified compression-field theory for reinforced concrete elements
subjected to shear”. ACI Journal, Vol. 83, No. 2, pp219-231.
Wong, Rita S.Y. and Frank J. Vecchio. (2003). “Towards modeling of reinforced concrete members with externally
bonded fiber-reinforced polymer composites." ACI Structure Journal Vol. 100, No. 1, pp 47-55.

394
Third International Conference on FRP Composites in Civil Engineering (CICE 2006)
December 13-15 2006, Miami, Florida, USA

NUMERICAL SIMULATION OF BOND DETERIORATION BETWEEN

CFRP PLATE AND CONCRETE IN MOISTURE ENVIRONMENT
Zhenyu Ouyang
(Ph.D. Student, Marquette Universit, Milwaukee, WI, USA)

Baolin Wan
(Assistant Professor, Marquette Universit, Milwaukee, WI, USA)

ABSTRACT
This research investigated the effect of interface region relative humidity (IRRH) on the bond between CFRP and
concrete using computer simulation technique. After FRP is peeled off concrete substrate, there is a layer of residue
concrete on the detached FRP if the bond is sound. The experimental program related to this research found that the
residual thickness of concrete (RTC) was directly related to the IRRH. A constitutive equation, which calculated the
RTC from IRRH, was proposed in this research based on the experimental data. The calculated RTCs were used to
build a series of finite element models and the virtual crack closure technique (VCCT) was used to calculate the
bond fracture energy Gf for debonding. Through the RTC, the bond fracture energy Gf was related to the interface
region relative humidity (IRRH). The model and FEM results had excellent agreement with the experimental data.
These models can be used to simulate the bond degradation for CFRP bonded concrete specimens suffering from
moisture attack.

KEYWORDS
Bond, CFRP, Deterioration, Moisture, Numerical simulation

1. INTRODUCTION
Externally bonding fiber reinforced polymer (FRP) composite materials to concrete beams in order to strengthen or
rehabilitate structures is receiving worldwide attention and application. However, many experimental studies
(Karbhari et al.,1997; Nguyen et al., 1998; Wan et al., 2006; Ouyang and Wan, 2006) conducted in the past decade
show that the water can seriously deteriorate the bond between FRP and concrete. Normally, the fracture mode
shifts from cohesive failure in concrete to adhesive failure in interface, where it is the weakest part of bonded
structures, in moist environment. Since the vast majority of advanced structural adhesives are epoxy based, they
have the propensity to absorb moisture, which can lead to undesirable changes in strength, stiffness and interfacial
adhesion. Therefore, the bond durability in moist environment is one of the most important issues for extensive
field application of FRP repairing technique in future. Although a lot of experimental researches have been
conducted, the mechanism for bond deterioration due to moisture attack is still not clear and how to model such
deterioration is still a problem needed to be solved.

Fracture mechanics has been widely used to study the debonding phenomena. The fracture mechanics parameter,
fracture energy Gf, includes both crack length and load information for a specimen with crack. Therefore, it is a
good criterion to study the bond performance of the FRP bonded concrete members. Besides of the material
properties, the critical fracture energy is also considered as an important parameter for modeling FRP debonding
from concrete (Coronado and Lopez, 2005). After FRP is peeled off concrete substrate, there is a layer of residue
concrete on the detached FRP if the bond is sound. The experimental program related to this research (Ouyang and
Wan, 2006) found that the residual thickness of concrete (RTC) remaining on the FRP after it was peeled off
concrete substrate was directly related to the interface region relative humidity (IRRH). It was also found that bond
fracture energy Gf of FRP-concrete system was directly related to the RTC. This research proposed a constitutive
equation and built a series of finite element models to connect IRRH, RTC and Gf.

395
2. RESEARCH SIGNIFICANCE
A constitutive equation, which calculated the RTC from IRRH, was proposed in this research based on the
experimental data. The calculated RTCs were used to build a series of finite element models to calculate the bond
fracture energy Gf for debonding. Through the RTC, the bond fracture energy Gf was related to the interface region
relative humidity (IRRH).

3. NUMERICAL MODELS
3.1 Constitutive Equation to Calculate RTC from IRRH

As observed in experimental program (Ouyang and Wan, 2006), the fracture energy Gf of control specimens was not
significantly affected by the residual thickness of concrete (RTC) in dry environment. After the specimens were
submerged in water for different durations, RTC decreased with the increase of interface region relative humidity
(IRRH). In this situation, even small change of RTC could cause significant decrease of bond fracture energy due to
the moisture at the interface region.

When FRP was peeled off concrete substrate, the crack tip actually located at the place with a distance of RTC from
the nominal interface bond line as shown in Figure 1. Since the bond fracture energy Gf is sensitive to RTC after
specimens are attacked by moisture, the deviation of crack tip location from nominal bond-line should not be
neglected in the debonding analysis models. It is shown in Figure 2 that RTC does not change significantly when the
interface is relatively dry. However, it decreases proportionally with the increase of IRRH after IRRH is higher than
certain value. Therefore, a bi-linear relation between RTC and IRRH is proposed as Equation 1.
⎧ IRRH − IRRHc
⎪RTCc − ( ) × RTCc For IRRH ≥ IRRHc (1)
RTC = ⎨ IRRH0 − IRRHc
⎪RTC For IRRH < IRRHc
⎩ c

where IRRHc is the critical value of IRRH to start decreasing the RTC; RTCc is the constant value of RTC when
IRRH is less than the IRRHc; and IRRH0 is the IRRH value when perfect interface adhesive failure happens (RTC =
0). This equation assumes that the change of RTC will cause the change of the fracture energy Gf only when IRRH
is greater than IRRHc.

Peel force 2.5

Bond-line
2
RTC (mm)

1.5
FRP layer
1

Crack tip Adhesive 0.5

RTC
Concrete
Substrate layer
0
Crack length a 45% 55% 65% 75% 85%
IRRH

Figure 1: Crack tip location determined by RTC Figure 2: RTC vs. IRRH

The RTC might be different for different loading modes and material properties. All specimens were tested by Mode
I (peeling) loading in this study. It is shown in Figure 2 that IRRHc was 55% and RTCc was 2.0 mm for the
specimens in this research. It can also be found in the figure that IRRH is 77% if the line is extended to intercept
with the IRRH axis where the RTC is 0. Thus, IRRH0 was set to be 77% in this study. Equation 2 was used to
calculate the RTC for finite element models in this study.
⎧ IRRH − 0.55
⎪2.0 − (
RTC = ⎨
) × 2.0 For IRRH ≥ 55% (2)
0.77 − 0.55
⎪2.0 For IRRH < 55%
⎩

396
3.2 Virtual Crack Closure Technique (VCCT)

The virtual crack closure technique (VCCT) is widely used to calculate the interfacial energy release rates from
finite element models through a single geometric model for crack propagation. A modified virtual crack closure
integral for square-root singularity elements was derived by Sethuraman and Maiti (1988) in order to compute the
strain energy release rate by moving the mid-side node position of the isoparametric quadratic element to the quarter
location. This method can increase the accuracy of strain energy calculation for linear elastic analysis. The strain
energy release rate, G, can be expressed as,
(u yk − u ′yk )
GI = [Fyj + (1.5π − 4)Fyi ] (3)
∆a
(u − u ′xk )
G II = xk [Fxj + (1.5π − 4)Fxi ] (4)
∆a
where (Fxi, Fyi) and (Fxj, Fyj) represent the nodal forces at nodes I and J, respectively; and (uxk, uyk) and (u'xk, u'yk)
represent the nodal displacements behind it as shown in Figure 3. The total energy release rate is the sum of GI and
GII. In this study, the modified virtual crack closure integral was used to calculate the bond fracture energy Gf of
FRP bonded concrete specimen.

Nodal displacements Node I forces

behind the crack tip (Fxi, Fyi)
(uxk, uyk)

Crack
Nodal displacements Node J forces
behind the crack tip (Fxj, Fyj)
(u'\xk,u'yk)

Figure 3: Nodes and elements generated around the crack tip

in FE model for energy release rate analysis.

Figure 1 was used to build the geometric model in FE analysis. The dimension of the concrete specimen was 76 x 76
x 191 mm. The width of CFRP plate was 51 mm and the thickness of it was 2 mm. The ultimate tensile strength and
tensile modulus in the principal fiber direction were 2.02 GPa and 139 GPa, respectively. The tensile modulus and
tensile strength of the epoxy adhesive used in the study were 3.18 GPa and 72.4 MPa respectively. The crack length
was set as 33 mm which was the typical first crack length in the tests. The load recorded in the test was applied to
the FE model. The RTC, which is the distance from nominal bond line to the crack in FE models, was calculated by
Equation 2 for different IRRH values. Ansys 10.0 was used to do finite element analysis and Plane 183 element was
selected to mesh the geometric model. All materials were assumed to be linear elastic. It was found that materials’
moduli did not have significant effect on Gf for linear elastic analysis (Wan and Ouyang, 2006). Therefore, the
material properties change due to moisture attack was not considered in this research.

4. NUMERICAL RESULTS AND DISCUSS

The FEM and experimental results of bond fracture energy Gf is plotted versus residual thickness of concrete as
shown in Figure 4. It is shown that the FEM results had excellent agreement with the experimental data. The Gf
value decreased with the decrease of RTC value. When the fracture location was moved closer to the bond line
(RTC = 0), Gf was significantly lower than that of control specimens. This phenomenon is consistent with the
common experimental result that adhesive failure has lower bond strength than cohesive failure. It can also be seen
in Figure 4 that slope of the trend line changed around the RTC value of 1.5 mm. When RTC was less than 1.5 mm,
the Gf value was very sensitive to the change of RTC value. However, the increase of RTC did not result the
significant increase of Gf when RTC was greater than 2 mm.

The relation between bond fracture energy and interface region relative humidity is shown in Figure 5. When IRRH
was smaller than 55%, there was no significant change in fracture energy. The FEM result with RTC value of 2.0
mm showed good approximation for this stage. When IRRH was larger than 55%, the reduction of fracture energy

397
with the increase of IRRH was very obvious. As shown in Figure 5, the FEM results agreed well with the
experimental data although the Gf values calculated from FEM were slightly larger than the corresponding
experimental results. This was due to the local nonlinearity at the crack tip while the finite element model was based
on linear elastic fracture mechanics. Figure 5 shows a nonlinear relation between IRRH and fracture energy
although the linear relation was assumed between RTC and IRRH in Equations 1 and 2. When the interface was
relatively dry (IRRH < 55%), the Gf kept relatively constant. Gf value decreased significantly with the increase of
IRRH when it was higher than 55%.

600 600
Experimental FEM
500 500
FEM Expermental
400 400

Gf (J/m2)
Gf (J/m2)

300 300

200 200

100 100

0 0
0 0.5 1 1.5 2 2.5 45% 55% 65% 75% 85%

RTC (mm) IRRH

Figure 4: Gf vs. RTC Figure 5: Gf vs. IRRH

5. CONCLUSION
A constitutive equation was proposed in this research to calculate the residual thickness of concrete (RTC) from the
value of interface region relative humidity (IRRH). Using this equation and finite element models, the relation
between bond fracture energy and the IRRH of the FRP bonded concrete specimens was successfully established.

6. REFERENCE
Coronado, C.A. and Lopez, M.M. (2005). “Modeling of FRP-concrete bond using nonlinear damage mechanics”,
Proceedings of the 7th Conference on Fiber Reinforced Polymers for Reinforced Concrete Construction
(FRPRCS7), Kansas City, U.S., pp 411-426.
Karbhari, V.M., Engineer, M. and Eckel II, D.A. (1997). “On the Durability of composite rehabilitation schemes for
concrete: use of a peel test”, Journal of Material Science, Vol. 32, 1997, pp147-156.
Kinloch, A.J. (1987). Adhesion and Adhesives - Science and Technology, Springer, Germany.
Nguyen, T., Byrd, W. E., Alsheh, D., Aouadi, K., Chin, J. W. (1998). “Water at the polymer/substrate interface and
its role in the durability of polymer/glass fiber composites”, Durability of Fibre Reinforced Polymer (FRP)
Composites for Construction (CDCC'98), 1st International Conference, Canada, pp451-462.
Ouyang Z. and Wan. B. (2006). “Deterioration mechanism of bond between CFRP plate and concrete in moisture
environment”, Third International Conference on FRP Composites in Civil Engineering (CICE 2006), Miami,
Florida, USA.
Sethuraman, R. and Maiti, S.K. (1988). “Finite element based computation of strain energy release rates by modified
crack closure integral,” Engineering Fracture Mechanics, Vol. 30, pp 227–231.
Wan, B., Petrou, M.F. and Harries, K.A. (2006). “Effect of the presence of water on the durability of bond between
CFRP and concrete”, Journal of Reinforced Plastics and Composites, Vol. 25, No. 8, pp875-890.
Wan. B. and Ouyang Z. (2006). “Finite element analysis of FRP debonding from concrete undergoing global mixed
mode I/II loading”, Third International Conference on FRP Composites in Civil Engineering (CICE 2006), Miami,
Florida, USA.

398
Third International Conference on FRP Composites in Civil Engineering (CICE 2006)
December 13-15 2006, Miami, Florida, USA

NUMERICAL STUDY ON MASONRY LOAD BEARING WALLS

RETROFITTED WITH GFRP COMPOSITES
Prakash, S.S
(Graduate Research Assistant, University of Missouri Rolla, Rolla, Missouri, USA)

Alagusundaramoorthy, P
(Associate Professor, Indian Insitute of Technology Madras, Chennai, Tamil Nadu,India)

ABSTRACT
The failure of masonry buildings due to earthquakes is because of its low tensile and shear resistance. To improve
the strength and enhance the seismic performance of masonry structures, retrofitting using fiber reinforced polymer
(FRP) composites is explored in this study. Characterization of materials and finite element analysis (FEA) are
carried out. Homogeneous macromodeling using the smeared crack concrete model is adopted for the analysis of
load bearing brick masonry walls under inplane loading. The resistance of the masonry walls is studied for a shear
specimen with an aspect ratio (length/height) of one and a flexure specimen with an aspect ratio of two. Effects of
different strengthening techniques like X-bracing, full-surface-bonding, and two-end-strap coating using FRP
composites are studied using the developed model based on the failure pattern of control specimens. Parametric
studies are also carried out to study the effect of vertical compression, and the results are discussed. It is concluded
that the effectiveness of FRP in increasing the lateral strength of masonry is based on the compressive strength of
masonry.

KEYWORDS
Masonry, load bearing walls, FRP composites, retrofitting, finite element modeling

1. INTRODUCTION
Brick masonry is a heterogeneous material with orthotropic mechanical characteristics that depend not only on the
properties of the brick unit and mortar but also on their interaction (Lourenco, 1996). In general, models for masonry
are categorized into homogeneous models and heterogeneous models. In a homogeneous material model, the
behavior of masonry is modeled by a fictitious material having properties that are equivalent to the material behavior
of brick masonry. In the heterogeneous model, the bricks and mortar are modeled through their respective
constitutive laws with the interaction in between them (Page, 1978). Both homogeneous and heterogeneous models
can be based on either smeared crack or discrete crack approaches (Lofti and Shing, 1991). However, the latter
involves large computational effort. Giordano et al (2002) studied the behavior of masonry with macromodeling
techniques using a smeared crack concrete model and found good correlation with the experimental results. In this
study, macro modeling is adopted for the analysis of load-bearing masonry due to its simplicity and practical
significance. The effect of different strengthening techniques like X-bracing, full-surface-bonding, and two-end-
strap coating using FRP composites on the behavior of masonry load bearing specimens are studied, and the results
are discussed.

2. CHARACTERIZATION OF MATERIALS
Tests were conducted to determine the strength and elastic properties of bricks, cement mortar cubes, brick masonry
wallettes, and shear triplets. The material parameters for FRP composites were found through testing of coupons for
tension, shear, and flexure. A FE model of a masonry load bearing wall is developed and calibrated with the
experimental test data of a load bearing wall subjected to a vertical compression of 0.55 N/mm2.

3. NUMERICAL STUDY
The smeared crack concrete model available in ABAQUS/Standard uses the concepts of oriented damaged elasticity
(smeared cracking) and isotropic compressive plasticity to represent the inelastic behavior of concrete. The model is
399
defined by using the uniaxial stress-strain relations, tension stiffening data from the uniaxial tension test results, and
optionally by defining the shear retention and the failure ratios options. This model can be used for plain concrete or
masonry, even though it is intended primarily for the analysis of reinforced concrete structures.

3.1 Material Properties for Modeling

Material properties of brick masonry are obtained from the testing of brick masonry wallettes of size 400 mm x 400
mm under compression. The modulus of elasticity for the numerical study is calculated from the test data as
168 N/mm2 in the direction normal to the bed joints and 405 N/mm2 parallel to the bed joints. The Poisson’s ratio for
the masonry material is normally around 0.15 to 0.2, and it is taken as 0.2. For the plastic part of the model, the
parameters are taken from the nonlinear part of the uniaxial compression stress-strain curve (stress and
corresponding absolute plastic strain). Tension stiffening data is the post cracking behavior under uniaxial tension,
and the parameters for input are a fraction of the remaining stress to stress at cracking and the corresponding
absolute value of the direct strain minus the direct strain at cracking. FRP material used for retrofitting is glass fiber
woven roving mat (WRM) with an area density of 360 gsm. FRP composite is modeled assuming elastic behavior
through lamina option (Hibbit et al, 2002). The tensile strength of the laminate was determined as per procedures
given in BS 2782 (1996). The average modulus of elasticity in tension is found to be Ex=12984 N/mm2 in the warp
direction and Ey=11500 N/mm2 in the weft direction. The Poisson’s ratios υxy and υyx are found to be 0.126 and
0.112 respectively. Perfect bond is assumed to exist between the FRP material and the masonry wall surface. FE
studies assume no debonding of the FRP from masonry. It is assumed that premature debonding failure of FRP will
be prevented by suitable anchorage of FRP composites with masonry walls.

3.2 Description of Finite Element Model

In the test setup, the wall is placed on a rigid concrete foundation beam, which is connected to the laboratory floor
by high strength steel anchored bolts. The vertical compression is applied through pretensioned high strength steel
strands. The cyclic or montonic lateral load is applied to the side of the loading beam by an actuator anchored
against a strong wall. Brick masonry wall panels of sizes 2500 mm x 2500 mm x 230 mm and 1250 mm x 2500 mm
x 230 mm are used as shear and flexural specimens in FE study. The shear specimen is supported on a concrete
beam foundation of size 2700 mm x 230 mm x 200mm. The vertical and horizontal loading is applied on the top
face of the wall. The flexural specimen is supported on a concrete foundation of size 1450 mm x 2500 mm x 240
mm. In the retrofitted and unretrofitted specimens, the concrete beams and the wall are modeled by eight noded
linear continuum elements (C3D8R) with reduced integration and hourglass control. The FRP material is modeled
using 3D shell elements (S4R) with both rotation and translations at each node. All specimens are loaded laterally on
the top face of masonry wall by giving linearly increasing displacements. The vertical compression load is applied
through different pressure loading of magnitudes 0.3 N/mm2, 0.55N/mm2, 1.0 N/mm2 and 2.0 N/mm2 on the top face
of the wall. The interface between the bottom foundation beam and the wall surface is modeled using surface-based
tie constraints. The bottom of the foundation beam is given a fixed-end condition by arresting all the degrees of the
freedom.

4. RESULTS AND DISCUSSION

The smeared crack concrete model adopted is verified with experimental data of a load bearing wall subjected to a
vertical compression of 0.55 N/mm2. The prediction using the macromodel was found to be accurate enough when
the uncertainties associated with the derived material properties and the advantages of reduction in the
computational cost compared to a detailed micromodel are taken into account. The FE result underestimated the
stiffness of the test specimen but the maximum load resistance predicted is in close agreement (Fig. 1). The failure
of the test specimen was by diagonal cracking, and the same was observed in FE analysis. After the validation of the
FE-model, as a parametric study, two types of load bearing walls (namely, flexure and shear specimens) are
analyzed for different constant vertical compressions with monotonic lateral loading. Based on the results of the FE
analysis, two types of retrofitting strategies are suggested for both the specimens (Fig. 2). The shear specimen is
retrofitted using (i) X-bracing of FRP strips and (ii) a full-surface bonding of FRP composites. The flexural
specimen is retrofitted using (i) two-end-strap coating and (ii) a full-surface bonding of FRP composites. After the
retrofitting, the increases in load resistance and displacement levels are calculated and compared with the behavior
of the control specimen. The effect of vertical compression with lateral loading on the load resistance is also studied.

400
 250

Load Resistance (kN)

200

150

100

50 Experiment
FE Analysis
0
0 2 4 6 8 10 12 14
Displacement (mm)

Fig. 1 Comparison of FE Results of a Load Bearing Wall with Experimental Data

Fig. 2 (a) Full Surface Bonding for Shear Specimen (b) X-Bracing for Shear Specimen (c) Full-Surface
Bonding for Flexure Specimen (d) Two-End-Strapping for Flexure Specimen

4.1 Shear Specimen

In the shear specimen, strut action along the compression diagonal is observed. The load resistance increases
linearly up to 26.8 kN (Fig. 3a). Thereafter, the load shedding is observed. With increase in the vertical
compression, the ultimate load resistance also significantly increases. In the case of vertical compression of
1.0 N/mm2, the load resistance increased from 26.8 kN to 160 kN. At vertical compression of 2.0 N/mm2, the
ultimate load resistance observed is 128 kN because of the compression failure of masonry. The ultimate load
resistance in retrofitted shear specimen with X-bracing, increased from 26.8 kN to 94 kN (Fig. 3d). The first
cracking load also increased from 25 kN to 44 kN. Significant increase exists in the load resistance for the X-

401
Bracing under vertical compression (Fig. 3b). For the vertical compression of 2.0 N/mm2, the increase in the load
resistance is linear up to the final compression failure of the masonry. The ultimate load resistance increased from
26.8 kN to 271 kN (Fig. 3d). A gradual increase happens in the load resistance in the case of full-surface bonding
due to the presence of vertical compression, unlike retrofitting with X-Bracing (Fig. 3c). In the case of a vertical
compression of 2.0 N/mm2, the full-surface bonding of FRP prevented the compression failure of masonry and
increased the ultimate deformation of masonry up to 60 mm, unlike in X-Bracing.

(a) (b)

(c) (d)
Fig. 3 Load Displacement Curves for (a) Control Shear Specimen (b) Retrofitted Specimen with X-Bracing
(c) Retrofitted Specimen with Full-Surface Bonding and (d) Comparison of X-Bracing and Full-Surface
Bonding with Control Shear Specimen

4.2 Flexure Specimen

The flexural specimen is retrofitted using (i) two end strap coating and (ii) a full surface bonding of FRP
composites. For the flexural specimen without vertical compression the load resistance linearly increased up to 8 kN
with corresponding displacement of 10 mm (Fig. 4a). Load shedding is observed after this load. The ultimate load
resistance increased from 8 kN to 40 kN for the increase in vertical compression from 0 to 1.0 N/mm2. In the case of
retrofitted flexural specimens with two end straps, the load resistance increased linearly up to a load of 18 kN with
the corresponding displacement of 10 mm (Fig. 4b). Due to retrofitting with end straps the ultimate resistance
increased from 18 kN to 36 kN, corresponding to the ultimate displacement of 48 mm for the unretrofitted specimen
(Fig. 4d). For two-end strapping, due to vertical compression, a gradual increase happens in the load resistance as
well as in the cracking displacement. In the case of the retrofitted flexural specimen with full-surface bonding of
FRP, the load displacement response is stiffer compared to the control specimen, and the ultimate load resistance
increased from 8 kN to 58 kN (Fig. 4d). Due to vertical compression, a gradual increase happens in the load
resistance as well as in the cracking displacement (Fig. 4c). The increase in the ultimate displacement (60 mm) was
higher compared to that of the control flexure specimen, which had ultimate displacement of 46 mm. This result
shows that full-surface retrofitting with FRP increases the ductile behavior of the flexure specimen.

402
 (a) (b)

(c) (d)
Fig. 4 Load Displacement Curves for (a) Control Flexure Specimen (b) Retrofitted Specimen with Two-End
Straps (c) Retrofitted Specimen with Full-Surface Bonding and (d) Comparison of Two-End Strapping and
Full-Surface Bonding with Control Flexure Specimen

5. SUMMARY AND CONCLUSIONS

Two types of load bearing wall specimens (namely, flexure and shear) were analyzed for different constant vertical
compressions with monotonic lateral loading. Based on the analysis, two types of retrofitting strategies were
suggested for both the specimens. The flexural specimen was retrofitted using (i) two-end strap coating and (ii) a
full-surface bonding of FRP composites. The shear specimen was retrofitted using (i) X-bracing and (ii) a full
surface bonding of FRP composites. Retrofitting using FRP composites was observed to increase the load at first
crack and the ultimate load. For the flexure specimen, a significant change was noticed in ultimate load resistance
when retrofitted with two-end straps and full-surface bonding. However, the increase in ultimate displacement was
high in case of full-surface bonding, thereby increasing the ductile behavior of the specimen. For shear specimen,
the ultimate load resistance increased from 26.8 kN to 271 kN, and the increase in ultimate displacement was from
50 mm to 60 mm for the shear specimen retrofitted with full surface bonding. Based on the results, the following
major conclusions are drawn: (i) FRP composites significantly increase the load at first crack, ultimate load and
deformational capacity, (ii) The effectiveness of FRP in increasing the lateral strength of masonry depends on
compressive strength of masonry, and (iii) Small amounts of vertical compression significantly increase the load at
first crack and ultimate load.

6. REFERENCES
1. Lourenco, P.B. (1996). “Computational strategies for masonry structures.” TU-Delft, ISBN 90-407-1221-2.
2. Page, A.W. (1978). “Finite element model for masonry.” Journal of structural division, 104(8), 1267-1285.
3. Lofti, H.R., Shing, P.B. (1991). “An appraisal of smeared crack models for masonry shear wall analysis”,
Computers and Structures 41(3), 413-425.
4. Giordano, A., Mele, E. and De Luca, A. (2002). “Modeling of historical masonry structures: comparison of
different approaches through a case study”, Engineering structures, 24(2002), 1057-1069.
5. Hibbit, D., Karlson, B. and Sorensen, P. (2002). ABAQUS/ Standard theory manual.
6. BS 2782-10: Method 1006. Methods of testing plastics, glass reinforced plastics: determination of tensile
properties, 1996”.
403
404
Third International Conference on FRP Composites in Civil Engineering (CICE 2006)
December 13-15 2006, Miami, Florida, USA

MODELLING OF RC SQUARE HOLLOW PIERS

WRAPPED WITH CFRP
Gian Piero Lignola, Andrea Prota, Gaetano Manfredi and Edoardo Cosenza

(Dept. of Structural Analysis and Design, University of Naples Federico II, Naples, ITALY)

ABSTRACT
A large number of concrete bridges built in Europe were characterized by Reinforced Concrete (RC) hollow piers;
many of them are now in need of a seismic upgrade to improve their response under severe earthquakes. Researchers
have mainly focused their attention on solid piers; few studies have been done about hollow cross sections. To study
the behaviour of rectangular hollow piers subjected to combined axial load and bending, a total of 7 specimens have
been tested. The strengthening was performed by Carbon Fiber Reinforced Polymers (FRP) wrapping. Planned tests
have allowed to improve the knowledge of unstrengthened hollow rectangular piers and to provide a contribution
towards the comprehension of the resistant mechanisms of hollow piers in presence of FRP confinement. The
experimental results and the related database has been used for validation of CFRP strengthening design methods for
hollow cross sections and for assessment of potential extension of the current models available for solid sections to
non-circular hollow sections. Strengthened and unstrengthened members have been numerically modelled. The
results of the present work suggest that a reliable numerical procedure to predict hollow cross section behaviour
under combination of flexure and compression should include appropriate models for compressed bars buckling and
concrete cover spalling.

KEYWORDS
Concrete columns, Buckling, FRP confinement, Hollow sections, Modelling

1. INTRODUCTION
To maximize structural efficiency of the strength-mass and stiffness-mass ratios and to reduce the mass contribution
of the pier to seismic response and high carrying demand on foundations, it is desirable to use hollow cross sections.
However modern codes of practice oriented to new design do not recognize any specific problem related to hollow
sections. A large number of existing bridges have hollow piers. Many of them are now in need of a seismic upgrade
but few researches have been done on hollow columns strengthened with Fiber Reinforced Polymers (FRP). Hollow
reinforced piers may be required to dissipate energy by forming ductile plastic hinges when they are subjected to
seismic forces while, due to its brittle nature, failure in shear of Reinforced Concrete (RC) bridge piers has in any
case to be avoided. Research programs have mainly addressed the behaviour of solid columns with either
rectangular or circular cross sections and have shown how FRP confinement can play a significant role in
constraining the concrete core and consequently improving flexural strength and ductility and adding additional
reinforcement in the hoop directions, increasing also shear strength, such reducing seismic vulnerability. It can be
seen that jacket thicknesses for shear, bar buckling restraint and lap splice clamping are driven by the modulus of the
jacket in the hoop direction, which favours higher modulus materials, whereas the flexural plastic hinge confinement
can also efficiently be achieved with a lower modulus material. Simple calculations procedures for the assessment of
strength and deformations capacity of existing bridge have been proposed in last years, but their reliability has been
checked mainly on columns with solid sections; their extension to those hollow (Fam and Rizkalla 2001), especially
rectangular, has been much less investigated.
Fam et al. (2005) tested under flexure rectangular concrete-filled FRP tubes with inner void and fibers oriented in
several directions (not only transverse). No reinforcement was provided other than the outer FRP shell. Buckling of
the FRP not only reduced the effectiveness in carrying compression forces, but eliminated any partial confinement
effect on the thin compression concrete inward buckling and cracking.

405
2. RESEARCH OBJECTIVES
This paper focuses on the upgrade and retrofit of existing RC piers with rectangular hollow cross section using FRP
composite materials applied in the transverse direction (only confinement) to enhance flexural strength and ductility.
The research aimed at investigating the behaviour of unstrengthened and FRP jacketed square hollow piers subjected
to combined axial load and bending un-coupled from shear; therefore, slender specimens were investigated. The
capability of external wrapping using FRP materials to change the failure mode was checked and modelled.

3. SUMMARY OF THE EXPERIMENTAL PROGRAM

A total of 7 hollow square cross section concrete columns were tested. Test specimens reproduce in scale 1:5 typical
bridge piers (Fig. 1a). The test matrix was designed in order to assess the FRP wrapping effectiveness in
correspondence of three P/M ratios. Accordingly three unstrengthened specimens (U1, U2 and U3) and other three
strengthened with CFRP laminates (S1, S2 and S3) were tested with a load eccentricity and kept constant during
each test. A total of 2 plies of CFRP wet lay-up unidirectional fabric (600 gr/m2) have been applied in the transverse
direction in all strengthened specimens for the entire specimen height. The number of installed plies was considered
an upper limit that could be derived from an economical and technical analysis, also accounting for the scale
reduction. However jacket thickness modestly affects the response of non-circular sections (Fam et al. 2005). One
unstrengthened specimen under pure axial load was also tested. Details on the experimental program and outcomes
can be found in (Lignola et al. 2006), herein only results relevant for the numerical comparison will be summarized.

4. REFINED ANALYSIS OF THE CROSS SECTION BEHAVIOUR

Through the use of a fiber model that meshes the concrete cross-sectional geometry into a series of discrete strips,
sections of completely arbitrary cross-sectional shape (including hollow prismatic cross sections) can be modeled.
Each discrete strip is assumed to have a constant stress. For a specified neutral axis location and a specified section
curvature (given a reference strain in a given point of the cross section and with neutral axis known) internal section
forces are computed (Fig.1b).
0

60

120

180

240

300

a) b) 360

Figure 1: Cross Section Geometry (a), Cross Section meshing and forces equilibrium (b)

The neutral axis position is changed until the net internal axial force in the section is in equilibrium with the
externally applied axial load acting on the section, then the internal section flexural moment is computed and the
corresponding moment curvature diagram is plotted. Perfect bond is assumed at the interfaces between concrete and
steel reinforcing bars; plate buckling (as local buckling of a thin compression flange) is not accounted, so that the
ultimate strength is generated by material failure and/or steel reinforcement bar buckling. Tension stiffening effect,
compressed bars buckling, concrete cover spalling and FRP confinement of concrete are included in the model.
When steel reinforcement reaches in compression the buckling stress, as it pushes outward surrounding concrete, the
concrete cover spalls out. The spalling of concrete cover and the buckling of the reinforcement is taken into account
by not considering a portion of the concrete cover area in the equilibrium. In the case of member wrapped with
CFRP the steel bars push internal concrete cover in the inward direction (in the hollow part) when they start
buckling and in the numerical model concrete cover spalling has been simulated by discarding a portion of the
internal concrete cover area.

406
4.1 Material Consitutive Laws

The numerical method uses nonlinear stress-strain relationships for concrete and steel. A reliable stress-strain
behaviour of concrete is necessary particularly when a member is subjected to combined bending and axial load.
The theoretical analyses have been carried considering four constitutive models for concrete. For the unstrengthened
members the Mander model (1988) and the Model Code (1990) were adopted for concrete combined with the size
effect theory after Hillerborg (1989). To simulate the effect of the FRP confinement a simplified approach was used
to obtain a preliminary assessment of the behaviour of the FRP confined members. Spoelstra-Monti model (1999)
and the model recommended by the CNR Italian Guidelines (2004) have been considered. The numerical stress-
strain model (after Cosenza and Prota 2006) was used for the reinforcing steel bars under compression. Analyzing
the failure mode of eccentrically loaded columns, the more compressed wall is considered confined by FRP. The
transverse dilation of the compressive concrete walls stretches the confining device, which along with the other
restrained walls applies an inward confining pressure. The effective pressure fl’ is reduced by a reduction factor keff
due to the so-called arching effect in the wall. The reduction factor is given by the ratio of the effective confinement
area, the core of the walls, to the total area of concrete enclosed by the FRP jacket; in the present case is keff=0.69
and confinement is evaluated on an equivalent circular column of a diameter D equal to the average (longer) side
length (D≈300mm). The ultimate compressive axial strain of FRP-confined concrete is considered to be attained
when lateral strain is equal to FRP ultimate strain; however experimental evidence shows that FRP failure did not
occur. It is noted that the Spoelstra-Monti model is rather more conservative than CNR. This algorithm allowed to
draw the theoretical P-M interaction diagram (Fig. 2a), moment-curvature diagrams (Fig. 2b) and theoretical strain
development during load history.

Strengthened
U0 S2 CNR Model
3000
Strengthened 200
CNR Model
P [kN]

U1 U2
M [kNm]
S1 EI
2000
Unstrengthened
Unstrengthened Model Code 1990
Model Code 1990 100
S2
U2 Experimental
1000
Sectional Analysis
S3
U3 Member FEM Analysis

0 0
a) 0 100 M [kNm] 200 b) 0.00E+00 1.00E-05 χ [mm-1] 2.00E-05 3.00E-05

Figure 2: P-M interaction diagram and experimental outcomes (a), Flexural Moment vs. Curvature Diagram
(experimental U2/S2 and theoretical e=200mm) (b)

4.2 Experimental-Theoretical Comparison

The developed refined methodology with the selected material models and the assumptions made seems to predict
both the sectional strength and deformability of the hollow columns. The strength increase of FRP confined columns
with respect to the unstrengthened columns is about 15%. Wrapping has delayed buckling of steel longitudinal
reinforcement and concrete cover spalling, allowing for the full development of the load capacity of the concrete.
The beneficial effect of confinement increases as the compressed part of the cross-section increases. The strength
increase is about 7% in the case of larger eccentricity and 19% in the case of smaller eccentricity (see also P-M
diagram in Fig. 2a), while a gradually more ductile behaviour when increasing the eccentricity has been observed.
The bar buckling brittle mechanism acting together with internal concrete cover spalling has been modelled and the
predicted post peak behaviour is consistent with the experimental observation. The ascending branch is very well
predicted by numerical model, and curve drop was generally attained (in the numerical analysis the concrete cover
spalls and pier response deteriorates when compressed steel reinforcement reaches buckling strain).

5. FEM ANALYSIS OF THE MEMBER BEHAVIOUR

One of the aims of this study was to develop finite-element models (FEM) that could simulate the behaviour of
tested specimens, evaluate and confirm the FRP confinement effect. An accurate analysis has been performed using

407
the multipurpose finite-element analysis software DIANA v9.1, which can handle non-linear concrete behaviour.
The columns were modelled by 15520 eight-node, three-dimensional solid brick element, and the steel embedded
reinforcement by 1568 two-node truss elements, while the FRP confinement by 6531 three-node plane bonded
elements (Fig. 3a). In the FEM the concrete is modelled according to a Total Strain Rotating crack model with linear
softening in tension and Thorenfeldt curve in compression. Steel reinforcement is ideally elasto-plastic, while FRP is
modelled as an elastic material. It is remarked that at date the FEM analysis does not take into account
reinforcement buckling and concrete cover spalling, but only concrete cracking and plasticity. A comparison of
FEM results and refined section modeling with experimental outcomes for both U2 and S2 is plotted in Fig. 2b. Fig.
3b shows the crack pattern at peak load for U2 specimen that is consistent to the experimental data. Fig. 3c depicts
the FRP strain ascending branch evolution for S2 specimen compared to experimental.
1250

1000

FEM
750 S2
FRP

P [kN]
500

250

0 FRP Strain
a) b) c) 0.00E+00 1.50E-04 3.00E-04 4.50E-04

Figure 3: F.E.M. model (a), Crack pattern at peak load U2 (b), S2 FRP strain development (exp. vs. FEM) (c)

6. CONCLUSIONS
The present work is included into a wider activity that aims to improve the knowledge and develop an effective
design method for fast strengthening of hollow bridge columns so that bridge function can be quickly restored.
The presented numerical procedure to predict hollow cross section behaviour under combination of flexure and
compression included appropriate models for compressed bars buckling and concrete cover spalling, because the
failure of unstrengthened members is strongly affected by the occurrence of these premature mechanisms; the FRP
confinement allows delaying these mechanisms thus resulting in both strength and ductility increases of members
even under large eccentricities. The occurrence of these mechanisms has been modelled and the results of
experiments and theoretical and FEM analyses show that a good agreement was achieved between the experimental
outcomes and the analytical calculated results.

7. REFERENCES
CEB-FIP Model Code 1990,“Design Code”. Comité Euro-Intermational du Béton, Lausanne,Switzerland, T Telford.
CNR-DT 200 (2004) “Guide for the Design and Construction of Externally Bonded FRP Systems for Strengthening
Existing Structures”. Roma – CNR
Cosenza E and Prota A (2006) “Experimental Behavior and Numerical Modelling of Smooth Steel Bars under
Compression", Journal of Earthquake Engineering. In press.
Fam, Amir Z. and Rizkalla, Sami H. (2001) “Confinement Model for Axially Loaded Concrete Confined by FRP
Tubes,” ACI Structural Journal, 98(4):251-461.
Fam, A., Schnerch, D. and Rizkalla, S. (2005) "Rectangular Filament-Wound GFRP Tubes Filled with Concrete
under Flexural and Axial Loading: Experimental Investigation", Journal of Composites for Construction,9(1):25-33.
Hillerborg A. (1989) “The compression stress-strain curve for design of reinforced concrete beams”. Fracture
Mechanics: Application to Concrete, ACI SP-118:281-294.
Lignola GP, Prota A, Manfredi G and Cosenza E. (2006) “Experimental performance of RC hollow columns
confined with CFRP”. ASCE Journal of Composites for Construction. In press.
Mander JB, Priestley MJN and Park R. (1988) “Theoretical stress-strain model for confined concrete” ASCE Journal
of Structural Engineering, 114(8):1804-1826.
Spoelstra MR, Monti G. (1999) “FRP-confined concrete model”. ASCE Journal of Composites for Construction,
3(3):143-150.

408
Third International Conference on FRP Composites in Civil Engineering (CICE 2006)
December 13-15 2006, Miami, Florida, USA

PREDICTION OF FAILURE LOADS OF FRP PULTRUDED PROFILES

USING CLOSED-FORM EQUATIONS
Linda M. Vanevenhoven
(Graduate Research Assistant, Dept. of Civil and Environmental Engineering, University of Wisconsin, Madison,
WI 53706, USA)

Carol K. Shield
(Professor, Dept. of Civil Engineering, University of Minnesota, Minneapolis, MN 55455, USA)

Lawrence C. Bank
(Professor, Dept. of Civil and Environmental Engineering, University of Wisconsin, Madison, WI 53706, USA)

ABSTRACT
FRP pultruded profiles having I-shaped sections are used as load bearing compression and flexural members in truss
and frame structures. For design purposes closed form expressions are desired for predicting both global and local
failure loads due to buckling or material rupture. A set of closed-form equations has been assembled from the
technical literature for the prediction of the critical local and global failure modes of pultruded columns needed for
design. A database of tests on pultruded columns conducted over the last 15 years and reported in the technical
literature was used to verify the accuracy of the proposed theoretical equations. Only those test data for which the
measured orthotropic material properties of the profiles tested were reported were considered. This paper presents
comparisons between the theoretical equations and test data for wide-flange pultruded columns for two failure
modes, local-flange buckling and global-flexural buckling. Insufficient test data exist in the literature on other
profiles (such as narrow flange I beams or box sections) or for failure modes other than those noted. However,
since most pultruded columns used are of the wide-flange profile type and have been shown to fail either in local or
global buckling modes, the results presented are of practical importance. The paper also addresses the phenomenon
of the interaction of local and global buckling modes for intermediate length columns. Ultimately, the results of this
investigation will be used to develop resistance factors for a load and resistance factor design (LRFD) basis for
pultruded columns.

KEYWORDS
Columns, Global buckling, Local buckling, Pultrusion, Wide-flange profiles.

INTRODUCTION
Wide-flange, pultruded, glass-reinforced FRP profiles are produced by a number of manufacturers in the US. They
have geometries, ranging from depths of 4 inches to 12 inches and wall thickness of ¼ inch to ½ inch. In these
commonly produced profiles the height of the section, h, is equal to the breadth, b, and the wall thickness, t, is the
same in the web and the flanges of the profile. The walls of these pultruded profiles typically contain alternating
layers of E-glass rovings and continuous filament mats (CFMs). Profiles of polyester and vinylester resin systems
are generally produced. Due to the proprietary fiber architectures used by different manufacturers, the profiles
generally have different mechanical properties from manufacturer to manufacturer even though their geometric
dimensions are nominally identical.
Over the last 15 years, a number of studies have been conducted to determine the failure modes of wide-flange
pultruded columns. Concurrently, studies have been conducted to develop theoretical equations to predict the
observed failure modes and loads. Recently, a set of these theoretical equations has been assembled and
recommended for use for the design of pultruded columns (Bank, 2006). The equations are based on orthotropic
plate theory and include the failure modes of local in-plane and shear buckling, global buckling, and material
rupture. In this paper the equations presented in Bank (2006) for local flange buckling, global flexural buckling,

409
and their interaction are presented and comparisons are made with test data presented in the literature to verify the
accuracy of these equations. The effect on the variability in the reported measured material properties of the
profiles tested is also discussed. In order to judge the validity of the proposed expressions for global and local
buckling of open pultruded sections, test results that did not include any measured material properties were
discarded. Ultimately, 75 buckling tests meeting the criteria on sections including W12×12×½, W10×10×½,
W10×10×3⁄8, W8×8×½, W8×8×3⁄8, W6×6×3⁄8, and W4×4×¼ were identified (Mottram et al., 2003; Lane and
Mottram, 2002; Scott, 1997; and Yoon, 1993).
To predict the global flexural buckling of pultruded columns the Euler buckling equation with a correction for shear
deformation is recommended (Bank, 2006). Numerous investigators have proposed this equation to predict the
global buckling load of “long” columns (Barbero and Tomblin, 1993; Zureick and Scott, 1997; Mottram et al.,
2003). For global buckling about the minor axis of the wide-flange section, the following equations are used:
Peuler π 2 E L I min Af
Pglobal = ; Peuler = ; ks =
1+
Peuler (kL ) 2
1.2 Ag
k s Ag GLT
where, Peuler is the Euler buckling load, ks the shear coefficient (Scott, 1997), Ag the gross area of the cross section,
GLT the in-plane shear modulus of the pultruded material in the flanges, EL the longitudinal modulus of the pultruded
material in the flanges, Imin the second moment about the minor axis of the profile, k the end-restraint coefficient,
and Af the area of both flanges of the profile.
In recent years, numerous numerical and approximate equations have been proposed to predict local flange buckling
in pultruded profiles (Mottram, 2004). The equation proposed by Kollár (2003) is recommended by Bank (2006).
This closed-form equation is based on classical orthotropic laminated plate theory and includes an explicit method
to determine the coefficient of restraint at the web-flange junction in the profile:
7 D Lf DTf DTf
Plocal = σ local Ag ; σ local = + 12 D Sf ; ζI =
(b f 2 tf)
2
1 + 4.12ζ I (
kI b f 2 )
⎛ ( )
σ ss Ew ⎞ 4t 2f f
( )
⎜1 − free f L ⎟ ; σ ss = 2π ⎛⎜ D w D w + D w + 2 D w ⎞⎟ ; σ ss ( )
2
Dw
kI = T =
( )
GLT
dw ⎜ σ ssss w E Lf ⎟
ss w
twd w ⎝
2 L T LT S
⎠ free f
b 2f
⎝ ⎠
where, σlocal is the in-plane buckling stress of one outstanding flange, bf is the breadth of the profile (twice the flange
breadth), tf or tw is the thickness of the flange or web, ζI is the flange coefficient of restraint, kI is the web-flange
junction rotational stiffness, dw is the depth of the profile, σ ssfree is the in-plane buckling stress of the flange
assuming no restraint at the web-flange junction, σ ssss is the in-plane buckling stress of the web assuming it is simply
supported at the web-flange junction. The wall flexural rigidities, DL, DT, DLT, and DS are given in terms of the
orthotropic material properties, EL, ET, GLT, νL, and νT, of either the web or the flange as,
ELt 3 ET t 3 G t3
DL = ; DT = ; DLT = ν T DL ; DS = LT
12(1 − ν L ν T ) 12(1 − ν L ν T ) 12
For columns of intermediate length, interaction between the local and global buckling modes (termed “combined
buckling”) is thought to occur and has been observed in a small number of the tests considered in this paper
(Mottram et al., 2003). To account for the interaction between local and global buckling, an equation has been
proposed by Barbero and Tomblin (1994) as follows,

Pcr = ki Plocal ; λ =
Plocal
Pglobal
; (2
)
k λ = 1 + (1 λ ) 2c ; ki = k λ − k λ2 − 2
cλ
1

where, Pcr is the critical buckling load, λ is a non-dimensional column length, and c is a curve-fitting parameter that
is used to calibrate the interaction equation with experimental data. When λ = 1 the critical length is obtained where
local and global buckling are predicted to occur at the same critical load. When c = 1 the second-order interaction
equation reduces back to the two separate equations, one for local buckling ( Pcr = Plocal which is length
independent) and one for global buckling (Pcr= Pglobal = Plocal/λ2).

DATA ANALYSIS
Three different failures were identified in the tests found in the literature: global buckling about the minor axis,
local buckling, and “combined” buckling. Predictions for global buckling depend on both EL and GLT. Predictions

410
for local buckling depend on these two Table 1 Values measured through experimental investigation
material properties, as well as ET and νL. Publication EL ET νL GLT t d b
Because some of these tests were performed Yoon (1993) x x x x x x x
mainly to investigate global buckling, not
Scott (1997) x - - x x x x
every researcher measured every material
Lane & Mottram (2002) x - - - x x x
property. Table 1 indicates which properties
Mottram et al (2003) x - - - - - -
and dimensions were measured and reported
by each researcher.
Material properties of pultruded sections vary among size and manufacturer, so different methods were used to
estimate the missing data from each of the studies (Vanevenhoven, 2006). The tests done by Yoon (1993) and Scott
(1997) were both performed at Georgia Institute of Technology, so it was assumed that the specimens were
reasonably similar. Therefore, the Yoon data was used to help estimate reasonable ET and νL for the Scott data.
Scott tested W6x6x3⁄8 sections and W4×4×¼ sections. Yoon also tested W6×6×3⁄8 sections; material properties for
the Yoon section that best matched Scott’s measured EL were used for Scott’s W6×6×3⁄8. The values for the
W4×4×¼ sections that Scott tested were estimated using the Yoon data assuming that the ratios between the needed
properties and EL were constant for the same manufacturer. Estimates for the non-measured properties for Lane &
Mottram (2002) and Mottram et al. (2003) were determined by using other data from the literature on tests of
similar sections (Bank et al., 1996) from the same manufacturer at approximately the same time period.
Figure 1 shows a plot of the
experimental buckling load 1.6
Lane and Mottram - global
normalized by the predicted local
1.4 Lane and Mottram - local
buckling load as a function of λ with Lane and Mottram - combined
the theoretical curves using values of 1.2 Mottram et al - global
c = 1 (no interaction between local Mottram et al - local
1 Scott - global
and global buckling), c = 0.8 as
Ptest/Plocal

Yoon - local
recommended by Bank (2006) and 0.8
c=1
c = 0.99. In general, the tests that c=0.99
0.6
failed in global buckling (solid c=0.8
points) follow the curve predicted 0.4
using c = 1 well, with the tests from
0.2
Lane and Mottram, and Mottram et
al. lying closer to the theoretical 0
curve than those from Scott. Test-to- 0 0.5 1 1.5 2 2.5 3 3.5 4
predicted ratios based on c = 1 for λ
each reference are given in Table 2. Figure 1 Nondimensional buckling load versus length
The tests failing in local buckling
(open symbols) show significantly
more scatter than those that failed in global buckling. For small λ, a small number of half wavelengths was
observed in the tests (typically 3 for the smallest λ and up to 7 for λ close to 1). When there are few half
wavelengths, the effect of unwanted test end restraint could increase the measured local buckling load. The local
buckle failures also have a higher test-to-predicted ratio. The other two tests, identified as “combined” failure,
occurred at λ ≈ 1 at a load close to the predicted local buckling load. These two tests have a lower test-to-predicted
ratio than the tests that failed in local buckling, thus it may be possible that an interaction between global and local
buckling caused the tests with the combined failures to fail prematurely. It also appears that the interaction between
these buckling modes is
limited to λ very close to 1, Table 2 Test-to-Predicted Ratios
so using values of c around Global Local Combined
0.99 would limit the affect ave COV n ave COV n ave COV n
of interaction on the Lane&Mottram 0.97 2.0% 9 1.13 2.5% 3 0.98 4.1% 2
theoretical predictions to Mottram et al 1.01 6.2% 17 1.11 5.0% 2 --- ---
λ very close to 1, as Scott 0.92 4.9% 11 --- --- --- ---
opposed to c = 0.8, which Yoon --- --- 1.07 12.9% 31 --- ---
affects the curve between
0.2<λ< 2.

411
Because measured material data from the exact columns tested was not available for all the tests, a sensitivity study
was conducted using the column test data from Lane and Mottram to determine the sensitivity of the results to the
non-measured material properties. Yoon (1993) found coefficients of variations (COVs) of 12.5% for ET, 3.6% for
νL and 12.7% for GLT based on coupon tests. Using these COVs, the Lane and Mottram data was re-evaluated using
transverse material properties two standard deviations higher and lower than the estimated mean. The results of this
study are shown in Table 3. For the tests that failed by local buckling, changing the transverse material properties
had a large impact on the predicted buckling loads. Increasing the ET, GLT, and νL by 2 standard deviations caused
the predictions to go from conservative to unconservative. As
expected, the effect of changing ET, νL and GLT on the Table 3 Test-to-Predicted Ratios for Lane
predictions for the global buckling tests was much less and Mottram Data
significant, as the global buckling load is only slightly Local Global
dependent on GLT and not at all dependent on the other two ET, νL, GLT ave COV ave COV
properties. Because of the large impact of transverse material
-2 std dev 1.42 2.5% 0.98 1.9%
properties on local buckling, it may be possible to attribute the
mean 1.13 2.5% 0.97 2.0%
large variation in the Yoon data in Figure 1 to the variation in
+2 std dev 0.94 2.5% 0.97 2.0%
measured material properties that Yoon found, even within a
single profile series.

CONCLUSIONS
In general, the presented design equations yield good agreement with the available test data for the global and local
buckling of pultruded GFRP columns, although very few complete sets of data exist. In the past, adequacy of the
proposed theoretical equations for global and local buckling and their interaction have been investigated using
measured buckling loads paired with nominal design material properties, or in some cases unspecified material
properties. In order to determine the professional factor for a reliability based calibration of the design equations,
the adequacy of the theoretical predictions must be evaluated using measured material properties. Different
approaches to achieving safety in the design equations are also possible. Some have suggested adjusting the value
of c in the interaction equation to achieve a lower bound design equation over the entire range of slendernesses
(Barbero and Tomblin, 1994). An alternative approach would be to select a large value of c to account only for
interaction between local and global buckling over a small range of slenderness (0.9<λ<1.1), and use a reliability
based calibration of a resistance factor to provide the required safety.

ACKNOWLEDGEMENTS
Support for the first author from Universities of Wisconsin-Madison and Minnesota is gratefully acknowledged.

REFERENCES
Bank, L.C. (2006). Composites for Construction: Structural Design with FRP Materials, Wiley, NY, Chapter 14
Bank, L.C, Gentry, T.R., and Nadipelli, M., (1996) “Local Buckling of Pultruded FRP Beams - Analysis and Design,” Journal
of Reinforced Plastics and Composites, V15, No. 3, pp. 283-294.
Barbero E.J. and Tomblin J. (1993) “Euler buckling of thin-walled composite columns,” Thin-Walled Structures. V17, 237-258.
Barbero E.J. and Tomblin J. (1994) “A phenomenological design equation for FRP columns with interaction between local and
global buckling,” Thin-Walled Structures. V18, 117-131.
Kollár, L.P. (2003) “Local buckling of fiber reinforced plastic composite structural members with open and closed cross
sections,” Journal of Structural Engineering, V129, No. 11, 1503-1513.
Lane A. and Mottram J.T. (2002) “Influence of modal coupling on the buckling of concentrically loaded pultruded fibre-
reinforced plastic columns,” Proc Inst Mech Engrs Part L: J Materials: Design and Applications, V216, 133-144.
Mottram, J.T. (2004) “Determination of critical load for flange buckling in concentrically loaded pultruded columns,”
Composites Part B: Engineering, V35, N1, 35-47.
Mottram J.T., Brown N.D. and Anderson D. (2003) “Physical testing for concentrically loaded columns of pultruded glass fibre
reinforced plastic profile,” Structures & Buildings 156 Issue 2, 205-219.
Scott D.W. (1997) Short- and Long-Term Behavior of Axially Compressed Slender Doubly Symmetric Fiber-Reinforced
Polymeric Composite Members, PhD Thesis, Georgia Institute of Technology.
Vanevenhoven, L. (2006) “Closed-form Equations for Predicting Failure Loads of Pultruded Profiles,” Independent Study
Report. University of Wisconsin-Madison.
Yoon S.J. (1993) Local Buckling of Pultruded I-Shape Columns, PhD Thesis. Georgia Institute of Technology.
Zureick, A. and Scott, D., (1997), “Short-term Behavior and Design of Fiber-reinforced Polymeric Slender Members Under Axial
Compression,” Journal of Composites for Construction, V1, pp. 140-149.

412
Third International Conference on FRP Composites in Civil Engineering (CICE 2006)
December 13-15 2006, Miami, Florida, USA

THEORETICAL STUDY ON SPALLING RESISTANCE OF FRP SHEETS

BONDED TO BENT CONCRETE SURFACE
Hong YUAN
(Professor, Department of Mechanics and Civil Engineering, Jinan University, Guangzhou 510632, P R China)

Ren-huai LIU
(Professor, Department of Mechanics and Civil Engineering, Jinan University, Guangzhou 510632, P R China)

ABSTRACT
In this paper, the peeling behavior and the spalling resistance effect of fiber reinforced polymer (FRP) sheets
externally bonded to bent concrete surface are firstly investigated theoretically. A peeling load is applied on the FRP
sheet by loading a circular rod placed into the central notch of beam, and the theoretical analysis is conducted for the
specimens. Load is expressed as the function of peel angle by geometry analysis and equilibrium of forces between
FRP sheets and circular rod. Interfacial fracture energy is calculated analytically using membrane peeling method.
Both initial loading and debonding propagation stages are explored. It is realized that only two material parameters,
the interfacial fracture energy of FRP-concrete interface and the tensile stiffness of FRP sheets, are necessary to
represent the interfacial spalling behavior. Finally, the peeling load – deflection curves for various radii of curvature
of concrete surface are demonstrated by means of a numerical example. It is found that radius of curvature has
remarkable influence on peeling load.

KEYWORDS
FRP sheet, interfacial fracture energy, debonding, spalling resistance, bent concrete surface

1. INTRODUCTION
Acceptance of high performance composites in the construction industry has grown at a rapid pace during the last
decade. One of the areas where composites are preferable is the rehabilitation of existing structures against spalling
failure. When FRP sheet is applied to a concrete surface to prevent pieces of the concrete from peeling and spalling,
acting force on FRP sheet is similar with the usual peel test to peel a sheet from a substrate. Nevertheless, there
exists an important difference between peel and spalling resistance test. In peel test, the peel angle can be adjusted
according the experimental need, and can be varied independently. But in the spalling resistance test, peel angle is
determined by the interfacial geometric and material properties, and is not an independent variable.

Kimpara et al. (1998) proposed firstly a spalling resistance test method of FRP sheets bonded to mortar and concrete
to characterize the peeling strength and also to examine the effects of different surface treatment and primer. Wu et
al. (2005a,b) studied experimentally and analytically the peeling behavior and spalling resistance effect of
unidirectional and bi-directional FRP sheets externally bonded to plane concrete surface. Simple expressions among
peeling load, interfacial energy and FRP sheet stiffness were obtained. The theoretical results were validated by
comparing with experimental results. In this paper, the peeling behavior and the spalling resistance effect of FRP
sheets externally bonded to bent concrete surface are investigated theoretically.

2. THEORETICAL ANALYSIS
Theoretical analysis is carried out on specimens representing the spalling resistance effect of unidirectional FRP
sheets externally bonded to bent concrete surface. A notch of 2a0×2a0 is made in the mid-span at the bottom side of
the beam specimens, where loading pin is situated. Figure 1 shows the detail dimensions of the specimen and the

413
loading arrangement. Both initial loading stage before debonding and debonding propagation stage are explored.
The bending energy of the FRP sheets is neglected to simplify the analysis.

h0 Concrete beam 2a0 a

h 2a0
ui L
Pin P u φ
FRP sheet
P
(a) Elevation R

θ
θ0
b

l
O

(b) Plan (c) Extension of debonding

Figure 1: Specimen Details and Loading Arrangement

Refer to Figure 1, where l is span of the beam, h is height of the beam, h0 is height of the beam at middle span, b is
width of FRP sheet, whereas t is the thickness of FRP sheet. ϕ is the angle between vertical line and FRP sheet. u is
the deflection of the loading pin. a is debonding length, whereas ao is the initial debonding length. T is peel force per
unit width of FRP sheet. P is the load on the circular pin and applied through the loading device. a0+a is total
debonding length. L is the deformed length of original FRP length L0=a0+a. R is curvature radius of bent surface,
and stands for the instant FRP debonding location. O is curvature center. θ0 is angle between vertical line and R at
initial debonding location. θ is angle between vertical line and R at instant debonding location.

2.1 Initial Loading Stage Before Debonding

When load is acted on the pin, the FRP sheet of length 2a0 is firstly subjected to deformation because of the
existence of initial debonding length a0 as shown in Fig. 1. Peel angle increases from zero to the maximum value
with the increase of loading. By assuming the diameter of pin is small compared to the debonding length,
equilibrium of the pin gives
P = 2Tb cos ϕ (1)
According to the geometrical deformation, normal strain in FRP sheet ε is obtained as
1
ε= −1 (2)
sin ϕ
Assume that FRP sheet is linear elastic and obey Hook’s law, the peel force per unit width of FRP sheet is
T = Etε (3)
where E is Young’s modulus of FRP sheet. Substituting (3) and (2) into (1) yields
⎛ 1 ⎞⎟ 1 u
P = 2 Etbx⎜⎜1 − , x = cot ϕ = = , for 0 ≤ u ≤ u m = a 0 cot ϕ m (4)
2 ⎟ tan ϕ a0
⎝ 1+ x ⎠
where ui is the deflection of the loading pin before debonding and um is the maximum deflection of the loading pin
before debonding. ϕ m is value of ϕ when debonding appears and begin extend.

2.2 Debonding Propagation Stage

414
When interfacial debonding between FRP sheet and concrete beam occurs. debonding length can be calculated as
a = R(θ − θ 0 ) , for θ ≥ θ 0 (5)
Original length of FRP sheet is
L0 = a 0 + a = a 0 + R (θ − θ 0 ) (6)
The deformed length of original length L0 is
L = R 2 + ( R cos θ 0 − u ) 2 − 2 R( R cos θ 0 − u ) cos θ = [u − R(cosθ 0 − cos θ )] + R 2 sin θ 2
2
(7)
Angle ϕ is obtained by geometrical analysis
u − R(cos θ 0 − cos θ )
cos ϕ = (8)
L
and normal strain in FRP sheet is
L − L0 R 2 + ( R cos θ 0 − u ) 2 − 2 R ( R cos θ 0 − u ) cos θ
ε= = −1 (9)
L0 a 0 + R (θ − θ 0 )
Similar derivation with Eq. (4) gives
⎛ L ⎞ u − R(cos θ 0 − cos θ )
P = 2 Etbε cos ϕ = 2 Etb⎜⎜ − 1⎟⎟ (10)
⎝ L0 ⎠ L
Namely
⎡ u − R (cos θ 0 − cos θ ) u − R (cos θ 0 − cos θ ) ⎤
P = 2 Etb ⎢ − ⎥ (11)
⎣ L0 L ⎦
As the observed load-deflection curve is not linear, the conventional compliance method is not appropriate to be
applied to evaluate the energy release rate. To evaluate the energy release rate, a modification of the compliance
method was proposed by Kimpara et al. (1998). Elastic strain energy U can be obtained by integrating equation (11)
with respect to u under the condition that debonding length a is constant.
⎡ 0.5u 2 − Ru (cos θ 0 − cos θ ) ⎤
U = 2 Etb ⎢ − L + R 2 (1 + cos 2 θ 0 ) − 2 R 2 cos θ cos θ 0 ⎥ (12)
⎣ L 0 ⎦

Finally, energy release rate G is derived by differentiating strain energy with respect to a under the condition that
u of FRP sheet is constant.
∂U ⎡ 0.5u 2 − Ru (cos θ 0 − cos θ ) u sin θ ( R cos θ 0 − u ) sin θ sin θ cos θ 0 ⎤
G=− = Et ⎢ + + − ⎥ (13)
2bR∂θ ⎣ L0
2
L0 L (1 + cos 2 θ 0 ) − 2 cos θ cos θ 0 ⎥⎦

Letting θ = θ 0 , L0 = a 0 , L = u m + a 0 , G = G f obtains deflection um when debonding initiates.

2 2

⎡ ⎤
⎢ 2 um ⎥
⎢ um um
cos θ − ⎥
R
0
G f = Et ⎢ + + − cos θ 0 ⎥ (14)
⎢ 2a 0
2
R um
2
⎥
⎢ +1 ⎥
a
2
⎣ 0 ⎦
where Gf is interfacial fracture energy. Initial debonding load is given by equation (4) as
⎛ ⎞
Pm = 2 Etbx m ⎜⎜1 −
1 ⎟ , x = um (15)
2 ⎟
⎜ ⎟
m
1 + x a0
⎝ m ⎠
Setting G = G f in equation (13) yields a relation between deflection u and θ. Finally, (11) gives load-deflection
curves at debonding propagation stage. Obviously, only two material parameters, the interfacial fracture energy of
FRP-concrete interface Gf and the tensile stiffness of FRP sheets Et, are necessary to represent the interfacial
spalling behavior.

3. NUMERICAL SIMULATIONS

415
The peeling load – deflection curves for various radii of curvature of concrete surface are shown in Figure 2.The
following parameters are used in the examples: t=0.111mm, b=120mm, a0=15mm, E=230GPa, Gf =0.40N/mm, and
R varies between 2000mm and ∞. It can be concluded that radius of curvature has remarkable influence on peeling
load. At initial loading stages before debonding, peeling load – deflection curves are the same linear lines. When
initial debonding occurs, peeling load decreases with increasing deflection. Peeling load decreases more rapidly with
deflection for small radii of curvature.

1.8
R=∞
1.6
R=50000
1.4
R=10000
1.2
Load P (kN)

R=6120
1.0

0.8 R=5700

0.6 R=5430

0.4 R=5140

0.2 R=2000

0.0
0 20 40 60 80 100 120
Deflection u (mm)

Figure 2: Peeling Load-Deflection Curves for Different Radii of Curvature

4. CONCLUSIONS
The peeling behavior and the spalling resistance effect of fiber reinforced polymer (FRP) sheets externally bonded
to bent concrete surface are firstly investigated theoretically. It is realized that only two material parameters, the
interfacial fracture energy of FRP-concrete interface Gf and the tensile stiffness of FRP sheets Et, are necessary to
represent the interfacial spalling behavior. It is found that radius of curvature and interfacial fracture energy have
remarkable influence on peeling load – deflection curves.

ACKNOWLEDGMENTS
The authors gratefully acknowledge the financial support provided by the Scheme of Science and Technology of
Guangdong Province (2005B32801002), Ministry of Construction (No. 05-K4-45), Guangzhou City (2005J1-C0251)
and Jinan University, China.

REFERENCES
Kimpara, I., Kageyama, K., Suzuki, T., Ohsawa, I. and Yamaguchi, K. (1998). “Characterization of peeling strength
of FRP sheets bonded on mortar and concrete”, Proceedings of the Eighth JAPAN-U.S. Conference on Composite
Materials, pp 1010-1019.
Wu, Z. S., Yuan, H., Kojima, Y., and Ahmed, E. (2005a). Experimental and analytical studies on peeling and
spalling resistance of unidirectional FRP sheets bonded to concrete, Composites Science and Technology, Vol. 65,
No. 7-8, pp. 1088-1097.
Wu, Z. S., Yuan, H., Asakura, T., Yoshizawa, H., Kobayashi, A., Kojima, Y., and Ahmed, E. (2005b). Peeling
behavior and spalling resistance of bonded bi-directional fiber reinforced polymer sheets, Journal of Composites for
Construction, Vol. 9, No. 3, pp. 214-226.

416
Third International Conference on FRP Composites in Civil Engineering (CICE 2006)
December 13-15 2006, Miami, Florida, USA

THEORY OF A NEW METHOD FOR EXTERNALLY ANCHORED FRP

BAND PRE-STRESSED REINFORCEMENT
Zhuo Jing
(Chongqing Communications Research & Design Institute, Chongqing, China 400067)

Li Tangning
(School of Civil Engineering, Chongqing University, Chongqing, China 400045)

ABSTRACT
After several years of development, FRP sheet pre-stress technology still remains in lab stage. The reason is that no
FRP sheet grips or anchors suitable for easy application on a reinforcement site, the most critical aspect of pre-stress
reinforcement technology, have been developed from previous experiments and researches. The invention of WSGG
(a new type anchorage of FRP, which is made of two or more wave shaped gear plates) anchor resolves the problem
of FRP sheet anchoring. After a series of researches, a brand-new method for externally anchored FRP sheet pre-
stress reinforcement (invention patent No. 02128047.9) is conceived, which makes full use of the features of WSGG
anchors by fixing both ends and then fastening in the middle to force FRP sheets to elongate geometrically and
generate a pre-tension. Worn bridge members (solid RC plates of 8m long) removed from Cheng-Yu Expressway is
used to take an experiment of externally anchored carbon fiber fabric pre-stress reinforcement. Results of the
experiment demonstrates that application of pre-stress by the method is very simple, easy, and technically
practicable, requires no special straining devices, and remarkably enhances effects of reinforcement, showing quite
broad prospects of generalization and application.

KEYWORDS
pre-stressed, carbon fiber fabric, reinforcement, WSGG anchor

1. INTRODUCTION
If an FRP sheet is simply bonded to the surface of a structural member for reinforcement, tensile stress in the FRP
sheet remains low until the member is damaged under a load; therefore, material strength cannot function well.
Especially in actual reinforcement engineering cases, where an initial load usually exists on the members to be
reinforced, strain in the FRP sheet will lag much behind that in steel bars in the beam under secondary load and
makes economic effect of reinforcement less evident if an ordinary method for FRP sheet reinforcement is used. For
the purpose of making more adequate and more effective use of the feature of high strength of FRP sheets and
achieving better effects of reinforcement, the technology of pre-tensioned FRP sheet reinforcement has significant
application value. Initial pre-stress generated by pre-tensioning of FRP sheets can be used to balance part of the dead
weight of or the load on the structure so that FRP sheets can fully exert its effects of reinforcement, such as
considerable deferment of crack development and reduction of crack width, effective enhancement of structural
rigidity, reduction of structural member deflection, mitigation of internal bar strain, and enhancement of bar yield
load and of ultimate bearing capacity of the structure (Stöcklin and Meier, 2001) (Triantafillou et al., 1992).
After several years of development, however, FRP sheet pre-stress technology remains in lab stage as there exist this
and that problems that prevent it from being utilized (Wu and Matsuzaki, 1998). One of the problems is the complex
arrangement, complicated operational procedure, and small tonnage of tensioning implements. To resolve existing
technical problems in relation to FRP sheet pre-tensioning and anchoring, WSGG anchorage is developed
specifically for FRP sheets (Zhuo and Li, 2004). The reliability of FRP sheets gripped and anchored with WSGG
anchors has been verified by a number of experiments which show their anchoring performance much superior to U-
shaped FRP sheets or bolts (Zhuo and Li, 2005). Based on the features of WSGG anchors, we specially conceive a
brand-new method for externally anchored carbon fiber band pre-stress reinforcement (invention patent No.:

417
02128047.9), which has very simple operating procedure and requires no special equipment but fastening operations
on bolts to perform pre-tensioning and anchoring of FRP sheets. The following is our study on the pre-stress theory
of the method as well as experimental verification by actual bridge members.

2. THEORY OF THE METHOD FOR EXTERNALLY ANCHORED FRP SHEET PRE-

STRESS REINFORCEMENT

2.1 Basic theory of the occurrence of pre-stress from gripping and anchoring of FRP sheets with WSGG
anchors

As shown in Fig 1 (a), an FRP sheet with straight length l is fixed and restrained at both ends. When a WSGG
anchor is used to grip the middle of the FRP sheet, the FRP sheet will be forced to elongate. Since there is difference
δ between the stretched length of the camber line of WSGG anchor’s wave-shaped gear faces and the projected
length of its plane, the FRP sheet will elongate by δ when upper and lower wave-shaped gear faces of the WSGG
anchor get fully held down in perfect fitness, as shown in Fig 1 (b). The value of such forced elongation causes the
occurrence of pre-stress in the FRP sheet:
δ
F = EA ⋅ (1)
l
F: Built-up pre-tension in the FRP sheet;
E: Tensile elastic modulus of the FRP sheet;
A: Sectional area of the FRP sheet;
L: Net length of the FRP sheet between two fixed ends;
Δ: Difference between the stretched length of the camber line of WSGG anchor’s wave-shaped gear faces and its
horizontally projected length

Fig 1 Schematic of the pre-stress applied to the FRP sheet by a WSGG anchor

During actual reinforcement of a structure, fixing restraint on both ends of an FRP sheet can be implemented by
WSGG anchors fixed to the structure. To achieve pre-stress tensioning and anchoring of said FRP sheet, therefore,
just 3 or more WSGG anchors will do. To be specific, both ends of the FRP sheet are anchored with fixed WSGG
anchors to form two anchor points before gripping and anchoring the middle of two anchor points with WSGG
anchors, so as to generate a pre-tension in the FRP sheet.

2.2 Theory of pre-stress in a multipoint anchored FRP band

Where multipoint anchoring of an FRP sheet is required for reinforcement of a longer structure, the basic pre-stress
theory detailed in Section 1.1 is still applicable to multipoint gripping and anchoring of the FRP sheet. For example,
the following technical program for pre-stress reinforcement can be used when 5 WSGG anchors are used for pre-
stress tensioning and anchoring of an FRP sheet, as shown in Fig 2:
Step 1: Fix in advance lower wave-shaped gear plates of 5 WSGG anchors to the structure to be reinforced and then
anchor both ends of the FRP sheet at WSGG anchors A and E to form two anchor points, as shown in Fig 2(a).
Step 2: Close upper wave-shaped gear plates of WSGG anchors B and D simultaneously and screw up the bolts
synchronously and uniformly such that upper and lower wave-shaped gear faces of WSGG anchors B and D get
completely held down in perfect fitness, as shown in Fig 2(b). For the purpose of simplicity, it’s assumed that all
pre-stress losses in the FRP sheet are ignored, including the assumption that the reinforced structure maintains its

418
rigidity or, in other words, the reinforced structure will not suffer any loss arising from deformation (bending and
compression) under eccentric load of FRP sheet pre-stress. Under such ideal conditions, then, the FRP sheet between
two anchor points of WSGG anchors A and E can be considered to be stretched forcibly by 2δ when upper and
lower wave-shaped gear faces of WSGG anchors B and D get completely held down in perfect fitness, giving rise to
the pre-tension shown in Formula (2):
2δ
F1 = EA ⋅ (2)
l AE

l AE

l AE

l BD
l

Fig 2 Schematic of the pre-stress applied to an FRP sheet by WSGG anchors in series

Under said conditions, pre-tension in the FRP sheet applies an eccentric pre-stress of magnitude F1 to the reinforced
structure via WSGG anchors A and E. Since the action of the FRP sheet on WSGG anchor B or D is vertical during
fastening with WSGG anchors B and D, longitudinal eccentric pre-stress of both WSGG anchors B and D on the
reinforced structure is 0.
Step 3: Next, close upper wave-shaped gear plate of WSGG anchor C and screw up the bolts uniformly such that
upper and lower wave-shaped gear faces of WSGG anchor C get completely held down in perfect fitness, as shown
in Fig 2(c). For the purpose of simplicity, the same assumptions as in Step 2 are made, plus further assumption that
the FRP sheet is completely anchored at WSGG anchors B and D or, in other words, that no slippage will arise to
the FRP sheet in WSGG anchor B or D. Under ideal conditions, then, the FRP sheet between two anchor points of
WSGG anchors B and D can be considered to be stretched forcibly by δ when upper and lower wave-shaped gear
faces of WSGG anchor C get completely held down in perfect fitness, giving rise to the pre-tension increment
shown in Formula (3):
δ
ΔF2 = EA ⋅ (3)
l BD
Under said ideal assumptions, various pre-stressing losses are ignored and Formula (2) is added to Formula (3) to
get the pre-tension in the FRP sheet between two anchor points of WSGG anchors B and D:
 2δ δ 
F2 = F1 +Δ F2 = EA ⋅  +  (4)
 l AE l BD 
Similarly under said conditions, pre-tension increment in the FRP sheet between WSGG anchors B and D applies
longitudinal eccentric pre-stress of magnitude DF2 to the reinforced structure via WSGG anchors B and D. WSGG
anchors A and E maintain application of longitudinal eccentric pre-stress of magnitude F1 to the reinforced
structure and longitudinal eccentric pre-stress of WSGG anchor C on the reinforced structure is 0.
or multipoint FRP sheet gripping and anchoring, different operational sequence corresponds to different pre-tension
arising to each section of the FRP sheet; therefore, reasonable pre-stressing construction program can be arranged
depending on actual case of reinforcement.

419
Of course, it’s taken into account that there will be a variety of pre-stressing losses during actual operation, such as
loss caused by eccentric pressure on a member, loss arising from possible slippage of a sheet between anchors at
both fixed ends, and loss arising from incomplete fitness of intermediate anchors. Research activities in such aspect
need to be further carried out.

3 DISCUSSION OF EXTERNALLY ANCHORED FRP BAND PRE-STRESS

REINFORCEMENT TECHNOLOGY
(1) An FRP sheet achieves effective multipoint anchorage on a reinforced structure and ensures reliable transfer of
FRP sheet pre-tension. Especially for greater pre-tensioning tonnage, ordinary U-shaped FRP sheet anchoring or
bolt anchoring cannot achieve this at all, whereas reliable multipoint anchorage can bear not only pre-tension but
also tension increment caused by later load.
(2) The technology provides simple, convenient FRP sheet pre-tensioning operations and requires no special
equipment but fastening operations on bolts to perform pre-stress tensioning and anchoring of an FRP sheet. This
feature makes the method easy to be applied to engineering practice.
(3) Pre-tension in an FRP sheet is allocated between several anchor points in operational sequence; each of the
anchor points bears part of FRP sheet pre-tension only. This feature is helpful in that anchor zone design for anchor
points becomes simpler; especially for reinforcement of an existing structure, the smaller the anchor point tonnage is,
the easier it is to accomplish anchor zone design and construction.
(4) Distribution of FRP sheet pre-tension is different for different anchor sections and different operational
sequences. This feature is significant for reinforcement of a bridge structure with larger span. For example, a simply
supported beam with larger span requires larger pre-stress for the middle section of its span and requires gradually
decreasing pre-stress towards its two sides; distribution of FRP sheet pre-stress after construction by the technology
can meet the requirement.
(5) When pre-stress reinforcement is carried out with the technology, it’s unnecessary for an FRP sheet to perform
bonding operations on the surface of the reinforced structure. This is mainly because WSGG anchors perform
concentrated anchoring of an FRP sheet, which has smaller width and usually larger thickness, and the function
exerted by bonding force of the bonded surface has gone down to a minor place on account of the influence of
various factors; in other words, there will be no significant influence on the effects of reinforcement even if the FRP
sheet is not bonded to the reinforced structure.
(6) Since an FRP sheet is pre-tensioned and anchored by section, its pre-stress reinforcement calculation is relatively
more complicated.

REFERENCES

Wu zhishen, Tanabe K, Matsuzaki T,et al (1998). A retrofitting method for concrete structures with externally
prestressed carbon fiber sheets[J]. Journal of Structural Engineering, 1998, 44A:1299-1308
Stöcklin I, Meier U (2001). Strengthening of concrete structures with prestressed and gradually anchored CFRP
strips[A]. Proceedings of the Fifth International Conference on Fiber-reinforced Plastics for reinforced concrete
Structures(FRPRCS-5) [C], Cambridge, UK, 16-18 July,2001.291-296
Triantafillou T C, Deskovic N, Deuring M (1992). Strengthening of concrete structures with prestressed fiber
reinforced plastic sheets[J]. ACI Structural Journal, 1992, 89(3):235-244
Zhuo Jing, Li Tang-ning (2004). An innovative WSGG anchor of FRP sheet[A]，ACMBS-IV, 2004.8:86
Zhuo Jing, Li Tang-ning (2004).The mechanism of an innovative WSGG anchor of FRP laminations[A],ISSEYE-8,
2004.7:805-811
Zhuo Jing, Li Tang-ning (2005). The mechanics of an innovative wave-shape-teeth-grip anchor of FRP laminations
or sheets[J], China Civil Engineering Journal,2005(10):49-53
Zhuo Jing, Li Tang-ning (2005). The Experimental Research of RC Beam Strengthened with Carbon Fiber Sheet
Which End Anchored by Wave-shaped-teeth-grip anchorage[J], Building Structrue,2005(7):63-66

420
Third International Conference on FRP Composites in Civil Engineering (CICE 2006)
December 13-15 2006, Miami, Florida, USA

POST-BUCKLING BEHAVIOR OF
FRP COMPOSITE THIN-WALLED MEMBERS

Nuno Freitas Silva1, Nuno Silvestre2, Dinar Camotim3

1
Ph.D. Student, 2Assistant Professor, 3Associate Professor
Dept. of Civil Engineering, ICIST/IST, Technical University of Lisbon, Av. Rovisco Pais, 1049-001 Lisboa, Portugal

ABSTRACT
This paper (i) provides a brief account of the steps and procedures involved in deriving and implementing a geometrically non-
linear orthotropic formulation based on Generalized Beam Theory (GBT) and (ii) illustrates its application and capabilities by
presenting and discussing numerical results concerning the distortional post-buckling behavior of lipped channel FRP composite
columns. Taking advantage of the GBT modal features, it is possible to provide fresh insight on the member post-buckling
behavior, namely to obtain (i) accurate “exact” and approximate (but with a “controlled approximation”, achieved by including
just a few selected deformation modes) post-buckling equilibrium paths, as well as (ii) the evolution, along those paths, of the
most relevant displacements and stresses. For validation purposes, some GBT-based results are compared with values yielded
by finite element analyses, performed in the code ABAQUS and adopting shell elements to discretize the thin-walled members.

KEYWORDS
Thin-walled members; FRP composites; Local buckling; Distortional post-buckling; Generalized Beam Theory (GBT)

1. INTRODUCTION
The use of composite material in civil engineering became significant in the last decade, as their well known structural
efficiency and excellent behavior in aggressive environments conditions were joined by sufficiently low fabrication costs. In
particular, these three features are responsible for the growing demand for thin-walled composite structural members in
the construction industry, namely in off-shore structures and chemical plants. On the other hand, most composite materials
exhibit linear elastic stress-strain relations (with fairly low moduli), almost no ductility (i.e., they remain elastic up to collapse)
and different types of orthotropy − these material properties clearly point towards a high susceptibility to (i) local and/or global
instability phenomena and (ii) brittle collapse modes. Since mastering these two aspects is very important to achieve efficient
(safe and economic) designs, engineers must have analytical/numerical tools to model them accurately, so that their effect on the
member structural response may be properly assessed. In order to illustrate the capabilities of Generalized Beam Theory (GBT),
concerning the first aspect, one studies the post-buckling behavior and strength of lipped channel FRP composite columns.

2. GBT POST-BUCKLING ANALYSIS

GBT may be viewed as either (i) a bar theory incorporating cross-section in-plane deformation or (ii) a folded-plate theory
including plate rigid-body motions. By (i) decomposing the member deformed configuration or buckling mode shape into a linear
combination of cross-section deformation modes and (ii) assessing their individual contributions, GBT provides a general
approach to obtain accurate, elegant and clarifying solutions for several structural problems. A GBT post-buckling analysis
comprises two main tasks: (i) a cross-section analysis and (ii) a member analysis. Next, one briefly addresses the most relevant
concepts and procedures involved in performing and numerically implementing these tasks − their application is subsequently
illustrated through the investigation of the post-buckling behavior of FRP composite lipped channel columns.

Take a reference system where x, s, z (u, v, w) are coordinates (displacement components) along the member length, cross-
section mid-line and wall thickness. For a displacement representation compatible with a classical beam theory, one considers

u( x , s ) = uk ( s ) φk ,x ( x ) v( x , s ) = vk ( s ) φk ( x ) w( x , s ) = wk ( s ) φk ( x ) , (1)

a variable separation where (i) (⋅),x≡d(⋅)/dx, (ii) the summation convention applies to subscript k, (iii) φ(x) is a “mode

421
displacement amplitude function”, defined along the member length, and (iv) uk(s), vk(s) and wk(s) are the displacement
profiles associated with deformation mode k. The whole set of GBT deformation modes can be divided into three families: (i)
conventional, (ii) shear and (iii) transverse extension modes (Silvestre and Camotim 2003, Silvestre 2005). The conventional
modes constitute the original core of GBT, were developed by Schardt (1989) and are based on Vlasov’s null membrane shear
strain (γM
xs=0) and transverse extension (εss=0) assumption − they exhibit warping displacements varying linearly within each
M

wall mid-line and comprise global (extension, major and minor axis bending, torsion), distortional and local-plate modes
− the last ones are obviously associated with null warping displacements. The shear modes deal exclusively with the non-
linear variation of the warping displacements along the various cross-section wall mid-lines and involve no cross-section in-
plane deformation − since the membrane shear strains are non-null in each wall mid-line (γM xs=∂u/∂s≠0), the shear modes do not
comply with Vlasov’s first assumption. Because they are included in a GBT analysis independently from the conventional
modes (all shear undeformable), their joint participation in the deformed configuration provides a very clear assessment
of the influence of shear deformation on the member post-buckling behavior. The transverse extension modes involve only
in-plane displacements (i.e., have null warping) and exhibit non-null membrane transverse extensions (εM ss=∂v/∂s≠0) − these
deformation modes are coupled with the conventional ones and their joint participation accounts for the “bowing effect”
due to wall transverse bending, a phenomenon linked to the in-plane (local) cross-section deformation and only relevant in the
advanced post-buckling stages. For illustration, consider the cross-section discretization shown in figure 1(b) (6 natural and 7
intermediate nodes: n=6 and m=7), later adopted to analyze the post-buckling behavior of “cross-ply orthotropic” lipped
channel members dealt with in this work (see figure 1(a)) − it leads to (i) 13 conventional (4 global, 2 distortional, 7 local-plate),
(ii) 10 shear and (iii) 15 transverse extension deformation modes − their most important features are displayed in figures 2-4.

5 6 7 8 9
0.3 cm E1 =45 GPa
E2 =8 GPa
6 cm
G12 =4 GPa
Natural
3×0.1 cm 4 10
[0º, 90º, 0º] ν12 =0.3 Intermediate
ρ =2.1 gcm
-3 d
8 cm
3 1;2 12 ; 13 11
1 cm (a) 1 cm (b)
Figure 1: Orthotropic lipped channel members: cross-section (a) geometry and (b) GBT discretization

1 2 3 4 5 6

(a) (b)

7 8 9 10 11 12 13

(c)
Figure 2: Conventional mode in-plane shapes: (a) global, (b) distortional and (c) local-plate

14 15 16 17 18

x(u)

19 20 21 22 23

Figure 3: Warping displacement profiles of the shear deformation modes

dx

24 25 26 27

28 29 30 31

Figure 4: In-plane shapes of the first 8 (out of 15) transverse extension deformation modes

422
After completing the cross-section discretization procedure (i.e., knowing the deformation mode displacements), determining
the member post-buckling behavior involves the solution of the one-dimensional problem defined by the non-linear system

C kh ( φ k − φ k ),xxxx + H kh ( φ k − φ k ),xxx − Dkh ( φ k − φ k ),xx + Fkh ( φ k − φ k ),x + Bkh ( φ k − φ k ) −

, (2)
− C kjh ( φ k ,xxφ j ,x − φ k ,xxφ j ,x ),x + 12 C hjk ( φ k ,xφ j ,x − φ k ,xφ j ,x ),xx + 12 C kijh ( φ k ,xφi ,xφ j ,x − φ k ,xφi ,xφ j ,x ),x + h.o.t . = 0

where φ−k are modal amplitude functions describing the member initial geometrical imperfections − similarly to eq. (1), any
initial imperfection shape can be expressed as a linear combination of products involving deformation mode displacements one-
dimensional amplitude functions. For example, the components of the 2nd, 3rd and 4th-order C tensors appearing in (2) read

C kh = ∫ ∫ Q11 ( u k u h − z( wk u h + wh u k ) + z
2
w k wh )dzds Ckjh = ∫ ∫ Q11( uk − zwk )( v j vh + w j wh )dzds
b t b t

Ckijh = ∫ ∫ Q11( vk vi + wk wi )( v j vh + w j wh )dzds (3)

b t

and are obtained through the integration, along the cross-section mid-line and thickness, of terms involving material constants
and deformation mode displacements and/or their derivatives. While the 2nd-order quantities (Ckh, Hkh, Dkh, Fkh, Bkh) are modal
mechanical properties characterizing the cross-section linear behavior, the 3rd and 4th-order ones (Ckjh, Ckijh), as well as additional
ones related to higher order terms (not explicitly shown in eqs. (2)), are associated with its geometrically non-linear behavior.
After establishing the system (2), which includes information about (i) the cross-section modal mechanical properties, (ii) the
member length and end support conditions and (iii) the applied load nature and values, one solves it numerically by means of
the finite element method − a non-linear beam finite element has been formulated, following the work of Silvestre and
Camotim (2003) developed in the context of cold-formed steel members, which adopts Hermite and Lagrange polynomials to
approximate the modal amplitude functions. The solution of the ensuing system of non-linear algebraic equations is obtained
through an incremental-iterative technique based on (i) the well-known Newton-Raphson method and (ii) an arc-length control
strategy. The post-buckling results include equilibrium paths describing the evolution of the member deformed configuration
(amplitude functions φk(x)) with the applied load parameter λ. A detailed account on the (beam finite element) numerical
implementation of the non-linear GBT formulation can be found in the recent work of Silvestre (2005).

3. ILLUSTRATIVE EXAMPLE
In order to illustrate the application and capabilities of the proposed GBT formulation, some numerical results are presented
and briefly discussed next. They concern the distortional post-buckling behavior of two identical FRP laminated plate lipped
channel members (i) length L=40 cm, (ii) submitted to uniform axial compression, (iii) having locally/globally pinned and
free-to-warp end sections, (iv) with the material properties and cross-section dimensions given in figure 1(a), (v) containing
critical-mode initial geometrical imperfections with amplitudes v0=±0.15·t (t is the wall thickness and v0 is the outward or
inward motion of the mid-span flange-lip corners), and (vi) discretized into 8 beam finite elements (the cross-section
discretization was already shown in figure 1(b)). It is worth noting that the column geometry selected ensures that (i)
buckling occurs in a single-wave distortional mode and that (ii) local-plate/distortional mode interaction effects are not
relevant − the ratio between the minimum local-plate and distortional buckling loads is 1.6.

Figure 5(a) shows the post-buckling equilibrium paths σ /σcr.D vs. v/t of the two columns (σcr.D=38.6 MPa is the critical
buckling stress and v is the additional outward/inward mid-span flange-lip corner motion − i.e., it does not include the initial
value), hereafter identified by the terms “outward” and “inward” − also included, for comparison purposes, are the results
yielded by shell finite element analyses performed in ABAQUS (HKS 2002). Figure 5(b), on the other hand, displays modal
participation diagrams providing the contributions of the various GBT deformations modes (depicted in figures 2-3 and 4) to
various column deformed configurations located along the post-buckling equilibrium paths. Finally, figure 5(c) provides the
post-buckling evolution of the mid-span normal stress distribution along the mid-lines of the outward and inward column outer
layers (longitudinally aligned fibers). A close observation of the above figures leads to the following conclusions:
(i) First of all, one instantly recognises the important role played by the initial imperfection “sign” − indeed, the inward column
post-buckling stiffness and strength are larger than their outward column counterparts by a non-negligible amount. Like in
isotropic members (Silvestre and Camotim 2003), these distortional post-buckling asymmetry stems mostly from the
different contributions of the shear modes 15+19+23 to the outward and inward column deformed configurations.
(ii) Then, one also notices the virtual coincidence between the GBT-based equilibrium paths and the post-buckling results
yielded by the ABAQUS analyses. Moreover, the fact that the GBT analyses never involved more than 450 degrees of
freedom provides clear evidence of the high computational efficiency of the GBT-based approach − the corresponding
(and similarly accurate) shell FEM results required column discretization into very refined shell element meshes.

423
(iii) In the initial pre-buckling stages, the normal stress distribution is uniform (i.e., mode 1 fully dominates). As post-
buckling progresses, and regardless of the v0 sign (outward or inward motions), the normal stress distribution becomes
non-linear in the web and flanges, mostly due to the influence of the shear modes 15+19+23.
(iv) In the outward column post-buckling stages, the contributions of modes 5 and 15 account for the fact that the compressive
stresses (iv1) increase near the web-flange corners and (iv2) decrease in the vicinity of the flange-lip corners. Moreover,
for σ /σcr.D ≥ 1.0 tensile stresses start to develop around the flange-lip corners and rather high compressive stresses appear
close to the lip free ends. Conversely, the inward column exhibits a compressive stress increase near the flange-lip corners
(the whole flanges are under compression) and high tensile stresses develop in the neighbourhood of the lip free ends −
however, these tensile stresses are lower than the compressive ones appearing in the outward column. This last fact is due
to the relevant participation of mode 19, which reinforces mode 5 in the outward column and opposes it in the inward one.

σ /σcr.D OUTWARD INWARD

100% 100%
1.4
FEM (ABAQUS)
1 5 15
7 1 5 7
1.2 19+23 15+19+23
0 26+28 0 26+28
0.6 0.9 1.2 1.5 0.6 0.9 1.2 1.5
σ /σcr.D (b) σ /σcr.D
1.0
σ (MPa) OUTWARD σ (MPa) INWARD
100
0.8 400
OUTWARD INWARD 0 40 80 120
0
300
0.6 -100 σ /σcr.D
σ /σcr.D 200 0.52
0.52 Z (cm)
0.87
-200 0.87 1.00
0.4 0.15
1.00 0.1 σ 100 1.17
1.17
-300 0.05 0 40 80 120
x 0 0
0.2
-400 -0.05
-100
-0.1
v/t -0.15
0.0 -500 -200
0 5 10 15 20
(a) (c)
Figure 5: Outward and inward column (a) distortional post-buckling equilibrium paths σ/σcr.D vs. v/t, (b) modal
participation diagrams providing the evolution of the deformed configurations and (c) evolution of the mid-
span normal stress distribution along the outer layer mid-line σ (x=20cm; z=0.1cm; s)

4. CONCLUDING REMARKS
The main steps and procedures involved in deriving and numerical implementing (beam finite elements) a geometrically non-
linear orthotropic formulation based on Generalized Beam Theory were briefly described. In order to illustrate its application
and capabilities, numerical results concerning the distortional post-buckling behavior of lipped channel FRP composite columns
were presented and discussed. For validation purposes, some GBT-based results were compared with values yielded by ABAQUS
shell finite element analyses (involving many more d.o.f.)− a virtually perfect match was found in all instances.

5. ACKNOWLEDGEMENTS
The first author gratefully acknowledges the financial support provided by FCT − Fundação para a Ciência e Tecnologia
(Portugal), through scholarship nº SFRH/BD/21439/2005.

6. REFERENCES
Hibbit, Karlsson and Sorensen Inc. (HKS) (2002). ABAQUS Standard (version 6.3-1).
Schardt, R. (1989). Verallgemeinerte Technische Biegetheorie, Springer Verlag, Berlin. (German)
Silvestre, N. (2005). Generalized Beam Theory: New Formulations, Numerical Implementation and Applications, Ph.D. Thesis,
Civil Engineering Department, IST, Technical University of Lisbon. (Portuguese)
Silvestre, N. and Camotim, D. (2003). “Non-linear generalised beam theory for cold-formed steel members”, International
Journal of Structural Stability and Dynamics, Vol. 3, No. 4, pp. 461-490.

424
Third International Conference on FRP Composites in Civil Engineering (CICE 2006)
December 13-15 2006, Miami, Florida, USA

FATIGUE CRACK GROWTH SIMULATION FOR CFRP BONDED

STEEL PLATES USING BOUNDARY ELEMENT METHOD
Hongbo Liu, Xiao-Ling Zhao, Riadh Al-Mahaidi
(Department of Civil Engineering, Monash University, Victoria 3800, Australia)

ABSTRACT
Composite repair has been verified to be an efficient and economical method to extend the service life of cracked
steel components. There is a need to develop an accurate tool for investigating the stress intensity factor in the
cracked steel structure after repair. In this paper the crack propagation of steel plates bonded with CFRP sheets is
studied numerically using the boundary element method. The composite patch and the cracked steel plate are
simulated using surface elements, whereas the adhesive layer is simulated as interface elements to connect the patch
and steel plate. The numerical results are compared with the experimental results from previous research. The
influences of the boundary conditions, CFRP modulus and the spring stiffness on the results are discussed.

KEYWORDS
CFRP (Carbon Fibre Reinforced Polymer), Crack Propagation, Fatigue Life, Boundary Element (BE) Method.

1. INTRODUCTION
Adhesively bonded composite repair has proved to be an efficient and cost-effective method to prevent or retard the
crack re-initiation or crack propagation in civil engineering structures (Domazet, 1996; Bassetti et al., 1998; Jones
and Civjan, 2003; Tavakkolizadeh and Saadatmanesh, 2003) Carbon fiber reinforced polymer (CFRP) materials
have a very high directional stiffness, high strength, low density, high failure strain and durability under cyclic
loading. These advantages make CFRP very effective in prolonging the fatigue life of structures. In Bassetti’s study
(2000), the CFRP patches could increase the fatigue life of the un-reinforced steel plate by three times. With the
wide application of composites there is a need for research towards a better understanding of repair effectiveness
and improvement of repair design. To reduce the high costs involved in experimental methods, investigations based
on numerical analysis, utilising the increase in computational power, can be used together with experimental
calibration. In this paper the BEM was used to analyse cracked steel plates repaired with CFRP patches and its
accuracy was assessed by comparing the results with experimental findings from previous research by Colombi et al.
(2005). A parametric study was also performed to investigate the sensitivity of the stress intensity factor at the crack
tip to variations in the patch modulus, adhesive stiffness and boundary conditions.

2. BOUNDARY ELEMENT MODEL

A schematic view of the specimens’ geometry reported in Colombi et al. (2005) is shown in Figure 1(a). The base
plate is of grade FeE 235-C steel, with a yield stress of 292MPa and an ultimate tensile strength of 374MPa. In the
centre of the plate there is a notch consisting of a 10mm hole and two 5mm long initial transverse cracks. Four
1.2mm thick CFRP strips were bonded on both sides of the plates. The carbon fibre used was Sika CarboDur S512,
which has a Young’s modulus of 174GPa in the fibre direction. Structural adhesive was used to bond the composites
to the steel plates. The adhesive is a two-component epoxy, with an elastic modulus Ea of 714MPa and thickness Ta
of 0.3mm. For this kind of structural adhesive, its shear modulus is about 1000MPa. All the specimens were tested
to failure by constant amplitude tensile loading with a stress range of 80MPa and stress ratio of 0.4 (Colombi et al.
2005).

425
 (a) (b)
Figure 1: (a) Crack emanating from a centre hole reinforced by composite patch (lengths in [mm])
(b) Boundary element model of cracked steel plates with repair patches

The numerical study was undertaken by using BEASY Fatigue and Crack Growth software, which was developed
on the basis of boundary element theory and proved well suited for simulating fracture and crack propagation (Beasy,
2005). The specimen was modelled three-dimensionally and due to symmetry only half of the specimen was meshed,
as shown in Figure 1(b). The steel plate and each CFRP strips were modelled as continuous plates composed of
several zones. Quadrilateral elements were placed on all external boundaries and interface surfaces. The centre hole
in the steel plate was meshed and the 0.1mm wide slot was defined as the initial crack using BEASY fracture wizard.
It provides a simple method of adding the crack into the model by simply selecting a crack type from the library and
supplying orientation data. Material properties used in the modelling were as reported previously for the test
specimens. According to the method introduced by BEASY, the structural adhesive used to bond the composites to
the steel plates was simulated as interface elements providing both in-plane and out-of-plane stiffness, as illustrated
in Figure 1(b). They were represented as zone interfaces with internal spring boundary conditions. The stiffnesses
are defined as follows:
G G E
Kt = a , Ku = a , Kn = a
Ta Ta Ta
where Ga is the adhesive shear modulus, Ta is the bond line thickness, Ea is the Young’s modulus of the adhesive. In
practice the adhesive materials are critical to performance and a lot of failures are caused by the interface debonding.
Therefore it is essential to define the shear stiffness to realise the patch effectiveness.

3. ANALYSIS
To simulate the notched steel plate without CFRP composites Model 0 was designed. It was shown in Figure 2 that
good agreement was achieved between the numerical and experimental results. Therefore the validity of the
boundary element method was established. In the parametric study the effect of the spring stiffness representing the
adhesive, the CFRP modulus and the boundary conditions were considered. In total, eight different models were
analysed. Models 1, 2, 3 were designed to analyse the influence of in-plane spring stiffness. Compared to Model 1,
Models 2 and 3 have in-plane stiffness two and three times that of Model 1. The elastic modulus of CFRP and
boundary conditions in Models 1, 2, 3 are the same as shown in Table 1. Model 4 is identical to Model 2 except that
the CFRP elastic modulus was doubled. In this study all the models were subjected to uniform cyclic loads
(representing the constant amplitude fatigue loading) applied over the cross section at two ends. Three models, 5, 6
and 7, were used to study the influence of boundary conditions. Models 5 and 6 were developed on the basis of
Model 1. In addition to the tractions in the x-direction, representing the cyclic loads, displacements restrains in the
y- and z- directions were applied at the end patches of Models 5 and 6. Models 6 and 7 were compared with Models
5 and 2 in order to investigate the effect of boundary conditions on the symmetry plane. There are two methods to
define the symmetrical condition. One is to use the symmetry command available in BESAY. The other is to define
a set of elements on the symmetry plane and enforce the symmetry by the use of displacement restrains in the y-
direction and springs in the z- direction. All the variables of the models are detailed in Table 1, in which x y and z
represent the longitudinal, transverse and thickness directions, respectively.

426
 Table 1: Variables of the Parametric Studies

Model Spring Stiffness CFRP Boundary Condition

No. Kt, Ku Kn Modulus On Symmetry Plane At the end patches
Model 0 - - - Symmetry command used Traction of 133.3MPa in x
Model 1 3333MPa 2380MPa 174GPa Zero displacements in y and Traction of 133.3MPa in x
spring of 5000MPa in z
Model 2 6666MPa 2380MPa 174GPa As above Traction of 133.3MPa in x
Model 3 13332MPa 2380MPa 174GPa As above Traction of 133.3MPa in x
Model 4 6666MPa 2380MPa 350GPa As above Traction of 133.3MPa in x
Model 5 3333MPa 2380MPa 174GPa As above Traction of 133.3MPa in x,
0 displacement in y and z
Model 6 3333MPa 2380MPa 174GPa Symmetry command used As above
Model 7 6666MPa 2380MPa 174GPa Symmetry command used Traction of 133.3MPa in x

4. RESULTS AND DISCUSSION

The curves of crack propagation versus fatigue cycle numbers are plotted in Figure 2. The curves without any
notations represent the experimental results. It can be concluded that the application of CFRP strips is effective in
prolonging the fatigue life of the notched steel plates and the results of boundary element analysis (Models 0, 1, 2, 3,
5, 7) match well with the experimental results.

Figure 2: Comparison of numerical and experimental results

Figure 2 shows that for the steel only model, the numerical curve is very close to the experimental one. The
boundary element method predicted its crack growth and fatigue life accurately. For the CFRP repaired steels, at the
initial stage of the crack growth, Models 1, 2, 3 and 7 provided very accurate predictions. With the crack
propagating, the curve of Model 5 closely follows the experimental results.

As mentioned previously, Model 5 was developed on the basis of Model 1. It had additionally applied displacement
restrains in the y- and z- directions. Their curves in Figure 2 showed that Model 5 has a much steeper curve than
Model 1. When the traction acts in the longitudinal direction of the model, the entire volume had a tendency of
contracting in the transverse direction, which was held back by the displacement restrains. Therefore more energy
was needed to get the same crack propagation. The stress intensity factor was increased and the crack grew faster
when there were displacement restrains. In Figure 2, Models 6 and 7 was designed to compare with Models 5 and 2
respectively. The slight difference in results revealed that the methods to apply the boundary conditions on

427
symmetry plane had no effect on the results. So in BEASY the displacement restrains was recommended to be
applied at the end patches and the symmetry command was proved to be a convenient way in setting up the model.

Comparing the curves of Model 1, 2 and 3, it was shown that as the in-plane stiffness of the spring increased more
energy was required for crack growth. This caused a reduction in crack growth rate. For the CFRP repaired steel
plates, the stresses were carried by both the CFRP and steel. The load transfer between them occurred through the
interface spring, representing the adhesive. As the stiffness value increased, the spring performed more efficiently
and thus additional loads were potentially transferred to the CFRP from the steel plate. Therefore the stresses in the
steel were reduced, which lead to a small value of stress intensity factor at the crack tip. As a consequence, the crack
growth rate was slowed down. For this application of structural adhesive, the in-plane spring stiffness of 6666, two
times the value of shear modulus divided by the bond line thickness, showed the best agreement with the
experimental curves.

From the comparison between Model 4 and Model 2, it was concluded that the model with higher elastic modulus of
CFRP materials had a slower growth rate. In general, when the loads are applied on the composites, both the fibres
and epoxy are subjected to the same strain while the stresses in the two phases differ depending on their volume
fraction and the elastic moduli. The high modulus CFRP reflects high stiffness. When the same strain was
introduced in the structural adhesive, high stiffness fibers carried more stress compared with low stiffness CFRP.
Therefore the steel plates repaired with high modulus CFRP can carry more loads without increasing the stress level
in the steel plate. Thus the stress intensity factor on the crack tip is effectively reduced and the crack growth rate is
slowed down.

5. CONCLUSIONS
Three-dimensional boundary element analyses of composite patch repaired steel plates were performed. The
variables that affect the fatigue life of the numerical models were also discussed. From the results the following
conclusions are made:
• The boundary element analysis could be used to simulate the crack propagation of the adhesive bonded steel
plates and its results agree well with the experimental results.
• Interface elements can simulate the action of structural adhesive efficiently. Different in-plane stiffness for
the interface elements results in different crack growth rates. The spring constants in the interface elements
evaluated in this study are limited to one type of adhesive and one thickness. Their properties were provided
in Colombi et al. (2005).
• When CFRP is used to repair the cracked steel plates, a higher value of elastic modulus of CFRP seems much
more efficient.
• Displacement restrains are recommended to be applied at the end patches when using the Boundary Element
Method.

6. REFERNCES
Bassetti, A., Liechti, P. and Nussbaumer, A. (1998). "Fatigue Resistance and Repairs of Riveted Bridge Members",
Fatigue Design 1998, Espoo, Finland, pp 535-546.
Bassetti, A., Nussbaumer, A. and Manfred, A. (2000). "Crack Repair and Fatigue Life Extension of Riveted Bridge
Members Using Composite Materials", Bridge Engineering Conference ESE-IABSE-FIB, pp 227-238.
Beasy (2005). Crack Growth Guide, Acoustic Guide, Corrosion and CP Guide, Computational Mechanics BEASY,
Ashurst, Southampton.
Colombi, P. (2005). "Plasticity induced fatigue crack growth retardation model for steel elements reinforced by
composite patch". Theoretical and Applied Fracture Mechanics, Vol. 43, No. 1, pp. 63-76.
Domazet, Z. (1996). "Comparison of Fatigue Crack Retardation Methods". Engineering Failure Analysis, Vol. 3, pp
137-147.
Jones, S.C. and Civjan, S.A. (2003). "Application of Fiber Reinforced Polymer Overlays to Extend Steel Fatigue
Life". Journal of Composites for Construction, Vol. 7, pp 331-338.
Tavakkolizadeh, M. and Saadatmanesh, H. (2003). "Fatigue Strength of Steel Girders Strengthened with Carbon
Fiber Reinforced Polymer Patch". Journal of Structural Engineering, ASCE, Vol. 129, pp 186-196.

428
 NONLINEAR FINITE ELEMENT MODELING OF RC BEAMS
STRENGTHENED WITH DIFFERENT CFRP SCHEMES
Rajai Z. Alrousan
(Ph.D. Student, University of Illinois at Chicago, Chicago, IL, USA)

Mohammad A. Alhassan
(Ph.D. Student, University of Illinois at Chicago, Chicago, IL, USA)

Mohsen A. Issa
(Professor of Structural and Materials Engineering, University of Illinois at Chicago, Chicago, IL, USA)

ABSTRACT
This paper presents the effect of the number and configuration of CFRP sheets on the flexural strength and ductility
of RC beams and provides useful relationships that can be effectively utilized to determine the required number of
CFRP sheets for a certain necessary increase in the beam flexural strength without a major loss in its ductility. The
reinforced concrete beams are identical in the geometric and reinforcement details. The variables in this paper are
the number and configuration of CFRP sheets. The RC beams were modeled and analyzed using the nonlinear finite
element ANSYS (Version 9) software package. The FEA results were validated with the experimental test results of
identical RC beams. The validation of the FEA results with the experimental test results concludes that the
nonlinear FEA modeling could predict accurately the ultimate load capacity, deflections, concrete strains, steel
strains, and CFRP strains of the CFRP strengthened slabs. Analysis of the FEA shows that the normalized flexural
moment capacity (M/Mo) of the beams strengthened with 1 layer (t/d = 0.00081) and 5 layers (t/d = 0.00406) of the
U-wrap scheme were 1.7 and 2.7 times that of the control beam, respectively. For the corresponding beams
strengthened with the tension face scheme, the M/Mo was 1.25 and 1.8 times that of the control beam, respectively.
Approximately 5 layers of the tension face scheme are equivalent to 1 layer of the U-wrap scheme. In both
schemes, it was confirmed that after a certain t/d value, there would be no further noticeable increase in the flexural
strength of the beam, while significant reduction in its ductility continues to occur.

KEYWORDS
CFRP, Strengthening, FEA, t/d ratio, RC beams, Flexural Strength Capacity, Ductility.

1. INTRODUCTION AND OBJECTIVES

Carbon fiber reinforced polymer (CFRP) composites are one of the most economical and reliable methods of
strengthening deficient structural members with high corrosion and fatigue resistant, high strength/weight ratio, and
ease of application. The critical factor for commercial applications of CFRP composites is the cost. Various studies
have been conducted to investigate the static and fatigue behaviors of CFRP-strengthened reinforced concrete
members (Issa et al., 2003, Shahway et al., 2001, and Kachlakev et al., 2000). The use of the finite element analysis
(FEA) to simulate the response of RC members provides substantial advantages mainly in terms of reduction in the
cost and time as well as the ability to model any strengthening scheme and sophisticated geometry. This paper
reports on nonlinear finite element analysis (FEA) of reinforced concrete (RC) beams strengthened with different
number and configurations of CFRP sheets. The main objectives were to assess the effect of the number and
configuration of CFRP sheets on the flexural strength and ductility of RC beams and to provide useful relationships
that can be effectively utilized to determine the required number of CFRP sheets for a necessary increase in the
beam flexural strength without major loss in its ductility. The FEA results were validated with experimental test
results of identical specimens. Detailed modeling methodology and practical results are provided in the following
sections.

429
2. FINITE ELEMENT ANALYSIS METHODOLOGY
2.1 Geometric and Reinforcement Details of the Beams

In this study, nonlinear finite element modeling and analysis was carried out using ANSYS software for eleven RC
beams identical in the geometric and reinforcement details while having different number and configuration of
CFRP sheets. The beams are 2.44 m (8 ft) long with a cross section of 150x230 mm (6x9 in.), and reinforced with 3
#4 steel bars at the bottom, 2-3/16" in diameter bars at the top, and 3/16" in diameter stirrups spaced at 75 mm (3
in.). This reinforcement design ensures flexural failure mode. The beams included a control beam (without CFRP)
and 2 groups of strengthened beams, each included 5 beams strengthened with 1, 2, 3, 4, and 5 layers of CFRP
sheets. The first group (CFRP applied on the entire tension face) was designated as tension face strengthening
scheme, while the second group (CFRP applied on the entire tension face and both sides) was designated as U-wrap
strengthening scheme. The exact geometric and reinforcement details of the beams were employed in the FEA
modeling.

2.2 Element Types and Material Properties

SOLID65 element was used to model the concrete. This element is typical for the 3-D modeling of solids with or
without reinforcing steel bars. The most important aspects of the SOLID65 are the treatment of nonlinear material
properties and the capability of cracking in 3 orthogonal directions, crushing, plastic deformation, and creep. The
ultimate tensile and compressive strengths are required to define a failure criterion for the concrete. The shear
transfer coefficient (βt) represents the condition of the crack face. The available literature showed a range of values
for βt (0.05-0.25), and the most common value of 0.2 was used in this study. The 28-day compressive strength of
concrete (f 'c) of 55 MPa (8000 psi), young's modulus (Ec) of 35063 MPa (5100 ksi), and Poisson’s ratio of 0.2 were
used. The steel reinforcement was assumed to be an elastic-perfectly plastic material identical in tension and
compression. The 3-D LINK8 element was used to model the steel reinforcement with Poisson’s ratio, elastic
modulus, and yield stress of 0.3, 200 GPa (29,000 ksi), and 413 MPa (60 ksi), respectively. The SOLID45 element
was used to model the supports and loading steel plates (75x50x25 mm). The layered SOLID46 element was used
to model the CFRP sheets (0.165 mm thick) and the epoxy (0.835 mm thick) materials. The SOLID46 element
allows up to 250 different layers of orthotropic materials with different orientations. The CFRP was assumed to be
an orthotropic material having a tensile strength of 4272 MPa (620 ksi), an elastic modulus of 228 GPa (33,000 ksi),
and an ultimate tensile strain of 0.0167 in the fibers direction. In the directions perpendicular to the fiber direction,
the elastic modulus of CFRP was assumed to be 10-6 times that of the main direction. The ultimate tensile strength
of the epoxy was 55 MPa (8000 psi), elastic modulus of 30 GPa (440 ksi), and ultimate tensile strain of 0.03. Linear
elastic properties for the CFRP and epoxy were assumed. The SOLID65, SOLID45, SOLID46 are defined by 8
nodes, while LINK8 is defined by 2 nodes, and all of the four elements have 3 degrees of freedom at each node
(translations in the nodal x, y, and z directions). Real constant set 1 was used for the SOLID65 element. It requires
material number (type of the reinforcing rebars), volume ratio (steel/concrete ratio in the element), and orientation
angles (orientation of the reinforcement). In this paper, the beams were modeled using discrete reinforcement, i.e. a
value of zero was entered for all of the real constants of SOLID65, which turned the reinforcement capability of the
SOLID65 element off. The cross sectional areas of the main steel and stirrups were entered for the LINK8 element
as real constant sets 2 and 3, respectively. Real constant sets 4 and 5 were used for the epoxy and the CFRP
materials, respectively.

2.3 Meshing and Boundary Conditions

Square and rectangular elements were created for the rectangular volumes (concrete, CFRP, epoxy, and steel plates)
using the volume-mapped command. This properly sets the width and length of the steel reinforcement elements to
be consistent with the elements and nodes of the concrete. A convergence study was carried out to determine the
appropriate mesh density as shown in Fig. 1. The meshing of the reinforcement was a special case and the
individual elements were created in the modeling process as shown in Fig. 1. However, the necessary mesh
attributes for the concrete were set before each section of the reinforcement was created. The SOLID46 elements
for epoxy and CFRP layers have the same meshing as SOLID65 elements for concrete to allocate the node over the
node of each element. The command merge item was used to merge separate entities that have the same location
into single entities. To reduce the computation time and disk space requirements and since there are symmetries in
the cross-section of the RC beams and in the applied load about 2 planes, a quarter of each bream was modeled with
proper boundary conditions. The models were 1.22 m (48 in.) long having a cross-section of 75x230 mm (3x9 in.).

430
To ensure proper modeling, displacement boundary conditions were applied at the planes of symmetry. The
symmetry boundary conditions were set first. Nodes defining a plane through the beam cross-section at the center of
the beam define one plane of symmetry. The nodes on this plane were constrained in the perpendicular direction.
These nodes, therefore, have a degree of freedom constraint UX = 0. Second, all nodes selected at Z = 0 define
another plane of symmetry. These nodes were given the constraint UZ = 0. The support was modeled as a roller
that allows the beam to rotate at the support. The applied force was applied across the entire centerline of the steel
plate. The beams were analyzed simulating 4-point loading case with the distance between the 2-point of loading is
610 mm (24 in.). The total applied load was divided into a series of small load increments, each 0.45 kN (0.1 kip).
The Modified Newton–Raphson equilibrium iterations were used to check the convergence at the end of each load
increment within a tolerance value of 0.001. The static analysis type was utilized to obtain the response of the
beams. The model failure was identified when the solution of 0.0045 kN (0.001kips) load increment was not
converging.
Tension face scheme

U-wrap scheme

Figure 1: Meshing and Reinforcement Configuration for Quarter of the Beam

3. FINITE ELEMENT ANALYSIS RESULTS

The FEA results were validated with experimental test results of identical beams as shown in Fig. 2. The load
deflection curves, load strain curves, ultimate flexural strengths, and ultimate displacements obtained from the FEA
were in good agreement with the experimental test results. Fig. 2 shows typical load deflection curves for control
beams and CFRP-strengthened beams with 1, 2, 3, 4, and 5 layers on the tension face and U-wrap strengthening
schemes. Additional tables and figures that confirm the good agreement between the FEA and the experimental
results will be presented at the conference. Fig. 2 reveals that for both strengthening schemes, increasing the
number of CFRP layers leads to a non-proportional increase in the flexural strength capacity of the beam and to a
reduction in the ultimate displacement.

Figure 2: Load-Deflection Curves for the Control and Strengthened RC Beams

For further illustration, the normalized thickness of the CFRP sheet layers/effective beam depth ratio (t/d) was
plotted versus the normalized flexural moment capacity (M/Mo) and versus the normalized ultimate displacement
(Δ/Δo) as shown in Fig. 3. Logarithmic relationships were observed with high correlation coefficients as shown in
Eqs. 1 and 2 below:

431
 M t
= 0.6078Ln( ) + 5.985 (R 2 = 0.9894) (1)
Mo d

M t
= 0.3493Ln( ) + 3.7398 (R 2 = 0.9862) (2)
Mo d
Depending on the effective depth (d) of a RC beam having similar materials properties, these relationships can be
utilized to predict the required thickness of CFRP sheet layers (t) for a required increase in the flexural moment
capacity (M/Mo). The M/Mo of the beams strengthened with 1 layer (t/d = 0.00081) and 5 layers (t/d = 0.00406) of
the U-wrap scheme were 1.7 and 2.7 times that of the control beam, respectively. For the corresponding beams
strengthened with the tension face scheme, the M/Mo values were 1.25 and 1.8 times that of the control beam,
respectively. The FEA results show that 5 layers of the tension face scheme are equivalent to 1 layer of the U-wrap
scheme in terms of the improvement in the flexural moment capacity. Also, the reduction in the ultimate
displacement is smaller and more stabilized for the beams strengthened with the U-wrap scheme than the beams
strengthened with the tension face scheme. For t/d = 0.00081, the ultimate displacement of the beam strengthened
using the U-wrap scheme was similar to the control beam, while it was about 90% of the ultimate displacement of
the control beam for the beam strengthened using the tension face scheme. Using t/d = 0.00406, the reduction in the
ultimate displacement was about 20% using the U-wrap scheme, while 35% using the tension face scheme. After a
certain t/d ratio, there would be no further significant increase in the flexural moment capacity of the beam, while
significant reduction in its ductility continues to occur.

Figure 3: Normalized Ultimate Flexural Moments and Ultimate Displacements

4. CONCLUSIONS
A good agreement was observed between the FEA and the experimental test results. This concludes that FEA
modeling could accurately predict the ultimate load capacity and deformation of CFRP strengthened slabs. The U-
wrap CFRP-strengthening scheme improves the flexural strength of RC beams while maintaining acceptable
ductility much better than the tension face scheme. After a certain t/d ratio, there would be no noticeable increase in
the flexural strength of the beam, while significant reduction in its ductility continues to occur.

5. REFERENCES
ANSYS (2005), ANSYS User’s Manual Revision 9.0, ANSYS, Inc.
Issa, M.I., Shabila, H., and Issa, Moussa A. (2003). “Structural behavior of reinforced concrete beams strengthened
with CFRP subjected to static and fatigue loading”. Advanced Composites for Concrete Repair-1, Proceedings of
the Structural Faults and Repair Conference, Commonwealth Institute, Kensington, London, UK.
Kachlakev, D.I., Miller, T.H., Yim, S., Chansawat, K., and Potisuk, T. (2000). “Linear and nonlinear finite element
modeling of reinforced concrete structures strengthened with FRP laminates,” United States Department of
Transportation, Federal Highway Administration.
Shahway, M., Beitelman, T., Arockiasamy, M., and Sowrirajan, R. (2001). “Flexural strengthening with carbon
fiber-reinforced polymer composites of preloaded full-scale girders”. ACI Structural Journal, pp. 735-742.

432
Third International Conference on FRP Composites in Civil Engineering (CICE 2006)
December 13-15 2006, Miami, Florida, USA

FINITE ELEMENT ANALYSIS OF FRP DEBONDING FROM

CONCRETE UNDERGOING GLOBAL MIXED MODE I/II LOADING
Baolin Wan
(Assistant Professor, Marquette Universit, Milwaukee, WI, USA)

Zhenyu Ouyang
(Ph.D. Student, Marquette Universit, Milwaukee, WI, USA)

ABSTRACT
This research studies FRP debonding from concrete substrate of Modified Double Cantilever Beam (MDCB)
specimen undergoing global mixed mode I/II loading. A series of 2D finite element models were built using
ANSYS 10.0 software package to simulate the MDCB test. A modified virtual crack closure technique (VCCT) was
used to extract the mode I and mode II interfacial energy release rates from the models. The finite element model
was validated and calibrated by experimental data. The relationship between loading angle and critical interfacial
energy release rate was obtained by a series of FEM analysis with different loading angles. Parametric studies were
conducted to investigate the characteristics of FRP bonded concrete system. Seven parameters were considered:
Young’s modulus of concrete strength, Young’s modulus and thickness of primer concrete layer, Young’s modulus
and thickness of adhesive layer and Young’s modulus and thickness of FRP. It was found that Young’s moduli of
the materials had relatively small effects, while the thicknesses of primer concrete, adhesive layer and FRP had
significant effects on the interfacial energy release rate while FRP debonding failure happened.

KEYWORDS
Bond, FRP, concrete, finite element, modified virtual crack closure (VCCT)

1. INTRODUCTION
One of major failure modes of the fiber reinforced polymer (FRP) composite materials repaired reinforced concrete
beams is the FRP debonding failure from concrete substrates. This debonding failure limits the strength contribution
from the FRP. There are five debonding mechanism (Oehlers, 2005) including plate end (PE), critical diagonal
crack (CDC), flexural intermediate crack (FIC), shear intermediate crack (SIC) and axial intermediate crack (AIC).
PE is concrete cover delamination starting from the end of FRP due to peeling force normal to the FRP. Such
peeling force can be considered as Mode I loading based on the fracture mechanics definition. FIC and SIC
propagate due to both shear and opening forces after concrete crack. Therefore, they are under mixed Mode I and II
loading. AIC happens under pure shear loading condition which can be considered as Mode II loading. Modified
Double Cantilever Beam (MDCB) test can be used to determine the energy release rate of FRP debonding from a
concrete or masonry substrate under pure Mode I, mixed modes and pure Mode II loadings (Wan, 2002; Wan et al.,
2004). Therefore, using MDCB test can simulate PE, FIC, SIC and AIC debonding.

The virtual crack closure technique (VCCT) has been widely used to extract the Mode I and Mode II interfacial
energy release rates from finite element models (Krueger, 2002). The advantage of VCCT is that energy release
rates can be calculated in a single geometric model instead of two complete analysises. Sethuraman and Maiti
(1988) derived a modified virtual crack closure integral for square-root singularity elements in order to compute the
strain energy release rate by moving the mid-side node position of the isoparametric quadratic element to the quarter
location. This method can increase the accuracy of strain energy calculation for linear elastic analysis. In this study,
the modified virtual crack closure integral was successfully used to model FRP debonding from concrete substrate
under mixed mode loading conditions of MDCB tests. A series of FE models were built to study the effect of
different parameters on the FRP debonding behaviors.

433
2. FINITE ELEMENT MODEL
Using the modified virtual crack closure integral, the strain energy release rate, G, can be expressed in terms of the
nodal forces ahead of the crack tip and the opening displacements behind it (Sethuraman and Maiti, 1988).
(u yk − u ′yk )
GI = [Fyj + (1.5π − 4)Fyi ] (1)
∆a
(u − u ′xk )
G II = xk [Fxj + (1.5π − 4)Fxi ] (2)
∆a
where (Fxi, Fyi) and (Fxj, Fyj) represent the nodal forces at the node of crack tip and the node ahead of it, respectively;
and (uxk, uyk) and (u'xk, u'yk) represent the nodal displacements behind it as shown in Figure 1.

∆a P
α
PI

(uxk, uyk) PII

(u'xk, u'yk) 0.025 mm

α

(Fxi, Fyi) (Fxj, Fyj)

Figure 1: Elements and Figure 2: MDCB test under Figure 3: Elements around
nodes around crack tip. mixed modes loading the crack tip

Wan et al. (2004) used MDCB test frame to study the bond between CFRP and concrete undergoing global mixed
mode I/II loading. Three loading angles were used in the test program and corresponding energy release rates, GI
and GII were calculated. The loading angle was defined as the angle between the bonding line and the horizontal
plane as shown in Figure 2. It was reported (Wan, 2002) that a layer of concrete residue remaining on the fracture
surface of the detached CFRP after MDCB test. Such layer of concrete contains the primer which penetrated into
concrete through the microcracks, void and capillary pores on the superficial concrete. It was defined as primer-
concrete in this research. The typical Young’s modulus of polymer modified concrete is 4.2 × 104 MPa (Blaga and
Beaudoin, 1985) and this value was used in the models in this research. ANSYS 10.0 was used to generate 2D finite
element models and perform parametric studies in this research. The geometric and material properties used in
experimental test (Wan,2002; Wan et al., 2004) were adopted. All materials were assumed to be linear elastic. The
crack was assumed to locate between primer-concrete and concrete substrate. The element length of the first group
of elements around crack tip was 0.025 mm as shown in Figure 3. The critical energy release rates GIC and GIIC
obtained in the experimental test with its linear least-squares best fit are shown in Figure 4. The loads used in FE
models were calibrated by referring the experimental results in order to calculate the critical energy release rates. It
is shown in Figure 4 that the FE results had excellent agreement with the experimental data.

3. COMPUTER SIMULATIONS
The validated FE model presented in previous section was used as control model. The calibrated load values were
applied in the models for parametric studies. Seven parameters were selected, i.e., (a) Young’s modulus of concrete,
Ec, (b) Young’s modulus of primer concrete, Epc, (c) thickness of primer concrete, tpc, (d) Young’s modulus of
adhesive, Ea, (e) thickness of adhesive, ta, (f) Young’s modulus of FRP, Efrp and (g) thickness of FRP, tfrp. The
following paragraphs introduce each study and discuss the findings. In Figures 5 to 11, the solid line represents the
critical G values for FRP debonding from concrete substrate in control specimen. If the calculated G values fall in
the left of and below the line, it means FRP will not debond under the calibrated loads and larger loads are needed to
debond for such cases.

The Young’s modulus of concrete, Ec, was modified by ±10% and ±20% to study its effect on FRP debonding. It is
shown in Figure 5 that total energy release rate increased slightly with the decrease of Ec. However, such change
was very small. The change of total G value was less than 1% when Ec was changed up to 20%.

434
 600 600
Critical G values
Ec -20%
Experimental data
500 500 Ec -10%
Trend of experimental data
Ec +10%
FEM results
Ec +20%
400 400

GIC (J/m )

GI (J/m )
2

2
300 300

200 200

100 100

0 0
0 100 200 300 400 500 600 0 100 200 300 400 500 600
2 2
GIIC (J/m ) GII (J/m )

Figure 4: Critical G values Figure 5: Effect of Ec

The Young’s modulus of primer-concrete, Epc, was modified by ±10% and ±20%. When Epc decreased, GII
component increased while GI component kept almost same as control specimen as shown in Figure 6. Total G
value increased with the decrease of Epc. It indicated that the specimen with lower value of Epc would be easier to
debond. The average thickness of primer-concrete in experimental test was 2 mm. Four different primer-concrete
thicknesses were selected for the parametric study: 0, 1, 3 and 4 mm. When primer-concrete thickness decreased,
the energy release rate values increased significantly if the load was kept same as control model (Figure 7). It means
that actual fracture load with thinner primer-concrete would be significantly lower.

600 800
Critical G valures Control: tpc = 2 mm
tpc = 0
Epc -20% 700
500 tpc = 1 mm
Epc -10%
tpc = 3 mm
Epc +10% 600 tpc = 4 mm
Epc +20%
400
500
GI (J/m )
GI (J/m )

2
2

300 400

300
200
200

100
100

0 0
0 100 200 300 400 500 600 0 100 200 300 400 500 600 700

GII (J/m )
2 GII (J/m2)

Figure 6: Effect of Epc Figure 7: Effect of tpc

600 600
Control: ta = 1.2 mm
Critical G values
ta = 0.4 mm
Ea -20%
500 500 ta = 2.0 mm
Ea -10%
ta = 2.8 mm
Ea +10%
Ea +20%
400 400
GI (J/m )

GI (J/m )
2

300 300

200 200

100 100

0 0
0 100 200 300 400 500 600 0 100 200 300 400 500 600
2
GII (J/m ) GII (J/m2)

Figure 8: Effect of Ea Figure 9: Effect of ta

The Young’s modulus of adhesive, Ea, was modified by ±10% and ±20%. It is shown in Figure 8 that there was
almost no change of G values for the ±20% change of Ea values. Therefore, Young’s modulus of adhesive did not
have significant effect on the FRP debonding phenomenon in the scope of this study. The average thickness of
adhesive in experimental tests was 1.2 mm. Three different primer-concrete thicknesses were selected for
parametric study: 0.4, 2.0 and 2.8 mm. The adhesive thickness effect is shown in Figure 9. In the region where GI

435
value was dominant, calculated G value increased with the decrease of ta value. It indicated that the specimen with
thinner adhesive layer was easier to fracture when Mode I component was dominant. In the region where Mode II
loading was dominant, thinner adhesive layer would result lower GII value but higher GI value. However, the total
G value located on the critical line for all selected thickness in this situation. It indicated that the fracture loads for
them would be same as those for control specimen.

600 600
Critical G value Control: FRP t = 4.8 mm
Efrp -20% FRP t = 3.4 mm
FRP t = 6.2 mm
500 Efrp -10% 500
Efrp +10%
Efrp +20%
400 400

GI (J/m2)
GI (J/m )
2

300 300

200 200

100 100

0 0
0 100 200 300 400 500 600 0 100 200 300 400 500 600
2
GII (J/m ) GII (J/m2)

Figure 10: Effect of Efrp Figure 11: Effect of tfrp

The Young’s modulus of FRP, Efrp, was modified by ±10% and ±20%. It is shown in Figure 10 that G values
increased with the decrease of Efrp. Therefore, the FRP with lower Young’s modulus would be easier to debond.
The average thickness of FRP in experimental tests was 4.8 mm. The FRP thicknesses of 3.4 and 6.2 mm were
selected for parametric study: 3.4 and 6.2 mm, which were ±30% of the control specimen. It is shown in Figure 11
that thinner FRP would be easier to debond if the stress in FRP itself did not reach its ultimate strength.

4. CONCLUSIONS
This research modeled the MDCB tests using finite element method and parametric study was performed. The
following conclusions can be drawn from this study:
1. Concrete Young’s modulus did not have significant effect on the energy release rate when FRP was
debonding.
2. The specimen with lower value of Epc and thinner primer-concrete would be easier to debond.
3. Young’s modulus of adhesive did not have significant effect on the FRP debonding phenomenon in the
scope of this study. The specimen with thinner adhesive layer was easier to fracture when Mode I
component was dominant. When there was significant Mode II loading component, the fracture loads for
the specimens with different thickness of adhesive layer were same as those for control specimen.
4. If FRP had lower Young’s modulus and smaller thickness, the specimen would be easier to debond.

5. REFERENCES
Blaga, A. and Beaudoin, J.J. (1985). “Polymer Modified Concrete”, Canadian Building Digests, CBD-241,
http://irc.nrc-cnrc.gc.ca/pubs/cbd/cbd241_e.html, 04/03/06.
Krueger, R. (2002). “The virtual crack closure technique: history, approach and applications”, NASA/CR-2002-
211628, ICASE Report No. 2002-10, Hampton, Virginia, 64 pp.
Oehlers, D.J. (2005). “Generic debonding mechanisms in FRP plated beams and slabs”, Proceedings of the
International Symposium on Bond Behaviour of FRP in Structures, December 7-9, 2005, Hong Kong. pp 35-44.
Sethuraman, R. and Maiti, S.K. (1988). “Finite element based computation of strain energy release rates by modified
crack closure integral,” Engineering Fracture Mechanics, Vol. 30, pp 227–231.
Wan, B. (2002). Study of the Bond Between FRP Composites and Concrete, Ph.D. Dissertation, Department of Civil
and Environmental Engineering, University of South Carolina, 205 pp.
Wan, B., Sutton, M., Petrou, M.F., Harries, K.A., and Li, N. (2004). “Investigation of bond between FRP and
concrete undergoing global mixed mode I/II loading”, ASCE Journal of Engineering Mechanics. Vol. 130 No. 12,
pp 1467-1475.

436
Third International Conference on FRP Composites in Civil Engineering (CICE 2006) 
December 13­15 2006, Miami, Florida, USA 

FINITE ELEMENT MODELLING OF 
FRP SHEAR­STRENGTHENED RC BEAMS 
O.F.A. Otoom 
(PhD Candidate, Centre for Built Infrastructure Research,University of Technology Sydney, Australia) 

S.T. Smith 
(Senior Lecturer, Centre for Built Infrastructure Research,University of Technology Sydney, Australia) 

S.J. Foster 
(Associate Professor, School of Civil and Environmental Engineering, 
The University of New South Wales, Australia) 

ABSTRACT 
Reinforced  concrete  (RC)  beams  can  be  strengthened  in  shear  with  externally  bonded  fibre  reinforced  polymer 
(FRP) composite strips. Despite the relatively large body of research available, there remains uncertainty regarding 
the  distribution  and  magnitude  of  strains  and  stresses  in  the  concrete,  and  in  the  externally  bonded  FRP.  Of 
particular importance is the contribution of the FRP reinforcement to the shear strength at ultimate.  In this paper, a 
finite element (FE)  investigation on RC  beams strengthened in shear  with epoxy  bonded FRP  strips  is  undertaken 
and  the  results  compared  with  test  data.  The  study  shows  that  the  FE  approach  is  able  to  capture  the  overall 
behaviour of the test specimens, including debonding of the FRP strips. 

KEYWORDS 
FRP, Reinforced Concrete, Strengthening, Shear, Finite Element Modelling, External Bonding 

1. INTRODUCTION 
The shear capacity of a reinforced concrete (RC) beam can be enhanced with the addition of externally bonded fibre 
reinforced  polymer  (FRP)  strips  or  sheets  as  has  been  reported  from numerous  experimental studies  (e.g.  Khalifa 
and  Nanni 2002, Bousselham and Chaallal 2006). Tests have  found anchored  FRP strengthening  to  predominantly 
fail by rupture of the FRP, while debonding was found to occur for unanchored strips. Only a handful of papers (e.g. 
Khalifa and Nanni 2002) describing tests on FRP shear­strengthened RC beams have reported  the geometrical and 
material  properties  of  the  concrete,  steel  reinforcement  and  FRP  strengthening,  as  well  as  the  failure  mode,  in 
enough detail for the results to be meaningfully computationally analysed. 

A limited number of finite element (FE) studies are reported on the modelling of shear­strengthened RC beams with 
the  study  by Wong  and  Vecchio  (2003) being  one  of  the  more  comprehensive.  Wong  and  Vecchio  compared  FE 
predictions  with  tests  on  RC  beams,  without  steel  shear  stirrups,  strengthened  with  side  bonded  FRP.  All  beams 
failed by debonding and the bond­slip behaviour of the FRP­to­concrete was a key component in  the modelling of 
debonding failures. More recently, Smith et al. (2006) used FE to analyse the distribution and magnitude of strains 
and  stresses  in  RC  beam  containing  internal  steel  stirrups  and  strengthened  in  shear  with  FRP  strips.  Tests  are 
urgently required in order to overcome the lack of reliable experimental data  that  is prohibiting further comparison 
with FE predictions. In such tests, it is imperative that all control and material data needed as input into an analytical 
or numerical model be collected. 

In this paper the results of a FE investigation are reported on RC beams, without steel shear stirrups, strengthened in 
shear with FRP U­jackets. The FE predictions are compared with experimental results of Khalifa and Nanni (2002) 
of a FRP U­strip shear­strengthened RC beam failing by FRP debonding.

437
2. FINITE ELEMENT MODELLING OF FRP­STRENGTHENED RC BEAMS 
The non­linear reinforced concrete FE program RECAP (Foster and Gilbert 1990), updated to account for externally 
bonded  FRP,  is  used  in  this  study.  A  brief  description  of  the  elements  used  for  modelling  of  FRP  strengthened 
members is given below with a full implementation of the model in Khomwan and Foster (2004a, 2004b). 

Modelling of repaired concrete beams where the repair is on the sides of the beam can be undertaken by attaching a 
layer of repair elements to the beam via  a  layer of interface elements (Figure 1a). Side face repaired beams can be 
modelled using either a 2D element formulation, e.g. as presented by Khomwan and Foster (2004a, 2004b), or using 
1D discrete bar elements with an area equal to the thickness of the FRP times the width over which it acts. 

Concrete 

+  + 
u n 
1  un 
2  t  b,max 

Bond Stress,  t  b
ut + 
1  ut + 2 
3  4  4  3 
-  - 
1D ­ Interface 
un 
1  -  un 
2  - 
3  4 1  ut 1  2  u t 2 
1  2  t b  s  n 
L  = ×
FRP  t b , max  s max  (n - 1 ) + (s / s max ) n 
1  2 
s max 
(a) Connection of FRP to  (b) interface degrees of  Slip, s 

concrete via interface element  freedom 
Figure 1. Concrete, steel and interface elements  Figure 2. Bond stress­slip model 
Bar elements are used to model the FRP and a 4­node interface  element, shown in Figure 1b, is used to model slip 
between  the  FRP  and  the  concrete.  To  maintain  compatibility  between  the  FRP  and  the  concrete  in  the  normal 
direction, a stiff elastic modulus is used. In the tangential direction Popovics (1973) equation, as shown in Figure 2, 
is used to represent the local bond stress tb  versus slip s relationship where tb,max  is the maximum local bond stress; 
smax  is the slip at t b,max; and n is a constant. The parameters tb,max, smax  and n in the current study were matched to the 
ascending portion of Lu et al.’s (2005) generic bond­slip relationship using concrete and FRP properties of the  test 
beam described in Section 3. The results of the FE model were not found to be overly sensitive to the shape of the 
descending curve. For stability of the solution process, the bond stiffness (in the tangential direction) is taken as the 
secant  stiffness  in  a  modified Newton­Raphson  solution  procedure  with  more  details  in  Chong  et  al.  (2004).  The 
cracked membrane model finite element (CMM­FE) formulation of Foster and Marti (2003) was used for the RC. 

3. COMPARISON OF FE RESULTS WITH TEST DATA 
The  FE  model  is  compared  with  tests  undertaken  by  Khalifa  and  Nanni  (2002)  on  RC  beams  strengthened  with 
externally bonded FRP U­jackets strips that failed by debonding of the FRP. 

3.1. Test Beams and FE Mesh 

Khalifa and Nanni (2002) tested 12 RC beams of which four were unstrengthened reference beams and seven were 
strengthened in shear with different arrangements of externally bonded FRP U­jacket strips or continuous U­jacket 
sheets. Other variables investigated were of different amounts of internal shear reinforcement (i.e. two with stirrups 
and  two without stirrups in the  critical shear span)  and  different  shear  span­to­effective depth ratios. Two of these 
beams have been selected for analysis; reference beam S03­1 which was unreinforced in shear in one shear span and 
beam S03­2 which was also unreinforced for shear but externally strengthened using CFRP U­jackets (Figure 3). 

The mean  cylinder compressive strength of  both beams was fcm  = 28  MPa,  and the concrete tensile  strength in the 

present paper was taken as 1.7 MPa (0.33Öfcm). The yield and fracture strengths of the 10 mm diameter stirrups were 
350  MPa  and  530  MPa,  respectively,  with  corresponding  yield  and  fracture  strains  of  1750 me  and  2650 me 
respectively.  The  yield  strength  and  elastic  modulus  of  the  longitudinal  bars  were  460 MPa  and  200 GPa 
respectively.  The  effective  depth  of  the  longitudinal  tension  steel  bars  was  taken  as  260 mm  with  clear  cover 
assumed  to  be  20 mm. Each  FRP  strip  was  50 mm wide  and  0.165 mm  thick  and  was  applied  using  a  wet  lay­up 
procedure. The ultimate tensile strength and corresponding fracture strain of the FRP were 3,790 MPa and 16,620 me 
respectively. Based on the concrete strength and geometric  properties of the FRP and concrete for beam S03­2, the 
values of parameters tb,max, smax  and n for use in Popovics equation were 4.5 MPa, 0.06 mm and 2.5, respectively.

438
 P/2  P/2 

(a) A 
D10 @ 125 mm c/c  D10 @ 125 mm c/c 
2 D 32 

305 

305 
D10 @ 125 

2 D 32 
4 D 32 
A  150 
1525  1125  400 

3050 

P/2  P/2 

(b) B 
FRP U­Jacket @125 mm c/c 

2 D 32 
305 

305 
1  2  3  4  5  6  D10 @ 125 

2 D 32 

B  150 
610  760  310  760  610 

Figure 3. Khalifa and Nanni (2002) specimens: (a) beam S03­1, and (b) beam S03­2. 

The FE model used in the study was based on a 25 mm square mesh using 4­node isoparametric elements. Cracking 
was modelled using a crack band approach (Bazant and Oh 1983) with the tensile fracture energy taken as 75 N/m. 
The longitudinal reinforcing steel was smeared within the concrete element and shear reinforcing steel was modelled 
using bar elements overlayed on the concrete mesh. In mesh sensitivity studies by Khomwan and Foster (2004a), the 
solution involving debonding failure, using the elements described, was shown to be mesh size independent. 

3.2. FE and Test Results 

3.2.1 Reference Beam, SO3­1 

The  load  versus  midspan deflection response  for  the  reference  beam  is  given  in  Fig.  4a.  A  reasonable  correlation 
exists between  the  experimental results (denoted  as S03­1  (Exp))  and the FE  calculation (denoted  as S03­1  (FE)). 
Although  the  FE  model  gave  a  higher  failure  load,  the  results  were  found  to  be  sensitive  to  the  concrete  tensile 
strength adopted. The FE model correctly predicted the beam to fail in a shear mode. 

3.2.2 FRP­Strengthened Beam, S03­2 

The  load  versus  midspan deflection response  for  beam  S03­2  is  given  in  Figure  4b.  Good  correlation is  observed 
between the experimental results  and FE calculations. Figure 5a shows the variation of slip along the length of the 
FRP  strips  at  the  maximum  applied  shear  (P  = 265kN). The  greatest slip  is  recorded  in  strips  labelled  3  and  4  in 
Figures 5b and 3b. These two strips correspond to the debonded strips in Figure 5b from Khalifa and Nanni (2002). 
Strips 2 and 5 also debonded as can be seen in Figure 5b and their slips also slightly exceeded the peak slip in Figure 
5a. It is likely the  debonding of strip 5  in Khalifa  and  Nanni’s (2002)  experiment occurred  from a  sudden shift  of 
load from debonding of strips 3 and 4. Strip 2, which debonded near its free end, may have debonded at quite a low 
load but it did not affect the load carrying capacity of the system 

4. CONCLUSIONS 

The behaviour of FRP­strengthened RC beams has been determined by FE modelling with the epoxy bond interface 
included  in the  model. The computed  load versus displacement response  compared  well with that  measured  in the 
test.  The  numerical  model  accurately  predicted  a  shear  failure  mechanism  for  the  reference  beam  and  FRP 
debonding for the strengthened beam. 

5. ACKNOWLEDGEMENTS 

This  project  was  funded  by  Australian  Research  Council  (ARC)  Discovery  Grant  DP0453096.  The  financial 
assistance of the ARC is gratefully acknowledged. 

439
 200  300 
250 

Load, P (KN) 
Load, P (KN)  150 
200 
100  150 
SO3­1 (Exp)  SO3­2 (Exp) 
100 
50 
SO3­1 (FE)  50  SO3­2 (FE) 
0  0 
0  2  4  6  8  0  2  4  6  8 
Midspan Deflection (mm)  Midspan Deflection (mm) 

(a) Reference Beam – S03­1  (b) Strengthened beam – S03­2 
Figure 4. FE and experimental results: load versus deflection 

Strip 1 
Distance from soffit (mm) 

300  Strip 2 
250  Strip 3 
200  Strip 4 
150  Strip 5 
Strip 6 
100 
Smax 
50  3  2
5  4 
0  6 
­0.2  ­0.1  0  0.1  0.2 
Slip (mm) 

(a) Distance from soffit versus FRP slip at maximum applied shear  (b) Test beam (Khalifa and Nanni 2002) 

6 
Figure 5. FE and experimental results, and behaviour of FRP 

6. REFERENCES 

Bazant, Z.P. and Oh, B.H. (1983). “Crack band theory for fracture of concrete”. Materials and Structures, Vol. 16, No. 

93, pp. 155­177. 
Bousselham, A. and Chaallal, O. (2006). “Effect of transverse steel and shear span on the performance of RC beams 
strengthened in shear with CFRP”. Composites: Part B, Vol. 37, pp. 37­46. 
Chong,  K.T.,  Gilbert, R.I.  and  Foster, S.J.  (2004). “Modelling  time­dependent  cracking  in reinforced concrete using 
bond­slip interface elements”. Computers and Concrete, Vol. 1, No. 2, pp. 151­168. 
Foster,  S.J.  and  Gilbert,  R.I.  (1990).  Non­linear  Finite  Element  Model  for  Reinforced  Concrete  Deep  Beams  and 
Panels.  UNICIV  Report  No.  R­275,  School  of  Civil  Engineering,  University  of  New  South  Wales,  Australia, 
December, 113 pp. 
Foster, S.J. and Marti, P. (2003). “Cracked Membrane Model: FE Implementation”. Journal of Structural Engineering, 
ASCE, Vol. 129, No. 9, pp. 1155­1163. 
Khalifa, A.  and Nanni, A. (2002). “Rehabilitation of rectangular simply  supported  RC beams with shear deficiencies 
using CFRP composites”. Construction and Building Materials, Vol. 16, pp. 135­146. 
Khomwan, N, and Foster, S.J., (2004a) Finite Element Modelling of FRP Strengthened Walls”, UNICIV Report R­432, 
The  University  of  New  South  Wales,  School  of  Civil  and  Environmental  Engineering,  Kensington,  Sydney, 
Australia, November 2004, ISBN: 85841 399 X, 68 pp. 
Khomwan, N. and Foster, S.J. (2004b). “FE Modelling of Bond in FRP­Repaired Plane Stress Members”. Proc., 18th 
Australasian Conf. on the Mech. of Struct. and Mat., ACMSM 18, 1­3 December, Perth, Australia, pp. 113­118. 
Lu,  X.Z.,  Teng,  J.G.,  Ye,  L.P.  and  Jiang,  J.  (2005).  “Bond­slip  models  for  FRP  sheets/plates  bonded  to  concrete”. 
Engineering Structures, Vol. 27, pp. 920­937. 
Popovics,  S.  (1973).  “A  numerical  approach  to  complete  stress­strain  curve  of  concrete”.  Cement  and  Concrete 
Research, Vol. 3, pp. 583­559. 
Smith, S.T., Otoom, O.F.A. and Foster, S.J. (2006). “Finite element modelling of RC beams strengthened in shear with 
FRP composites”. Proc. (CD Rom), 2 nd  International fib Congress, Naples, Italy, 5­8 June. 
Wong, R.S.Y. and Vecchio, F.J. (2003). “Towards modelling of reinforced concrete members with externally bonded 
fiber­reinforced polymer composites”. ACI Structural Journal, Vol. 100, No. 1, pp. 47­55. 

440
Third International Conference on FRP Composites in Civil Engineering (CICE 2006)
December 13-15 2006, Miami, Florida, USA

A FINITE ELEMENT ANALYSIS ON SHEAR AND NORMAL STRESSES

IN ADHESIVELY-BONDED JOINTS OF COMPOSITE STRUCTURES
Luciano Feo
(Associate Professor, Department of Civil Engineering, University of Salerno, Fisciano (Sa), Italy)

Francesco Ascione
(PhD Student, Department of Civil Engineering, University of Rome “Tor Vergata”, Rome, Italy)

ABSTRACT
A two-dimensional finite element analysis is developed in this paper to study the effect of adherends transverse
moduli on the distributions of shear and normal stresses in adhesively double-lap joints of composite structures. The
numerical results are compared with those of analytical solutions.

KEYWORDS
Adhesion, Bonded joints, Pultrusion, Composite structures, FEA.

1. INTRODUCTION
Fiber-reinforced composite (FRP) materials are being used more and more in civil engineering structures due to
their positive mechanical properties such as elevated rigidity/weight and resistance/weight ratios, high corrosion
resistance and rapid installation of components. The advances in pultrusion technology allow large-scale structural
profiles with acceptable cost for civil infrastructure applications [Barbero, 1998, Head, 1996, Keller, 2001] to be
produced.
As well-known, structural FRP components are still difficult to connect due to the brittle fibrous and anisotropic
nature of the materials. The current practice of bolting leads, in most cases, to an over-sizing of the components.
Adhesive bonding is therefore more appropriate for FRP composites. Adhesive joints show higher joint efficiencies
and are much stiffer compared to bolted joints. Furthermore, the load transfer in adhesive joints is more uniform
with fewer stress concentrations when compared to bolted joints.
Joint strength must be predictable as a function of the material properties, the joint geometry and the type of loading,
for adhesive bonding to be used in the design of FRP structures. The understanding of the stress-strain state in the
joint is a prerequisite for the successful prediction of joint strength. Stress analyses are readily performed throught
numerical analyses such as finite element method. In literature one of the first theories about adhesive joints was
proposed by [Volkersen, 1938] who studied a simple shear lag model based on the assumption of one dimensional
bar-like adherends with only shear deformation in the adhesive layer. Over the years, this theory has been modified
by incorporating with adherend shear deformation [Hart-Smith, 1987, Hart-Smith, 2002, Tsai, Oplinger and Morton,
1998].
The aim of this paper is to evaluate the effect of adherends shear deformation on the shear and normal stress
distributions along the overlap of double-lap joints made from FRP.

2. DOUBLE-LAP JOINT
The geometric profile of a double-lap joint is shown in Fig.1, where: 2c is the length of the overlap, to and ti are the
thicknesses of the outer and inner adherends, respectively, Eo and Go are the elastic modulus (in the longitudinal
direction) and the shear modulus (in the transverse direction) for the outer adherends, respectively, Ei and Gi are the
corresponding properties of the inner adherend, Gc and η are the adhesive shear modulus and the corresponding
thickness, T is an applied force per unit width.

441
 Gc c c

Go , Eo to T
2
T ti Gi , Ei x

Go , Eo to T
2

Figure 1: Geometric profile of a double-lap joint

In [Tsai, Oplinger and Morton, 1998] it is possible to find the integral version of the double-lap joint (TOM theory).
The shear stresses, τc, in the adhesive layer can be expressed as the following form:

τ c = A sinh ( βx ) + B cosh ( βx ) . (1)

The expressions of the constant coefficients A and B are:

⎡ Et ⎤
⎢ 1− i i ⎥
βT 2E o t o βT
A= ⎢ ⎥, B = , (2)
4 cosh ( βc ) ⎢ Ei t i ⎥ 4sinh ( βc )
⎢ 1 + 2E t ⎥
⎣ o o ⎦

where the parameter β is defined as:

β2 = λ 2 α 2 , (3)

λ and α being, respectively, the elongation parameter and the shear deformation parameter:

Gc ⎛ 2 1 ⎞
λ2 = ⎜ + ⎟, (4)
η E t
⎝ i i E o to ⎠

−1
⎡ G ⎛ ti t ⎞⎤
α = ⎢1 + c
2
⎜ + o ⎟⎥ . (5)
⎣⎢ η 6G
⎝ i 3G ⎥
o ⎠⎦

The closed form solution proposed by [Volkersen, 1938] can be recovered by assuming that adherends shear
deformations are zero, or that adherends shear moduli, Gi and Go, are infinitely large, therefore α =1.

3. THE FINITE ELEMENT ANALYSIS

A balanced double-lap joint, with graphite-epoxy laminated adherends, has been considered in this study. The 2D
finite element analysis has been performed by using the commercial program Straus7. Fig.2 shows the finite element
mesh, where the six-node triangle is employed. In Table 1 the values of the used geometrical and mechanical
parameters are summarized.
Fig.3 shows the comparison between FEM results and those of the analytical solutions, both TOM theory (α <1) as
well as Volkersen (VT) (α =1), in terms of shear stress distribution in the adhesive layer.
As it can be seen, the shear stress distribution, obtained by FEM, is characterized by higher values than the
analytical ones in the central part of the overlap region. Moreover, it can be observed that in the finite element
analysis the maximum shear stress does not occur on both ends simultaneously in contrast to the analytical solutions.
Furthermore, the maximum values of shear stress, at the ends, are also lower.

442
 Figure 2: Finite element mesh of the double-lap joint.

Table 1: Geometrical and mechanical parameters.

Shear Mod. Young Mod.

Parameters Length [mm] Thickness [mm]
[GPa] [GPa]
Adhesive 12.7 0.15 0.910 2.51
Outer adherend 12.7 1.00 137.0 4.83
Inner adherend 12.7 2.00 137.0 4.83

5.0
c [MPa]
4.5
VT ( =1)
4.0
TOM ( <1)
3.5

3.0

2.5

2.0
FEM
1.5

1.0

0.5

-1.5 -1 -0.5 0 0.5 1 1.5

x/c

Figure 3: Comparison between FEM results and analytical solutions: VT (α=1), TOM (α<1).

4. PARAMETRIC STUDY
In the classical theory [Volkersen, 1938], adherend shear deformations are ignored, either possibly due to the
relatively small values compared to longitudinal normal deformations in some cases, or to the complexity of
formulations.
The maximum shear distribution in the adhesive is over-estimated if adherend shear modulus is neglected, as
highlighted in [Tsai, Oplinger and Morton, 1998]. This effect is more accentuated especially for adherends with
relatively low transverse shear modulus as in the case of laminate composite adherends.

443
In order to investigate the effect of shear adherends modulus on the adhesive shear and normal stress distributions, a
parametric study, with respect to the below parameter, has been developed in this work:

kG
, ( G = Go = Gi ) . (6)
Gc

Figs. 4,5 show, respectively, the influence of shear adherends modulus on the adhesive shear and normal stress
distributions along the overlap length. It can be observed, in particular, that as k increases the adhesive shear stress
belongs to VT solution (α=1).

5.0
c [MPa]

4.5 10
VT
c[MPa]
4.0
8

3.5
6
TOM

3.0
4

2.5
K=100
2
K=10
2.0 K=2
K=1
0
K=0.5 z/c
1.5
-1.5 -1 -0.5 0.5 1 1.5

K=0.1 k
-2
1.0
K=0.01
-4
0.5
k
-6

-1.5 -1 -0.5 0 0.5 1 1.5

z/c

Figure 4: Shear stress distributions. Figure 5: Normal stress distributions.

5. CONCLUSIONS
A two-dimensional finite element analysis has been developed to study adhesive bonding for double-lap joints of
FRP materials. The effect of shear adherend modulus on the shear and normal stress adhesive distributions has been
investigated, showing that such stresses depend on the adherend shear modulus, in particular for laminated
composites with low shear modulus. When comparing FEM results to classical theory it is possible to note that the
latter over-estimates them.

6. REFERENCES
Barbero E , 1998, “Introduction to composite material design”, Taylor & Francis.
Hart-Smith LJ. , 1987, in: Matthews FL Editor., “Design of adhesively bonded joints. Joining fiber-reinforced
plastics”. Elsevier Applied Science; 271–311.
Hart-Smith LJ., 2002, “Adhesive bonding of composite structures—progress to date and some remaining
challenges”. J Compos Technol Res; 24(3):133–53.
Head P. R., 1996, “Advanced composites in civil engineering - A critical overview at this high interest, low use
stage of development”, II International conference on advanced composite materials in bridges and structures,
Montréal, Québec, Canada; 3-15.
Keller T. , 2001, “Recent all-composite and hybrid fiber reinforced polymer bridges and buildings”. Progress Struct
Eng Mater; 3/2:132–40.
Tsai MY, Oplinger DW, Morton J., 1998, “Improved theoretical solutions for adhesive lap joints”. Int J of Solids
Struct; 35(12):1163–85.
Volkersen O., Die Nietkraftverteilung, 1938, in “Zugbeanspruchten“ Konstanten Laschenquerschnitten.
Luftfahrtforschung; 15:41–7.

444
Third International Conference on FRP Composites in Civil Engineering (CICE 2006)
December 13-15 2006, Miami, Florida, USA

FINIT ELEMENT MODELING OF FRP-FILAMENT WINDED POLES

R.Masmoudi
(Professor, University of Sherbrooke, Sherbrooke, QC, Canada)

H. Mohamed
(Ph.D Student, University of Sherbrooke, Sherbrooke, QC, Canada)

ABSTRACT
A large-scale research program is currently undertaken at the Department of Civil Engineering, University of
Sherbrooke, Canada, to assess the general behaviour of light weight, glass fibre reinforced polymer (GFRP), pole
structures. A finite element (FE) program was used to perform a nonlinear numerical analysis to model the static
flexural behaviour of GFRP poles. The results of the FE analysis are compared to the experimental results conducted
on full scale identical GFRP poles. A parametric study on 12 m (40 ft) GFRP poles was carried out to show the
effect of fibre orientation and the number of circumferential layers on the load carrying capacity and deflection
behaviour. The results show a good agreement between the FE analysis and the experimental data. The theoretical
model is used to evaluate the performance of a GFRP pole and to determine the optimum cross section dimensions
at three different zones along the height of the GFRP pole structure.

KEYWORDS
Fibre Reinforced Polymers, FRP Structural Shapes, FRP poles, Flexural behaviour, Filament Winding.

1. INTRODUCTION
The GFRP poles technology has been used for over thirty years in the application of the light poles and electrical
transmission tower element, as a replacement for the conventional materials, due to their high strength-to-weight
ratio and corrosion resistance (Ibrahim et al 2000 and Ibrahim et al., 1999). High quality surface coating and the
ultraviolet radiation resistance treatment give the FRP pole a long service life, beyond eighty (80) years (Miller et
al., 1995). A limited number of experimental and theoretical studies have been conducted on the behaviour of the
tapered GFRP poles structure under lateral load (Lin 1995, Crozier et al., 1995, Derrick 1996). Due to the existence
of a service opening at 2.7 m from the bottom of the FRP pole, and also due to small thickness-to-radius ratio, local
buckling failure can reduce significantly the load carrying capacity. Therefore the part which includes this service
opening must be addressed. It is important to find the optimum geometrical details for the region of the service
opening, in order to be compatible with the upper and lower zones over the length of the pole. t
In this paper, a finite element program with a nonlinear numerical analysis was used to optimize the design of a 12
m (40ft), GFRP poles having a service opening.

2. FINITE ELEMENT ANALYSIS

A nonlinear finite element model was developed using the software ADINA finite element program. The finite
element analysis was verified through comparison with the experimental data obtained from the static testing of full-
scale GFRP poles, according to the recommendations described in ASTM and ANSI standard (Masmoudi and
Metiche, 2006). The specimens were tapered hollow sections, and divided through the height into three zones, I, II
and III. The 101.6 x 304.8mm-(width x length) service opening is located at the center of the middle zone II and was
in the compression side, when loaded. The typical specimen dimensions and details of the three zones are shown in
Figure 1. GFRP poles are fabricated using the filament winding technique. E-glass fibres and Epoxy resin were
used for these poles. The pole modeled with a total number of elements equal to 2288 (16 and 143 in the

445
circumference and longitudinal direction, respectively). The mesh layout was fine for zones I and II, where
maximum stresses and failure are expected to occur, and gradually becomes coarse at the top. This was made by the
automatic mesh density option of the program. The general layout of the mesh distribution and the used finite
element models are shown in Figure 1.

114

8890 mm
Zone III

12090 mm

one II with
Pressure load on
the top of zone III
Hand - hole

Zone II
1000

Zone I with
two opposite
faces supports
1500
Zone I
2200

265
Zone II with open
Dimension of FRP pole Model

Fig. 1: The typical specimen dimensions and the details of the finite element model.

The under ground length of the GFRP poles were restraint along two opposite half circumference area, the first area
at the end of the base and the second area at the ground line. This Configuration of restraints was to simulate the
support condition described in standards ASTM D 4923-01 and ANSI C 136.20-2005. The GFRP pole was
subjected to a horizontal pressure load (W) from the top of the pole edge by 300mm according to the ANSI C
136.20-2005 recommendations. The pole was incrementally loaded using 100-200 time steps. To avoid local failure
under the applied load, a pressure-load has been used to simulate the same effect as in the experimental testing.
An eight-node quadrilateral multilayered shell element is used in the model; each node has six degrees of freedom,
three translations (Ux, Uy, and Uz) and three rotations (Rx, Ry, and Rz). The composite shell elements are
kinematically formulated in the same way as the single layer shell elements, but an arbitrary N number of layers can
be used to make up the total thickness of the shell. The basic equations used in the formulation of the Multilayered
Shell Element are given in the reference Bathe 1996. The material model used with the shell element is elastic-
orthotropic with large displacement /small strain. The mechanical properties of the FRP laminate were obtained
from the material properties of the E-glass fibre and the epoxy resin. Orthotropic material properties in the fibre and
transverse to the fibre direction were defined. Fibre orientation for each layer was specified by defining the fibre
angle with respect to the element axes. The effective materials properties were taken as follows: E1 = 45.5 GPa,
E2 = 10.5 GPa, G12 = 4.875 GPa,, 12 =0.31, where, E1, E2, G12 and 12 are the Young’s modulus in the fibre
direction, the Young’s modulus in the transverse direction, the shear modulus, and the Poisson’s ratio, respectively.
Table 1 presents the fiber orientations and the stacking sequences for the actual and new designs.

Table 1. Stacking sequences for the tested poles and proposed new designs
Masmoudi and al. (2006) New design
Zone Fibre orientation Thickness Fibre orientation
(Degrees) (mm) (Degrees)
I [±70, ±30, ±90] 5.56 [90, (±10)5, 90]
II [90, ±20, ±70] 13.62 {90, ±45, [90, (±10)5, 90]±45, 90 }
III [±70, ±10, ±70] 6.3 [90, (±10)5, 90]

446
 3. RESULTS AND DISCUSSION

Failure of the modeled FRP poles was determined when the divergence of the solution was achieved or when the
Tsai-Wu failure criterion value reached unity. A comparison between the finite element analysis and the results
obtained from experimental testing of full-sale prototypes obtained by Masmoudi and al., 2006, was in terms of the
load-deflection relationship and the ultimate load carrying capacity. Figure 2 represents the load deflection
relationship for the experimental and finite element analysis. It is evident from this figure that there is a strong
correlation between the results obtained from the finite element analysis and the experimental results.

9000 3.00

8000
2.75
7000

Stiffness (N /m m )
6000 2.50
Load (N)

5000
2.25
4000

3000 2.00

2000 Finite Element (actual design)

Experimental (Masmoudi and Metiche) 1.75
1000 New design 1
New design 2
0 1.50
0 500 1000 1500 2000 2500 3000 10 20 30 40 50 60 70 80 90 100
Top deflection (mm) Fibre orintation of circumferencial layers

Figure 2: Comparison between experimental Figure 3: Effect of fibre orientations of

and FE analysis Load-deflection circumferential layers on the stiffness
behaviour of GFRP pole

Because of this agreement, the same finite element analysis was used to extend the study and examine the effect of
fibre orientation of circumferential layers on the flexural behaviour of GFRP poles. To study this parameter the
circumferential layers with fiber orientation of 70 degree were changed to 90, 60, 50 and 30 degree, with the same
details of wall thickness, dimension of the GFRP pole and material properties. Figure 3 shows the relation between
fibre orientations of circumferential layers and stiffness of GFRP pole for different models. It is clear that 90 degree
of the fiber orientation gives the highest stiffness.

Description of the proposed GFRP pole design to reach an equivalent performance of classes 2 and 4 of
wooden pole
The design criteria to satisfy the requirement strength criteria as given in the ASTM D 4923-01, AASHTO, and the
general standard specifications of ANSI C 136.20-2005 for maximum deflection, ultimate load capacity and
minimum embedment depth. The different combinations of parameters such as fibre orientation, number of
longitudinal and circumferential layers and layer thickness are considered. It was also assumed that the total number
of layers to be constant equal to 12 layers, (10 longitudinal and two circumferential layers in zone I, II, III).
However, in order to reinforce zone II (service opening), 6 additional layers are used (3 inside the 12 previous layers
and 3 outside the 12 previous layers), which make the total layers of this zone equal 18. The fibre angles assumed to
be +10°/-10° for longitudinal layers and 90° for circumferential layers. The (90°, +45°/-45°) orientation are used for
the 6 additional layers. The thickness of each layer was varied from 0.40 mm to 0.8 mm and for the thickness of
each additional was assumed to be twice the previous thickness.

The finite element analysis was employed to optimize the design of the 12 m (40 ft) FRP pole to achieve an
equivalent ultimate load capacity required for class 4 and class 2 wooden poles. After several FE simulations with
different thickness, the results indicate that; the requirements for the FRP pole class 2 and class 4 achieved when the
thickness of each layer were to be equal 0.5 mm and 0.75 mm, respectively. Thus, the total thickness of the laminate
in zone I and zone III for class 4 and class 2 equal 6 mm and 9 mm, respectively, where the total thickness in zone II
will be over by the thickness of the 6 additional layers which is equal to 12 mm and 18 mm for FRP class 4 and class
2, respectively. The mode of failure for these model occurred at the ground level due to the local buckling. Figure 2

447
shows the load deflection relationships for GFRP poles equivalent to class 2 (New design 1) and class 4 (New
design 2) wooden poles.

4. CONCLUSION
The finite element program ADINA (version 8.2) was used to perform a nonlinear numerical analysis of tapered
GFRP poles. Layered composite shell elements were used in this finite element analysis. The program accounts for
the nonlinear behaviour of the poles and includes a strength failure check by applying the Tsai-Wu failure criterion.
The results were in an excellent agreement with the experimental results. The finite element method used in this
investigation provided an excellent prediction of the critical buckling and material failure loads, as well as the
corresponding modes of failure for thin-walled GFRP poles. The load-deflection curve of GFRP poles under lateral
loading can be considered linear up to failure. The fibre orientation of circumferential layer with 90 degree for the
tapered GFRP poles gives the higher load capacity and stiffness. The internal and external additional three layers
(90, ±45) for the laminate at the middle zone II, improved the flexural behaviour due to the existed hand-service
opening. The proposed models with 12 layers with two circumferential with 90 degree and ten with 10 degree give
excellent result. Optimum designs for 12 m (40 ft) GFRP poles equivalent to class 4 and class 2 wood poles were
obtained using the finite element model, to satisfy the requirement strength criteria as specified in the ASTM,
AASHTO and ANSI standards for the maximum deflection and ultimate load capacity.

5. REFERENCES

ADINA (2004), Theory and Modeling Guide, Volume I, Chapter 3, Version 8.2. ADINA R&D Inc., Watertown, MA,
USA.
American Association of State Highway and Transportation Officials, AASHTO (2001), “Standard Specifications
for Structural Supports for Highway Signs, Luminaires and Traffic signals”.
American National Standard Institute. (2005). “Fiber-Reinforced Plastic (FRP) Lighting Poles, American National
Standard for Roadway Lighting Equipment”, USA, ANSI C 136.
American Society for Testing and Materials. (2001). “Standard Specification for Reinforced Thermosetting Plastic
Poles”, Annual book of ASTM Standards, D 4923 – 01, USA.
Bathe, K.J. (1996). Finite Element Procedures. Prentice-Hall.
Crozier, W., Dussel, J.P., Bushey, R., West, J. (1995). “Evaluation of deflection and bending strength characteristics
of fibre-reinforced plastic lighting standards”. Department of Transportation, New York, State of California, USA.
Derrick, G. L. (1996). “Fiberglass Composite Distribution and Transmission Poles”, Manufactured Distribution and
Transmission Pole Structures Workshop Proceeding, July 25-26, Eclectic Power Research Institute, pp. 55-61.
Ibrahim, S., and Polyzois, D., and Hassan, S. (2000). “Development of glass fiber reinforced plastic poles for
transmission and distribution lines”. Can. J. Civ. Eng. Vol. 27, pp. 850-858.
Ibrahim, S., and Polyzois, D. (1999). “Ovalization analysis of fiber reinforced plastic poles”. Composite Structures,
Vol.45, pp 7–12.
Lin ZM. (1995). “Analysis of pole-type structures of fibre-reinforced Plastics by finite element method”, Ph.D.
Thesis, University of Manitoba, Winnipeg, Manitoba, Canada.
Masmoudi, R., Metiche, S., (2006). “Experimental and Theoretical Evaluation of the Flexural Behaviour
of Fibre-Reinforced Polymer (FRP) Poles”, Accepted for publication in the Journal of Composite
Materials, 26p.
Miller, M. F., Hosford, G.S. and Boozer III, J.F. (1995). “Fiberglass Distribution Poles: A Case of Study" IEEE
transactions on Power delivery, Vol. 10, No. 1, pp. 497-503.
Tsai, S. W., and Wu, E. M. (1971). “A General Theory of Strength for Anisotropic Materials” Journal of
Composites materials, Vol. 5, pp. 58-80.

448
Part XV. Novel Applications
Third International Conference on FRP Composites in Civil Engineering (CICE 2006)
December 13-15 2006, Miami, Florida, USA

ADVANCED GRID STIFFENED FRP TUBE ENCASED CONCRETE CYLINDERS

Guoqiang Li
Assistant Professor, Department of Mechanical Engineering, Louisiana State University, Baton Rouge, Louisiana,
USA-70803

Dinesh Maricherla
Graduate Student, Department of Mechanical Engineering, Louisiana State University, Baton Rouge, Louisiana,
USA-70803

ABSTRACT
In this study, an advanced grid stiffened (AGS) composite tube, which was made of a lattice of interlaced fiber
reinforced polymer (FRP) ribs that was wrapped by a FRP skin, was used to encase concrete cylinders. Two types of
cylinders were prepared. One was a circular cylinder and the other was a square cylinder. For the circular cylinders,
a uniaxial compression test was conducted; for the square cylinders, both uniaxial compression test and four-point
bending test were conducted. The test results showed that the AGS tube confined concrete cylinders displayed a
considerable positive composite action due to the mechanical interlocking. The AGS tube provided an “active”
confinement to the concrete core. The elastic range and the specific axial strength of the circular AGS cylinders, and
the specific bending strength of the square AGS beams were higher than those encased by solid FRP tubes, solid
steel tubes, or steel grid tubes. The square AGS tubes were not as effective as the circular AGS tubes in confining
concrete.

KEYWORDS
Grid, stiffener, rib, FRP, concrete, cylinder, beam, confinement.

1. INTRODUCTION

In recent years, an ever-increasing attention has been paid to use fiber reinforced polymer (FRP) tube-encased
concrete columns for new construction and rebuilding of engineering structures [1,2]. The mechanism in FRP tube-
encased concrete columns is to replace the steel rebar by a corrosion-resistant laminated FRP shell. The FRP tube
serves as a stay-in-place formwork during construction; during service, the tube confines the lateral expansion of the
concrete core and transfers the core to a triaxial compressive stress condition. As a result, the compressive strength
and ductility of the column are enhanced significantly.

Subjected to an axial compressive load, sufficiently confined concrete cylinders behave bi-linearly [3-6]. The
confinement provided by the FRP tube is passive. The FRP tube cannot confine the concrete core until the concrete
has sufficient lateral expansion, i.e., when the concrete is cracked or crushed. The higher ultimate compressive
strength of the confined concrete cannot be realized unless the concrete is damaged and there is an excessive axial
strain (an axial strain in the order of 0.02 or higher), which is much larger than the currently accepted design strain
of 0.003, as specified by the American Concrete Institute (ACI318-99). Since confined concrete cylinders are
designed to work in the elastic region, i.e., the first linear region in the axial stress-axial strain curve, the
strengthening potential of the FRP tube cannot be realized and utilized in practice. In order to utilize the
strengthening potential of the FRP, the first region must be enlarged.

The reasons for this have been summarized in [7]. They include (1) a FRP tube usually has a much larger radial
(out-of-plane) Poisson’s ratio than that of concrete core before the concrete cracks and (2) For most FRP tubes,
fibers are aligned towards the hoop direction in order to provide higher confinement. As a result, the axial stiffness
of the FRP tube may be lower than that of the concrete core. These result in a transverse expansion of the FRP tube
larger than that of the concrete core before the concrete cracks or crushes. Since these are intrinsic properties to FRP
and concrete, innovative ways must be found to solve this problem.

449
A viable way of solving this problem is to use grid stiffened composite tube. It was proved that, by using a hybrid
tube – a lattice of steel grids that were wrapped by a FRP skin, a mechanical interlocking between the tube and the
core was developed [8,9]. This insured that any additional transverse expansion by the concrete core be resisted by
the confining tube, leading to early activation or engagement of the tube in confining the concrete core. As a result,
the confined concrete cylinders had a higher elastic region than the solid FRP tube confined counterparts. Because
FRPs behave linearly elastic while steel behaves elastic-perfectly plastic, it is interesting to find how FRP grid tube
encased concrete cylinders behave. The purpose of this study is thus to use advanced grid stiffened (AGS) tubes to
encase concrete cylinders. Both column and beam behaviors will be investigated.

2. SPECIMENS PREPARATION AND TESTING

2.1 Raw Materials

Type I Portland cement, gravel, natural sand, water, and a superplasticizer DAVA 170 were used to prepare the
concrete. The mix design followed ACI Standard 211.1 (“Standard” 1991). The mix ratio by weight was cement :
water : coarse aggregate : fine aggregate : admixture = 1 : 0.56 : 3.80 : 2.19 : 0.001. The maximum coarse aggregate
diameter was 25.4mm. The slump was 140.2mm; the air content was 6.8%; and the 28-day cylinder compressive
strength was 30 MPa. The same concrete has been used previously [8]. The AGS composite tubes were fabricated
using a Fiberex 503 E-glass fiber roving and a Sunrez UV curing vinyl ester resin. The same fiber and resin were
used to manufacture circular AGS tubes and square AGS tubes. For the FRP skin, the same resin was used. The fiber
was a unidirectional E-glass 7715 style plain woven fabric.

2.2 AGS Tube Fabrication

While a number of manufacturing methods have been explored in the literature [10,11], a hand lay-up process based
on a mandrel-pin system, which simulated the filament winding technology, was used to fabricate circular grid
tubes. The fiber rovings with resin were first wound along the pin-guidance to form the grid skeleton. Once the
designed thickness was achieved, the fiber was cut and the specimen was exposed to an UV-A light source (or
sunlight) for curing. Typical curing time was about half an hour for each specimen. Once the specimen was fully
cured, the pins were removed and the mold was pushed out of the cured grid. The plain woven fabric with resin was
then wrapped around the skeleton to form the skin. The cured circular grid tube had an inner radius 50.8mm and
height 304.8mm; the ribs had a thickness 6.25mm and width 5.08mm. The bays were squares with a side length
25.4mm. Two types of square AGS tubes were fabricated. The first type had a height 304.8mm and the second type
had a height 558.8mm. Both types had the same square cross-section with a side length 114.3mm. Each bay was
again a square with a side length 25.4mm. A total of three circular AGS tubes, three shorter square AGS tubes with
a height 304.8mm, and three taller square AGS tubes with a height 558.8mm were prepared. The average weight of
the circular tubes was 0.58kg; it was 0.64kg for the shorter square AGS tubes and 1.31kg for the taller square AGS
tubes. A burn-off test per ASTM D2584 was used to determine the fiber volume fraction in the grids. It was found
that the fiber volume fraction was about 30%. The cured thickness of the skin was 0.738mm.

2.3 Instrumentation and Testing

2.3.1 Compression test

Two pairs of strain gages were used for each specimen. Each pair contained two strain gages mounted at the mid-
height of the cylinder aligned in the hoop and axial directions, respectively. The two pairs were spaced at 180o in the
hoop direction. The strain gages were mounted on the surface of the FRP skin above the ribs. During the axial
compression test, each specimen was loaded uniaxially to about 40% of the compressive strength of the unconfined
concrete (30.0MPa) and unloaded to guarantee close contact between each component. Then, the specimen was
reloaded until failure. The compression testing was conducted using a FORNEY machine, with a capacity of
2,688kN. The assembled computer data acquisition system can directly record the load-displacement curves. The
strain was recorded using a Yokogawa DC100 Data Acquisition Unit. Compression testing was conducted per
ASTM C 39. The loading rate was 0.23MPa/s.

2.3.2 Four-point bending test

Four-point bending tests were conducted on each second-type square specimen. This was a simply supported
bending test with a span length 457.0mm. Each specimen was first loaded with a preload of 20KN to guarantee close

450
contact between each component. The specimen was then unloaded and reloaded until failure. The tests were
conducted using the same FORNEY machine. The test was conducted per ASTM C 78. The loading rate was
230N/s.
140

3. RESULTS AND DISCUSSION 120

Specific axial stress (MPa/kg)

100
3.1 Axial stress-axial strain behavior
Solid FRP tube [8]
80 Axial steel grid tube [8]
Solid steel tube [8]
Typical axial stress-axial strain behaviors of an AGS Square FRP grid tube
circular tube encased cylinder, an AGS square tube 60 Circular FRP grid tube

encased cylinder, a solid FRP tube encased cylinder, a 40

solid steel tube encased cylinder, and an axial steel grid
tube encased cylinder are shown in Fig. 1. The test 20

results of the solid FRP tube, solid steel tube, and axial 0
steel grid tube encased cylinders are obtained from [8]
by dividing the axial stress by their respective tube -20
0 2 4 6 8 10 12
weight of 1.32kg, 5.08kg, and 2.52kg. It is seen that the
circular AGS tube behaves differently from the solid Axial strain (%)

FRP tube. The axial stress-axial strain curve before the Fig. 1 Comparison of specific axial stress-axial strain behavior
peak stress can be represented by a linear section of various types of tubes encased concrete cylinders
followed by a non-linear section, instead of the typical
bi-linear curve connected by a transition zone. The specific stress at the end of the linear section (about 75MPa/kg)
for the circular AGS cylinder is much higher than that at the end of the first linear section (about 30MPa/kg) for the
solid FRP cylinder. This suggests that the elastic region of the AGS tube encased concrete cylinder is enlarged. The
square AGS tube and the axial steel grid tube behaved similarly to that of the circular AGS tube. The difference is
that the axial steel grid cylinder has a higher stiffness in the elastic region probably due to the higher axial stiffness
of steel than that of FRP. The axial stiffness of the square AGS cylinder is the lowest due to the larger lateral and
thus axial deformation of the AGS tube caused by the stress concentration at the four sharp corners of the tube. For
the solid steel tube confined cylinders, the specific strength is the lowest among all the encasing devices. This is due
to the larger density of steel and the lack of mechanical interlocking between the tube and the concrete core.

3.2 Specific compressive strength

From Fig. 1, the specific compressive strength is the highest for the cylinder encased by the circular AGS tube,
followed by the solid FRP tube, square AGS tube, axial steel grid tube, and solid steel tube. This means that, for an
ideal case, i.e., assuming liner relationship between the tube weight and the confinement strength, the circular AGS
tube encased concrete cylinder would have the highest compressive strength if the same amount of materials is used.
Therefore, the circular AGS tube not only has a larger elastic region, but also has a higher ultimate specific strength.
Comparing the circular AGS tube with the square AGS tube, the square AGS tube shows a much smaller specific
strength and ductility due to the stress concentrations at the four sharp corners.

3.3 Bending strength 80

Steel grid tube [9]

Figure 2 shows the transverse bending load-deflection FRP grid tube
Specific transverse load (KN/kg)

60
curves from the four-point bending test. One was encased
by the taller square AGS tube and the other was encased
40
by the same size steel grid tube from [9]. Three
observations can be made. (1) The AGS tube encased
beam shows a linear elastic behavior up to the peak load, 20

while the steel grid tube encased beam shows a linear

section followed by a non-linear section with a continuous 0
reduction in stiffness. (2) The specific load and deflection
of the AGS beam is higher than those of the steel grid
beam. Together with the higher specific compressive 0 5 10 15 20 25
strength, it is concluded that, using the same amount of Transverse deflection (mm)
materials, AGS tube encased cylinders would have a Fig. 1 Comparison of steel grid tube confined concrete beam
higher load carrying capacity than the steel grid tube with FRP grid tube confined concrete beam under bending

451
encased concrete counterparts. (3) Both before and after the peak load, the AGS beam shows some fluctuations of
load. This is because of the load redistribution through the network once the skin or some ribs fail.

3.4 Failure mode

The failure modes of the circular AGS cylinder, the square AGS cylinder, and the square AGS beam are shown
respectively in Fig. 3 (a) – (c). From Fig. 3 (a), both the axial ribs and hoop ribs have fractured, close to the nodal
area. The very even fractured surface suggests that the failure is due to shear stress. For the square AGS cylinder, the
failure is primarily due to the fracture of
the hoop ribs at the corner, as shown in
Fig. 3 (b). This is due to the stress
concentration at the sharp corner of the
square AGS tube. For the AGS beam,
Fig. 3 (c), the failure is due to the
fracture of the longitudinal ribs.

4. CONCLUSIONS
(a) (b) (c)
An innovative FRP grid tube confined Fig. 3 Failure modes
concrete cylinders have been developed
and experimentally tested subjected to
both compression and bending. It is found that the AGS tube confined concrete cylinders or beams have the highest
specific strength than counterparts confined by steel tubes, FRP tubes, and steel grid tubes.

ACKNOWLEDGMENT
This study was partially sponsored by the Louisiana Board of Regents and EDO Fiber Science under contract
number LEQSF(2004-07)-RD-B-05. The encased concrete cylinders and beams were prepared and the tests were
conducted in the Concrete Laboratory at the Louisiana Transportation Research Center (LTRC). The support by Mr.
Randy Young, Mr. John Eggers, Mr. Sadi Torres, and Mr. Walid Alaywan from LTRC is greatly appreciated.

REFERENCES
1. L.C. Hollaway, “The evolution of and the way forward for advanced polymer composites in civil
infrastructures,” Construction and Building Materials, 17, pp. 365-378, (2003).
2. A. Mirmiran, L. C. Bank, K. W. Neale, J. T. Mottram, T. Ueda, and J. F. Davalos, “World Survey of Civil
Engineering Programs on Fiber Reinforced Polymer Composites for Construction,” Journal of Professional
Issues in Engineering Education and Practice, 129, pp. 155-160, (2003).
3. F. Seible, A. Davol, R. Burgueno, R.J. Nuismer, and M.G. Abdallah, “Structural behavior of concrete filled
carbon fiber composite tubular columns,” Proceedings of 28th International SAMPE Conference, Seattle,
WA, pp. 1258-1269, (1996).
4. A. Mirmiran and M. Shahawy, “A new concrete-filled hollow FRP composite column,” Composite Part B:
Engineering, 27B, pp. 263–68, (1996).
5. A.Z. Fam and S.H. Rizkalla, “Behavior of axially loaded concrete-filled circular fiber-reinforced polymer
tubes,” ACI Structural Journal, 98, pp.280-289, (2001).
6. G. Li, S. Torres, W. Alaywan, and C. Abadie, “Experimental Study of FRP Tube-Encased Concrete
Columns,” Journal of Composite Materials, 39, pp. 1131-1145, (2005).
7. G. Li, “Experimental Study of FRP Confined Concrete Cylinders,” Engineering Structures, 28, pp. 1001-
1008, (2006).
8. G. Li, “Experimental Study of Hybrid Composite Cylinders,” Composite Structures, (2005), (in press)
doi:10.1016/j.compstruct.2005.08.028.
9. G. Li, M. John, and D. Maricherla, “Experimental Study of Hybrid Composite Beams,” Construction and
Building Materials, 2005 (in press) doi:10.1016/j.conbuildmat.2005.11.003.
10. S.M. Huybrechts, T.E. Meink, P.M. Wegner, G.M. Ganley, “Manufacturing theory for advanced grid
stiffened structures,” Composite Part A: Applied Science & Manufacturing, 33, pp. 155-161, (2002).
11. DY Han and SW Tsai, “Interlocked composite grids design and manufacturing,” Journal of Composite
Materials, 37, pp. 287-316, (2003).

452
Third International Conference on FRP Composites in Civil Engineering (CICE 2006)
December 13-15 2006, Miami, Florida, USA

APPLICABILITY OF FIBER REINFORCED PLASTICS

TO HYDRAULIC GATES
Tomonori TOMIYAMA
(Research Staff, Public Works Research Institute, Tsukuba, Ibaraki, Japan)

Itaru NISHIZAKI
(Team Leader, Public Works Research Institute, Tsukuba, Ibaraki, Japan)

ABSTRACT
Following the results of surveys targeting related government ministries and agencies, industry organizations, local
governments, etc., it has become clear that Fiber Reinforced Plastics (FRP) have been applied to members of
hydraulic gates since the 1960s with over 400 structures now existing in Japan. Fifty FRP gates were visually
inspected. Though they have been only lightly maintained over 30 years, little deterioration is evident. Results of
durability tests with FRP specimens in the laboratory indicate that FRP for hydraulic gates has satisfactory water and
corrosion resistance, though appearance gradually deteriorates from exposure to ultraviolet light.

KEYWORDS
Fiber Reinforced Plastics (FRP), Hydraulic Gate, Durability, Weather Resistance, Corrosion Resistance

1. INTRODUCTION
Hydraulic gate facilities are main structures in river-ways and play key roles in flood control and water utilization
works. At present, mainly steel materials have been used for these facilities. Though these structures are always
exposed to severely corrosive environmental elements, there are few effective anticorrosion engineering options.
Hence, it is desirable to develop advanced materials having excellent corrosion resistance in order to reduce the
labor and cost for maintaining hydraulic gates. In this paper, present situations of FRP that have been utilized as
hydraulic gates in Japan were investigated and problems for bringing FRP to market were clarified in order to
promote the application of FRP for hydraulic gates. In addition, field studies were carried out on 53 in-service FRP
hydraulic gates in order to understand their actual conditions and performance. Furthermore, long-term durability of
FRP for hydraulic gates was examined by accelerated weathering and water immersion tests.

2. MARKET RESEARCH OF FRP HYDRAULIC GATES IN JAPAN

In association with the main Japanese manufacturer, construction results of FRP hydraulic gates in Japan were
investigated. Results of these investigations showed that 438 FRP gates were initiated from 1961 to 2002 in Japan
but of these, the existence of only 59 FRP gates was confirmed. The reason for this low number of verified FRP
gates is because many of them were delivered to general contractors and are therefore now impossible to track.
Figure 1 shows initial construction data of FRP hydraulic gates arranged by construction year. This data shows that
more than 80% of total FRP gates were constructed before 1990, whereas construction in recent years has
significantly declined. The styles of FRP gate adopted in Japan are slide gate, flap gate, roller gate, swing gate, miter
gate, sliding gate and angle chute; most gates are compact in size. The area is smaller than 4.0 m2 on almost 90% of
adopted FRP door bodies. It is assumed that FRP was difficult to apply to large door bodies because FRP has low
elastic modulus and is flexible. In the present state, it may be effective to promote the advantages of FRP by limiting
targets to comparatively small-scale (such as 10 m2 or less) hydraulic gate facilities.

453
 over 80% of total FRP gates 1990

Construction starts of FRP gates

40 37 Total = 438 gates
30 30 29
30 25
24 24 24
20 19 20
20 16
13 12 12
10 10 11
9 8 7 9
10 6 6
4 5 4
3 3 3 2
1 1 1
0

1990

1995
1975

1980

2000
1985
1969
Before
Construction year

Figure 1: Construction Starts of FRP Hydraulic Gates in Japan

In order to examine the marketability of FRP hydraulic gates, questionnaire surveys were carried out targeting
related government ministries and agencies, industry organizations, local governments, and so on. The
questionnaires consisted of questions about durability, abrasion resistance, ease of installation, maintenance, weather
resistance, strength, corrosion resistance, initial cost, and running cost for FRP hydraulic gates. The results of the
questionnaire showed that low visibility of FRP within the FRP hydraulic gates market has been one of the obstacles
against its increased use.

3. LONG-TERM DURABILITY OF FRP IN RIVER ENVIRONMENTS

In order to confirm the applicability of FRP to members of hydraulic gate facilities, age deterioration of in-service
FRP gates was investigated.

3.1 Field Studies on Existing FRP Hydraulic Gates

Field studies were carried out on 53 FRP hydraulic gates which were in service. Material degradations of FRP
members caused by river water or ultraviolet rays were evaluated visually. Results of the visual inspections show all
FRP gates had no major deterioration except for slight discoloration and water stain despite the fact that they were
hardly maintained for 30 years. Thus, FRP gates seem to be more durable than steel gates after an equivalent period
of time.

3.2 Disassembling Investigations on an FRP Door Body

An FRP door body which had been used as a sluice gate on an agricultural waterway was selected and examined
(see Table 1).

Table 1: Overview of FRP Hydraulic Gate Selected for Strength Tests

Construction Year before 1969

In-service Period more than 35 years
Gate Style slide gate
Dimensions 1.15m x 1.00m
Applied FRP door body, door stop
Forming Method hand lay-up
Operating Situation full-time operating
Transformation by
N/A
Water Absorption
Erosion Damage N/A
water stain
Degradation Panoramic View
discoloration

Tensile tests were carried out by using a main girder channel and a skin plate of a dismantled door body. From the
channel and sections of both the submerged area and non-submerged area of the skin plate, five test specimens were

454
cut into pieces 8 mm thick, 25 mm wide, and 250 mm long respectively. For comparison, tensile tests were also
similarly carried out on another FRP channel and laminate that were newly fabricated with exactly the same
laminate composition as that of the obtained old door body. The sampling method of the FRP door body for the
tensile test is shown in Figure 2. After testing, specimens were examined by scanning electron microscope (SEM),
energy dispersive X-ray spectrometer (EDS), Fourier transform infrared spectrophotometer (FT-IR), and so on.

Downstream Side Upstream Side

Non-submerged
Area

Submerged
Area

(a) Main Girder (b) Skin Plate

Figure 2: Sampling Methods of FRP Door Body for Tensile Test

Table 2 shows a part of the results. Some data spread on tensile strength, which is a weakness of FRP material, was
recognized. A future challenge is to reduce such data spread. In any case, it was confirmed that the mean values of
tensile strength and modulus were almost the same between the two FRP laminates. Results of instrumental analyses
for specimens after tensile tests did not indicate significant deterioration of the FRP door body.

Table 2: Comparison of Tensile Properties between FRP Skin Plate after Use for 35 Years and Newly
Fabricated Laminate

After Use for 35 Years

Newly Fabricated
Non-Submerged Area Submerged Area
Specimen
No. Tensile Tensile Tensile Tensile Tensile Tensile
Strength Modulus Strength Modulus Strength Modulus
[MPa] [GPa] [MPa] [GPa] [MPa] [GPa]
1 168.49 17.10 152.35 15.62 161.69 15.04
2 156.71 15.25 149.60 15.31 168.82 15.08
3 144.87 15.57 149.91 23.61 162.90 17.66
4 135.63 16.49 148.65 15.25 167.19 16.43
5 135.63 15.83 149.72 15.55 173.62 15.86
Average 148.27 16.05 150.04 17.07 166.84 16.02

4. DURABILITY TESTS OF FRP FOR HYDRAULIC GATES

Generally, FRP for hydraulic gate members is produced using the hand lay-up, resin injection, or pultrusion molding
methods. Matrix resins of those FRP are epoxy resins, unsaturated polyester resins, or vinylester resins, with glass
fibers used as reinforcement. General characteristics of FRP for hydraulic gate members are shown in Table 3. In
order to obtain basic knowledge about the long-term durability of FRP, accelerated weathering tests and water
immersion tests were carried out. FRP laminates manufactured by hand lay-up, resin injection, and pultrusion were
used as testing samples. Dimensions of test specimens were 4–10 mm thick, 15 mm wide, and 200 mm long. In
accelerated weathering tests, specimens were put into an accelerated weathering tester in the laboratory and exposed
to xenon-arc sources. Changes of properties with UV irradiation were evaluated by glossiness and color difference
on specimen surface, flexural strength, and flexural modulus. In water immersion tests, the same specimens as the
accelerated weathering tests were immersed in water at 20–25°C. Changes of properties with penetration of water
were evaluated by weight change, flexural strength, and flexural modulus.

In Figure 3 (a), gloss retentions in all cases were significantly reduced from the beginning of exposure and then
decreased to less than 20% of the original glossiness at about 2500 hours exposure. On the other hand, Figure 3 (b)
shows that the flexural strength of each FRP sample barely changed from before and after exposure to ultraviolet
light for 2500 hours, though pultrusion FRP strength showed some loss. From these results, it was determined that
degradation of FRP by ultraviolet rays had occurred only at the surface after 2500 hours exposure, indicating that
FRP’s strength remained virtually intact. It is supposed that such degradation in FRP surface by ultraviolet light can

455
be chemically improved by coating the FRP surface with, for example, a gel coating with high resistance to UV
radiation.

Table 3: General Characteristics of FRP* for Hydraulic Gate Members

Mechanical Properties Thermal Properties

Vf Coefficient
Molding Method Tensile Tensile Flexural Flexural Compression Thermal Specific
[%] of Linear
Strength Modulus Strength Modulus Strength Conductivity Heat
Expansion
[MPa] [GPa] [MPa] [GPa] [MPa] [W/mK] [J/gK]
[10-5/°C]
Hand Lay-up 25 - 40 80 - 120 7 - 10 120 - 180 7-9 120 - 170 3-4 0.17 - 0.23 1.38
Resin Injection 25 - 40 80 - 120 7 - 10 120 - 180 7-9 120 - 170 3-4 0.17 - 0.23 1.38
Pultrusion 55 - 85 500 - 900 20 - 40 590 - 900 20 - 35 250 - 500 0.6 - 0.8 0.29 - 0.52 1.05 - 1.26
*matrix: unsaturated polyester resin, reinforcement: glass fiber

120 140

Retention of flexural strength [%]

hand lay-up 120
100
Gloss retention [%]

resin injection
100
80 pultrusion
80
60
60
hand lay-up
40
40 resin injection
20 20 pultrusion
0 0
0 500 1000 1500 2000 2500 0 500 1000 1500 2000 2500
Exposure time [hr] Exposure time [hr]

(a) (b)

Figure 3: Results of Accelerated Weathering Test; Gloss Retention (a) and Retention of Flexural Strength (b)
of FRP Exposed to Laboratory Xenon-arc Sources

In the immersion tests for 2500 hours, the largest value of material swelling rates was +0.61% and that of weight
changes was about +1.4%. In three-point bending tests, the flexural strength of FRP decreased slightly. It seemed
that the strength degradation of FRP was caused by swelling due to water absorption. In this immersion test,
swelling and strength degradation of FRP did not exceed the permissible limit, since water absorption of FRP was
very small.

The results of durability tests clarified that FRP for hydraulic gates has satisfactory water resistance and corrosion
resistance, while appearance gradually deteriorates from exposure to ultraviolet light. Therefore, it seems that FRP is
especially suitable for hydraulic gates which are usually submerged in water.

5. CONCLUSION
The present study revealed that FRP is especially suitable for hydraulic gates usually submerged in water since they
have excellent durability and water or corrosion resistance. For the future, it will be necessary to formulate
guidelines for the application of FRP to hydraulic gate members in order to improve the visibility of FRP and
promote its utilization in the hydraulic gates industry.

REFERENCES
Chowdhury, M., Hall, R. (1998): “Performance Evaluation of Full-Scale Composite and Steel Wickets for Use at
Olmsted Locks & Dam”, Proceedings of the 2nd International Conference on Composites in Infrastructure
(ICCI’98), Vol. 1, pp. 667-681
Xue, X., Jiang, S., Fang, X. (1990): “Application of resin coating on hydraulic construction”, Proceedings of the 6th
International Congress on Polymers in Concrete, pp. 634-641
Yoshida, M., Gocho, M., Noguchi, T., Hoshino, J., Momojima, H., Saeki, N. (1990): “Development of slide sluice
gate by carbon hybrid FRP (In Japanese)”, Reinforced Plastics, JRPS, Vol. 36, No. 5, pp. 159-162

456
Third International Conference on FRP Composites in Civil Engineering (CICE 2006)
December 13-15 2006, Miami, Florida, USA

STRESSED ARCH MODULAR DEPLOYABLE COMPOSITE SHELTERS

CONCEPT AND DEVELOPMENT
T Omar
(PhD Candidate, Fibre Composites Design and Development, University of Southern Queensland, Toowoomba,
Australia)

G Van Erp
(Professor Fibre Composites Design and Development, University of Southern Queensland, Toowoomba, Australia)

P W Key
(Manager & Engineering director, Kencana Strarch Asia Pacific, Kuala Lumpur, Malaysia)

ABSTRACT
Deployable shelters of various forms have been utilised since ancient civilisation. The need for these systems has
not diminished over time and development continues for military forces, civilian humanitarian aid, and post-natural
disaster scenarios. Recent developments have focussed mainly on tent type structures, air beam technology and steel
frames supporting soft fabric, yet none of these systems have fully satisfied the deployability requirements.

The Military Modular Shelter System (M2S2) initiative is a research project that aims to develop a fibre composite
re-deployable arched shelter system with rigid or fabric cladding. The main frames are formed from modular fibre
composite panels that are connected and stressed in position by prestressing cables. Different geometries can be
obtained using this system by changing the number of panels per frame and the packer sizes between panels.

This paper presents the concept of M2S2 with background about existing systems followed by the development and
testing of an innovative, simple to manufacture, truss module that was investigated as part of this project. The test
results showed good characteristics. These include having alternate load paths and failure initiated and propagated in
the web with no, undesirable, failure observed in the adhesive layers.

KEYWORDS
Shelters, Composites, Deployable, Hanger, Strarch, Truss.

1. INTRODUCTION
The need for deployable structures has existed since ancient times. Deployable structures are similar to normal
structures in that they have to be stable and able to carry designated loads in their deployed status (Gantes, 2001). In
addition, they should satisfy the deployability requirements of being able to be dismantled, stored and transported in
a compact form. They should also have an inherent deploying mechanism that allows the transition between the
deployed status and the dismantled status and vice versa. Deployable shelters are an important application of
deployable structures. They are needed for military applications, aircraft maintenance hangers, aid relief and
temporary and/or remote structures.

Over time, the performance requirements of modern deployable shelters have become more demanding, which has
driven the development of more sophisticated structural forms and solutions. The basic components of deployable
shelters are the structural system (primary load transfer) and the cladding system. The cladding system can have
different functions depending on its inherent properties and those of the structural system used. For example,
cladding systems can be used to stabilize the structural system, assist in carrying primary loads, or can be integrated
with the overall load carrying system. Consequently the two sub-systems are generally dependent on each other. The
recent developments of deployable shelter technology can be categorized as air-inflated shelters; rigid frames
supporting soft fabric shelters and stressed arch systems. These developments were reviewed by Verge

457
(www.natick.army.mil) and Omar et. al. (2006). These reviews showed that the currently used deployable shelter
systems do not fully satisfy the shelter deployability requirements. Air beam technology has not satisfied the needs
for deployable shelters in spite of being under development for a significant period. Low pressure air beams can
only be used for short spans. High pressure air beams store significant amounts of energy and still can not be used
for large spans. The state-of-the-art deployable shelter system may well be the Widespan by Weatherhaven. The
system can be used for large free spans. However, steel frames are used for the main panels with a dimension of
3.66m in length and a weight of 68kg (http://www.weatherhaven.com/). Both the size and the weight of these panels
are more than the legal carrying capacity of two persons, in Australia. This may necessitate using some form of
cranage to erect this system. The Weatherhaven systems use soft fabric for the cladding which creates the
impression of a temporary structure that accommodates large deformations.

The M2S2 research project aims to investigate the behavior of a composite arch deployable shelter system that uses
the post-tensioning prestressed technology as a deploying mechanism. One of the core components of this system
was its modular panel. An innovative, simple to manufacture, adhesively bonded truss panel system was developed
for this purpose. In this paper, the M2S2 concept is presented followed by the development and testing of its panel.

2. M2S2 CONCEPT
The concept of the prestressed arch technique has been implemented successfully in Strarch steel frames (Strarch
1999). The continuous nature of the top chord, the plastic deformation during stress erection and the strength to
weight ratio associated with the steel trusses all provide challenges to the deployable functionality of conventional
Strarch frame systems.

The M2S2 concept is similarly based on the stressed arch concept. However, it is adapted to the requirements of
deployability by using more manageable (approx. 1500mm square), light-weight truss panels that do not require
plastic deformation. The top chord deformation is concentrated at discrete joints (Figure 1). The concept of M2S2
can be summarized as follows:
1. Frames are manufactured, mostly, from identical standard panels;
2. Standard panels are stacked to form each frame on the ground;
3. Panels are then connected by inter-panel top hinged joints. The difference in dimension between the top
chord and the bottom chord allows having initial gaps at the bottom chord;
4. One side of the frames is fixed to the foundation, while the other is free to move horizontally during
erection. The prestressing cables are threaded through the bottom chord;
5. Roof sheeting (rigid type) and other services are assembled while the frames are still on the ground, prior to
carrying out any prestressing (Assembly stage);
6. Finishing the installation of services; frames are stressed using prestressing cables. The bottom chord gaps
allow the geometry to change to an arch shape during the prestressing process (Erection stage);
7. The prestressing cables are stressed to the level that allows for losses and/or relaxation in addition to
ensuring that the bottom chord will be in compression under any serviceability loads. The cables are then
blocked and the moveable frame support is fixed. The shelter is complete and ready to use (Deployed
stage).

The concept of M2S2 is quite flexible. The number of panels per frame and the packer sizes define the frame span
and height in the deployed position. Table 1 shows the effect of increasing the packer sizes from 200mm to 220mm
on the frame geometry (frames are based on 32 standard panels). Increasing the packer size increases the frame
span, reduces the rise/span ratio and the subtended angle. This flexibility should be accounted for when investigating
the behavior of such frames.

Table 1 Effect of packer size on frame geometry

Frame Alternative A1 A2 A3
Packer Size(mm) 200 210 220
Rise/Span – Radius(m) 12.1/36.7– 19.9 11.1/38.4- 22.1 10.1/40.0- 24.8
Rise/Span Ratio 0.330 0.289 0.252
Subtended Angle (deg) 133.3 120.3 107.2

458
 Figure 1 M2S2 system – main components

3. THE M2S2 MODULAR PANEL

The panel system is the core of the M2S2 concept. Accordingly, it was the first investigated component. As per the
requirement of the system, the panel should have hollow sections for the chords to accommodate the prestressing
cables. Previous investigations showed that it is recommended to have panels with flat-sided standard components;
extended joint area and alternate load path after failure (Omar et al, 2006). Using a multi-pultruded panel with
diaphragm bracing system satisfied these requirements. In addition, it was simple to manufacture and allows the use
of multi-cables for prestressing.

Due to the arch geometry, the shelter frames are mainly subjected to axial forces. In addition, shear and bending
moments are generated due to the un-symmetric loading of wind and live loads. In using standard pultrusion
sections for the chords and vertical members, the member capacity under axial loads can be predicted using design
codes. However, the diagonal skins and connections to the chord and vertical members need further investigations.
These necessitated conducting prototype panel testing by applying shear forces on the panel to investigate its
behavior. The panel was tested in beam mode with loads applied at mid span. The tested configuration had two
panels of 650mm centerline dimension, with 50mm gap at the centre (Figure 2). The overall structure consists of
three identical frames of 50mmx50mmx5mm hollow square pultrusions which were adhesively bonded to the two
web laminates using a toughened epoxy adhesive. Load, deflection and strains were recorded at locations shown in
Figure 2. Strain gauges located across the panel thickness were used to locate any differential stress and strain
distributions. Gauges located on the laminated web were used to measure the tensile and compressive strains in the
+45deg direction. Loads were applied using an Instron loading ram with a 500kN capacity. The structure was loaded
using a displacement controlled loading rate of 2mm/min.

4. PANEL TEST RESULTS

The tested panel showed good performance with a load-deflection curve shown in Figure 3. The load-deflection
curve indicates that the panel still reserves partial load carrying capacity (about 50% of its ultimate capacity) after
failure in spite of continuous increase in the applied displacement. Failure initiated at the top corner of the
diaphragm, due to the combined tension in the diagonal direction and compression in the perpendicular direction.
Compressive forces were due to the confinement of the web with the tendency of the angle between the vertical and
top chord members to reduce under the applied loads. Failure propagated along the inner faces of the vertical and
top chord following the pattern of the formed wave of the buckled web (Figure 3). No failure was observed in the
adhesive layers. This was another desirable feature as adhesive failure is inherently brittle. In releasing the applied
load, the panel restored most of its deflection (in spite of rupturing of the web). More detailed behavioral
investigation of this panel will be the subject of future publication where a finite element model was developed at
the micro level to provide an insight of the panel is good attributes.

5. CONCLUSIONS
Combining the effectiveness of the arch as a structural form, the post tensioning prestressing technology and
composite light weight materials has significant potential for developing effective deployable shelter systems. The
M2S2 concept seems able to deliver a flexible deployable shelter system that satisfies the deployment requirements

459
and the flexibility needed by the end users. The flexibility is achieved by using modular panels of manageable size
and weight which are within the carrying capacity of two persons. In using packers at the bottom chord, different
structural configurations can be obtained with different span to height ratio.

Using a multi-pultrusion panel system with diaphragm web showed good characteristics with the advantages of
having alternate load paths, simplicity of manufacturing and allowing the usage of multi-prestressing cables. For the
panel tested, failure initiated and propagated in the laminated web with no failure observed in the adhesive layers.
The laminated web was able to restore the panel original geometry after releasing the applied loads.

Figure 2 Test panel geometry and strain gauge positions

200

180

160

140
Crack
120
initiation &
Load (kN)

100 propagation
80

60

40

20
Defl (mm)
0
0 5 10 15 20 25

Figure 3 MK III load deflection curve and failure mode

5. References
Clarke, M. J. and Hancock, G. J. (1994). Behavior and design of stressed-arch (Strarch) frames. IASS-ASCE
International Symposium 1994 on spatial, lattice and tension structures, Atlanta, ASCE.
Gantes, C. J. (2001). Deployable structures: Analysis and design. Southhampton, WIT Press, UK.
Omar, T., Heldt, T., and al, e. (2006). " M2S2 Modular Deployable Composite Shelters - Concept and loading
criteria." Australian J of Structural Engineering, 6(3), 217-226.
Strarch (1991). Analysis of US military requirements for large deployable shelters, Sydney, Australia.
Strarch (1999). The Strarch building system – Technical discussion. Sydney, Australia.
Verge, A. S. “Rapidly deployable structures in collective protection systems”, Massachusetts, U.S. Army Natick
Soldier Center: (www.natick.army.mil).
Weatherhaven Home page, http://www.weatherhaven.com/

460
Third International Conference on FRP Composites in Civil Engineering (CICE 2006)
December 13-15 2006, Miami, Florida, USA

EVALUATION OF EPOXY ENCAPSULATED TIMBER PILES

Nakin Suksawang
(Assistant Professor, Florida International University, Miami, FL, USA)

Hani Nassif
(Associate Professor, Rutgers, The State University of New Jersey, Piscataway, NJ, USA)

Ozgul Ozkul
(Structural Engineer, Leslie E. Robertson Associates, New York, NY, USA)

Ali Maher
(Professor, Rutgers, The State University of New Jersey, Piscataway, NJ, USA)

Abbas Sarmad
(Vice President, DMJM+Harris, New York, New York, USA)

ABSTRACT
Many bridges are supported on timber piles that have significant deterioration due to fungal decay of the wood. A
method of repair is proposed to encase the damaged timber pile in Glass Fiber Reinforced Polymer (GFRP) wraps
that are injected with special epoxy mortar. The composite repaired piles need to be tested to determine their
ultimate load capacity under combined shear and flexure. This paper evaluates the behavior of extracted repaired
timber piles under lateral loads. Sixteen repaired timber piles are extracted from actual conditions and tested as
cantilever beams. Special experimental setup is designed to test the piles under flexure and shear. Various types of
measurements such as deflections and strains are measured at different locations of the pile. Results show that all
repairing systems increase the stiffness and lateral load capacity of existing timber piles (Douglas fir) regardless of
the dimensions of the cross section. The stiffness and lateral load capacity of the repaired timber pile could be as
high as that of the Greenheart but requires more epoxy mortar thickness.

KEYWORDS
Glass Fiber Reinforced Polymer, Timber Pile, Epoxy Mortar, Repairng Techniques.

1. INTRODUCTION
Timber piles are one of the oldest and most frequently used pile foundations in North America. They still remain
one of the most cost-effective solutions for foundation. However, one of the problems with timber piles is the
deterioration above ground water table where the pile is liable to attack by marine organisms, fungi, and wood
eating insects (Chellis, 1961, Lopez et al., 2004). Overtime, the deterioration is severe enough that the pile looses its
bearing capacity, and it could eventually lead to foundation failure. A strong commitment by the New York City
Economic Development Corporation (NYCEDC) is now underway to prevent further losses. A new repairing
technique using epoxy encapsulation to provide protection and preservation is proposed by DMJM+Harris, New
York and conducted by Trevcon Construction Co. Since, this technique is relatively new; its performance needs to
be evaluated (Menzel et al., 2004).

A total of sixteen timber piles were tested laterally using a 1780 kN loading frame at the Rutgers University
Structural Engineering Laboratory (Nassif et al., 2005). The project compares the lateral load carrying capacity of
repaired timber piles using glass fiber reinforced polymer (GFRP) jacket and epoxy mortar obtained from three
manufacturers (i.e., Sika Corp./MFG Construction Products, Master Builders/Watson Bowman, and Fox Industry,
Inc.) to existing timber piles (Douglas fir), and replacement alternative (Greenheart). In addition, the long-term
performance is also investigated. Figure 1a and 1b illustrate the cross section of the repaired timber pile and test

461
setup, respectively. The epoxy mortar is a mixture of sand and epoxy that is injected through an access port at the
bottom of the GFRP jacket to fill the space between the wood and the GFRP, as well as the burrows.

Gross
Outer Timber
Jacket

Damaged
Timber Epoxy
Mortar
Thickness Thickness
N
Core
W E
S
(a) (b)

Figure 1: (a) Typical Cross Section of a Pile Specimen and (b) Test Setup

2. EXPERIMENTAL PROGRAM
The sixteen timber piles consisted of twelve GFRP encapsulated Douglas fir timber piles, two deteriorate Douglas
fir timber pile, and two new Greenheart timber pile. The twelve timber piles were encapsulated using GFRP and
epoxy mortar from different manufacturer. Three manufacturer combinations were used: (1) Sika epoxy mortar and
MFG Construction Product GFRP (Sika), (2) Watson Bowman epoxy mortar and GFRP (WB), and (3) Fox
Industries epoxy mortar and GFRP (Fox). Table 1 summarizes the cross section of each timber pile specimen used
in this study.

Table 1—Cross Section at failure location of the Tested Timber Piles

Undamaged Timber Gross Wood Outer Jacket Ave.

Test Diameter (mm) Diameter (mm) Diameter (mm) Epoxy
Manuf.
No Thickness
E-W N-S Ave. E-W N-S Ave. E-W N-S Ave.
(mm)
1 (S1) Sika 292 279 286 394 375 384 438 445 441 25
2 (S2) Sika 349 330 340 349 330 340 445 445 445 49
3 (F1) Fox 241 229 235 343 324 334 445 445 445 52
4 (S3) Sika 298 305 302 368 375 372 432 438 435 29
5 (W1) WB 279 279 279 394 394 394 438 445 441 21
6 (S4) Sika 337 324 330 362 387 375 438 438 438 29
7 (W2) WB 248 241 245 343 362 349 419 432 425 35
8 (W3) WB 279 248 264 356 330 343 432 432 432 41
9 (F2) Fox 279 279 279 343 356 349 470 470 470 57
10 (W4) WB 298 279 289 381 375 378 445 438 441 29
11 (D2) N/A 273 292 283 311 318 314 N/A
12 (F3) Fox 254 248 251 343 356 349 445 445 445 44
13 (S5) Sika 305 279 292 381 368 375 432 425 429 24
14 (G1) 305 305 305 305 305 305
15 (G2) N/A 330 330 330 330 330 330 N/A
16 (D1) 356 356 356 356 356 356
Notes: S, F, W, D, and G denotes Sika, Fox, Watson Bowman, deteriorated Douglas fir, and new Greenheart timber
piles.

462
3. RESULTS
Table 2 summarizes the results from tests performed on the pile specimens. The cracking load is the load when a
crack is initiated, as observed during the tests, on either the tension or compression side of the GFRP jacket. The
ultimate load is the peak load, whereas the failure load is the instantaneous load before the specimen completely
fails. If the failure of the timber pile is brittle, then the cracking, ultimate, and failure loads are very close.

A summary of the failure modes of each timber piles is also reported. Majority of repaired timber pile specimens
failed in a similar failure mode: first by having a local cracking of the GFRP jacket near the support and
compressive failure of the epoxy mortar, leading to a tensile cracking of the GFRP jacket, denoted as “A” on the
table. Figure 2 illustrates typical failure mode “A”. The other failure modes consisted of tensile failure, denoted as
“B”, where GFRP jacket fail in tension first followed by compression failure of the epoxy mortar. This failure mode
is more desirable since it yield higher ultimate strength and provide more toughness to the repaired timber piles.
The other two failure modes, denoted as “C” and “D”, are also compression failure similar to “A” but the failure is
located not at the position of maximum moment but at the weakest location along the repaired timber piles, i.e.,
GFRP splice and injection port, respectively.

Table 2—Timber Pile Loading Capacity at Ultimate, Failure, and Cracking

Test Months Load (kN) Deflection (mm) Failure
No In Place Cracking Ultimate Failure Cracking Ultimate Failure Mode
1 (S1) 15 62.3 81.6 75.9 102 208 230 A
2 (S2) 4 111.2 112.2 100.9 152 152 207 B
3 (F1) 23 75.6 84.9 57.1 137 229 291 C
4 (S3) 15 71.2 116.0 -- 64 171 -- A
5 (W1) 31 86.8 -- -- 224 -- -- D
6 (S4) 18 111.2 115.5 -- 127 5.20 -- A
7 (W2) 31 53.4 49.5 127 184 212 212 A
8 (W3) 31 106.8 83.9 140 189 264 264 B
9 (F2) 23 115.7 133.0 127 187 212 212 A
10 (W4) 31 106.8 84.4 119 120 213 213 A
11 (D2) >25 21.7 -- -- 132 -- -- A
12 (F3) 23 71.2 96.7 93.9 102 213 237 A
13 (S5) 15 75.6 77.8 -- 254 258 -- A
14 (G1) 0 115.7 145.7 -- 157 210 -- A
15 (G2) 0 124.6 159.1 -- 135 180 -- A
16 (D1) >25 80.1 91.2 -- 178 261 -- A

Note:
The following are the description of the failure mode:
A = Compressive failure in the GFRP jacket and epoxy mortar, followed by a tensile failure in the GFRP jacket.
B = Tensile failure in the GFRP jacket, followed by compressive failure in the GFPR jacket and epoxy mortar.
C = Failure mode similar to A but the tensile failure is located at the splice of the GFRP jackets.
D = Failure mode similar to A, but the tensile failure is located at injection port.

In order to perform a comparison between the three repairing systems, an attempt to normalize the applied load to
the corresponding cross section is made. The normalization is achieved by dividing the applied load by the moment
of inertia of the pile specimens. The moment of inertia is based on intact specimen using the secant modulus of
elasticity and Hook’s Law. The reason for using the moment of inertia is to incorporate the cross section properties,
which are highly variable in each group, as a factor in deciding the efficiency of each system.

Figure 3 illustrates the relationship between the load-section efficiency factors (i.e., the ratio of the load capacity to
the elastic moment of inertia of the composite section) versus the tip deflection of the pile. As expected, the new
Greenheart specimens outperformed the repaired timber piles. However, all the repaired timber piles outperformed
the existing deteriorated timber piles (Douglas fir). This shows that, in general, the repair techniques used in this

463
study are highly effective in adding to the capacity of the existing Douglass fir timber piles. Moreover, comparing
the absolute load carrying capacities without considering the section properties shows that Sika has outperformed all
other specimens. However, the best performing specimens among all groups, the WB specimen (W3) has shown
comparable performance as the Sika specimen (S3). Thus, all epoxy and GFRP jacket manufacturers provide a good
protective system with an added benefit of increasing the lateral load carrying capacity of the existing deteriorated
timber piles.

Figure 2: Typical Failure Mode (Mode A)

Tip Deflection, mm
0 50 100 150 200 250 300 350 400
40
Core Epoxy 4.0
4

4
S3 Pile # Diameter Thick.
Load/Moment of Inertia, lbs/in

Load/Moment of Inertia, N/cm

F2 (mm) (mm) 3.5
30 W3 S3 302 29
G1 F2 279 57 3.0
G2 W3 264 41
D1 G1 305 N/A 2.5
20 D2 G2 330 N/A
2.0
D2 356 N/A
D2 283 N/A 1.5
10 1.0
0.5
0 0.0
0 2 4 6 8 10 12 14 16
Tip Deflection, in.

Figure 3: Relationships of Load and Moment of Inertia of Composite Section on Pile Specimens

4. CONCLUSIONS
The following conclusions can be made from the testing program:

1. Overall, all repairing systems increase the stiffness and lateral load capacity of existing timber piles (Douglas
fir) regardless of the dimensions of the cross section. The stiffness and lateral load capacity of the repaired
timber pile could be as high as that of the Greenheart but requires more epoxy mortar thickness. However, it
should be noted that the repaired timber will generally have larger diameters than the Greenheart timber pile.

464
2. The cracking and ultimate load carrying capacities of the piles are highly dependant on the undamaged timber
radius as well as the damaged timber thickness. Repaired timber piles with larger undamaged timber radius
have exhibited higher cracking and ultimate load carrying capacities. However, since the cross section of the
existing pile is uncontrollable, it is recommended that thicker epoxy mortar be used.
3. Quality control process, especially with the production of epoxy mortar, plays a significant role in the pile
lateral load capacity.

5. ACKNOWLEDGEMENTS

The authors would like to thank Paul Menzel from DMJM+Harris inc., New York, Keith Neuscheler from
Turner Construction Co., New York, and Dmitri Konon and Patrick Askew from New York City
Economic Development Corporation (NYCEDC) for their help and assistance during the period of the
project. Also, the assistance of laboratory technician, Edward Wass, for fabricating the testing setup, is
thankfully acknowledged.

6. REFERENCES
Chellis, R. D. 1961. “Deterioration and Preservation of Piles, Pile Foundations”, McGraw-Hill, New York: 339-372.
Lopez-Anido, R., Michael, A.P., Goodell, B. & Sanford, T.C. 2004. “Assessment of Wood Pile Deterioration due to
Marine Organisms”, Journal of waterway, port, coastal and ocean engineering, ASCE, March-April: 70-76.
Menzel, P., Sarmad, A., and Carel, J. 2004 “Epoxy Encapsulation Systems & Construction Methods for Timber
Repairs,” Ports 2004, Houston, TX, May 23-26.
Nassif, H., Suksawang, N. & Ozkul, O., Maher, A. 2004. “Test Evaluation of Repair Techniques for Timber Piles-
Final Report”. Department of Civil and Environmental Engineering, Rutgers University, RU-CEE-No. 2004-02,
p.30.

465
466
Part XVI. Prestressing Applications
Third International Conference on FRP Composites in Civil Engineering (CICE 2006)
December 13-15 2006, Miami, Florida, USA

CFRP PRESTRESSING TENDONS: SOLUTION UPTAKE AND

SWELLING EFFECTS
Paul Scott
(PhD Candidate, University of Cambridge, Cambridge, United Kingdom)

Dr Janet Lees
(Senior Lecturer, University of Cambridge, Cambridge, United Kingdom)

ABSTRACT
Experimental work has been undertaken to assess the nature of water, salt water and concrete pore solution uptake,
and corresponding swelling effects, in Carbon Fibre Reinforced Polymer tendons. A method is presented to predict
the long-term diffusion behaviour of aqueous solutions in CFRP tendons using short-term tests and Fickian models.
Year long uptake tests have shown Fickian diffusive behaviour that is insensitive to the solution types considered.
After one year, tendon mass and volume increases in the order of 0.3 % and 0.5 % respectively have been observed
and the two phenomena appear to increase proportionally with respect to each other. Compared to thermal expansion
effects, such swelling may be significant: potentially affecting the integrity of prestressed concrete structures.

KEYWORDS
CFRP, Prestressing Tendons, Durability, Solution Uptake, Swelling

1. INTRODUCTION
Carbon Fibre Reinforced Polymer (CFRP) materials are permeable to aqueous solutions which have been shown, in
epoxy matrices, to affect their chemical and mechanical properties (Springer, 1988). The modelling of phenomena
such as the diffusion of solutions into CFRP tendons and associated swelling effects have yet to be fully addressed
but are of significance when designing CFRP prestressed concrete for durability in aggressive environments. This
paper characterizes the possible behaviour of CFRP tendons in the marine environment through studying the
diffusion of water, salt water and synthetic concrete pore solution (CPS) in tendon samples.

2. CFRP TENDONS: THEORETICAL DIFFUSIVE BEHAVIOUR AND SWELLING

Space between molecular chains in epoxies makes them permeable to aqueous solutions. Solution uptake into CFRP
tendons with epoxy matrices is a dual mode process: whilst the solution diffuses into the tendon, it also interacts
chemically with the epoxy. Models proposed to characterize diffusion into polymers include Fickian, Dual Mode
Sorption (considering both diffusion and chemical interaction) (Barrer, 1984) and Langmuir (for anomalous
moisture diffusion) (Carter and Kibler, 1978). This paper considers Fickian diffusion only, given its robustness and
reported suitability for use when considering diffusion in both matrix and composite materials (Springer, 1988).

Fick’s first and second laws, equations (1) and (2), relate the rate of transfer of diffusing substance per unit area, F,
to its concentration, C, in the direction of diffusion, x, perpendicular to the unit area considered. The diffusion
coefficient, D, is a function of how permeable a given material is to a given solution over time, t. Diffusion
behaviour in materials can be measured using gravimetric sorption: periodically weighing samples exposed to
solutions to assess the rate of uptake. Equation (3) is used to calculate the percentage increase in mass of a sample
(mass increase), Mt, based on its mass at time t, mt, compared to its initial dry mass, mo, prior to solution uptake.

∂C (1) ∂C ∂ 2C (2)  m − mo  (3)

F = −D =D M t = 100  t 
∂x ∂t ∂ x2  mo 

467
Fickian diffusion can be used to model aqueous solution uptake in CFRP in the absence of matrix cracking, moisture
propagation along the fibre/matrix interface, voids or non-Fickian matrix behaviour due to, for example, significant
relaxation as a result of solution uptake (Springer, 1988). When studying the diffusion of solutions into materials,
the diffusion coefficient and the mass increase at saturation, M∞, are key parameters. For CFRP, suitable diffusion
parameters must be calculated bearing in mind the heterogeneous nature of the material. The permeability of carbon
fibres is insignificant compared to that of the surrounding epoxy matrix so the mass increase in a saturated
composite, Mc∞, is a function of the volume fraction, Vf, and the mass increase in the saturated neat resin, Me∞, as
shown in equation (4) (Springer, 1988). Similarly, the fibres act as geometric barriers to diffusion, so the diffusion
coefficient for a solution to pass through CFRP perpendicular to the fibre direction, Dc, is a function of the diffusion
coefficient of the epoxy, De, and the volume fraction as shown in equation (5) (Springer, 1988). Crank (1975)
derived equation (6) that shows how data from solution uptake experiments measuring the mass increase at time t,
Met, and at time of saturation in thin films of epoxy material of thickness l, could be used to calculate the Fickian
diffusion coefficient for the respective solution-epoxy interaction.

 Vf 
( )
Mc ∞ = 1− V f Me ∞ (4) Dc =  1 − 2 D (5) Me t
=
4 Det (6)
 π  e Me ∞ l π

It is possible to predict solution uptake in prestressing tendons, considered as long cylinders, using a Fickian radial
diffusion model. From equations (1) and (2), using suitable boundary conditions, Crank (1975) derived equation (7)
that calculates the mass increase of an isotropic cylinder of constant radius, r, when submerged in solution by
summing n roots, αnr, of a Bessel function of the first kind, of order zero. As the number of summation terms
evaluated in equation (7) approaches infinity, the solution tends to a linear relationship between mass increase and
the square root of time, which tails off as uptake reaches 80 % saturation. By reasonably assuming that the models
leading to equations (4) and (5) are valid when considering radial diffusion in CFRP tendons, mass increase at time
t, Mct, can therefore be predicted by specifying suitable parameters for the composite, as shown by equation (8).

∞ 
 4  2 
( )
∞
Mct
Mt
M∞ n =1  r n
( )
= 1 − ∑  2 2 exp − Dα n 2 t 
α 
(7)
Mc ∞
4
= 1 − ∑  2 2 exp − Dcα n t 
n =1  r α n 
(8)

Uptake of aqueous solutions into epoxies causes swelling due to the formation of hydrogen bonds between the
solutions and polar groups on the polymer chains. These bonds cause water molecules to be adsorbed onto the
polymer chains, causing the physical structure of the polymer to open up and swell. Wright (1981) reports a number
of studies that have shown linear relationships between mass increase and volume change of CFRP laminates.

3. EXPERIMENTAL WORK
Experiments were carried out to measure the rate of radial solution uptake and diameter increase of CFRP tendons
immersed in water, salt water and CPS. The tendons were made of Tenax UTS 5131 fibres, volume fraction 0.63, in
a Bakelite EPR 4539 epoxy matrix. In each experiment, five specimens were prepared for testing in each solution.
The silica coating was removed from the tendons, which were then manually sanded smooth to a diameter of 4.2
mm. The solutions were formulated as described by Chin et al. (1999), using deionised water in all cases. The salt
water solution was of 3.5 % by mass sodium chloride in deionised water. The CPS was of 1.8 % by mass potassium
hydroxide, 0.68 % by mass sodium hydroxide and 0.5 % by mass calcium hydroxide in deionised water. All
solutions were maintained at 22 oC in airtight polypropylene flasks throughout the experiments.

In the uptake experiment, samples were cut to 150 mm length, dry weighed then submerged. Given the large ratio of
tendon length to diameter, axial solution uptake in the tendons can be considered negligible (Broughton and Lodeiro,
2000). Gravimetric sorption was used to measure the solution uptake: the tendons were removed from solution,
rinsed in deionised water, blotted dry, then weighed before being returned into solution. Readings were taken using a
Mettler AE160 balance of 0.1 mg resolution and the mass increase was calculated periodically using equation (3).

In the swelling experiment, samples were cut to 50 mm length. For each reading, three diameter measurements were
made of each tendon at distinct positions about the centre of the tendon length using a Universal Horizontal
Metroscope of 100 nm resolution. Dry readings were taken before submerging five tendons in each solution.
Tendons were rinsed in deionised water and blotted dry prior to taking each set of subsequent measurements.

468
4. EXPERIMENTAL AND MODELLING RESULTS
Experimental and modelling results are shown in Figures 1 to 6, where each data point represents the mean of five
readings. Error bars where shown give 95 % confidence limits based on the standard error of the mean, assuming a
normal distribution of data. Water, salt water and concrete pore solution are abbreviated to W, SW and CPS
respectively in all figures. Diameter increase, Øct, and volume increase assuming constant length, Vct, of the tendons
were calculated in the same way as mass increase. Figures 1 and 4 show observed increases in tendon mass and
diameter respectively over time. Figures 2 and 5 take logarithms of positive experimental data values, facilitating the
hypothesis of relevant relationships by considering lines of best fit. Increases in tendon mass and volume are directly
compared using lines of best fit in Figure 6. By summing the terms of equation (8) until the solutions converged,
curves predicting mass increase over the test period and up to saturation are shown in Figures 1 and 3 respectively.
Without data specific to the epoxy used; data for water, salt water and CPS uptake into epoxies at room temperature
studied by Chin et al. (1999) was modified using equations (4) and (5). Values of Mc∞ used were 0.525, 0.662 and
0.607 % respectively. Values of Dc used were 55.3 x 10-12, 109 x 10-12 and 69.9 x 10-12 cm2/s respectively.
0.4
0.7
W: experimental W: Ln(M ct ) = 0.471Ln(t ) - 3.91
0
0.35 SW: experimental SW: Ln(M ct ) = 0.501Ln(t ) - 4.07
CPS: experimental 0.6
CPS: Ln(M ct ) = 0.473Ln(t ) - 3.87
0.3 -1
Ln[Mass increase (%)]

Mass increase (%)

0.5
Mass increase (%)

0.25
-2 0.4
0.2 400 day duration
0.3 of test period
0.15 -3

0.1
0.2
W: prediction
W: prediction -4 W
SW: prediction SW 0.1 SW: prediction
0.05
CPS: prediction CPS CPS: prediction
0 -5 0
0 100 200 300 400 -2 -1 0 1 2 3 4 5 6 0 2 4 6 8 10 12 14 16 18 20
Time (days) Ln[time (days)] Time (years)

Fig. 1. Tendon mass increase Fig. 2. Log-log plot of tendon Fig. 3. Predicted long-term
with time mass increase with time tendon mass increase with time
1 0.7
W W: V ct = 1.22M ct - 0.0487
0.4 SW W: Ln(Ø ct ) = 0.274Ln(t ) - 4.05
0 0.6 SW: V ct = 1.20M ct + 0.0766
Ln[Diameter increase (%)]

CPS SW: Ln(Ø ct ) = 0.162Ln(t ) - 2.69

CPS: Ln(Ø ct ) = 0.194Ln(t ) - 2.90 CPS: V ct = 1.09M ct + 0.082
Diameter increase (%)

Volume increase (%)

0.3 0.5
-1
0.4
0.2 -2
0.3
-3
0.1 0.2
-4 W
0.1 SW
0.0 W CPS
-5 SW 0.0
CPS
-0.1 -6 -0.1
0 100 200 300 400 -4 -3 -2 -1 0 1 2 3 4 5 6 0.0 0.1 0.2 0.3
Time (days) Ln[time (days)] Mass increase (%)

Fig. 4. Tendon diameter Fig. 5. Log-log plot of tendon Fig. 6. Tendon volume
increase with time diameter increase with time increase with mass increase

5. DISCUSSION
The precise and repeatable results from the uptake experiment, shown in Figure 1, indicate relatively small increases
in tendon mass, in the order of 0.3 %, over a one year period. Figure 2 shows mass increase is proportional to the
square root of time in all cases: an indication of Fickian diffusion as predicted by equation (8). The predicted
diffusion behaviour shown in Figures 1 and 3 differs from the observed experimental results. This discrepancy is due
to the fact that different epoxies may have significantly different values of diffusion coefficient and mass increase at
saturation (Wright, 1981). Given that the modelling predictions were made based on data for a different epoxy to the
one used in this study, these discrepancies are to be expected. The difference in diffusion behaviour due to the type
of aqueous solution used are most likely attributable to the difference in specific chemical composition of each

469
epoxy. Whereas solutions containing greater concentrations of polar molecules diffused more favourably through the
material studied by Chin et. al. (1999), the solution type made little difference to the diffusion behaviour in this
study. Despite the discrepancies between the sets of data in Figure 1; given their similar behaviour, the model shows
it is possible to predict the rate of solution uptake in CFRP tendons, and hence corresponding effects, using diffusion
models and data that can be calculated from short-term thin film tests. When saturation of such tendons could take
over ten years to occur, as shown in Figure 3, this method facilitates the prediction of long-term behaviour by using
equations (4), (5), (6) and (8) to manipulate thin film diffusion data that can be obtained in less than one month.

In the swelling experiment, limited repeatability when locating tendons in the Metroscope, combined with deviations
in tendon roundness resulted in considerable data scatter. Best fit lines and equations in Figures 5 and 6 are therefore
indicative rather than absolute. Figure 5 shows diameter increases occurring between rates proportional to t 0.16 and
t0.27. A rate proportional to t0.25, within the bounds of results obtained, indicates a proportional relationship between
solution uptake and volume change: as suggested in Figure 6 and reported in the literature for laminate behaviour
(Wright, 1981). The type of aqueous solution made little difference to the rates of swelling observed. Figure 6 shows
volume increases in the order of 0.5 % over one year. If the tendons proceed to absorb solution as predicted in
Figures 1 and 3, and the relationship between volume and mass increase is linear, they may ultimately swell in
volume by 1 %. The significance of this swelling can be assessed in comparison with thermal effects. Transverse
coefficients of thermal expansion of CFRP tendons are in the order of 50 x 10-6 /oC (ACI Committee 440, 2003). A
temperature change of 50 oC, a theoretical seasonal variation, would cause a tendon volume change of 0.51 %. Thus,
swelling effects due to solution uptake may ultimately be greater than due to thermal effects: potentially affecting
the micromechanics of tendon failure, the tendon-concrete bond integrity, and placing a lower bound on feasible
thicknesses of concrete cover.

6. CONCLUSIONS
1. Solution uptake into CFRP tendons studied has been observed to behave according to Fickian diffusion models
2. A method has been presented to predict the long-term diffusion behaviour of solutions into CFRP tendons using
short-term thin film tests and Fickian diffusion models
3. Tendon volume appears to increase linearly with solution uptake in the CFRP tendons considered
4. The specific nature of the aqueous solutions considered had little effect on the diffusion or swelling behaviour
5. Tendon volume increases of approximately 0.5 % have been observed over one year and may ultimately be
greater than thermal effects, potentially compromising the integrity of prestressed concrete structures

ACKNOWLEDGEMENTS
The authors wish to thank Dr Michael Sutcliffe, University of Cambridge, for guidance given to the research project.
The support of Dr Giovanni Terrasi and SACAC, Switzerland, is much appreciated. The UK Engineering and
Physical Sciences Research Council (EPSRC) is gratefully acknowledged for sponsoring the work undertaken.

REFERENCES
ACI Committee 440 (2003). Guide for the Design and Construction of Concrete Reinforced with FRP bars, ACI,
Farmington Hills.
Barrer, R.M. (1984). "Diffusivities in glassy polymers for the dual mode sorption model". Journal of Membrane
Science, Vol. 18, pp 25-35.
Broughton, W.R. and Lodeiro, M.J. (2000). Techniques for Monitoring Water Absorption in Fibre-Reinforced
Polymer Composites, HMSO, London.
Carter, H.G. and Kibler, K.G. (1978). "Langmuir-type model for anomalous moisture diffusion in composite resins".
Journal of Composite Materials, Vol. 12, pp 118-131.
Chin, J.W., Nguyen, T. and Aouadi, K. (1999). "Sorption and diffusion of water, salt water and concrete pore
solution in composite matrices". Journal of Applied Polymer Science, Vol. 71, No. 3, pp 483-492.
Crank, J. (1975). The Mathematics of Diffusion, Oxford University Press, New York.
Springer, G.S. (1988). Environmental Effects on Composite Materials, Technomic Publishing co., Westport.
Wright, W.W. (1981). "The effect of diffusion of water into epoxy resins and their carbon-fibre reinforced
composites". Composites, Vol. 12, No. 3, pp. 201-205.

470
 Third International Conference on FRP Composites in Civil Engineering (CICE 2006)
December 13-15 2006, Miami, Florida, USA

EXPERIMENTAL STUDY ON BEHAVIORS OF CONCRETE BEAMS

STRENGTHENED WITH EXTERNAL PRESTRESSING CFRP TENDONS
Zhi Fang
(Professor, Hunan University, Changsha,Hunan,China)

Hongfang Li
(PhD candidate , Hunan University, Changsha,Hunan,China)

ABSTRACT
The purpose of present paper was to study the flexural behaviors of concrete T-beams prestressed with external
carbon fiber reinforced polymer(CFRP) tendons. Tests were carried out on 5 beams under four-point load, in which
four beams were reinforced with hybrid external CFRP tendons and steel rebar and another one only reinforced with
steel rebar. Such parameters as the different initial loading states at external CFRP tendon jacking and different
jacking stress in tendon were considered. Based on the test results, the loading capacity, deformation, ductility index,
cracking pattern, failure mode of those beams were studied. It was showed that the initial loading state on the beam
at tendon jacking is insignificant to the ultimate behaviors of the beams. Non-prestressing reinforcement can
improve the flexural behaviors effectively since it can lead to a more rational crack distribution and better ductility.
The ductility index of those beams prestressed with external CFRP tendons can reach about 2.5, which can meet the
requirement of engineering application.

KEYWORDS
prestressed concrete, beam, strengthen, Carbon Fiber Reinforced Polymer/Plastic(CFRP), externally prestressed,
flexural behavior

1. INTRODUCTION
The external prestressing is a post-tensioning technique which is becoming more popular both in new construction
and in strengthening of existing structures. The stress lag in external tendons can be prevented and material benefits
can be maximized using the technique to strengthen the existing structures. Corrosion free, high strength-to-weight
ratio enable CFRP composites to be a potential replacement for ordindary steel external tendons in engineering
applications. In contrast to the wide investigation on behaviors of concrete beams externally prestressed with steel
tendon(Tan and Ng, 1997; Ng, 2003; Tan and Tiandra, 2003; Wollmann et al., 2005), even though a number of
studies on flexural behaviors of concrete beams with external CFRP tendons have been carried out in recent
years(Mutsuyoshi and Machida, 1993; Grace and Sayed, 1998; Grace, 1999), it is insufficient to fully understand the
performance of the structures. The objective of this paper focus on the flexural behaviors of concrete beams
strengthened with external prestressing CFRP tendons.

2. EXPERIMENTAL PROGRAM

The test specimens included one ordinary reinforced concrete beam and four prestressed concrete beams with
external CFRP tendons, All beams were 4.2m in length and 3.9m in span. The rising angle of external tendon to base
line of the beam at deviators was 2o. 2 deviators with 5cm wide, 31cm deep for each tendon were located at the one-
third span separately and a stainless steel plate was placed at a saddle-like interface between the tendon and deviator
to reduce the friction. The jacking stress in tendons was about 0.30~0.45 fp, here fp referred to the guaranteed tensile
strength of tendon. Of 4 prestressed beams, the two of EB-45-N and EB-30-N were jacked at 0 applied load and
called new beams hereafter; another two of EB-35-S and EB-30-S jacked at 40kN applied load and called
stengthened beam hereafter. Because those strengthened reinforced concrete beams with short or middle span are

471
usually cracking and under dead load of about 50% total action effect, a 40kN pre-load of about 50% ultimate
capacity of beam RCB was applied on those strengthened beam EB-35-S and EB-30-S to simulate the initial loading
state of the beam at strengthening . All beams were tested under four-piont load. The details about cross section and
reinforcement were shown in Fig.1, test set-up and external tendon profile were shown in Fig.2, material properties
and effective prestress in tendons were listed in table 1.

Fig.1 Reforcement and Cross Section of Beams and Deviators Fig.2 Test set-up

Table 1 Material Properties and Effective Prestress in Tendons

σcon/fp PPR fe CFRP tendon Reinforcement Concrete

Beam No. % % MPa
Ap/mm2 Ep/GPa fp/MPa fy/MPa Es/GPa fcu/MPa Ec/GPa
RCB / / / / / / 347.3 201.3 53.5 35.1
EB-45-N 44.9 70.3 919.9 157.1 120 2100 347.3 201.3 62.5 36.3
EB-35-S 32.3 72.1 673.5 143.6 147 2550 347.3 201.3 62.5 36.3
EB-30-N 29.5 72.1 747.9 143.6 147 2550 347.3 201.3 56.6 35.5
EB-30-S 29.6 72.1 754.1 143.6 147 2550 347.3 201.3 54.8 35.3
Note：RCB is a reinforced concrete beam, EB-45-N and EB-30-N are prestressed concrete beams jacked at 0 applied loaded, EB-35-S and EB-30-S are prestressed concrete
beams jacked at 40kN applied load; partial prestress ratio PPR=Ap fp /(Ap fp +As fy )，Ap and As are the area of two CFRP tendons and reinforcement respectively, fp and fy are the
strength of CFRP tendon and reinforcement; σ con is jacking stress in CFRP tendon; fe is effective stress; Ep , Es and Ec are the modulus of elasticity of CFRP tendon,
reinforcement and concrete respectively; fcu is the cube strength of concrete.

3. TEST RESULTS AND DISCUSSION

3.1 Load-deflection and load-prestress increment response and failure mode in the beams

The measured load-mid-span deflection and load-prestress increment curves of the beams were shown in Fig.3, and
crack distribution on the beams at ultimate shown in Fig. 4, some results were listed in table 2.

The typical load-deflection curves of new beams were of nearly tri-linear type, The point A, B and C on the curves
were corresponding to the concrete cracking, steel yielding and concrete crushing or tendon rupturing, respectively.
The typical load-deflection curves of those strengthened beams demonstrated bi-linear nearly for they had cracked at
jacking. The mid-span deflection both at loading and after unloading was relative small before reinforcement
yielding since reinforcement and CFRP tendons were at linearly elastic state in this stage. After steel yielding, the
deflection increased rapidly with applied load increasing and the residual deflection also became larger. The load-
prestress increment curves were in the similar manner to those load-deflection relationships. The prestress increment
in external CFRP tendon was nearly proportional to the mid-span deflection, which also could be found in those
beams internally prestressed with unbounded steel strands (Burns et al., 1978).

It also could be seen from fig.3 there almost existed an identical post cracking manner both in new and strengthened
beams, which indicated that the initial load state at the tendon jacking was insignificant to the flexural behaviors of
the beams if only those beams were not beyond yielding at the initial applied load.

All beams failed with concrete crushing except for beam EB-45-N failed suddenly with CFRP tendon rupture near
the deviator since the CFRP tendon in the latter was of lower strength, higher effective stress and worse quality. It is
expected that failure mode of the beams is concrete crushing rather than tendon rupturing for engineering safety.

472
The crack distribution in fig.4 showed that the crack space in the beams prestressed with external CFRP tendons was
larger than that in reinforced concrete beam. Compared with the crack distribution in new beam EB-30-N, the crack
space in strengthened beam EB-30-S was smaller and similar to that in reinforced concrete beam RCB, which
resulted from the strengthened beam cracked at initial applied load before external tendon jacking. Non-prestressing
reinforcement could lead to a more rational crack distribution in externally prestressed concrete beams.
1000
200 200

Prestress Increment/MPa
C
800
160 RCB 160 B
EB-30-N EB-45-N
EB-30-S EB-30-N EB-45-N

Load/kN
120 600
120
Load/kN

EB-45-N EB-35-S EB-30-N

A
EB-35-S A EB-30-S EB-35-S
80 80 400 EB-30-S

40 200
40
0 0
00 20 40 60 80 100 120 0 200 400 600 800 1000 0 20 40 60 80 100 120 140
Mid-span Deflection/mm Prestress Increment/MPa Mid-Span Deflection/mm

(a) Load-deflection curves (b) Load-prestress increment curves (c) Deflection-prestress increment curves
Fig.3 Relationship of deflection, load and prestresse increment

a) RCB

b) EB-45-N c) EB-35-S

d) EB-30- N e) EB-30-S
Fig 4 Crack Distribution of Beams at Ultimate

Table 2 Comparison between test and computed values

Concrete Cracking Reinforcement Yielding Ultimate State

Beam Prestress Prestress Prestress
Load Deflection Load Deflection Load Deflection
increment increment increment
/kN /mm /kN /mm /kN /mm
/MPa /MPa /MPa
RCB 19 2.23 - 65 11.06 - 82 146.2 -
EB-45-N 68 4.32 28.27 140 17.62 120.86 (158) (51.9) (335.86)
EB-35-S - - - 127 17.50 109.25 168 168 461.99
EB-30-N 60 3.63 32.00 120 13.15 133.31 186 98.70 886.00
EB-30-S - - - 120 10.82 113.00 185 110.68 854.00
Not: 1. EB-45-N failed accidently with CFRP tendon rupture near the deviator, the data in bracket was measured at the test terminating state.

2. EB-35-S and EB-30-S were strengthened after cracking, so there was no data at cracking state.

3.2 Ductility of the beams

The ductility index µ of beams prestressed with external CFRP tendons is defined on the energy concept
as(NAAMAN and JEONG, 1995):
µ = ( Etot / Eel +1) / 2 (1)
Where Etot=Epl+ Eel, Etot refers to the total energy, Epl to the plastic energy, Eel to the elastic energy, and they are
shown schematically in Fig.5. The ductility indexes of the beam RCB, EB-45-N, EB-35-S, EB-30-N and EB-30-S
are 5.86, 2.19, 2.54, 2.45 and 2.25, respectively. The index of EB-45-N is smaller than those of other beams for
beam EB-45-N failed suddenly with CFRP tendon rupture near the deviator. It could be seen that the ductility index
for beams prestressed with external CFRP tendons would be larger than that for the beams prestressed with internal
CFRP tendons which was less than 2 generally(Fang and Yang, 2005). So the ductility of the beams prestressed with

473
external CFRP tendons can meet the requirement of engineering application if only appropriate non-prestressing
reinforcement is placed in the beams.

Fig.5 Sketch of load-deflection curve and energy component

4. CONCLUSIONS
The following conclusions can be drawn from the above investigations on the flexural behaviors of concrete beams
prestressed with external CFRP tendons:
(1) There almost existed an identical post cracking manner both in new and strengthened beams, which indicated
that the initial load state at the tendon jacking was insignificant to the flexural behaviors of the beams if only those
beams were not beyond yielding at the initial applied load.
(2) The load-prestress increment in tendon curves were in the similar manner to those load-deflection relationships,
and the prestress increment in external CFRP tendon was nearly proportional to the mid-span deflection, which also
could be found in those beams internally prestressed with unbounded tendons
(3) The appropriate non-prestressing reinforcement can improve the flexural behaviors of beams prestressed with
external CFRP tendons effectively since it can lead to a more rational crack distribution and better ductility.
(4) The ductility index of those beams prestressed with external CFRP tendons can reach about 2.5, which can
meet the requirement of engineering application.

REFERENCES
Burns, N.H., Charney, F.A. and Vines, W.R.(1978). “Tests of One-Way Psot-Tensioned Slabs With Unbonded
Tendons”. PCI JOURNAL, Vo. 23, No. 5, pp. 66-83.
Fang, Z and Yang, J (2005). “Behaviors of Concrete T-beams Prestressed with Partially Bonded CFRP Tendons”,
11th International Conference on Fracture, Editors: Alberto Carpinteri, Turin [ Italy], pp. 448.
Grace, N.F. and Sayed, G.A.(1998). “Ductility of Prestressed Bridges Using CFRP Strands”. Concrete International,
Vo.20, No. 6, pp.25-30.
Grace, N.F. and Sayed, G.A.(1998). “Behavior of Externally Draped CFRP Tendons in Prestressed Concrete
Bridges”. PCI JOURNAL, V. 43, N. 5, pp. 88-101.
Grace, N.F. (1999). “CONTINUOUS CFRP PRESTRESSED CONCRETE BRIDGES”. Concrete International,
Vo.21, No. 10, pp.42-48.
Mutsuyoshi, H. and Machida, A. (1993). “Behavior of Prestressed Concrete Beams using FRP as External Cable”.
SP-138, American Concrete Institute, 401- 418.
Ng, C.K.(2003). “Tendon Stress and Flexural Strength of Externally Prestressed Beams”. ACI Structural Journal,
V.100, N.5, pp. 644-653.
Tan, K.H. and Ng, C.K.(1997). “Effects of Deviators and Tendon Configuration on Behavior of Externally
Prestressed Beams”. ACI Structural Journal, Vol. 94, No. 8, pp. 13-22.
Tan, K.H. and Tjandra, R.A.(2003). “Shear Deficiency in Reinforced Concrete Continuous Beams Strengthened
with External Tendons”. ACI Structural Journal, V.100, N. 5, pp. 565-572.
Wollmann, C.L.R., Kreger, M.E., Rogowsky, D.M. and Breen, J.E. (2005). “Stresses in External Tendons at
Ultimate”. ACI STRUCTURAL JOURNAL,V.102, No.2, pp. 206-213.
NAAMAN, A. E. and JEONG, S. M. (1995). “Structural ductility of concrete beams prestressed with FRP tendons”.
Non-metallic (FRP) reinforcement for concretes structures, Proceeding of the 2th international RILEM symposium
(FRPRCS-2)], Editor: Taerwe, E& FN Spon, London, pp. 379 –386.

474
Third International Conference on FRP Composites in Civil Engineering (CICE 2006)
December 13-15 2006, Miami, Florida, USA

POST-STRENGTHENING OF PRESTRESSED CONCRETE BEAMS

USING UNSTRESSED AND PRESTRESSED CFRP STRIPS
Mohammad Reza Aram
(PhD student, Faculty of Civil Engineering, University of Tehran, Tehran, Iran)

Masoud Motavalli
(Assistent Professor,University of Tehran and Empa, Swiss Federal Laboratories for Materials Testing and
Research, Dübendorf, Zürich, Switzerland)

Christoph Czaderski
(Project leader, Empa, Swiss Federal Laboratories for Materials Testing and Research, Dübendorf, Zürich,
Switzerland)

ABSTRACT
Four prestressed concrete beams were constructed and tested to investigate the effectiveness of flexural post-
strengthening with prestressed CFRP strips. One of the beams served as reference beam, another was bonded with
unstressed CFRP strip, and two specimens were strengthened with prestressed CFRP strips in two prestressing
levels. Gradient method was used for the anchorage of the prestressed CFRP strips. Experimental results and failure
modes are explained in this paper.

KEYWORDS
Strengthening, Prestressed CFRP, Prestressed concrete beams, Gradient method

1. INTRODUCTION
Today there is a great demand on post-strengthening of concrete structures. A well-known material for post-
strengthening is CFRP, Carbon Fiber Reinforced Polymer. To take full advantage of the material, it is beneficial to
apply prestressed strips. Relatively little work has been carried out on the strengthening of internally prestressed
concrete structures using prestressed CFRP strips (e.g. El-Hacha et al., 2004). Most of references investigated the
application of prestressed CFRP strips by mechanical anchorage system. An anchorage system was developed at
Swiss Federal Laboratories for Materials Research (Empa) without using any mechanical devices for anchoring of
the end parts of the CFRP strips (Stöcklin and Meier, 2003). The prestressing force is gradually reduced to zero at
the end parts of the strips. At Empa a project for strengthening of large-scale concrete bridge girders with
prestressed CFRP strips anchored using gradient method (Czaderski and Motavalli, 2005) was performed. The
project was the motivation to carry out a new set of small scale tests which aims at investigation of debonding of
prestressed CFRP strips anchored with gradient method. In this paper, results of this small scale tests are presented.

2. EXPERIMENTS
Test specimens

Four 2.4m-long prestressed RC beams were constructed with the dimensions shown in Figure 1. All the beams were
loaded in a four-point bending test scheme with a span of 2.0m. In attempt to examine the effect of prestressed
CFRP strips, one beam was not strengthened and served as a control to the other three beams, which were
strengthened with either unstressed or prestressed (two different stress levels) CFRP strips. A short description of
each beam appears in Table 1. The concrete compressive and tensile strength tests were conducted using cylindrical

475
and cubic samples respectively at the ages of 28 days and at the time of the experiments. Steel bars with a nominal
yield strength of 500 MPa and elastic modulus of 205 GPa were used for the longitudinal and shear reinforcement.
Prestressing strands had a yield strength of 1618 MPa and an ultimate strength of 1869 MPa with an assumed elastic
modulus of 195 GPa.

Figure 1: Dimensions, test setup and monitoring devices of the tested beams

Sika CarboDur unidirectional carbon/epoxy strips with a nominal width of 50 mm and thickness of 1.2 mm were
used for strengthening. The flexural elastic modulus, Ea , and the flexural tensile strength, fta , of the epoxy adhesive
were determined through tests of adhesive samples at the ages of 7 days. The compressive strength of the adhesive,
fca , was evaluated through cubic compression tests. The mechanical properties of the CFRP composite strips, the
adhesive and the concrete are summarized in Table 1.

Table 1. Material properties of test beams

Concrete Adhesive Strip

Beam fc+ (MPa) ft+ (MPa) Ec+ (MPa) Ea (MPa) fta (MPa) fca (MPa) Ef* (GPa) ff* (MPa)
Pb1 (reference beam) 48.9 3.1 33400 --- --- --- --- ---
Pb2 (strengthened with
48.1 3.3 33400 12600 42.6 105.6 165 >2800
unstressed CFRP)
Pb3 (strengthened with
51.5 2.8 37450 12200 41.8 128.9 165 >2800
prestressed CFRP, 32% of ff)
Pb4 (strengthened with
47.1 2.3 34400 12200 41.8 128.9 165 >2800
prestressed CFRP, 16% of ff)
®
+ at the time of the experiments, * specified by Sika (Ef: mean value of tensile elastic modulus, ff: minimum value
of tensile strength)

Prestressing technique

To prevent premature failure of prestressed CFRP strips due to high shear stresses at the ends, the strips have to be
anchored with special devices. This mechanical anchorage is expensive, difficult to install and often subjected to
corrosion. To overcome these anchorage problems, the prestressing force can be anchored using gradient method
developed at Empa (Stöcklin and Meier, 2003). In this method the prestressing force is reduced gradually towards
both ends of the strip in steps. The CFRP strips were bonded to the concrete by using an epoxy based adhesive
(Sikadur-30 LP) which cures faster at elevated temperatures (80˚C). A special computer controlled device was used
to control the prestressing and heating process. After heating of the strip and therefore curing of the adhesive in
sector 1, the prestress force was reduced. Then, sector 2 was cured, followed by a reduction in prestress force and so
on. Approximately 3½ hours were required to finish the gradient and the strip was then cut at the end. Lastly, the
free length of the strip between the gradients was also cured by resistance heating, to ensure good adhesive quality.
Each strip with a cross-section of 60 mm2 was prestressed to approximately 60 kN (Beam Pb3) and 30 kN (Beam
Pb4) in the same gradient length and number of steps. These prestressing forces correspond to two prestress levels of
32% (Beam Pb3) and 16% (Beam Pb4) of the nominal tensile strength of the strip (2800 MPa).

Test setup

476
The test setup, the various monitoring devices and their locations along the beams are shown in Figure 1. The beams
were loaded mid-span displacement controlled using a testing machine with two 150 kN hydraulic actuators. The
loading rate was 0.02 mm/s. The monitoring devices include linear voltage displacement transducer (LVDT), dial
gauge, thermometer, strain gauges (SG) and mechanical strain gauges on CFRP and concrete. Slippage between
CFRP and concrete is obtained from difference between mechanical strain gauge measurements on CFRP and
concrete. The manual mechanical strain gauge measurements were performed during load steps where the
deformation at mid-span kept constant and the load was adapted. From mechanical strain gauge measurements
length of 100mm, the mean value strain along each 100mm length of CFRP can be determined. Cracks width were
also measured in some load steps.

Results

The load versus midspan deflection curves for the beams are illustrated in Figure 2. These curves include a linear
response up to the cracking, a cracked behaviour up to the yielding of the internal unstressed rebars and an inelastic
response up to ultimate load. The curves reveal that the strengthening process has increased the load resistance
capacity of the beam up to 17% for unstressed CFRP application (Pb2). For Pb3 and Pb4, in spite of compression
effect of externally prestressing force on the concrete cross section, the load capacity increase is less than unstressed
case due to premature debonding of strips. It can also be found that beams strengthened with prestressed strip have
lower deflections at the service load and at failure (ductility). Another effect of prestressing was that cracks width
decreased enormously in comparison to the unstressed beam.
Failure of control beam, Pb1, occurred due to crushing of the concrete in compression. The maximum compressive
strain measured on the top of midspan was 0.00296. The cracking pattern, as shown in Figure 3, includes vertical
flexural cracks around midspan and diagonal shear-flexure cracks in the shear spans.
The failure of beam Pb2 was associated with a rapid and unstable debonding of the CFRP strip followed by concrete
crushing at top. This debonding was initiated at a large flexural crack in the midspan which is named C in Figure 3.
The failure of beam Pb3 was a sudden failure due to delamination of the CFRP strip along with the concrete cover
beneath the internal reinforcement in the midspan (Figure 3). Additional CFRP strain from loading was 3 ‰ (SG4)
at failure added to the initial prestressing strain results in a total CFRP strain of 9 ‰.
The failure of beam Pb4 was a rapid, sudden separation of the CFRP strip. The separation started at a shear crack in
the midspan which was formed suddenly near crack M (Figure 3). Maximum measured slip is for this shear crack.
Such a crack is usually associated with discontinuity in the vertical deflections and yields different deflections left
and right of the cracked section. In most cases, the thin adhesive layer cannot accommodate such vertical
displacements and cannot resist the corresponding vertical normal stresses developed. The beam after failure appears
in Figure 3, which clearly shows the discontinuous vertical deflections at the lower face of the beam and the
delamination of the CFRP strip. For Pb4, additional CFRP strain from loading was 4.9 ‰ at failure which results in
a total CFRP strain of 7.9 ‰ including the prestressing strain. Based on the strain gauges SG1, SG4, it appears that
the concrete didn’t exceed its crushing strain and the CFRP strip didn’t reach its rupture strain. It means that beams
Pb3 and Pb4 didn’t exceed its potential flexural capacity due to the premature debonding failure.

120.00
110.2 kN 110.4 kN
100 kN
100.00 94 kN

80.00
Pb1
Load F (kN)

Pb2
60.00
Pb3
Pb4
40.00

20.00

0.00
0.00 5.00 10.00 15.00 20.00 25.00
Deflection at midspan measured by LVDT2 (mm)

Figure 2: Experimental load deflection diagrams; Pb1: control beam, Pb2: strengthened with unstressed
CFRP, Pb3 and Pb4: strengthened with prestressed CFRP

477
 Pb1

Pb2

Pb3 Pb4

Figure 3: Test beams after failure

A summary of measured results can be found in Table 2. This table shows that for unstressed strip, 6 ‰ strain and
3 MPa bond shear stress were critical values for CFRP. For prestressed strips anchored with gradient method, the
maximum achievable CFRP strain depends on prestress level and anchorage length on uncracked region. The
gradient anchorage part of these small beams was in cracked region with high shear stresses and total CFRP strain
was limited to 9 ‰ for Pb3 and 7.9 for Pb4. Maximum additional slip generated from loading was 0.165 mm and
0.193 for Pb3 and Pb4 respectively. Gradient force of strip at the anchorage zone produces an initial shear stress
between CFRP and concrete. In Table 2 total maximum bond shear stress including this initial stress is presented.

Table 2: Maximum measured values

Beam Failure load (kN) Max. CFRP strain (‰) Max. slip (mm) Max. bond shear stress (MPa)
Pb1 94 --- --- ---
Pb2 110.4 6.7 not measured 3.1
Pb3 100 9.0 0.165 4.3
Pb4 110.2 7.9 0.193 3.2

3. Conclusion

An experimental investigation of four prestressed concrete beams strengthened using unstressed and prestressed
CFRP strips was presented. Prestressing the strip caused decrease in beam deflection and crack width if compared to
the unstressed strips. However, the failure load could not be increased, it was even smaller. Furthermore, the
ductility was clearly smaller. The results of the experiments point out that gradient anchorage method was not
effective because the gradient anchorage was in the region of high shear stresses. The short beams provided not
enough anchorage length on uncracked regions. This method would be more effective for large span beams like
bridge girders. Strengthening using prestressed strips or plates has to be planned very carefully and the gradient
anchorage should not be in regions with high shear stress. More research is needed in order to determine appropriate
anchorage length as well as minimum beam length for applying this novel prestressing technique more efficiently.

4. References

El-Hacha, R., Wight, R.G., and Green, M.F. (2004). “Prestressed carbon fiber reinforced polymer sheets for
strengthening concrete beams at room and low temperatures”. Journal of Composites for Construction, Vol. 8, No.
1, pp 3-13.
Stöcklin, I., and Meier, U. (2003). “Strengthening of concrete structures with prestressed and gradually anchored
CFRP strips”. Proceeding of FRPRCS-6, Singapore.
Czaderski, C., and Motavalli, M. (2005). “Large-scale concrete bridge girder strengthened with prestressed CFRP
plates anchored using gradient method”. Composites Part B: engineering, submitted for consideration for
publication.

478
Third International Conference on FRP Composites in Civil Engineering (CICE 2006)
December 13-15 2006, Miami, Florida, USA

REINFORCEMENT TECHNOLOGY BY PRESTRESSED CARBON

FIBER REINFORCED PLASTICS (CFRP)
Shouping Shang, Yongjun Jin, Hui Peng, Ming Wang, Qiaojuan Leng
（College of Civil Engineering, Hunan University, Changsha, PR China）

ABSTRACT
The carbon fiber reinforced plastics (CFRP)，which has the great characteristics of inoxidizability, durability as
well as high specific tenacity (unit quality intensity), is among the most ideal reinforcement materials. However,
If CFRP is used directly in strengthening building and bridge structures; it can make certain enhancement to the
load, but seldom enhance the rigidity distinctly. Moreover, this way cannot fully develop the CFRP’s
superiority of high strength, and may lead to component’s debonding failures.
It is indicated by some related experiments that using prestressed carbon fiber reinforced plastics (CFRP)
technique not only can effectively improve the rigidity of the reinforced concrete component, but also can
greatly restrain crack development. The reinforced concrete structure which is reinforced by prestressed CFRP is
not easy to appear debonding failures, and can fully reach its high intensity. It takes less expenditure compared
to other reinforcement techniques also.
In conclusion, Reinforcement technology by prestressed CFRP has wide applicable prospect.

KEYWORDS
Prestressing force，CFRP，Strengthening，Stretching

1. INTRODUCTION

As compared to conventional strengthening technology, carbon fiber strengthening technology possesses

obviously synthetic advantages. For example, it has extensive applicability，facilitative construction，swift
construction velocity, small disturbance，high reinforcement efficiency, easily guaranteed quality，long service
time，low synthetic construction cost and it doesn't increase the deadweight and volume of the structural
component and so on.
With the continual development of the research work and the practical engineering applications, researcher
found its shortcomings at the same time of taking cognizance of its advantages. The high-intensity
predominance of CFRP can not exert fully and that will cause the material strength waste. The bonded amount
of CFRP is limited restricted by the adhesion agent and destruction of bond easily occurs. In addition, the
rigidity reinforcement effort of CFRP is indistinct and the deformation of the structural member can't be
restrained availably.
The application of prestressing force in CFRP for strengthening concrete structure members is strengthening
method which has emerged in the last few years. Compared with the common concrete strengthening
technology, its technologic advantage is very obvious. Besides the merit of the common CFRP strengthening
technology, it improves the deficiency of the common CFRP strengthening technology.

1.1 Utilize the characteristic of the CFRP effectively and give full play to the high-intensity advantage of
CFRP.

Imposing prestressing force on CFRP will raise the starting point of bearing load. The CFRP can participate in
working immediately and play a greater role at the beginning of the strengthening member carrying load.
Compared with the non-prestressed CFRP reinforcement member, the high-intensity advantage of CFRP can be
exerted ahead of time. Moreover, bonding a layer of CFRP cloth with high level initial stress is equivalent to
bonding several layers of non-prestressed CFRP and makes the reinforcement technology with CFRP have
preferable economic benefit.

1.2 The obvious effect of the rigidity strengthening restrains the structural member deformation and the

479
appearance and development of the cracks.

The prestressing force can restrain the appearance and the development of the crack of the structure member,
and so improve the sectional rigidity remarkably. Prestressing force can also improve and adjust the stress of the
original structure, eliminate the phenomenon of stress stain lagging which is caused by the initial load of the
structure and reduce the deflection of the flexural member obviously though allowance for camber.

1.3 Seldom emergence of destruction of bond and desquamation destroy

First of all, because the increment of the stress and strain in the CFRP of the prestressed member is less than that
in the CFRP of the non-prestressed member, the shear deformation the of epoxy resin layer and the shear stress
transmitted by the epoxy resin layer is smaller than these of the non-prestressed member and the prestressed
member is not apt to appear destruction of bond. Secondly, because the CFRP has been stretched while making
prestressed test piece, it is extraordinary flat while being bonded. The CFRP will not deform with the test
specimen surface, and not follow the surface undulation of structural member, which reduce the peel stress
caused by curvature under combined tension and bending condition.

2. THE DIFFICULT AND THE KEY TECHNOLOGY OF STRENGTHENING TECHNOLOGY

WITH PRESTRESSED CFRP.

2.1 Technical difficulty

CFRP is one kind of material with high tensile strength and low shear strength. Because of that special character
of the CFRP, it is usually to paste the CFRP and jig together though adhesion agent, and the CFRP is anchored
by the cohesive force of the adhesion agent. There are a lot of problems in this stretching mode.
First of all, it can't guarantee uniformly stretching. The CFRP is very apt to tearing failure during the stretching
process. In addition, because the shearing strength of the adhesion agent is limited, the prestressing force value
applied can't be too high. Though the CFRP is stretched successfully and pasted to the surface of concrete,
there will be some CFRP silk which reach high stress value in advance and be destroyed, thereby influence the
effective exertion of the material strength. Secondly, the adhesion agent can't function without solidification,
thus it adds a solidification cycle in the whole process, and prolong the construction cycle. Furthermore, if we
use one-off anchorage to apply prestressing force, the cost of reinforcement will increase.
The application of prestressing force to the CFRP will make the inherent advantages of the CFRP strengthening
technology lost, including simple process and convenient construction. Therefore, how to ensure the whole rang
of process of stretching and plastering simple and practical enough while exerting prestressing force to the
CFRP uniformly and effectively has become the difficulty of the study of this technology.
Strengthening concrete flexural member with prestressed CFRP technology differed widely from the traditional
CFRP strengthening concrete flexural member technology, and the flexural performance of reinforcement
member will inevitably have great difference, so the traditional design calculation theory of the concrete beam
strengthening with CFRP can't supervise the design of concrete beam strengthening with prestressed CFRP. In
order to make strengthening concrete flexural member with prestressed CFRP to be a practical engineering
technology, we must study and grasp the flexural performance of the concrete beam strengthening with CFRP,
then bring forward corresponding design calculation theory and practical formula. This is another difficulty of
the strengthening technology with prestressed CFRP.

2.2 Key technology

In order to solve the difficulty above-mentioned, though iterative experimental study, we have developed the
prestressing force strengthening construction machinery which is suitable for prestressed CFRP cloth and plate
materials
The stretching rig of the CFRP cloth adopts friction self-lock mechanism. The hardcore is a jig composed of two
semi-cylinders. The CFRP cloth enwind on the semi-cylindrical jig, such design make the CFRP cloth could
self-regulate the stress distribution under the function of the tensile force. The actuation force in the tensioner of
CFRP cloth is induced by the handle. The impetus of transmission system guide screw rotation, which causes the nut to
move. When the nut moves along the steering bar to the limiting position, the carbon fiber cloth material starts to be pulled.
When the carbon fiber cloth material is stretched to predetermined condition, stop inputting actuation external force. The
implement is promoted by the equipment, then the carbon fiber cloth materialist glued strictly onto the member which will
be reinforced. The practices indicate that the design make the CFRP cloth could reach a higher prestressing level

480
when being stretched.
The stretching rig of the CFRP plate is different from the stretching rig of CFRP cloth, because of the special
physical property of the CFRP plate compared with cloth material. The construction machinery is composed of
tensioner and jig. The tensioner consists of a fixed end with mobile rod and a movable clamp which could move
along the mobile rod. And the jig consists of a piece of soleplate with tooth shape grain and four pieces of clamp
splice. When the CFRP plate is being stretched by the implement. The plate and the member, the rubber, and the bottom
surface must be the strictly contacted. And that is realized by the making slope on the member surface and gelatinizing
satiety. The partial tiny slit may be improved by the external angle steel.

Figure 1: The schematic diagram of CFRP sheets rig and CFRP cloth machine

The prestressed CFRP strengthening construction process only add two links compared with the traditional
CFRP strengthening process, including fixing construction machinery and stretching, and is convenient as
traditional CFRP strengthening process. The actual engineering application proves that the construction process
is advanced and simple and can obtain favorable effect.
In addition, in order to put forward the design theories to guide the application of the prestressed CFRP
strengthening technology to practice in the engineering, it is necessary to study and grasp the stressing
performance of the concrete flexural member strengthened with prestressed CFRP. According to experiment
result and the mechanics model built up, we put forward a design method of the strengthen technology with
prestressed CFRP.

3. THE MAIN TEST RESULTS

In order to ascertain the performance of flexural member strengthened with the CFRP(Carbon Fiber Reinforced
Plastic ) sheet material nine piece of small flexural member (100×150×2000㎜) strengthened with the CFRP
cloth material are tested . The primary test result is showed in table 3-1.
According to the test result, the cracking load of concrete and the yield load of steel bar will increase remarkably
in the member strengthened by prestressed CFRP material, which will make the flexural rigidity of the member
increase according to the prestressing force level of test piece. However, the using of prestressed CFRP material
has on effect on the ultimate bearing resistance of structural member. The different midspan flexural deflections
of two kinds of test pieces subjected to the yield load of the standard test piece reflect the influence of the
prestreesing force level on the flexural rigidity of test pieces. According to the data display，the strengthen
technology with prestressed CFRP adequately bodied forth the predominance on rigidity strengthening.
It can be observed from the table that the tensile deformation of CFRP material in non-prestessed CFRP cloth
reinforced member is restricted to 10.6% of its theoretical limits deformation (0.018) when the tension
reinforcement yield. After stressing, but the tensile deformation of CFRP with maximal initial stress level
reaches 60% of the theoretical limits deformation when the tension reinforcement yield，the increase degree is
five times of non-prestressing force CFRP ，it is observed that prestressing force cause the high performance of
CFRP bring into full play.
Experimental study on the performance of flexural members reinforced by CFRP plates has started recently. In

481
the previous period, we have developed a stretching rig and jig which are applicable to prestressed CFRP plates
strengthening，judged from the test result, the effect was beyond compare. In the next step, we should take the
study on fatigue performance of CFRP reinforcement member.

Test Result of Prestressed CFRP Reinforcement Member 3-1

Serial Initial Cracking Percentage Yield Percentage Ultimate Ratio Midspan Ratio
Number Strain Load Of Load Of Load Deflection
εPi /KN Increase /KN Increase /KN In Force
（×10-6） Of
12.3KN
/mm
B1~4
B 0 3.5 - 12.25 - 22.1 100% 7.5 100%
B5 B 6000 - - 16.0 30.6% 18.0 81.1% 4.01 53.5%
B6 B 6000 9 157% 17.0 38.8% 28.0 126.1% 3.81 50.8%
B7 B 6000 8 129% 20.0 63.3% 26.0 117.1% 4.23 56.3%
B8 B 3300 5 43% 14.0 14.3% 21.5 96.8% 4.16 55.4%
B9 B 8400 11 214% 20.0 63.3% 25.0 112.6% 2.24 29.8%
Notes: B5（prestressing test piece）have loading history before strengthening, so have no cracking load indices.

4. EXAMPLE OF ENGINEERING APPLICATION (PHOTOGRAPHS)

Not long ago,bridge strengthening apply these strengthening technology in Liuyang County,Hunan province,
China. Because of long service time of this bridge, the traffic capacity and traffic loads exceed the, design index.
The main girder emerge quite a lot and wide cracks, some cracks already exceed the standard of governing code.
According to the appraisal of related department,the bridge was in danger. Owing to better quality of concrete
and well condition of foundation works, strengthening this bridge is greatly fit for applying ,prestressed CFRP.
Strengthening 10 pieces of T-section beam rib which the length is 19.5m,the height is 1.245m,and the width is
0.175m, the practical engineering time was about 20 days, not includeing the time of preparing material and
corollary equipment. As seen from the detection result of static test, the strengthening effect was beyond
compare.

5. CONCLUSIONS

As seen from ongoing experimental study and practical engineering application, the performance of prestressing
force CFRP strengthening technology greatly superior. but we do not know well of this strengthening
technology, at the same time, for the sake of preferably exerting the predominance of this strengthening
technology, it remains to require the researchers to take the study about durability, fatigue performance,
interfacial blocking property and the influence of prestressing force, concrete creep, stress loss, and so on to
reinforcement member performance, and so on.

6. REFERENCES

Technical Specification for Strengthening Concrete Structure with Carbon Fiber Reiforced polymer
Laminate(CECS146:2003). Beijing：China Planning Press. 2003
Feiwei, Jiang shiyong, Peng feifei, et al. (2003). The experimental study of prestressing carbon cloth
strengthening concrete flexural member. Sichuan Building Research,Vol. 29, No.2.
Shang shouping, Peng hui,Tong hua, et al. (2003). The flexing resistance study of the prestressing carbon
cloth material strengthening concrete flexural member. Architecture Structures Journal, Vol. 24, No. 5.
Zhang tanxian, Lv xilin, Xiao dan, et al. (2005). Experimental study on primary and secondary load for RC
beam strengthened with Prestressed CFRP sheet. Structural Engineer, Vol. 21, No.2.

482
Third International Conference on FRP Composites in Civil Engineering (CICE 2006)
December 13-15 2006, Miami, Florida, USA

WEDGE ANCHORAGE FOR LOADED OR PRESTRESSED FRP

Stefan L. Burtscher
Research Associate, Vienna University of Technology, Vienna, Austria

ABSTRACT
Wedge systems are used for steel tendons since many years with great success. The application of conventional
wedge anchorages for CFRP tendons leads to a high lateral stress peak at the tip of the wedge that destroys the
tensile element before the tensile strength is reached. In the paper the patented “Composite Wedge anchorage”
(Burtscher, 2004) is described. The idea is that the wedge is formed of two materials with high and low modulus of
elasticity. The material with low modulus of elasticity is arranged parallel to the tensile element and the thickness
increases from the rear to the tip of the wedge. This influences the wedge stiffness normal to the CFRP element and
distributes the pressure and shear stresses in the CFRP-wedge interface. The pressure and shear stress distributions
determined by numerical simulations are continuous without stress peaks. The experimental results clearly show that
the anchorage is very efficient and is able to anchor CFRP strips with thicknesses of 1.2 mm with an efficiency of
more than 100 %. All 1.2 mm strips failed by exceeding the tensile strength of the strip and the failure did not occur
in the anchor region. The tests on 2.5 mm thick CFRP that were bonded to the wedge reached efficiencies of 84 and
87 %. Tests on the same CFRP strips, but with a friction interface even gave an efficiency of 93 %: The failure in all
tests occurred by slipping of the strips and not due to exceeding the transversal strength. The improved friction
surface of the next generation of wedges will allow an even higher efficiency of the anchoring system.

KEYWORDS
Wedge anchorage, composite wedge, prestressing, CFRP

1. INTRODUCTION
In recent years strengthening of concrete structures has become a large application field for carbon fiber reinforced
plastics (CFRP). The structures are strengthened by bonding or prestressing the strips on the tensed surface. Bonded
strips increase the load bearing capacity of the structure. When the strips or rods will be prestressed an additional
benefit is that the deflections under load can be reduced and less CFRP strips are often necessary. In some cases it is
also possible to increase the durability of the structure. Anchorages developed for steel tendons cannot be applied
for FRP tendons because of the sensitivity to transversal pressures. Therefore, special anchorages are necessary. In
the passed decades investigations on cast anchors with resin or grout, anchors with expansive material, spike
anchors, wedge anchors, other clamping mechanisms and various combinations were found in literature, see e.g.
Nanni et al. (1996), Pincheira et al. (2001), Stresshead (2005), Horvatits et al (2003), Meier (1996), Andrä (2005).
High anchoring loads were found for cast anchorages and other systems, where the CFRP tendons are fixed by
bonding using resin. For easily installable wedge anchors the transferable loads were low.

For economic purposes it is important that the anchorage is able to anchor high loads, easy to apply, produced at low
cost, easy to handle on the construction site and necessitates only a short time for the anchoring process. These
requirements can be met by applying the Composite Wedge System (Burtscher, 2004), which is presented in this
paper. The anchorage combines the advantages of the wedge anchors used for prestressing high strength steel
tendons, with the ability to anchor high loads coming from CFRP strips or rods.

First numerical simulations on conventional wedge anchors and the composite wedge anchor are presented to
explain the working principle of the Composite Wedge anchorage. Next, experiments on 1.2 mm and 2.5 mm thick
CFRP strips are presented, that confirm the high efficiency of the anchorage system. In the experiments the interface
between wedge and CFRP strip was established by bonding using an epoxy adhesive, but also by a simple friction
interface. Several tests with the friction interface were performed. Here only one test on a 2.5 mm thick CFRP strip
is presented. Finally, some further economic designs that will be investigated in the future will be presented.

483
2. STRESS DISTRIBUTION FOR THE CONVENTIONAL AND THE COMPOSITE
WEDGE ANCHORAGE
The load applied to a tensile element (rod or strip) has to be transferred first to the wedges and then to the anchor
body. The wedges and the anchor body are sliding along each other on an inclined plane. When a tensile force F is
applied to the rod or to the strip the wedges are pressed to the tensile element. Figure 1 shows the composite wedge
system, the wedges are made of two components, epoxy and steel. The epoxy component is in parallel to the CFRP
strip and its thickness decreases from the wedge tip to the rear part of the wedge. The epoxy component has a much
lower elastic modulus than the steel component. Therefore, the stiffness of the Composite Wedge is primarily
determined by the thickness and elastic modulus of the epoxy component. Due to the higher epoxy thickness at the
wedge tip than in the rear part, the transversal stiffness increases from the wedge tip to the rear part.

A traditional wedge is made of one material only. The transversal stiffness of a traditional wedge is therefore
increasing in opposite direction, from rear part to wedge tip. In figure 1 the transversal pressure and shear stress
distribution determined by finite element simulation at mean ultimate load (ft= 3100 N/mm²) of a 1.2 mm thick
CFRP strip are shown for a traditional wedge (dashed line) and the Composite Wedge System (solid line). The
traditional wedge design leads to a high value of transversal pressure and a high gradient at the wedge tip, therefore
the CFRP strip fails by exceeding the transversal compression strength, before the ultimate tensile load will be
reached.

Figure 1: Transversal stresses on CFRP element and shear stresses in the wedge-CFRP interface
determined by elastic finite element simulation. The dashed line is for traditional wedges and solid
line for the Composite Wedge System.

The Composite Wedge System distributes the transversal pressure uniformly, no high gradients occur and the
ultimate load of the CFRP strip can be exploited, see figure 1. The same holds for the shear stresses. With the
Composite Wedge System it is possible to adapt the pressure and shear stress distributions on the CFRP tendon as
required. The geometric parameters are the angle of the wedge (angle between axis of the tensile element and the
sliding surface) and the thickness ratio of the epoxy component in the rear region and wedge tip. With the variation
of the angle, the total lateral force acting on the CFRP tendon can be changed. The total lateral force corresponds to
the area below the pressure curve in figure 1. The smaller the wedge angle the higher is the total lateral force and the
area under the pressure curve. The stress distribution can be adjusted by changing the thickness ratio of the epoxy
component in the rear region and the wedge tip. The pressure stress can be constant from the wedge tip to the rear
region or increasing with a desired gradient. Due to this smooth transitions, low gradients and low maximum values
it is possible to activate the whole wedge length, which leads to high ultimate anchoring loads.

4. EXPERIMENTAL VERIFICATION
In this experimental series the components of the Composite Wedges were of steel and epoxy resin. The length of
the wedge was 150 mm, the width 60 mm and the angle to the axis of the strip 8°, see drawing in figure 1. The steel

484
component of the Composite Wedge was made with smooth sliding surfaces and a structured surface that will be in
connection with the epoxy layer. For the tests with the friction interface the wedges were produced with a surface
that was rough. The load was applied using a hydraulic jack with a hollow piston. Identical anchor bodies and
Composite Wedges were used on both tendon ends. The load was monotonically increased and measured with load
cells arranged between the hydraulic jack and an additional loading plate. The displacement was measured as the
stroke of the hydraulic jack. For the tests on the CFRP strip with thickness 1.2 mm (T12V1 to T12V4) and 2.5 mm
(T25V1, T25V2) the wedges were bonded to the strip before testing. The adhesive used was Sikadur 30, see Table 1
for details. Several tests with a friction surface (Type FT1), without adhesive, were performed successfully with
different test configurations and strips. Here only one test on 2.5 mm thick CFRP strips (T25FT1) is presented.

4.1 Experiments on 1.2 mm thick CFRP Strips

The tests on 1.2 mm thick strips showed the same failure load, which was equal to the mean tensile stress (ft=3100
N/mm²) times the area. In Table 1 Fcalc and the efficiency are determined with the guaranteed tensile strength of ft=
2800 N/mm² (Sika, 2003). The efficiency was between 110 to 114 %. The failure occurred due to failure of the
CFRP strip. Figure 2 shows an image of one test series after removal from the testing machine. The wedges were
still bonded together and no indication of slippage or failure due to high transversal pressure occurred. The
conclusion is that the wedge anchorage system in this configuration is able to anchor higher loads.

Figure 2. Two Composite Wedges after failure test.

Table 1: Test specimens and experimental results.

Ftest bxt Fcalc Efficiency Interface

(kN) (mm) (kN) (%)
T12V1 186.3 50x1.2 168 111 Adhesive
T12V2 186.5 50x1.2 168 111 Adhesive
T12V3 184.3 50x1.2 168 110 Adhesive
T12V4 191.5 50x1.2 168 114 Adhesive
T25V1 152.9 24.8x2.5 181 84 Adhesive
T25V2 149.2 23.7x2.5 171 87 Adhesive
T25F1 160.7 24.1x2.5 173 93 Friction

4.2 Experiments on 2.5 mm thick strips using bond and friction interfaces

To determine the limits of this anchorage tests on 2.5 mm thick CFRP strips were prepared. The strips were cut to a
width of approximately 25 mm, see Table 1. Although the width of the 2.5 mm thick CFRP strip is smaller than the
width of the 1.2 mm one, the transversal and shear stress in the interface are the same for different width of strips,
but increase strongly with higher strip thickness. The reason for that is that the whole anchorage including strip is a
plain stress problem. Therefore, for strips with the same thickness it is possible to reduce the width of the strip and
end up with the same shear stress and pressure normal to the strip axis.

485
For the 2.5 mm thick CFRP strip the manufacturer guaranties 2800 N/mm². The efficiency reached in the tests with
adhesive in the interface was 84 and 87 %. The observation of the specimens after testing showed that the CFRP
strips slipped in the anchor region, without causing any damage due to transversal stress.

The tests with the friction interface were performed on the same CFRP strips, but the wedges were prepared with a
rough surface and sandpaper was used in the interface. Again the failure was initiated by slipping in the wedge
CFRP strip interface. As before no damage due to excessive transversal pressure was observed. The results are given
in Table 1. By comparing the efficiencies of the tests on 2.5 mm thick strips it can be observed that the friction
interface showed the best performance with an efficiency of 93 %.

4. SPECIAL LAYOUTS OF THE COMPOSITE WEDGE SYSTEM FOR STRIPS

The standard layout of the Composite Wedge System is shown in figure 1. An advanced system that exploits
symmetry conditions leads to an anchorage for near surface mounted prestressing strips, see figure 3 (left). Here
only one wedge is necessary. The short distance from the structure surface reduces the requirements for bonding and
bolting. One important application of this special Composite Wedge layout could be for the strengthening of
bridges. In figure 3 (right) two strips are anchored in one anchor body. This reduces the size of the anchor body and
the number of wedges necessary. Another advantage is that two strips can be prestressed in one step.

Figure 3: Asymmetric Composite Wedge System for near surface mounted strips and anchorage
for two strips.

2. REFERENCES
Andrä H.-P., Maier M. (2005). “Instandsetzung von Brücken mit einer neuen Generation von Spanngliedern auf
Basis von CFK-Bändern”, Bauingenieur, Vol. 80, pp. 7-15.
Burtscher, S.L. (2004). “Keilverankerung für vorgespannte und/oder belastete Zugelemente (Wedge anchorage for
prestressed and/or loaded tensile elements“, Austrian Patent.
Harper, C.A. (2002). “Handbook of Plastic, Elastomers & Composites”, MacGraw-Hill, New York.
Horvatits J., Kollegger J. (2003). “Anchorage Advances”, Bride Design & Engineering, Vol. 33, No. 4, pp. 69-71.
Meier U. (1996). “Zwei CFK-Kabel für die Storchenbrücke“, Schweizer Ingenieur und Architekt, Vol. 44, pp. 980-
985.
Nanni A., Bakis C.E., O’Neil E.F., Dixon T.O. (1996). “Performance of FRP Tendon-Anchor Systems for
Prestressed Concrete Structures”, PCI Journal, Jannuary-February, pp. 34-44.
Pincheira J.A. (2001). “Anchorage of Carbon Fiber Reinforced Polymer (CFRP) Tendons using Cold-Swaged
Sleeves”, PCI Journal, November-December, pp. 100-111.
SIKA (2003). „Construction: Sicher bauen mit System”, Technische Merkblätter, Ausgabe 5.
Stresshead (2005). Technische Dokumentation.

486
Part XVII. Repair of Columns
Third International Conference on FRP Composites in Civil Engineering (CICE 2006)
December 13-15 2006, Miami, Florida, USA

EVALUATION OF THE CARRYING CAPACITY OF RC COLUMNS STRENGTHENED WITH

COMPOSITE MATERIALS

E. Ferrier
(Assistant professor,University Lyon 1, Lyon, France)

M. Quiertant
(Researcher, Structural Engineering Unit - French Public Works Research Laboratory - LCPC, Paris, France)

P. Hamelin
(Professor,University Lyon 1, Lyon, France)

ABSTRACT
This paper presents a simple evaluation procedure of the ultimate capacity of RC columns strengthened with
composite materials. This method, which is based on the French design guidelines (BAEL) permits to take into
account the influence of the CFRP confinement on the mechanical behavior of columns. The procedure is a quite
general frame that allows the evaluation of different combinations of reinforcement with continuous or
discontinuous confinement associated or not with flexural reinforcement using FRP laminates or sheets. After a
detailed presentation, results obtained by the proposed relations are compared to experimental data.

KEYWORDS
Reinforcement, CFRP, RC columns, Confinnement, Design.

1. INTRODUCTION
The confinement of RC columns is an application where the external wrapping by fiber-reinforced polymer (FRP)
sheets is particularly effective and offer numerous advantages as related to speed and easiness of application, high
strength-to-weight ratio, durability and corrosion resistance. This technique, which is applied either to the repair of
damaged columns or to the rehabilitation of healthy structures, is increasingly used in USA, Canada and Japan. It
permits to extend the life expectancy of ageing structures as well as to answer seismic requirements. The purpose of
this work is to propose a design method with simple calculation procedure of the ultimate capacity of RC columns
externally strengthened with bonded composite materials.

2. DESIGN METHOD
1.1 Evaluation of the carrying capacity of a confined column

In the proposed study, four combinations of external reinforcement of columns are considered: continuous or
discontinuous rings for the confinement, associated or not with flexural reinforcement using FRP laminates or
sheets. External reinforcement has to be carried out according to the recommendations of the French Civil
Engineering Association (AFGC, 2003). The proposed calculation take into account the confinement effect and the
axial reinforcement on the carrying capacity of columns, but the procedure is restricted to the following conditions
related to:
¾ Slenderness ratio (λ): Only short columns, with slenderness ratio limited to fifty are considered ( λ ≤ 50 ). Sections
can be circular rectangular or square.

487
¾ Loading condition: The applied compressive load must be applied with an ecentricity lower or equal to emax,
where emax is the upper value comparing 2 cm (this absolute value is referred to acceptable construction deviation) to
10% of the smallest side of the cross section (or 10% of the diameter in case of circular section). Columns with the
smallest side of the cross section smaller than 200 mm are not considered here (at date this point is still debated in
the task group).
¾ Conditions of continuity: Columns have to respect the constructive dispositions proposed by AFGC (AFGC,
2003), in particular the conditions on the covering lengths and the curvature radius of the corners. For square or
rectangular section, it is necessary to round off angles, to prevent a premature local failure of the composite. Indeed,
the tensile strength of the composite significantly decreases for low corner radius (Yang et al., 2001).

1.2 Calculation procedure for confined columns

The compressive strength ( f cc' ) of the confined concrete is evaluated using the following general form:
f cc' = f c' + k1 (ψ f k c k h f pu + ke f pa ) (1)

The performance coefficient ψf depends on the strengthening technique, the kind of reinforcement and the curvature
radius of the composite. We consider in a first approach:
ψ f = 0.8 Circular section ψ f = 0.6 Square section (with rc ≥ 35 mm)

This coefficient can be corrected from tests on composites adopting the specified curvature radius (Yang et al.,
2001). The confinement pressure fpu resulting from the composite is expressed as follows:
f pu = E p ⋅ ε fu (2)

With the ultimate strain of the composite εfu (measured from a tensile test on a composite) and the confinement
modulus E p , which traduces the confinement stiffness and given as follows:
t f ⋅ np 2t f ⋅ n p
Ep = ⋅ Ef For circular section (3) Ep = ⋅ Ef For rectangular section (3’)
r b

With Ef the tensile modulus of the composite, tf the thickness of one layer and np the number of layers. b and r are
respectively the biggest side (in the case of square or rectangular section) and the radius of the concrete core for a
circular section. Many studies show that FRP-confinement can significantly enhance the ultimate strength and strain
of the concrete. Ultimate compressive strength directly depends on confinement ratio (fig. 1).

Depending on the concrete strength, the coefficient k1 is given by (Berthet et al., 2005a, b):
k1 = 3.45 With f c' ≤ 50 MPa (4)

In circular sections, the confinement pressure is uniform in the concrete core whereas in rectangular columns, the
distribution of the confinement pressure is not homogeneous (fig. 1). In these cases, only a part of the concrete core
is effectively confined, that considerably reduces the confinement efficiency. Square sections generate parabolic
distribution for confinement pressure with an initial angle of 45° with the jacket, depending only on the geometry of
the section (Mander et al., 1988).

Figure 1: Confinement pressure on a cross section Figure 2: Confinement pressure between two rings

488
For the calculation, we shall consider an effective confinement pressure. So, the nominal confinement pressure is
corrected using a coefficient kc, which takes into account the geometry of the columns (CSTB-Freyssinet, 2001):
a ' 2 +b ' 2
kc = 1 − for square section (5) k c =1 for circular section (5’)
3 Ab

For a discontinuous confinement, the distribution of the confinement pressure is not homogeneous on the height of
the column (fig. 2). fpu is then combined with a coefficient kh which integrates the width of the composites rings bf,
their spacing Sf and their orientation θf, kh is given by the produce of two coefficients η1 and η2, which traduce
respectively the influences of the orientation and the spacing of the composite rings and defined using:
kh = η1η 2 (6)

Rectangular section: Circular section:

θf
η1 = 1 (7’) η1 =
1
(7)
θf
Sf 1+
bf ⎛ ⎞
2
πD
⎜1 − S f − b f ⎟

η2 =
⎜
⎝
2D ⎟
⎠
(8) kh = (1 −
((a − 2 ⋅ r )
c
2
)(
+ (b − 2 ⋅ rc ) 2 S f − b f )) ⋅ (1 − (a + b) (S f − bf ))
1 − ρl 3 ⋅ a ⋅ b ⋅ (1 − ρl ) 3 ⋅ a ⋅ b ⋅ (1 − ρl )
b ou D

With ρl longitudinal steel ratio (As / B), θf winding angle of the reinforcement (between fibers axis and transverse
direction of the column), a and b , the straight lengths of the sides (a’= a - 2rc and b’= b - 2rc).

The confinement pressure fpa generated by the steel stirrups is estimated (Mander et al., 1988) using:
Ast Ast
f pa = fy for circular section (9) f pa = fy for square section (9’)
st d s t bt

The effective confinement pressure induced by the stirrups is estimated by associating to the nominal confinement
pressure a coefficient ke which takes into account the geometry of the section and stirrups spacing (Mander et al.,
1988):
2
⎛ ⎞ ⎛ 2 ⎞⎛ ⎞⎛ ⎞
⎜1 − s t ⎟ ⎜1 − at + bt ⎟⎜1 − st
2
⎜ 2D ⎟ ⎟⎜1 − st ⎟
⎝ ⎠ ⎜ 3 Abt ⎟⎠⎜⎝ 2at ⎟⎜
⎠⎝ 2bt ⎟⎠
ke = ⎝
t
ke = (Circular section) (10) (Square section) (10’)
1− ρl 1 − ρl

1.3 New carrying capacity for strengthened columns

Developed method is an extension of the French design guidelines (BAEL) and takes into account the influence of
the confinement effect on the mechanical behavior of the concrete. In these relations, confined concrete strength f cc'
is substituted to plain concrete strength f c' . A significant reduction of the confinement efficiency in slender column
was demonstrated from experimental studies (Mirmiran et al., 2001; Thériault, et al., 2001) and this effect is
integrated in the parameter α .

The maximum axial load for a confined column is given by:

[
N u ≤ α Ab f cc' + As f a ] (10)

With α , a coefficient depending on the slenderness ratio and given under the following form:
0.85
α= for λ ≤50 with λ ' slenderness ratio (11)
2
⎛ λ' ⎞
1 + 0.2 ⎜ ⎟
⎝ 35 ⎠

489
3. COMPARISON WITH EXPERIMENTAL RESULTS
Comparison of predicted results and those experimentally obtained by others authors (Quiertant et al., 2004) is
reported in table 1. The experimental program was based on tests on twenty columns with a 200 x 200 mm2 square
cross section and an overall height of 2,500 mm. Two series of specimens were cast, with two different concrete (40
and 55 MPa strength) and two different internal reinforcements (low and medium). Finally columns were externally
strengthened with different combinations of longitudinal and transverse CFRP layers. From table 1, it can be seen,
that the model slightly underestimates the ultimate capacity of the reinforced columns. Indeed, the average
differences between experimental values and those calculated from the model are inferior to 10 %.

Table 1: Comparison between calculated and experimental data

Columns with low internal Columns with medium
CFRP for Flexural CFRP wrapping
reinforcement internal reinforcement
reinforcement tf /Ef confinement tf /Ef
Ultimate load Predicted Ultimate load Predicted
None None 1215* kN 1088 kN 1254* kN 1295 kN
6 CFRP plates (2cm 1 layer of CFRP
1578* kN 1407 kN 1711* kN 1594 kN
width) 1.2 mm/180 GPa 0.43 mm /105 GPa
1 layer of CFRP 0.117 1 layer of CFRP
1194* kN 1326 kN 1653* kN 1540 kN
mm /240GPa 0.117 mm /240 GPa
2 CFRP plates (2cm 1 layer of CFRP
1422* kN 1334 kN 1689* kN 1545 kN
width) 1.2 mm /165 GPa 0.13 mm /230 GPa
1 layer of CFRP 1 layer of CFRP (angle of
1462* kN 1493 kN 1628* kN 1654 kN
1 mm /62-70 GPa 20°) 1 mm /62-70 GPa
(*) Mean value of two tests

4. CONCLUSION
A design method, based on an extension of the French design guidelines (BAEL) and modified from experimental
study, is presented. This method take into account different parameters such as confinement level, steel bars and
stirrups ratio, longitudinal reinforcement ratio and concrete core strength,. Based on comparison with experimental
results, it can be concluded that the developed approach provides a good estimate of the carrying capacity of RC
columns strengthened with composite materials.

5. REFERENCES
AFGC. (2003) Documents scientifiques et techniques, fascicule « Réparation et renforcement des structures en béton
au moyen des matériaux composites ».
Berthet J.F., Ferrier E., Hamelin P. (2005a). “Compressive behavior of concrete externally confined by composite
jackets. Part A: experimental study” Construction & building materials, Elsevier ed. Vol. 19, Issue 3, pp. 223-232.
Berthet J.F., Ferrier E., Hamelin P. (2005b). “Compressive behavior of concrete externally confined by composite
jackets. Part B: modeling” Construction & building materials, Elsevier ed. On line.
CSTB-Freyssinet (2001). « Éléments de structure renforcés par un procédé de collage de fibres de carbone », Avis
technique 3/04-424.
Mander J.B., Priestley M.J.N. and Park R. (1988). “Theorical stress-strain model for confined concrete”. Journal of
structural engineering, Vol 114, n° 8, pp. 1804-1849.
Mirmiran A., Shahawy M. and Beitleman T.(2001) “Slenderness limit for hybrid FRP-concrete columns”. Journal of
composites for construction, vol. 5 N°1, pp. 26-34.
Quiertant M., Toutlemonde F., Clément J.-L. (2004) “Combined flexure-compression loading for RC columns
externally strengthened with longitudinal and transverse CFRP retrofitting” Proceeding of the fib Symposium 2004 –
April 26-28 – Avignon, France.
Thériault, M., Claude, S. & Neale, K.W. (2001). “Effect of size and slenderness ratio on the behaviour of FRP-
wrapped columns”. Fibre-reinforced plastics for reinforced concrete structures, ed by C.J. Burgoyne, London, pp.
765-771.

490
Third International Conference on FRP Composites in Civil Engineering (CICE 2006)
December 13-15 2006, Miami, Florida, USA

LARGE-SIZE REINFORCED CONCRETE COLUMNS STRENGTHENED

WITH CARBON FRP: EXPERIMENTAL EVALUATION
Silvia Rocca
(Ph D. Candidate in Civil Engineering, University of Missouri - Rolla, Rolla, Missouri, USA)

Nestore Galati
(Research Engineer, University of Missouri - Rolla, Rolla, Missouri, USA)

Antonio Nanni
(Chair & Professor of Civil Engineering, University of Miami – Coral Gables, Coral Gables, Florida, USA)

ABSTRACT
The use of Fiber Reinforced Polymer wrapping technique has been extensively studied; in particular, the behavior of
confined elements of circular cross-sections subjected to pure axial loads has been studied. However, the available
models are based on small-scale specimens. Limited studies are found for the cases of prismatic members,
especially on large-size ones.
To analyze the behavior of axially loaded large-size Reinforced Concrete (RC) columns confined by means of
Carbon FRP (CFRP) wrapping, a test matrix was designed to investigate the effect of different variables, such as the
geometry of the specimen (circular, square, and rectangular), the area aspect ratio, the side aspect ratio, and a height-
to-width aspect ratio. A total of 22 specimens were divided into six series of three specimens each and two series of
two specimens. The largest column tested had a cross-sectional area of 0.8 m2 (9 ft2) and the smallest one an area of
0.1 m2 (1 ft2). The experimental results are compared and contrasted with current available data on reinforced
concrete specimens with one minimum dimension of the cross-section of 300 mm (12 in). This evaluation allowed
concluding and confirming that among circular and square specimens of the same cross-sectional area, the
confinement effect of the FRP is less effective for the latter. It was observed that within specimens of circular and
prismatic cross-sections with size-aspect-ratio less or equal than 2.0, the size effect could be negligible.

KEYWORDS
Confinement, Ductility, FRP-Strengthening, Prismatic Columns, Reinforced Concrete.

1. INTRODUCTION
Confinement of Reinforced Concrete (RC) columns by means of Fiber Reinforced Polymer (FRP) jackets is a
technique being used with growing frequency to seek the increment of load carrying capacity and/or ductility of
such compression members.
The confinement of prismatic columns is generally acknowledged to be less efficient than the confinement of
circular columns, since in the latter case, the wrapping provides circumferentially uniform confining pressure to the
radial expansion of the compression member. In prismatic columns, the confinement is concentrated at the corners
rather than over the entire perimeter.
Extensive work in both the experimental and analytical areas has been conducted on RC columns of circular cross-
sections confined with FRP and subjected to pure axial compressive loading. Studies focused on RC columns of
non-circular cross-sections have also been conducted; however, such work is limited because the experimental
research has primarily been on small specimens of plain concrete due to high cost and lack of high-capacity testing
equipment. This situation has been the main reason for overlooking the following important effects on the element
performance that are not accounted for in most of the available models: (a) the size of the cross-sectional area, (b)
the dimensional aspect ratio of the cross-sectional area, (c) the presence and possible detrimental effect of
longitudinal steel reinforcement instability, (d) the concrete dilation dependant on a pseudo-Poisson ratio, and (e) the

491
contribution of the internal transverse steel reinforcement. In spite of these obstacles, several models have been
proposed (Wang and Restrepo 2001, Lam and Teng 2003, Maalej et al. 2003) and have become the basis for design
provisions.

2. EXPERIMENTAL PROGRAM
2.1. Test Matrix

The test matrix (Table 1) was designed to investigate the influence of different variables: side aspect ratio (h/b), area
aspect ratio (based on an area of 457 x 457 mm [18 x 18 in]), and height-to-side aspect ratio (H/h). The experimental
program was divided into two sub-matrices based on the laboratories where the experiments were carried out:
CALTRANS Seismic Response Modification Device Testing Laboratory (SRMD) at the University of California
San Diego (UCSD) with 18 specimens (six series of three specimens each: A, B, C, D, E, and F), and the Building
and Fire Research Laboratory at the National Institute of Standards and Technology (NIST) with four specimens
(two series of two specimens each: G and H). Table 1 presents the characteristics of the test units in the following
order: specimens acronym, diameter D of the circular columns (Series A), side dimensions of the prismatic columns
(b, h), side-aspect-ratio h/b, overall specimens height H, height-to-side-aspect ratio (H/h), gross section area Ag,
area-aspect-ratio Ag/Ag(C) (Ag(C) is the gross area of specimens series C), longitudinal steel reinforcement ratio ρl,
yield strength of longitudinal steel reinforcement fy, FRP volumetric ratio ρf, and average characteristic concrete
compressive strength at the corresponding age of testing of each column f’c.

Table 1: Test Matrix

Ag
Test D b h h H H Ag ρl fy ρf f'c
Unit (mm) (mm) (mm) b (m) h (cm2) A g(C) (%) (MPa) (%) (MPa)
A1 0.00 31.7
A2 508 NA NA 1.1 NA 2027 NA 1.53 446 0.26 31.9
A3 0.33 31.9
B1 0.00 30.2
B2 NA 313 635 2.0 1.4 2.2 1984 1 1.56 447 1.12 30.4
B3 0.32 30.4
C1 0.00 32.1
C2 NA 457 457 1.0 1.0 2.2 2090 1 1.48 446 0.58 32.3
C3 0.29 32.1
D1 0.00 30.7
D2 NA 648 648 1.0 1.4 2.1 4195 2 1.48 446 0.52 30.9
D3 0.21 30.7
E1 0.00 32.3
E2 NA 324 324 1.0 0.7 2.1 1049 1 1.53 447 0.41 33.0
E3 0.53 33.2
F1 0.00 31.5
F2 NA 324 324 1.0 1.4 4.2 1049 1 1.53 447 0.41 31.5
F3 0.53 31.7
G1 0.00 31.6
NA 914 914 1.0 2.0 2.2 8361 4 1.50 690
G2 0.58 31.6
H1 0.00 30.3
NA 635 1270 2.0 2.7 2.2 8065 4 1.52 690
H2 1.50 30.3

Series A, B, C, D, E, and F consist of three specimens each: one control unit (A1, B1, C1, D1, E1, and F1), one unit
strengthened to achieve an increment of 30 percent of load carrying capacity featuring a full wrapping scheme (A2,
B2, C2, D2, E2, and F2), and a third unit whose thickness of FRP jacket matched the same number of plies used in
the specimen A2 (B3, C3, and D3). Specimens A3, E3, and F3 were partially strengthened to attain 30 percent

492
increment of carrying capacity as well. Series G and H were composed of two test units each: one control (G1 and
H1), and one strengthened to gain the same level of increase in axial capacity (G2 and H2). Regarding the wrapping
scheme of all the strengthened specimens, a gap of about 7-13 mm (0.25-0.5 in) was left at the top and bottom ends
between the edge itself and the fabric to avoid axial compressive loading of the FRP jacket. The partially wrapped
specimens featured strips of 133 mm (5.25 in) wide and a pitch of 76 mm (8.25 in).
All the specimens featured a clear concrete cover of 38 mm (1.5 in), and the prismatic specimens were designed
with a corner radius of 30 mm (1.2 in). Further details on the construction and strengthening of the specimens can
be found in Rocca et al. (2006).

2.2. Material Properties

All the specimens were designed with a nominal concrete compressive strength of 28 MPa (4,000 psi) and steel
Grade 60 (420 MPa). The concrete and steel characteristics strengths (fy and f’c) shown in Table 1 were determined
according to ASTM C39-04 and ASTM A370, respectively.
Unidirectional Carbon FRP (CFRP) of ply nominal thickness (tf) of 0.167 mm (0.0066 in) was the wrapping material
used for the entire research project. Tensile coupon tests were performed to determine the mechanical properties of
the CFRP material used in the evaluation of the test results. For the preliminary design, the mechanical properties
provided by the manufacturer were used. This characterization was conducted according to ASTM D3039-00 and
yielded an ultimate tensile strain εfu of 0.93 percent, an ultimate tensile strength ffu of 2668 MPa (387 ksi), and a
modulus of elasticity Ef of 291 GPa (42,200 ksi).

2.3. Instrumentation and Test Setup

The instrumentation consisted of internal sensors (strain gages) located on both longitudinal and transverse steel
reinforcement and external sensors to measure axial deformation (linear potentiometers in UCSD specimens and
LVDTs in NIST specimens), fixed to the faces of the columns at about mid-height. Additionally, strain gages were
installed on the FRP jacket at critical locations (corner areas and middle distance on each side) along the perimeter
of the cross-section on the central region of the strengthened specimens.
The equipment at both UCSD and NIST laboratories is capable of applying an axial compressive force of 53 MN.
However, because of the height limitation of the former (1.5 m [5 ft]), the larger specimens (groups G and H) were
tested at NIST.

3. TEST RESULTS
The results from this experimental program are presented along with collected relevant available data on RC
columns of circular and prismatic cross-sections (h/b ≤ 2) in terms of trends of the strengthening ratio f’cc/f’co versus
the parameter ρfEf /Ec (See Figure 1). f’cc and f’co represent the peak concrete strengths corresponding to the
maximum load carried by the RC column for confined and unconfined cases, respectively. In ρfEf /Ec, the FRP
volumetric ratio ρf considers the thickness of the FRP jacket and the geometry of the cross-section, and the relative
stiffness of the confining FRP to the axial stiffness of the concrete is represented by the ratio Ef /Ec. Figure 1(a), (b),
(c) refer to cases of specimens of circular, square and rectangular cross-sections, respectively. Figure 1(d) presents
the linear trends of the types of cross-sections and their reliability indexes obtained by regression analysis
corresponding to each dataset. The specimens’ acronyms correspond to the following references: CH to Carey and
Harries (2003), DN to Demers and Neale (1999), KE to Kestner et al. (1997), MA to Matthys et al. (2005), RO to
specimens of the presented study (Rocca et al., 2006), and WR to Wang and Restrepo (2001). Each acronym is
followed by a number(s), which in Figure 1(a) indicates the diameter of the cross-section, and in Figure 1(b) and (c)
indicate the dimensions of the cross-sections. With respect to Figure 1(a), circular cross-section dataset, note the
uniformity of the trend and minor scattering. No pattern reflecting the effect of cross-sectional area size is identified
leading to the establishment of the lack of such effect on this type of cross-section. Regarding Figure 1(b), square
cross-section dataset, the scatter of data is more pronounced. Concerning the specimens of rectangular cross-
sections (Figure 1(c)), no definite observation can be concluded because of the high level of scatter of the data and
the limited number of data points. The linear trends of the three datasets presented in Figure 1(d) reflect the level of
effectiveness of the FRP confinement in the axial strengthening. The slopes of the trends corresponding to the
prismatic specimens are similar, and their strengthening performance is less effective than in the case of specimens
of circular cross-sections, which confirms the generally accepted notion of confinement of different cross-section
shapes.

493
 1.8 1.8
DN(305) WR(300)
1.7 MA(400) 1.7 KE(457)

Strengthening Ratio f'cc / f'co

Strengthening Ratio f'cc / f'co

KE(508) CH(540)
1.6 1.6
RO(508) RO(457)
1.5 CH(610) 1.5 RO(648)
RO(324)
1.4 1.4 RO(324)
1.3 1.3 RO(914)

1.2 1.2

1.1 1.1

1.0 1.0
0 2 4 6 8 10 12 14 16 18 0 2 4 6 8 10 12 14 16 18
Note: 2.2 ≤ H/h ≤ 5.0 ρf*Ef / Ec (%) Note: 2.1 ≤ H/h ≤ 4.0 ρf*Ef / Ec (%)

(a) Circular Cross-Section (b) Square Cross-Section

1.8
WR(300x450) 1.8 Circular Circular
1.7 RO(318x635) y = 0.15x + 1.00 Square
Strengthening Ratio f'cc / f'co

2
RO(635x1270) 1.7 R = 0.91
Rectangular

Strengthening Ratio f'cc / f'co

1.6
1.6 Linear (Circular)
1.5 Square
1.5 Linear (Square)
y = 0.03x + 1.00
1.4 2 Linear (Rectangular)
1.4 R = 0.59
1.3 Rectangular
1.3
y = 0.01x + 1.00
1.2
1.2 2
R = 0.72
1.1 1.1
1.0 1.0
0 2 4 6 8 10 12 14 16 18 0 2 4 6 8 10 12 14 16 18
Note: 2.0 ≤ H/h ≤ 2.2 ρf*Ef / Ec (%) ρf*Ef / Ec (%) Note: 2.2 ≤ H/h ≤ 5.0
(c) Rectangular Cross-Section (d) Varied Cross-Section

Figure 1: Strengthening Performance of FRP-Confined RC Columns

4. CONCLUSSIONS
Based on experimental observations on RC specimens of varied circular cross-section areas it can be concluded the
lack of size effect. However, for the case of RC prismatic specimens, the high scattering and limitation of data-
points does not allow at the present time to draw a definite concluding remark with respect to the size effect.

5. REFERENCES
Carey, S. and Harries, K. (2003). “The Effects of Shape, ‘Gap’, and Scale on the Behavior and Modeling of
Variably Confined Concrete,” Report No. ST03-05, University of South Carolina, Columbia, SC.
Demers, M. and Neale, K. (1999). “Confinement of Reinforced Concrete Columns with Fibre Reinforced
Composites Sheets - An Experimental Study,” Canadian Journal of Civil Engineering, No. 26, pp. 226-241
Kestner, J. T., Harries, K. A., Pessiki, S. P., Sause, R., and Ricles, J. M. (1997). “Rehabilitation of Reinforced
Concrete Columns using Fiber Reinforced Polymer Composite Jackets,” ATLSS Report No. 97-07, Lehigh
University, Bethlehem, PA.
Lam, L., and Teng, J. (2003). “Design-oriented Stress-Strain Model for FRP-confined Concrete in Rectangular
Columns,” Journal of Reinforced Plastics and Composites, Vol. 22, No. 13, pp. 1149-1186.
Maalej, M., Tanwongsval, S., and Paramasivam, P. (2003). “Modeling of Rectangular RC Columns Strengthened
with FRP,” Cement & Concrete Composites, Vol. 25, pp. 263-276.
Matthys, S., Toutanji, H., Audenaert, K., and Taerwe, L. (2005). “Axial Load Behavior of Large-Scale Columns
Confined with Fiber-Reinforced Polymer Composites,” ACI Structural Journal, Vol. 102, No. 2, pp. 258-267.
Rocca, S., Galati, N., and Nanni, A. (2006). “Experimental Evaluation of FRP Strengthening of Large-Size
Reinforced Concrete Columns,” CIES Report No. 06-63, University of Missouri – Rolla, Rolla, MO.
Wang, Y. C., and Restrepo, J. I. (2001). “Investigation of Concentrically Loaded Reinforced Concrete Columns
Confined with Glass Fiber-Reinforced Polymer Jackets,” ACI Structural Journal, V. 98, No. 3, pp. 377-385.

494
Third International Conference on FRP Composites in Civil Engineering (CICE 2006)
December 13-15 2006, Miami, Florida, USA

PARAMETRIC ANALYSIS OF SHAPE EFFECTS IN CONCRETE

COLUMNS CONFINED WITH FRP
Ricardo Carrazedo
(PhD in Structural Engineering, São Carlos School of Engineering / University of São Paulo, São Paulo, SP, Brazil)

João Bento de Hanai

(Professor of Civil Engineering, São Carlos School of Engineering /University of São Paulo, São Carlos, SP, Brazil)

ABSTRACT
This work presents the results of a parametric study on shape effects in concrete columns confined with FRP. The
study was developed through numerical simulation with the Finite Element Method (FEM) and experimental
analysis. Parametric input files were used to model square columns with different corner radius wrapped by Carbon
Fiber Reinforced Polymers. A 3D non-associated plasticity model was used to model concrete behavior.
Comparison of numerical x experimental results showed a good correspondence. The results showed that the
variable 2r/D (where r is the corner radius and D is the column cross section size) can be used as an effectiveness
coefficient to reduce the nominal lateral pressure and estimate strength gains.

KEYWORDS
Concrete, confinement, FRP, plasticity.

1. INTRODUCTION

It is well known that the cross section geometry plays a main role in defining the stress x strain behavior of concrete
columns confined with FRP. Rochette (1996) showed that the ratio 2r/D can be used to evaluate the confinement
effectiveness.

Alternatively, FIB (2001) suggested the use of an effectiveness parameter based on the shape of the effectively
confined core. The effectively confined core was defined by subtracting the unconfined areas near to the flat sides of
the column. These areas were defined by parabolas with initial and final slope of 45 degrees. For a square column
with an external dimension (D) and a corner radius (r) this effectiveness coefficient (Acc/Ac) is given by equation 1:

Acc 2
⋅ (D − 2r )
2
= 1− Equation 1
Ac 3 Ac

where Ac is the gross cross-sectional area and Acc is the effectively confined area.

In order to evaluate the performance of the existing equations to predict the strength increase of concrete columns
wrapped by CFRP with different cross section geometries, a parametric analysis was developed using the Finite
Element Method. This approach was based in the fact that the constitutive model used for concrete showed a very
good match to experimental behavior in uniaxial compression and confinement tests. Consequently, in addition to
some test results of concrete columns confined by CFRP wraps, in this study several columns were numerically
modeled, expanding the range of results for analysis.

2. EXPERIMENTAL ANALYSIS
Nine concrete columns were tested under uniaxial compression loading. Three different geometries were used: a
circular cross section column with 150 mm of diameter, a square cross section (150 mm x 150 mm) column with

495
rounded corners of 30 mm of radius and a square cross section column with rounded corners of 10 mm of radius. All
columns had a length of 450 mm. Three columns of each geometry were tested wrapped by 0, 1 and 2 CFRP layers.

The concrete compressive strength determined by testing of 100 mm x 200 mm cylindrical specimens was of 42.5
MPa and the elastic modulus was of 28.5 GPa. The mechanical properties of the composite (CFRP) in the direction
of the fibers were tensile strength of 872 MPa and modulus of elasticity of 71 GPa, respectively. These properties
were evaluated with a thickness per layer of the composite equal to 0.5 mm.

3. NUMERICAL ANALYSIS
The numerical models were developed using the software ABAQUS release 6.2 and parametric input files which
allowed creating a parametric mesh, using some variables whose values were defined in the parametric study. These
variables were the cross section sides length hx and hy (which assumed constant values of 150 mm in this study) and
the corner radius r, as shown in Figure 1.

(hx / 2) - r r

r r = D = 75 mm r = 50 mm r = 30 mm
hy / 2

(hy / 2) - r

r = 10 mm r = 5 mm
Figure 1: Parametric shape variation

Concrete behavior was modeled using an Umat subroutine based on Grassl et al. (2002) model. Some new equations
were added to this 3D non-associated plasticity constitutive model, and the complete code of the subroutine was
presented in Carrazedo (2005). CFRP was modeled using orthotropic material properties for which the maximum
strain criterion was used, with a value of 0.0106 for the ultimate strain in fibers direction.

4. RESULTS AND DISCUSSION

As shown in Figure 1, the numerical and experimental stress x strain curves generated were similar. As expected, the
confinement effects were higher as the cross section shape was closer to circular. For r = 10 mm the confined
specimens showed a descending stress x strain diagram after a small strength increase was obtained. The columns
with r = 30 mm showed a higher strength increase with an almost horizontal second branch. The circular columns
showed a second branch with a positive slope, resulting in high strength and ultimate strain increases.

Both experimental and numerical failures were caused by FRP rupture. However, the numerical response in the
stress x strain curve continued further than the experimental one. This occurred mainly due to three factors: the
ultimate strains measured during the experiments, through strain gages attached to the perimeter of the square
columns, were in some cases smaller than the adopted in the simulations (0.0106); the experiments showed an
important concentration of stresses in the limit between the flat sides and the rounded corners, not captured properly
by the numerical model; the dilation of concrete in the numerical model during the descending branch was lower
than the experimental. However, as can be noticed in Fig. 2, the shapes of the numerical and experimental curves
were very similar, indicating that a proper representation of the problem was obtained in the numerical modeling.

In addition to the columns whose results are shown in Figure 2, the numerical parametric study included columns
wrapped by four CFRP layers and with different round off radius. The description of columns characteristics and
main results are shown in Table 1. Complete axial stress (σ3) x strain (ε3) curves are shown in Figure 3.

496
 - 5 0 ,0 -6 0

-5 0
- 4 0 ,0 n = 2

Stress (MPa)
Stress (MPa)
n = 2
-4 0
- 3 0 ,0 n = 1
-3 0 n = 1
- 2 0 ,0 n = 0 n = 0
N u m e r ic a l -2 0 N u m e r ic a l
E x p e r im e n t a l E x p e r im e n t a l
- 1 0 ,0 -1 0
S q u a r e s e c t io n ; r = 1 c m S q u a r e s e c t io n ; r = 3 c m
0 ,0 0
0 ,0 0 0 -0 ,0 0 5 -0 ,0 1 0 - 0 ,0 1 5 - 0 ,0 2 0 - 0 ,0 2 5 0 ,0 0 0 -0 ,0 0 5 -0 ,0 1 0 - 0 ,0 1 5 -0 ,0 2 0
S t r a in S tr a in

-9 0 n = 2
-8 0
-7 0

Stress (MPa)
-6 0
n = 1
-5 0
-4 0
-3 0 N u m e r ic a l
n = 0
-2 0 E x p e r im e n ta l
-1 0 C ir c u la r c r o s s s e c t io n
0
0 ,0 0 0 - 0 ,0 0 5 - 0 ,0 1 0 - 0 ,0 1 5 - 0 ,0 2 0
S tr a in

Figure 2: Comparison of experimental x numerical stress x strain diagrams

3 ,5 n = 1 3 ,5 n = 2
3 ,0 f l /f c o = 0 .1 3 8 3 ,0
f l/ f c o = 0 . 2 7 7
2 ,5 k 1 = 4 .7 6 2 ,5 k 1 = 5 .2 4

2 ,0 2 r /D 2 ,0 2 r /D
co

co

1 ,0 0 0 1 ,0 0 0
σ /f

σ /f

1 ,5 1 ,5 0 ,6 6 7
0 ,6 6 7
3

1 ,0 0 ,4 0 0 0 ,4 0 0
1 ,0
0 ,1 3 3 0 ,1 3 3
0 ,5 0 ,0 6 7 0 ,5 0 ,0 6 7
u n ia x ia l u n ia x ia l
0 ,0 0 ,0
0 5 10 15 0 5 10 15
ε3 / εco ε 3 / ε co

3 ,5 n = 4
3 ,0 f l /f c o = 0 . 5 5 4
k 1 = 4 .4 7
2 ,5

2 ,0 2 r/D
co

1 ,0 0 0
σ /f

1 ,5 0 ,6 6 7
3

0 ,4 0 0
1 ,0
0 ,1 3 3
0 ,5 0 ,0 6 7
u n ia x ia l
0 ,0
0 5 10 15
ε 3 / ε co

Figure 3: Stress x strain diagrams generated in the numerical parametric study

The shape effectiveness coefficient (ke) from the tests and numerical simulations was isolated using Equation 2. As
for circular columns ke=1 by definition, the coefficient k1 was determined from the results of these columns. As k1 is
a concrete property also dependent on the variable fl/fco, one value of k1 was evaluated for each number of applied
CFRP layers. The obtained results are shown in Table 1 and graphically in Figure 4. In Figure 4 it is possible to
notice that the variable 2r/D has a closer match to ke than Acc/Ac. Consequently 2r/D or a function of this variable
can be used to evaluate the effective lateral pressure fle=ke.fl in non-circular concrete columns wrapped by CFRP.
Finally the increase on the ultimate stress of confined concrete can be estimated as k1.fle.

497
 f cc f
= 1 + k1 ⋅ k e ⋅ l Equation 3
f co f co

Table 1: Columns characteristics and main results

Numerical Numerical Experimental

fcc / fcc / fcc /
Column n 2r/D Acc/Ac ke Column n 2r/D Acc/Ac ke Column n 2r/D Acc/Ac ke
fco fco fco
P1 1 0,667 0,918 0,539 1,355 P9 4 0,667 0,918 0,719 2,780 S10r3 1 0,400 0,751 0,321 1,150
P2 1 0,400 0,751 0,225 1,148 P10 4 0,400 0,751 0,528 2,308 S10r1 1 0,133 0,497 0,132 1,062
P3 1 0,133 0,497 0,135 1,089 P11 4 0,133 0,497 0,270 1,668 S20r3 2 0,400 0,751 0,333 1,430
P4 1 0,067 0,419 0,131 1,086 P12 4 0,067 0,419 0,153 1,380 S20r1 2 0,133 0,497 0,057 1,074
P5 2 0,667 0,918 0,661 1,922 P13 1 1,000 1,000 0,999 1,658 C10 1 1,000 1,000 1,000 1,468
P6 2 0,400 0,751 0,443 1,618 P14 2 1,000 1,000 1,000 2,395 C20 2 1,000 1,000 1,000 2,290
P7 2 0,133 0,497 0,256 1,357 P15 4 1,000 1,000 1,000 3,477
P8 2 0,067 0,419 0,115 1,161

1,0 1,0

0,8 0,8

0,6 0,6

Acc/Ac
numerical numerical
2r/D

experimental experimental
0,4 0,4

0,2 0,2

0,0 0,0
0,0 0,5 1,0 0,0 0,2 0,4 0,6 0,8 1,0
ke ke

Figure 4: Comparison of ke x 2r/D and Acc/Ac

5. CONCLUSIONS
This paper showed the results of numerical and experimental investigations on shape effects in confinement of
concrete with CFRP. A parametric study was developed using a 3D non-associated plasticity model through an
Umat subroutine, obtaining a good match to experimental results. The numerical and experimental results were used
to evaluate the influence of 2r/D and Acc/Ac on strength gains of confined concrete, where r is the round off corner
radius; D is the side of a square column; Acc is the effectively confined area and Ac is the gross cross section area.
The investigation showed that the variable 2r/D is more adequate to estimate the shape effectiveness coefficient (ke)
and consequently the effective lateral confinement pressure fle.

6. ACKOWLEDGEMENTS
The authors express their gratitude to FAPESP (São Paulo State Foundation for Research) for the financial support
and scholarship.

7. REFERENCES
Carrazedo, R. (2005). Confinement mechanisms in concrete columns wrapped by carbon fiber reinforced polymers
subjected to flexural compression. Ph.D. Thesis – São Carlos School of Eng. / Univ. of São Paulo, São Carlos, 2005.
Féderation Internationale du Betón (2001). FIB – Externally bonded FRP reinforcement for RC structures. Bulletin
14, July, 2001.
Grassl, P.; Lundgren, K.; Gylltoft, K. (2002). “Concrete in compression: a plasticity theory with a novel hardening
law”. International Journal of Solids and Structures. v. 39, p. 5205–5223.
Rochette, P. (1996). Confinement of short square and rectangular columns with composite materials, MS Thesis,
Univ. of Sherbrooke, Quebec, Canada.

498
Third International Conference on FRP Composites in Civil Engineering (CICE 2006)
December 13-15 2006, Miami, Florida, USA

SLENDERNESS EFFECTS ON CIRCULAR FRP-WRAPPED

REINFORCED CONCRETE COLUMNS
Jason Fitzwilliam
(M.Sc. Student, Queen's University, Kingston, Ontario, Canada)

Luke Bisby, Ph.D., P.Eng.

(Assistant Professor, Queen's University, Kingston, Ontario, Canada)

ABSTRACT
Externally-applied circumferential FRP wraps have proven to be an efficient and effective technique to repair and
strengthen structurally deficient reinforced concrete columns. Guidelines for the design of FRP-wrapped concrete
members are currently available and field applications of this technique have been implemented around the world.
However, two specific and related issues that require research attention are eccentric loading and slenderness effects
on FRP-wrapped concrete columns. To date, the vast majority of tests on FRP-confined concrete have been
performed on short, un-reinforced, small-scale concrete cylinders with height-to-diameter ratios of less than 3, and
tested under concentric, monotonic axial load. These tests have been useful as a means of accurately characterizing
the stress-strain behaviour of FRP-confined concrete such that it is now reasonably well understood. In practice
however, most columns are subjected to eccentric loads and have considerably larger height-to-diameter ratios. The
potential for increased susceptibility of FRP-wrapped columns to slenderness effects has received relatively little
research attention, both for columns that become slender as a consequence of wrapping and increased service loads,
and for strengthening of pre-existing slender members. This paper presents the results of an experimental program to
investigate the effects of slenderness on FRP-wrapped circular reinforced concrete columns.

KEYWORDS
Fibre reinforced polymers, FRP wraps, reinforced concrete columns, confinement, slenderness, eccentric loads.

1. INTRODUCTION AND BACKGROUND

FRP-wraps for axial strengthening of reinforced concrete columns are an effective and efficient application of FRP
materials in Civil Engineering (Bisby et al., 2005). However, most existing design guidelines are applicable only to
short columns under concentric axial loads. In reality, concrete columns are often subjected to eccentric loads and
may be susceptible to increased instability resulting from their slenderness. Furthermore, FRP-wrapping of short
columns has the potential to change the behaviour of from “short”, under existing loads, to “slender” under
increased (strengthened) loads. To date, relatively few studies have investigated the behaviour of slender FRP-
confined concrete members under eccentric axial loading (Fam et al., 2003; Hadi, 2006; Mirmiran et al., 2001; Tao
et al., 2004). In reinforced concrete column wrapping applications, FRP wraps are typically applied in the hoop
direction and thus provide confining reinforcement only – they cannot be relied upon for additional bending or
buckling strength. Furthermore, while several theoretical studies have been performed to develop axial load-moment
(M-N) interaction diagrams for FRP-confined concrete columns (Teng et al., 2002), these have not generally been
validated through comparison with test data. Since current design recommendations for FRP-wraps are limited to the
design of short concentrically loaded columns (ACI, 2002; ISIS, 2001), they cannot be used in cases where
slenderness or load eccentricity are concerns, and additional experimental data and analysis are therefore required.

With the above points in mind, the objectives of the research program are: (1) to experimentally evaluate the
performance and effectiveness of circumferential FRP-wraps for circular concrete columns of increasing
slenderness; (2) to study the effectiveness of longitudinal FRP wraps in reducing the susceptibility FRP-
wrapped circular concrete columns to slenderness effects; (3) to develop and validate procedures for the

499
development of theoretical bending moment-axial load interaction diagrams for circular FRP-wrapped
reinforced concrete columns and subsequently conduct parametric studies; and (4) to suggest rational
procedures for the design of FRP-wrapped circular columns that account for the effects of slenderness and/or
load eccentricities.

2. EXPERIMENTAL PROGRAM
As outlined in Table 1, eighteen small-scale circular reinforced concrete columns of varying slenderness were tested
to failure in monotonic, eccentric axial compression. All columns were 152 mm in diameter and were reinforced
with four 6.4 mm diameter bars longitudinally and with 6.4 mm diameter circular ties spaced at 100 mm center-to-
center in the transverse direction with a cover of 25 mm to the longitudinal reinforcement. Parameters varied among
the specimens included the length (i.e., slenderness), number of hoop wraps (i.e., level of confinement), and number
of longitudinal wraps (i.e., level of increase in flexural rigidity). The reader should note that kL/r in Table 1 has been
calculated based on a transformation of the initial tangent elastic moduli of the component materials. Clearly, at high
levels of axial load, the elastic modulus of the concrete is significantly decreased and the effects of the longitudinal
FRP wraps on the effective kL/r values are much more significant than implied by the data in Table 1. The
SikaWrap® Hex 230C carbon/epoxy FRP strengthening system was used for all wraps and in all cases the
longitudinal wraps were installed before the hoop wraps. The FRP system has a manufacturer specified design
thickness of 0.381 mm, and ultimate strength and strain of 894 MPa and 1.33 % respectively. The columns were
tested using end conditions that were pinned-pinned about one horizontal axis and effectively fixed-fixed about the
other. All columns were tested under a constant load eccentricity of 20 mm to simulate inevitable field eccentricities
and to promote bending in the desired direction. All columns were instrumented with four Pi-type strain gauges in
the axial direction and eight electrical resistance strain gauges in the hoop direction at mid-height. Lateral deflection
was monitored using three linear potentiometers.

Table 1: Overview of Experimental Program and Summary of Test Results

Ave. Peak Ave. Peak

Lateral
No. No. Ult. axial axial hoop hoop
a defl. @
Name kL/r hoop long. load strain @ strain @ strain @ strain @
ultimate
wraps wraps (kN) ultimate ultimate ultimate ultimate
(mm)
(%) (%) (%) (%)
300U-A 9.74 0 0 471 0.149 0.495 - - 0.0
300U-B 9.74 0 0 462 0.135 0.480 - - 0.4
300C-1-0-A 9.74 1 0 675 0.292 1.930 0.660 1.20 1.9
300C-1-0-B 9.74 1 0 679 0.300 1.660 0.810 1.67 1.4
300C-1-2-A 9.51 1 2 681 0.601 2.000 0.464 0.734 0.6
300C-2-0-Ab 9.74 2 0 671 0.566 1.820 0.476 0.981 0.0
300C-2-0-B 9.74 2 0 911 0.858 3.140 N/Ac 1.37 N/Ac
600U-A 17.63 0 0 428 0.217 0.611 - - 1.4
600C-1-0-A 17.63 1 0 563 0.309 1.210 0.235 0.373 5.3
900U-A 25.53 0 0 398 0.100 0.199 - - 2.1
900C-1-0-A 25.53 1 0 549 0.198 1.280 0.483 0.939 9.7
1200U-A 33.42 0 0 389 0.113 0.267 - - 2.7
1200U-B 33.42 0 0 411 0.102 0.315 - - 6.2
1200C-1-0-A 33.42 1 0 451 0.152 2.420 0.326 0.781 27.4
1200C-1-0-B 33.42 1 0 481 0.125 1.630 0.263 0.627 23.3
1200C-1-2-A 32.65 1 2 584 0.463 2.010 0.413 0.820 20.3
1200C-1-4-A 31.91 1 4 673 0.975 2.540 0.419 0.847 19.1
1200C-2-0-A 33.42 2 0 539 0.175 3.100 0.572 1.51 45.7
a
For example, 1200C-1-4-A is 1200mm long, wrapped with carbon, 1 layer hoop FRP, 4 layers longitudinal FRP
b
Failure of restraining collar caused premature failure of specimen at lower loads than expected
c
Some data unavailable due to malfunctioning of lateral displacement and hoop strain instrumentation

500
3. EXPERIMENTAL RESULTS AND DISCUSSION
The test data provided in Table 1 show the observed trends in ultimate load capacity, average and peak axial strain,
and average and peak hoop strain for all of the columns. As expected, column strength increases with increasing
numbers of hoop wraps and with decreasing slenderness. Also as expected, the effects of hoop wraps appear more
significant for shorter columns, whereas the effect of longitudinal wraps is more significant for slender columns. In
fact, the data indicate that the addition of four longitudinal wraps allowed slender column 1200C-1-4-A to achieve
the same strength level as columns 300C-1-0-A and 300C-1-0-B, whereas without the longitudinal wraps the slender
columns experienced strength reductions in the order of 31% due to slenderness effects. For columns of the same
slenderness, the data of Table 1 also show that the lateral deflection at failure is increased by increasing levels of
confinement (which tends to increase the failure strain of the concrete in compression), and decreased by the
addition of longitudinal wraps (which tend to increase the columns’ flexural rigidity, as expected). In all cases,
confinement with FRP hoop wraps drastically increased the columns’ deformation capacity.

Figure 1(a) gives applied load versus mid-height deflection plots for all single-wrapped columns, and clearly shows
the effects of increasing slenderness on load carrying capacity and lateral deformation of the FRP-wrapped columns.
Figure 1(b) also shows the effect of increasing slenderness on the ultimate strength of both wrapped and unwrapped
columns (including least-squares regression trends), where it appears that slenderness effects are more significant for
FRP-wrapped columns. Of course, this is due to the flexure-dominated behaviour of the slender columns and the fact
that circumferential FRP wraps’ effects on the strength of confined concrete are more significant than their effects
on stiffness. Again, it appears that the addition of longitudinal wraps (increasing flexural rigidity) significantly
improves the behaviour of the slender FRP-wrapped columns (but has no significant effect on short columns).
1.6
300C-1-0-A 1.5
600 300C-1-0-B
1.4
Total Applied Load (kN)

600C-1-0-A
900C-1-0-A Normalized Ultimate Load 1.3
1200C-1-0-A
1200C-1-0-B 1.2
400
1.1
1

200 0.9 Unconfined

0.8 1 hoop
1 hoop 2 long
0.7
1 hoop 4 long
0 0.6
-5 0 5 10 15 20 25 30 0 10 20 30 40
(a) Lateral Deflection @ Mid-Height (mm) (b) Slenderness ratio (kL/r )

Figure 1: (a) Experimental Load versus Mid-Height Deflection Curves for all Single Wrapped Columns and
(b) Effect of Slenderness on Ultimate Load Capacity
1400
1200U-A
1200U-A
1200U-B
600 1200U-B 1200
1200C-1-0-A 1200C-1-0-A
Distance from bottom hinge (mm)
Total Applied Load (kN)

1200C-1-0-B 1200C-1-0-B
1200C-2-0-A 1000 1200C-1-2-A
1200C-1-2-A
1200C-1-4-A
1200C-1-4-A
400 800 1200C-2-0-A

600

200 400

200

0 0
0 10 20 30 40 50 -1 0 1 2 3 4

(a) Lateral Deflection @ Mid-Height (mm) (b)

Lateral Deflection (mm)

Figure 2: (a) Experimental Load versus Mid-Height Deflection Curves for all 1200mm-Long Columns and (b)
Deflected Shapes of all 1200mm-Long Columns at 390kN Applied Load

501
Figure 2(a) provides a plot of applied load versus mid-height deflection for all 1200 mm long columns with various
wrapping schemes. This figure shows the beneficial effects of both increasing levels of confinement and addition of
longitudinal FRP wraps on the load carrying capacity and deformation response of slender FRP-wrapped concrete
columns. Indeed, adding four longitudinal wraps allows the slender FRP-wrapped column to achieve the same
strength as the short wrapped columns. Figure 2(b) shows deflected shapes recorded for all of the 1200 mm long
columns at 391 kN (the ultimate load for the unwrapped 1200 mm long columns) and clearly illustrates the
stiffening effect of both longitudinal FRP wraps and increasing levels of confinement on slender FRP-wrapped
circular concrete members.

5. CONCLUSIONS
This paper has very briefly presented and discussed the results of a series of tests on small-scale FRP-wrapped
reinforced concrete columns, conducted to study the effects of both circumferential and longitudinal FRP wraps on
the performance of circular reinforced concrete columns of increasing slenderness under a constant eccentric axial
load. On the basis of the test observations presented in this paper, the following conclusions can be drawn:
• Circumferential FRP wraps increase strength and deformation capacity of both short and slender circular
concrete columns, although the effects on strength are more significant for short columns.
• Column strength increases with increasing numbers of hoop wraps (i.e., increasing levels of confinement) and
with decreasing slenderness.
• Adding longitudinal FRP wraps to slender FRP-wrapped circular concrete columns can improve their behaviour
and allow them to achieve strengths similar to equivalent short FRP-wrapped columns. The longitudinal FRP
wraps had no significant effect on the strength of deformation capacity of short columns.
Research is ongoing in this area to better understand the mechanics of confinement for slender and eccentrically-
loaded FRP-wrapped circular concrete columns. Procedures are also being developed for prediction of theoretical
axial load-moment interaction diagrams and appropriate slenderness limits for FRP-wrapped columns. It his hoped
that experimental data obtained in this study can be used to at least partially validate theoretical and design
calculations for these types of members.

6. ACKNOWLEDGEMENTS
The authors are members of the Intelligent Sensing for Innovative Structures Network (ISIS Canada) and wish to
acknowledge the support of the Networks of Centres of Excellence Program of the Government of Canada and the
Natural Sciences and Engineering Research Council of Canada. We would also like to thank Queen's University and
Sika Canada Inc.

7. REFERENCES
ACI (2002). ACI 440.2R-02, American Concrete Institute, Farmington Hills, USA.
Bisby, L.A., Dent, A.J.S., and Green, M.F. (2005). “Comparison of confinement models for fiber-reinforced-
polymer-wrapped concrete”. ACI Structural Journal, Vol. 102, No. 1, pp. 62-72.
Fam, A., Flisak, B., and Rizkalla, S. (2003). “Experimental and analytical modeling of concrete-filled fiber-
reinforced polymer tubes subjected to combined bending and axial loads”. ACI Structural Journal, Vol. 100, No. 4,
pp. 499-509.
Hadi, M.N.S. (2006). “Comparative study of eccentrically loaded FRP wrapped columns”. Composite Structures,
Vol. 74, pp. 127-135.
ISIS (2001). Design Manual No. 4, ISIS Canada, Winnipeg, Canada.
Mirmiran, A., Shahawy, M., and Beitleman, T. (2001). “Slenderness limit for hybrid FRP-concrete columns”.
Journal of Composites for Construction, Vol. 5, No. 1, pp. 26-34.
Tao, Z., Teng, J.G., Han, L-H., and Lam, L. (2004). “Experimental behaviour of FRP-confined slender RC columns
under eccentric loading”. Proceedings of the Second International Conference on Advanced Polymer Composites for
Structural Applications in Construction, Editors: L.C. Hollaway, M.K. Chryssanthopoulos, and S.S.J. Moy,
University of Surrey, Guildford, UK, pp. 203-212.
Teng, J., Chen, J., Smith, S., and Lam, L. (2002). FRP Strengthened RC Structures, Wiley, UK.

502
Third International Conference on FRP Composites in Civil Engineering (CICE 2006)
December 13-15 2006, Miami, Florida, USA

THEORETICAL MODEL FOR FRP-CONFINED CIRCULAR

CONCRETE-FILLED STEEL TUBES UNDER AXIAL COMPRESSION
J. G. Teng
(Chair Professor of Structural Engineering, Department of Civil and Structral Engineering,
The Hong Kong Polytechnic University, Hong Kong, China)

Y. M. Hu
(PhD Student, Department of Civil and Structral Engineering,
The Hong Kong Polytechnic University, Hong Kong, China)

ABSTRACT
Fibre-reinforced polymer (FRP) jackets have been widely used to confine reinforced concrete (RC) columns for
enhancement in both strength and ductility. More recently, the benefit of FRP confinement of concrete-filled steel tubes
has been explored by researchers. The main aim of this paper is to present and verify a theoretical model for FRP-
confined concrete-filled steel tubes based on a recent analysis-oriented model for FRP-confined concrete. To this end,
results of a series of tests on FRP-confined concrete-filled steel tubes are first presented. The theoretical model is next
presented and verified using these test results. Both the test and the theoretical results show that FRP jacketing can
significantly enhance the ultimate load but may reduce the ductility of concrete-filled FRP columns.

KEYWORDS
Fibre-reinforced polymer (FRP), concrete-filled steel tubes, confinement, theoretical model, tests

1. INTRODUCTION
In concrete-filled steel tubes, the concrete and the steel tube interact in a beneficial manner: the steel tube confines
the concrete and the concrete delays the occurrence of local buckling in the steel tube. Concrete-filled steel tubes are
thus an economic form of structural members. Although inward buckling deformations are now prevented,
degradation in steel confinement, strength and ductility can result from inelastic outward local buckling. Xiao (2004)
and Xiao et al. (2005) recently explored the use of FRP jackets for the confinement of the critical regions of
concrete-filled steel tubes. Although his work was directed at new construction, the same concept can be employed
in the retrofit of columns. In such concrete-filled steel tubular columns, the inward buckling deformation of the steel
tube is prevented by the concrete core while the outward buckling deformation is prevented by the FRP jacket. FRP
jacketing therefore provides an effective means of suppressing local buckling failures at the column ends. Teng and
Hu (2005) confirmed the effectiveness of this technique.

The main aim of this paper is to present and verify a theoretical model for FRP-confined concrete-filled steel tubes based
on a recent analysis-oriented model for FRP-confined concrete developed by Teng et al. (2006). To this end, results of a
series of tests on FRP-confined concrete-filled steel tubes are first presented. The theoretical model is next presented and
verified using these test results. Both the test and the theoretical results show that FRP jacketing can significantly
enhance the ultimate load but may reduce the ductility of concrete-filled FRP columns.

2. EXPERIMENTS
2.1 Test Specimens and Procedure

The existing test data of FRP-confined concrete-filled steel tubes is very limited (Xiao 2004; Xiao et al. 2005).
Several series of tests have therefore been conducted at The Hong Kong Polytechnic University to supplement the

503
existing test data, to provide further insight into the experimental behavior, and provide test data for the verification
of theoretical models in the future. Only the first series (series I) of tests are presented here. This series consisted of
four specimens: I-F0, I-F1, I-F2, and I-F3, where “I” indicates the series they below to, “F” represents the FRP
jacket with the number that follows representing the number of plies of the FRP jacket. The steel tubes all had an
outer diameter of 165 mm, a thickness of 2.75 mm (i.e. a D/t ratio of 60), and a length of 450 mm and were filled
with concrete with a cube compressive strength of 56 MPa determined from three cube tests. The average values of
the elastic modulus, yield stress and ultimate strength of the steel from tensile tests of three coupons taken from the
same long tube which provided the four steel tubes in the test specimens were 201.3 GPa, 385.9 MPa, and 486.8
MPa respectively. The average values of the elastic modulus and tensile strength from five coupon tests for the
GFRP, calculated on the basis of the nominal ply thickness of 0.17 mm, were 80.1 GPa and 1,825.5 MPa
respectively, leading to an ultimate tensile strain of 0.0228.

The compression tests were all conducted using an MTS machine with displacement control at a constant rate of
0.5mm/min until failure. Four bidirectional strain gauges with a gauge length of 20mm were installed evenly at 900
apart at the mid-height to measure the axial and hoop strains of the FRP jacket. The total axial shortening was
measured by using three transducers at approximately 1200 apart.

2.2 Test Results

Specimen I-F0 (concrete filled steel tube without FRP jacketing) failed by local outward buckling, as has been
observed in many tests by other researchers. Specimens I-F1, I-F2 and I-F3 (FRP-confined concrete-filled steel
tubes) all failed by the rupture of the FRP jacket due to hoop tension (Figure 1). Once rupture of the FRP jacket
occurred, the confinement effect of the FRP jacket disappeared and the load carried by the tube reduced immediately
and rapidly. With further loading, local buckling occurred where FRP ruptured. The axial load-axial shortening
curves of these four specimens are shown in Figure 2, where the axial shortening was the mean reading of the three
transducers located on the loading platens. The curve of the concrete-filled steel tube without FRP confinement
features a slowly descending branch after reaching the peak load, while those of the FRP-confined tubes all exhibit a
monotonically increasing bilinear shape before the rupture of the FRP at peak load. Following the rupture of the
FRP jacket, the load reduced immediately and rapidly and the specimen then was able to follow a ductile path at a
lower load level. The results from these tests illustrate two important points: (a) the FRP jacket can effectively
enhance the strength of the column but the strain capacity is not always enhanced; (b) if ductility enhancement is the
aim, then a suitable gap between the steel tube and the FRP jacket should be provided as has been explored by Xiao
et al. (2005).
2400

2000

1600
Axial Load (kN)

1200
I - F0 fcu = 56 MPa
800 I - F1
I - F2 D / t = 60
I - F3
400 FRP Rupture
I-F0 I-F1 I-F2 I-F3
0
0 5 10 15 20
Axial Shortening (mm)
Figure 1: Failure modes of test specimens Figure 2: Experimental load-axial
shortening curves of all four specimens
3. THEORETICAL MODEL
3.1 General

Teng et al. (2006) proposed a theoretical model for FRP-confined concrete for which the confining pressure is
related to the hoop strain in the FRP jacket by the following simple expression:
2 E frp t frpε θ , frp
σl = (1)
D

504
where σ l is the lateral confining pressure, D is the diameter of the concrete core, and E frp , t frp and ε θ , frp are the
elastic modulus, thickness and hoop strain (equal to the lateral strain in magnitude) of the FRP jacket. Teng et al.
(2006) showed that their model can also be used to concrete confined with any other material, provided an equation
equivalent to Eq. 1 to predict the confining pressure-hoop strain relationship can be established for the confining
material. Teng et al. (2006) has shown that their model provides close predictions of test results. It is easy to see that
if the confining pressure-hoop strain equation (Eq. 1) is modified to reflect the combination of confinement provided
by the steel tube and the FRP jacket, this model can be expected to be capable of predicting the behavior of the
present tests.

3.2 Evaluation of the Stress-Strain Curve

The evaluation of confining pressure is based on the assumption that the steel tube does not experience any bending
or buckling deformations. This assumption is later shown to be reasonable for FRP-confined steel tubes before the
rupture failure of the FRP jacket. The total confining pressure is equal to the sum of that from the steel tube and that
from the FRP jacket. The confining pressure from the FRP jacket can be easily found using Eq. 1, but the evaluation
of the confining pressure from the steel tube is slightly more involved.

Based on the assumption mentioned above, the steel tube is subjected to membrane stresses only. As strain
hardening of the steel is generally insignificant before the tensile rupture of the FRP jacket, the stress-strain curves
of the steel are assumed to be elastic-perfectly plastic, which approximate the test stress-strain curves closely. The
concrete, the steel tube and the FRP jacket are all assumed to be perfectly bonded at the interfaces in the analysis.
The effect of permitting separations to occur between the steel tube and the concrete in the initial stage of loading on
the predicted axial stress-strain response was investigated and was found to be small, so this assumption is
acceptable. The evaluation of the axial stress-axial strain curve of the concrete in an FRP-confined steel tube
involves an incremental process.

In each increment, a small increment of hoop strain is specified first, from which the total lateral strain, the hoop
stress in the FRP jacket and the confining pressure supplied by the FRP jacket can be calculated. The total axial
strain that satisfies the lateral-axial strain equation of Teng et al. (2006) corresponding to the total lateral strain is
then determined using the bi-section method, starting with suitable lower and upper bounds. The axial strain
increment can then be easily deduced. In this iterative process, the J2 flow theory of plasticity for a plane stress
condition is employed to determine the axial and hoop stress increments for known hoop (lateral) and axial strain
increments. A tolerance of 0.1% of the axial strain was used in obtaining the numerical results presented later in this
paper. Once this tolerance is satisfied, the hoop strain, the axial strain and the confining pressure are employed in
Teng et al.’s (2006) model to generate a point on the stress-strain curve for the concrete in the steel tube and the
analysis process moves to the next lateral increment. The evaluation of the stress-stain curve for concrete in FRP-
confined steel tubes is therefore an incremental iterative process.

3.3 Comparison with Test Data 2500

In Figure 3, the test axial load-axial strain curves are

compared with the predictions of the theoretical 2000

model. In these comparisons, the nominal axial strain

was obtained from the transducer readings as the
Axial Load (kN)

1500
average shortening divided by the length of the
specimen, and the concrete cylinder compressive
I-F0 (prediction)
strength f c' was taken as 0.8 times the cube 1000 I-F0 (test)
compressive strength f cu . The theoretical axial load I-F1 (prediction)
I-F1 (test)
resisted by the specimen is a sum of that resisted by I-F2 (prediction)
the steel tube and that resisted by the confined 500
I-F2 (test)
concrete. It is seen that for specimens I-F2 and I-F3, I-F3 (prediction)
which had a two-ply FRP jacket and a three-ply FRP I-F3 (test)
0
jacket respectively, the test results are in close 0.00 0.01 0.02 0.03 0.04
agreement with the theoretical predictions. However,
Nominal Axial Strain
for specimen I-F0 which had no FRP jacket and
specimen I-F1 which had a one-ply jacket, the Figure 3: Comparison of load-strain curves between
test results and theoretical predictions

505
theoretical predictions significantly overestimate the resistance of the specimens. These comparisons clearly show
that the model is limited by its assumption of no bending or buckling deformations in the steel tube and provides
accurate predictions of test results only before significant bucking deformations develop in the steel tube. If a thick
steel tube is used so that local buckling does not occur or if a sufficiently strong FRP jacket is employed to suppress
buckling deformations in the steel tube, then the theoretical model can be expected to deliver accurate predictions. In
specimen I-F0, no FRP jacket was provided and local buckling of the steel tube was obvious during the test. In
specimen I-F1, the comparison given in Fig. 3 indicates that the one-ply FRP jacket was insufficient to suppress the
local buckling deformations in the steel tube.

As most of the strain gauges had been damaged before the FRP jacket ruptured, in making the theoretical predictions,
the hoop rupture strain of the jacket was assumed to be equal to that obtained from flat coupon tensile tests (ie
0.0228). The resulting theoretical stress-strain curves show rupture failures of the FRP jackets at significantly higher
axial strain levels than observed in the tests. This observation indicates that the FRP jackets in the tests ruptured at
hoop strains lower than the rupture strain from tensile coupon tests. This phenomenon of premature hoop rupture
failure has been widely observed in FRP-confined concrete cylinders (Teng and Lam 2004; Lam and Teng 2004).
Further research is needed to evaluate the extent of reduction in the ultimate strain of the FRP jacket when used to
confine a steel tube.

4. CONCLUSIONS
This paper has been concerned with the modeling of the axial compressive behavior of FRP-confined concrete-filled
steel tubes. Results of a series of axial compression tests on FRP-confined circular concrete-filled steel tubes were first
presented. A theoretical model for FRP-confined concrete-filled circular steel tubes modified from the recent analysis-
oriented model for FRP-confined concrete proposed by Teng et al.’s (2006) was next presented and compared with these
test results. It can be concluded that the theoretical model, based on the assumption of no bending/buckling deformations
in the steel tube, provides close predictions of test results if the steel tubes does not buckle. This condition is satisfied
when the steel tube is sufficiently thick or the FRP jacket is sufficiently strong to suppress local bucking deformations.
The comparisons also indicate that FRP jackets confining a concrete-filled circular steel tube rupture at a hoop strain
lower than the ultimate tensile strain from flat coupon tensile tests, similar to what has been widely observed in tests
on FRP-confined plain concrete cylinders. The success of the proposed theoretical model lends further support to
Teng et al.’s (2006) model as an accurate analysis-oriented model for concrete confined by different materials. The
proposed model can be deployed to again further insights into the effect of various parameters, such as the inclusion
of a gap between the steel tube and the FRP jacket to increase the ductility of the FRP-confined tube (Xiao et al.
2005). The authors will report their research on the effect of such a gap in a future paper.

5. ACKNOWLEDGEMENTS

The authors are grateful to the Research Grants Council of the Hong Kong Special Administrative Region (Project
No: PolyU 5269/05E) and The Hong Kong Polytechnic University (Project Code: RGU4) for their financial support.

6. REFERENCES

Lam, L. and Teng, J.G. (2004). “Ultimate condition of fiber reinforced polymer-confined concrete”. Journal of
Composites for Construction, ASCE, Vol. 8, No. 6, pp. 539-548.
Teng, J.G. and Lam, L. (2004). “Behavior and modeling of fiber reinforced polymer-confined concrete”. Journal of
Structural Engineering, ASCE, Vol. 130, No. 11, pp. 1713-1723.
Teng, J.G. and Hu, Y.M. (2005). “Enhancement of seismic resistance of steel tubular columns by FRP jacketing”.
Proceedings, 3rd International Conference on Composites in Construction, 11-13 July, Lyon, France, pp. 307-314.
Teng J.G., Huang Y. L., Lam L., Ye L.P. (2006). “Theoretical model for fiber reinforced polymer-confined
concrete”. Journal of Composites for Construction, ASCE, accepted for publication.
Xiao, Y. (2004). “Application of FRP composites in concrete columns”. Advances in Structural Engineering, Vol. 7,
No. 4, pp. 335-341.
Xiao, Y., He, W.H. and Choi, K.K. (2005). “Confined concrete-filled tubular columns”. Journal of Structural
Engineering, ASCE, Vol. 131, No. 3, pp. 488-497.

506
Part XVIII. Repair Techniques
Third International Conference on FRP Composites in Civil Engineering (CICE 2006)
December 13-15 2006, Miami, Florida, USA

METHODOLOGICAL CONSIDERATIONS FOR REINFORCED

CONCRETE STRUCTURES REPAIR USING POSTENSIONED CFRP
STRIPS.
José Antonio Bellido de Luna del Rosario
(Researche, Universidad Central de Chile, Santiago de Chile, Chile,)

ABSTRACT
Fiber Bond composites (Carbon Fiber Reinforced Polymer) CFRP, are used like an external reinforcing of structural
elements. This system is completely studied and has been applied with much success to a great number of structures
with problems of resistance and camber

The paper shows a methodologic study conducted to determine the CFRP area and the tension to which it will have
to be submitted to resist the additional ultimate state loads to which the analyzed structure will be exposed, or for the
deflection decrease and the excessive cracks limiting the demanded minimum level.

KEYWORDS
Reinforced, concrete, carbon, postensioned

1. INTRODUCTION

Fiber Bond composites (Carbon Fiber Reinforced Polymer) CFRP, are used like an external reinforcing of structural
elements. This system is completely studied and has been applied with much success to a great number of structures
with problems of resistance and camber

The paper shows a methodology study conducted to determine the CFRP area and the tension to which it will have
to be submitted to resist the additional ultimate state loads to which the analyzed structure will be exposed, or for the
deflection decrease and the excessive cracks limiting the demanded minimum level.

2. MINIMAL REINFORCEMENT CRITERIA.

It is important, in case of FRP use, to take into account different considerations of the internal characteristics of
materials that could determine certain limits in reinforcement. Reinforcement limits are imposed to protect against
structure collapse and also in case of bond failure or other FRP failures due to fire vandalism or other causes. This
criterion tries to avoid a surprising collapse of the structure.

If this condition is not fulfilled, FRP reinforcement is not recommendable and other type of reinforcement should be
selected.

3. MAXIMUM REINFORCEMENT CRITERIA.

Maximum reinforcement criterion is focused in keeping a ductile behavior of CFRP reinforced element. Limit
amount of reinforcement should be taken , the reinforced element would have a ductile behavior at the moment of
failure.

4. CONFIGURATION AT ULTIMATE LIMIT STATE.

507
Figure 1 represents the equilibrium condition for ultimate limit state of a beam at flexure with ordinary and CFRP
reinforcement.

Figure 1 : Force, stress and strain diagram for a rectangular section reinforced with ordinary and CFRP
reinforcement.

From equilibrium diagram shown in Fig. 1, the value of the force acting on CFRP is obtained as:

TCFRP = 0,85. f ' c .b.a − Ts (3)

From moment equilibrium represented in Fig. 1 is seen:

Mu ⎛ a⎞ ⎛ a⎞
= Ts .⎜ d − ⎟ + TCFRP ⎜ h − ⎟ (4)
φ .ψ CFRP ⎝ 2⎠ ⎝ 2⎠

Where ψ CFRP is a CFRP strength reduction factor depending on the exposition conditions, being ψ CFRP = 0,85 for
exterior and aggressive environments and ψ CFRP = 0,95 for exposition in interior locals.

Values of compressions block “a” and force TCFRP can be obtained combining adequately equations (3) and (4).

5. DETERMINATION OF THE REQUIRED PRESTRESSING FORCE.

Existing section at the time of prestressing is strained and cracked, loads acting on the element at the moment of
repair should be determined. What is normal is to have the solely action of permanent load; thus the required
prestressing force to nullify tension in the ordinary reinforcement should be determined for the time of
rehabilitation.

To achieve the above mentioned condition, it is necessary to determine the initial strain on the steel due to
permanent loads at the instant just before putting the reinforcement. With this strain, it is possible to obtain the stress
on the steel to be counteracted with the prestressing force, achieving the closure of cracks in the element to be
reinforced and that the beginning of flexure development of steel and CFRP from a known point for the steel, εs0 =
0.
From the compatibility of deflections, the following equation for ordinary deflection for steel under permanent
loads:

508
 ε 1a =
[(M CP ).10 −5 (d − k1d )]
(5)
E c .I cr1

Where Mcp; is the moment due to permanent loads at the cross section at the time
of reinforcement.
Icr1; is the cracked moment of inertia at the cross section according to Chapter 9 of ACI 318-2002.
k1; coefficient defining the depth of compression block, obtained as:

k1 = (ρ s .ns )2 + 2.(ρ s .ns ) − ρ s .ns (6)

Being: ρs: Ordinary reinforcing ratio.

Es
ns: Elasticity modulus ratio; ns =
Ec

Figure 2 : Strain configuration due to pretressing force.

Strain diagram when the prestressing force acts is shown in Fig. 2, obtaining this force as:
ε 1a. E s
P= (7)
⎛ 1 ⎞ [(h − k1 .d )(
. d − k1 .d )]
⎜⎜ ⎟⎟ +
⎝ As ⎠ I cr1

When the element is tensioned and the load are acting, there will be generated on the CFRP strip two kinds of
deformations: the first one induced by the strip prestressing itself, and the second due to loads action, producing the
deflection diagram indicated in Fig. 3, from which the required CFRP area can be obtained with the following
analysis:

Figure 3 : Tension diagram at ultimate limit state.

ε CFRP = ε fCFRP + ε pCFRP (8)

Where: ε fCFRP ; is the CFRP strip deflection due to load action, ε pCFRP ; is the CFRP strip deflection due to
prestressing action.

509
The required CFRP strip area can be obtained as:
(TCFRP − P )
ACFRP = (12)
ε fCFRP .ECFRP

This area should be brought to a discrete amount of strips and then determine the CFRP ratio according to:
AdisCFRP
ρ CFRP = (13)
b.d

6. PRESTRESSING FORCE LOSSES.

In CFRP tension there are also some losses, some of them by the same causes as in a normal prestressed element.
Each one of these losses should be evaluated to determine the effective prestressing force to provide to the strips,
obtained as:
Pefec = P + ΔP (14)
Being ΔP the value of losses.

7. VERIFICATION OF MAXIMUM CFRP DEFLECTIONS.

After obtaining the effective prestressing force to apply to the CFRP area, it is possible to determine and control the
limit of maximum deflection due to prestressing and the ultimate deformation of CFRP strips.

8. CONCLUSIONS.
An important condition at the time of structural reinforcement design with post-tensioned CFRP strips is “existing
reinforcement limit”, which conditions the reinforced concrete element to have a minimum bearing capacity for the
new loads to which it is going to be submitted.

Though post-tensioned CFRP strips have mechanical properties, regarding stress and strain, quite high in
comparison to those of the steel, the previously mentioned “ductility” and “minimum reinforcement criteria”, allows
a greater design factor with the application of this new technology. Remember that the study of composite materials
and their applications is relatively new; therefore these criteria are valid in this methodology.

The methodology developed here is focused on the application of post-tensioned CFRP strips; however, it could be
applicable to any kind of material susceptible of being used as an active external reinforcement. This is due to the
fact that this theory is based on basic theories of structural design.

This methodology still requires multiple tests that allow corroborating its theoretical components.

9. REFERENCES

Aparicio, C. y Ramos, G. (1993) “Estado actual de la técnica del pretensado exterior aplicado a puentes de
carretera”, MOPT MA, Madrid.
Bellido de Luna, J.A. (2002) “Elementos Pre y Post tensados de Hormigón, Bases de Diseño. Control de Obra”. 4
edición, Universidad de Santiago de Chile, Santiago 2002.
Bellido de Luna, J.A. (1997)“La tecnología del Pretensado Parcial en Elementos de Hormigón”. Calculo y Diseño.
Universidad de Santiago de Chile. Santiago.
Brito, M.(1997) “Caracterizaçao do comportamento dos plásticos reforçados com vista a apliaçoes estructurais”,
LNEC, Departamento de materiais de contruçao, núcleo de cerámica e plásticos, Lisboa, Outubro.
Monteiro, L. y Gomes A. (1996) “Reforço à flexao de vigas de betão armado- modelos de dimensionamiento e
verificaçao de segurança, betão estructural”, 6º encuentro Nacional de la asociación de preesfrozado Portugues,
LNEC Noviembre,.
Sika Colombia S.A.(2003) “Guía de Diseño e Instalación Platinas Sika CarboDur, Tejidos SikaWrap, Platinas en
forma de L Sika CarboShear Platinas Preesforzadas Sika Leoba CarboDur y Sika StressHead”, Bogotá D.C.,
Colombia, Enero.

510
Third International Conference on FRP Composites in Civil Engineering (CICE 2006)
December 13-15 2006, Miami, Florida, USA

Physico-chemical and mechanical behaviors of fiber reinforced repair

mortars
Amjad Mallat
(PhD student, MSSMAT Laboratory, Ecole Centrale Paris, Châtenay-Malabry, France)

Abdenour Alliche
(Professor,LM2S Laboratory, Université Pierre et Marie Curie, Paris, France)

ABSTRACT
We have investigated the mechanical behavior of two fiber reinforced repair mortars. A lime-based mortar and a
cement-based mortar containing a small quantity of silica fume. Physico-chemical characterization of the materials
is carried out by measurements of porosity and shrinkage strains, and by Infra-red spectrometry and X-ray
diffraction analysis. The microstructure of the mortars is investigated by microscopic observations. Mechanical
characterization is based on compression, three-point bending and tensile tests. For that, we propose a new post-peak
tensile testing configuration to study the damaged behavior of the materials. The evolution of the mechanical
properties and the influence of the cure conditions of the mortars are also investigated in this study. The bond
between the repair mortars and substrate is studied by slant shear and three points bending tests. Results obtained
have demonstrated the influence of the surface roughness and moisture conditions on the bond strength. Finally,
SEM observations show the morphology of the substrate-to-repair mortar interfacial zone. They allow us to observe
and understand the phenomena occurred at the interface, affecting the bond between substrate and repair mortar.

KEYWORDS
Fiber reinforced repair mortars, physico-chemical characterization, mechanical characterization, bond strength.

1. INTRODUCTION
In recent years, repair, refurbishment and maintenance of concrete structures have become a significant part of the
total cost of construction worldwide (Mangat and Limbachiya, 1997). Various repair methods and materials are cur-
rently used to overcome damage in deteriorated structures. Fiber-reinforced mortar (FRM) is one of materials widely
used to repair old concrete. Under current practice, there are no standard procedures for the design of patch repairs.
Design is usually based on the experience of specialist contractors and when selection of repair materials is made,
emphasis is normally given to their relative short-term properties such as strength, bond and early age plastic
shrinkage/expansion etc. Although these properties indicate the immediate performance of the repair, they give little
information on its long-term performance. Therefore, there is an important need for recognizing and understanding
the properties of repair materials, which are of significance to the subsequent structural behaviors of repaired con-
crete members. The present study presents an experimental method to investigate the mechanical behavior of two
fiber-reinforced repair mortars, both in elastic and damaging regime.

2. MATERIALS AND EXPERIMENTAL TESTS

Two repair mortars (FLM and FOM) reinforced with polyacrylonitryle fibers were used, together with an ordinary
mortar (OM), which it was used to provide control specimens for comparison and to replace the substrate for bond
tests. FLM is a lime-based mortar containing thickening agent, limestone and additives. FOM is a non-shrinkable
cement-based mortar containing silica fume and additives. OM is prepared using CEMII cement. The mix propor-
tions (by weight) were 1:3.68:0.48 (cement:sand:water), to achieve 28 days strength of 25 MPa and of 20 GPa
Young’s modulus. All samples were cured and stocked at 23°C and 50% RH. Physico-chemical characterization of
the materials is carried out by measurements of porosity and shrinkage strains, XRD and FT-IR analysis, and by
SEM exams. Mechanical characterization was carried out by compression, tensile and three-point bending tests. The

511
bond strength is tested by slant-shear and three-point bending tests. For that a range of surface roughnesses with
various substrate moisture contents was used. Table 1 summarizes some physical properties of the mortars.

Table 1: Some physical properties of the mortars.

Mortar Grain size slump pH Porosity Density Fiber length Fiber content
FOM 0 – 2 mm 1-5 cm 12.5 23% 2.08 8 mm 0.2
FLM 0 - 1.25 mm 2 cm 12.5 31% 1.82 4 mm 0.3
OM 0 - 0.8 mm – 12.5 24% 2 No fiber No fiber

3. RESULTS
Shrinkage strains measurements show that the repair mortars exhibit greater shrinkage compared with the OM.
Although the FOM was designed to be non-shrink, however it displays higher shrinkage than the OM. The highest
shrinkage is accorded to the FLM. The shrinkage is fast during the first two weeks then it becomes relatively slow
with time. SEM observations of the FOM did not reveal the presence of added silica fume in the hardened paste,
which is probably due to a possible pozzolanic effect. The morphology of thickening agent within the FLM was not
observed by the SEM. Infra-red spectrometry and X-ray diffraction analyses of the anhydrous and hydrated
materials showed that the synthetic fibers and thickening agent would not modify the nature of the cement hydration
products even if the thickening agent seems to slow down the kinetic of the reaction. Fig. 1 presents the variation in
Young’s moduli and compression strengths of the repair mortars according to the age of materials. It is obvious that
both Young's modulus and compression strength of the FLM increase almost linearly with time. This evolution may
be due to: (1) the reduction of the hydration rate due to the water retention by the thickening agent and, (2) the slow
carbonation of the portlandite in the lime-based mortar. FOM modulus increases slightly with time. Its compressive
strength increases strongly at early age until 28 days, and hardly evolves beyond this age, what indicates that the
mortar has achieved most of its hydration during the first month.

Figure 1: Evolution of compression strengths and Young moduli of the repair mortars.

FLM flexural load increases with time, however the mortar becomes more brittle; at 210 days mortar age, FLM
samples fail brutally by three-point bending test. FOM flexural strength increases until 28 days. Then, a decrease of
strength has been noted. Three reasons can explain the evolution of the flexural strength of the materials: (1) the
porosity, (2) a possible difference in cement hydration, and (3) the percolation of the thickening agent over the FLM
samples, which may limit these samples from drying out and, therefore, from surface micro-cracking (Mallat, 2006).
The flexural strength is strongly influenced by the cure conditions. FLM strength increases with humid cure (100%
HR), contrary to the FLM whose the strength decreases with humidity. Both FOM and FLM flexural strengths
decrease when the samples were exposed to UV light which seem to degrade synthetic fibers.
Two testing configurations (Mallat and Alliche, 2006) were used to study tensile behavior (Fig. 3): (1) Direct tensile
test showing a linear behavior of the mortars until failure which occurs by an unstable way and, (2) Post-peak tensile
test giving complete tensile response. The highest tensile strength is accorded to the FOM containing silica fume.
Also, silica fume has increased the tensile strain capacity of the FOM compared with the OM having similar
Young's modulus. FLM shows the highest tensile strain capacity which may be due to the thickening agent and the
microstructure. The Young's moduli and tensile strengths of the repair materials increase with time. The FLM
Young's modulus evolves very slowly at early age and increases significantly at 210 days.

512
 Figure 2: Envelope of load – strain curves in three-point bending test of the repair mortars.

Figure 3: Tensile tests of the repair mortars

Figure 4: Bond strength vs. surface roughness and moisture content of the substrate by slant-shear tests

FOM shows the highest bond strength, in both slant shear and three points bending tests. Slant shear test results
show that wet substrates with dry surfaces gave the highest bond strengths compared with the other moisture condi-
tions. All FOM-to-substrate specimens underwent a monolithic behavior. However, debonding was occurred for all
FLM-to-substrate specimens. Three points bending tests show (Fig. 2) that the bond strengths of FOM-to-substrate
specimens were between the flexural strengths of the two basic mortars (FOM and substrate). This is not true for
FLM-to-substrate specimens, where the bond strengths were lower than the flexural strengths of the basic mortars.
All FOM-to-substrate samples have undergone failure in the substrate or in the repair mortar (cohesive failure), con-
trary to FLM-to-substrate samples where only debonding (adhesive failure) was occurred.
SEM observations of the repair mortars-to-substrate interfacial zones highlight a high micro-cracking in the FLM,
probably due to differential shrinkage between the hardened substrate and the freshly laid plastic overlay. Under an
applied load, these flaws cause stress concentrations and weakens the interface. Low bond strength of the FLM may
also be related to the presence of a film of thickening agent at the interface and the bond depends mainly on
glutinous nature of the thickening agent (molecular force). FOM-to-substrate interfacial zone is more compact and
uniform. Chemically, the silica fume reacts pozzolanically with portlandite to produce a greater solids volume of

513
CSH gel, leading to an additional reduction in capillary porosity. Physically, the silica fume particles fill the weak
spaces of interfacial and transition zones making them denser and more homogeneous. C-S-H type IV (clear gray
crown, Fig. 6) was also observed around the clinker grains in the FLM hardened paste. These hydrates or “inner-
products” appear tardily during the hydration, which indicates than the hydration process will continue longer.

Figure 5: Bond strength according to the surface roughness of the substrate by three points bending tests.

Figure 6: SEM exams of the repair mortars-to-substrate interfacial zone.

4. CONCLUSION
Synthetic fibers and thickening agent would not modify the nature of the cement hydration products. The slow
carbonation of portlandite and the thickening agent slow down the evolution of the FLM mechanical properties.
FOM flexural strength increases until 28 days and decreases after. FLM flexural strength remains increasing with
time. Since the influence of porosity is difficult to verify and a difference in cement hydration cannot explain the
evolution of the materials, it is suggested that the percolation of the thickening agent over the sample increases its
flexural strength. Tensile tests show that fibers and silica fume enhance the tensile strength and strain capacity of
mortars. The tensile strain capacity seems also to be increasing with the thickening agent. Silica fume increases the
mechanical properties of the repair mortars-to-substrate interfacial zone and interfacial transition zones, leading thus
to a better bond. Cracking due to differential shrinkage between the substrate and the FLM weakens the bond
strongly influenced by the moisture content and the surface roughness of the substrate. The bond may also be
dependent on the glutinous nature of the thickening agent. Finally, a repair mortar with Young's modulus close to
that of the substrate is recommended.

4. REFERENCES
Mangat, P.S. and Limbachiya, M.C. (1997), “Repair material properties for effective structural application”, Cement
and Concrete Research, Vol. 27, No. 4, pp 601-617.
Mallat, A. (2006). “Etude des phénomène de dégradation des monuments anciens. Techniques et matériaux de
réhabilitation”, Ph.D. thesis, Ecole Centrale Paris, Châtenay-Malabry, France.
Mallat, A. and Alliche, A. (2006). “Mechanical behavior and bond characterization of fiber-reinforced repair
mortars”, Concrete Solutions 2006, Proceedings of the Second International Conference on Concrete Repair,
Editors: M. G. Grantham, GR Technologie Ltd. London, UK, pp. 568-580.

514
Third International Conference on FRP Composites in Civil Engineering (CICE 2006)
December 13-15 2006, Miami, Florida, USA

SEVERELY DAMAGED URM WALLS RETROFITTED WITH FRP

STRIPS SUBJECTED TO OUT-OF-PLANE LOADING
C. R. Willis
(Postdoctoral Fellow, School of Civil and Environmental Engineering, The University of Adelaide, S.A., Australia)

Q. Yang
(Masters Candidate, School of Civil and Environmental Engineering, The University of Adelaide, S.A., Australia)

R. Seracino
(Senior Lecturer, School of Civil and Environmental Engineering, The University of Adelaide, S.A., Australia)

S. H. Xia
(Lecturer, School of Civil and Environmental Engineering, The University of Adelaide, S.A., Australia)

M. C. Griffith
(Associate Professor, School of Civil and Environmental Engineering, The University of Adelaide, S.A., Australia)

ABSTRACT
Unreinforced masonry (URM) structures comprise a significant proportion of the building stock in many countries
worldwide. However, URM walls do not behave well when subjected to out-of-plane loading, such as that
experienced under seismic events. Consequently, many existing URM structures require some form of retrofit to
comply with existing codes. As part of ongoing research at The University of Adelaide on the behaviour of URM
walls subject to out-of-plane loading, three full-scale walls (with window openings) were tested under reversed-
cyclic loading. The severely damaged walls were subsequently retrofitted using externally bonded fibre-reinforced
polymer (FRP) strips to quantify the increase in strength and ductility relative to the original capacities. This paper
presents the results of the wall tests and observations leading to future work. It was observed that the FRP strips may
act as boundaries such that failure of the wall is governed by the capacity of sub-panels bound by the strips.

KEYWORDS
Fibre-Reinforced Polymer, Externally Bonded, Masonry Walls, Out-of-plane Loading.

1. INTRODUCTION
Recent catastrophic failures of unreinforced masonry (URM) walls during seismic events worldwide (e.g.
Newcastle, Australia in 1989 and Kocaeli, Turkey in 1999) have driven the development of new techniques for
strengthening URM structures. Externally bonded (EB) and near surface mounted (NSM) fibre-reinforced polymer
(FRP) strips have been successfully used to increase the flexural capacity of reinforced concrete structures (Oehlers
and Seracino, 2004). Due to the brittle nature of both masonry and concrete, the debonding mechanisms for
retrofitted URM are expected to be similar to those characteristic of retrofitted RC behaviour. As a result, the use of
FRP retrofitting techniques has been extended to URM in recent times.

Test results reported in the literature have demonstrated the effectiveness of using FRP to strengthen URM walls by
improving the flexural capacity and ductility under out-of-plane loading (Triantafillou, 1998; Velazquez-Dimas and
Ehsani, 2000; Albert et al., 2001; Hamoush et al., 2001; Kuzik et al., 2003; Ghobarah and Galal, 2005). The current
Australian Masonry Code (Standards Australia, 2001) uses the virtual work method for the design and analysis of
URM walls subjected to out-of-plane loading. This design procedure assumes that the initial horizontal crack at wall
mid-height would develop at a low level of applied loading such that the vertical bending moment capacity does not

515
contribute to the ultimate wall capacity. This can lead to overly conservative estimations of strength for walls with
high levels of axial loading. However, the consideration that vertical bending is the weak link in the two-way
bending of URM walls has provided the impetus for the application of vertical FRP strips to strengthen such walls,
which is anticipated to improve the vertical bending capacity and thus the ultimate wall capacity.

2. WALL TESTS
2.1 URM Wall Specimens

As part of ongoing research at The University of Adelaide investigating the out-of-plane behaviour of URM walls,
three full-scale walls (with window openings) were tested under two-way reversed-cyclic loading (Griffith and
Vaculik, 2005). The walls (height, H = 2.5 m and length, L = 4.0 m) each had a 1.0 m × 1.2 m window opening and
two 480 mm long return walls. The failure modes for the wall tests are shown in Figure 1. The severely damaged
walls were subsequently retrofitted with vertical EB CFRP and GFRP strips (of cross-sectional dimensions 1.2 mm
× 77 mm and 2.0 mm × 154 mm, respectively) and tested under two-way monotonic static out-of-plane loading to
determine the increase in strength and ductility compared to the original capacities. A fourth wall (H = 2.5 m ×
L = 2.5 m) was also tested and strengthened using vertical NSM CFRP strips (of cross-sectional dimensions 1.4 mm
× 20 mm). For completeness, the failure pattern is presented in Figure 1 (iv), however due to space limitations,
discussion of this test is beyond the scope of this paper.

V1 V2 V3 V4 V1 V2 V3 V4

750

1000

750

400 1600 650 650 650 400 1600 650 650 650
s s
(i) Test 1 (ii) Test 2
V1 V2 V3 V4 V5 V1 V2

750

1000

750

400 1550 500 500 500 500 289 1922 289

s
(iii) Test 3 (iv) Test 4
s Strip spacing Vi Label for Pre-existing crack New crack
vertical strip pattern pattern
FRP Initiation of Masonry collapse Location of ε max
debonding debonding

Figure 1: Failure Patterns For Wall Tests

516
2.2 Material Properties

Tests were conducted to determine the material properties of the masonry. The mean values of the compressive
strength, fmc, flexural tensile strength of the masonry, fmt, modulus of elasticity of the masonry, Em, and lateral
modulus of rupture of the brick unit, fut, were 16 MPa, 0.61 MPa, 3539 MPa and 3.55 MPa with coefficients of
variation of 0.14, 0.19, 0.41 and 0.27, respectively. The FRP material properties were provided by the manufacturer
(corresponding experimental data is given in brackets). For the CFRP, the modulus of elasticity, E, ultimate strain at
failure, εult, and ultimate stress at failure, ft, were 165 (162) GPa, 1.4 (1.7)% and 2700 (2799) MPa. For the GFRP
plate, the corresponding values were (16) GPa, (1.4)% and (202) MPa. For the adhesive, E = (6.7) GPa and
ft = (13.9) MPa.

2.3 Test Set-up And Instrumentation

All but one of the walls tested were left unrepaired before strengthening. The damaged mortar joints of wall 3 were
repaired to compare the initial stiffness of the strengthened wall and the impact of pre-existing cracks. Prior to
application of the FRP strips, the wall surfaces were prepared by sanding. The FRP strips were applied vertically to
the tension side of each wall as indicated in Figure 1. The walls were simply supported on four sides and out-of-
plane loading was applied using air bags. A displacement transducer was placed at the centre of each wall to
measure the out-of-plane displacement. Several strain gauges were attached along the FRP strips to provide a strain
distribution profile.

For walls 1 to 3, the FRP material type (carbon or glass) and strip spacing, s, were varied while the theoretical value
of the maximum tensile force that can be developed in an FRP strip, FT = AFRP . EFRP . εdb, was kept constant (where
AFRP is the cross-sectional area of the strip, EFRP is the modulus of elasticity of the strip and εdb is the strain at
debonding). The value of FT required for equilibrium was determined by cross-sectional analysis at the position of
maximum moment, assuming masonry crushing and FRP debonding, i.e. a balanced section. The intermediate crack
(IC) debonding strain of the FRP used in the analysis was based on push-pull tests (Yang et al., 2006). From this, the
geometry of the strip was calculated. The values of AFRP, EFRP and εdb for the CFRP were 92.4 mm2 (i.e. 1.2 mm ×
77 mm), 160 GPa and 5000 µε respectively, while for the GFRP the values were 308 mm2 (i.e. 2.0 mm × 154 mm),
20 GPa and 12000 µε. For these combinations of parameters, the corresponding value of FT was 73.92 kN for both
CFRP and GFRP.

3. RESULTS AND CONCLUSIONS

Results of the wall tests are given in Table 1. Two main types of failure occurred, i.e. FRP debonding and masonry
collapse. For wall 3, the ultimate displacement, ∆u, did not occur at the ultimate load, Pu. Figure 2 shows the typical
static load-displacement behaviour of the original unreinforced wall (wall 1), the envelope of its cyclic behaviour
after damage and the static behaviour of the retrofitted wall. The application of vertical FRP strips allowed the
formation of new vertical cracks in the wall segments and as a result, the cracking patterns and failure modes
changed compared to the unreinforced walls (Figure 1) leading to increased load-carrying capacity and a more
ductile response. For example, comparison of the cyclic envelope and the retrofitted wall behaviour at a level of
displacement of approximately 50 mm (Figure 2) indicates that the retrofitted wall has a corresponding lateral load
carrying capacity approximately 2.5 times that of the damaged wall.

Table 1: Test Results

Test H L Retrofitting s Pu ∆u ε max Failure Mode

(m) (m) Technique (mm) (kPa) (mm) (µε)
Control A 2.5 4.0 - - 5.1 25.2 - -
Wall 1 2.5 4.0 CFRP / EB 650 9.7 72.7 4729 Debonding V4
Wall 2 2.5 4.0 GFRP / EB 650 10.5 143.5 12412 Collapse (No FRP Failure)
Wall 3 2.5 4.0 GFRP / EB 500 12.1 119.4 10117 Debonding V5 + Collapse
Control B 2.5 2.5 - - 8.7 26.5 - -
Wall 4 2.5 2.5 CFRP / NSM - 8.2 101.7 5897 Debonding V2

517
For the retrofitted wall tests compared to the unreinforced wall response, the results may be summarised as follows:
• Ultimate load, Pu, increased by 90 to 137% (average of 111%); and,
• Ultimate displacement, ∆u, increased by 188 to 469% (average of 344%).

For the retrofitted wall tests, the following outcomes were observed:
• GFRP is less stiff than CFRP, hence the maximum strain, εmax, and thus the ultimate displacement, ∆u, is
greater (i.e. comparison of tests 1 and 2) which increases the level of energy absorption;
• By increasing the number of strips (i.e. comparison of tests 2 and 3) and thus reducing the strip spacing, s, by
23% while keeping the FRP material type constant, the ultimate load, Pu, increased by 15% and as the rigidity
of the wall was increased, the ultimate deflection, ∆u, decreased by 16%;
• As expected, the maximum strain, εmax, generally occurred at the locations of cracks; and,
• The values of εmax (Table 1) approached εdb, but IC debonding did not appear to be the form of debonding
observed. As this was the basis for the design of the strengthening scheme further work is required.

10

CFRP retrofitted
8 damaged wall
(static behaviour)
Lateral load (kPa)

Original damaged
4 wall (envelope of
Original wall cyclic behaviour)
(static behaviour)
2

0
0 20 40 60 80 100
Central wall displacement (mm)

Figure 2: Typical Load-Displacement Behaviour

4. REFERENCES
Albert, M.L., Elwi, A.E., and Cheng, J.J.R. (2001). “Strengthening of unreinforced masonry walls using FRPs.”
Journal of Composites for Construction, Vol. 5, No. 2, pp. 76-84.
Ghobarah, A., and Galal, K.E.M. (2004). “Out-of-plane strengthening of unreinforced masonry walls with
openings.” Journal of Composites for Construction, Vol. 8, No. 4, pp. 298-305.
Griffith, M.C., and Vaculik, J. (2005). “Flexural strength of unreinforced clay brick masonry walls”, Proceedings of
the 10th Canadian Masonry Symposium, Banff, Alberta.
Hamoush, S.A., McGinley, M.W., Mlakar, P., Scott, D., and Murray, K. (2001). “Out-of-plane strengthening of
masonry walls with reinforced composites.” Journal of Composites for Construction, Vol. 5, No. 3, pp. 139-145.
Kuzik, M.D., Elwi, A.E., and Cheng, J.J.R. (2003). “Cyclic flexure tests of masonry walls reinforced with glass fiber
reinforced polymer sheets.” Journal of Composites for Construction, Vol. 7, No. 1, pp. 20-30.
Oehlers, D.J., and Seracino, R. (2004). Design of FRP and Steel Plated RC Structures: Retrofitting of Beams and
Slabs for Strength, Stiffness and Ductility, Elsevier, Kidlington, Oxford, UK.
Standards Australia (2001). AS 3700-2001: Masonry Structures, Standards Australia, Sydney.
Triantafillou, T.C. (1998). “Strengthening of masonry structures using epoxy-bonded FRP laminates.” Journal of
Composites for Construction, Vol. 2, No. 2, pp. 96-104.
Velazquez-Dimas, J.I., and Ehsani, M.R. (2000). “Modeling out-of-plane behavior of URM walls retrofitted with
fiber composites.” Journal of Composites for Construction, Vol. 4, No. 4, pp. 172-181.
Yang, Q., Willis, C.R., Seracino, R., Xia, S. H., and Griffith, M. C. (2006). “Push-pull tests on FRP retrofitted URM
units.” Proceedings of the 19th Annual Conference on the Mechanics of Structures and Materials, Christchurch, New
Zealand (under review).

518
Third International Conference on FRP Composites in Civil Engineering (CICE 2006)
December 13-15 2006, Miami, Florida, USA

SUPPORTING CONDITIONS EFFECTS IN THE OUT-OF-PLANE

BEHAVIOR OF URM WALLS STRENGTHENED WITH EXTERNALLY
BONDED FRP STRIPS
Ehab Hamed
(Ph.D Student, Technion-Israel Institute of Technology, Haifa, Israel)

Oded Rabinovitch
(Senior Lecturer, Technion-Israel Institute of Technology, Haifa, Israel)

ABSTRACT
The effect of the supporting conditions on the overall and the localized out-of-plane behavior of un-reinforced
masonry (URM) walls strengthened with externally bonded composite materials is analytically investigated. Four
combinations of supporting conditions with different restrictions of the longitudinal elongation and rotation are
examined. These cases reflect some of the possible realistic supporting conditions that may exist in practical
applications. A theoretical model that is based on variational principles, equilibrium equations, and compatibility
requirements between the structural components (masonry, mortar, FRP, adhesive) is used for the investigation. The
effect of cracking of the mortar joints and the interaction of the existing wall with the composite strengthening
system through the adhesive layer are considered. The results reveal the critical influence of the various supporting
conditions on the response of the strengthened wall and shed some light on aspects of the design of the externally
bonded strengthening system.

KEYWORDS
Bonding, FRP, Masonry, Strengthening, Supporting conditions.

1. INTRODUCTION
Externally bonded fiber reinforced polymers (FRP) provide an excellent solution for strengthening and upgrading of
existing unreinforced masonry (URM) walls. The results of experimental investigations conducted on FRP
strengthened URM walls and subjected to out-of-plane loading indicate that the externally bonded strips allow an
increase of up to 10-50 times in the strength of the URM wall, enhance its stiffness, and provide the URM wall with
the ability to resist bending moments beyond the cracking point (Hamilton and Dolan 2001; Albert at al. 2001).
However, while most of the laboratory tests have focused on simply supported masonry walls that are free to rotate
and longitudinally elongate at their edges, in practice, masonry walls are usually built within a surrounding rigid
frame. The realistic supporting conditions provided by the supporting frame partially or totally restrict the rotations
and the longitudinal deformations at the edges of the wall and allow the arching effects to develop (McDowell et al.
1956). These effects significantly increase the stiffness of the wall, give rise to stability issues, and affect the
structural role of the strengthening system. Field applications of the strengthening technique indicated that the
increase in the strength of the wall due to the externally bonded FRP reinforcement is limited to a factor of about 1.4
(compared to a factor of about 10-50 in simply supported laboratory specimens) (Tumialan et al. 2003; Davidson at
al. 2005). A dominant contributor to the large differences between the laboratory and the field results is the
influence of the supporting conditions on the structural behavior of the strengthened wall. Another aspect of the
strengthening and upgrading task is the uncertainty regarding the condition of the existing structure at hand. In
particular, strengthening of masonry walls is conducted under some level of uncertainty with regards to the
supporting conditions of the existing wall. In these cases, the assessment of the influence of the range of possible
supporting conditions is essential for the design of the strengthening system.

In this paper, the effect of the supporting conditions on the out-of-plane bending behavior of URM walls
strengthened with bonded FRP strips is studied. The goals of the paper are to examine the role of the supporting

519
conditions in the structural response of the strengthened wall and to shed some light on the influence of the
uncertainty regarding these conditions on the design of the strengthening system. To achieve this goal, four types of
supporting conditions with different restrictions of the longitudinal elongation and the rotation are investigated. The
investigation uses the theoretical model developed by Hamed and Rabinovitch 2006. This model is based on a one-
way flexural response of the strengthened wall; the first order shear deformations theory for the modeling of the
masonry units, the mortar joints, and the FRP strips; and the 2D elasticity theory for the modeling of the adhesive
layer. The cracking of the mortar joints and the debonding of the FRP strips near the cracked joint are considered.
The four cases are investigated in the numerical study presented next. A summary and conclusions close the paper.

2. NUMERICAL STUDY
Four URM walls strengthened with GFRP strips and subjected to out-of-plane loading are examined. The walls
differ only in their supporting conditions. The geometry of the walls, the strengthening system, the material
properties, the loads, and the four supporting conditions appear in Fig. 1. It is assumed that the FRP reinforcement is
fully bonded through the height of the masonry wall. However, due to the inability of the cracked joints to transfer
shear stresses, it is assumed that debonded regions are formed near the cracked joints. The length of the debonded
regions is estimated as the height of the mortar joint plus twice the thickness of the adhesive layer (see Fig. 1b). It is
also assumed that the debonded interfaces are free of shear stresses. However, in case the debonded interfaces are in
contact, they can transfer compressive out-of-plane normal stresses. The supporting conditions (Fig. 1e) include two
types of simply supported conditions with and without restriction of the longitudinal elongation (cases II and I
respectively), and two types of clamped conditions with and without restriction of the longitudinal elongation (cases
IV and III respectively).
(I) (II) (III) (IV)
170

Cross Section
q=4.2kN/m 2
Adhesive:
240

Ea=1.5 GPa
Ga=0.55 GPa
6.0 FRP:
240

E frp=185.2 GPa
3994mm

194

Gfrp=9.26 GPa
231

Masonry unit:
26
Debonding Emu=22 GPa
240

Gmu=9.17 GPa
10mm
1.27mm Mortar:
E mj=12.4 GPa
240

Gmj=6.2 GPa
170

ft=0.42 MPa
dc =194 (bond strength)
(a) (b) (c) (d) (e)
Figure 1: Geometry, material properties, and loads: (a) Geometry and loads; (b) Cracked mortar joint and
debonded regions; (c) Cross section; (d) Mechanical properties; (e) Supporting conditions.

The distributions of the out-of-plane deflections, the bending moments, and the axial forces in the masonry walls
and the FRP reinforcement under the different supporting conditions appear in Fig. 2. The comparison between the
two simply supported cases (I and II) reveals that the restraint of the longitudinal elongation significantly decreases
the out-of-plane deflection (Fig. 2a). This is due to the development of compressive "arching" forces that increase
the cracking load and provide the wall with the ability to resist bending moment by means of eccentric thrust forces
forming an "arching action" (Fig. 2b). The results also show that the variation of the flexural rigidity of the masonry
wall between the masonry unit section and the cracked joint section influences the distributions of the bending
moments and the axial forces through the height of the wall (see Hamed and Rabinovitch 2006). Note that under
supporting condition (I), almost all mortar joints are cracked. On the other hand, under supporting condition (II),
only the critical joints at midspan are cracked. Due to the inability of the longitudinally free wall to resist bending
moments by means of the arching action, the tensile forces in the FRP and thus the portion of the moment that is

520
carried by means of a tension-compression couple ("composite action") are much higher under condition (I) than
under condition (II) (Fig. 2d). These observations indicate that rupture failure of the FRP reinforcement is more
likely to occur in walls with free elongation edges (Hamilton and Dolan 2001). However, due to the magnified
compressive axial forces in the longitudinally restrained walls, crushing failure of the masonry may control the
behavior of these walls (Tumialan et al. 2003). In slender walls, the thrust forces may also lead to lost of stability
and a snap-through type of failure (Carney and Myers 2005). The notable differences between the behavior of the
wall under conditions (I) and (II) and the uncertainty regarding the ability of the adjacent components to restrain the
longitudinal deformation of the wall imply that the design of the strengthening system has to take into account both
supporting conditions.

The comparison between supporting conditions (III) and (IV) further highlights the influences of the arching action
and the associated cracking pattern on the behavior of the strengthened wall. Fig. 2e shows that the ratio between the
negative moment at the edges and the positive one at midspan is highly affected by the longitudinal supporting
conditions. Under conditions (III), the peak negative moment is smaller than the positive one (in absolute value) due
to cracking of almost all joints. In the fully constrained wall (conditions (IV)), the ratio equals about 2. This is
attributed to the stiffening effect of the arching forces and to the restrained cracking of the joints in conditions (IV)
(see Figs. 2b,c). As a result, the negative moment zone is much smaller in conditions (III) than (IV) (Fig. 2c). This
observation is critical for the design of the strengthening system and implies that strengthening on both sides of the
wall may be needed. Fig. 2d shows that the restriction of the longitudinal elongation yield compressive stresses in
the bonded FRP strips. This effect may lead to buckling/wrinkling of the FRP reinforcement and to total debonding
failure of the strengthening system (Hamed and Rabinovitch 2006).
7 20
(a) (b)
6 I
0
III
5
III -20
4 I
IV
-40
3
-60
2 II
1 -80 II
IV
0 -100
0 0.5 1.0 1.5 2.0 2.5 3.0 3.5 4.0 0 0.5 1.0 1.5 2.0 2.5 3.0 3.5 4.0
10 50
(c) (d)
8 I II 40 I
6
30
4
2 20
III
0 IV III 10 II
-2
0
-4 IV
x[m] x[m]
-6 -10
0 0.5 1.0 1.5 2.0 2.5 3.0 3.5 4.0 0 0.5 1.0 1.5 2.0 2.5 3.0 3.5 4.0
Figure 2: Response of the strengthened wall with different supporting conditions: (a) Out-of-plane
deflections; (b) Axial forces in the masonry; (c) Moments in the masonry; (d) Axial forces in the FRP.

The comparison between all four conditions reveals that the restriction of the elongation plays a more critical role
than the restriction of the rotations. Note that the deflections observed in the longitudinally constrained simply
supported wall (conditions II) are smaller than the ones observed in the longitudinally free clamped wall (conditions
III) (Fig. 2a). Also, due to the cracking of almost all joints in the longitudinally free walls (conditions (I), (III)), the
behavior of these two walls in terms of the bending moments and axial forces is qualitatively similar and not
affected by the restriction of the rotation in case (III) (Figs. 2b-d). These results also imply that the ability to restrict
the elongation of the wall may affect the design and the detailing of the FRP strengthening system.

The use of externally bonded strengthening system is usually characterized by local effects in the form of shear and
normal out-of-plane (peeling) stresses within the adhesive layer. The distribution of the shear stresses through the
height of the wall and the distribution of the peeling stresses near midspan under supporting conditions (I) and (II)

521
appear in Fig. 3 and reveal the development of the stress concentrations near the cracked mortar joints. Due to the
absence of the arching action under supporting conditions (I), the contribution of the strengthening system to the
global moment resistance mechanism is much larger than under supporting conditions (II) (see Fig. 2d). As a result,
the shear and peeling stresses that develop under supporting condition (I) are much higher than the ones observed
under conditions (II). This observation implies that debonding mechanisms of the strengthening system as well as
localized diagonal shear cracks at the edges of the masonry units (Albert et al. 2001), which are both governed by
the stress concentration near the joints, are more likely to occur in walls that are free to elongate.

3 8 2
7 0
2 6
5 -2
1
4 -4
0 3
2 -6
-1 1 -8
-2 0
-10
-1
-3 -2 -12
0.5 1.0 1.5 2.0 2.5 3.0 3.5 1.7 1.75 1.8 1.85 1.9 1.95 2.0 2.05 2.1 1.7 1.75 1.8 1.85 1.9 1.95 2.0 2.05 2.1
Fig. 3: Stresses in the adhesive under supporting conditions (I) and (II): (a) Shear stresses; (b) Peeling stresses
at the adhesive-FRP interface; (c) Peeling stresses at the adhesive-masonry interface.

3. SUMMARY AND CONCLUSIONS

The influence of the supporting conditions on the flexural response of masonry walls strengthened with composite
materials has been investigated. It has been shown that the potential restriction of the longitudinal deformation and
thus the development of the arching action significantly affect the local and global behavior of the wall. As a result,
it plays a major role in the detailing of the strengthening system and in the expected failure mode of the strengthened
wall. These observations imply that in cases of uncertainty regarding the actual supporting conditions of the existing
wall, the design must account for a response envelope obtained using different supporting conditions rather than be
limited to a single characteristic case.

4. ACKNOWLEDGMENT
The financial support for this study by the Ministry of Construction and Housing is gratefully acknowledged.

5. REFERENCES
Albert, M.L., Elwi, A.E. and Roger Cheng, J.J. (2001). "Strengthening of Unreinforced Masonry Walls using FRPs".
Journal of Composite for Construction, 5(2): 76-84.
Carney, P. and Myers, J.J. (2005). "Out-of-Plane Static and Blast Resistance of Unreinforced Masonry Walls
Connections Strengthened with FRP". 7th International Symposium on Fiber Reinforced Polymer (FRP)
Reinforcement for Concrete Structures (FRPRCS-7), Carol K. Shield, John P. Busel, Stephanie L. Walkup, Doug D.
Gremel, pp: 229-248, Kansas, MI..
Davidson, J.S., Fisher, J.W., Hammons, M.I., Porter, J.R., and Dinan, R.J. (2005). "Failure Mechanisms of Polymer-
Reinforced Concrete Masonry Walls Subjected to Blast". Journal of Structural Engineering, 131(8): 1194-1205.
Hamed, E. and Rabinovitch, O. (2006). "Out-of-Plane Behavior of Unreinforced Masonry Walls Strengthened with
FRP Strips ". Accepted for publication in Composites Science and Technology
Hamilton, H.R. and Dolan, C.W. (2001). "Flexural Capacity of Glass FRP Strengthened Concrete Masonry Walls”.
Journal of Composites for Construction, 5(3): 170-178.
McDowell, E.L., Mckee, K.E. and Sevin, E. (1956), “Arching Action Theory of Masonry Walls”. Journal of
Structural Division, 82(ST2): 915-1: 915-18.
Tumialan, J.G., Galati, N. and Nanni, A. (2003). “Field Assessment of Unreinforced Masonry Walls Strengthened
with Fiber Reinforced Polymer Laminates”. Journal of Structural Engineering, 129(8): 1047-1056.

522
Third International Conference on FRP Composites in Civil Engineering (CICE 2006)
December 13-15 2006, Miami, Florida, USA

TWO DIMENSIONAL EVALUATION OF PATCH REPAIRS OF

STRUCTURAL ELEMENTS
Dr. Sameer Hamoush
(Professor,North Carolina A&T State University,Greensboro NC, USA)

Maurice McNeal
(Student,North Carolina A&T State University,Greensboro NC, USA)

Hisham Abdul-Fattah
(Assistant Professor, University of Sharjah, P. O. Box 27272, Sharjah, United Arab Emirates)

ABSTRACT
The recent advances in the field of composite materials make the use of Fiber Reinforced Plastic (FRP) an
effective solution to upgrade and retrofit of deficient structural members. Due to the wide range of
bonding process of the patch to the structural elements, there is a need to evaluate performance of the
patch under load application especially if imperfection occurs during the bonding process. A two
dimensional evaluation of both good and defective patch repairs of structural elements was performed in
this paper. The analysis is carried out on 0.6-inch wide strips extracted from a full scale panel repaired in
accordance with an approved structural engineering standard. The experimental program was extended to
evaluate elements with a perfect bond patch and to patch with internal defect between the patch and the
parent materials. The considered defects were 1-inch wide flaw between the sixth ply and the parent
material. A finite element model was developed to evaluate the stress field within and adjacent to the
patch area. Due to the thin nature of the patch, the developed model is based on a geometric nonlinear
formulation. Experimental results are being used to validate and guide the analytical approach. Results
of the program will be presented to quantify the effectiveness of the patch repair in structural upgrading
applications.

KEYWORDS
Scarf, Two-dimensional, Repair, Tensile, Disbond

1. INTRODUCTION

There are numerous studies related to the field of composite materials and their investigation, maintenance, and
repair. In 1973, Findik et al., [1] reported that engineers proposed the repair of aircraft structures using composite
material patches which would be adhesively bonded over cracks in the metal of the aircraft structures. Analytical
and numerical studies of an isotropic cracked plate repaired with bonded composite patches were conducted by
Tsamasphyros et al., [2]. An investigation into the tensile failure of sandwich joints was carried out by Qian and
Akisanya [3], who established that for a given scarf joint the failure was initiated along the edge of the repair and
was regulated by the free edge fracture toughness. One of the typical failure modes of composite plates is the
interlayer crack and delamination which are investigated by Luo and Hanagud [4]. Schipperen and Lingen [5]
investigated free edge delamination through two-dimensional calculations. Odi and Friend [6] improved upon the
two-dimensional model for composite joints. Found and Friend [7] evaluated two wet lay-up and one prepreg repair
systems. Their investigation was focused on the improvement of repairs which would take place on the structure and
the problems associated with in situ repairs. Ahn and Springer [8] also showed that the prepreg load at failure was
higher than that of wet lay-up repairs in both scarf repairs and stepped lap repairs. Baker et al., [9] demonstrated that
scarf repairs can be used to mend highly strained graphite/epoxy structures by repairing the stabilator of an F/A-18

523
Hornet fighter aircraft. Oztelcan et al., [10] also demonstrated the effectiveness of scarf repair by designing test
coupons for the scarf repair of composite helicopter blades. Hamoush, et al., [11] experimented with the evaluation
of scarf repair of solid ((-60/60/0)s) composites.

2. MATERIALS SELECTION AND SPECIMEN FABRICATION

The material used to construct the panels consists of carbon fibers which are pre-impregnated (prepreg)
with epoxy resin. The Hexcell AS4/3501-6 prepreg is composed of continuous AS4C carbon (graphite) fibers and an
amine-cured epoxy resin system. The orientation in which the pieces of prepreg are cut must match the desired fiber
orientation (-60/60/0/0/60/-60) of the stacked panels.
After curing and trimming the panel the defect area is made using a drill press with a 1” hole cutting saw. A
7” slot was removed along the bottom two thirds of the panel. This circular drilling is a rough cut which will
obviously not leave the investigator with a straight edge. The edges of the slot are smoothed by hand by filing the
extruding corners away carefully until the edge of the 1” boundaries is reached. The ends of the slot are also filed in
order to produce a rectangular (1x 7 inch) slot. Each side of this slot will be scarfed to reveal 0.5” of each successive
layer. A 22,500 revolution per minute (RPM) Chicago Pneumatic die grinder (CP875) is the scarfing tool used for
this process. The first scarfing or sanding disk used is a 1” Powerlock 80 grit sanding disk. This disk is used at an
angle along the cut edge of the slot to carefully expose all six layers. After all layers have been exposed they are
scarfed away from the opening to expose 0.5” of each layer. After the initial exposure with the 80 grit disk, the disk
is changed to a finer 120 grit 1” disk to continue the operation. The bottom -60° ply will be exposed 0.5” from the
edge of the slot. This layer is very thin and a very fine 240 grit 1” disk is used to remove the fifth ply material from
this bottommost ply. The lines previously marked along the perpendicular tape are extended by pencil over the scarf
area to ensure that the proper 0.5” area of each ply is removed. After the scarf was completed the sample was
cleaned and oven dried for the patching process. The patch material is AS4/3501-6 in the same prepreg form as the
scarfed laminate. The patch layers must match the fiber orientations of the layer that it is adhered to. The first layer
of the patch is the Spec BMS 5-154C structural adhesive film which was manufactured by Critical Materials
Incorporated. A 1” 90º filler ply follow the adhesive film and the remaining six layers of prepreg are laid to cover
the 0.5” exposed ply and the fiber angles. The panel is debulked after application of each layer of the patch. Once
the patch has been cured, the panel is ready to be cut into fifteen, 0.6” wide strips which serve as the specimens used
for testing.

3. TESTING

The specimens will be subjected to axial tensile loading using an Instron testing machine. The test is
performed under displacement control of 0.5in/min. The strain values are recorded from the five pristine strips
through the use of an Instron extensiometer. This device is designed to be spring attached to a specimen that is to
undergo testing. The extensiometer calculates how long the material has stretched relative to its original position of
1” between the leads. The values for strain and load are imputed into a data recovery computer system. The
extensiometer system was sufficient for the pristine strips which were expected to experience a constant strain
throughout the length. According to the raw data collected however, there is evidence that the leads on these strain
gages may have begun to slip as the material reached closer the ultimate failure load.
The repaired strips would have to have strain gages attached to the specimen in order to monitor strains
along their axis since the strains for this repaired system are not the same throughout the length. For the good repair
strips, strain gages were adhered to the specimen over the same area which has a through width delamination in the
defective repair strips to enable the comparison of the strain values. A total of two strain gages were used for each
good repair strip; one in the center on the top side of the patch which would be numbered as gage 1 for data
collection and one over the area which corresponds to the area which would have the defect in the defective repair.
In order to perform stress calculations from the strain data collected, the areas of the samples were taken by
measuring the width and thickness of the strips in four areas. The patch area or the middle of the strips, each end,
and the mid distance between the center of the strip and the end of the strips were measured and recorded. The
defective repair strips used a minimum of three strain gages to record the strain values along the length of the
sample. One gage was placed in the center of the top surface of the patch area and two others were placed along the
rear of the engineered defect area. For one strip of the defective repair samples, an additional strain gage was placed
in the pristine area of the sample in order to make a comparison with the pristine strips. The load values at failure
were recorded and the broken strips were inspected to search for similarities in the failure modes. The pristine strips

524
failed explosively with much of the material destroyed during the failure stage. Both types of repair strips failed
similarly with the defective repair strips failing at both through width delamination areas leaving much of the center
of the patch intact. The good repair strips failed at varying locations within the patch area with greater size of broken
pieces than the good repair.
The load values recorded were in accordance with expected failure levels as calculated from the testing
data for the previous test [11] of whole panels and dividing the failure loads recorded there by the cross sectional
area of the 0.6” strips. Examination of the results shows that the pristine strips carried loads closest to the 2000lb.
expected failure load. The highest ultimate load recorded of 2231lb.belonged to pristine strip 4 which also had the
largest cross sectional area of the pristine strips. The width of the pristine strip which had the lowest ultimate load of
1802lb. was also the smallest width of the fifteen tested strips.
The two scarf repaired samples indicate that the repair was a successful process due to the capability of the
repairs to carry over 95% of the pristine failure load. This ability to carry loading is promising to the repair industry.
These static loading conditions may not accurately represent service loading conditions and fatigue loading or
hot/wet loading conditions may be needed to better evaluate the performance of the repair. The engineered defect’s
impact upon the static tension load characteristics of the repairs was minimal due to the location and orientation of
the plies in which it was inserted. The break in the top -60° ply caused a stress concentration in the plies below it
since the same load was supported by a smaller cross section. The 0° fiber orientation plies carry the greatest tensile
force due to the increased stiffness of these two layers. The loss of one -60° ply from the delamination caused a
slight drop in the ultimate load, but the amount of load which would typically carried by this layer is smaller than
the load carried by the center plies and could be more easily distributed to the other layers. These scarf repairs with
imbedded defects were still able to carry an average load of 1998lb. which is still over 95% of the average failure
load of the pristine samples. The failure load was reduced in the scarf repair specimens and further reduced for the
engineered defect repairs.

4. NUMERICAL EVALUATION
The repaired panel was modeled by 8 node elements (see Figure 1). Because of symmetry, one half of the panel with
the appropriate boundary conditions was modeled. All layers of the patch and parent panels and the adhesive layer
were modeled as 2-D elements with their respective material properties. A commercial FE software ANSYS was
used to perform the analysis. Due the geometric nature of the panel, the geometric nonlinearity was used in the
analysis. The model considers each ply separately with the appropriate fiber orientations. The material properties
used is listed in Table 1. A reduction of 5 percent was implemented for the patch properties for both the strength
and stiffness to account for the difference in fabrication process between the parent materials and the patch. The
parent material was fabricated by autoclave while the patch was added with the assistant of the vacuum bag.
Systematic convergence study was made to confirm the convergence of the mesh and results. Analysis was
conducted by incrementing the displacements till the overall panel strain in the x-direction reaches one percent. The
complete displacements, strains, and stresses in the model at various load levels were analyzed. The peel and shear
stresses in the resin layer are shown in Figures 2 and 3.

Table 1: The Used Materials Properties.

AS4/3501-6
E11 142 GPa, E22 = E33 = 8.27 GPa, G12 = G13 = 4.82 GPa, G23 = 2.27 GPa, Poisson’s Ratio in the direction 1-2 = ν12
= 0.33, and Poisson’s Ratio in the direction 2-3 = ν 23 = 0.33
Adhesive Properties Exx= 4.24 GPa and νxy = 0.3650

5. CONCLUSIONS AND RECOMMENDATIONS

The objectives of the present paper are to evaluate the response of the patch repair to tension static loading
conditions and to investigate the performance of defectively repaired panels under unidirectional tensile load. Based
on the test performed in this study, the following can be concluded:
1. It appears that stress concentration is experienced at the connection of the patch and the parent materials.
2 Scarf repairs are an effective retrofitting technique of defected members. It restores 85 to 95 percent
original strength.

525
 Figure 1: Finite Element Model

Figure 2: The Peel Stress at Resin Line Figure 3: The Shear Stress at Resin Line

Acknowledgements
The work in this research is supported by FAA through the Air awareness center of excellence Iowa State university
grant number 4-48453.
4 REFERENCES

Findik F., Mrad N., Johnson A., “Strain Monitoring in Composite Patched Structures”,The Journal of Composite
Structures, Vol. 49, 2000.
Tsamasphyros G. J., Kanderakas G. N., Karalekas D., Rapti D.,.Gdoutos E .E, D. Zacharopoulus, and Z. P. Marioli-
Riga, “Study of Composite Patch Repair by Analytical and Numerical Methods”, Fatigue Fracture Engineering
Material Structures, Vol. 24, 2001.
Qain Z. and Akisanya A. R., “An Experimental Investigation of Failure Initiation in Bonded Joints”, Acta Materials,
Vol. 46, 1998.
Luo H., and Hanagud S., “Delamination Modes in Composite Plates”, Journal of Aerospace Engineering, Vol. 9,
1997.
Schipperen J. H. A. and Lingen F. J., “Validation of Two-dimensional Calculations of Free Edge Delamination in
Laminated Composites”, The Journal of Composite Structures, Vol. 45, 1999.
Odi R. A. and Friend C. M., “An Improved 2D Model for Bonded Composite Joints”, International Journal of
Adhesion and Adhesives, Vol. 24, 2004.
Found M. S. and Friend M. J, “Evaluation of CFRP Panels with Scarf Repair Patches”, Journal of Composite
Structures, Vol. 32, 1995.
Sung-Hoon A. and Springer G., “Repair of Composite Laminates-I: Test Results”, Stanford University Department
of Aeronautics and Astronautics.
Baker A. A., Chester R. J., Hugo G. R., and Radtke T. C., “Scarf Repairs to Highly Strained Graphite/epoxy
Structure”, International Journal of Adhesion and Adhesives, Vol. 19, 1999.
Oztelcan C., Ochoa O. O., Martin J., and K. Sem, “Design and Analysis of Test Coupons for Composite Blade
Repairs”, Journal of Composite Structures, Vol. 37, 1997.
Hamoush S., Shivakumar K., Darwish F. H. and Sharpe M., “Defective Repairs of Laminated Solid Composites”,
Department of Civil and Arch Engineering, Department of Mechanical Engineering, Center for Composite Materials
Research, North Carolina A&T State University

526
Part XIX. Retrofit of Slabs
Third International Conference on FRP Composites in Civil Engineering (CICE 2006)
December 13-15 2006, Miami, Florida, USA

ANALYSIS OF DEBONDING FAILURES IN FRP-STRENGTHENED

CONCRETE BEAMS AND SLABS
K.W. Neale
(Professor, University of Sherbrooke, Sherbrooke, QC, Canada )

U.A. Ebead
(Research Associate, University of Sherbrooke, Sherbrooke, QC, Canada)

W. Elsayed, H. Abdel Baky, and A. Godat

(PhD Students, University of Sherbrooke, Sherbrooke, QC, Canada)

ABSTRACT
This paper deals with the nonlinear finite element analysis of the load−deformation behaviour of reinforced concrete
beams and slabs strengthened with externally bonded fibre reinforced polymers (FRPs). The particular focus is on
the implementation of appropriate constitutive models for the FRP/concrete interface that are capable of properly
simulating interfacial stresses and strains, as well as characterizing possible debonding failures. Various applications
are considered including the flexural and shear strengthening of beams and slabs. The proposed numerical models
are validated against experimental data. The numerical analyses are shown to provide useful insight into phenomena
that are virtually impossible to observe experimentally.

KEYWORDS
FRP, Strengthening, Reinforced concrete, Slabs, Beams, Flexure, Shear

1. INTRODUCTION
In this paper, we review some of our research work on the finite element simulations of FRP strengthened concrete
beams and slabs in Neale et al. (2006), Abdel Baky et al. (2006), Godat et al. (2006), and Elsayed et al. (2006). For
the beam applications, the finite element models address the FRP/concrete interfacial responses and are capable of
simulating the various failure modes, including FRP debonding at either the plate end or at intermediate cracks.
With regard to slab applications, various FRP configurations, with both passive as well as prestressed FRP
strengthening, are considered. We present typical results in terms of load–deflection relationships, ultimate load
capacities, and interfacial slip and stress distributions. To accurately predict the ultimate load carrying capacities and
capture the debonding failures, it is necessary to properly model the bond behaviour at the FRP/concrete interface.
For this purpose, interface elements that are able to properly represent the local shear−slip characteristics and failure
are utilized.

2. NONLINEAR FINITE ELEMENT MODELS

The numerical analyses are carried out using the finite element software package ADINA (2004a). This allows us to
simulate the entire nonlinear load−deformation behaviour of the structure under displacement-controlled loading
conditions. In the analyses, the software formulations for the concrete, steel and FRP are employed. These are
described in detail in the ADINA theory and modelling guide (ADINA, 2004b), and are briefly summarized below.
To model the FRP/concrete interface, we introduce appropriate bond stress−slip relations via truss elements for the
direct shear and flexurally-strengthened beam applications, and spring elements for the slabs and shear-strengthened
beam applications. Both truss and spring elements are used to link the FRP laminates to the concrete.

527
The constitutive law used to model the concrete has the following features (ADINA, 2004b): (i) a nonlinear
stress−strain relation to allow for the weakening of the material under increasing compressive stresses, (ii) failure
envelopes that define both failure in tension and crushing in compression, and (iii) a strategy to model the post-
cracking and post-crushing behaviour of the material. The general multiaxial stress−strain relations are derived from
a nonlinear uniaxial stress−strain relation. The cracked concrete is assumed to be orthotropic, with the directions of
orthotropy being defined by the principal stress directions. Failure envelopes are used to establish the uniaxial
stress−strain law accounting for multiaxial stress conditions, and to identify whether tensile or crushing failure of the
concrete has occurred. The post failure material behaviours account for post-tensile cracking, post-compression
crushing, and strain softening.

For the steel reinforcement, a uniaxial elastic-plastic stress−strain law is employed. A linear elastic orthotropic
constitutive relation is adopted for the FRP composites. The mechanical behaviour of the FRP/concrete interface is
represented by a relationship between the local shear stress, τ, and the relative displacement, s, between the FRP
laminate and the concrete. The area under the τ−s curve represents the interfacial fracture energy, Gf, which
corresponds to the energy per unit bond area required for complete debonding of the laminate. The details of the
formulations of the bond−slip model can be found in (Lu et al., 2005; Neale et al., 2006).

In the characterization of the interfacial behaviour, the FRP nodes are connected to the concrete nodes using
nonlinear translational interface elements, as shown in Figure 1. It is necessary to emphasize that these elements
represent the overall interfacial behaviour between the concrete and the FRP, and not the adhesive as such. The
aforementioned constitutive model (the bond–slip model) thus represents the overall contribution of the FRP
composites, adhesive and concrete. The relative displacement between Point 1 and Point 2 of the interface element
(Figure 1) represents the interfacial slip, and the stress in the two-node interface element represents the interfacial
shear stress. Full strain compatibility is assumed between the FRP nodes and the concrete nodes in the peeling-off
direction by enforcing suitable constraint equations.

Point 1 (master)- concrete node

2- node truss element
Point 1 (slave)
Point 2 (master)

Point 2 (slave)-FRP node

Figure 1: Interface element

For the flexurally-strengthened beam application, plane stress elements are used to simulate the concrete. For the
shear-strengthened beams and the strengthened slabs, 3-D analyses are employed. Due to the geometrical and the
loading symmetry; only one half of the beam or one quarter of the slab is modelled. Detailed descriptions of the
geometrical parameters of the beams and the slabs are given in Neale et al. (2006), Abdel Baky et al. (2006), Godat
et al. (2006), and Elsayed et al. (2006).

3. NUMERICAL RESULTS AND DISCUSSION

Typical results are presented in the subsequent sections in terms of the ultimate load carrying capacities and load–
deflection relationships for the different applications simulated in this study. Special emphasis is placed on the
results of the interfacial behaviour between the FRP laminates and the concrete in terms of the interfacial stress
distributions and slip profiles. The specimen notations here correspond to those employed in the original references.

Ultimate load carrying capacities and load–deflection relationship

The experimental results of 25 flexurally-strengthened beams were used to assess the validity of the finite element
model (Abdel Baky, et al., 2006). In addition, 15 shear-strengthened beams are considered (Godat et al., 2006).
There is a very good agreement between the numerical predictions of the ultimate load capacities and the
experimental data. The average numerical-to-experimental load capacity ratio for the flexurally-strengthened beam
applications is 100.3% with a standard deviation of 6.5%. The corresponding values of the average and the standard

528
deviation in the case of the shear-strengthened beams are 102% and 2.31%, respectively, thus indicating an excellent
agreement. With regard to the slab application, we obtained a very good agreement when comparing the numerical
results with the experimental data of 12 different specimens with an average numerical-to-experimental load
capacity ratio of 97% with a standard deviation of 7.3% (Elsayed, et al., 2006). Detailed numerical-to-experimental
comparisons can be found in Abdel Baky et al. (2006), Godat, et al. (2006), and Elsayed et al. (2006).

The proposed models are able to simulate the entire load–deflection relationships, including the descending and post
failure profiles, in view of the displacement-controlled solution adopted in these analyses. Figures 2a and 2b show
typical numerical versus experimental comparisons in terms of the load–deflection relationships for beam and slab
specimens (M’Bazaa, 1995; Longworth et al., 2004). As seen here, we were able not only to capture the debonding
load, but also the complete post-debonding plateau until complete failure.

120 250

100 Debonding 200

80

Load (kN)
150
Load (kN)

60
100
B2-SL1-exp.
40 B2-SL1-num.
Control-Exp.
Control-Num. 50
B2-SL4-exp
20 Po-Exp.
Po-Num. B2-SL4-num
0
0
0 20 40 60 80 100
0 20 40 60 80 100
Deflection (mm) Deflection (mm)

a. M’Bazaa (1995) b. Longworth et al. (2004)

Figure 2: Typical load–deflection relationships for selected specimens

Interfacial shear stresses and slip profiles

A distinct advantage of having reliable numerical tools is that they can provide valuable insight into phenomena that
are very difficult to assess experimentally. For example, knowing the values of the interfacial stresses and slips
between the bonded FRPs and concrete can be very helpful for a better understanding of the FRP/concrete interfacial
behaviour and bond performance. These observations are virtually impossible to detect in experiments. The
numerical results shown in Figures 3a and 3b represent the slip profiles along the beam depth for an FRP side-
bonded beam tested by Pellegrino and Modena (2002) and a U-shaped bonded beam tested by Adhikary and
Mutsuyoshi (2004). The slip profiles depicted in these figures correspond to load increments up to failure for a
section midway the shear span. As seen in Figure 3a for the side-bonded beam, the predicted slip values are
significantly higher at the bottom of the beam than those at the top; this is due to the tensile stresses in this region.
This suggests that this specimen should experience debonding of the FRP laminate at the bottom edge of the beam,
which is in accordance with the experimental results. For the specimen strengthened using U-shaped FRP wraps the
interfacial slips are higher at the top edge and shift to negative values around the mid-depth. With an increase of the
applied shear force up to failure, the interfacial positive and negative values of slip are increased. It is thus obvious
that using U-shaped FRP wraps rather than side-bonded laminates is more efficient in mitigating debonding, as seen
from comparing the corresponding slip values.

4. CONCLUSION

A review has been presented on finite element analyses to address the interfacial behaviour of FRP-strengthened
reinforced concrete structures. A nonlinear constitutive model was incorporated to represent the interfacial
behaviour between the bonded FRP laminates and concrete substrate. In order to investigate the validity of the
numerical models, theoretical predictions have been calibrated against published experimental data. The
comparisons between the numerical and experimental results showed very good correlations in terms of the ultimate
carrying capacities and load–deflection relationships. Our studies have shown the importance of appropriately
modelling the FRP/concrete interface if accurate predictions of the behaviour of externally FRP-strengthened

529
members are to be obtained. These studies has also demonstrated that reliable numerical models represent very
valuable tools for gaining insight into phenomena that are extremely difficult to assess experimentally.

300 200
188 kN
291 kN
150

Beam depth (mm)

Beam depth (mm)

332 kN
200
410 kN
59 kN
100
109 kN
100 150 kN
50 166 kN

0 0
0.0 0.4 0.8 1.2 1.6 -0.4 0.0 0.4
Slip (mm) Slip (mm)

a. Side-bonded (Pellegrino and Modena 2002) b. U-wraps (Adhikary and Mutsuyoshi 2004)

Figure 3: Typical slip profiles for FRP shear-strengthened beams

5. ACKNOWLEDGEMENTS

This research was funded by the Natural Sciences and Engineering Research Council of Canada (NSERC), and the
Canadian Network of Centres of Excellence on Intelligent Sensing for Innovative Structures (ISIS Canada). KWN is
Canada Research Chair in Advanced Engineered Material Systems and the support of this program is gratefully
acknowledged.

6. REFERENCES
Abdel Baky, H., Ebead, U.A. and Neale, K.W. (2006). “Flexural and interfacial behaviour of FRP-strengthened
reinforced concrete beams”. Journal of Composites for Construction, ASCE, submitted.
Adhikary, B. and Mutsuyoshi, H. (2004). “Behaviour of concrete beams strengthened in shear with carbon-fibre
sheets”. Journal of Composites for Construction, ASCE, Vol. 8, No. 3, pp. 258-264.
ADINA (2004a). Automatic Dynamic Incremental Nonlinear Analysis, Finite Element Software, Version 8.2.
ADINA R&D Inc., Watertown, MA, USA.
ADINA (2004b). Theory and Modeling Guide, Volume I, Chapter 3, Version 8.2. ADINA R&D Inc., Watertown,
MA, USA.
Elsayed, W., Ebead, U.A. and Neale, K.W. (2006). “Interfacial behaviour and debonding failures in FRP-
strengthened concrete slabs”. Journal of Composites for Construction, ASCE, submitted.
Godat, A., Neale, K.W., and Labossière, P. (2006) “Numerical modelling of FRP shear-strengthened reinforced
concrete beams”. Journal of Composites for Construction, ASCE, submitted.
Longworth, J., Bizindavyi, L., Wight, R.G. and Erki, A. (2004). “Prestressed CFRP sheets for strengthening two-
way slabs in flexure”. Advanced Composite Materials in Bridges and Structures, El-Badry, M. and Dunaszegi, L.,
Eds., Canadian Society for Civil Engineering, 8 p.
Lu, X. Z., Teng, J. G., Ye, L. P. and Jiang, J. J. (2005). “Bond−slip models for sheets/plates bonded to concrete”.
Engineering Structures, Vol. 27, pp. 920-937.
M’Bazaa, I. (1995). “Renforcement en flexion de poutres en béton armé a l’aide de lamelles en matériaux
composites: optimisation de la longueur des lamelles”. Mémoire de maîtrise, Département de génie civil, Université
de Sherbrooke, Canada.
Neale, K.W., Ebead, U.A., Abdel Baky, H., Elsayed, W., Godat, A. (2006). “Towards understanding the load–
deformation behaviour and debonding for FRP-strengthened concrete structures”. Advances in Structural
Engineering, Special Issue, submitted.
Pellegrino, C. and Modena, C. (2002). ‘‘Fibre reinforced polymer shear strengthening of reinforced concrete beams
with transverse steel reinforcement’’. Journal of Composites for Construction, 6(2), 104-111.

530
Third International Conference on FRP Composites in Civil Engineering (CICE 2006)
December 13-15 2006, Miami, Florida, USA

AUUSTRALIAN GENERIC RETROFITTING GUIDELINE FOR

PLATING RC BEAMS and SLABS
Deric John Oehlers
(Associate Professor, The University of Adelaide, Adelaide, South Australia, Australia)

ABSTRACT
Strengthening reinforced concrete beams and slabs using adhesively bonded or bolted plates is well established
because the fundamental plate debonding and failure mechanisms are now generally understood and recognised in
most published guidelines. Furthermore, research is well advanced in quantifying these failure mechanisms, so that
plating for strength can be applied in practice and generally with confidence. The aim of the Australian guideline, as
with established generic national standards such as those for reinforced concrete design, is to provide engineers with
comprehensive and, in particular, generic design tools that cover all forms of plating, that will allow them to find
their own best solutions, to develop their own plating techniques and to encourage current and future developments
in this new and rapidly developing retrofitting technique. The generic nature of this guideline and philosophy behind
this guideline is described in this paper as it is felt that a generic approach is needed if FRP plating is to reach the
same status and application as other structural forms.

KEYWORDS
FRP Guideline, FRP, retrofiiting, reinforced concrete, adhesive bonding, bolting.

1. INTRODUCTION AND SCOPE OF GUIDELINE

The philosophy behind the Australian Guideline (Oehlers et al 2006) is to provide the choice of a wide variety of
plating techniques, and their associate design rules, that covers all forms of failure, in order to allow engineers the
freedom to find their own solutions; furthermore, to develop a generic approach to encourage the development of
new plating techniques. For example, the Australian Guideline covers externally bonded (EB) plates that can be
adhesively bonded to any surface of an RC member as in Fig.1. This form of retrofitting is now well established,
however, it is known to be prone to brittle failure at low strains and, hence has a limited application even though
cost effective. If this approach is found by the engineer to be unsuitable, then design rules for near surface mounted
(NSM) plates are provided where the plate is adhesively bonded within the concrete cover as it is now known that
much larger strains can be developed, hence, greater ductility. Finally if this approach is also found to be unsuitable,
then FRP plates can be bolted to the RC beam and the bolt extended well beyond the cover into the confined region
of the concrete to form a strong and ductile plating technique, although, this approach can be expensive.
guideline applies to
plates on any surface

bolts bolted plate anchored

adhesively bonded near within confined region
surface mounted (NSM)
plates within concrete adhesively bonded
cover externally bonded (EB)
plates on surface

Figure 1: Forms of plating

The Australian plating guideline “Design Guideline for RC Structures Retrofitted with FRP and Metal Plates: beams
and slabs”: first identifies the generic deformations and stress resultants that can induce failure and which are
applicable to all forms of plating; then generic failure mechanisms are developed from these generic stress
resultants; and then, where possible, generic design rules are given. Hence, much of the Guideline is applicable to
new techniques and not just the established techniques in Fig.1 It is hoped that this approach will encourage the

531
development of new techniques and, hence, further increase the application of FRP plating. In summary, the
Australian Guideline covers: the techniques of either adhesive bonding or bolting plates; any type of plate material;
prestressed or unprestressed beams; plates on any surface of the structural member; generic stress resultants and
capacities that have to be designed for; generic debonding mechanisms in adhesively bonded EB and NSM plates;
generic debonding mechanisms in mechanically fastened that is bolted plates; longitudinal shear capacities; plate
buckling; strength design philosophies for longitudinal plates; and generic ductility principles.

2. GENERIC STRESS RESULTANTS

Of fundamental importance to the Australian Guideline is the identification of the stress resultants within a plated
beam as it is these stress resultants that cause the design failure or premature failure. The flexural analysis of a side
plate beam is shown in Fig.2 where full details are given elsewhere (Oehlers and Seracino 2004). Of importance are:
the various failure strains shown in Fig.2(b) that have to be catered for; the fact that part of the concrete may remain
elastic at failure as in Fig.2(c); and, very importantly, that the plate is subject to a moment and axial force in
Fig.2(d). Hence, all longitudinal plates are subjected to an axial force Pplate and moment Mplate, no matter what the
position of the longitudinal plate, that have to be resisted by the bond between the plate and the RC beam. For
example, Mplate is resisted by transverse forces in the bond, Vplate in Fig.3, which applies to both adhesive bonds and
bolted bonds and which should not be ignored in a generic design approach even though they may be shown to be
minor under certain circumstances such as for thin FRP plates on the tension face.
possible
pivotal strains plastic
σ (stress) Plate stress
point of position of point of
resultants
zone contraflexure Vplate maximum moment contraflexure
εconc side plate continuous T-beam
elastic
zone Mplate
εrebar
Pplate Pplate Pplate
εdebond εfracture (c) (d)
(a) (b)

Figure 2: Flexural Analysis of Side Plated Beam Figure 3: Plate Debonding Forces

The shear deformation that causes failure in an RC beam is the formation of a critical diagonal crack (CDC) in
Fig.4. Shear is transferred by aggregate interlock across the CDC, that requires normal forces to the CDC provided
by the reinforcement. It is the rigid body deformation across the CDC that controls the concrete component Vc of the
shear capacity of the RC beam and this form of crack and deformation should not be confused with inclined
flexural/shear cracks. It is common practice in RC design to assume that the longitudinal reinforcing bars contribute
to Vc, hence for this reason it can be assumed that longitudinal plates as in Figs.1 and 3 affect Vc and that the
horizontal component of the plate force (Pplate)h in Fig.4 also contributes to Vc. Furthermore, it is common practice to
assume that the internal steel stirrups resist the shear force directly Vs and, hence, transverse plates resist the shear
directly or the vertical component (Pplate)v of inclined plates.

3. GENERIC DEBONDING MECHANISMS IN ADHESIVELY BONDED PLATES

The generic deformations and associated stress resultants described previously have to be resisted by the adhesive
bond (CNR 2005). Any crack of width w as in Fig.5 that intercepts a plate induces additional interface shear due to
the stress concentration about the crack. This is often referred to as IC debonding and the shear/slip (τ/δ) of this
partial-interaction interface bond has been well researched (Teng et al 2002). When the crack w in Fig.5 is induced
and widened by flexure, then the flexural IC debonding resistance (FIC) of the plate must be able to resist Pplate in
Figs.2and 3. When the crack width w in Fig.6 is widened through shear deformations and aggregate interlock, then
the force in the plate is still controlled by the IC debonding resistance but in this case it is referred to as CDC
debonding as it affects Vc. For vertical plates crossing a CDC crack, the force in the plate is also controlled by the IC
debonding resistance and this is referred to as SIC in the Australian Guideline. It can be seen that three distinct
debonding failures (CNR 2005) that depend on the IC debonding resistance, that is FIC, CDC and SIC, have been
identified.

The transverse forces Vplate in Fig.3 induce debonding from the stress concentration at the plate end as in Fig.7 and
this is referred to as plate end (PE) debonding. This form of debonding starts at the plate end and propagates inwards

532
in contrast to IC debonding and, hence, is totally different from IC debonding and is not dependent on the τ/δ
characteristics but on the tensile strength of the concrete. This form of debonding should not be confused with the IC
debonding of short plates where debonding starts at the position of maximum plate strain and not at the position of
zero plate strain as occurs in PE debonding. The Australian Guideline also recognises that away from stress
concentrations the elementary interface shear VAY/Ib still exists and depends on the tensile strength of the concrete.

(Pplate)2 internal intermediate crack

inclined steel stirrup
plate or CDC w2
(Pplate)2 IC interface crack
inclined(P )
plate 1
fibres (Pplate)v
τ w1
fibre/plate
(Pplate)h orientation
δ w
τ τ
δ Pplate

Figure 4: Shear contribution from inclined plates Figure 5: Intermediate Crack Debonding
IC interface crack critical diagonal crack

direction of debonding
crack propagation PE debonding crack
Pplate w Pplate

Nplate Nplate
interface shear τ
rigid body displacement τ
δ Pplate Mplate

Figure 6: Critical Diagonal Crack Debonding Figure 7: Plate End Debonding

4. GENERIC DEBONDING MECHANISMS IN BOLTEDPLATES

The Australian Guideline gives comprehensive design rules for the bolt shear connectors that are based on extensive
research on the behaviour of stud shear connectors in composite steel and concrete beams (Oehlers and Bradford
1995). The transverse forces Vplate and axial forces Pplate in Fig.3 are resisted directly by the bolts either through their
dowel or shear resistance or through their axial or embedment resistance depending on the configuration of the plate
being bolted. The bolt dowel forces, that resist the longitudinal force Pplate in Fig.3, apply concentrated forces to the
plate as well as to the concrete beam shown as D in Fig.8. The dispersal of these concentrated forces induce both
lateral tensile and compressive forces P. These stresses can cause the plate or concrete elements to split and have to
be designed against. Furthermore, the lateral force Vplate in Fig.3 may be resisted by the RC beam in regions where
there are already flexural cracks as in Fig.9, in which case design rules are given for the post-splitting resistance.

5. LONGITUDINAL SHEAR PLANE CAPACITIES AND PLATE BUCKLING

Combinations of shear connections can also cause the concrete element to fail along shear planes that surround the
shear connection and this applies to both adhesively bonded plates as well as bolted plates as in Fig.10. Failure can
occur along shear planes that surround individual or groups of shear connectors as shown and it is necessary to find
the weakest failure plane by considering all possible failure planes. This form of failure also occurs in beams with
transverse plates. Adhesively bonded plates and bolted plates can also buckle as in Fig.11and design rules are given
in the Australian Standard for the buckling resistances for both FRP and steel plates that are either adhesively
bonded or bolted.

6. PHILOSOPHY BEHIND GUIDELINE

Plated RC structures are a new and unique form of structure which has similar failure mechanisms or behaviours as
in both RC structures and composite steel and concrete structures. However plated structures also have many new

533
failure mechanisms that are not covered in RC and composite design manuals. As with all new forms of structures
and because plating is a very efficient retrofitting technique, plating is being applied concurrently with the
development of design rules. Hence, it is not possible at this stage of development of this new and unique technique
to formulate prescriptive design rules that cover all situations. This should not hinder the application of plating but it
does require a deep understanding of the behaviour of plated structures to ensure a safe design which consequently
requires a deep understanding of the behaviour of both RC structures and composite steel and concrete structures.
The Australian Guideline consists of a Guideline and a Commentary. The Guideline covers the generic and
fundamental behaviour of both plated beams and plated slabs and it is these behaviours that have to be designed for.
The more advanced design rules that quantify the generic fundamental behaviours in the Guideline are given in the
Commentary. Hence the designer needs to be aware and design for the generic behaviours described in the
Guideline. However, the Commentary is only meant to assist in the design and the designer is free to use any
approach that satisfies the Guideline. It is recognised that these rules are improving and developing rapidly. In the
long run, it is the intention of the committee to gradually transfer information from the Commentary to the Guideline
as design rules not only become established but more importantly are proven to be correct and safe under all design
circumstances. The Guideline requires fundamental structural mechanics principles to be adhered to. It is recognised
that design rules are improving and developing rapidly. In the long run, it is the intention to gradually transfer
information from the Commentary to the Guideline as design rules become established.

tension lateral stress distribution

P P splitting
P P
zone
Vplate
D D

direction of thrust bolt shear

along longitudinal connector
axis
compression flexural crack tension face

Figure 8: Longitudinal Splitting Figure 9: Transverse Post-Splitting Resistance

shear planes transverse

steel
near
surface
mounted bolt
plates shear
connector

Figure 10: Shear Plane Perimeters Figure 11: Plate Buckling

7. REFERENCES
CNR (2005) Instructions for the design, execution and control of strengthening measures through fibre-reinforced
composites. CNR-DT 200/04, Italian Research Council, Rome, Italy
Teng, J.G., Chen, J.F., Smith, S.T. and Lam, L. (2002). FRP Strengthened RC Structures. John Wiley and Sons Ltd.
England.
Oehlers, D.J., Seracino, R., and Smith, S (2006) Design Guideline for RC structures retrofitted with FRP and metal
plates: beams and slabs. SAI Global Limited Australia, Standards Australia, In press.
Oehlers, D.J. and Seracino, R. (2004) Design of FRP and Steel Plated RC Structures: retrofitting beams and slabs
for strength, stiffness and ductility. Elsevier England.
Oehlers, D. J. and Bradford, M. A. (1995) Composite Steel and Concrete Structural Members: Fundamental
Behaviour. Pergamon Press, England

534
 !"#!$ %& % % '()

* %+ ,)+)& -% . /% .)% '()

% - ( & % / ( ' /% ' / 0% )% '()

! "
# $ %&' ! " (
( ) ( "
( "
( "
! ! (
" ( # $ **+ & +,
" ( !

(! ! !- "! -

!" #! " #!$ % ! " &' (

# ( .&///0! "
1

 
1=2 + 
2
β0
2
2
.&0

( ! ( ! "
!β
!
! (. ( 0 0
!β
&3 + ++%& ! . ! &///0 ) & (
( "

4 ! # $ %&' &/// (
(! 5 . 0!
( ( " ( !
6 ( "7 5 ( ! (
) # $ %&'8 +9

535
  2'+   2'+ 
= %'+  −2 9
 ≤ %++  

   
. 8 0 .20

: ;
) &. ) 20 (! ( ) . ) %0
" !1

 1 
= 2   −
2


2

 
.%0
2 β 0

# ( # $ %&'8 +9 " ) 2 ! .&///0 ) 2

" +* .+ +&, 0 + 92 .+ +2 0 <
%+= $ ! + ,> /
+, / # $ %&' # $ %&'
( .+ * 0.+ ,>?+ ,0 @ + ** .+ +&' 0

( & ) % + ** + 9'
( ) 2 ) % (

#) * + , -. !" /#(# !(

0 !" #! #!$ % ! " &' (

$ # $ **+ & +,! " " ( ) &

) +9 .+ +2+ 0 +> .+ +2' 0! ! "
!0 ) !0
!0 ( $ 0 ! &*
# $ **+ & +, A( " B (
.2+++0 (

<
" (

536
( ) 2! ( " " # $ %&'8 +9! (
(

0
C ( ) 2 ( ! ( "
1

 2'+   1  &  2'+   1  &

= %'+     −2 9 ≤ %++    
   + **  0    + **  0
.*0
2++! +++ 2++! +++

8 1 5 " 2++!+++ 8 + ** ! "!

( $ ! ) *

1 1
=& 2 −2 9 ≤ + /9 . 8 0 .90
0 0

" " .2++&0 " "

( ! (1@+> ) 9! " (!

=+' −2 9 ≤+> . 8 0 .,0

0 0

( 2 ( ) % 9 6
@ *+ 6 ! @ '+ 8 ! 0 @ & *! ( +> + /& $ (
1 @ +> ! β 1 β = & + + ++,% . ! @ 2%+
0! β = & + + +++' . ! @ &'++ 0! β = & + + ++%& ! "
β ( ) >! (0@+%

β = & + + +++'

β = & + + ++%&

β = & + + ++,%

#) 0* 1 ( % !" /#(# !( $ 2 #!$ % ! " &' (

537
β = &+
(& − 0 )
.>0

β " ! ! " ( "

( 1 @ + /& β = & + + ++%& "

( 2 ! "! " β (
A ! "! 4 β
$ (
" β = & + + ++%& ( 2 ) 9
) %

+
( ! ) 9 "
! 0 (" $ !
! " " (

< "
( ) 9 "
! ) # $ %&'8 +9 "
" +* !
( < ( )
9 ( ( ( "
! " 6
" " (

( # $ **+ & +, ! ) 9
) %
C ( " ( $ !# $
**+ & +, "

3
( "
(
( ) " (
"

.
# $ %&'! .2++90! :8 ( ) " .# $
%&'8 +90!; # $ ! ( A !8 ! *%,

# $ **+! .2++,0! :6 B ( -
.# $ **+ & +,0!; # $ ! ( A ! 8 ! **

- # ! .2+++0! : A( " ( B ( . #D? -# -, ++0!; -#

$ ! !

! E ! .&///0! :# 7 ( !; ) ( 2 2
3 2 ! F /,! D %! 8 " E ! *%> **2

! E ! .2++&0 : !; 2
& 0 ! - 2+*! # $ ! ( A !8 ! &%9 &9*

538
Third International Conference on FRP Composites in Civil Engineering (CICE 2006)
T T

December 13-15 2006, Miami, Florida, USA

STRENGTHENING OF PRECAST PRESTRESSED HOLLOW CORE

SLABS WITH CFRP SHEETS
M. Kalimur Rahman
(Research Engineer, Research Institute, K.F.U.P.M, Dhahran, Saudi Arabia)

A. H. Al-Gadhib
(Associate Professor, Civil Engineering, K.F.U.P.M, Dhahran, Saudi Arabia)

Jaffer Mohiuddin
(Graduate Student, K.F.U.P.M, Dhahran, Saudi Arabia)

M. H. Baluch
(Professor, Civil Engineering, K.F.U.P.M, Dhahran, Saudi Arabia)

ABSTRACT
Precast prestressed hollow core slabs are very commonly used as floor system in Saudi Arabia in the booming
precast concrete construction industry. Strengthening of these slabs may be required for several reasons including
increased load due to higher dead or live loads, architectural modifications in locations of walls, damage due to
corrosion, installation of heavy machinery, opening cut through slabs and errors in planning, construction and
design. This paper presents the results of flexural strengthening of hollow core slabs using CFRP sheets. Seven full-
scale hollow core slabs (5.0 m span, 1.2 m width and 0.2 m thickness) were tested under flexure using four point
loading up to failure, with one slab being tested as control specimen and the remaining six slabs strengthened with
CFRP. Experimental results show that the flexural load carrying capacity of precast prestressed hollow core slabs
increased in the range of 14 to 40% depending on the width/layers of CFRP sheets used. Interesting transition in the
behavior of the hollow core panels from flexure to shear failure was noted as the number of layers of CFRP sheets
was increased. A good correlation was observed between the experimental and the predicted values obtained using
a strain compatibility approach.

KEYWORDS
Strengthening, CFRP sheets, prestressed slab, hollow core slab, full-scale test

1. INTRODUCTION
There has been a surge in the precast concrete construction of commercial and residential structures in the recent
years in Saudi Arabia. Precast prestressed hollow core (PPHC) slabs are primarily used as floor or roof system in the
precast construction. Prestressed hollow core slabs also have application as spandrel members and bridge deck units
and as wall panels. Strengthening of PPHC slabs may be needed to upgrade the load-carrying capacity, changes in
locations of walls, damage due to corrosion, installation of heavy machinery, opening a cut through slabs for
heating/ventilation units and errors due to insufficient design and prestressing steel.

CFRP composite sheets are now being used for strengthening, repair and extension of service life of reinforced
concrete elements including beams, columns, one-way and two-way slabs and wall panels (Bakis et. al. 2002, Nanni,
1995, El-Maaddawy et al., 2005). CFRP has excellent properties of high tensile strength, lightweight, and corrosion
resistance which make it amenable for retrofitting of structures. The ultimate flexural strength of reinforced concrete
beams can be increased significantly by application of CFRP sheets on the tension face of these elements (Arduini
and Nanni, 1997). CFRP sheet has also been used to strengthen beams in shear, by placing CFRP sheets in high
shear stress zones (Li et al., 2001). Some research has been conducted on the strengthening of one and two-way

539
reinforced concrete slabs using CFRP laminates and sheets (Arduini et al.,2004, Mosallam and Mosallam, 2002) but
research on prestressed and hollow core slabs is limited (Hosny et al., 2003). This paper presents the results of full-
scale load test on PPHC slabs strengthened by CFRP sheets in flexure and comparison made with the predicted
flexural load capacity using a mechanistic model.

2. EXPERIMENTAL DESIGN
A series of seven full scale hollow core concrete slabs (Table 1: Designated as S1 to S7) having a 5.0 m span, 1.2 m
width and 0.2 m thickness and made from concrete having a compressive strength of 45 MPa, were procured from a
commercial precast factory. Each PPHC slab has 44% voids and is reinforced with four 7-wire pretensioned steel
strands of 12.7 mm diameter.

The PPHC slabs were strengthened with CFRP sheets (SIKA wrap hex 230C with unidirectional carbon fiber fabric)
applied at the bottom of the slabs using Sikadur 330 epoxy resins. The CFRP sheets, 0.3 m in width and 0.13 mm
thickness, has a tensile strength of 3.45 GPa, tensile modulus of 230 GPa and an ultimate strain of 1.5%. The
simply supported PPHC slabs were tested in flexure using four point loading up to failure. An array of five LVDT’s
was used at different points on the slab to measure the deflection. Strain gauges were applied on concrete and CFRP
sheets and all data was captured using a data logger. The testing arrangement is shown schematically in Figure 1.

Figure 1: Details and Isometric View of Experimental Setup

Table 1: Designation and Description of Strengthening for Hollow Core Slabs

S. No Slab Specimens ID Description

1 S1 Control specimen
2 S2 Single layer CFRP sheet on 300 mm width
3 S3 Single layer CFRP sheet of 300 mm width (Identical to S2)
4 S4 Single layer CFRP sheet of 600 mm width
5 S5 Single layer CFRP sheet of 900 mm width
6 S6 Two Layers CFRP sheet of 600 mm width
Damaged slab strengthened with single layer CFRP sheet of
7 S7
300 mm width

3. RESULTS AND DISCUSSIONS

The cracking load, ultimate load, deflection at maximum load and failure mode of all the slabs are summarized in
Table 2. The virgin slabs strengthened using various widths of CFRP strips (S2 to S5) showed a brittle rupture of the
CFRP sheet at failure (Figure 2). The Slab S6 which was strengthened with two layers of CFRP sheets showed a
sudden failure in shear with diagonal crack emanating from outside the point of load application as shown in Figure
3. The flexural strength enhancement of PPHC slabs is limited by shear strength of these slabs. There is a transition
in the failure mode from flexural to shear failure once the shear capacity of these slabs is exceeded. It is therefore
important to strengthen these slabs in shear if the flexural capacity needs to be increased further. The load deflection
curves for the virgin slabs strengthened by CFRP sheets are shown in Figure 4. This figure shows identical linear

540
behavior of all specimens until cracking occurred at a load of about 65-70 kN. A non-linear behavior ensues
subsequently till the brittle failure of CFRP. The flexural capacity of the PPHC slab increases by 14-36% depending
on the cross sectional area of the sheets applied. Slab damaged by loading up to cracking load and subsequently
strengthened by CFRP showed 18% increase in ultimate moment capacity (Figure 5).

Figure 2: Brittle Rupture of CFRP Strip Figure 3: Shear Failure of Slab S6

160

160

120
120
Load (kN)

Load (kN)

80
Load vs Displacement 80
Slab S1 Load vs Displacement
Slab S2 Slab S1
Slab S3 Slab S7
Slab S4
Slab S5
40
Slab S6 40

0 0

0 20 40 60 80 100 0 20 40 60 80 100
Deflection (mm) Deflection (mm)

Figure 4: Load vs. Deflection of Virgin Strengthened Slabs Figure 5: : Load vs. Deflection of Cracked Slab

Table 2: Experimental Results for Flexural Strengthening of PPHC Slabs with CFRP sheets

Cracking Maximum Deflection

%
Slab Load Load at Max Failure Mode
Increase
(kN) (kN) Load (mm)
S1 (control) 65 110 - 97 Flexural Failure
S2 64 125 14% 89 Rupture of CFRP
S3 68 130 18% 87 Rupture of CFRP
S4 69 145 32% 84 Rupture of CFRP
S5 67 148 36% 83 Rupture of CFRP
S6 66 154 40% 79 Shear failure,
S7 64 130 18% 61 Flexural and Shear Cracks

541
Cracking load of PPHC slabs was calculated using the method outlined in the PCI manual for the design of hollow
core slabs (PCI, 1998). Strain compatibility approach (ISIS M05-00, 2001) was used to predict the flexural failure
loads of PPHC slabs strengthened with CFRP sheets. A good correlation between the experimental and predicted
values was obtained for slabs S1 to S4 (Percent variation 2 % to 5 % - Table 3). For slabs S5 and S6 as the
percentage of CFRP sheet was increased beyond the balanced level, the experimental value of ultimate flexural load
was found to be significantly smaller than the predicted capacity. The pattern of cracking especially in Slab S6 was
predominantly diagonal, showing shear failure rather than flexural failure in the slabs.

Table 3: Comparison of Cracking Load and Ultimate Load

Cracking Load (kN) Ultimate Flexural Load (kN)

Slab
Experimental Predicted Experimental Predicted
S1 65 75 110 105
S2 64 75 125 123
S3 68 75 130 123
S4 69 75 145 143
S5 67 75 148 162
S6 66 75 154 180

4. CONCLUSIONS
The present study showed that CFRP sheets can be successfully used for flexural strengthening of virgin PPHC with
an increase in flexural capacity ranging from 14 % to 36% depending on the cross-sectional area of CFRP used.
Slabs with initial damage, repaired and strengthened with CFRP sheet, showed a significant increase in ultimate
flexural capacity. A good correlation was obtained between the experimental and predicted values of failure loads
using strain compatibility approach for under-reinforced slabs. Over-reinforced slabs showed a transition to shear
failure (in contrast to flexural failure) due to the limited shear capacity of the hollow core slabs. As an engineering
guideline, it is recommended to ensure that ρ cfrp < 0.75ρ (cfrp) bal in strengthening hollow core slabs.
B B B B

5. ACKNOWLEDGEMENTS
The support of Civil Engineering Department and Research Institute at King Fahd University of Petroleum &
Minerals (KFUPM) is deeply acknowledged. Thanks to INCO Precast Company and SIKA for providing materials.

6. REFERENCES
Bakis, C. E. et al. (2002). “Fibre-Reinforced Polymer Composites for Construction-State of the Art Review,”
Journal of Composites for Construction, ASCE, Vol. 6, No. 2, pp. 73-87.
Nanni, A.(1995). “Concrete Repair with Externally Bonded FRP Reinforcement,” Concrete International, Vol. 17,
No. 6, pp. 22-26.
El Maaddawy, T. and Soudki, K. (2005). “Carbon-Fiber-Reinforced Polymer Repair to Extend Service Life of
Corroded Reinforced Concrete Beams”, Journal of Composites for Construction, ASCE, Vol. 9, No. 2, pp. 187-194.
Arduini, M. and Nanni, A. (1997). “Behavior of Precracked RC Beams Strengthened with Carbon FRP Sheets”,
Journal of Composites for Construction, ASCE, Vol. 1, No. 2, pp 63-70.
Li, A., Assih. J, and Delmas, Y. (2001). “Shear Strengthening of RC Beams with Externally Bonded CFRP Sheets”,
Journal of Structural Engineering, Vol. 127, No. 4, 2001, pp. 374-380.
Arduini, M., Nanni, A., and Ramagnolo, M. (2004) “Performance of One-Way Reinforced Concrete Slabs with
Externally Bonded Fiber-reinforced Polymer Strengthening”, ACI Structural Journal, Vol. 101, No. 2, pp. 193-201.
Mosallam, A.S., and Mosallam, K.M. (2002). “Strengthening of Two-way Slabs with FRP Composite Laminates.”
Construction and Building Materials, Vol. 17, pp. 43-54.
Hosny, A., Abdelrahman, A. and Elarabi, A. (2003) “Strengthening of Prestressed Concrete Slabs Using CFRP”,
Proceedings International Conference on. Composites in Construction CCC2003, Cosenza, Italy, pp. 355-359.
PCI. (1998) “Manual for the Design of Hollow Core Slabs” Second Edition, Precast Concrete Institute.
ISIS M05-00 (2001). “Strengthening Reinforced Concrete Structures with Externally Bonded Fiber Reinforced
Polymers”, ISIS Canada, University of Manitoba, Winnipeg, Manitoba, Canada.

542
Third International Conference on FRP Composites in Civil Engineering (CICE 2006)
December 13-15 2006, Miami, Florida, USA

USE OF CARBON-FIBER REINFORCED POLYMERS

IN SLAB-COLUMN CONNECTION REPAIRS

Widianto
(Structural Engineer, Bechtel Corporation, Houston, TX 77056, USA)

Ying Tian
(Graduate Research Assistant, University of Texas at Austin, Austin, Texas 78712, USA)

Jaime Argudo
(Project Manager, Engineering Diagnostics, Inc., Austin, Texas 78701, USA)

Prof. Oguzhan Bayrak

(Professor, Dept. of Civil Engineering, University of Texas at Austin, Austin, Texas 78712, USA)

Prof. James O. Jirsa

(Professor, Dept. of Civil Engineering, University of Texas at Austin, Austin, Texas 78712, USA)

ABSTRACT
Five 2/3-scale slab-column connections were tested to compare the efficiency of different rehabilitation techniques
for repair and strengthening earthquake-damaged connections. Two strengthening and repair techniques using CFRP
were studied: (i) externally installed CFRP stirrups, (ii) externally installed CFRP sheets on the tension surface and
anchored into the slab using CFRP anchors. Both techniques increased the connection strength and improved the
residual capacity after punching failure.

KEYWORDS
Repair, strengthening, slab-column connection, two-way shear, carbon-fiber reinforced polymers sheets.

1. INTRODUCTION
Flat-plate structural systems may be prone to shear failure at slab-column connections during or after strong ground
motions. Such failures may result in a progressive collapse of a building. Therefore, slab-column connections with
insufficient punching shear strength or connections damaged during strong ground motions may require
rehabilitation. This paper focuses on rehabilitation of typical flat-plate structures, built in the mid 1900s, that have
been subjected to an earthquake.

Typical flat-plate structures do not have a concentration of the slab top steel near the column as required by the
current building code (ACI 318-05). Structural drawings of several flat-plate structures located in Western US show
that those structures have roughly 0.5% reinforcement ratio in the column strip and no shear reinforcement. Since
low percentage of longitudinal reinforcement in the column strip results in low two-way shear strength (Marzouk
and Hussein, 1991), the rehabilitation of the earthquake-damaged connections in the existing structures usually
involves strengthening for flexure and punching shear. In order to preserve the advantage of flat-plate system (i.e.
maximizing clear space for given story heights), Carbon-Fiber Reinforced Polymer (CFRP) materials were used in
rehabilitation because the CFRP materials do not change the useable space in the building.

ACI 318-05 states that the nominal two-way shear strength Vc of an interior slab-column connection with a square
column is the lesser of Eqs. 1 and 2: Vc = (40 × d bo + 2) × f c ' × bo × d (Eq.(1)) ; Vc = 4 × f c ' × bo × d (Eq.(2)),
where d is the average depth of slab reinforcement, bo is the critical shear perimeter located at a distance d/2 away
from the edge of the column or from the outermost shear reinforcement, and fc′ is the concrete compressive strength.

543
2. TEST PROGRAM, MATERIAL PROPERTIES, AND TEST SETUP
Five two-third-scale specimens representing interior slab-column connections were tested (Table 1). The prototype
structure for all specimens, except G1.0, was assumed to have office occupancy, a live load of 50 psf, partition and
additional dead load of 20 psf, 21′ span length, 24″ square column, and 9″ slab thickness. All slabs, except G1.0, had
0.5% top reinforcement in the column strip, 0.25% reinforcement elsewhere, and no shear reinforcement. G1.0 had
1.0% top reinforcement between lines that are 1.5×(slab thickness) outside opposite faces of the column (a width of
(c+3h)), which is typical in the flat-plate structures designed using ACI 318-05. All slabs had the same bottom
reinforcement.

Grade 60 deformed reinforcement satisfying ASTM A 706-06 requirements and 4000-psi concrete were used in the
experimental program. The actual concrete compressive strengths for all specimens are shown in Table 1. The
details of slab reinforcement are given elsewhere (Widianto et al., 2006). 0.04″ thick, unidirectional CFRP sheet
with aramid cross fibers (Tyfo SCH-41S) was used. The specified elastic modulus, ultimate tensile stress, and
ultimate tensile strain of the CFRP sheet are 10500 ksi, 127 ksi, and 0.012, respectively.

Table 1. Test program, specimens, and results

fc' ρ top V ACI b o

v = V / (b o d )
Specimen Test program
(psi) within (c +3h )* (kip) (in)
G0.5 Punching shear loading only 4550 0.5 69.9 84 2.47 √(f c ' )
G1.0 Punching shear loading only 4070 1.0 90.2 84 3.37 √(f c ' )
LG0.5 Lateral loading up to 1.25%, then punched 4860 0.5 72.7 84 2.48 √(f c ' )
Lateral loading up to 1.25%,
LRstG0.5 4930 0.5 86.5 84** / 135*** 2.93 √(f c ' ) / 1.83 √(f c ' )
rehabilitated by external CFRP stirrups, then punched
Lateral loading up to 1.25%, 3.41 √(f c ' )
LRshG0.5 4630 0.5 97.5 84
rehabilitated by CFRP sheet, then punched
Specimen notation: G: Gravity load up to failure (concentric punching shear test) L: Reverse cyclic lateral load up to 1.25% drift
Rst: Rehabilitation using externally installed CFRP stirrups Rsh: Rehabilitation using externally installed CFRP sheets
0.5: 0.5% slab top steel within (c+3h) 1.0: 1.0% slab top steel within (c+3h)
* c=16" (column dimension of the specimen), h=6" (slab thickness of the specimen) ρ top : % slab top steel
** b o =84" was calculated for the critical perimeter d/2 away from the column face (d=5"). V : Punching load
*** b o =135" was calculated for the critical perimeter d/2 away from the outermost CFRP stirrup (d=5"). v : Failure shear stress

Specimens G0.5 and G1.0 served as control specimens to provide information on punching shear strength of
undamaged and unstrengthened specimens. Specimens LG0.5, LRstG0.5, and LRshG0.5 were first subjected to
reversed cyclic lateral displacements up to 1.25% to produce damage in the slab-column connection, and then the
slab was subjected to a concentric punching shear loading up to failure. At the end of the simulated seismic loading,
specimens LRstG0.5 and LRshG0.5 were rehabilitated. In order to mimic the rehabilitation process in a flat-plate
building that is unshored, the column axial load producing a shear force of 0.23Vc (representing dead load plus 25%
of live load on the prototype structure) on the critical shear perimeter was maintained during rehabilitation.

Figure 1 shows the test setup for simulated seismic loading and for concentric punching shear loading. The positions
of the struts were selected to reflect results of finite element analyses conducted on the prototype structure subjected
to lateral and gravity loads. More details of test setup are given elsewhere (Widianto et al., 2006).
16″×16″
14′×14′×6″ slab
Horizontal column
Torsional actuator
bracing

Vertical
Horizontal strut
strut

(a) Simulated seismic test Hydraulic jack (b) Concentric punching shear test

Figure 1: Test setup

544
3. REHABILITATION PROCESS
3.1 Installation of CFRP Stirrups

LRstG0.5 was repaired by installing external CFRP stirrups around the column (Figure 2). This rehabilitation
technique was previously studied by Binici (2003). Surface preparation involved (i) locating the slab reinforcement
using non destructive testing, (ii) drilling 3/4–inch holes, (iii) grinding the slab surface, and (iv) chamfering the edge
of the hole to minimize stress concentrations. 3/4-inch wide CFRP strips were cut from a roll of CFRP fabric. The
first row of CFRP stirrups were located as close to the column face as possible (d/4 away from column face) in order
to intercept the shear crack. The other rows of CFRP stirrups were spaced at about d/2. After being impregnated
with epoxy and passed through a saturator to remove excessive epoxy, the CFRP strips were stitched (once or twice)
through the holes and wrapped to form closed stirrups. After the completion of CFRP stirrup placement, the bottom
of the vertical holes were plugged and the holes were filled with epoxy.

3.2 Installation of Well-anchored CFRP Sheets

LRshG0.5 was repaired by applying 12″-wide CFRP sheets around the column (Figure 2), on the tension surface of
the slab to increase the flexural capacity of the slab that had only 0.5% steel within (c+3h) critical section. The
amount of CFRP reinforcement was selected to produce the same flexural capacity as that of the connection with
1.0% steel. The installation of CFRP sheets involved drilling four holes (¾-inch diameter and 4½-inch deep) at each
corner of the column so that CFRP anchors could be inserted. The edges of the holes were chamfered and the
concrete surface was ground smooth and cleaned. The CFRP placement was done as follows: (i) Epoxy was poured
into cracks and holes, (ii) concrete surface was coated with the epoxy, (iii) an epoxy-impregnated CFRP sheet was
inserted into a saturator to remove excessive epoxy, (iv) the sheet was placed on the concrete surface and a paint
roller was used to remove air pockets below the sheet, (v) CFRP anchors were inserted into the holes, and (vi) the
protruding ends of the anchors were splayed over the CFRP sheet.

4. TEST RESULTS AND DISCUSSIONS

Figure 2 shows the gravity load versus vertical displacement (at the column) curves for a punching shear test. The
gravity load capacity and the shear stress at the critical shear perimeter vc are summarized in Table 1. For the
connections tested in this study (all specimens except LRstG0.5), Eq. 2 governs the design. The critical shear
perimeter of LRstG0.5, which is at d/2 away from the outermost CFRP stirrup, is larger than that of the other
specimens. Thus, Eq. 1 governs and gives vc of 3.48 f c ' for LRstG0.5. The measured vc of LRstG0.5 was 1.83 f c ' .

Table 1 shows that all of the specimens tested in this study failed at shear stress levels that were lower than 4 f c ' ,
the value used in most designs. Punching shear failure was initiated by large flexural cracks that reduced concrete
contribution to shear strength and caused early punching shear failure. At failure, the vc inferred from the
measurements of specimens G0.5 and G1.0 were 2.47 f c ' and 3.37 f c ' , respectively. The measured strengths
were only 63% and 85% of the strength estimated using ACI 318-05 expression (Eq. 2). This observation is
consistent with the test results of a 45-foot square flat-plate structure (Guralnick and LaFraugh, 1963).

4.1 Effect of Externally Installed CFRP Stirrups

Comparing the behavior of LRstG0.5 with that of LG0.5 shows that the externally installed CFRP stirrups improved
the punching shear strength and deformation capacity of an earthquake-damaged connection. A tightly knit array of
CFRP stirrups shifted the failure surface away from the column (increased the critical shear perimeter) and also
engaged more flexural reinforcement into the punching cone for improved dowel resistance following a punching
shear failure. Therefore, the CFRP stirrups increased the punching shear capacity and residual capacity after
punching shear failure. A peak shear stress of 2.93 f c ' on the critical perimeter d/2 away from the column face was
inferred from measurements. On the critical section d/2 away from the CFRP–reinforced zone, the shear strength of
concrete was equal to 1.83 f c ' , which is equivalent to 53% of the capacity calculated using Eq. 1.

545
 100
90
LRshG0.5 G1.0 LRstG0.5
80 LRstG0.5
70 LG0.5
60
Load (kip)

G0.5
50
40
30
20 Service load
LRshG0.5
10
0
0 0.5 1 1.5 2 2.5
Displacement (inch)
4 CFRP anchors

Figure 2: Test results

4.2 Effect of Well-anchored CFRP Sheets

Comparing the behavior of LRshG0.5 with that of LG0.5 shows that the installation of well-anchored CFRP sheets
on the tension side of the slab improved the punching shear strength of an earthquake-damaged connection. Table 1
shows that the CFRP installation improved the punching shear strength of the connection by 38% (from 2.47 f c ' to
3.41 f c ' ). The area of the CFRP sheets was selected to match the flexural capacity of a connection with 1.0% steel
within (c+3h) region. Figure 2 shows that the ultimate capacity of LRshG0.5 was about the same as that of G1.0,
implying that external CFRP sheets were just as effective as the steel reinforcement. Four CFRP anchors per column
corner (Figure 2) were very effective in preventing early delamination. At failure, the CFRP sheets close to the
middle of the column face ruptured (causing significant decrease in load-carrying capacity as shown in Figure 2),
whereas the sheets elsewhere were still attached to the slab. After punching shear failure occurred, the well-
anchored CFRP sheets acted as tension bands and allowed the slab to carry substantial shear force through larger
deformations. The CFRP sheets did not change the location of the failure surface but limited the width of flexural
cracks (that occurred due to simulated seismic displacements) so that the area of concrete resisting shear was
maintained and the connection was able to carry more load before a punching shear failure occurred.

5. CONCLUSIONS
Two-way shear strength was sensitive to the slab top reinforcement ratio within (c+3h). Two-way shear strength of
the undamaged control specimens with 0.5% and 1.0% top reinforcement ratio was only 63% and 85% of that
estimated using ACI 318-05 expression ( Vc = 4 × f c ' × bo × d ). Both the installation of the external CFRP stirrups
and the application of the well-anchored CFRP sheets on the tension surface of the slab increased the two-way shear
strength of the earthquake-damaged slab-column connections and improved the residual capacity after punching
shear failure. CFRP anchors were very effective in preventing early delamination of CFRP sheets.

6. REFERENCES
Binici, B. (2003). “Punching shear strengthening of reinforced concrete slabs using fiber reinforced polymers”,
Ph.D. dissertation, University of Texas at Austin, Texas, USA.
Guralnick, S.A., and LaFraugh, R.W. (1963). “Laboratory study of a 45-foot square flat plate structure”, Journal of
the American Concrete Institute, Vol. 60, pp 1107-1185.
Marzouk, H., and Hussein, A. (1991). “Experimental investigation on the behavior of high-strength concrete slabs”,
ACI Structural Journal, Vol. 88, No. 6, pp 701-713.
Widianto, Tian, Y., Argudo, J., Bayrak, O., and Jirsa, J.O. (2006). “Rehabilitation of earthquake-damaged reinforced
concrete flat-plate slab-column connections for two-way shear”, 8th US National Conference on Earthquake
Engineering, San Francisco, California, April 18-22, 2006.

546
Part XX. Seismic Applications
Third International Conference on FRP Composites in Civil Engineering (CICE 2006)
December 13-15 2006, Miami, Florida, USA

SEISMIC RETROFIT OF LARGE SCALE CIRCULAR RC COLUMNS

WRAPPED WITH CFRP SHEETS
G. Wu
Associate professor, College of Civil Engineering, Southeast University, Nan Jing, China

Z.S. Wu
Professor, Department of Urban & Civil Engineering, Ibaraki University, Hitachi, Japan
Cheung Kong Scholar, College of Civil Engineering, Southeast University, Nan Jing, China

Z.T. Lü
Professor, College of Civil Engineering, Southeast University, Nan Jing, China

D.S. Gu
Doctoral Candidate, College of Civil Engineering, Southeast University, Nan Jing, China

ABSTRACT
An experimental investigation was conducted to study the seismic behavior of large scale circular reinforced
concrete columns strengthened with carbon fiber reinforced polymer (CFRP). Results from the experimental
program are presented in which five 360mm diameter and 1000mm long columns were tested under constant axial
load and reversed cyclic lateral load that simulated forces from an earthquake. Each specimen consisted of a column
cast integrally with a 1550mm×600mm×700mm stub that represented a beam-column joint area or a footing. The
results indicate that the failure models of FRP-wrapped circular RC columns may be shear failure or flexural failure
and the cyclic behavior progressively improved as the number of CFRP layers increased. New methods for
predicting shear and flexural capacities of FRP-wrapped circular RC columns under seismic loading were proposed.

KEYWORDS
Columns, Concrete, CFRP, Seismic loading, Strength

INTRODUCTION
It is well known that column performance in the inelastic mode is of utmost importance for the safety of a structure
during an earthquake. FRP composites has a lot of advantages and has already been widely used for improving the
seismic performance of circular reinforced concrete columns. As a result, numerous papers on various aspects
related to the subject have been published recently. Most of the work on FRP-wrapped columns was undertaken on
small cylinders or columns mainly subjected to axial loading. A very limited amount of experimental data exists on
the seismic behavior of realistically sized circular concrete columns confined with CFRP wraps. This research
investigates the seismic performance of near full-scale CFRP-retrofitted columns typical of existing buildings and
highways. The main variable of the study was the layers of CFRP. New methods for predicting shear and flexural
capacities of FRP-wrapped circular RC columns under seismic loading are proposed.

EXPERIMENTAL PROGRAM
Specimens

Five large-scale circular reinforced concrete columns were constructed. Each specimen consisted of a circular
column of 360mm diameters cast integrally with a stub of dimensions 1550mm×600mm×700mm. All columns
contained fourteen D25mm longitudinal bars uniformly distributed around the core. The D6mm ties were placed at a
spacing of 150 mm within the test region. Reinforcement of columns is shown in Fig.1. Properties of steel bars are
given in Table 1. One unwrapped column was used as control specimen to evaluate the effects of FRP retrofitting.

547
Four of these columns were strengthened with CFRP wraps. The main variable of the study was the number of
CFRP layers. The region of 100mm from the stub face was strengthened with additional eight D25mm bars, and the
D10mm ties were placed at a spacing of 30 mm within this region to minimize the chances of failure at the section
of the stub face. All the specimens were tested under constant axial load and cyclic lateral excursions simulating
seismic loading conditions. A commercially available CFRP wrap system was used to retrofit the designated
specimens. The thickness of one layer of CFRP sheet was 0.167 mm, the average tensile strength, the modulus of
elasticity and the tensile rupture strain were 3945MPa, 249.6GPa and 1.52% respectively.

Table 1 Details of test specimens

CFRP Axial Longitudinal steel Transver steel ρf

Specimen f’c treatmen load Ds fy fult Ds S fy fult
No.
(MPa) t (kN) (mm) (MPa) (MPa) (mm) (mm) (MPa) (MPa) (%)
CL0 34.9 Control 1200 12 25 382.4 571.1 6 150 319.8 474.4 0
CL1 34.9 1.0 layer 1200 12 25 382.4 571.1 6 150 319.8 474.4 0.093
CL2 34.9 2.5layers 1200 12 25 382.4 571.1 6 150 319.8 474.4 0.2325
CL3 34.9 3.5layers 1200 12 25 382.4 571.1 6 150 319.8 474.4 0.3255
CL4 34.9 4.5layers 1200 12 25 382.4 571.1 6 150 319.8 474.4 0.4185

Testing

The specimens were tested as shown in Fig. 2. The axial load was applied through a hydraulic jack having the
capacity of 3000 kN and was measured using a load cell of similar capacity. To apply reverse lateral load, an
actuator with a load capacity of 1000 kN and a stroke capacity of ±150 mm was used. The load cycles were divided
into two phases: load control and displacement control. Load control phase was used up to yielding of the
longitudinal bars, beyond that point, a displacement control sequence was used.
550
65
400

φ25
A-A
φ25

φ25
500

A A
φ25
100

B B

φ25
B-B

φ25

250
1550

Fig.1 Reinforcement of columns Fig.2 Test setup

Table 2 Test results

Fy （kN） Δy Fmax（ Δu （ Δu/Δy

Specimen CFRP (mm） kN） mm） Failure modes
treatment
CL0 Control column 380 - 424.9 6.0 - Shear failure
CL1 1 layer CFRP 400 5．6 569.1 24.0 4．3 Shear failure, FRP rupture
CL2 2.5 layers CFRP 360 6．5 634.4 48.0 7．4 Flexural failure, FRP rupture
CL3 3.5 layers CFRP 400 6．2 664.8 48.0 7．7 Flexural failure, FRP rupture
CL4 4.5 layers CFRP 400 7．1 679.8 48.0 6．8 Flexural failure, FRP rupture

Analysis of results
The test results are shown in Table 2. The curves of applied lateral load versus displacement are shown in Fig.3. The
maximum lateral load of the control column was 424.9kN, and it was controlled by shear failure. The Column CL1

548
strengthened with one layer CFRP, the maximum lateral load of the Column CL1 strengthened with one layer CFRP
sheet was 569.1kN, 33.9% higher than the control column. Examination of lateral load indicates that columns
strengthened with more than 2.5 layers CFRP attained Fmax that were 49.3% to 60% higher than the Fmax for the
control column. And all these columns had good ductility and failed by flexure. The results also showed that the
maximum lateral loads were improved with the increase the layers of CFRP, while the ultimate lateral displacements
were almost the same for the Columns CL2, Column CL3 and Column CL4 due to flexural capacities of specimens.
It is worth being researched further.

Fig. 3 Lateral load-displacement responses

SHEAR CAPACITY OF CFRP-WRAPPED COLUMNS

Existing researches show that it is difficult to accurately estimate the contribution of FRP sheets to the shear
capacity Vf of beams. Detailed investigations on the strengthening of RC members against shear using externally
bonded FRP sheets have been relatively limited and, to a certain degree, controversial. Due to the lack of adequate
test data, it is difficult to standardize a design equation that takes into account all of the factors affecting Vf .
Many parameter were not considered by the existing models(Nanni 2002; Triantafillou 2000; Chaallal 2004): (1)
most of existing models for predicting Vf were suggested based on the test data of concrete beams; (2) the influence
of axial load on the ultimate effective strain of FRP sheet was not analogized; (3) some methods were suggested to
estimate the contribution of FRP sheets to the shear capacity of rectangular columns, while few on circular RC
columns; and (4) the improvement of the confinement of FRP on concrete strength was not considered on the
contribution of concrete to the shear capacity.
The following assumptions are made: (1) if the stress-strain curve of FRP-confined concrete circular columns has
no strain-softening response, the influence of the confinement of FRP on concrete strength will be considered on the
contribution of concrete to the shear capacity; (2) the nominal shear strength of RC columns strengthened using
externally bonded FRP sheets can be computed accumulating the contribution of concrete, steel stirrup and CFRP
composites; (3) the shear contribution of concrete and steel stirrup can be predicted by the model suggested by Ghae
et al.(1989); and (4) the efficiency factor of the ultimate effective strain of CFRP sheet can be temporarily taken as
0.58 according to the Lam&Teng (2003)’s research based on the analysis of CFRP-wrapped circular columns.
Table 3 shows that calculated shear forces of Vcv predicted by this paper compare well with the test results.
Table 3 Predicted and experimental values of columns (Unit: kN)
Spec. CFRP Treatment Vc Vs Vf Vmax,t Vcv Vcv/ Vmax,t Failure modes
CL0 No CFRP 357.6 31.7 0 424.9 389.3 0.92 Shear failure
CL1 1.0 layers 359.5 31.7 169.9 569.1 561.1 0.99 Shear failure, FRP rupture

549
FLEXURAL CAPACITY OF CFRP-WRAPPED COLUMNS
A finite element model is developed for the sectional analysis of CFRP-confined columns. The model satisfies force
equilibrium and strain compatibity, and utilizes uniaxial constitutive models of concrete and FRP. The following
assumptions are made: (1) plane sections remain plane after bending; (2) perfect bond exists between concrete and
FRP tube; (3) FRP only provides lateral confinement, without any stiffness in the longitudinal direction; (4) the
stress-strain relationship of concrete in compression is defined by Wu et al. (2003; 2006) models; (5) the tensile
strength of concrete is ignored; (6) the steel reinforcing bars are elastic-perfectly plastic; (7) any confinement effect
of steel hoop reinforcement is ignored; and (8) the ultimate limit state is reached when the strain of the extreme
compression fiber of concrete attains the ultimate strain.
Table 4 shows the calculated flexural capacities of Vcf predicted by this paper compare well with the test results.

Table 4 Predicted and experimental values of columns

Specimen CFRP treatment Vmax,t Vcf Vcf/ Vmax,t Failure modes

CL2 2.5 layers 634.4 536 0.85 Flexural failure, FRP rupture
CL3 3.5 layers 664.8 546 0.82 Flexural failure, FRP rupture
CL4 4.5 layers 679.8 557 0.82 Flexural failure, FRP rupture

SUMMARY AND CONCLUSIONS

The effectiveness of CFRP confined large scale circular column was verified. The seismic behavior of columns
progressively improved as the number of CFRP layers increased. The new design methods suggested by this paper
can be used to predict the shear and flexural capacities of FRP-wrapped circular RC columns under seismic loading
well.

ACKNOWLEDGMENTS
The project sponsored by the Scientific Research Foundation for the Returned Overseas Chinese Scholars, State
Education Ministry. The authors also would like to thank the support from Scientific Research Foundation from
Qing Hai Science & Technology Department of China (2005G155) and Nanjing Construction Committee of China.

REFERENCES
A. Khalifa and A. Nanni (2002). “Rehabilitation of rectangular simply supported RC beams with shear deficiencies
using CFRP ”. Construction and Building Materials ,Vol.16 : 135 -146.
T.C. Triantafillou and C.P. Antonopoulos ( 2000) .“Design of Concrete Flexural Members Strengthened in Shear
with FRP”. Journal of Composites Construction, ASCE, Vol.(4), No. 4, pp. 198-205.
A.Bousselham and O. Chaallal (2004). “Shear Strengthening Reinforced Concrete Beams with Fiber-Reinforced
Polymer: Assessment of Influencing Parameters and Required Research”, ACI Structural Journal , Vol.(101), No .2,
pp.219 - 227.
G. Wu, Z.T. Lu¨ and Z.S. Wu, (2006). “Strength and ductility of concrete cylinders con.ned with FRP composites”,
Construction and Building Materials ,Vol,20, pp.134–148.
G. Wu, Z.T. Lu and Z.S. Wu, (2003). “Stres-strain relationship for FRP-confined Concrete Cylinders”.Proceedings
of the sixth International Symposium on FRPRCS, pp.552-560.
L. Lam and J.G. Teng, (2003). “Hoop rupture strains of FRP Jackets in FRP Confined Concrete”.Proceedings of the
sixth International Symposium on FRPRCS, pp. 601-612.
S.A. Sheikh and G. Yau, (2002). “Seismic Behavior of Concrete Columns Confined with Steel and Fiber-Reinforced
Polymers”. ACI Structural Journal, Vol.99, No.1, pp.72-80.
J.B.Mander, M.J.N. Priestley, and R.Park.(1988).“Theoretical Stress-strain Model for Confined Concrete ”. Journal
of Structural Engineering, ASCE, Vol.114, No.8, pp.1804-1826.
A.B. Ghae, M.J.N. Priestley and T. Paulay.(1989). “Seismic Shear Strength of Circular Reinforced Concrete
Columns”. ACI structural Journal, Vol.(86),No.1, pp. 45-59.

550
Third International Conference on FRP Composites in Civil Engineering (CICE 2006)
December 13-15 2006, Miami, Florida, USA

SEISMIC RETROFITTING OF REINFORCED CONCRETE COLUMNS

BY CONTINUOUS FIBER ROPE WITH CONCRETE JACKET
Takumi Shimomura
(Associate Professor, Nagaoka University of Technology, Nagaoka, Niigata, Japan)

Nguyen Hung Phong

(Former Graduate Student, Nagaoka University of Technology, Nagaoka, Niigata, Japan)

Akihiro Matsumoto
(Former Graduate Student, Nagaoka University of Technology, Nagaoka, Niigata, Japan)

Kenzo Sekijima
(Tokyo Branch, Kurasoku Kensetsu Consultant Co.Ltd., Tokyo, Japan)

Kyuichi Maruyama
(Vice President, Nagaoka University of Technology, Nagaoka, Niigata, Japan)

ABSTRACT
A new seismic retrofitting method for existing reinforced concrete columns using continuous fiber rope, CF rope,
coupled with concrete jacket is developed. CF rope is enough flexible to be easily arranged in structural members by
hand and, apart from other FRP materials for concrete such as FRP rod, grid and sheet, CF rope is applied without
epoxy resin. This paper presents tests of reinforced concrete column specimens retrofitted with CF rope. Aramid and
vinylon CF rope was wound continuously around the column periphery. Instead of epoxy resin, concrete jacket was
cast to bond CF rope with the original member and to protect CF rope from environmental attacks. The column
specimens were tested under reversed cyclic load. It was experimentally verified that the proposed retrofitting
method can successfully increase seismic resistance of the member.

KEYWORDS
continuous fiber rope, seismic retrofitting, ductility, concrete jacketing

1. INTRODUCTION
Continuous fiber materials, such as carbon fiber sheet, are often used in strengthening existing concrete structures.
This paper introduces a new seismic retrofitting method for existing reinforced concrete piers and columns using
continuous fiber material made in the form of rope. This retrofitting method is a very simple handling method since
it makes use of continuous fiber rope, CF rope. CF rope is distinguished from other types of continuous fiber
reinforcement for concrete, such as rod, grid and sheet, by its non-epoxy usage. In the previous research, the authors
have verified that CF rope can be used as shear reinforcement initially embedded in concrete (Phong et al., 2005). In
the present study, CF rope is applied on the surface of existing RC column with concrete jacket. A series of
reinforced concrete column specimens are strengthened by this method and tested under reversed cyclic load. Based
on the experimental results, load-carrying characteristics and failure behavior of the column specimens are
investigated and the effectiveness of CF rope in ductility improvement is verified. Test results are compared with
those of RC columns retrofitted with CF sheet that had been done by the authors (Maruyama et al., 2003).

2. CONTINUOUS FIBER ROPE

551
Two types of CF rope made of aramid and vinylon fiber are used in this study. The CF ropes consist of several
bundles of fiber twisted together. Without epoxy resin, CF rope is enough flexible to be easily arranged and adapted
to various shapes of existing structural RC members. Tensile properties of the CF ropes were determined based on
the uniaxial tensile test that had been developed for this material (Phong et al., 2005). Five test pieces were tested in
total for each type of CF rope. Test results are presented in Table 1. The nominal cross sectional area of CF rope was
determined by dividing the weight of 1m long of rope by the density of fiber.

Table 1: Tensile Properties of CF ropes

Cross
Name Type Tensile Elastic Ultimate
sectional
of of strength modulus strain
area
CF rope fiber (MPa) (MPa) (%)
(mm2)
CF rope 1 aramid 17.19 1660 53300 3.1
Figure 1: Continuous Fiber Rope CF rope 2 vinylon 12.66 734 15857 4.6

3. RETROFITTING METHOD FOR EXISTING RC COLUMNS

3.1 The Original Column Specimens

The series of test specimens consists of four identical RC columns. Specimen No.0 is a control specimen without
being retrofitted, while three other specimens were to be retrofitted with CF rope and concrete jacket. The column
specimens were designed to fail in shear mode. Details of the original column specimens before retrofitting are
shown in Figure 2. Compressive strength of concrete and yielding strength of steel reinforcement were 26 MPa and
384 MPa, respectively.

350
4D13
500

7D19 x 2
7D19 x 2
500

1250
4D6
@250
7D22 x 2 7D22 x 2
@180 @180

9D16 x 2 2D16 x 2
700

@210

1200 1800

Figure 2: Details of the Original Column Specimens

3.2 Retrofitting Procedure

Winding CF rope
Before winding rope, surface of the columns was ground by a grinding machine to ensure bonding between the
original concrete in column and the concrete jacket. The corners of the columns were rounded also by a grinding
machine to prevent stress concentration in CF rope. Then, CF rope was wound by hand around the outside periphery
of the columns. The rope was wound in a closed line: firstly from top to bottom of column and then back from
bottom to top. The ends of rope were anchored simply by making a knot. No epoxy resin was used to bond or anchor
CF rope on the concrete surface.

Concrete jacketing
After winding rope, formworks were set around the column and concrete whose maximum aggregate size was 13
mm was cast between the column and the formwork as a concrete jacket of 40 mm thickness. Concrete jacket was

552
cast from the footing up to the height close to the loading point of the column. Retrofitting procedure is shown in
Figure 3. Concrete jacket is expected to have multi functions, which are to assure bonding between original concrete
and CF rope, to protect CF rope from collision and environmental attack under service, and to make exterior of the
member beautiful. Details of the retrofitted column specimens are presented in Table 2. Compressive strength of
concrete used in jacket was examined on the day of the loading test of the column specimen.

Grinding Surface → Marking Rope Positions → Winding Rope → Setting Formworks → Jacketing
Figure 3: Retrofitting Procedure

Table 2: Details of the Retrofitted Column Specimens

Specimen No.1 No.2 No.3

Type of CF rope CF rope 1 CF rope 1 CF rope 2
Spacing of CF rope (mm) 100 x 2 200 x 2 60 x 2
Strength of concrete jacket (MPa) 26 22 25

4. REVERSED CYCLIC LOADING TEST

4.1 Test Setup and Method

The column specimens were tested under reversed cyclic load to examine their seismic resistance. Firstly, they were
loaded until the yielding of the longitudinal reinforcement was observed. The displacement of the column at the
loading point at this stage was defined as the yield displacement (δy). Then, reversed cyclic load was applied
incrementally under deformation control, such as +1δ y , −1δ y , +2δ y , −2δ y , +3δ y , −3δ y. At each deformation step,
three cycles of loading were repeated. Failure of the member is defined when its bearing capacity is reduced to 80%
of the maximum load.

4.2 Test Results and Discussions

Load – deflection characteristics

Test results of the four column specimens are shown in the first four columns in Table 3. The second row of the
table indicates retrofitting ratio of rope or sheet, taking into account their tensile strength. In the last row, ductility
ratio of column is defined as the ratio of the maximum displacement and the yield one. An example of the load-
deflection curve of the column is shown in Figure 4. For comparison, test results of RC column specimens
retrofitted with CF sheet in the previous research by the authors (Maruyama et al., 2003) are also shown in Table 3.
These specimens, having the same dimensions and concrete strength with the specimens in this study, were wrapped
by CF sheet bonded to concrete surface by epoxy resin.

The maximum load of the column specimens in this study is governed by the yielding of longitudinal reinforcement
in spite of retrofitting of CF rope. On the other hand, the maximum displacements are significantly different for the
four specimens; the member ductility is improved by retrofitting of CF rope and concrete jacket. The obtained
ductility ratio depends on the amount of CF rope retrofitted. Comparing the specimens with approximately the same
CF ratio, Column No.2 and No.3, the specimen with smaller rope spacing, Column No.3, had higher ductility ratio.

553
 Table 3: Test results of RC columns retrofitted with CF rope and CF Sheet

none CF rope CF Sheet

Retrofitting method
No.0 No.1 No.2 No.3 n5 n6
Type of fiber – aramid Aramid vinylon aramid aramid
CF ratio x tensile strength (MPa) – 228 114 124 64.0 129
304 309 315 315 297 294
Maximum load (kN)
-263 -300 -296 -300 -287 -303
13.0 55.2 15.9 32.1 20.7 44.0
Maximum displacement (mm)
-13.3 -25.8 -15.9 -32.1 -21.2 -38.5
Ductility ratio 2 3.3-7 2 5 4 7

Specimen No.3 Specimen No.1 Specimen No.2 Specimen No.3

Figure 4: Load-Displacement Curves Figure 5: Crack Patterns at failure stage

Failure mode
Under the reversed cyclic load, cracks were formed in concrete jacket and bonding was lost gradually. The crack
patterns of the specimens at failure state are shown in Figure 5. At the failure stage, most part of the jacket on the
loading surfaces lost bonding with the original column concrete as well as CF rope and could easily be removed
from the column. However, CF rope was not ruptured even at the failure stage in all the specimens with CF rope.

Similar to the case of the specimens retrofitted with CF sheet (Maruyama et al., 2003), it was also observed in this
study that the specimens having more distributed cracks, specimen No.1 and No.3, show greater deformation
capability or ductility. The difference between the case of CF sheet and CF rope lies in the fact that CF sheet was
bonded to concrete surface by epoxy resin, while CF rope was covered with concrete jacket instead of epoxy resin.
In the case of CF rope, bonding between rope and concrete was lost at the earlier stage than in the case of CF sheet.
This is why strain of CF rope was uniformly distributed and CF rope was not ruptured at the failure stage of the
specimen, whereas crush of concrete in compressive zone was observed.

6. CONCLUSIONS
A new seismic retrofitting method for existing RC columns using CF rope coupled with concrete jacket was
developed. It was experimentally verified that seismic resistance of RC columns could be improved by this
retrofitting method as effectively as CF sheet.

7. REFERENCES
Maruyama, K. et al. (2003). “Ductility Improvement Mechanism of Concrete Columns by Wrapping of FRP Sheets”.
Proceedings of the International Conference on Advances in Structures, Sydney, Australia, pp.1391-1397.
Phong, N.H. et al. (2005). “Shear Strengthening Effectiveness of Continuous Fiber Rope in RC Members”.
Proceeding of the Third International Structural Engineering and Construction Conference, Shunan, Japan, pp.57-64.

554
Third International Conference on FRP Composites in Civil Engineering (CICE 2006)
December 13-15 2006, Miami, Florida, USA

SEISMIC PERFORMANCE OF HYBRID FRP-CONCRETE

PIER COLUMNS
Yilei Shi
(Ph.D. Student, Florida International University, Miami, FL, USA)

Bin Li
(Ph.D. Student, Florida International University, Miami, FL, USA)

Amir Mirmiran
(Professor and Chair, Florida International University, Miami, FL, USA)

M. Saiid Saiidi
(Professor, University of Nevada, Reno, NV, USA)

ABSTRACT
One of the most cost-effective applications of fiber reinforced polymers (FRPs) is in hybrid construction with
concrete, where FRP provides the pour form, protective jacket, and shear and flexural reinforcement for concrete.
Seismic performance of concrete-filled FRP tubes (CFFT) is the subject of this study. While previous studies have
shown the feasibility of the proposed system both at the member level and at the connection level, especially under
reverse cyclic loading, this study focuses on the following issues that have not been addressed before: (a) effect of
the type of fibers and hybrid lay-up of the tube on the performance of the system, (b) effect of combined shear and
moment, especially for shorter columns, (c) lateral performance of two-column bent systems, and (d) holistic
performance of the entire bridge structure on shake tables. This paper reports on an on-going project to design the
next generation of seismic-resistant bridge substructure, as part of a multi-university project, which is funded by the
NSF-Network of Earthquake Engineering Simulation Research (NEESR) program.

KEYWORDS
Cyclic Loading, Fiber Reinforced Polymers, Hybrid Columns, Reinforced Concrete

1. INTRODUCTION
One of the most cost-effective applications of fiber reinforced polymers (FRPs) is in hybrid construction with
concrete, where FRP provides the pour form, protective jacket, and shear and flexural reinforcement for concrete.
Concrete-filled FRP tubes (CFFT) have been used in Virginia, California, and Florida (Zhu et al. 2006). Seible et al.
(1996) studied the feasibility of carbon CFFT columns under simulated seismic actions, and concluded that CFFT
without any starter bars would fail prematurely under combined compressive and inter-laminar shear stresses. On
the other hand, when carbon tubes are augmented with ample number of starter bars, ductility of the column is
increased significantly. A number of more recent studies (as cited in Zhu et al. 2006) have provided evidence as to
the advantages of glass over carbon in providing additional ductility for the column. This may be attributed to the
lower modulus of elasticity of glass FRP. Most recently, Zhu et al. (2006) showed construction feasibility and
superior performance of cast-in-place or precast CFFT columns with reinforced or prestressed connections.

This paper describes an on-going project on the use of FRP tube to enhance seismic performance of reinforced
concrete bridge substructure, as part of a multi-university project, which is funded by the NSF-Network of
Earthquake Engineering Simulation Research (NEESR) program. The FRP component of the project includes three
phases: (1) Single-column tests; (2) Two-column bent tests; and (3) Large-scale four-span bridge tests on shake
tables. The focus of this study is to (a) compare the performance of glass, carbon, and hybrid FRP tubes, (b)

555
compare flexure-dominant versus shear-dominant behavior of CFFT columns, (c) lateral performance of two-
column bent systems, and (d) holistic performance of the entire system on shake tables.

2. SINGLE COLUMN TEST PROGRAM

A total of six specimens are prepared in this stage with five CFFT columns and one control RC column. The
formwork for the RC control specimen consisted of a 305 mm (12 in.) diameter sonotube, while the other five CFFT
columns used FRP tubes as the cast-in-place formwork. One of the tubes was made by the industry using filament-
winding of 17 layers of +55o E-glass fibers and epoxy resin, with an inside diameter of 315 mm (12.4 in.) and a wall
thickness of 5.1 mm (0.2 in.). The other four FRP tubes were prepared in the Structures and Construction Lab (SCL)
at the Florida International University (FIU), by wrapping saturated FRP fabric around sonotubes of the same
diameter as that used for the control RC specimen. The sonotubes were first covered with a layer of wax paper, so
that the FRP can be easily detached later. All FRP tubes consisted of 152 mm (6 in.) overlaps in the hoop direction
and 305 mm (12 in.) overlap in the longitudinal direction. Figure 1 illustrates the preparation of one carbon FRP
tube.

Among the four FRP tubes made in SCL, one glass FRP tube was made with 3 layers of bi-directional glass fiber
sheets. The cured laminate thickness per layer is 0.33 mm (0.013 in.). Two carbon FRP tubes with shear spans of
1,295 mm (51 in.) and 2,210 mm (87 in.), respectively, were made with 2 layers of bi-directional carbon fiber sheets.
The cured laminate thickness per layer is 0.25 mm (0.01 in.). In the hybrid FRP tube, 3 layers of unidirectional glass
fiber sheets were wrapped in the transverse direction over the two layers of unidirectional carbon fabric. It may be of
interest to note that the layer numbers were designed based on equivalence comparison of the tensile, compressive
and shear strengths of cured laminate properties of glass and carbon fabrics. Apart from the CFFT with carbon tube
of 2,210 mm (51 in.) shear span (hence long column), all the other five columns have the same 1,295 mm (51 in.)
shear spans (hence short columns). Figure 2 shows the five short specimens. Table 1 shows the test matrix of single
column experiments.

Figure 1: Preparation of Carbon FRP Tube Figure 2: Five Short Column Specimens

All six columns had the same longitudinal reinforcement of sixteen 10 mm (No. 3) steel bars of Grade 414 MPa (60
ksi) along the entire length of the columns, with adequate embedment into the footing and the column head. This
corresponded to a 1.5% of reinforcement ratio. The RC column additionally featured a spiral reinforcement with 5.3
mm (0.207 in.) steel wire of Grade 414 MPa (60 ksi) with 279 mm (11in.) outside diameter placed at a pitch of 32
mm (1.25 in.). CFFT columns had no transverse steel reinforcement, except for four or five 279 mm (11in.) diameter
hoops placed at a spacing of 305mm (12 in.) only to hold the longitudinal reinforcement cage together before casting
concrete. Figure 3 shows the column reinforcements of the RC and the long carbon column.

The FRP tubes were embedded 305 mm (12 in.) into the footings to provide sufficient development length, while the
embedment into the column heads was only 152 mm (6 in.). Considering the horizontal testing configuration of the
columns, the specimens were cast horizontally in two batches with 44.8 MPa (6.5 ksi) and 31.0 MPa (4.5 ksi) 28-
day compressive strength. The ready-mix concrete achieved a slump of 254 mm (10 in.), equivalent of a pump mix,
to ensure proper placement of concrete along the entire length of the tubes.

A pedestal was designed to accommodate the height of the reaction frame and the actuator, and also to span over the
tie-down pattern. Each specimen is placed on top of the pedestal, and post-tensioned with eight threaded rods to the

556
strong floor through the pedestal and four threaded rods in the middle to the pedestal. All threaded rods are 25.4 mm
(1 in.) diameter. Figure 4 depicts the test set-up of specimen CL, i.e., the long column with carbon FRP tube. Tests
are currently in progress at FIU.

Table 1: Test Matrix of Single Column Experiments

Specimen Shear Span Core Diameter f’c Longitudinal Transverse

Name mm (in.) mm (in.) MPa (ksi) FRP FRP
RC (Control) 1,295 (51) 310 (12.2) 44.8 (6.5) None
Y (Yellow tube) 1,295 (51) 315 (12.4) 44.8 (6.5) 17 Layers of +55o E-Glass
G (Glass) 1,295 (51) 318 (12.5) 44.8 (6.5) 3 Layers of Bi-directional Glass
CS (Carbon,
1,295 (51) 318 (12.5) 31.0 (4.5) 2 Layers of Bi-directional Carbon
Short Column)
2 Layers of 3 Layers of
H (Hybrid) 1,295 (51) 318 (12.5) 31.0 (4.5) Unidirectional Unidirectional
Carbon Glass
CL (Carbon,
2,210 (87) 318 (12.5) 31.0 (4.5) 2 Layers of Bi-directional Carbon
Long Column)

Figure 3: Column Reinforcement for RC and CL Figure 4: Test Set-up for Carbon Long Column

3. TWO-COLUMN BENT TEST PROGRAM

A total of four (4) bent specimens are being prepared with three CFFT and one control RC (see Table 2). The three
CFFT bents are made of yellow tube, carbon and hybrid FRP, which are the same as those described in the single
column test program. In the bent specimens, the columns are 203 mm (8 in.) diameter, while the reinforcement is
scaled down accordingly. The FRP tubes are made using the same technique as that described for the single column
test program. The pier cap beams for all CFFT bents are cast into carbon FRP formwork, which in turn is made by
wrapping bi-directional carbon FRP fabric onto a wooden mold. Figure 5 shows the bent specimen design. Figure 6
shows the test layout for the two-column bent specimens. The lateral load is transferred to the pier cap beam through
an adapter and four rods outside of the pier cap beam. This phase is currently at the stage of specimen preparation.

Table 2: Test Matrix of Two-Column Bent Experiments

Specimens Columns Pier Cap Beam

RCF (Reinforced Concrete Frame) RC RC
CFF (Carbon FRP-Concrete Frame) Carbon FRP Carbon FRP
GFF (Glass FRP-Concrete Frame) Glass FRP Carbon FRP
HFF (Hybrid FRP-Concrete Frame) Hybrid FRP Carbon FRP

557
 Figure 5: Two-Column Bent Specimen Configuration

Figure 6: Plan View of Test Layout of Two-Column Bent Specimens

4. FOUR-SPAN BRIDGE MODEL TEST PROGRAM

A large-scaled four-span bridge model will be tested on three bi-directional shake tables at the University of Nevada,
Reno to assess the holistic performance of the proposed system. Figure 7 shows the tentative layout of the shake
table test of the bridge model.

Figure 7: Tentative Layout of Shake Table Test of the Bridge Model

5. REFERENCES
Seible, F., Burgueño, R., Abdallah, M. G., and Nuismer, R. (1996). “Development of advanced composite carbon
shell systems for concrete columns in seismic zones”, Proceedings of the 11th World Conference on Earthquake
Engineering, Pergamon, Elsevier Science Ltd. Oxford, England, Paper No. 1375.
Zhu, Z., Ahmad, I., Mirmiran, A. (2006). “Seismic Performance of Concrete-Filled FRP Tube Columns for Bridge
Substructure”, Journal of Bridge Engineering, ASCE, Vol. 11, No. 3, pp 359-370.

558
Third International Conference on FRP Composites in Civil Engineering (CICE 2006)
December 13-15 2006, Miami, Florida, USA

Experimental study of FRP wrapping for RC bridge under seismic loads

Éric St-Georges
(Teknika-HBA, Granby, Québec, Canada)

Nathalie Roy
(Ph.D. candidate, Université de Sherbrooke, Sherbrooke, Québec, Canada)

Pierre Labossière
(Professor, Université de Sherbrooke, Sherbrooke, Québec, Canada)

Jean Proulx
(Professor, Université de Sherbrooke, Sherbrooke, Québec, Canada)

Patrick Paultre
(Professor, Université de Sherbrooke, Sherbrooke, Québec, Canada)

ABSTRACT
An ongoing research project at Université de Sherbrooke has shown that fibre reinforced polymers (FRPs) enhance
ductility as well as flexural capacity and shear resistance. An experimental program was completed on six 2.15-m
high column specimens which were tested under combined axial and cyclic lateral loads. The objective of the
experimental project presented herein was to test a new strengthening configuration of reinforced-concrete bridge
columns using FRP wrapping. The retrofitted columns were subjected to cyclic lateral loads with two varying axial
load levels representative of service conditions for bridges and for building, respectively, as well as two wrapping
configurations, one based on “conventional” design and the other on displacement-based design. The results are
analysed in this paper in order to: (i) evaluate the extent of seismic-related damages in FRP retrofitted bridge
columns; (ii) compare the resistance and ductility of each column, before and after the retrofit; and (iii) compare the
efficiency of conventional and performance-based wrapping configurations. The paper provides a description of the
specimens, rehabilitation techniques, and testing apparatus. Results for three of the six columns under service axial
load conditions representative of bridge piers are presented and discussed.

KEYWORDS
CFRP, reinforced concrete bridge column, cyclic loading, performance-based design, seismic retrofit

1. INTRODUCTION
This study is part of a bridge pier seismic retrofit project currently underway at the Earthquake Engineering and
Structural Dynamics Research Centre (CRGP) of the Université de Sherbrooke. The objectives are to optimize a
retrofitting methodology of bridge columns with carbon fiber reinforced polymers and to evaluate the increase in
earthquake resistance by means of reverse-cyclic testing.

2. EXPERIMENTAL PROGRAM

Cyclic tests were performed on six large scale circular columns with properties similar to typical piers of short-span
bridge located in eastern Canada. The 2.15-m high columns have a 300-mm diameter and are embedded in a
massive stub that was anchored to a strong floor for the tests. The columns were cast vertically with a 30-MPa
compressive strength concrete. The specified steel yield strength for the longitudinal reinforcement and the

559
transverse reinforcement is 400 MPa. The longitudinal reinforcement volumetric ratio, Dl=2.5%, and the transverse
reinforcement volumetric ratio, Dh=0.8%, are typical of columns that were designed and constructed before recent
modifications to code requirements for minimum transverse reinforcement. The columns are thus expected to
dissipate very little energy during seismic events. The axial force (corresponding to 0.1 Ag f c' and 0.35 Ag f c' ) was
applied by means of 2 hydraulic actuators (see test setup on figure 1) and was maintained constant during the test.
Prescribed displacement histories were imposed on the column by means of a 500-kN double-acting displacement-
controlled dynamic-rated servo-hydraulic actuator reacting on a large-capacity vertical reaction wall. For the first
cycle, whose purpose was to crack the member and obtain elastic characteristics, the horizontal load reached 75% of
the expected yield load. The second cycle reached the yield load and the yield displacement. Each subsequent cycle
was repeated twice with a maximum displacement equal to 1.5, 2, 3, …, times the measured yield displacement up
to failure. The columns were fully instrumented with strain gauges. One set of four linear variable displacement
transducers (LVDTs) and two sets of four potentiometers were placed in the plastic hinge region on steel rings at
respectively 25 mm, 325 mm and 625 mm from the bottom of the column.

Figure 1: Test Setup

2.1 Performance Based vs Conventional Confinement

The performance-based retrofitting methodology of bridge columns with CFRP optimized in this project was
reported by Tian and Chaturvedi, 2004 and is based on performance criteria. In this project, the performance criteria
were the following: the retrofitted structure must meet prescribed ductility and drift requirements corresponding to
given seismic events having respectively low (1/2500 p.a.), medium (1/475 p.a.) and high (1/100 p.a) probability of
exceedance. The capacity spectrum method was utilized for this approach and the non-linear behavior of the
columns was accounted for. The conventional methodology consisted of a simplified relationship between ductility
capacity and column aspect ratio suitable for force-based design of simple structures proposed by Priestley et al.,
1996. The calculations to meet the required ductility level in each case led to two levels of confinement that
corresponds to: a) 2 layers of CFRP for the performance-based approach ( µ ∆ = 3.5 ), and b) 4 layers of CFRP for
the conventional approach ( µ ∆ = 5.0 ). The confinement was limited to a 635-mm length corresponding to twice the
estimated plastic hinge length. Table 1 contains the details of the experimental program for the three columns tested
under an axial load of 0.1 Ag f c' which corresponds to typical service load for bridge piers.

Table 1: Description of Specimens

'
Specimen Axial load ( % Ag fc ) # layers of CFRP
S1 10 0
S2 10 2
S3 10 4

560
3. TEST RESULTS
The left column of figure 2 shows the appearance of the specimens at the end of the test and the lateral load-tip
deflection responses are presented in the column on the right.

a)

b)

c)

Figure 2: Specimen after testing and lateral load-tip displacement diagram for columns with axial load
corresponding to 10% A f and: a) 0 layer of CFRP (S1), b) 2 layers of CFRP (S2) c) 4 layers of CFRP (S3)
g c
'

As defined in Paultre et al., 2001, seismic response indicators are generally quantified by curvature and structural
ductility and by energy-dissipation capacity. The indicators - namely the ultimate displacement ductility µ ∆u , the
ultimate curvature ductility µϕu , the maximum drift ratio δ θ u , the normalized dissipated energy EN and the work
index Iw proposed by Gosain et al., 1977 - are presented in Table 2 enabling a rational comparison of the columns
behavior.

561
 1 n
EN =
Fmax ∆ yI
∑E i
i =1
n
Fi ∆ i
Iw = ∑
i =1 Fmax ∆ yI
F is the applied horizontal load, ∆ is the tip displacement of the column, ∆ yI is the ideal yield displacement, Ei
is the energy dissipated for cycle i and n is the number of cycles before conventional failure. Results show that
structural ductility criteria were exceeded in both the performance-based and the conventional design. While the
sectional ductility is almost the same for the three columns, a significant increase in structural ductility and energy
dissipation for the confined columns is observed and this phenomenom is more pronounced between the columns
having 0 and 2 layers of CFRP.

Table 2: Summary of Results

Specimen µ∆u µϕu δθ u (%) EN Iw

1
S1 3.5 7.7 8.89 13.56 11.90
S2 5.1 8.0 12.88 34.18 25.05
S3 5.8 7.6 11.72 37.69 31.98
1
calculated at 325 mm from column’s base

4. CONCLUSION
Six large-scale reinforced concrete circular columns were tested under cyclic loading. The confinement varied from
0 to 2 and 4 layers of CFRP. The columns were subjected to constant axial loads corresponding to 10 and 35 % of
the columns’ axial load capacity and to a cyclic bending moment. It is shown that, at constant level of axial
compression of 10% of the column capacity, the confinement influences the flexural behavior of the columns and
significantly improves the seismic behavior. It is of interest to mention that, in a next phase of the project, the test
results will be used to validate a new confinement model that was developed by Eid et al., 2006. This new model
features a unique prediction of the confinement effect due to both the transverse steel reinforcement and the FRP.

5. ACKNOWLEDGEMENTS
The authors would like to acknowledge the financial support of ISIS Canada, the Natural Sciences and Engineering
Research Council of Canada (NSERC), the Quebec Fonds pour la recherche sur la nature et les technologies
(FQRNT) and Sika-Canada. They would also like to thank Marc Demers, Claude Aubé, Sébastien Gauthier and
Laurent Thibodeau at the Université de Sherbrooke for their collaboration.

6. REFERENCES
Eid, R., Roy, N. and Paultre, P. (2006). “Behaviour of Circular Reinforced Concrete Columns Confined with
Transverse Steel Reinforcement and Fiber-Reinforced Composite Sheets”, Proceedings of the 2nd Intl FIB Congress.
Gosain N.K., Brown, R.H. and Jirsa, J.O. (1977). “Shear Requirements for Load Reversals on RC Members”.
Journal of the Structural Division, Proceedings of the ASCE, Vol. 103, No. ST7, pp. 1461-1475.
Paultre, P., Légeron, F. and Mongeau, D. (2001). “Influence of Concrete Strength and Transverse Reinforcement
Yield Strength on Behavior of High-Strength Concrete Columns”. ACI Structural Journal, Vol. 98, No. 4, pp. 490-
501.
Priestley, M.J.N., Seible, F. and Calvi G.M. (1996). “Seismic Design and Retrofit of Bridges”. John Wiley & Sons,
New York, 686 p.
Tian, Y. and Chaturvedi, S.K. (2004). “A Seismic Retrofit Design Methodology for R/C Bridge Columns Using
Fiber Composites”. Earthquake Spectra, Vol. 20, No. 2, pp 483-502.

562
Third International Conference on FRP Composites in Civil Engineering (CICE 2006)
December 13-15 2006, Miami, Florida, USA

USE OF CFRP TO STRENGTHEN POORLY DETAILED REINFORECED

CONCRETE BEAMS
InSung Kim
(Graduate Reaserch Assistant,The University of Texas at Austin,USA)

James O. Jirsa
(Professor, The University of Texas at Austin,USA)

Oguzhan Bayrak
(Assistant Professor, The University of Texas at Austin,USA)

ABSTRACT
In many reinforced concrete structures built in the 1970’s and earlier, bottom beam reinforcement is not continuous
and if a column support is lost due to terrorist attack or other unexpected action, the structure could be vulnerable to
progressive collapse. The beams may not develop catenary action if the reinforcement is not continuous. The use of
CFRP material may provide a solution for rehabilitating such structures. CFRP materials can not develop full
tensile capacity unless they are properly anchored to the reinforced concrete structure. The intent of this study is to
find an effective method of anchoring CFRP material to a reinforced concrete beam so that the ultimate tensile
strength of the CFRP is realized. In this study, four reinforced concrete beams rehabilitated using different
configurations of anchors were tested to assess the effectiveness of the anchors. All the beams were undamaged
before they were rehabilitated. Both CFRP anchors and CFRP U-wraps were investigated. The rehabilitated beams
were loaded until failure of the CFRP material or anchor occurred. Different failure modes, strengths and
deformation capacities of the rehabilitated beams were observed depending on the configurations of anchors. The
maximum capacity was obtained using a combination of CFRP anchors and U-wraps.

KEYWORDS
CFRP sheets, CFRP anchors, reinforced concrete, rehabilitation

1. INTRODUCTION
The basic rehabilitation technique for this program was installation of CFRP materials on the sides of the beam to
provide continuity to bottom reinforcement (Figure 1). A CFRP sheet was attached to the concrete surface by epoxy
resin and through the use of CFRP anchors and CFRP U-wraps in order to develop full tensile capacity of CFRP
sheet after delamination of CFRP sheet occurs. This work was supported under a grant from the NSF.

Span A Span B Span A Span B

Strengthening by CFRP Continuity with CFRP

Discontinuity in Portion of structure

bottom reinforcement represented by the test specimens
Column removed due
to blast or impact

Figure 1: Rehabilitation Technique

563
2. TEST SPECIMENS
The dimensions of the beams tested are shown below. The length of specimen corresponded to the distance between
inflection points. The 12 in. long deepened section at mid span represented a portion of the supporting column.
Bottom beam reinforcement extended 4.5 in. from column face with a 3 in. discontinuity existed at the middle of the
column (Figure 2).

All specimens were loaded at “mid span” to represent the loading when the column below is removed (Figure 2).
The target strength was P=32 kip which corresponded to nominal strength of the beam with two #6 continuous
bottom bars. If full tensile capacity of the CFRP sheet can be developed, the target strength will be achieved.
Applied load and deflection of the column (center of the specimen) were measured.

P
CFRP U-wrap
96 in.
8 in.
7 in. 12 in.

16 in.

2 in.

CFRP sheet CFRP anchor

3 in. fc’=3,500 psi
P/2 P/2 5-#6 reinforcement, GR60
66 in. #3 ties @ 7 in

Figure 2: Test Specimen (Specimen No. 4)

3. APPLICATIONS OF CFRP MATERIALS

A CFRP sheet was used as a tensile element to provide continuity to the bottom reinforcement (Figure 3-(a)). CFRP
anchors are shown in Figure 3-(b) and consist of a roll of CFRP sheet inserted into the concrete and splayed out over
the CFRP sheet on the side face of beam. CFRP U-wraps were sheets of CFRP attached transverse to the main sheet
as shown in Figure 3-(c).

(a) CFRP sheet (b) CFRP anchor (c) CFRP U-wrap

Figure 3: Applications of CFRP Material

4. MATERIAL PROPERTIES
Compressive strength of concrete was 3,500 psi, and GR60 reinforcements are used for the tests. Material
properties of CFRP are shown in Table 1.

Table 1: Material Properties of CFRP

Ultimate Tensile Elongation Tensile Laminate

Properties
Strength at Break modulus thickness
Typical Test Value 143,700 psi 1.26 % 11.4 x 106 psi 0.035 in.
* Tyfo® SCH-35 composites with Tyfo® S Epoxy

564
5. REHABILITATION METHODS USING CFRP MATERIALS
All CFRP materials were installed in symmetric way on both sides of the beams. The CFRP was attached using two
different sheet arrangements. In one case, a CFRP sheet 5.5 in. wide x 66 in. long was attached on the both sides of
the beam. In the other case, two CFRP sheets 2.75 x 66 in. were placed on top of one another. The area of CFRP
materials in both cases was the same, and was selected to provide an ultimate tensile force the same as that at yield
using two #6 bottom bars. For the specimen with one CFRP sheet, only epoxy resin was used to attach it to the
beam. Two layers of CFRP were used in the other three specimens.

A CFRP anchor was made using the same width of CFRP, 5.5 in., sheet that was attached to the beam. The length
of the anchor was 9.5 in. with 4 in. of the anchor inserted into a 5/8 in. hole drilled into the concrete, and the rest of
the anchor was spread out in a fan shape on the CFRP sheet. Eight anchor holes were installed in each beam, four
located at 12 in., and the other four located at 24 in. from the center of the column.

CFRP U-wrap was also made of the same 5.5 in. width of CFRP sheet, and a total length of 26 in.. The CFRP U-
wrap was attached on each side of the beam over the CFRP sheet, and extended 9 in. from the bottom face of the
beam. Four CFRP U-wraps were installed in a specimen, and two of them were located 10.5 in. and the others were
located 22.5 in. from center of the specimen.

Four different rehabilitation methods using CFRP sheet, CFRP anchor and CFRP U-wrap are shown in Table 2.

Table 2: Rehabilitation Methods

Number of
Maximum
Configuration of CFRP Layers of Type of
No. Failure Mode Applied
Materials CFRP Anchorage
Load
sheet
Delamination of
1 5.5 in.
1 layer None 14.57 kip
66 in. CFRP sheet
5.5 in.
9 in.
2.75 in. Delamination of
2 2 layers CFRP U-wrap 15.38 kip
10.5 in. 22.5 in.
CFRP U-wrap
66 in.

5.5 in. Anchor hole

Concrete
2.75 in.
3 12 in. 24 in. 2 layers CFRP Anchor Failure around 25.78 kip
66 in. Anchor Holes
CFRP Anchor
2.75 in.
Fracture of
4 66 in. 2 layers & CFRP U- 31.94 kip
CFRP Sheet
* Location of the anchors and U-wraps are the same
as specimen No. 2 and No. 3
wrap

(a) Specimen No.1 (b) Specimen No.2 (c) Specimen No.3 (d) Specimen No.4
Figure 4: Failure Modes of Specimens

6. TEST RESULTS: FAILURE MODES

Failure mode of each specimen is shown in Table 2 and Figure 4. The most desirable failure mode is fracture of the
CFRP sheet after developing full tensile capacity. However, failure of the anchorage system occurred prior to
failure of CFRP sheet in all the specimens except No. 4 which had both CFRP anchors and CFRP U-wraps.

565
One layer of CFRP sheet was only used in specimen No.1 in order to maximize surface area of CFRP sheet attached
to the concrete because it did not have anchors, so the area is a critical factor for bonding capacity of CFRP sheet to
concrete. Delamination of CFRP sheet was the failure mode of specimen No.1 (Figure 4-(a)). Although CFRP U-
wraps were installed in specimen No.2 to provide anchorage to the CFRP sheet, they did not hold the CFRP sheet
effectively after delamination of the sheet occurred. CFRP U-wraps delaminated right after delamination of CFRP
sheet (Figure 4-(b)). CFRP anchors in specimen No.3 held CFRP sheet to concrete effectively after delamination of
CFRP sheet occurred. However, concrete cover around anchor hole crushed before any material failure of CFRP
materials (Figure 4-(c)). In specimen No. 4, CFRP U-wrap was installed on CFRP sheet before CFRP anchors were
applied to prevent concrete crushing around the anchor hole by confining concrete around anchor hole. Full tensile
strength of the CFRP sheet was developed in specimen No. 4, and material failure of the CFRP sheet was observed
in specimen No. 4 (Figure 4-(d)).

7. TEST RESULTS: LOAD-DEFLECTION RELATIONSHIPS

As shown in Figure 5, not only strength but also deformation capacity increased an anchorage improved. Strength
of specimen No. 4 was 31.94 kip (target strength). The CFRP sheet developed its full tensile capacity with a
combination of CFRP anchor and U-wrap.
load (kip)
35
31.94 kip

30

25.78 kip
25

20

14.57 kip 15.38 kip

15

10

0
0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2
deflection (in.)

Figure 5: Load-Deflection Relationships

8. CONCLUSION
Rehabilitation using CFRP materials to correct poor details was successful only if proper anchorage was provided.
It was necessary to use both CFRP anchors and CFRP U-wraps to achieve full strength of the CFRP sheet. If either
CFRP anchors or CFRP U-wraps were used alone, delamination of the sheet leads to failure of anchor or U-wrap.

Further research into various geometries and quantities of CFRP anchors and U-wraps are required before design
guidelines can be developed. Moreover, application may be limited because the method requires flat side surfaces
and many beam-column connections consist of columns that are wider than the beams. However, it is possible to
use the method in other applications in which CFRP sheets are used to provide continuity in the same plane.

9. References
ACI committee 440. (2002). “Guide for the Design and Construction of Externally bonded FRP Systems for
Strengthening Concrete Structures”, American Concrete Institute, Farmington Hills, Michigan.
Brena, S. (2000), “Strengthening Reinforced Concrete Bridges Using Carbon Fiber Reinforced Polymer Composite”,
Ph.D. dissertation, University of Texas at Austin, Texas, USA
Kobayashi, K., et al., (2001). “Advanced Wrapping System with CF-Anchor”, Proceedings of the 5th International
Conference on Fibre Reinforced Plastics for Reinforced Concrete Structures, Volume 1, pages 379-388
Ozdemir, G., Akyuz, U. (2006) “Tensile Capacities of CFRP Anchors”, Advances in Earthquake Engineering for
Urban Risk Reduction(Proceedings of the NATO Science for Peace Workshop on Advances in Earthquake
Engineering for Urban Risk Reduction, Istanbul, Turkey), Springer, pages 471-488
Teng, J.G., et al., (2002). FRP-strengthened RC Structures, Willy, New York

566
Third International Conference on FRP Composites in Civil Engineering (CICE 2006)
December 13-15 2006, Miami, Florida, USA

RESIDUAL BEHAVIOR OF FRP RETROFITTED RC COLUMNS

AFTER BEING SUBJECTED TO EARTHQUAKE DAMAGE
Shan, Bo
(Research Assistant,Civil Engineering College of Hunan Universty, Changsha, Hunan Province, China)

Xiao, Yan
(Associate Professor, University of Southern California, Los Angeles, California, USA)
(Cheung Kong Scholar , Civil Engineering College of Hunan Universty, Changsha, Hunan Province, China)

Huang, Zhengyu
(Professor, Civil Engineering College of Hunan Universty, Changsha, Hunan Province, China)

ABSTRACT
Despite the significant importance, residual seismic behavior of reinforced concrete (RC) columns retrofitted with
fiber-reinforced plastic (FRP) has received little attention. In this paper, the residual performance of FRP retrofitted
columns damaged after simulated seismic loading is studied. Eight model columns with a shear aspect ratio of 5.0
were tested first under cyclic lateral force and a constant load equal to 20% of the column gross axial load capacity.
The main parameters considered were the type of FRP jacket and peak drift ratio where the lateral loading was
interrupted. Five of the model columns were subjected to long-term axial loading after subjected to limited damage
by lateral cyclic loading. The deformation of retrofitted columns under long-term axial loading related to the
previous damage intensity and the modulus of FRP. The post damage long-term axial strain corresponding to time
increment can be simply estimated using the ACI 209 creep model. The effective creep Poisson’s ratios of the
retrofitted columns were much smaller than the as-built column but identical for GFRP and CFRP retrofitted
columns. Under the testing condition of this study, the long-term axial deformation of retrofitted column tends to be
stable, despite subjected simulated earthquake damage.

KEYWORDS
Seismic retrofit; FRP; Earthquake damage; Residual performance; Creep

1. INTRODUCTION
Performance of bridges and buildings during past earthquakes such as Loma Prieta (1989), Northridge (1994) and
Kobe (1995) repeatedly demonstrated the vulnerabilities of older reinforced concrete structures. For reinforced
concrete buildings and bridges, a significant majority of the structural failure could be attributed to inadequate
seismic design of the columns, such as the lack of transverse reinforcement and the poor splicing details of
longitudinal reinforcement in potential plastic hinge regions. Recently, fiber reinforced plastics (FRP) have been
widely studied and found growing applications, due to the expectation of its high-strength, light-weight, quick and
easy installation on site, high resistance against corrosion and easy fabrication.
Numerous existing structures, particularly bridges in California and elsewhere, have been seismically upgraded.
Retrofit design is based on the philosophy of enabling the structures to survive essentially one extreme seismic event.
However, several questions related to the post-earthquake health of the structure in relation to the residual behavior
of the retrofitted bridge columns remain unanswered. These issues are of significant importance. It has been
suspected that the rupture of several GFRP jackets installed on the columns of the I-5 and Freeway 2 interchange in
Los Angeles was due to the insufficient residual capacity after the columns were subjected to the shaking of the
1994 Northridge earthquake. In this case, prestressing the FRP for active confinement of the concrete also
compounded the problem (Hipley 2004). The primary focus of this paper is on the sustained long-term axial loading
behavior of FRP retrofitted columns after being subjected to limited damage by simulated earthquake loading.

567
Research on the long-term performance of FRP retrofitted columns is rare. Naguib and Mirmiran (2002) studied the
time-dependence behavior of concrete column retrofitted with FRP jacket under axial load. It should be indicated
that while the long-term loading used in this study is similar to that used by Naguib and Mirmiran (2002), the
research objectives are distinctly different. The test of Naguib and Mirmiran (2002) mainly studied the creep of FRP
confined concrete columns. But the main objective in this paper is emphasis on the residual performance of
retrofitted columns with earthquake damage.

2. EXPERIMENTAL PROGRAM
Eight RC model columns were designed to simulate the 1970’s design and detailing practice according to the
specifications for RC bridges in China. Similar details are also seen in older bridges in the United States and
elsewhere. The typical specimen was designed as a vertical cantilever column with a strong footing, as shown in
Figure 1. The testing matrix is shown in Table 1. FRP jackets were wrapped for a 400mm long segment near the
column bottom end. The FRP retrofit together with the existing transverse reinforcement provides the columns with
an equivalent transverse reinforcement slightly exceeding the current ACI 318 code requirement.

Table 1: Testing Matrix

f c’ Lateral
Specimen Retrofit Testing Condition
(MPa) Displacement Ratio
CA-1 34.0 － 6% Lateral Loading Only
CA-2L 35.0 － 3% Lateral Loading and Long-term Test
CCR-1 30.9 4-layer CFRP 8% Lateral Loading Only
CCR-2L 38.6 4-layer CFRP 3% Lateral Loading and Long-term Test
CCR-3L 41.4 4-layer CFRP 6% Lateral Loading and Long-term Test
CGR-1 38.7 5-layer GFRP 10% Lateral Loading Only
CGR-2L 34.9 5-layer GFRP 3% Lateral Loading and Long-term Test
CGR-3L 37.9 5-layer GFRP 6% Lateral Loading and Long-term Test

During pseudo-static test, a constant axial load, equal to 20% of the static capacity Agfc’, was applied to the column.
The lateral force was cycled under a lateral displacement control condition. Lateral loading program is shown in
Figure 2. Three model columns, representing the as built condition, GFRP and CFRP retrofits, respectively, were
first tested to failure to establish the benchmark testing data.
Steel beam

12
10 Lateral loadingterminatedfor creep High-strength steel rod
Lateral force
8 test for CCR-3LandCGR-3L Hydraulic jack Vibrating chord sensor
6
Drift ratio(%).

1 1 4
2
0 Test column
-2
FRP jacket Measuring staff for
-4 Lateral loadingterminatedfor axial deformation
Strain guage locations -6 creeptest for CA-2L, CCR-2L,
-8 andCGR-2L Measuring staff for Footing stub
Footing Stub
-10 radial deformation
Width=700mm
-12
0 2 4 6 8 10 12 14 16 18 20 22 24
Cycle

Figure 1:Details of Test Column Figure2: Lateral Loading Program Figure 3:Long-term Test Setup

Five sets of long-term axial loading setups as shown in Figure 3 were manufactured for such testing purpose. The
test setup included a 2000kN hydraulic jack, a cross beam and two high-strength steel rods with a diameter of 50mm.
The fluctuation of the load application was monitored using vibrating wire gauges, which were mounted on the
high-strength steel rods. Several measurement target staffs were fixed on the test columns as shown in Figure 3.
Through measuring the varying distance of the measuring staffs on test column using a manual dial gauge, the radial
deformation and axial deformation of the test column can be obtained. All the specimens were initially loaded at an
axial load level equal to 0.2 Agfc’, and observed for 30 days when the deformation became sufficiently stable. The
axial load was then removed and reapplied at a higher axial load level of 0.4 Agfc’, for a period of 60 days. This

568
adjustment of axial load level was for an attempt to intensify the effects of axial loading, and did not necessarily
reflect any actual loading condition.

3. EXPERIMENTAL RESULTS
3.1 Presentation of Data of Long-Term Axial Loading Tests

The characteristic points of Figure 4 and Figure 5 such as A, B, C, D, E, F, G and H denote the strain value after
applying axial load, the strain value at the end of pseudo-static test, the strain value removal at axial load, the strain
value at the start of long-term loading test, the strain value at the 30th day, the strain value at the end of empty load,
the strain value after applying 0.4 axial load ratio and the strain value at the 100th day, respectively.
0.3
1st Stage 2nd 3rd Stage 1st Stage 2nd 3rd Stage

Effective Creep Poisson's Ratio

4000 Stage 2000 Stage
H
0.25
G
H G
3000 1500 0.2
F
D
Strain ( 10 )

E F
-6

Strain ( 10 )
B D E 0.15

-6
2000 C 1000 B CA-2L
C
0.1 CCR-2L
CCR-3L
1000 500 0.05 CGR-2L
A CGR-3L
A 0
0 0
0 20 40 60 80 100 0 20 40 60 80 100
0 20 40 60 80 100
Time(d) Time(d) Time (day)

Figure 4:Axial Strain of CA-2L Figure5: Axial Strain of CCR-3L Figure 6:Effective Creep Poisson’s Ratio

3.2 Effective Creep Poisson’s Ratio

From circumferential strain and axial strain, the effective creep Poisson’s ratio over time can be obtained and shown
in Figure 6. Note that the effective creep Poisson’s ratios are calculated based on the net creep. It is interesting to
note that the trends of the effective creep Poisson’s ratios for all model columns are very stable during the entire
period of long-term axial loading, except for the interruption for changing the axial load level. There is little
difference among the effective creep Poisson’s ratios of the retrofitted columns using CFRP or GFRP jacketing. This
is considered as the consequence of an equal target strength design with different jackets. The average value of the
effective creep Poisson’s ratios of the as built model CA-2L is approximately 0.23, which is higher than the average
value 0.15 for the FRP retrofitted columns. This reflects the confinement effect of FRP jacketing. Moreover, the test
results also indicate that despite having experienced different levels of lateral loading damage, the effective creep
Poisson’s ratios are identical.

3.3 Axial Deformation under Post-damage Long-term Loading

The axial strain developments of CA-2L and CCR-3L are shown in Figure 4 and Figure 5, respectively. The axial
strain stabilized after approximately 7 days during the first 30 days of axial loading. However, the axial strains took
a longer period of about 20 to 30 days to cease the increase and stabilize under the higher axial load level of 0.4 Agfc’.
Thereafter, because there was no significantly meaningful time dependant augmentation, the loading was terminated
after 60 days.
Behaviors during the first 30 days for all the CFRP retrofitted columns, GFRP retrofitted columns and the as built
column are shown by the solid curves in from Figure 7 to Figure 11, respectively, along with the creep models based
on ACI 209. Comparing Figure 8 with Figure 7, it can be seen that the axial strain of CCR-3L, which experienced
the lateral cyclic loading up to a drift ratio of 6%, was about 15% higher than its counterpart model CCR-2L which
experienced lateral loading up to a drift ratio of 3%. This indicated that post damage long-term deformation was
influenced by the damage level. On the other hand, by comparing the axial deformations of CFRP retrofitted
columns shown in Figure 8 with the results of GFRP retrofitted columns shown in Figure 10, it can be observed that
the long-term deformation was also affected by the type of FRP jacket, essentially meaning the modulus of elasticity
of the FRP jacket. Despite having experienced identical level of lateral loading, the GFRP retrofitted column
developed larger strain compared with the counterpart CFRP retrofitted column.
From Figure 11, the lower limit curve based on the ACI 209 creep model approaches the experimental curve for the
as-build column. The experimental curves of retrofitted columns are included in the range between lower limit and
the upper limit of creep coefficient Cu, and reasonably close to the curves using the average value, Cu = 2.35, as
shown in from Figure 7 to Figure 10. However, the predicted curves using average value of Cu tend to over-estimate
the test results for later loading stages beyond 15 days.

569
 2000 2000

1800 1800 Axial strain

Axial strain ACI 209
1600 ACI 209 1600
Cu=4.15

Strain(10 )
-6
1400 1400
Cu=4.15

Strain(10 )
-6
Cu=2.35
1200 Cu=2.35 1200

1000
Cu=1.3
Cu=1.3 1000

800 800

600 600
0 5 10 15 20 25 30 0 5 10 15 20 25 30
Time(day) Time (day)

Figure 7: CCR-2L vs. ACI 209 Figure8: CCR-3L vs. ACI 209

2000 2000 5000

1800 Axial strain

Axial strain 1800
ACI 209 Axial strain
ACI 209 4500
ACI 209
1600 1600
Cu=4.15

Strain (10 )
Cu=4.15 4000
Strain (10 )

-6
1400 1400 Cu=2.35
-6

Strain(10 )
-6
Cu=4.15
Cu=2.35
1200 1200
Cu=1.3 3500
Cu=1.3 Cu=2.35
1000 1000
Cu=1.3
800 800 3000

600 600
2500
0 5 10 15 20 25 30 0 5 10 15 20 25 30 0 5 10 15 20 25 30
Time (day) Time (day) Time(day)

Figure 9: CGR-2L vs. ACI 209 Figure10: CGR-3L vs. ACI 209 Figure 11:CA-2L vs. ACI 209

4. CONCLUSIONS
During long-term axial loading test after being subjected to some degree of lateral loading damage, the deformation
of the FRP retrofitted columns was lower when compared with the non-retrofitted column subjected to similar
loading condition. The development of long-term loading deformation of retrofitted column was found to relate to
the previous damage level and the modulus of elasticity of the FRP. The post damage long-term axial strain
corresponding to time increment can be simply estimated using the simple creep model recommended by ACI
Committee 209 committee. The effective creep Poisson’s ratio for the retrofitted columns with various testing
conditions has an average value of about 0.15.The study essentially confirmed that within the testing range of the
axial load ratios between 0.2~0.4 Agfc’, the post damage long-term axial loading would not cause any significant
creep effects to the FRP retrofitted columns.

REFERENCES
ACI Committee 209, (1992). “Prediction of Creep, Shrinkage and Temperature Effects in Concrete Structures (ACI
209R-92)”. American Concrete Institute, Farmington Hills, Mich., 47pp.
ACI Committee 318 (2002). “Building Code Requirements for Reinforced Concrete and Commentary (ACI 318-
02/ACI 418R-02) ”. American Concrete Institute, Farmington Hill.
Hipley, P. (2004). “Seismic Retrofit for Columns/Piers Composite Bridge Column Retrofitting in California”.
Federal HighwayAdministration, FHWA, Bridge Technology,June,http://www.fhwa.dot.gov/bridge/frp/frpdatar.htm
Naguib, W. and Mirmiran, A. (2002). “Time-dependent behavior of Fiber-Reinfoced Polymer confined columns
under axial loads”. ACI Structural Journal, V.99, No.2, pp.142-148.
Saadatmanesh, h.; and Ehsani, M.R., (1994), “Strength and ductility of concrete columns reinforced with fiber
composite straps,” ACI Structural Journal, Vol.91, No.4, pp.434-447.
Xiao, Y.(1997). “Seismic retrofit of RC circular columns using prefabricated composite jacking”. ASCE, Journal of
Structural Engineering, Vol.123, No.10, pp.1357-1364.
Xiao, Y. and Wu, H. (2000). “Compressive Behavior of Concrete Cylinders Confined by Carbon Fiber Composite
Jackets”. ASCE, Journal of Materials in Civil Engineering, Vol. 12, No. 2, May, pp. 139-146.
Xiao, Y.; Sheikh, S.A.; and Li, Z.X. (2001), “Applications of FRP composites in concrete columns,” Proceedings
FRP Composites in Civil Engineering, Hong Kong, China, December, pp.731-740.
Xiao, Y. (2004). “Applications of FRP Composites in Concrete Columns”. International Journal of Advances in
Structural Engineering, A Multi-Science Publication, UK, Vol.7, No.4, August, pp.335-344.

570
Third International Conference on FRP Composites in Civil Engineering (CICE 2006)
December 13-15 2006, Miami, Florida, USA

ESTIMATION OF THE DUCTILITY OF WEB-BONDED FRP BEAMS

FOR ASSESSMENT OF STRENGTHENED RC EXTERIOR JOINTS
S. S. Mahini
(Asst. Professor, Department of Civil Engineering, Yazd University, Yazd, Iran)

H.R. Ronagh
(Senior Lecturer, Division of Civil Eng. School of Engineering, The University of Queensland,
Brisbane, QLD 4072, Australia)

ABSTRACT
Beam-column joints in Reinforced Concrete (RC) moment resisting frame (MRF) subjected to lateral loads can be
critical regions. They must therefore be designed adequately in order to dissipate large amounts of energy without a
significant loss of strength and ductility. When the joint are inadequately designed or damaged under unanticipated
loads, web-bonded FRP is one of the possible strengthening methods that can be used. In this paper, a computer
model is presented in order to estimate the ductility of strengthened beams and a range of possible design charts is
produced to select the type and amount of FRP required for upgrading exterior beam-column joints. Finally,
displacement ductility of a subassembly is assessed using these design charts and the result is compared with
experiment. The comparison proves the model is acceptable.

KEYWORDS
Ductility, wed-bonded, RC, joint, FRP

1. INTRODUCTION
Since 1970s, a significant amount of research studies are carried out studying various aspects of the behaviour of
beam-column joints in moment resisting frames (MRF) subjected to severe lateral loads. Shear and flexural failure
and penetration of cracks into the joints core have been identified as failure mechanisms resulted from inadequately
designed joints (Paulay and Priestley, 1992). Rehabilitation may become necessary when members and joints lack
sufficient strength, or ductility. Strengthening the beam end of the joints to improve the joint strength is an area of
research that has been over looked. Traiantafillou and Plevris (1992) studied the strengthening of RC beams with
epoxy-bonded fibre-composite materials including CFRP and GFRP sheets. They considered several failure
mechanisms such as FRP rupture, steel yield, concrete crushing and de-bonding using an analytical model.
Traiantafillou and Plevris produced diagrams showing the beam design for which each failure mechanism is
dominant and examined the effect of FRP sheets on the ductility of strengthened components. They confirmed their
analysis with testing beams under four-point bending. Traiantafillou and Plevris concluded that unless failure is
governed by FRP rupture, the results by no means indicate that GFRP should be preferred to CFRP for ductility
considerations due to the fact that the GFRP ratio needed to achieve a given moment capacity is much higher than
when CFRP composite is used. Duthinh and Starnes (2001) conducted an experimental study on seven pre-cracked
concrete beams reinforced internally with steel and externally with an FRP laminate under four-point bending.
Duthinh and Starnes concluded that in comparison with a beam reinforced heavily with steel only, CFRP-
strengthened RC beams have adequate deformation capacity in spite of their brittle failure.
In this paper, computer models are presented in order to estimate curvature ductility of web-bonded FRP beams.
This model can be used to assess the ductility of the strengthened exterior beam-column joints. Design charts are
also obtained to facilitate the calculations. Displacement ductility of a beam-column subassembly retrofitted by web
bonded FRP, tested by Mahini et al. (2004), is then obtained and the result is compared with the design charts in
order to show their reliability.

571
2. COMPUTER MODELS

The cross-section of a typical rectangular RC beam strengthened with web-bonded FRP is shown in Fig. 1. As is
seen, the section has a width b and a height h. The fibre composite has a thickness t f ; the tensile reinforcement has
an area, As and compressive reinforcement an area, As′ . d and d ′ are distances from the extreme compression
fibre to the centroid of tension and compression reinforcements respectively. Modulus of elasticity of steel, concrete
and FRP are denoted as E s , E c and E f respectively. Steel bars have a yield stress and strain of f y and ε y
respectively. The concrete has a compressive strength, f c′ and an ultimate compressive strain ε cu . The ultimate
stress and strain of fibre composite at rupture are f fr and ε fr respectively.
This method of strengthening (web-bonded FRP sheets) can be used: 1) for controlling the location of plastic hinge
by upgrading the strength of the beam close to the column which may help in moving the plastic hinge away from
the edge of the column and 2) for repair of moderately damaged rectangular RC beams in moment resisting frames.
There, the best way to repair the damaged beam is by web-bonded FRP, since wrapping the top of the beams in
existing buildings is hard to apply due to the presence of the floor slab.
Usually, one of three major flexural failure modes may be observed in reinforced concrete beams strengthened with
web-bonded FRP reinforcement. These are: 1) Yielding of reinforcing steel in tension followed by the rupture of
FRP (FRP rupture); 2) Yielding of reinforcing steel in tension followed by concrete crushing (tension failure) and 3)
Crushing of concrete in compression before yielding of the reinforcing steel (compression failure). In these modes,
the quantity of energy absorbance of the web-bonded FRP beam can be evaluated through its curvature ductility.
In order to simplify the analysis of web-bonded FRP beams in different failure modes and to calculate the first yield
neutral axes of the beam, a computer procedure has been developed that is presented in the following. The procedure
for calculating the depth of neutral axis of the beam section (when first yield occurs, c y and also at ultimate
moment, c) starts with the estimation of the neutral axis depth and ends with a calculated value for c y and c based
on the force equilibrium using the constitutive laws of the materials and the strain compatibility in the section. If the
difference between the estimated and calculated values is small, the estimated values are accepted otherwise
modified using the bisection method until convergence occurs. For the first yield calculations, compatibility of the
strain and stress as is shown in Fig. 1 is used. Based on the yield strain of steel reinforcement ε y , the strain in the
concrete, FRP and the compressive steel reinforcements can be determined. The maximum curvature ductility is
expressed as μφ = φ u / φ y , where φ u the maximum curvature is expected to be attained or relied on and φ y is the
yield curvature and can be calculated from strain distribution as shown in Fig. 1.

εc fc εcu .85 fc´

A ś d´ εs´ fs´ εs´ β1 c fs´
φ
cy c φ
y u
h d
h-cy h-c
As εy fy εy fy
εf ff ε fr f fr
tf b tf
(a) (b)

Figure 1: Strain and stress distributions of web-bonded FRP beams (a) at first yield (b) at ultimate

The stresses in the concrete, FRP and the compressive steel reinforcements are f c = E c ε c , f f = E f ε f
and f s′ = E s ε s′ respectively, where εc and ε f are the strains in the extreme fibre of concrete and FRP respectively
and the ε s′ , is the strain at compressive steel level.
Applying the force equilibrium in the section leads to:

c y = ( As f y − As′ f s′ + f f (h − c y )t f ) / 0.5 f c b (1)

572
In order to calculate the neutral axes at ultimate, the strain distributions shown in Fig. 1 is used. The following
criteria are then checked in order to obtain the mechanism governing the failure. If ε fr + ε f o < ε cu (h − c) / c then
the failure is governed by “FRP rupture”. Otherwise the failure is governed by “tension failure”. In that, ε fr and
ε f o are the ultimate strain of FRP and initial strain of unstrengthened beam at the extreme tensile level of the RC
section respectively and ε cu is the compressive strain of concrete at ultimate. When failure is governed by “FRP
rupture”, the strain in the FRP at failure is equal to the maximum ultimate strain, ε fr and therefore the strain values
at the compressive concrete, tensile and compressive steel reinforcements can be determined based on this strain
level. The stresses in the steel reinforcements may be calculated assuming an elastic perfectly plastic condition and
the stress in the FRP, f f may be taken as the ultimate tensile strength, f fr . The process when the failure is
governed by “tension failure” is similar to FRP rupture mode. In that, the maximum strain of concrete is equal
to ε cu . Based on the concrete strain level and the assumed value of c, the strains in the tension and compression steel
reinforcements can then be determined. The stresses in the steel reinforcements can be calculated similar to the FRP
rupture mode. The FRP behaves linearly to failure, therefore f f = E f ε f .
Using force equilibrium into the section leads to the following equation:

c = ( As f y − As′ f s′ + f f (h − c)t f ) / 0.85 f c′β 1b (2)

3. SAMPLE DESIGN CHARTS

Based on the presented computer model and the trail and error procedure, design charts are developed using a
computer program. The concrete had a compressive strength around 48.20 MPa and a modulus of elasticity around
35.16 GPa. Yield strength of the main steel reinforcements was around 500MPa and the modulus of elasticity was
equal to 200 GPa. CFRP and GFRP possessed a tensile strength of about 3900 MPa and 1700 MPa, a modulus of
elasticity of 240GPa and 65 GPa, and an ultimate tensile elongation of 1.55% and 2.80% respectively. The
maximum dependable concrete compression strain in the extreme fibre of unconfined beam sections is assumed to
be 0.003, when normal strength concrete ( f c′ < 45 MPa) is used.
The effect of the FRP reinforcement ratio, ρ f = t f h / bd , on the curvature ductility of the member is shown in Fig.
2 for a wide range of tensile steel reinforcement ratios, ρ s = As / bd and the compressive steel ratio, ρ s′ = As′ / bd =
0.0067. As is seen, when CFRP sheet is used, the ductility increases as ρ f decreases, unless ρ f becomes small
enough so that failure is caused by FRP rupture. In the case of GFRP-strengthening, the tension failure is governed
for any amount of FRP ratios, ρ f .
5
fy = 500 MPa, f'c = 48.2 MPa
7
fy = 500 MPa, f'c = 48.2 MPa
4 ρ's = 0.0067, h/d = 1.23 6
Curvature ductility
Curvature ductility

h/b = 1.28, CFRP

ρ 's = 0.0067, h/d = 1.23
5 h/b = 1.28, GFRP
3
Tension failure 4
μφ

μφ

2 FRP rubture
ρ s = 0.005
3
Tension failure
0.015
2
1 0.03 0.025 0.02 0.01
1 0.03 0.025 0.02 0.015 0.01 ρ s = 0.005
0 0
0 0.01 0.02 0.03 0 0.02 0.04 0.06 0.08 0.1 0.12
Ratio of FRP reinforcement, ρ f Ratio of FRP reinforcement, ρ f

Figure 2: Curvature ductility values

4. DESIGN EXAMPLE

As an example, it is intended to calculate the displacement ductility of an exterior beam-column subassembly

strengthened with CFRP web-bonded system shown in Fig. 3 reported by Mahini et al. (2004). Their results showed
that the failure is a flexural one which is mainly concentrated in the beam, therefore the ultimate capacity of the
subassembly is close to the ultimate capacity of the strengthened beam. The beam-tip and displacement curve of this
specimen is shown in Fig. 3. The most convenient quantity to evaluate the structure’s capacity to develop ductility

573
μ Δ is displacement as define μ Δ = Δ u / Δ y , where Δ u and Δ y are the lateral tip displacements at ultimate and at
yield respectively. As post peak response is often accompanied by strength degradation, a specified limit to this can
be considered for the ductility to be calculated. A percentage reduction from the peak load Pu is often assumed
(Park, 1989). As a conservative estimate, a 10% reduction in the load can be considered. Consequently from Fig. 3,
the experimental displacement ductility can be calculated as μ Δ ,exp = 3.3. The curvature ductility of web-bonded and
the plain section can be found directly from Fig. 2 as μ φ =3.15 and μ φ = 4.3 respectively. In order to calculate the
displacement ductility factor of the subassembly from curvature ductility, the following relationship presented by
Paulay and Priestley (1992), can be used:

μ φ = 1 + ( μ Δ − 1) /[3(l p / l )[1 − 0.5(l p / l )]] (3)

220 mm
2N12 2N12
180

Tie R6.5

Plastic hinge
location Applied Load
P
Web bonded
180
CFRP tf tf 10% reduction in
1402 mm

Beam tip load, P (kN)

2N12 30 load

230
Tie R6.5 Pu
2N12
Retrofitted 20
Plain
350 Section Section
t f = 0.33 mm
10
Maximum
strength
0
Constant axial Δy Δu
load (N = 305 kN) 0 30 60 90 120
1246 mm Beam tip displacement, Δ (mm)

Figure 3: Failure mechanism and load vs. displacement of subassembly tested by Mahini et al. (2004)

where l p is the plastic hinge length of the beam and l is the cantilever length both in m. Due to the fact that the
failure occurs at the cut-off point of FRP, curvature ductility of this plain section is calculated and the displacement
ductility of the subassembly is estimated using Eq. 3. Applying l = 0.896 m, μ φ = 4.3 and l p = 0.204 m in that, the
analytical value of the displacement ductility for this subassembly is obtained as μ Δ , ana = 2.99.

5. CONCLUSION

Computer models for calculating the neutral axis of the strengthened section at ultimate and yield
conditions leading to the calculation of the curvature were presented and the design charts were
then extracted from these models using a computer program. The displacement ductility of a
web-bonded CFRP beam-column subassembly using the design charts was calculated and the
results were compared with the test. Closeness of these results proved that the presented charts
can be used by design engineers with confidence.

6. REFRENCES
Paulay, T. and Priestley, M. J. N. (1992). Seismic Design of Reinforced Concrete and Masonry Buildings, John
Wiley and Sons, INC.
Triantafillou, T. C. and Plevris, N. (1992). “ Strengthening of RC beams with Epoxy-Bonded Fibre-Composite
Materials”. Materials and Structures, Vol. 25, pp 201-211
Duthinh, D., and Starnes, M. (2001). Strength and Ductility of Concrete Beams Reinforced with Carbon FRP and
Steel, National Institute of Standards and Technology, Technology Administration U.S. Department of Commerce.

574
Mahini, S.S., Ronagh, H.R. and Dux P.F. (2004). “Flexural Repair of RC Exterior Beam-Column Joints Using
CFRP Sheets”, Proc. of the Second International Conference on FRP Composites in Civil Engineering, CICE 2004,
8-10 Dec. Adelaide, Australia, pp 653-658.
Park, R. (1989) “Evaluation of Ductility of Structures and Structural Assemblages from Laboratory Testing”,
Bulletin of the New Zealand National Society for Earthquake Engineering, Vol. 22, No.3, pp 155-166.

575
576
Part XXI. Shear Retrofit
Third International Conference on FRP Composites in Civil Engineering (CICE 2006)
December 13-15 2006, Miami, Florida, USA

PARAMETERS INFLUENCING THE BEHAVIOR OF RC BEAMS

STRENGTHENED IN SHEAR WITH EXTERNALLY BONDED FRP
Omar CHAALLAL
(Professor, Université du Québec, ETS, Montreal, Canada)

Abdelhak BOUSSELHAM
(Research Fellow, Université du Québec, ETS, Montreal, Canada)

ABSTRACT
The identification of parameters and criteria that influence shear resistance mechanisms is conducted using in-depth
analysis of experimental data reported in the literature on shear strengthening with FRP. This identification is based
on knowledge of the shear behavior of conventional RC structures on the one hand, and on the state of the art
regarding the behavior of RC elements retrofitted in shear with FRP on the other.

KEYWORDS
RC beams, strengthening, shear, FRP sheet, parameters, behavior, static loads, research need.

1. INTRODUCTION
The following parameters and criteria are successively assessed and discussed: (a) the influence of mechanical and
geometric properties of FRP, (b) the effect of transverse steel ratio, (c) the effect of longitudinal steel, (d) the effect
of shear span ratio or type of beams (slender or regular versus deep), and (e) the scale factor effect or size of the
specimens. Additionally, other parameters and criteria which are not sufficiently documented, but which may be
important, are also discussed in light of the few research studies available in the literature. These include the effect
of concrete strength, the effect of fatigue, the effect of pre-cracking, and the effect of prestress.

2. INFLUENCE OF FRP PROPERTIES

Figure 1 shows, for slender beams ( a d ≥ 2.5 ) without transverse steel reinforcement, the variation of effective
FRP strain (ε ) as a function of FRP rigidity, expressed with respect to the compressive strength of concrete
f ,e

(E f ρf f ′ ) . Figure 1 illustrates the behavior of beams that have failed by debonding. The same trend was
c
23

observed for those that have failed in shear by modes other than debonding. From Figure 1, it can be observed that
the effective FRP strain decreases as FRP rigidity increases. This confirms the results reported by other researchers
(e.g. Bousselham and Chaallal 2004; Triantafillou et al. 2000). In addition, results reveal that the average effective
strains attained by the FRP are generally smaller when failure occurs by debonding, compared to other modes of
shear failure. However, in all cases, these strains represent only a modest fraction of the ultimate FRP strain
capacity: the ratio (ε f ,e ε f ,u ) is generally less than 0.30 for failure by debonding and approximately 0.40 for
other modes of shear failure. This tendency also holds true for the case of slender beams with transverse steel. In
addition, in the presence of transverse steel reinforcement, the FRP tends to attain higher strain values compared to
those attained in the absence of transverse steel. However, this does not mean that the gain in shear resistance due to
FRP is greater, a conclusion confirmed by numerous studies (Bousselham and Chaallal 2006-a; 2006-b; Chaallal et
al. 2002; Pellegrino et al. 2002; Li et al. 2001).

3. EFFECT OF INTERNAL TRANSVERSE STEEL REINFORCEMENT

The influence of transverse steel on the contribution of FRP to shear resistance has been documented by many
research studies (Bousselham and Chaallal 2004; Pellegrino et al. 2002; Chaallal et al. 2002; Czaderski 2002; Li et
al. 2001). More recently, this influence has been further demonstrated through monitoring of the variation of strains
in FRP and in transverse steel under increasing load (Bousselham and Chaallal 2006-a; 2006-b). This influence is

577
confirmed, at a much larger scale, in the present study which encompasses the data from all research studies
reported in the literature. Figure 2 presents, for slender beams, the variation of the gain due to FRP as a function of
the ratio ( ρs,t f s,t ρ f f f ,e ) , where ρs,t and ρf represent the transverse steel and the FRP ratios respectively, and fs,t
and ff,e the yield stress of the transverse steel and the effective stress of FRP respectively. Figure 2, which refers to
beams that have failed by debonding, clearly indicates that the gain in shear resistance due to FRP decreases as the
ratio ( ρs,t f s,t ρ f f f ,e ) increases. The same tendency was observed for deep beams ( a d < 2.5 ) that have
failed by debonding and those that have ruptured by other modes of shear failure. However, the gains in shear
resistance due to FRP which were recorded for beams with transverse steel reinforcement are substantially greater in
slender beams (more than 90% gain in shear resistance) than in deep beams (less than 30%). The influence of
transverse steel on the contribution of FRP to shear resistance is now an accepted fact well-documented by
experimental evidence. However, the resistance mechanisms associated with the phenomenon remain to be
explained.

0,020 150

Sides Sides

0,016 U-Wrap 120 U-Wrap

Total-Wrap

Shear force gain (%)

Total-Wrap

0,012 90
εf, e

0,008 60

0,004 30

0,000 0
0,00 0,03 0,06 0,09 0,12 0,15 0,18 0,0 0,5 1,0 1,5 2,0 2,5 3,0 3,5 4,0
2/3
Efρf/fc fs,tρs,t/ff,eρf

Figure 1: Effective strain of FRP in terms of Figure 2: Shear Force Gain in terms of
Ef ρ f f c′2 3 f s , t ρ s ,t f f , e ρ f

4. EFFECT OF SHEAR SPAN TO DEPTH RATIO (SLENDER VS. DEEP BEAM)

Figure 3 typically shows the variation of the gain in shear resistance due to FRP as a function of the so-called a/d
ratio involving the shear length (a) and the effective beam depth (d). This figure refers to beams without transverse
steel that have failed by debonding. The tendency that is apparent in this figure is the same as that reported in the
literature (Bousselham and Chaallal 2004; 2006-b): the gain in shear resistance due to FRP seems in general to be
greater in slender beams (a/d ≤2.5) than in deep beams. This is attributed to the strut-and-tie behavior exhibited by
the deep beams. It follows that the contribution of external FRP, like that of internal transverse steel, is less
significant in deep beams than in slender beams. However, a possible effect of enhanced confinement of concrete in
compression due to FRP wrap cannot be excluded (Chaallal et al. 2002). Also, results revealed that failure by
debonding of FRP occurs more in slender beams than in deep beams.

5. SCALE FACTOR EFFECT

Figure 4 typically presents the variation of the gain in shear resistance due to FRP as a function of the effective
depth of beam section, d. This figure refers to slender beams without transverse steel that have failed by debonding.
Figure 4 shows a very clear tendency for gain due to FRP to decrease with increasing effective beam section depth.
This confirms the result of a very recent experimental study by Bousselham and Chaallal (2006-c). It must be
remembered that the relevance of this result stems from the fact that the experimental data used in this study are
derived largely from tests performed on beams of relatively small size, particularly in comparison to bridge girders.

6. EFFECT OF LONGITUDINAL STEEL REINFORCEMENT

Examination of results from the database seems to indicate that the gain due to FRP decreases as the longitudinal
steel reinforcement ratio increases. This leads one to suspect the existence of an interaction between the FRP and the
longitudinal steel reinforcement. The state of knowledge regarding shear behavior of conventional RC beams
indicates that beams with a small longitudinal reinforcement ratio exhibit lower shear resistance than beams with a

578
larger ratio (ASCE-ACI, 1998). Indeed, the flexural cracks that develop in the latter case are wider and progress at a
greater rate towards the compression zone of the beam. In addition, the formation of diagonal cracks occurs earlier
(MacGregor and Bartlett, 2000). This affects the shear stress in uncracked concrete, and consequently the
contribution of concrete to the shear resistance. Increasing the longitudinal steel reinforcement ratio will translate
into a greater contribution of concrete to the shear resistance. Will this increase also translate into a decrease in the
gain due to FRP? Clarification of this type of questions may require further experimental and theoretical
investigations.

240 240
Sides
Sides
200 U-Wrap 200
U-Wrap
Total-Wrap
Total-Wrap
Shear force gain (%)

Shear force gain (%)

160 160

120 120

80 80

40 40

0 0
0,0 0,6 1,2 1,8 2,4 3,0 3,6 4,2 4,8 0 100 200 300 400 500 600 700

a/d d (mm)

Figure 3: Shear Force Gain in terms of a d Figure 4: Shear force gain in terms of d

7. EFFECT OF OTHER PARAMETERS AND RESEARCH NEED

Concrete strength has an influence on the performance of shear strengthening with FRP from both the local and the
global point of view. From the local point of view, this influence impacts the bonding performance at the FRP-
concrete interface. A higher concrete strength will delay, if not inhibit altogether, failure by debonding. From the
global point of view, this influence is linked to the failure scenarios proposed, that can differ depending on the level
of concrete strength. A high concrete strength (relative to the transverse steel ratio if any, or to the FRP ratio) will
inhibit premature crushing of concrete in the compression zone or in the web. Additionally, it will enhance bonding
characteristics at the FRP-concrete interface (Bousselham and Chaallal 2006-a). However, despite its importance
with regards to the performance of shear strengthening with FRP, the effect of concrete strength has not been
systematically studied. It is interesting to note, though, that the guidelines for the design of RC structures
strengthened with externally applied FRP take into account the concrete strength when calculating the contribution
of FRP to shear resistance (ACI-440.2R 2002; fib TG 9.3 2001), either for the determination of the effective FRP or
to prevent premature crushing of concrete. Therefore, it may be useful to document this influence analytically and
experimentally.

The ACI-215R-05 (2005) Committee has written: “Special attention should be given to the shear fatigue strength of
beams subjected to high nominal shearing stresses”. This is in accordance with the recommendations formulated by
Regan (1993), who, in an exhaustive state-of-the-art report on the shear behavior of RC beams, clearly underlined
the need to study this aspect. However, compared to retrofitted RC beams subjected to static loading, the behavior of
RC beams retrofitted in shear with FRP and subjected to cyclic loading is much less well-documented. In addition,
all the studies on the behavior of strengthened beams under fatigue loading are related either to strengthening of
beams in flexure or to the FRP-concrete interface In contrast, shear strengthening under fatigue is yet to be
investigated, since so far there are only two very recent studies reported in the literature dealing with this specific
subject (Czaderski 2002; Carolin et al. 2005). It is noted in particular, in the latter study: “There is a tendency for
fatigue-loaded beams to have a higher load-bearing capacity when tested to failure compared to beams without a
fatigue history.” Nevertheless, additional investigations appear to be necessary, with particular emphasis on the
fatigue performance of bonds in the anchorage zone.

Almost all of the experimental investigations conducted so far are related to the shear performance of strengthened
RC beams that had not been pre-loaded (pre-cracked) prior to their retrofit. However, external strengthening with
FRP is most suitable for existing in-service structures that often are pre-cracked, if not pre-damaged. The few
investigations carried out so far on RC beams that were pre-cracked prior to their strengthening indicate that pre-

579
loading does not affect the shear performance of retrofitted beams (Czaderski 2002, Carolin et al. 2005, Hassan
Dirar et al. 2006).

According to fib T.G 9.3 (2001), “less than 10% of the bridges that have been FRP-strengthened so far are pre-
stressed”. This seems rather surprising given the number of bridge superstructures made of pre-stressed concrete
(PC). To our knowledge, the only study dealing with PC beams strengthened in shear with FRP was performed by
Hutchinson and Rizkalla (1999). The latter reported that the prediction by the shear equation in ACI-318 was in
good agreement with the test results of seven pre-stressed concrete beams strengthened with CFRP strips. However,
there is not a single study in the existing literature dedicated specifically to the effect of pre-stressing on the
performance of shear strengthening with externally bonded FRP. It must be borne in mind that this is a very
complex problem, because on the one hand, the behavior of PC structures is mostly characterized by long-term
phenomena, and on the other hand, strengthening is often applied to structures that have already been subjected to
long-term effects such as creep, relaxation, and shrinkage. Therefore, the impact of such effects on the state of
deformation of the retrofitted structure remains to be understood and adequately evaluated through simulations or
targeted tests.

8. REFERENCES
ASCE-ACI Task Committee 445. (1998). Recent Approaches to Shear Design of Structural Concrete. J. of
Structural Engineering, ASCE, 124(12), pp. 1375-1417.
Bousselham, A., Chaallal, O. (2006-a). Behavior of RC T-Beams Strengthened in Shear with CFRP: An
Experimental Study. ACI Structural Journal, Vol. 103, No.3, May-June, pp. 339-347.
Bousselham, A., Chaallal, O. (2006-b). Effect of Transverse Steel and Shear Span on the Performance of RC Beams
Strengthened in Shear with CFRP. Composites: Part B, Elsevier, Vol. 37, pp 37-46.
Bousselham, A., Chaallal, O. (2006-c). Influence of the Beam Size on the Shear Strength of RC T-Beams Retrofitted
in Shear with CFRP fabrics. To appear in ACI Structural Journal, 40 p.
Bousselham, A., Chaallal, O. (2004). Shear Strengthening Reinforced Concrete Beams with Fiber-Reinforced
Polymer: Assessment of Influencing Parameters and Required Research. ACI Str. J., Vol. 101, No.2, pp. 219-227.
Chaallal, O., Shahawy, M., Hassan, M. (2002). Performance of Reinforced Concrete T-Girders Strengthened in
Shear with CFRP Fabrics. ACI Structural Journal, 99(3), pp. 335-343.
Czaderski, C. (2002). Shear Strengthening with Prefabricated CFRP L-Shaped plates. Test Beams S1 to S6, EMPA.
Report No. 116/7, Switzerland, 78 p.
Carolin, A., Taljsten, B. (2005). Experimental Study of Strengthening for Increased Shear Bearing Capacity. Journal
of Composites for Construction, 9(6), pp. 488-496.
fib-TG9.3. (2001). Design and Use of Externally Bonded Fiber Polymer Reinforcement (FRP EBR) for Reinforced
Concrete Structures. Technical Report Prepared by EBR Task Group 9.3, Bulletin 14.
Hassan Dirar, S.M.O., Hoult, N.A., Morley, C.T., Lees, J.M. (2006). Shear Strengthening of Pre-cracked Reinforced
Concrete Beams Using CFRP Straps. Fédération Internationale de Béton (FIB), Proceedings of the 2nd International
Congress, Naples, Italy, June 5-8, ID 10-70
Li, A., Assih, J., Delmas, Y. (2001). Shear Strengthening of RC Beams with Externally Bonded CFRP Sheets. J. of
Structural Enginnering, ASCE, 27(4), pp. 374-380.
Pellegrino, C., Modena, C. (2002). Fiber Reinforced Polymer Shear Stengthening of RC Beams with Transverse
Steel Reinforcement. J. of Composites for Construction, ASCE, 6(2), pp. 104-111.
Regan, P.E. (1993). Research on Shear: A Benefit to Humanity or a Waste of Time? The Structural Engineer,
71(19), pp. 337-347.
Rizkalla, S., Hassan, T., Hassan, N. (2003). Design Recommendations for the Use of FRP for Reinforcement and
Strengthening of Concrete Structures.Prog. Struct. Engng. Mater. Wiley, 5, pp. 16-28.

580
Third International Conference on FRP Composites in Civil Engineering (CICE 2006)
December 13-15 2006, Miami, Florida, USA

Calculating the Thickness of FRP Jacket for Shear and Torsion Strengthening
of RC T-Girders
A. Deifalla
(Ph.D Candidate, McMaster University, Hamilton, ON, Canada)

A. Ghobarah
(Professor, United Arab Eimrates University, Al-Ein, UAE)

ABSTRACT
Failure of a structural element under diagonal compression is brittle in nature and could compromise the structure
ductility in seismic design. In many cases, strengthening using FRP Jackets might be the optimum solution to
prevent such failures. A procedure for calculating the thickness of the FRP jacket required to resist combined shear
and torsion is proposed. The proposed procedure was used to calculate the FRP thickness required for the
strengthening of three T-girders subjected to combined shear and torsion. The girders were constructed and tested
using a test setup designed to subject the specimen to combined shear and torsion with relatively low moment up to
failure. Three strengthening techniques namely; U-jacket, extended U-jacket and full jacket were implemented using
± 45° carbon fiber reinforced polymers (CFRP). The theoretical predications were compared to the experimental
results. The procedure was found to provide reasonable conservative estimates.

KEYWORDS
T-girders, shear, torsion, strengthening, FRP experimental

1. INTRODUCTION
In elevated expressways and bridges, girders are often subjected to significant torsion which could result from
eccentricity of lane loading, torsionaly restrained joints and supports and curved alignments in plan. The complexity
and disadvantages of strengthening using materials such as steel and concrete jackets and the high cost of structures
replacement have prompted the research into the area of strengthening structures using composite materials. Fiber
reinforced polymers (FRP) laminates and fabrics have been used successfully in various applications. Due to the
properties of the FRP such as high strength and low weight as well as ease of installation, these materials are
becoming of interest in the civil engineering applications. Research efforts were directed towards strengthening
rectangular beams using fully wrapped jackets. In this paper, a simple procedure was proposed to calculate the
thickness of FRP jacket required to provide additional combined shear and torsion strength for the RC T-girders.
The procedure is an extension of the simplified methodology developed by Deifalla and Ghobarah (2005) for torsion
strengthening. The procedure is verified using the results of a pilot experimental program that is conducted to
evaluate the influence of externally applied FRP fabrics on the behavior of the strengthened reinforced concrete
(RC) T-girders subjected to combined shear and torsion (Deifalla and Ghobarah 2006). Where four T-girders TG1,
TG2, TG3 and TG4 were tested under a constant torque to shear ratio of 0.5 m which was chosen as to represent the
case of significant torsion combined with shear. TG1 is the control girder while TG2 to TG4 are the strengthened
girder.

3. CALCULATION OF THE FRP JACKET THICKNESS

The proposed procedure calculates the thickness required at the critical section where it is reasonable to assume that
the effect of the torsion and shear can be added together. The FRP thickness required to resist torsion (tt) is
calculated using the hollow tube analogy, space truss theory such that:

581
 T − To
tt = (1)
2 E f ε f Ao [cos β + sin β ]
where T is the total torsion capacity of the strengthened girder and To is the total torsion capacity of the un-
strengthened girder, Ef is Young’s modulus of FRP sheets, εf is the effective strain in the FRP sheet, β is angle of
fiber orientation for the FRP sheets and Ao is the area enclosed inside the critical shear flow path. The threaded
anchors were used and were to resist torsion by creating a closed loop for shear flow path. Based on the design
provisions CSA-S806-02 (2002), the FRP thickness required to resist the additional shear (tv) may be computed as:
V − Vo
tv = (2)
2 E f ε f d [sin β + cos β ]
where V is the total shear resistance of the strengthened girder, Vo is the total shear resistance of the un-strengthened
girder and d is the depth of the FRP sheet resisting shear. Assuming linear interaction between the FRP torsion and
shear contribution, the FRP area per unit length required to resist both shear and torsion (tf) will be taken as the
summation of the FRP area per unit length required to resist shear (tv) and the FRP thickness required to resist the
torsion (tt). Thus, the total FRP thickness required to resist the applied combined torsion and shear (tf) can be
calculated as:
t f = tv + tt (3)
The procedure requires that the value of the FRP strain at failure is known which is dependent on several parameters
including the strengthening scheme used, the concrete, FRP dimensions and properties. No previous studieds were
conducted on the area of combined shear and torsion strengthening. However, Very limited number of investigations
were conducted on the area of torsion strengthening. Deifalla and Ghobarah (2005) proposed three different
formulas for calculating the strain for torsion strengthening which were verified using an extensive database
gathered from various sources. Although these formulas were developed for torsion strengthening, the T-girders
were tested under significant torsion. Hence these formulas can be adopted in the analysis.

2. EXPERIMENTAL RESULTS
The proposed procedure was verified using the results of an experimental investigation conducted for the purpose of
investigating the behavior and strengthening of RC T-girders under combined actions. A brief summary of the
program and the results is provided. Four T-girders TG1, TG2, TG3 and TG4 were constructed and tested. The
compressive strength of 25.6 MPa was obtained from the compression testing of three standard concrete cylinders at
28 days. The CFRP used in this experimental program is Tyfo BCC composite, which includes Tyfo BCC
reinforcing fabric and Tyfo S Epoxy (Fyfe Co. 2002). The primary fibers are continuous in the ±45 directions and
the composites provide ultimate strength in the direction of the fibers. The properties of the CFRP as provided by
the supplier were a tensile strength of 609 MPa, modulus of elasticity of 63.3 GPa, maximum elongation of 9.6
mm/m, and thickness of 0.86 mm.

The T-girders were tested under significant torsion, shear with a ratio of 0.5 m. The applied bending moment is
considered small. The typical cross section of the T-girders is shown in figure 1. The girders were heavily reinforced
in the longitudinal direction within the test zone to minimize the effect of the flexure focusing on the shear and
torsion behavior. To insure failure occurring within the test zone, the girders were heavily reinforced in the
longitudinal direction as well as in the transverse direction. Fig. 2 shows a schematic of the applied loads (L1, L2
and L3) and restraint reactions (R1, R2 and R3) in the test setup.

During the strengthening procedure, special attention was paid to insure the bond between the concrete surface and
the CFRP sheets. In addition, the concrete edges were rounded in order to minimize the stress concentration in the
fibers at the edges. The strengthening techniques implemented covered most of the different situations. The first
technique used consisted of one FRP sheet bonded to the web and anchored below the inner intersection of the web
and the flange as shown in Fig. 3.a. The second technique shown in Fig. 3.b consisted of FRP sheet bonded to the
web and extended to the bottom of the flange. The FRP sheet was prevented from pulling out at the inner
intersection of the web and the flange by a steel angle. The third technique consisted of two FRP sheets, the first
piece was similar to the second technique and the second sheet was bonded to the flange. The two sheets overlapped
to insure continuation of the forces resisting torsion. A steel angle as well as anchor rod was provided.

582
 450 mm 2#10
2#15

350 mm
100
2#10
#10@200mm 4#20
150 mm

Figure 1: Typical T-girder cross section

500 mm

900 mm
1400 mm
700 mm
Figure 2: Test Setup

Threaded anchor rod 15 mm diameter

Theraded anchor rod 15 mm diameter
15 mm diameter
Angle 125x50x10
FRP sheet
Angle 100x100x10
FRP sheet FRP sheet
a) b) c)

Figure 3: Strengthening Schemes a) U-Jacket, TG2, b) Extended U-Jacket, TG3 and c) Full wrapping, TG4

Table 1 Summary of the experimental results

Girder Shear* Mode of failure

(kN)
TG1 46 Steel stirrup yield followed by concrete crushing
TG2 68 FRP intermediate debonding followed by flange concrete crushing
TG3 76 FRP intermediate debonding

TG4 80 FRP intermediate debonding

* Torsion to shear ratio is 0.5 m.

583
Table 1 shows a summary of the maximum shear resistance and the failure mode for the tested girders TG1, TG2,
TG3 and TG4. The corresponding torsion resistance can be calculated as 0.5 m multiplied by the maximum shear
resistance. All strengthened girders failed due to FRP intermediated diagonal debonding similar to the concrete
diagonal cracks due to diagonal tension resulting from applied torsion and shear.

4. VERIFICATION OF THE PROCEDURE

The FRP thickness was predicted using equations 1, 2 and 3. The strain was calculated by the formula developed by
Deifalla and Ghobarah (2005) for the FRP intermediate debonding due to concrete cracking. The FRP strain is
calculated as follows:
0.33 w f
εf = (4)
Le s f
where wf is the width of the FRP strips and sf is the spacing between the centerline of the FRP strips. For the case of
continuous FRP sheets wf = sf and Le is calculated using the formula:
Eftf
Le = (5)
`
fc
Where Ef is the FRP Young’s modulus and tf is the thickness of the FRP sheets and fc` is the compressive strength of
the concrete. The values of Vo and To are taken the same as the strength of TG1. Table 2 shows the comparison
between the predicted and the actual FRP thickness. The thicknesses calculated from the proposed procedure are in a
good agreement with the experimentally used thicknesses.

Table 2 Comparison between actual and the calculated FRP thickness

Strengthening Technique
Ratio between Calculated and
Beam
Actual FRP thickness

TG2 Anchored U-jacket 0.74

TG3 Anchored Extended U-Jacket 0.85
TG4 Anchored Full wrapping 0.91
Average 0.83
Standard deviation 0.09

5. CONCLUSIONS
A procedure for calculating the thickness of the FRP jacket for shear and torsion strengthening of RC T-girders was
proposed and verified against the results of an experimental program. The proposed procedure provides conservative
estimates for the FRP thickness required to resist the applied torsion and shear. Further experimental testing is
required to generalize the procedure, in particular, the predication of the effective FRP strain.

6. REFERENCES
CSA-S806-02. (2002). “Design and Construction of Building Components with Fiber-Reinforced Polymers”,
Canadian Standards Association, Rexdale, Canada, 206 pages.
Fyfe Co. (2002). “http://www.Fyfeco.com/products /compositesystems /Bcc.html”, accessed January 2002.
Ghobarah, A., Ghorbel, M., and Chidiac, S. (2002). “Upgrading Torsional Resistance of RC Beams Using FRP”,
Journal of Composites for Construction, ASCE, Vol. 6, No. 4, pp. 257-263.
Deifalla, A., Ghobarah, A. (2005). "Simplified Analysis of RC Beams Torsionaly strengthened Using FRP",
Proceedings of International Symposium on Bond Behavior of FRP in Structures (BBFS 2005), Hong Kong, China,
edited by Chen and Teng, December 7-9, pp. 381-386.
Deifalla, A., Ghobarah, A. (2006). "Shear and torsion strengthening of RC T-beams", Submitted, Composite
structures, Elsevier.

584
Third International Conference on FRP Composites in Civil Engineering (CICE 2006)
December 13-15 2006, Miami, Florida, USA

CALIBRATION OF EUROCODE-LIKE EQUATION FOR THE

CONCRETE CONTRIBUTION TO THE SHEAR CAPACITY OF FRP RC
MEMBERS
Raffaello Fico, Andrea Prota, Renato Parretti, Gaetano Manfredi and Antonio Nanni

(Department of Structural Analysis and Design, University of Naples “Federico II”, Naples, Italy)

ABSTRACT
Fiber Reinforced Polymer (FRP) bars represent an interesting alternative to conventional steel as internal
reinforcement of Reinforced Concrete (RC) members where some properties such as durability, magnetic
transparency, insulation, are of primary concern. Critical aspects of the design of FRP RC members are
serviceability issues, bond, and the reduced concrete contribution to the shear capacity compared to equivalent steel
reinforced members. The present paper focuses on the calibration of Eurocode-like equation for the evaluation of the
shear strength of FRP RC members without shear reinforcement. The calibration is based on an extensive database
of experimental data available in the literature. The concrete contribution to shear capacity is analyzed considering
tested members without shear reinforcement and, with reference to Eurocode equation for shear capacity of steel RC
members, a modified equation is proposed. A parametric analysis is presented comparing the results of the proposed
equation with those given by both CSA and ACI guidelines.

KEYWORDS
Design, FRP, Reinforced Concrete, Shear.

1 INTRODUCTION
FRP has much potential as longitudinal reinforcement in concrete structures exposed to reinforcement corrosion and
stressed primarily in bending. Bridge decks, footings, floor slabs, soft eyes applications and wall type structures are
some examples of such structural components. In these elements, flexural strength is mainly provided by the
longitudinal reinforcement, and the shear strength is provided by the concrete alone because of the lack of transverse
reinforcement. Hence, it is critical that an accurate assessment of concrete contribution to the shear strength of
members reinforced with FRP bars as flexural reinforcement, Vc,f , is performed. Test results have shown that the
shear strength of FRP RC beams is significantly lower than that predicted using equations developed for steel
reinforcement. Nevertheless, it is definitely recognized that the shear strength due to concrete, Vc,f , can be
calculated according to the same principles as for steel reinforced concrete after accounting for the different
mechanical properties between FRP and steel reinforcement. This is the approach taken herein to calibrate the
equation given by Eurocode 2 for the calculation of the concrete contribution to shear capacity of steel RC members.
Based on this equation, a new formula accounting for the concrete contribution to shear capacity of FRP RC
elements was derived; this expression has been included in the lately issued guidelines of the Italian Research
Council CNR-DT 203/2006.

2 REVIEW OF CURRENT DESIGN PROVISIONS

Findings from experimental investigations on concrete beams without stirrups and longitudinally reinforced with
carbon and glass FRP bars show that the shear strength reduction experienced by such members when compared to
the shear strength of those reinforced with the same amount of steel reinforcement, is mainly due to the relatively
lower modulus of elasticity. Such investigations also reveal that the axial stiffness of the reinforcing bars is a key

585
parameter when evaluating the concrete shear strength of flexural members reinforced with FRP bars. Most of the
current international Standard Codes developed methods to compute Vc,f that are based on these assumptions. This
section summarizes the design equations to compute Vc,f , as recommended by the American Concrete Institute (ACI
440.1R-03 2003), the Canadian Standard Association (CAN/CSA-S806_02 – 2002) and by Tureyen and Frosh
(Tureyen and Frosh, 2003).

2.1 ACI 440.1R-03 Design Guidelines

ACI Committee 440 recommends the following equation for calculating Vc,f :
ρ f Ef
Vc,f = V (1)
ρ s Es c
where ρ f and ρ s are the flexural FRP and steel reinforcement ratio, respectively, Ef and Es are the modulus of
elasticity of FRP and steel reinforcement, respectively, and Vc is the design shear resistance provided by concrete
when steel reinforcement is used. It is clear that such equation accounts for the axial stiffness of the FRP
reinforcement ( Af Ef ) as compared to that of steel reinforcement ( As Es ) .

2.2 CAN/CSA-S806_02 Design Guidelines

The CSA S806-02 gives the following expression to compute Vc,f for sections having either the minimum amount of
transverse reinforcement required or d < 300mm:
1
⎛ V ⎞3
Vc,f = 0.035λφc ⎜ f c' ρ f E f f d ⎟ bw d (2)
⎜ M f ⎟⎠
⎝
where λ accounts for concrete density (set equal to 1 herein), φc is the resistance factor for concrete, f c ' is the
specified compressive strength of concrete, Vf and M f are the factored shear force and moment at the section of
interest, bw is the web width and d is the distance from the compression fiber to the centroid of the main tension
reinforcement. For sections with d > 300mm and with no transverse shear reinforcement, Vc,f is calculated using:
⎛ 130 ⎞
⎟ λφc f c bw d ≥ 0.08λφc f c bw d
' '
Vc,f = ⎜ (3)
⎝ 1000 + d ⎠
The latter equation is derived from the corresponding formula given for steel reinforced sections, multiplied by 0.5 .

2.3 Equation Proposed by Tureyen and Frosh

Due to the conservative predictions of equation (1), Tureyen and Frosh (2003) proposed a different design method to
compute Vc,f , as follows:
2
Vc,f = k f c' bw d (4)
5
where
k = 2 ρ f nf + ( ρ f nf )2 − ρ f nf (5)
and nf = Ef Ec , Ec being the elastic modulus of concrete. It can be observed that through factor k , eq. (4) is a
function of the axial stiffness Af Ef .

3 PROPOSED SHEAR EQUATION

The proposed formula is a modified version of the Eurocode 2 (Eurocode 2, 1992) shear equation recommended for
conventional steel RC members, which is:
V c = τ Rd ⋅ c ⋅ (1.2 + 40 ρ s ) ⋅ bw ⋅ d (6)

586
where τ Rd is the design shear stress per unit area (equal to 0.053 f 'c γ c , γ c being the strength reduction factor for
concrete) and c is a factor depending on d .
A calibration was conducted in order to modify eq. (6) and extend it to FRP RC members; the following expression
for Vc,f has been proposed:
1/ 2
⎛E ⎞
V c,f = 1.3 ⋅ ⎜ f ⎟ ⋅τ Rd ⋅ c ⋅ (1.2 + 40 ρ f ) ⋅ bw ⋅ d (7)
⎝ Es ⎠
Eq. (7) is also included in the Italian guidelines CNR-DT 203/2006.

4 VERIFICATION OF THE PROPOSED EQUATION

4.1 Experimental Database

In order to verify the proposed equation, a database composed of test results related to 79 specimens tested (beams
and one way slabs) was used for comparisons, as given in Table 1.
Four specimens were reinforced with aramid FRP bars, 22 specimens reinforced with carbon FRP, and 53 specimens
reinforced using glass FRP bars. All specimens had no transverse reinforcement and failed in shear. The concrete
compressive strength, f c ' , ranged between 24.1 and 50.0 MPa (specimens with f c ' > 50 MPa were neglected not
being typical of FRP RC members). The reinforcement ratio of tensile FRP bars, ρ f , ranged between 0.0025 and
0.03; however, since CNR-DT 203 prescribes a minimum ρ f equal to 0.01, experimental points below this threshold
(dashed line in Figure 1) were not significant for the sake of the calibration. The effective depth, d , ranged between
150 and 970 mm, and the shear span to depth ratio, a d , ranged between 1.82 and 5.80.

4.2 Comparison with Major Design Provisions

The predictions from equation (7) were compared with the values derived using the equations reported in section 2,
after setting the strength reduction factors equal to the unity. It can be observed from Table 1 and Figure 1 that the
trend line of CNR-DT 203 equation is very similar to that of CSA S-802 equation; the CNR-DT 203 line has the
least mean value of Vexp Vpred (i.e., 1.33) and coefficient of variation (i.e., 26 %), which is defined as the ratio of the
standard deviation to the mean and is a measure of dispersion of a probability distribution. Thus, the proposed
equation is found to be reliable for predicting the shear capacity of FRP RC members without shear reinforcement.

5 CONCLUSIONS
The paper presents the work at the basis of the Eurocode-like shear equation suggested by the CNR-DT 203 to
compute the concrete contribution to the shear capacity of FRP RC members. Based on a wide experimental
database, a verification of the proposed equation is carried out and a comparison with American and Canadian
provisions is discussed. It is concluded that the proposed equation could give accurate predictions and yet enough
conservative.

6 REFERENCES
ACI Committee 440. (2003). “Guide for the Design and Construction of Concrete Reinforced with FRP Bars,” ACI
440.1R-03, American Concrete Institute, Farmington Hills, Mich..
CAN/CSA S806–02. (2002). “Design and Construction of Building Components with Fibre Reinforced Polymers”,
Canadian Standards Association, Rexdale, Ontario.
CNR-DT 203/2006. (2006). “Istruzioni per la Progettazione, l’Esecuzione ed il Controllo di Strutture di
Calcestruzzo armato con Barre di Materiale Composito Fibrorinforzato,” National Research Council, Rome, Italy.
El Sayed, A.K. et al. (2006). “Shear Strength of Concrete Beams Reinforced with FRP Bars: Design Method”, ACI
Structural Journal, V. 103, No 2, pp. 235-243.
EN 1992-1-1 Eurocode 2 (1992). Design of concrete structures - Part 1-1: General rules and rules for buildings.

587
Razaqpur, A.G., and Isgor, O.B. (2006). “Proposed Shear Design Method for FRP-Reinforced Concrete Members
without Stirrups”, ACI Structural Journal, V. 103, No 1, pp. 93-102.
Tureyen, A. K., and Frosch, R. J. (2003). “Concrete Shear Strength: Another Perspective”, ACI Structural Journal,
V. 100, No. 5, pp. 609-615.

Table 1: Characteristics of Specimens Used and Comparison of Design Methods

Reference f 'c bw d ρf Ef Vexp / Vpred Reference f 'c bw d ρf Ef Vexp / Vpred
[Mpa] [mm] [mm] [Gpa] CNR ACI CSA Tur.-Fr. [Mpa] [mm] [mm] [Gpa] CNR ACI CSA Tur.-Fr.
40,5 200 225 0,0025 145 0,87 5,80 1,30 2,19 Lubell et al. 40,0 450 970 0,0046 40 1,33 4,43 0,75 1,17
40,5 200 225 0,0050 145 1,05 3,77 1,34 2,08 Yost et al. 38,0 305 192 0,0036 41,4 0,91 7,94 1,17 1,92
40,5 200 225 0,0063 145 1,02 3,00 1,25 1,89 Mizukawa et al. 34,7 200 260 0,0130 130 1,18 1,81 1,22 1,73
Razaqpur
40,5 200 225 0,0050 145 2,16 7,72 2,41 4,26 38,1 150 210 0,0131 45 1,27 3,70 1,31 1,87
Duranovic et al.
40,5 200 225 0,0050 145 1,05 3,77 1,47 2,08 32,9 150 210 0,0131 45 1,17 2,99 1,14 1,61
40,5 200 225 0,0050 145 0,86 3,09 1,28 1,70 Swamy-Aburawi 34,0 254 222 0,0155 34 1,24 3,34 1,06 1,67
36,3 229 225 0,0131 40,3 1,22 3,56 1,24 1,73 40,0 1000 165 0,0039 114 0,96 4,99 1,31 2,11
36,3 178 225 0,0170 40,3 1,21 2,96 1,23 1,66 40,0 1000 165 0,0078 114 1,03 2,98 1,24 1,85
36,3 229 225 0,0197 40,3 1,25 2,80 1,28 1,70 40,0 1000 161 0,0118 114 1,08 2,30 1,07 1,80
Yost et al.
36,3 279 225 0,0215 40,3 1,01 2,15 1,04 1,37 Benmokrane 40,0 1000 162 0,0086 40 1,17 5,30 1,15 1,96
2
36,3 254 225 0,0243 40,3 1,05 2,08 1,09 1,42 et al. (2004) 40,0 1000 159 0,0170 40 1,23 3,44 1,15 1,84
36,3 229 225 0,0264 40,3 1,05 1,98 1,10 1,42 40,0 1000 162 0,0171 40 1,39 3,85 1,32 2,06
Zhao & 34,3 150 250 0,0151 105,0 1,26 1,93 1,30 1,79 40,0 1000 159 0,0244 40 1,22 2,75 1,17 1,80
Maruyama 34,3 150 250 0,0302 105,0 0,96 0,99 1,06 1,38 40,0 1000 154 0,0263 40 1,25 2,71 1,18 1,85
34,3 150 250 0,0227 105,0 0,97 1,15 1,03 1,36 36,3 229 225 0,0111 40,3 1,33 4,37 1,24 1,94
Alkhajardi 24,1 178 279 0,0230 40,0 2,01 2,72 1,48 2,23 36,3 229 225 0,0111 40,3 1,31 4,30 1,22 1,91
et al. 24,1 178 287 0,0077 40,0 1,86 5,33 1,37 2,39 36,3 229 225 0,0111 40,3 1,25 4,11 1,17 1,82
24,1 178 287 0,0134 40,0 1,80 3,40 1,27 2,06 36,3 178 225 0,0142 40,3 1,14 3,16 1,06 1,60
37,3 160 346 0,0072 42 2,01 8,43 1,67 2,99 36,3 178 225 0,0142 40,3 1,43 3,93 1,32 1,99
37,3 160 346 0,0072 42 2,35 9,85 1,95 3,49 36,3 178 225 0,0142 40,3 1,31 3,61 1,21 1,83
43,2 160 346 0,0110 42 1,30 4,41 1,22 1,85 36,3 229 225 0,0166 40,3 1,20 2,99 1,11 1,65
Tariq & Newhook

43,2 160 346 0,0110 42 1,38 4,70 1,29 1,97 36,3 229 225 0,0166 40,3 1,46 3,63 1,35 2,01
34,1 160 325 0,0154 42 1,64 3,69 1,63 2,06 36,3 229 225 0,0166 40,3 1,34 3,34 1,24 1,85
Yost et al. (2001)
34,1 160 325 0,0154 42 1,51 3,40 1,51 1,90 36,3 279 225 0,0181 40,3 1,05 2,46 0,97 1,43
37,3 130 310 0,0072 120 1,44 3,66 2,01 2,30 36,3 279 225 0,0181 40,3 1,10 2,58 1,02 1,50
37,3 130 310 0,0072 120 1,34 3,40 1,88 2,14 36,3 279 225 0,0181 40,3 1,10 2,59 1,02 1,50
43,2 130 310 0,0110 120 1,14 2,36 1,81 1,78 36,3 254 224 0,0205 40,3 0,95 2,07 0,89 1,28
43,2 130 310 0,0110 120 1,27 2,61 2,00 1,97 36,3 254 224 0,0205 40,3 1,28 2,80 1,2 1,74
34,1 130 310 0,0154 120 1,42 1,91 2,39 1,94 36,3 254 224 0,0205 40,3 1,17 2,55 1,1 1,59
34,1 130 310 0,0154 120 1,48 1,99 2,50 2,03 36,3 229 224 0,0227 40,3 1,16 2,39 1,1 1,57
50,0 250 326 0,0087 130 0,86 2,23 1,37 1,47 36,3 229 224 0,0227 40,3 1,11 2,29 1,05 1,51
50,0 250 326 0,0087 40,0 1,41 6,61 1,25 2,27 36,3 229 224 0,0227 40,3 1,10 2,27 1,04 1,49
Benmokrane 44,6 250 326 0,0124 130 1,14 2,09 1,95 1,75 28,6 305 157,5 0,0073 40 1,21 4,67 1,19 1,85
et al.1 (2004) 44,6 250 326 0,0124 40 1,18 3,93 1,12 1,69 30,1 305 157,5 0,0073 40 1,23 4,99 1,38 1,92
43,6 250 326 0,0172 130 1,24 1,80 2,36 1,84 Deitz et al. (1999) 27 305 157,5 0,0073 40 1,37 5,01 1,47 2,04
43,6 250 326 0,0172 40 1,39 3,65 1,47 1,90 28,2 305 157,5 0,0073 40 1,30 4,94 1,42 1,97
39,7 457 360 0,0096 40,5 1,25 4,42 1,07 1,74 30,8 305 157,5 0,0073 40 1,19 4,90 1,33 1,86
39,9 457 360 0,0096 37,6 1,13 4,17 0,96 1,58
Mean 1,33 3,59 1,49 1,96
40,3 457 360 0,0096 47,1 1,22 4,04 1,15 1,72
Tur-Frosh
42,3 457 360 0,0192 40,5 1,22 2,82 1,34 1,59
Standard deviation 0,35 1,85 0,41 0,54
42,5 457 360 0,0192 37,6 1,40 3,39 1,49 1,83
42,6 457 360 0,0192 47,1 1,45 3,14 1,72 1,92 Coefficient of variation 26% 52% 28% 28%

4,50
CSA
ACI 440
CNR
4,00
Tureyen-Frosh
ACI
3,50

3,00
Vexp / Vpred

2,50 TUREYEN-FROSH

2,00

1,50
CSA S-802

CNR
1,00

0,50

0,00
0,0025 0,005 0,0075 0,01 0,0125 0,015 0,0175 0,02 0,0225 0,025 0,0275 0,03

Reinforcement ratio of reinforcing FRP bars, ρf

Figure 1: Comparison with Major Design Provisions

588
Third International Conference on FRP Composites in Civil Engineering (CICE 2006)
December 13-15 2006, Miami, Florida, USA

DUCTILITY AND SHEAR STRENGTH ENHANCEMENT BY FIBER

SHEET WITH LARGE FRACTURING STRAIN
UEDA Tamon
(Professor, Hokkaido University, Sapporo, Japan)

Dhannyanto ANGGAWIDJAJA
(Graduate Student, Hokkaido University, Sapporo, Japan)

SENDA Mineo
(Graduate Student, Hokkaido University, Sapporo, Japan)

NAKAI Hiroshi
(Manager, Maeda Kosen Co. Ltd., Tokyo, Japan)

DAI Jianguo
(Project Researcher, Port and Airport Research Institute, Yokosuka, Japan)

ABSTRACT
Recently the authors have developed a jacketing method with fiber sheet whose fracturing strain is higher and
stiffness is lower than those of commonly used fibers namely carbon aramid and glass. The fiber is Polyethylene
Terephthalate (PET) and Polyethylene Naphthalate (PEN). In this paper results from a series of experiments on
ductility and shear strength enhancement are presented. Fifteen specimens, modelled after a typical bridge pier, with
PET and PEN sheet jacketing were tested under a reversed cyclic loading until the ultimate ductility was attained.
The stiffness of PET and PEN sheet is smaller than that of carbon sheet to achieve a similar ultimate ductility. Two
specimens tested for shear strength enhancement showed a higher shear strength and more ductile manner than a
companion specimen with carbon fiber sheets. A good control of shear deformation by the un-fractured sheet is a
primary reason to enhance the ductility and shear strength which cannot be predicted by the existing formula derived
from results of tests with carbon and aramid fibers. Due to its good features and lower cost, the jacketing with PET
and PEN sheet is adopted as a seismic retrofitting method for railway structures in Japan.

KEYWORDS
FRP sheet, jacketing, ductility, shear strength, large fracturing strain

1. INTRODUCTION
Major drawbacks with FRP strengthening for concrete structures are high cost and brittleness of the material. One
reason why we have been unable to remove these drawbacks is the fact that we have been emphasizing on strength
and stiffness of the material. To fully utilize concrete material properties besides strength/stiffness high fracturing
strain is also vital to the reinforcing materials. A required strength/stiffness is not necessarily supplied by high
strength/stiffness materials. However a high fracturing strain can be achieved only by materials with a high
fracturing strain. Materials of a low strength/stiffness are generally equipped with a high fracturing strain at a low
cost, while materials of a high strength/stiffness usually with a low fracturing strain are available at a high cost (see
Figure 1). This is the background of our recent project to develop a seismic retrofitting method by jacketing fiber
sheets whose fracturing strain is large. The fiber used in this study was Polyethylene Terephthalate (PET) and
Polyethylene Naphthalate (PEN). The ultimate strength, fracturing strain, and elastic modulus in tension of PET are
923 MPa, 0.123 % and 6.7 GPa, while those of PEN are 1028 MPa, 0.045 % and 22.6 GPa, respectively.

589
 5000
Carbon

4000 High Low

Aramid (Kevlar) strength/stiffness strength/stiffness
Stress(Mpa)

3000 Aramid (Technora)

Glass Low fracturing High fracturing

2000 Polyacetal
strain strain
PET
1000 PEN
Steel High cost Low cost
0
0 3 6 9 12 15 18 21
Strain (%)

Figure 1: Comparison of strength, fracturing strain and cost of materials

2. EXPERIMENTAL OUTLINE
Two series of tests were conducted -- one for ductility enhancement and another for shear strength enhancement.
The first series consists of 15 bridge pier specimens whose hinge zone was jacketed with fiber sheets with PET or
PEN and the rest was jacketed with aramid fiber sheet (A-P Jacketing as shown in Figure 2 (a)) except a reference
specimen. Ten and five specimens were with a cross section of 400 x 400 mm and 600 x 600 mm, respectively (see
Table 1 and Figure 2 (b))(Anggawidjaja, et al. 2006a). Two specimens with a cross section of 250 x 250 mm were
tested for the second series (see Figure 2 (c))(Anggawidjaja, et al. 2006b). Load-deformation curves (envelopes) of
some of the specimens are shown in Figure 3.

D 100
700 400

JA +1
JA +3
JA +8

JA +5
JA +7
JA +6
JA +4
JA +2
JA -1
JA -3
JA -5
JA -2
JA -4
400
Shear strengthening

Aramid fiber
Ø6
400

D19
strengthening

1-1.5D

T1
T3
T8

T5
T7
T6
T4
T2
Ductility

PET fiber
700

1500 600

(a) A-P Jacketing (b) Specimens SP1 to SP10

250

225
1000

38
75
650
650
650

500

500

Unit: mm 150

250 250
500

500

500

SP2ｓ SP3ｓ

1000 1000 1000

(c) Specimens SP2s and SP3s
Figure 2: Specimens

590
 Table 1: Specimens

ρt ρw ρf Vc Vs Vf Vmu Vc + V s Cross-
Specimen fc' a/d Fiber µ1)
(%) (%) (%) (kN) (kN) (kN) (kN) Vmu section
SP1 29.5 3 2.87 0.16 - - 151 79 - 288 0.8 5.09 12)
5)
SP2 29.5 3 2.87 0.16 0.13 A2 151 79 213 288 0.8 11.84 1
SP3 29.5 3 2.87 0.16 0.38 PEN 151 79 201 288 0.8 10.65 1
SP4 29.5 3 2.87 0.16 0.37 PET 151 79 184 288 0.8 11.42 1
SP5 31.7 3 2.87 0.16 0.19 PET 155 79 90 290 0.8 7.98 1
SP6 31.7 4 2.87 0.16 0.12 PET 155 79 60 223 1.05 9.05 1
SP7 31.7 4 2.87 0.16 0.06 PET 155 79 30 223 1.05 8.46 1
SP8 31.7 4 2.87 0.16 - - 155 79 - 223 1.05 7.40 1
SP9 31.7 4 3.59 0.16 0.12 PET 169 79 60 267 0.93 8.76 1
SP10 31.7 4 2.15 0.16 0.06 PET 151 79 30 177 1.3 10.41 1
SP11 31.7 4 2.82 0.2 0.25 PET 318 206 264 463 1.13 8.52 23)
SP12 31.7 4 2.82 0.2 0.125 PET 318 206 132 463 1.13 7.54 2
SP13 34.5 3 2.82 0.2 0.29 PET 327 105 308 637 0.84 7.76 2
SP14 23.7 3 2.82 0.09 0.42 PET 289 83 441 612 0.61 4.12 2
SP15 31.1 3 2.82 0.09 0.42 PEN 316 83 469 641 0.62 6.87 2
SC1s 28.4 2.9 4.5 0.15 - - 80 29 - 255 0.43 - 34)
SC3s 29.0 2.9 4.5 0.15 0.032 Carbon 81.5 29 44 255 0.43 - 3
SP2s 35.4 2.9 4.5 0.15 0.67 PET 87 29 97 257 0.45 5.19 3
SP3s 36.7 2.9 4.5 0.15 0.35 PET 88 29 50 257 0.45 2.97 3
Note: 1) Ductility ratio (=δu/δy), 2) 400x400 mm, 3) 600x600 mm, 4) 250x250 mm, 5) Aramid of high strength type
whose ultimate strength is 3246 MPa, fracturing strain is 4.1 % and elastic modulus is 79.5 GPa. 6) Notations: fc' is
concrete strength, a/d is shear span to depth ratio, ρt, ρw, ρf are ratios of tension reinforcement, stirrup and fiber sheet,
Vc, Vs, Vf are concrete, stirrup and fiber sheet contribution in shear, Vmu is bending strength in terms of shear force.

3. ENHANCEMENT MECHANISM OF DUCTILITY AND SHEAR STRENGTH

The ductility enhanced by PET jacketing increases with an increase in PET fiber ratio (comparing SP1, SP4 and SP5
in Table 1 and Figure 3 (a), and SP8, SP7 and SP6 in Table 1). PEN jacketing also increases the ductility
(comparing SP1 with SP3 in Table 1). At ultimate deformation δ u , no fracture was observed with PET and PEN
fiber sheets, while the aramid fiber sheet fractured in SP2. No fracture or yielding of jacketed sheets can be
considered to not only improve the ductility ratio, but also reduce negative slope of the falling branch in the load-
deformation curve (see Figure 3 (a)).

A previous study on shear strength of concrete beams with shear reinforcement (Sato et al. 1997) indicates that shear
strength depends on stiffness of both tension and shear reinforcement. If we apply this fact to reinforced concrete
columns in which the flexural yielding takes place before the shear strength is reached, the following can be said.
Once yielding of tension reinforcement, which means the reduction in stiffness, takes place, the potential shear
strength starts to decrease. Yielding of shear reinforcement, which means not only reduction in the stiffness but also
no increase in shear reinforcement component to carry shear force, further decreases the potential shear strength.
Figure 4 (a) shows the shear force contribution of concrete, steel shear reinforcement and PET fiber sheet, the last
two of which were calculated using their measured strains. The concrete component starts to decrease after the
flexural yielding and decreases even faster after the shear reinforcement yielding. The load-carrying capacity
decreases because the load-carrying capacity in shear becomes smaller than that in flexure. Small contribution of
PET fiber sheet is visible only after the yielding of steel shear reinforcement in Figure 4 (a).

It seems that stiffness of both tension and shear reinforcement controls the potential shear strength. This means that
FRP jacketing, which adds the stiffness of shear reinforcement, increases the potential shear strength resulting in
enhancement of ductility and more ductile manner with falling load-carrying capacity. However, fracture of FRP
would instantly eliminate the FRP contribution. PET fiber with a large fracturing strain can keep its contribution
and contribute better than steel reinforcement which is likely to yield at ultimate deformation.

591
Similar observations can be made with specimens SP2s and SP3s, which were originally designed to fail in shear
based on the JSCE formula for carbon and aramid fiber sheet jacketing. Both specimens showed shear failure after
flexural yielding around 220 kN (see Figure 3 (b)). The load-deformation curves of SP2s and SP3s are compared
with those of companion specimens with no jacketing (SC1s) and carbon fiber sheet (SC3s) with stiffness greater
than those in SP2s and SP3s, which show rather brittle behavior and smaller shear strength. In specimen SC3s the
carbon fiber sheet fractured. Shear force components in specimen SP2s are shown in Figure 4 (b). After yielding of
steel shear reinforcement the concrete component increases with a smaller rate and the component of PET fiber
sheet becomes more significant. The concrete component starts to decrease after the flexural yielding.

In order to estimate the ultimate deformation we have to predict shear deformation in hinge zone. The experimental
results indicate that shear deformation increases with total deformation and more quickly after yielding of tension
and shear reinforcement. It can be more than 10 % of the total deformation at ultimate deformation.

400
300
Load-Envelope Curve
200
300

Shear Force (kN)

100
V (kN)

0 200
SP1
-100 SP2 100 SC1s
SC1 SC3s
SC3
-200 SP3
SP2
SP2s SP3
SP3s
-300 SP4
0
SP5
-400 0 50 100
Deformation (mm)
-15 -10 -5 0 5 10 15
δ /δ y
(a) Specimens SP1 to SP5 (b) Specimens SC1s, SC3s, SP2s and SP3s
Figure 3: Load-Deformation Curve (Envelope)

SP2s
SP 2
Yielding of shear Yielding of shear
reinforcement 300 reinforcement
250
Shear Force (kN)

200
150 200
Shear Force (kN)

100 150
50 Vtotal
0 100
Vc+Vs
-150 -100 -50 -50 0 50 100 150 50
-100 SP10 Vtotal Vc
-150 SP10 Vc+Vs 0
-200 SP10 Vc 0 20 40 60 80 100 120
Displacement (mm) Deformation (mm)
(a) Specimen SP10 (b) Specimen SP2s
Figure 4: Shear Force Contributions

5. REFERENCES
Anggawidjaja, D., Ueda, T., Dai, J.G. and Nakai, H. (2006a). “Deformation capacity of RC wrapped with new
reinforced-fiber polymer with large fracture strain”. Cement and Concrete Composites (to be printed).
Anggawidjaja, D., Senda, M., Ueda, T., and Dai, J.G. (2006b). “Shear capacity strengthening using high fracturing
strain fiber material”, Proceedings of Tenth East Asia and Pacific Conference on Structural Engineering and
Construction (to be printed).
Sato, Y., Ueda, T. and Kakuta, Y. (1997) “Shear strength of reinforced and prestressed concrete beams with shear
reinforcement”, Concrete Library International of JSCE, No.29, pp.233-247.

592
Third International Conference on FRP Composites in Civil Engineering (CICE 2006)
December 13-15 2006, Miami, Florida, USA

APPLICATION OF NEWLY DEVELOPED GFRP STIRRUPS FOR SHEAR

REINFORCEMENT OF CONCRETE BEAMS
Jongsung Sim
(Professor, Hanyang University, Ansan, Kyeonggi-do, Korea)

Cheolwoo Park
(Assistant Professor, Kangwon National University, Samcheok, Korea)

Sungjae Park
(Ph. D Candidate, Hanyang University, Ansan, Kyeonggi-do, Korea)

Minkwan Ju
(Ph. D Candidate, Hanyang University, Ansan, Kyeonggi-do, Korea)

ABSTRACT
Recently, researches on high-durability concrete structure have remarkably been studied by adopting new
construction material, fiber reinforced polymer (FRP) composites. In conjunction with these research trends, the
shear capacity of RC beams reinforced with GFRP stirrup, which is developed in this study, was verified.
Experimental variables were shear span to depth ratio and spacing of shear reinforcement. From the results, similar
structural performance was found between the specimens reinforced with steel and GFRP stirrups with respect to
crack pattern, failure mode, and shear capacity. It is anticipated that the adaptability of concrete structure shear-
reinforced with GFRP stirrups can be considered in the future high performance concrete structures.

KEYWORDS
FRP composites, GFRP stirrup, high durability, shear capacity

1. INTRODUCTION
In recent years, fiber reinforced polymer (FRP) composite materials have been widely used in the field of
construction for concrete structures because FRP reinforcement has many advantages, as it offers a high-strength,
low-weight, superior resistant on the deteriorations such as corrosion, alkali reaction, de-icing salt, and freeze-thaw.
Especially, the use of FRP bars to replace steel reinforcement in concrete structures is a new technique, which is a
most effective way to prevent and to solve problems caused by the deteriorations.
FRP bars possess superior mechanical properties different from steel bars, for example, high tensile strength
combined with low elastic modulus and elastic brittle stress-strain relationship. These properties might be reflected
in structural design. Several experimental studies have been carried to verify the flexural and shear behavior of RC
beams reinforced with FRP bars. As a result, few design guidelines have been published for the design and
construction of concrete structures reinforced with FRP bars. However, these design guidelines emphasized the need
for more research to verify the performance and behavior of RC beams reinforced with FRP bars for shear and
flexure. Particularly the use of FRP as shear reinforcement for concrete structures has not yet been explored enough
to establish a rational model to predict the performance and behavior of RC member with FRP stirrup.
In this paper, in order to verify shear performance and behavior of RC beams reinforced with GFRP stirrup, which is
developed in this study, static four points loading tests were carried. Shear span to depth ratio (a/d), spacing of shear
stirrup were variables.

593
2. EXPERIMENTS
2.1 Material properties and test variables

Fig. 1 shows the GFRP rebar developed in the study herein and the nominal diameter of both steel and GFRP
stirrups are 10mm. Table 1 show mechanical properties of the materials used. A total of 12 reinforced concrete
beams (i.e., 9 stirrup reinforced beams and 3 control beams) have been tested. The dimension of test beams is
180mm×230mm×2000mm (W×H×L). The stirrups are symmetrically placed as shown in Fig. 2. Designed
compressive strength of concrete was 27MPa. In order to investigate the effects of shear reinforcing, three different
reinforcing materials (non-stirrup, steel-stirrup, and GFRP-stirrup) and stir-up spacing (100mm and 200mm) were
considered. In the testing shear span to depth ratios varied as a/d = 1.7, 2.0, and 2.3. More details on the specimens
and variables are provided in Table 2 and Fig. 2.

(all dimension in mm)

50 3@100 3@100
50
25 25

400 1200 400

(a) Non-Stirrup reinforcing

9@100 9@100
50 50
25 25

Fig. 1 Developed GFRP stirrup 400 1200 400

(b) Steel and GFRP Stirrup reinforcing (Stirrup Spacing;
100mm)
Table 1 Mechanical properties of materials
2@100 3@200 3@200 2@100
Materials σtensile [MPa] εultimate E [GPa] 50 50

Glass fiber 400.0 4.8 - 25 25

Epoxy resin 73.8 5.0 -

GFRP rebar 750.0 1.2 49 400 1200 400

(c) GFRP Stirrup reinforcing (Stirrup Spacing; 200mm)
Steel rebar 300.0 - 200 Fig. 2 Test specimen details

Table 2 Test variables

Specimens Span to depth ratio (a/d) Spacing of shear stirrup (mm) Type of shear strengthening
S1.7-CON (None) - None
S1.7-CON 100 Steel
1.7
S1.7-100 100 GFRP
S1.7-200 200 GFRP
S2.0-CON (None) - None
S2.0-CON 100 Steel
2
S2.0-100 100 GFRP
S2.0-200 200 GFRP
S2.3-CON (None) - None
S2.3-CON 100 Steel
2.3
S2.3-100 100 GFRP
S2.3-200 200 GFRP

2.2 Test set-up

Fig. 3 shows the experimental test set-up. Each RC beams was simply supported and loading was applied under the
four-point bending configuration. The loading was controlled at a speed of 1 mm/min. One LVDT were used to
measure the displacement of the specimens at mid-span during testing. Fig. 4 shows the loading condition according
to the changing of the shear span to depth ratio (a/d = 1.7, 2.0, and 2.3).

594
 A ctuator

C oncrete
gauge

H indg e LVD T H indge

Fig. 3 Experimental test set-up

P P P P P P
420 420 370 370 320 320

230 230 230

400 1200 400 400 1200 400 400 1200 400

(a) a/d: 1.7 (b) a/d: 2.0 (c) a/d: 2.3

Fig. 4 Details of specimens according to the shear span to depth ratio

2.3 Test results and discussion

Though it is very difficult to compare directly the shear reinforcing capacity of the specimens since the material
properties are completely different, in the study herein, the authors are willing to compare the test results as an
initiative study on the application of GFRP stirrups. Table 3 summarizes the test results and Fig. 5 shows load-
deflection curves according to the shear span to depth ratios. During the test, the GFRP stirrup of all specimens was
not fractured and no debonding between the GFRP stirrups and concrete was observed. From the test results, as the
shear span to depth ratio (a/d) of specimens increased shear failure load and shear failure angles decreased. Shear
failure load of GFRP stirrup reinforcement specimen is similar to the specimens with steel stirrup that implies the
application of the developed GFRP stirrup may provide as comparable shear reinforcing capacity as the steel stirrups.
As the a/d ratio increased, the failure mode became governed by the interaction between bending and shear.

Table 3 Test results

Specimens CON (NONE) CON Spacing (100mm) Spacing (200mm)
Load (kN) 96.3 135.9 141.4 135.4
Displacement (mm) 5.7 11.6 16.5 9.1
S1.7
Failure mode S S S S
Failure angle 42˚ 46˚ 47˚ 48˚
Load (kN) 95.2 127 119.8 91.4
Displacement (mm) 5.22 13 17.5 5.2
S2.0
Failure mode S BS S S
Failure angle 37˚ - 45˚ 42˚
Load (kN) 88.9 107.9 105 91.9
Displacement (mm) 10.6 10.4 12.6 8.1
S2.3
Failure mode S BS FS S
Failure angle 36˚ 44˚ 43˚ 37˚
* S: Shear failure mode BS: Bending and Shear Failure mode

595
 (a) S 1.7 (a/d=1.7) (b) S 2.0 (a/d=2.0) (c) S 2.3 (a/d=2.3)
Fig. 5 Load-deflection curves according to a/d ratio

3. Conclusions
An experimental investigation was conducted on the application of the developed GFRP stirrups for the shear
reinforcement in concrete beams. According to the results from this investigation, the following conclusions can be
made:

1. From the static loading test, the crack pattern and failure behavior of RC beam specimens reinforced with
GFRP stirrup were similar with that of RC beam specimens used steel stirrup. No obvious fractures and
debonding of GFRP stirrups were observed during the test.

2. Below 2.0 of the a/d ratio, the ultimate loading of the specimens were similar. However, after the peak load,
the strength drop of the specimens reinforced with the GFRP stirrup was more significant. This may be due
to the bonding between concrete and GFRP stirrups and different material properties of GFRP comparing
to steel.

3. When a/d was 2.3, the ultimate strengths of the specimens were decreased compared to the other a/d ratio.
Particularly the specimen with GFRP stirrup and 200mm spacing the strength was even worse than the
Non-stirrup specimens.

4. From this study, the application of the newly developed GFRP stirrup was investigated. Even though the
mechanical properties of the GFRP is completely different from the steel, when it is used as a shear
reinforcement in concrete beams based on the current design specifications, its shear reinforcing capacity
seems to be applicable to the concrete structures. It is very obvious that definitely required are further
researches on the design method and improvement of GFRP stirrups in manufacturing.

ACKNOWLEDGEMENTS
This research is funded from Korea Institute of Construction & Transportation Technology Evaluation and Planning
[KICTTEP]. Authors thank their support.

REFERENCES
A. Clader, A.R. Mari, "Shear design procedure for reinforced normal and high-strength concrete beams using
artificial neural networks. Part Ι: beam without stirrups", Engineering Structures 26 (2004) 917~926.
A. Clader, A.R. Mari, "Shear design procedure for reinforced normal and high-strength concrete beams using
artificial neural networks. Part II: beam without stirrups", Engineering Structures 26 (2004) 927~936.

596
Third International Conference on FRP Composites in Civil Engineering (CICE 2006)
December 13-15 2006, Miami, Florida, USA

REPAIR OF CORROSION DAMAGED CONCRETE BEAMS WITHOUT

SHEAR REINFORCEMENT USING CARBON FIBER REINFORCED
POLYMER SHEETS

Abdullah Al-Saidy
(Assistant professor, Sultan Qaboos University, Muscat, Sultanate of Oman)

ABSTRACT
Strengthening/repair of existing reinforced concrete structures has become one of the important issues in the field of
civil engineering. Corrosion of reinforcement in Oman and the Middle East region is a serious problem and is the
main cause of concrete structures deterioration costing millions of dollars even though the majority of such
structures are at the early age of their expected service life. This paper presents the experimental results of
damaged/repaired reinforced concrete beams. The experimental program consisted of reinforced concrete
rectangular beam specimens reinforced with flexural reinforcement only without shear reinforcement and exposed to
accelerated corrosion. The corrosion rate was varied between 5 to 7.5% which represents loss in cross sectional area
of the steel reinforcement in the tension side. Damaged beams were repaired by bonding Carbon Fiber Reinforced
Polymer (CFRP) U-straps and in some by CFRP sheets to the tension side to restore the strength loss due to
corrosion. Test results showed that it is possible to use external shear reinforcement to enhance the strength of
damaged concrete beams. In addition, combining U-straps and CFRP sheets is very effective to restore the lost
strength in corrosion damaged beams.

KEYWORDS
Strengthening, rehabilitation, retrofitting, corrosion, advanced composite materials, CFRP sheets.

1. INTRODUCTION
The deterioration of reinforced concrete structures resulting from corrosion of steel reinforcement is a worldwide
problem, and the cost of repairs is substantial. Reinforcement corrosion is induced primarily by the ingress of
chloride (Cl-), water, and oxygen into uncontaminated concrete. Fresh concrete protects the steel reinforcement
from corrosion by the formation of a passive layer in the highly alkaline (pH>13) environment of fresh concrete.
However, this passive layer will be destroyed by Cl- ions which initiate the corrosion process. Corrosion damages
reinforced concrete in three ways. First, it results in the reduction of effective cross – sectional area of the
reinforcing bars. Second, it produces expansive corrosion products, causing cracking and spalling of the concrete
cover. Third, it causes loss of bond between the concrete and the reinforcing steel. Hence, the serviceability as well
as the ultimate capacity of the damaged reinforced concrete member is affected (Almusallam et al., 1996; Baweja et
al., 1999; Umoto et al. 1984).

Repair or strengthening with fiber reinforced polymers (FRP) has gained some acceptance in recent years. It
involves the external bonding of FRP sheets or plates to RC beams and slabs, or confinement of RC columns.
Strengthening with FRP is simple and does not involve heavy equipments. Numerous studies have shown that
repair and strengthening of corrosion damaged RC beams with FRP sheets or plates is efficient in restoring the
strength of concrete members (Bonacci et al., 2000; Kutarba et al., 2004, Soudki and Sherwood, 2000).

597
This paper presents the research findings of an experimental study involving a case of damage due corrosion in the
flexural reinforcement coupled with absence of shear reinforcement. This case represents a severe case of damage
where it assumes that there is total damage in the stirrups and partial damage in the flexural reinforcement.

2. TEST PROGRAM
A total of 7 reinforced concrete beams were tested in this study as summarized in Table 1. Beam C0 was a control
beam with no corrosion, while beams C (5%) and C (7.5%) were control beams with 5% and 7.5% corrosion (mass
loss in reinforcement). Beams RU5 and RU7.5 were damaged beams with corrosion of 5% and 7.5%, respectively
and then were repaired by applying six CFRP U-straps as shown in Figure 1. The remaining two beams, beams
SUL5 and SUL7.5 were damaged with corrosion of 5% and 7.5%; then were strengthened by bonding one layer of
CFRP sheet along the tension side of the beam followed by attaching CFRP U-straps as shown in Figure 1

Table 1 Beams description

Specimen Corrosion level Remark
designation (mass loss%)
C0 0% Control beam
C5 5 % Control 5% corrosion beam
C7.5 7.5 % Control 7.5% corrosion beam
RU5 5% 5% corrosion with CFRP U-straps
RU7.5 7.5 % 7.5% corrosion with CFRP U-straps
SUL 5 5% 5% corrosion with CFRP straps and sheet
SUL7.5 7.5 % 7.5% corrosion with CFRP straps and sheet

The specimens were 2.7 m long, 100 mm wide and 150 mm high. All beams were reinforced with two 10 mm
diameter bars Grade 400 tensile reinforcement (Area of steel = 157 mm2). The reinforcing steel was extended 60
mm beyond the end of the concrete for the purpose of making external electrical connections for the accelerated
corrosion process. The clear concrete cover was 20 mm on all sides of the specimen. No stirrups were used in all
specimens.

Unidirectional carbon fiber sheets were used for the U-straps and the bottom longitudinal sheets. Thickness of the
sheet was 0.11 mm (dry fibers), tensile strength of 3800 MPa, modulus of elasticity of 240 GPa, and ultimate
elongation of 1.55%. The composite (fiber and epoxy) thickness of the CFRP sheet was 1 mm on average. The
concrete had a 28-day compressive strength of 38 MPa with a maximum aggregate size of 10 mm. The concrete
mix was proportioned as follows, aggregate : sand : water : cement = 60 : 67 : 16: 25, with a water to cement ratio of
0.64. Reinforcing steel had yield strength of 460 MPa.

Bonding of the CFRP to the concrete was achieved by using epoxy adhesive. Prior to applying the epoxy and
CFRP, the surface of the concrete was prepared by grinding the concrete in the area to receive the CFRP. The
beams were tested after one week from applying the CFRP.

598
 Figure 1. Test set-up and dimensions

The casting of each beam was done in three layers; after placing the first layer (at the level of the tensile steel bars)
salt was spread along this layer except one beam (beam C0 -control beam). The amount of salt was approximately
1% by weights of cement. This was used to simulate chloride ions contamination and to accelerate corrosion. After
28 days curing in room conditions, the six beams were placed inside a tank which has salted water; the salt
concentration was about 3% by weight of water. To induce corrosion in the reinforcement, the rebars were
connected to a power (voltage) source where a current was applied to accelerate the corrosion process as shown in
Figure 2. Stainless steel rebars were placed parallel to the beams in the tank to act as cathode and were connected
to the negative charge of the power source. To obtain a theoretical 5% corrosion (or 5% mass loss in reinforcing
bar) it was found that the time required to produce this mass loss was 14 days of continuous application of 487 mA
current in each beam according to Farady's law.

Figure 2. Schematic of accelerated corrosion set-up

3. TEST RESULTS AND DISCUSSION

3.1 Beams With 5% Corrosion
Following the accelerated corrosion phase, the beams were left for two days to dry. The beams were then repaired
with CFRP and were left for a week for the CFRP to cure under room temperature. A four-point flexural test was
carried out to all beams up to failure. The load - deflection curve for beams with 5% corrosion beams are shown in
Figure3a. Behavior of beam C0 (control beam with 0% corrosion) is of a typical under-reinforced beam exhibiting
large deformation beyond the yield point before it failed by crushing of concrete. Beam C5 (5% corrosion) failed
prematurely due to bond failure along the interface at the reinforcement and concrete (see Figure 4a). This clearly
shows the importance of confinement provided by stirrups especially when corrosion is present. Note that the
predicted load was approximately 14.0 kN. It is clear that bond failure caused this beam to fail prematurely when
compared to the theoretical failure load that assumes complete bond between the concrete and steel reinforcement.
Beam RU5 had 5% corrosion and repaired by external U-shaped CFRP straps. Adding U-straps improved the
strength and ductility (deflection) as observed from the response of beam RU5 which failed by crushing of concrete.
Adding CFRP sheet and U-shaped straps (beam SUL5) increased both strength and the stiffness of the repaired
beam. The strength of beam SUL5 was higher by 37% above the uncorroded control beam (C0). Beam SUL5 also
failed by crushing of concrete after a large deflection beyond the yielding load.

3.2 Beams With 7.5% Corrosion

The load - deflection curves for beams with 7.5% corrosion beams are shown in Figure3b. Beam C7.5 had 7.5%
corrosion, this beam failed prematurely due to bond failure for similar reasons as discussed earlier in beam C5. On
the other hand, beam RU7.5 (repaired by external U-straps) showed an improved behavior by attaining adequate
strength and ductility as a result of the confining effect of the U-straps. Failure of beam RU7.5 was due crushing of
concrete. Following the accelerated corrosion phase, Beam SUL7.5 was severely cracked along the reinforcement
on the side where the rebars were connected to the voltage source. It is believed that more moisture penetrated the
beam through the cracks from the partially immersed bar ends. This produced higher concentration of corrosion on
one side of the beam. This beam failed prematurely in a brittle manner as one of U-straps debonded and the crack
along the beam opened up as shown in Figure 4b. This beam failed in shear-compression mode.

599
 25 25
SUL5
20 20

C0

Load (kN)
Load (kN)

15 15 C7.5
C0
RU5 SUL7.5 RU7.5
C5
10 10

5 5

0
0
0 10 20 30 40 50 60 70 80
0 10 20 30 40 50 60 70 80
Deflection (mm) Deflection (mm)
(a) (b)
Figure 3. Load-deflection (a) 5% corrosion beams ; (b) 7.5% corrosion beams

(a) (b)
Figure 4. (a) bond failure of beam C5 ; (b) shear-compression failure of beam SUL7.5

4. CONCLUSIONS
Corrosion of reinforcement weakens the strength of RC members as well as the bond between the reinforcing steel
and the surrounding concrete. It was shown that stirrups are very critical in corrosion damaged beams to enhance
the strength. The use of external CFRP U-straps can improve the performance of corrosion damaged beams by
applying confining action. In addition, applying CFRP sheets to the tension side of beam increases the strength and
stiffness considerably provided bond failure is prevented. However, if the cracks are wide and extensive probably
continuous shear reinforcement is needed to prevent any brittle failure due to the lack of internal stirrups.

REFERENCES
Almusallam, A., Al-Gahtani, A., Aziz, A. , Dakhil, F. and Rasheeduzzafar (1996). "Effect of reinforcement
corrosion on flexural behavior of concrete slabs", Journal of Materials in Civil Engineering, pp 123 -127.
Baweja, D., Ropert, H and Sirivivatnanan (1999)."Chloride-induced steel corrosion in concrete." ACI Materials
Journal, V.96, No.3, pp 306-313.
Bonacci, J. F. and Maaleej, M. (2000)."Externally bonded fiber reinforcement polymer for rehabilitation of
corrosion damaged concrete beams." ACI Structural Journal, September –October, pp 703-711.
Kutarba, M. P., Brown, J. R. and. Hamilton, H. R (2004). "Repair of Corrosion Damaged Concrete Beams with
Carbon Fiber-Reinforced Polymer Composites." Proceedings of COMPOSITES 2004, Tampa, Florida USA.
Soudki, K., and Sherwood, T. (2000). " Behaviour of reinforced concrete beams strengthened with carbon fiber
reinforced polymer laminates subjected to corrosion damage." Canadian Journal of Civil Engineering, 27, pp 1005-
1010.
Umoto, T, Tsuji, K., and Kakizawa, T. (1984). "Deterioration mechanism of concrete structures caused by corrosion
of reinforcing bars". Transactions of the Japan Concrete Institute, 6, pp 163-177.

600
Third International Conference on FRP Composites in Civil Engineering (CICE 2006)
December 13-15 2006, Miami, Florida, USA

SHEAR RETROFIT OF LOW STRENGTH REINFORCED CONCRETE

SHORT COLUMNS WITH GFRP COMPOSITES
Alper Ilki
(Assoc.Prof.Dr., Istanbul Technical University, Istanbul, Turkey)

Idris Bedirhanoglu, Ismail Hakki Basegmez, Cem Demir

(Graduate Student, Istanbul Technical University, Istanbul, Turkey)

Nahit Kumbasar
(Emeritus Prof.Dr.,Istanbul Technical University, Istanbul, Turkey)

ABSTRACT
In this experimental study, the behavior of shear critical reinforced concrete columns under constant axial load and
reversed cyclic lateral loads is investigated. The specimens were constructed using low quality concrete, and
insufficient transverse reinforcement, intentionally for reflecting the characteristics of relatively older existing
buildings. Test results showed that the brittle shear behavior of the reference specimen was enhanced in different
extents depending on the GFRP (glass fiber reinforced polymer) sheet retrofitting scheme. The retrofitted
specimens, while not being able to reach flexural capacity due to the effects of diagonal compression stresses,
exhibited much better performance with respect to reference specimen both in terms of strength and deformability.

KEYWORDS
Columns, Concrete, Fiber, Retrofit, Shear.

1. INTRODUCTION
Many existing buildings suffer severe seismic damage during earthquakes as a consequence of poor structural
characteristics, such as low quality concrete and inadequate transverse reinforcement. The deficiencies related with
low quality of concrete and inadequate transverse reinforcement, causing insufficent shear capacity, can be
overcome by jacketing the structural members using fiber reinforced polymer composite sheets. Experimental
research work on shear retrofit of concrete members with externally bonded fiber reinforced polymer composites are
generally on beam specimens, which do not resist axial forces as well as shear stresses. Reversed cyclic shear tests
on columns are limited, (Maruyama et al., 2001, Harmon et al., 2002, Furuta et al., 2003). In this experimental
study, the behavior of shear critical reinforced concrete members under constant axial load and reversed cyclic
lateral loads is investigated. Four specimens were tested after jacketing with GFRP sheets of different thickness and
details, while one specimen was tested before retrofitting as the reference specimen. The specimens were
constructed using low strength concrete (f′c~10 MPa), and insufficient transverse reinforcement (R8/200)
intentionally for reflecting the characteristics of relatively older existing structures. The experimental findings are
outlined by the damage patterns, load-displacement relationships, ductility characteristics, as well as attained
transverse strains on FRP sheets, together with analytical shear strength predictions of the specimens.

2. TESTS AND COMPARISON WITH ANALYTICAL PREDICTIONS

The specimens, representing the column parts between the mid-heights of succeeding stories, were 3 m high and 150
mm × 200 mm in cross-section. The longitudinal bars were 4D20 and transverse reinforcement in the testing region
was R8/200. The net concrete cover thickness was 20 mm from outside of transverse reinforcement. The geometry
and reinforcement details of the specimens and construction phases are presented in Figure 1 and 2, respectively.

601
The standard cylinder concrete compressive strength at the day of testing, number of GFRP plies, level of axial load
(ν), analytical and experimental shear strengths of the specimens are shown in Table 1. ν is calculated by Eq. (1).
Characteristics of steel reinforcement and GFRP sheets are given in Table 2. While LS-G-1/2 was jacketed with
straps, all other retrofitted specimens were jacketed continuously in the shear span. It should be noted that for both
cases, jackets were continuous around four sides.
R8 Testing zone
A
R8 2D20 140 2D20 R8
2 FLA-3 2 FLA-3 D20
50 100 50 200 140 R8

200

200
2D20 150
2D20 YFLA-5
A
Section A-A
1200 600 1200
3000
* All Dimensions are in mm

Figure 1: Geometry and Reinforcement Details of the Specimens

Figure 2: Specimen Construction and Retrofitting Phases

Table 1: Specimen Characteristics, Analytical and Experimental Shear Capacities

j f'cj f'ccj1
FRP ν Vc2 Vs3 Vf4 Vr,analytical5 Vmax6 Vr,experimental7
Specimen
(days) (MPa) plies (MPa) (kN) (kN) (kN) (kN) (kN) (kN)
LS-0-1 332 10.9 0 10.9 0.19 18.3 31.1 0 49.4 63.3 46.5
LS-G-1/2 347 11.0 18 11.4 0.19 18.3 31.1 12.0 61.3 66.2 66.0
LS-G-1 360 11.0 1 11.8 0.19 18.3 31.1 23.9 73.3 68.5 73.5
LS-G-2 370 11.0 2 12.9 0.19 18.3 31.1 47.8 97.2 74.9 80.0
LS-G-3 375 11.0 3 14.1 0.19 18.3 31.1 71.8 121.1 81.9 83.5
1
calculated as proposed by Ilki et al., 2004, εhrup is assumed as 0.004, 2calculated by Eq. (2), TS500, 2000,
3
calculated by Eq. (3), 4calculated by Eq. (4), 5calculated by Eq. (5), 6Vmax=0.22f'ccjbd, TS500, 2000, 7shear force
corresponding to flexural failure is 106 kN, 8100 mm wide strips with 100 mm clear spacing between them

Table 2: Mechanical Characteristics of Steel and GFRP Reinforcement

Reinforcement Diam. Yield strength Thickness Tensile strength of Elasticity Ultimate

(mm) of steel bars (mm) GFRP modulus of GFRP elongation
fy (MPa) ff (MPa) Ef (MPa) of GFRP
Longitudinal bars 19.8 536 - - - -
Transverse bars 7.8 370 - - - -
GFRP sheets - - 0.23 1700 65000 0.028

N
ν= (1)
bhf cj′
 N
′ bd1 + 0.07  , f ctj
Vc = 0.52f ctj ′ = 0.35 f cj′ (2)
 bh 

602
 A sw df yw
Vs = (3)
s

Vf = 2n f t f 0.004E f h (4)

Vr = Vc + Vs + Vf (5)

In these tables and equations N, b, h, d, f'cj, f'ctj are axial load, width, depth and effective depth of cross-section,
concrete compressive and tensile strengths at the day of testing. Vc, Vs and Vf are contributions of concrete,
transverse reinforcement and GFRP sheets to the shear resistance, and Vmax is the maximum shear force permitted to
prevent diagonal compression failure as given by TS500, 2000. Asw, fyw and s are cross-sectional area, yield strength
and spacing of transverse reinforcement. nf, tf, ff, εhrup and Ef are the number of plies, effective thickness, tensile
strength, rupture strain and elasticity modulus of GFRP, and f'ccj is the compressive strength of GFRP confined
concrete. The measuring system included diagonal deformation measurements, GFRP strains, strains of longitudinal
and transverse reinforcement as well as curvatures in potential plastic hinging zones, Figure 3. The envelopes of
shear force-drift ratio relationships and appearance of specimens around drift ratio 0.035-0.040 are presented in
Figure 4 and 5. As seen in these figures, significant enhancement was obtained in the behavior in terms of strength
and deformability. However, although longitudinal bars yielded in all retrofitted specimens, none of them could
reach their analytical flexural strength. The reference specimen prematurely failed in shear before longitudinal bars
yielded. The transverse bars also yielded in all specimens. The yielding of transverse bars was retarded, with the
increasing of number of GFRP plies. As it can be understood from Table 1 and Figure 4, the failure of retrofitted
specimens was due to a combination of flexural and diagonal compression stresses.

100
75 LS-0-1
Shear Force (kN)

50 LS-G-2
25 LS-G-3
0
-25
-50
-75
-100
-0.02 0.00 0.02 0.04 0.06 0.08 0.10

Average Shear Deformation (mm/mm)

Figure 3: Loading and Measuring Setup and Average Strains Measured on Shear Span

120
LS-0-1 100
LS-G-1/2 80
LS-G-1
60
LS-G-2*
40
Shear Force (kN)

LS-G-3
Flexural Capacity 20
0
-0.100 -0.075 -0.050 -0.025 -200.000 0.025 0.050 0.075 0.100
-40 LS-G-3
-60
-80
-100 * Loaded monotonically from drift ratio -0.05 to -0.10
-120
Drift ratio

Figure 4: Shear Force-Drift Ratio Envelopes and Shear Force Corresponding to Flexural Capacity

603
 (a) LS-0-1 (δ/L=-3.5, δ=-15.75 mm, V=-27.5 kN) (b) LS-G-1/2 (δ/L=-3.5, δ=-15.75 mm, V=-63 kN)

c) LS-G-1 (δ/L=-4.0, δ=-18.00 mm, V=-69 kN) (d) LS-G-3 (δ/L=-3.5, δ=-15.75 mm, V=-81 kN)

Figure 5: Damages of Specimens at δ/L=-3.5 and 4.0 (δ: drift, L: shear span)

At the end of the tests, the transverse GFRP sheets of LS-G-1/2 were totally fractured, while an approximately 50
mm wide part of GFRP jacket in the vicinity of stub was fractured in LS-G-1. The GFRP sheets were not fractured
in LS-G-2 and LS-G-3. However, there was an apparent swelling in the cross-section close to the stub due to plastic
compressive flexural and shear deformations.

3. CONCLUSIONS
Reinforced concrete columns representing typical deficiencies of relatively older buildings in Turkey were tested
under constant axial and reversed cyclic shear forces before and after retrofitting with externally bonded FRP sheets.
The original members were constructed using low strength concrete and inadequate transverse reinforcement.
Consequently, the reference specimen failed prematurely in a brittle manner, due to shear effects. The retrofitted
specimens, while not being able to reach flexural capacity, exhibited much better performance both in terms of
strength and deformability. The failure of retrofitted specimens was due to combined effect of flexure and shear,
particularly the diagonal compression stresses.

4. REFERENCES
Furuta, T., Kanakubo, T., and Fukuyama, H. (2003). “Evaluation of shear capacity of RC columns strengthened by
continuous fiber”, Proceedings of Sixth International Symposium on FRP Reinforcement for Concrete Structures,
Editor: K.H. Tan, National University of Singapore, Singapore, pp. 507-516.
Harmon, T.G., Gould, N.C., Ramakrishnan, S., and Wang, E.H. (2002). “Confined concrete columns subjected to
axial load, cyclic shear, and cyclic flexure-part I: analytical models”. ACI Structural Journal, Vol. 99,No. 1, pp 32-
41.
Ilki, A., Kumbasar, N., and Koc, V. (2004). “Low strength concrete members externally confined with FRP sheets”.
Structural Engineering and Mechanics, Vol. 18, No. 2, pp 167-194.
Maruyama, K., Nakai, H., Katsuki, F., and Shimomura, T. (2001). “Improvement of shear and ductility of reinforced
concrete columns by wrapping of continuous fiber-reinforced polymer sheet”. Advanced Composite Materials, Vol.
10, No. 2-3, pp 119-126.
Turkish Standards Institute (2000). TS500, Requirements for Design and Construction of Reinforced Concrete
Structures, Ankara, Turkey.

604
Third International Conference on FRP Composites in Civil Engineering (CICE 2006)
December 13-15 2006, Miami, Florida, USA

STATE-OF-PRACTICE OF FRP STRENGHTHENED RC GIRDERS IN

SHEAR
Abdeldjelil Belarbi
(Distinguished Professor, University of Missouri-Rolla, Rolla, Missouri, United States)

Ashraf Ayoub
(Assistant Professor, University of Missouri-Rolla, Rolla, Missouri, United States)

Sang-Wook Bae
(Research Associate, University of Missouri-Rolla, Rolla, Missouri, United States)

ABSTRACT
FRP systems are now becoming a widely accepted method of strengthening concrete structures. The acceptance and
utilization of these new strengthening techniques depend on the availability of clear design guidelines, installation
procedures and construction specifications. Standard specifications exist for all commonly used traditional materials
in civil engineering structures. At this time, design specifications for FRP use are still under development. This
paper presents an overview of the state-of practice of existing analytical and experimental studies on the use of FRP
for shear strengthening of RC and PC girders. A summary of current design guidelines as a result of these studies
will also be outlined. A statement of current research needs will then be provided.

KEYWORDS
Design guidelines, FRP, Shear, Strengthening.

1. INTRODUCTION

A significant portion of the U.S. infrastructure is in urgent need of strengthening and rehabilitation. There is an
urgent need for innovative solutions that offer real advantages over traditional methods, particularly in terms of
performance, simplicity of application, speed of execution, and ease of handling. FRP systems have been developed
through extensive materials and structural testing under the guidance of practicing engineers, government officials,
and university researchers. FRP systems have been used on a project-specific basis for the last two decades. They
are now becoming a widely accepted method of strengthening concrete structures. The acceptance and utilization of
these new strengthening techniques depend on the availability of clear design guidelines, installation procedures and
construction specifications. Standard specifications exist for all commonly used traditional materials in civil
engineering structures. At this time, design specifications for FRP use are still under development.

The results of several experimental investigations have shown that FRP systems can be effective for increasing
ductility and strength to structural members such as columns and girders. As most of the research focused on
strengthening of axial or flexural members, there are less experimental and analytical data on the use of FRP
systems for shear strengthening of girders. Shear strengthening with FRP is still under investigation and the results
obtained thus far are scarce and sometimes controversial. Even in traditional reinforced concrete members without
FRP, the shear design is a complex challenge and uses more empirical methods as compared to axial and flexural
design methods. Adding FRP to the equation, with its specific design issues, would bring another level of
complication to the design. These FRP-related shear design issues and lack of comprehensive analytical and
experimental models are the main motivation of several research studies worldwide. A thorough understanding of
the shear design problem along with the development of design method for FRP shear strengthening is needed.

605
2. DESIGN ISSUES OF FRP STRENGHTHENED RC AND PC GIRDERS IN SHEAR

It is well known that the shear behavior of RC girders is influenced by various factors such as tensile strength of
concrete, longitudinal and transverse reinforcement indices, shear-span to depth ratio, member size, and axial forces.
The number of parameters affecting the shear behavior increases if the RC/PC girder is strengthened with FRP
composite materials. This is because FRP-strengthened girders show various types of failure modes including
debonding, delamination, fracture of FRP, as well as peeling off of concrete cover with wide inclined cracks. It is,
therefore, rather difficult, because of the failure mechanisms, to identify the material parameters controlling the
behavior of FRP-strengthened girders at ultimate conditions.

Recently, a comprehensive research database was presented by Triantafillou and Antonopoulos (2000) and has been
updated and enriched later by Bousselham and Chaallal (2004). This database includes more than 160 test results.
The database was analyzed in terms of the following parameters: (a) properties of FRP composites, (b) shear-span to
depth ratio a/d, (c) shear steel reinforcement ratio, (d) longitudinal steel reinforcement, and (e) scale effect. The
analytical work was published with the objectives to synthesize the findings of the studies carried and to examine
and analyze the parameters that have the greatest influence on the shear behavior of RC members strengthened with
FRP laminates. The collected data was analyzed in order to assess the investigated test parameters. It was found that
most of the experimental studies focused on determining the most effective shear strengthening scheme; while other
influencing parameters, such as the longitudinal and transverse reinforcement ratio, the concrete strength, and the
effect of pre-damage, has been investigated but only in a limited number of studies. Furthermore, in many studies,
several parameters were varied simultaneously which made it difficult to decouple the effect of each parameter.

2.1. Existing Analytical Models

In current design methods, the total shear resistance of an RC beam, Vn , is expressed as the sum of the shear resisted
by concrete, Vc , and that resisted by the transverse steel reinforcement, Vs . If the member is strengthened for shear
with FRP laminates, an additional term, representing the contribution of FRP is added to the equation. To-date
about thirteen analytical models have been developed to determine the shear resistance of RC member strengthened
in shear with FRP sheets. The contribution of FRP to the shear resistance is often idealized as analogous to that of
the steel shear reinforcement. However, it should be noted that FRP seldom reach their full strength due to the
inherently different failure mechanisms. Most available models express the contribution of FRP to the shear
capacity as a function of the effective strain of FRP, which is typically expressed as a fraction of the ultimate strain.
The effective strain of FRP is largely dependent on the failure modes. The performance of the existing models has
not yet been fully verified through a comprehensive comparison with available experimental data. Typically, each
model tends to show a good agreement with the experimental data that was used to develop it, but is not verified
against other data to identify the most promising model. Furthermore, current data do not address all of the
parameters influencing the shear behavior of girders. Lastly, data derived from full-scale tests is not available to
calibrate the current models for large-scale girders such as AASHTO-type structure.

2.2 Existing Design Guidelines and Standards

Currently, design guidelines for the use of FRP for strengthening of structurally deficient RC members are available
in several countries. A brief discussion about all design guidelines mentioned above is presented in this section.
ACI 4402R-02 (2002): The ACI guidelines are the most thorough and complete specifications to date. The ACI
440-2R document provides strength reduction factors based on the expected failure mode, and is consistent with
ACI 318-99. The document also specifically addresses environmental effects through reduction factors associated
with various exposures. Fatigue and creep effects are addressed using prescribed FRP stress limits.
Canadian CAN/CSA S806-02 (2000): The Canadian specifications are the only formalized design codes addressing
externally bonded FRP reinforcement for concrete. For strengthening applications, CSA S806 considers all possible
failure modes rather than just concrete crushing. Durability is addressed through reference to CAN/CSA S478-95
Guideline of Durability in Buildings. Although a single limiting FRP tensile strain (0.007) is provided,
delamination and debonding are addressed by directing the designers to “currently available information appropriate
to the combination of sheets and adhesive”.
Japanese Recommendations for upgrading of concrete structures with use of CFRP sheets (JSCE, 2000): JSCE
adopted a performance-based approach to the design of externally bonded FRP materials. In addition to verifying

606
flexural and shear capacity, flexural crack width and protection of the concrete substrate from chloride ion
penetration are considered explicitly in the JSCE Recommendations.
European fib-TG 9.3 Bulletin 14 (2001): This document was produced by fib Task Group 9.3 and also represents a
combination of guidelines and state-of-the-art research. The fib-Bulletin 14 recognizes the difference in expected
performance between FRP material types and between preformed and wet lay-up FRP systems. Delamination and
debonding are addressed using a simplified bilinear bond model and by also addressing the effects of the loss of
composite actions between the FRP and concrete substrate.
ISIS Design Manual 4 (2001): This document provides considerable guidance and a number of design examples
for the use of externally bonded FRP. The document, however, is written as a state-of-the-art report, referring to the
recommendations of others rather than making its own design recommendations. The ISIS Design Manual 4
typically references to the recommendations of ACI 440.2R.
Great Britain Technical Report 55 (Concrete Society 2000): This report is similar to ISIS Design Manual 4 and fib
Bulletin 14 in its approach and scope. The report, however, addresses more practical construction issues associated
with the use of externally bonded FRP materials.

3. CURRENT PRACTICE AND FUTURE RESEARCH NEEDS

Based on the current state of knowledge, it is evident that further research investigations are needed to develop
rational design guidelines for RC beams externally strengthened with FRP composite materials. Among others, the
following surface to be the main controlling parameters that need further studies.
Properties of FRP Composite Materials: The determination of effective strain of FRP and its maximum limits are
keys to the accurate prediction of its shear contribution. Although many empirical equations currently exist, they
need to be evaluated through a larger set of experimental data.
Shear-span to Depth Ratio: The shear resistance of FRP can be affected by the shear-span to depth ratio. Several
research studies showed that current empirical equations could not accurately predict the shear resistance of FRP for
cases with low shear-span to depth ratio.
Transverse and Longitudinal Steel Reinforcement: Recent research studies showed that the interaction between
transverse steel reinforcement and FRP exist. Bousselham and Chaallal, (2004) reported that the effect of increasing
number of layers of FRP sheets decreases as the transverse steel reinforcement ratio increases. In addition, the
effect of longitudinal steel reinforcement may also affect the contribution of FRP in shear resistance. To further
investigate this, research studies must be carefully conducted to account for the effects of both transverse and
longitudinal steel reinforcement.
Scale effect and Geometry: Since most previous experimental studies were performed on small scale RC beams,
current design equations may not be accurate due to scale effects. In addition, failure modes of rectangular RC
beams and T-beams may be different as briefly discussed. Thus, it is important to conduct additional experimental
studies on full-scale T-sections.
FRP Strengthening Schemes and Control of Failure Mode: Numerous experimental works have been performed
to investigate the effectiveness of the different types of FRP strengthening schemes, and the corresponding
anchorage systems. As a result, it was found that the failure mode was dependent on the configurations of FRP
composite materials. Zhang and Hsu (2005) recently reported that the R factor (the ratio of the effective strain to the
ultimate strain of FRP) is more complex to define for FRP-rupture failure mode as compared to the case of FRP-
debonding. Thus, using the concept of effective strain without information regarding failure mode may lead to
unsafe design. Therefore, it may be necessary to control the failure mode of concrete structures with FRP by
providing appropriate detailing of strengthening schemes at the design level, for example; anchorage provisions,
minimum bond development length, minimum space of FRP strips.
Durability: Environmental factors also affect the failure mode since several research studies suggested that severe
environmental conditions can degrade the mechanical properties of FRP as well as the interfacial bond strength,
which may in turn affect the failure mode. Therefore, the long-term behavior of FRP-strengthened concrete
structures is an important issue, which needs be addressed in the new design guidelines.
Fatigue: Research studies on RC structural elements retrofitted with FRP under large-amplitude cyclic loading or
repeated low-amplitude fatigue loading are very few in comparison to monotonic loading. In addition, the research

607
work dedicated to fatigue performance is exclusively related to flexural strengthening. In contrast, limited studies
have been carried out on the fatigue performance and behavior of shear strengthening. Research work to address the
fatigue effect on shear behavior is therefore needed.
Prestressed vs. Non-Prestressed: A number of experimental research studies have been conducted with a focus on
the behavior of RC members; however, there is very limited experimental data regarding the PC members
strengthened in shear with FRP. Although Hutchinson and Rizkalla (1999) reported that the prediction by the
analytical model proposed by the authors was in good agreement with the test results of 7 prestressed concrete
beams strengthened with CFRP strips, the research regarding this parameter is still very limited and needs to be
conducted.
Anchorage Details: It is known that for both flexure and shear strengthening of girders using FRP sheets without a
carefully designed anchorage details at the ends of bonded FRP sheets, premature failure may occur due to
debonding or peeling of the FRP sheets at its ends. An enormous amount of research effort has focused on the
determination of the types, causes, and prevention of premature failure. To-date, several approaches have been tried
to avoid peeling failures such as the use of mechanical anchorages at the ends of the FRP sheet, the use of additional
wrapped sheets placed longitudinally at the top and bottom of girder web, and use of near surface mounted (NSM)
FRP rebars. Some of these methods are simple and some are labor intensive. The current available design equations
incorporate a limit on the effective FRP strains, which correspond to the maximum value at one of the following
situations: (1) the control crack opening, (2) shear failure due to FRP debonding, and (3) shear failure combined
with or followed by FRP rupture. The anchorage system should be reflected in newly developed design equation.

4. CONCLUDING REMARKS
FRP systems have shown great potential to be used for rehabilitation and retrofit of existing structures. They can be
used to provide increased ductility, and shear and flexural strength to structural elements such as columns,
beams/girders, slabs/decks and walls. The acceptance and utilization of these new strengthening techniques depend
on the availability of clear design guidelines, installation procedures and construction specifications. A significant
progress has been made in the last two decades and several design guidelines and documents have been developed
worldwide, but there are still issues and research needs to be addressed to bring the comfort level for practicing
engineers to use this FRP material as any other conventional material.

5. REFERENCES

ACI Committee 440. (2002) “ACI 440.2R-02-Guide for the design and construction of externally bonded FRP
systems for strengthening concrete structures, Farmington Hills, MI.
Bousselham, A. and Chaallal, O. (2004). “Shear strengthening reinforced concrete beams with fiber-reinforced
polymer: assessment of influencing parameters and required research.” ACI struc. J., 101(2), 219-227.
Concrete Society (2000). “Design guidance on strengthening concrete structures using fibre composite materials:
Technical Report 55, The Concrete Society, London.
CSA-S806-02 (2002). “Design and construction of building components with fibre-reinforced polymer.” Canadian
standards Association, Rexdale, Ontario.
fib-TG9.3 (2001). “Design and use of externally bonded fiber polymer reinforcement (FRP EBR) for reinforced
concrete structures.” Bulletin 14, July.
Hutchinson, R. L., and Rizkalla, SH. (1999). “Shear strengthening of AASHTO bridge girders using carbon fiber
reinforced polymer sheets.” Proc., 4th Int. Symp. on Fiber reinforced polymer reinforcement for reinforced concrete
structures. ACI publications SP-188, 945-56.
ISIS Canada (2001). “Design Manual 4: Strengthening reinforced concrete structures with externally-bonded fiber
reinforced polymers.” The Canadian Network of Centers of Excellence on Intelligent Sensing for Innovative
Structures, University of Manitoba, Winnipeg, Manitoba, Canada, Sep.
JSCE (2000) “Recommendations for upgrading of concrete structures with use of CFRP sheet.” Japanese Society of
Civil Engineers.
Triantafillou, T.C., and Antonopoulos, C.P. (2000). “Design of concrete flexural members strengthened in shear
with FRP.” J. Compos. Construct, 4(4), 198-205.
Zhang, Z. and Hsu, C.-T. (2005). “Shear strengthening of reinforced concrete beams using carbon-fiber-reinforced
polymer laminates.” J. of Comp. for Constr., 9(2), 158-169.

608
Third International Conference on FRP Composites in Civil Engineering (CICE 2006)
December 13-15 2006, Miami, Florida, USA

Strengthening of Concrete Beams in Shear with Mineral Based Composites

Laboratory Tests and Theory
Björn Täljsten
(Professor, Technical University of Denmark, Lyngby, Denmark
Luleå University of Technology, Luleå, Sweden)

Katalin Orosz
(PhD Student, Technical University of Denmark and Norut Teknologi AS, Norway)

Thomas Blanksvärd
(PhD Student, Luleå University of Technology, Sweden)

ABSTRACT

Today, there are many different repair and strengthening methods that might be used to upgrade a concrete structure.
One such method involves CFRP (Carbon Fibre Reinforced Polymer) bonding. This method has proven to be usable
for many different types of retrofitting applications. Even so, there are some disadvantages while using epoxy resins
as a bonding agent, i.e. diffusion closeness, thermal compatibility, working environment and the minimum
temperature of assemble. It is therefore of interest to replace the epoxy adhesive with a mineral based bonding agent,
e.g. polymer modified mortars with similar properties as the base concrete that also is more working environmental
friendly. A combination between the polymer modified mortar and fibre reinforced polymers (FRP) can be used for
repair and strengthening of civil structures. This paper presents a pilot study of RC beams strengthened in shear with
mineral based bonding agents and CFRP grids. The project is a collaboration project among Luleå University of
Technology, Norut Teknologi AS and Denmark Technical University and is also a part of the European funded
project “Sustainable Bridges”. The results so far show that comparable strengthening results as for epoxy bonded
systems can be achieved with MBC strengthening systems. The strengthening effect of the beams was 40 – 100 %
compared to the unstrengthened reference beam. The theoretical model describes the load carrying capacity fairly
well.

KEYWORDS
CFRP grids, Strengthening, Shear, Laboratory tests, MBC, Mineral Based Strengthening, Carbon Fibre

1. INTRODUCTION

Research with the use of short FRP fibres and cement based materials has been going on for some time now, see for
example (Kesner et. al., 2003). However, experience with the use of long FRP fibres is limited. Research studying
cement overlays with textiles of carbon fabrics embedded in cement based matrix to strengthen masonry walls has
been carried out by (Kolsch, 1998). The strengthening system prevents partial or complete collapse of masonry
walls in the critical out-of-plane direction during a seismic event. A study to improve the bond between carbon
fibres and cementitious matrices has been done by (Badanoiu, 2001), where dry fibre fabrics were used. It was found
that a pre-treatment with silica fume and high amounts of polymers improved the bond behaviour of carbon fibre to
the cement. However, it was also stressed that more research is needed in this field. A very interesting pioneering
work has been presented by (Wiberg, 2003). Large-scale tests of ordinary concrete beams strengthened with a
cementitious fibre composite were reported. The composite used was made of polymer-modified mortar and a
unidirectional sheet of continuous dry carbon fibres applied by hand. Both flexural and shear strengthening were
tested. From the tests it was concluded that the method works, and that considerable strengthening effects can be
achieved. In comparison with epoxy bonded carbon fibre sheets, the amount of carbon fibre needed to reach the

609
same strengthening effect for the cementitious strengthening system was more than double. The reason for this is
mainly due to problems with wetting the carbon fibre. This is also emphasised by (Badanoiu & Holmgren, 2003),
where it was found that the load capacity of the cementitious carbon fibre composite is influenced by the amount of
fibres in the tow. If the cementitious matrix can penetrate into the interior of the carbon fibre tow, a higher number
of filaments will be active during loading, and this will lead to an increase in load carrying capacity. To overcome
this problem use of CFRP grids and a cementitious matrix might be used. In this paper a brief presentation of the
tests carried out at Technical University of Denmark is presented.

2. TEST SET-UP
The test set up for the beam test is shown in Fig. 1. Five beams with the same geometry, concrete quality (average
compressive strength 38 MPa), and steel reinforcement (average tensile strength 517 MPa for the rebars and 530
MPa for the stirrups) were tested, four of them strengthened with CFRP grids, while the first beam served as
reference beam without CFRP strengthening. The properties of the grids are given in Fig. 1. The load was applied by
two cylinders standing on the floor therefore the beams were turned upside down. Each of the cylinders provided a
load of maximum 500 kN. The load was increased by approximately 10 kN/min/cylinder and the tests were load
controlled. The pressure was translated into voltage by the data-logger and the optical measuring equipment. The
measuring equipment comprised of transducers, strain gauges (on DTU2 and DTU5) and photogrammetric strain
measuring equipment to measure the strains and crack propagation on the strengthened surfaces of each test beam.
At failure, both data-logger and photo equipment were stopped, and the failure load was recorded. Numerous
pictures were taken with a digital camera to keep track on the crack development during loading. However, due to
limited space this is not presented in this paper. A great amount of longitudinal steel reinforcement ensured that the
beams would not fail in bending. All the beams had a steel reinforcement for shear only in one side. The CFRP grids
were applied using two types of mortars in two layers with a thickness of 10 mm on both sides of the beams. The
CFRP grid was placed between the two layers of the mortar. Before applying the first layer of mortar a primer was
applied to the sandblasted concrete surface to optimize the bond between concrete and mortar. In Fig. 1 the material
properties for the mortar and CFRP grid used are presented.

Steel column (vertical and

horizontal restraint) Steel rod
(vertical
Steel beam restraint)
(vertical restraint)

Steel rod
(horizontal
restraint)

Hydraulic load cylinder

MBC strengthening materials:

Mortar modulus of elasticity:
Cement I. E1=26.5 GPa, with short glass fibres Beam CFRP Mortar
Cement II. E2=18.0 GPa DTU1 Ref. - -
DTU2 Grid 3 Cement I
CFRP Grid properties (x: horisontal, y: vertical):
DTU3 Grid 2 Cement I
Grid I. 70x72 mm; 230 g/m2,
DTU4 Grid 1 Cement II
Ex = 341 GPa, εux = 1.3%, Ey = 390 GPa, εuy = 0.8 %
Grid 2. 24x25 mm; 150 g/m2, DTU5 Grid 1 Cement I
Ex = 281 GPa, εux = 1.3%, Ey = 380 GPa, εuy = 1.4 %
Grid 3. 42x43 mm; 390 g/m2,
Ex = 407 GPa, εux = 1.1%, Ey = 417 GPa, εuy = 1.1 %

Figure 1. Test set-up

610
3. EVALUATION

3.1. Theory
Studies on the shear strengthening of a RC beam by bonding FRP composites have been carried out since the early
1990’s (Chen & Teng, 2003). In the early studies the shear capacity of the FRP strengthening was based on a very
simplified stress distribution. In recent years a more advanced theory has been developed (Carolin & Täljsten,
2005). The theory provides a more detailed specification of the strain distribution in the bonded FRP. This is, of
course, a decisive factor in the study since the strains in the CFRP material are proportional to the stresses. A well
adopted approach for shear design is to use the truss or strut and tie model:

Vn = Vc + Vs + V f (1)
N N N
Concrete Steel FRP

The derivation of Vc and Vc refers often on national codes or standards. For the term Vf the same approach as for the
truss analogy may be used, with special consideration to the compatibility relationships for the studied FRP system.
In this paper a simplified theory is presented for the contribution of the CFRP grid. A CFRP grid usually consists of
a vertical and a horizontal tow. Both these tows contribute to the load carrying capacity, which in this paper is
presented as a result vector, see Fig. 2.

Figure 2. Relation between vertical and horizontal tow - left. The truss model - right

The direction and strength properties of the grid resultant are depending on the properties of each tow direction. This
is expressed in equation (2)-(5).

ε hor ⋅ Ehor ⋅ Ahor

N hor = (2)
shor
ε ⋅E ⋅A
N ver = ver ver ver (3)
sver
N res = N hor 2 + N ver 2 (4)
⎛ N hor ⎞
β = arc tan ⎜ ⎟ (5)
⎝ N ver ⎠

Based on these considerations some of the existing design models can be used to determine the contribution from
the CFRP grid, (Täljsten & Carolin, 2005). This equation is rewritten with the properties of the resulting tow, using
η = 0.4 as the modification factor due to the parabolic for of the strain contribution over the section:

2 ⋅η ⋅ N res ⋅ z cos(θ − β )
VFRP = ⋅ (6)
sres sin θ
sres = sver ⋅ cos β = shor ⋅ sin β (7)

611
3.2. Results from tests
The load deflection curves from the tests are shown in figure 3 together with the calculated and experimental values,
the calculation is carried out at a strain level of 10 ‰.

Beam Exp. Calc. Vexp Vcal

[kN] [kN]

DTU2 54 58 0.93

DTU3 54 42 1.29

DTU4 58 67 0.87

DTU5 54 45 1.20

Figure 3a. Load-Deflection curves Figure 3b. Experimental and calculated values

4. SUMMARY AND CONCLUSIONS

All five beams failed in shear. The strengthening effect was significant, the increase in load carrying capacity for the
strengthened beams was approximately 40-100 % compared to the reference beam. The largest increase was
achieved using the grid with the densest fibre area per cross-section and the mortar with short glass fibres. The
theoretical approach gave a reasonable estimation of the shear strengthening effect, however it was difficult to
exactly measure the strain in the tows and the scattering was large, therefore the theoretical evaluation is imperfect
and further laboratory research together with more detailed analytical and numerical analysis is needed to improve
the design model. Furthermore, a large test series is now ongoing at Luleå University of Technology in collaboration
with Technical University of Denmark and Norut Teknologi A/S in Norway. Here we are not only investigating the
structural behaviour of the strengthening system but also the effect of shrinkage and temperature.

ACKNOWLEDGEMENTS

The research presented in this paper has been funded by several organisations. Here Sto Scandinavia AB, the
Norwegian Research Council through the strategic institute program RECON and Skanska Sverige AB should be
acknowledged. Special thanks goes to the Master Students Morten Christiansen and Thorsten Jürgensen carrying out
their work at Technical University of Denmark.

REFERENCES

Badanoiu A., 2001, Improvement of the bond between carbon fibres and cementitious matrices, Technical report
2001:1, Concrete Structures, ISSN 1404-8450, Royal Institute of Technology, Department of Structural
Engineering, 100 44 Stockholm, Sweden, p 44.
Badanoiu A. and Holmgren J., Cementitious composites reinforced with continuous carbon fibres for strengthening
of concrete structures, Journal of Cement & Concrete Composites, vol. 25, 2003, pp 387-394.
Carolin A. and Täljsten B., Theoretical Study of Strengthening for Increase Shear Capacity, Journal of Composites
for Construction, November/December 2005, pp 497-506.
Chen, J. F. and Teng, J. G. (2003) Shear Capacity of Fiber-Reinforced Polymer-Strengthened Reinforced Concrete
Beams: Fiber Reinforced Polymer Rupture, Journal of Composites for Construction, ASCE / May 2003, pp 615-
625.
Kolsch H., 1998, Carbon fibre cement matrix (CFCM) overlay system for masonry strengthening, Journal for
composites for construction, Vol. 2, No. 2, May 1998, pp105-109.
Kesner K.E., Billington S.L and Douglas K.S., 2003, Cyclic Response of Highly Ductile Fibre-reinforced Cement-
Based Composites, ACI Materials and Journals - Technical Paper, September -October 2003, pp 381-390.
Wiberg A., 2003, Strengthening of concrete beams using cementitious carbon fibre composites, Doctoral Thesis,
ISSN 1103-4270, Royal Institute of Technology, Structural Engineering, 100 44 Stockholm, Sweden, p 140.

612
Third International Conference on FRP Composites in Civil Engineering (CICE 2006)
December 13-15 2006, Miami, Florida, USA

ANALYTICAL PREDICTION OF DEBONDING FAILURES IN RC

BEAMS STRENGTHENED IN SHEAR WITH NSM FRP
REINFORCEMENT
Andrea Rizzo
(PhD Student, Department of Innovation Engineering, University of Lecce, Lecce, Italy)

Laura De Lorenzis
(Assistant Professor, Department of Innovation Engineering, University of Lecce, Lecce, Italy)

ABSTRACT
The shear capacity of reinforced concrete (RC) members can be successfully increased using near-surface mounted
(NSM) fiber-reinforced polymer (FRP) reinforcement. Tests on NSM-strengthened beams have shown that failure is
controlled by diagonal tension associated to debonding between the NSM reinforcement and the concrete substrate.
In absence of steel stirrups and/or when the spacing of the NSM reinforcement is large, debonding involves
separately each of the bars crossed by the critical shear crack (type-I failure). The presence of steel stirrups,
combined with a relatively small spacing of the reinforcement, may originate a debonding failure mechanism
involving the lateral concrete covers of the steel stirrups (type-II failure). Thus, an analytical model able to
encompass both failure modes must be developed. This paper extends a previous simplified model to predict the
FRP contribution to the shear capacity when type-I failure occurs. The model, suitable for immediate design use,
assumes a complete redistribution of the bond stresses along the failure interface at ultimate. Experimental results
from previous test programs are compared to the model predictions.

KEYWORDS
Debonding failures, FRP, near-surface mounted reinforcement, RC beams, shear strengthening.

1. INTRODUCTION
An emerging method for shear strengthening of reinforced concrete (RC) members is the use of near-surface
mounted (NSM) fiber-reinforced polymer (FRP) reinforcement, usually in the form of round bars or strips.
Tests conducted thus far on NSM-shear-strengthened beams have shown that failure is controlled by diagonal
tension associated to debonding between the NSM reinforcement and the concrete substrate. In absence of steel
stirrups and/or when the spacing of the NSM reinforcement is large, debonding involves separately each of the bars
crossed by the critical shear crack (type-I failure). Recent experiments have also shown that the presence of steel
stirrups, combined with a relatively small spacing of the reinforcement, may originate a debonding failure
mechanism in which the lateral concrete covers of the steel stirrups detach from the core of the beam with the NSM
reinforcement still embedded (type-II failure) (De Lorenzis and Rizzo 2006).
Few research programs on NSM FRP shear strengthening of RC beams are available in literature, as summarized in
Table 1. Table 2 reports the main tests results. For all the cases reported, type-I failure occurred. Type-II failure is
not treated in this paper and needs appropriate investigation.
At present, different approaches to compute the capacity of shear-strengthened beams are available in the literature,
most of which are based on the generalization of the truss model. A model for beams strengthened in shear with
NSM reinforcement, based on the model by De Lorenzis and Nanni (2001), is here formulated and applied to predict
the test results reported in literature, when type-I failure occurs.

613
2. ANALYTICAL MODEL
The shear capacity (Vtot) of RC beams strengthened with NSM FRP systems can be calculated as the sum of the
contributions of concrete (Vc), steel stirrups (Vs) and FRP system (VFRP). The analysis developed herein focuses on
the prediction of the FRP contribution.

2.1 Shear contribution for debonding failure mode

As follows, the simplified approach proposed in De Lorenzis and Nanni (2001) is generalized for any value of the
angles formed by the shear crack and by the FRP strengthening with the horizontal direction. The formulation is
suitable to account for any possible debonding failure mode, provided that it involves each NSM bar separately. A
uniform shear stress τf at the failure interface is assumed in all the FRP bars intersected by the shear crack at
ultimate, thus the FRP shear contribution VFRP can be calculated multiplying τf by the total lateral surface of the
minimum embedment lengths of all the bars crossed by the crack. This results in Eq (1)

VFRP = 2lemb pτ f sin α (1)

where lemb is the sum of the minimum embedment lengths of the bars intersected by the shear crack, p is the
perimeter along which the bond stress acts (e.g. in the case of round bars, p = πφ where φ is the diameter of the bar),
α is the angle of the FRP bars to the horizontal axis z (Figure 1), factor 2 accounts for the bars on both sides of the
beam. Eq. (1) is valid provided that each embedment length lemb,i (i.e. the minimum length between ls,i – cs/sinα and
li,i – ci/sinα as shown in Figure 1) is smaller than the development length of the bars ldev, i.e. of the value of bond
length sufficient to cause failure in tension of the FRP bar prior to its debonding.. The next formulation assumes
lemb,i < ldev for each bar. For any given geometry of the beam, lemb depends on α, θ (θ being the angle between the z-
axis and the critical shear crack, see Figure 1) and the relative position between the shear crack and the
strengthening system. For design purposes, the value of VFRP corresponding to the most unfavourable crack position
must be calculated, i.e. the minimum of lemb and VFRP (lemb min and VFRP min) must be found.
It is possible to demonstrate that lemb min is given by Eq. (2):

⎪⎧ s f ⎫⎪ sin θ
( )
2
lemb min = ⎨n* [ cot θ + cot α ] − n* ⎬ hnet (2)
⎩⎪ hnet ⎭⎪ sin (α + θ )

where hnet is the reduced height of the beam (Eq. (3)) obtained by subtracting the concrete cover thicknesses cs and ci
(see Figure 1), sf is the spacing of the bars measured along the longitudinal axis of the beam, and n* is the number of
the bars for which the minimum embedment length coincides with the one located above the crack, given by Eq. (4)
where ⎜x⎟ indicates the integer part of x; r* is provided by Eq. (5). Eq. (6) gives the minimum FRP shear
contribution VFRP min.

hnet = h − ci − cs (3)
*
r
n* = r * − (4)
2
hnet
r * = [ cot θ + cot α ] (5)
sf
VFRP min = 2lemb min pτ f sin α . (6)

2.2 Determination of the shear stress τf

Given the local bond -slip relationship, generally it is not conservative to use the maximum value of the bond stress
as τf since the local bond stress-slip relationship is not sufficiently ductile. On the other hand, τf should be chosen
properly in order to avoid an excessive crack opening with consequent loss of aggregate interlock. This control can
be indirectly done by limiting the maximum strain in the bars as proposed in De Lorenzis and Nanni (2001). As
follows, τf is determined based on the limitation of the ultimate slip su. This method allows to control the crack
opening and to take the local bond -slip relationship into account.

614
When a bar is intersected by a shear crack, the crack width can be calculated if the slip at both loaded ends of the bar
is known. At ultimate, the crack width is approximately given by the slip of the side of the bar that collapses first. In
order to translate the limitation to this slip in terms of bond stress along the failure interface at ultimate, a reduced
constant tangential stress, τf, red, must be found, and this value can be reasonably obtained starting from the local
bond-slip relationship and maintaining constant the fracture energy Gf (see Eq. (7)).
The obtained value can be directly used in Eq. (6) instead of τf to find VFRP min. To sum up, the computation of
VFRP min with the proposed model consists in the following steps: computation of hnet from Eq. (3); computation of r*
and n* from Eq.s (4) and (5); computation of lemb min from Eq. (2); computation of τf, red from Eq. (7); computation of
VFRP min from Eq. (6).

su
Gf 1
τ f , red =
su
=
su ∫ τ ( s ) ds
0
(7)

ls,i
Shear Crack FRP Reinforcement
li,i cs
α
d h
ci
sf
Bar i Longitudinal
θ
Steel Rebars

Figure 1: Calculation of the Embedment Lengths of the FRP Bars

3. COMPARISON BETWEEN THEORETICAL AND EXPERIMENTAL RESULTS

In this section, the model is applied to the beams tested by De Lorenzis and Nanni (2001), Barros and Dias (2006),
Dias and Barros (2006) and Barros et al. (2006). Table 1 summarizes the main parameters. Figure 2a shows the
bond-slip relationships reported in De Lorenzis (2002) and in Sena Cruz (2004) (curve 1 and 2-3, respectively);
Figure 2b plots the reduced bond stress τf, red versus the maximum slip su, in the range of slips from 0 mm to 1.0 mm.
There is a value of su, su*, for which τf, red is maximum. This can be physically explained as follows. For slips
smaller than su*, the crack opening is limited and the bars are not stressed for their entire length; thus, a smaller
value τf, red must be used. Conversely, for slips higher than su*, part of the length of the bars, corresponding to the
loaded ends, can result debonded; thus, τf, red must be reduced to take this into account.
Table 1 and Figure 3 compare the experimental results with predictions of the model. τf, red was found using the
bond-slip relationship proposed by each author and assuming su = 0.2 mm. Figure 3a shows the results for the beams
of Table 1. With few exceptions, predictions are conservative. Figure 3a also shows the results obtained using for τf,
the local bond strength, which is evidently unconservative. Figure 3b analyses the influence on results of su and hnet
for the beam tests reported by De Lorenzis and Nanni (2001). It can be noted that the choice of su does not strongly
affect the results as hnet does.

4. CONCLUSIONS
From results of the reported investigation, the following conclusions can be drawn:
- the proposed model uses simplifying assumptions that make it very easy to use for design purposes unlike more
accurate but more onerous models accounting for bond-slip and evolution of bond stresses during opening of the
critical shear crack;
- the proposed model with a properly reduced value of design bond strength (for example, maintaining constant the
fracture energy and limiting the ultimate slip of the real bond-slip model in order to control the shear crack width)
gives in most cases conservative predictions on the FRP shear contribution.
Obviously, further experimental investigations are necessary to validate the proposed formulation and extend it to
the cases for which type-II failure occurs.

615
 (a) (b)
Figure 2: (a) Local τ-s Relationships Found by De Lorenzis (2002; curve 1), and Sena Cruz (2004; curve 2-3)
for Joints Having the Same Characteristics of Those in the Shear-Strengthened Beams; (b)
Reduced Bond Stress τf, red (eq. 7) versus Maximum Slip su.

(a) (b)
Figure 3: (a) Experimental and Theoretical Results; (b) Experimental and Theoretical Results for the Beam
Tests Reported by De Lorenzis and Nanni (2001) Using Different Values of su and hnet

5. REFERENCES
Barros, J. A. O., and Dias, S., (2006), “Near surface mounted CFRP laminates for shear strengthening of concrete
beams”, Cement and Concrete Composites, 28 (3), pp. 272-292.
Barros, J. A. O., Ferreira, D. R. S. .M., Fortes, A. S. and Dias, S. J. E., (2006) “Assessing the effectiveness of
embedding CFRP laminates in the near surface for structural strengthening”, Construction and Building Materials,
20, pp. 478-491.
Sena Cruz, J. M., (2004). “Strengthening of concrete structures with near-surface mounted CFRP laminate strips”,
PhD Thesis, Escola de Engenharia, Universidade do Minho, Portugal.
De Lorenzis, L., and Nanni, A., (2001), “Shear strengthening of reinforced concrete beams with NSM fiber-
reinforced polymer rods”, ACI Structural Journal, 98(1), pp. 60-68.
De Lorenzis, L., (2002), “Strengthening of RC structures with near-surface mounted FRP rods”, PhD Thesis,
University of Lecce, Italy.
De Lorenzis, L., and Rizzo, A., (2006). “Behavior and capacity of RC beams strengthened in shear with NSM FRP
reinforcement”. Proceedings of Second fib Congress, Naples, Italy, CD-ROM
Dias, S. J. E., and Barros, J. A. O., (2006), “NSM laminates for the shear strengthening of T Section RC Beams”,
private communication.
Sena Cruz, J. M., and Barros, J.A.O. (2004), “Modeling of bond between near-surface mounted CFRP laminate
strips and concrete”, Computers and Structures, 82, pp. 1513-1521.

616
 Table 1: RC Beams Strengthened in Shear with NSM Systems

Beam Steel Stirrups NSM System Angle, α Spacing, sf Reduced Height, hnet
Diameter/Spacing
(mm)/(mm) (degrees) (mm) (mm)
Source: De Lorenzis and Nanni (2001)
Section: T (web width/height = 152/305 mm; flange width/height = 381/102 mm; ci/cs = 50.8/0 mm)
NSM CFRP Shear Strengthening System: Round CFRP Bar (9.525 mm diameter; Ef = 104.8 GPa) + Epoxy
Local bond parameters: τf = 10.96 MPa; sm = 0.191 mm (from Table 4.2 in De Lorenzis (2002))
BV None None - - -
B90-7 None CFRP bars 90 177.8 254.0
B90-5 None CFRP bars 90 127.0 254.0
B90-5A None CFRP bars anchored in the flange 90 127.0 355.6
B45-7 None CFRP bars 45 177.8 254.0
B45-5 None CFRP bars 45 127.0 254.0
BSV 9.525/355.6 None - - -
BS90-7A 9.525/355.6 CFRP bars anchored in the flange 90 177.8 355.6
Source: Barros et al. (2006)
Section: Rectangular (width/height = 150/150 mm; ci/cs = 31.0/29.0 mm)
NSM CFRP Shear Strengthening System: CFRP Strip (width/thickness = 9.59/1.45 mm; Ef = 158.0 GPa) + Epoxy
Local bond parameters: τf = 19.81 MPa; sm = 0.25 mm (from Table 1 in Cruz and Barros (2004))
VB10 None None - - -
VBCV-10 None CFRP strips 90 100.0 119.0
VBCI-15 None CFRP strips 45 150.0 119.0
Source: Dias and Barros (2006)
Section: T (web width/height = 180/300 mm; flange width/height = 450/100 mm; ci/cs = 44.0/34.0 mm)
NSM CFRP Shear Strengthening System: CFRP Strip (width/thickness = 9.59/1.45 mm; Ef = 166.6 GPa) + Epoxy
Local bond parameters: τf = 15.80 MPa; sm = 0.39 mm (from Table 3.3 in Sena Cruz (2004))
2S-R 6/300 None - - -
2S-3LV 6/300 CFRP strips 90 267.0 256.0
2S-5LV 6/300 CFRP strips 90 160.0 256.0
2S-8LV 6/300 CFRP strips 90 100.0 256.0
2S-3LI45 6/300 CFRP strips 45 367.0 256.0
2S-5LI45 6/300 CFRP strips 45 220.0 256.0
2S-8LI45 6/300 CFRP strips 45 138.0 256.0
2S-3LI60 6/300 CFRP strips 60 325.0 256.0
2S-5LI60 6/300 CFRP strips 60 195.0 256.0
2S-7LI60 6/300 CFRP strips 60 139.0 256.0
Source: Barros and Dias (2006)
Section: Rectangular A-Beam (width/height = 150/300 mm; ci/cs = (15.0+0.5φlongitudinal steel rebar (10 or 12 mm))/18.0 mm)
Section: Rectangular B-Beam (width/height = 150/150 mm; ci/cs = (15.0+0.5φlongitudinal steel rebar (10 or 12 mm))/18.0 mm)
NSM CFRP Shear Strengthening System: CFRP Strip (width/thickness = 9.59/1.45 mm; Ef = 166.6 GPa) + Epoxy
Local bond parameters: τf = 15.80 MPa; sm = 0.39 mm (from Table 3.3 in Sena Cruz (2004))
A10_C None None - - -
A10_VL None CFRP strips 90 200.0 262.0
A10_IL None CFRP strips 45 300.0 262.0
A12_C None None - - -
A12_VL None CFRP strips 90 100.0 261.0
A12_IL None CFRP strips 45 150.0 261.0
B10_C None None - - -
B10_VL None CFRP strips 90 100.0 112.0
B10_IL None CFRP strips 45 150.0 112.0
B12_C None None - - -
B12_VL None CFRP strips 90 50.0 111.0
B12_IL None CFRP strips 45 75.0 111.0

617
 Table 2: Experimental and Theoretical Results of the Beams of Table 1

Beam Experimental Experimental Analytical ∆min3

Ultimate Shear FRP Contribution FRP Contribution
Vexp1 VFRP exp1 VFRP d2
(kN) (kN) (kN) (%)
BV 90.29 - - -
B90-7 115.20 24.91 10.45 138.4
B90-5 127.66 37.36 31.35 19.2
B90-5A 185.70 95.41 73.15 30.4
B45-7 165.47 75.17 47.02 59.9
B45-5 177.92 87.63 62.70 39.8
BSV 153.23 - - -
BS90-7A 206.83 53.60 52.25 2.6

VB10 37.01 - - -
VBCV-10 65.61 28.60 0.00 -
VBCI-15 60.22 23.21 10.35 124.3

2S-R 246.00 - - -
2S-3LV 189.60 0.60 0.00 -
2S-5LV 214.20 25.20 23.88 5.5
2S-8LV 237.60 48.60 45.26 7.4
2S-3LI45 196.80 7.80 15.50 -49.7
2S-5LI45 230.40 41.40 41.70 -0.7
2S-8LI45 229.20 40.20 63.44 -36.6
2S-3LI60 224.40 35.40 7.47 373.9
2S-5LI60 235.20 46.20 36.84 25.4
2S-7LI60 243.60 54.60 49.50 10.3

A10_C 50.20 - - -
A10_VL 79.32 29.12 35.64 -18.3
A10_IL 78.95 28.75 53.46 -46.2
A12_C 58.25 - - -
A12_VL 117.56 59.31 71.28 -16.8
A12_IL 131.19 72.94 106.92 -31.8
B10_C 37.01 - - -
B10_VL 65.61 28.60 17.82 60.5
B10_IL 60.22 23.21 26.73 -13.2
B12_C 37.85 - - -
B12_VL 69.60 31.75 35.64 -10.9
B12_IL 74.25 36.40 53.46 -31.9
1
Experimental data.
2
su = 0.2 mm.
3
VFRP exp − VFRP min
∆ min = .
VFRP min

618
Part XXII. Strengthening
Third International Conference on FRP Composites in Civil Engineering (CICE 2006)
December 13-15 2006, Miami, Florida, USA

FLEXURAL BEHAVIOUR OF REINFORCED CONCRETE BEAMS

STRENGTHENED WITH NEAR SURFACE MOUNTED CFRP STRIPS
Renata Kotynia
(Assistant Professor, Technical University of Lodz, Lodz, Poland)

ABSTRACT
The paper presents test results of full-size simply supported reinforced concrete beams with rectangular cross
section, strengthened for flexure with NSM CFRP strips. The specimens with different steel and CFRP
reinforcement ratio were tested under four point monotonic loading. The aim of the test was to obtain the same
flexural capacity for all tested beams. The structural performance and modes of failure of the tested beams are
presented and discussed in this paper. Full composite action between the NSM CFRP strips and concrete was
achieved. Results of the test indicated that the lower was steel reinforcement ratio and higher was CFRP ratio the
lower was stiffness and ultimate load of strengthened beams. It was due to lower the axial stiffness EfAf of NSMR
CFRP strips than the stiffness of the steel reinforcement. If the elasticity modulus of NSMR strip is similar to the
steel, the stiffness and load capacity of beams will be higher.

KEYWORDS
NSM, CFRP Strips, Flexural Strengthening, Debonding

1. INTRODUCTION
The research carried out up to now on RC members strengthened in flexure has indicated that externally bonded
reinforcement (EBR) technique cannot utilize the full strength of the fiber reinforced polymer (FRP) reinforcement
(Kotynia, 1999). Brittle failure due to premature FRP debonding and concrete cover sepapration has been the most
dominante mode of failure (Teng et al., 2002). In order to improve the efficiency of this technique additional
anchorage systems with U-jacket FRP sheets have been proposed (Kotynia and Kaminska, 2003). Moreover EBR
FRP technique is susceptible to damage from collision, high temeperature, fire and ultraviolet rays. To overcome
these drawbacks, near surface mounted (NSM) strengthening technique has been proposed. Due to better anchorage
of NSM FRP reinforcement bonded into pre-cut grooves opened in the concrete cover of the RC element, this
technique is significantly more efficient than EBR system (Täljsten, 2005). Recent tests have pointed out several
benefits of NSM technique such as: increase in the load carring capacity of concrete structures in bending and shear,
easy and cost effective (Barros and Fortes, 2005). The bond characteristic between NSM FRP reinforcement and
concrete has been widely analysed (De Lorenzis and Nanni, 2001). Using the same axial stiffness of FRP to
strengthen reinforced concrete beams, higher ultimate load has been achieved by the beams strengthened with the
NSM CFRP strips than by the beams strengthened with EBR (El-Hacha and Rizkalla, 2004).

2. EXPERIMENTAL PROGRAM
Test carried out on RC beams strengthened with the NSM CFRP strips indicated that the average strain of the strip
during debonding was 0.011 (Kotynia, 2005). This value has been assumed as a limit strain of the strip in the
ultimate state. Based on this assumption, the test program of four beams has been proposed. The aim of the test was
to obtain the same flexural capacity for all tested beams. Four rectangular concrete beams simply supported over
4200 mm clear span were tested in four-point flexure with a shear span of 1400 mm and a constant moment region
of 1400 mm. Four different tension steel reinforcement ratios and three CFRP reinforcement ratios
(ρs=As/bh, ρf=Af/bh, where As and Af are the total tension steel and CFRP reinforcement respectively, b and h are
the width and height of the beam respectively) were used (Table 1). The beam P0 with the highest steel ratio

619
(ρs=1.57%) was unstrengthened in flexure and three other beams were strengthened with one, two or three CFRP
strips 2.4 x 15 x 4050 mm. Shear reinforcement consisted of the steel stirrups made of the deformed steel bars of
nominal diameter 6 mm spaced at 100 mm and 200 mm. To install the NSM CFRP strips, longitudinal grooves (4
mm x 19 mm) were cut into the concrete cover on the tension side of the beams with a diamond blade. Details of the
beams are shown in Figure 1. Experimental strength characteristic of steel and CFRP is shown in Table 2. Strains at
the level of the longitudinal tension and compression steel were measured by the LVDT gauges, attached to the
concrete side. Deflections were measured in five points of the beam’s span. All measurements were recorded at
every load level.

Table 1: Details of tested beams

Steel tension reinforcement CFRP ratio Concrete

Beam
Bars ρs, (-) Strips ρf, (-) fc,cube (MPa) fc (MPa)
P0 2#20 + 1#10 1.57 - - 48.0 41.3
P1 2#18 1.13 1 strip 0.08 48.7 41.5
P2 2#12 + 1#6 0.57 2 strips 0.16 43.0 37.7
P3 2#12 0.50 3 strips 0.24 49.2 43.5

Figure 1: Details of tested beams, steel reinforcement and strengthening modes

Table 2: Properties of steel and CFRP reinforcement Table 3: Composition of concrete mix

Type A, (mm) fy, (MPa) E, (GPa) fu, (MPa) Components Amount (kg/m3)
Steel Cement 32,5R 350
#10 80.9 425 210 655 Fly ash 60
#12 114.1 542 220 630 Sand 0/2 711
#18 252.9 563 210 669 Gravel 2/8 573
#20 317.5 541 205 638 Gravel 8/16 451
#6 29.5 437 207 501 Water 185
CFRP Total sum 2330
XS1.52
29.5 - 163 2250
4
E – Elastic modulus; fy – Yielding strength;
fu – Ultimate tensile strength

Composition of concrete mix is given in Table 3. Compressive strength of concrete determined on the cubic fc,cube
(150 mm x 150 mm) and cylindrical fc specimens is presented in Table 1.

620
3. TEST RESULTS
Two beams with the high steel reinforcement ratio (P0 and P1) failed due to concrete crushing (CC) in the
compression zone (marked with a circle in Figure 2). Debonding of the NSM CFRP strips (SD) with the detached
concrete cover below the steel was observed in the other two beams P3 and P4 (Figure 2). Full composite action
between NSM CFRP strips and concrete was achieved throughout the tests since no slip was observed. The strip
debonding was preceded by a typical flexure vertical crack of the beam and the internal concrete cracking followed
by the formation of inclined and longitudinal cracks in the concrete surrounding the groove. The NSM CFRP strips
debonded from the beam with the detached concrete cover below the steel. The ultimate loads, the average
compression (εc) and CFRP strip’s strains (εfdb) obtained during the beam’s failure are summarized in Table 4.

P1

P1

P2

P3

Figure 2. Failure of tested beams due to concrete crushing (P0, P1) and the strip debonding (P2, P3)

Table 4: Details of tested beams

Strain in
Ultimate load Failure CFRP strain Compression Ultimate CFRP CFRP utilization
Beam tension steel
2Fu, (kN) mode εfdb, (‰]) strain εc, (‰) strain εfu, (‰) εfdb / εfu
level εt, (‰)
P0 136.0 CC 6.38 - 1.93 - -
P1 123.4 CC 11.18 12.72 1.36 0.92
P2 107.0 SD 9.94 11.2 1.20 13.8 0.81
P3 111.2 SD 8.05 9.10 1.58 0.66

The test results indicated that the lower was the steel reinforcement the lower was the ultimate load. The reason of
this decrease was lower modulus of elasticity of the CFRP strips than the modulus of the steel. The CFRP and steel
elasticity modulus ratio was Ef/Es = 0.7. Hence, the strengthening ratio should be analyzed in respect of total tension
steel cross section area (As) and reduced FRP cross section area of the FRP strips (Af × Ef/Es). To increase the
strengthening ratio, NSM CFRP strips with higher elasticity modulus should be used. In case of the beam P0 a real
stress-strain response of the steel bars (#20) turned out different than that assumed in the test program. Hence, the
ultimate load for this beam was different than the remaining three beams. Tension strain measurements (εt) indicated
that the higher was the CFRP ratio the lower was the strain of the debonded strip. It confirms the thesis about the
effect of the axial stiffness of the NSMR FRP strip (EfAf) on the limit strain during its debonding (εflim). Hence, the
strain utilization of the NSM CFRP (εfdb / εfu) increased from 66% to 92%. This is a satisfactory result in comparison
with the FRP utilization in the EB FRP strengthened beams that was only 35% (Kotynia, 1999).

621
Comparison of the load-deflection response for all tested beams is shown in Figure 4. The deflection measurements
confirm a decrease in the stiffness of the beams caused by the lower elasticity modulus of the NSM CFRP than the
steel modulus. Beams with the lower steel ratio indicated higher deflections in failure.

140
P0
120 P1
P3
100 P2
Load 2F [kN]

80 Beam Deflection vmax, (mm)

P0 49,71
60
P1 52,92
40 P2 89,53
20 P3 76,00
0
0 20 40 60 80 100
Deflection [mm]

Figure 4. Load-deflection responses of tested beams

4. CONCLUSIONS
The test results indicated different ultimate loads for all tested beams. The reason of this effect was due to the low
elasticity modulus of the NSM CFRP strips that caused decrease in the stiffness and ultimate loads of the
strengthened beams. The higher was NSM CFRP ratio the lower was limit strain during its debonding and the strain
utilization of the FRP. In order to increase the strengthening effect the NSM reinforbcement with higher elasticity
moduluds should be used.

5. REFERENCES
Barros, J.A.O. and Fortes, A.S. (2005). “Flexural strengthening of concrete beams with CFRP laminates bonded into
slits”. Journal of Cement & Concrete Composites, Vol. 27, pp. 471-480.
De Lorenzis, L. and Nanni, A. (2001). “A Bond Between Near-Surface Mounted FRP Rods and Concrete in
Structural Strengthening”. ACI Structural Journal, Vol. 99, No. 2, pp. 123-132.
El-Hacha, R. and Rizkalla, S. (2004). “Near-Surface-Mounted Fiber-Reinforced Polymer Reinforcements for
Flexural Strengthening of Concrete Structures”. ACI Structural Journal, Vol. 101, No. 5, pp. 717-726.
Kotynia, R. (1999). “Ductility and Load Capacity of Reinforced Concrete Members Strengthened with CFRP
Strips”, Ph.D. thesis, Technical University of Lodz, Lodz, Poland.
Kotynia, R. and Kaminska, M.E. (2003). Ductility and failure mode of RC beams strengthened for flexure with
CFRP, Report No. 13, Technical University of Lodz, Lodz, Poland.
Kotynia, R. (2005). Effectiveness of Near Surface Mounted CFRP Reinforcement for Strengthening of Reinforced
Concrete Structures Proceedings of the International Conference COBRAE 2005, Bridge Engineering with Polymer
Composites, EMPA, Dubendorf.
Täljsten, B. (2005). Flexural Strengthening of Concrete Beams with Prestressed CFRP Near Surface Mounted
Reinfprcement (NSMR), Proceedings of Third International Conference, Editors: P. Hamelin, D. Biguard, E.
Ferrier, E. Jacquelin, Lyon, pp. 3-10.
Teng, J.G., Chen, J.F., Smith, S.T. and Lam, L. (2002). FRP strengthened RC Structures, Wiley, Chichester, U.K.

622
Third International Conference on FRP composites in Civil Engineering (CICE 2006)
December 13-15 2006, Miami, Florida, USA

STRENGTHENING CHOICES FOR THE REPAIR & RETROFIT OF

CONCRETE BRIDGE STRUCTURES WITH FRP
J. M. Eyre
(Graduate Structural Engineer, Costain Ltd, Maidenhead, UK)

T. J. Ibell
(Professor of Civil & Architectural Engineering, University of Bath, Bath, UK)

A. Nanni
(Jones Professor of Civil Engineering, University of Missouri-Rolla, Rolla, MO, USA)

ABSTRACT
At present there exist numerous, approved documents and guidelines across Europe, the US, Japan and Canada,
giving detailed information on the various applications available that provide for the strengthening of concrete
structures with Fibre Reinforced Polymers (FRP). Such guidelines offer a detailed overview of the principles
involved in the design of strengthening for deficient beams, columns and slabs, but do not offer codified approaches
for individual materials and techniques, or the necessary specialist bridge related criteria. The information and
understanding gained from this study is intended for use in directing the generic approach to future FRP design
strategies, through the unification of material, system & installation procedure selection and inspection &
monitoring regimes.

Software has been developed by the authors to provide a rule-based, logic-programmed ‘FRP Strengthening System
– Selection Toolkit’ defining the extent of required superstructure strengthening following diagnosis and
recommendation for pre-installation treatment of the defects and degradation established from analysis of
assessment and inspection results. The design process protocol incorporated in the software, consciously targets the
concerns of largely inexperienced engineers in this field, particularly those of graduate level training, whilst also
attempting to meet the knowledge demands of the Project Engineer or Department of Transportation Investigating
Official, responsible for compilation of a preliminary strengthening scheme proposal, for later submission to tender.

KEYWORDS
Fibre Reinforced Polymer, Strengthening, Expert-system, Selection-criteria, Diagnosis.

1. INTRODUCTION
The failings of many of the guidelines, from the extensive range of those produced by the major European (Concrete
Society Technical Report 55, Fib TG9.3 Committee 2001), US (ACI 440.2R-02) and Canadian Research
committees/ national bodies, in providing truly useable and practical design assistance is in the provision of
information relating to practical cases of ‘whole’ bridge or building strengthening/repair or retrofit. Particularly
lacking are guidelines for the undertaking of adequate specification detailing.

Understandably, the industry still remains highly sceptical of new technologies, which further limits the take-up of
FRP solutions. Most significantly because degradation cannot be fully anticipated and the associated risk is borne by
clients, they are becoming increasingly aware of the durability and maintenance implications of the solutions they
adopt. Experience with steel plate bonding and more than 20 years of use of carbon fibre in highly stressed
applications for the aviation and automotive industries, has provided accepted criteria for comparable FRP
techniques. Adhesives, for instance, have been shown to be equally effective for use with FRP in construction as in
the above areas. Ultimately, the limitation of repair lies in the fatigue life of the embedded reinforcement and thus
the success of externally bonded FRP relies on its ability to increase the stress range of the reinforcement.

623
2. AIMS & OBJECTIVES
If FRP techniques are to receive continued use and development, reliable performance of demonstration schemes
must be assured and maintained. The foremost objective of this research work was therefore, to develop a design
process protocol for the selection of suitable strengthening systems, for the retrofit of structurally defective and/or
deteriorating concrete bridges. This objective has been reached through the production of a series of decision-tree’s/
flow diagrams which are presented as design aids, along with a rule-based, logic-programmed, Expert-system,
(referred to as the ‘Toolkit’ throughout this report where appropriate) titled; ‘FRP Strengthening System –
Selection Toolkit’, covering the following phases of selection and repair:
• Deterioration and degradation processes – Damage and crack analysis for selection of associated repair or
protective treatment.
• Strengthening option selection and elimination – Balanced assessment of the merits or limitations of FRP over
more traditional counterpart solutions, such as bonded steel plate, increased section size or additional prestress.
• Structural re-calculation – Analysis of the requirements for enhanced capacity – in shear, flexure, axial
strengthening and/or ductility.
• FRP material and system technique choice – Suitability of material type. Namely carbon, aramid or glass, with
epoxy, vinylester or polyester laminating resins/adhesives. Appropriate methods of application, principally wet-
lay-up sheet, pre-cured plate or Near Surface Mounted reinforcement.
• Structural Design – Rule-based sequencing for shear, flexural, axial-strengthening or ductility retrofit.
Extension of above phases, to include lengths, laps and splices, layering and the required geometry on the
surface of the structure. Criteria for the selection of pre-stressed and mechanically anchored FRP are included to
allow exploration for their suitability as appropriate solutions for less conventional defects. Flexibility is
however in-built for future extension.

Cost is arguably one of the most influential factors when assessing the advantages of alternative methods and
schemes. Detailed costing however is difficult to provide impartially and can become quickly outdated in a highly
changeable environment, where new technology is continually emerging. For the benefit of preliminary design
clarification, final unit costs are provided, developed from an accumulation of typical material-supply, labour and
inclusive construction costs, for each of the phases of development highlighted above. Design costs cannot be
accurately predicted under the scope of this work and thus are not included in the final cost report.

Work to tailor the expert system to cater directly for the needs of the Project Engineer, was also carried out after
close consultation with MoDOT (Missouri Department of Transportation) area engineers, responsible for the Rolla
precincts. Efforts have been made to manipulate criteria for selection throughout each stage, with refinements made
to the Toolkits’ user-form ordering, questioning and where appropriate, linking procedures and recommendations
with MoDOT documentation for standard detailing, ‘Special Job Provisions’ and contractual requirements.

3. DEGRADATION DIAGNOSIS
The design of strengthening systems for concrete structures is largely achieved through the application of externally
bonded reinforcement (abbreviated to EBR where appropriate). Such design is usually carried out as an iterative
process of generation, evaluation and modification of trial and error solutions. In such early stages of development
of composite technology, there exists little comprehensive guidance for use by inexperienced engineers for
preliminary design of systems with FRP. The need to provide interactive, step-by-step guidance to this effect was
considered from the project outset.

The most important use of the ‘FRP Strengthening System – Selection Toolkit’ is to initiate appropriate thinking
and direct/control preliminary design rationale and the generic approach. Overall, the suitability of a candidate
bridge superstructure for flexural, shear or axial strengthening, is gauged, having provided an evaluation of the
extent of the required strengthening and the sequential impacts of chosen options relating to construction,
manpower, material and cost resources. The software does not attempt to prescribe a fixed solution to every
eventuality at this stage, but does provide important and fundamental design considerations and practicality issues,
specific to local and global conditions, environment and loading. The intention is for the programme to receive
continued update to account for future modification and refinement to analysis approaches and as techniques and
technologies advance.

624
Various alternative protective and partially curative techniques may be used to combat each form of chemical attack,
based on the severity of their effects and the location and the nature of strengthening required. For diagnosis of
observed defects and/or subsequent selection of appropriate remedial and compatible crack treatment; ‘Pre-
Installation Treatment 1,2 &3’, ‘Weaknesses’, ‘Cracks’ and ‘Substrate’ of the ‘FRP Strengthening System -
Selection Toolkit’ should be consulted. Questioning on the user-forms in the software, associated with the a
foregoing categories, corresponds to the stages of pre-installation treatment investigation and requires responses to
establish the overall scheme priorities and limitations, progressively refined throughout each stage of treatment
selection. Additionally the environmental and existing structural criteria is defined through responses to the
‘Degradation Diagnosis’ phase of questioning, giving options and reasoning for the potential environmental,
chemical and mechanical defects present.

4. SELECTION
The knowledge base of the ‘Toolkit’ consists of two main knowledge modules pertaining to diagnosis & selection
(‘Pre-installation Treatment’) and preliminary design (‘FRP Strengthening’). Diagnostic information is stored in a
spreadsheet format, database search being undertaken through ‘If and Then’ ruling and ‘lookup’ systems, written in
Visual Basic. The benefit of relative simplicity in this form of data handling is in the ability for all routes through
the program to be predefined by the author once a workable decision tree has been established. Problems of changes
of the state of the variables with time are not of concern to this stage.

Taking the example of ‘System’ selection, a decision tree/network was used to create an Excel spreadsheet by
compiling a rudimentary series of questions, asked in the associated user-form, to delineate the overall geometry,
structure and construction methods used, but also to ensure compatibility with the previously selected repair,
preparation and pre-installation conditions. Further questioning is designed to determine whether a bond or contact
critical repair is necessary and hence what the stress-strain behaviour, eventual failure mode and ductility
requirements are likely to be. ‘System’ options of Plate (Pre-cured Laminate), NSM or Sheet are subsequently
progressively eliminated with applicable reasoning, based on their performance under the strengthening constraints
presented. The results obtained at this point are entirely qualitative, with built in subjective preference by the single
domain author. It is quite possible that substantial differences may be encountered in the rule base for different
experts and indeed facilities are provided, encouraging customary changes to be made, through the ‘expert
inclusions’ option at the base of each user-form.

5. STRUCTURAL ANALYSIS
Numerical calculations are now necessitated, some of which require iteration, whilst others require the solution of
sets of equations. Again built on decision flow/ network diagrams, responses to most questioning and execution
sequences are limited to ‘True’ and ‘False’, facilitating a rapid breakdown of the intricacies of design and division of
the various routes to completion. Manufacturer’s data is input by the user, allowing for testing and analysis with
contemporary material solutions. The final result is an economical cross-section, which satisfies the design criteria.
Responses given on user-forms called from the main-menu orientation page are written directly to the corresponding
spreadsheet/ flow chart in the Excel platform, activating the execution of subsequent stages of calculation within the
worksheet. Solutions and limits are then returned via message box captions to the screen or via the final report
textboxes, giving details of the final system.

Where seemingly complex or involved questions arise and are queried, or guidelines are sought for a series of such
questions by clicking on the available ‘checkbox’, numerous quick reference checklists have been compiled which
offer elaborations on the type of response demanded, or underlying nature of the conditions attempting to be
configured. Similarly where numerical input values are called for, options are given where possible, via a ‘see chart’
button function, which links the user to the worksheets corresponding to the calculation being undertaken, this
option allows for more accurate initial estimations, however optimisation may be achieved through continual trail
and error for all design parameters if complications arise.

American and British guidance is separated for the strengthening design stages, owing to the sometimes very
different principles adopted. The highly prescriptive approach used in ACI-440.2R-02 generates a very concise
program routine that has been adapted and applied to TR55 guidance by making alterations to the order of design
development officially recommended. Mixing the views held in the separate codes, or differential schools of thought
indiscriminately would almost certainly result in major shortcomings and inconsistencies.

625
6. COST ESTIMATION
The level of cost control included within the selection ‘Toolkit’ at this stage concentrates on the initial costs of
repair. Included within this categorisation are; material, labour (supervisors, skilled and unskilled) and UK
Highways Agency recommendations for start-up and preliminary design costs. It is argued that the single event
repair, using correctly designed and constructed FRP systems together with the negligible probability of failure,
results in near insignificant repair cost.

(Source: cited in Throft-Christensen (2000) – Repair cost model)

The limited level of allowable strengthening or repair of structures with FRP, demanding an ultimately service-
sustainable structural capacity following potential FRP failure, additionally eliminates failure costs of the
consequences of injury or loss of life from bridge damage. It is also argued that, particularly in the US, where repair
under live loading is permitted with nominal speed restrictions, user costs are minor factors of ultimate life cycle
pricing. Hence:
H

= Benefit that can be gained from existing bridge after rehabilitation

The final cost report, at this stage ultimately follows the recommendations generated through project RI 02-022,
‘Cost-Effectiveness Analysis of FRP Strengthening’. Firstly a cost per square foot is extracted, based on a budget
estimated per bridge span. Unit costs for material, labour, preparation and concrete repair are respectively adjusted
by a corresponding ‘factor ratio’ equated to the; amount of FRP material for element in question divisible by the
gross amount of material applied across the entire span.

7. CONCLUSIONS, POTENTIAL & FURTHER WORK

Preliminary design, involves the overall structural form of the artefact, satisfying a few key design constraints. This
early stage of the design process for repair and retrofit works, is perhaps more closely linked with the conceptual
design of a new structure, but is largely based around the adoption of key decisions, based on the criteria to be
satisfied at the detailed design stage. As such, an expert system and the particular method chosen for the ‘Toolkit’ is
entirely appropriate for providing a platform for integration of these related stages.

With the help of the ‘Toolkit’, design should now evolve with choices being made in a logical sequence. Where a
stage in the design process is reached, requiring a complete analysis to be carried out, the expert designer will do
this with the expectation that the structure is safe. The potential of the ‘Toolkit’ as a Value-Engineering tool and
device used for ‘Early-warning’ of project shortcomings, is also significant, in providing advanced notification of
potential or consequential constructional, technical or resourcing conflicts.

8. REFERENCES
ACI 440.2R-02, (2002), Farmington Hills, MI: American Concrete Institute, 2002. “Guide for the Design and
Construction of Externally Bonded Systems for Strengthening Concrete Structures,”
American Association of State Highway and Transportation Officials (2003), 17th ed, AASHTO, Washington, D.C.
“Standard Specifications for Highway Bridges”
FibTG9.3, (2001), “Externally bonded FRP reinforcement for RC structures”
Technical Report No.55, (2001), The Concrete Society, Berkshire RG45 6YS UK, (TR55) “Externally bonded FRP
Reinforcement for RC structures”
De-Lorenzis L.A.and Nanni A. (2005), Project RI 02-022, Preservation of Missouri transportation infrastructure
Validation of FRP composite technology through field testing, “Cost-Effectiveness Analysis of FRP Strengthening”
Lynch R.J (2004), Master Thesis, University of Missouri-Rolla, Rolla MO, U.S.A, “Provisional Design Guide in
AASHTO Language for FRP Bridge Strengthening”.

626
Third International Conference on FRP Composites in Civil Engineering (CICE 2006)
December 13-15 2006, Miami, Florida, USA

NATURAL FIBRE COMPOSITES FOR STRENGTHENING OF GLUED-

LAMINATED TIMBER IN TENSION PERPENDICULAR TO THE GRAIN
Alann André
(Ph.D. student, Civil and Environmental Engineering –Structural Engineering
Luleå University of Technology, SE-971 87 Luleå, Sweden)

Helena Johnsson
( Ph.D., Civil and Environmental Engineering –Timber Structure

Anders Carolin
(Ph.D., Civil and Environmental Engineering –Structural Engineering

ABSTRACT
This paper discusses strengthening of glued-laminated timber in tension perpendicular to the grain. Wood properties
are often inappropriate for high performance structural applications. Major drawbacks like durability and high
variability can be reduced by using glued-laminated timber. A further step to decrease the variability is to strengthen
the cross-section to prevent tensile failure perpendicular to the grain. This has been widely investigated during the
last decades by bonding fibre reinforced polymer (carbon, aramid and glass fibres) to timber or glulam beams, with
mostly promising results. However, a great concern about environmental friendly materials showed up a few years
ago. Mineral and petrol-based fibres are difficult to recycle and increase the amount of carbon dioxide in the
atmosphere leading, for instance, to the preoccupant greenhouse effect. Natural fibres such as flax are on the
contrary, recyclable and CO2 neutral. Their low density and high specific mechanical properties provide great
advantages for timber construction.
Specimens of glued-laminated timber with a volume of 0.01 m3 have been reinforced with fibre reinforced polymer
(FRP) composites, and tested in tension perpendicular to the grain. Unidirectional laminates have been prepared
with flax and glass fibres. Epoxy resin, which provides good bonding with hydroxyl groups in natural fibres, has
been used. The results show that there is a potential in using natural fibres for strengthening timber structures.

KEYWORDS
Natural fibres, Strengthening, Tension perpendicular to the grain, Flax fibres composites, FRP composites

1. INTRODUCTION
Wood mechanical properties perpendicular to the grain are weak and dominate failure initiations while loading a
timber member in bending. Besides, an important issue for a timber member is the failure mode. Compression
failure in wood is ductile. Unfortunately, most of the failures occur in tension perpendicular to the grain and are
brittle. This brittleness generates safety problems since it occurs suddenly. Indeed, brittle failures do not give
adequate warning signals that may initiate evacuation.
On the timber front, significant benefits of combining FRPs with wood-based composites in a variety of end-use
applications, including beams, has been demonstrated (Johnsson et al., 2005). With increased strength, reduced
variability, and elimination of size effects, high-performance FRP-wood composites may present new commercial
opportunities for both the FRP and the timber industry.

Natural fibres are light, renewable, CO2-neutral and possess interesting specific mechanical properties (Bledzki and
Gassan, 1998) which make them suitable to be used as polymer resins reinforcement for strengthening of structural
elements made of wood (André, 2006). Flax is grown in abundance in Europe, from Finland to Italy, and in many
other countries around the world. Temperate climate fits well to flax and European countries have focused their

627
natural fibre composite research principally on flax fibres because of its availability in Europe and its high
performance in term of mechanical properties (Oksman, 2001).

2. EXPERIMENTAL METHODS AND MATERIALS

2.1 Materials

All glulam specimens were cut from spruce glulam beams supplied by Martinsons AB, Sweden. The beams were of
strength class LT40 (BFS 2003:6, Swedish Construction Regulation), which is slightly similar to GL32c in ENV
1995-1-1: 2003 (Eurocode 5 – Design of timber structures). All specimens had the same volume V = 0,01 m³,
which is the reference volume use in EN 408 to characterize the tensile strength perpendicular to grain of glued
laminated timber.
Glass fibers were used to produce a reference FRP composite for the experiment. The glass fibre weave was
unidirectional with a weight of 250 g/m² (GFRP-250). Flax fibers from Finflax, Finland, have been manufactured by
Engtex, Sweden, to get two technical textiles (These products were derived from Project Texflax Contract N°:
G5ST-CT-2001-50111). The first flax fibre weave was 185 g/m² (FFRP-185) and the other 230 g/m² (FFRP-230).
FRP composites were manufactured using epoxy as matrix (NM infusion 664) and the resin infusion processing
technique. The FRP composites have been glued to the glulam specimens using the epoxy used previously. The
glulam specimens have been randomly chosen before being glued to the FRP composites to evenly distribute the
probability to have large defects. The steel plates, providing the interface specimen/tensile machine, have been glued
to the wood with epoxy (StoBPE 465) as in figure 1.

2.2 Experimental methods

The test method is based on EN 408 and was modified to prevent failure at the connected area between the glulam
reinforced FRP and the tensile test machine. The test set-up is shown in figure 1.

Figure 1: Test set-up

A total of 13 glulam specimens (405 mm × 223 mm × 115 mm) were tensile tested to failure in this experimental
program. The tensile testing equipment used was a Dartec Ltd RE 8991 machine with a 600kN load cell. The rate
was 0,15mm/min for all specimens. Two Linear Variable Displacement Transducers (LVDT) were used to measure
the displacement in the loading direction. The gauge length was 240mm.
All glued-laminated timber specimens had been previously conditioned at (20±1) °C and (30±10) % relative
humidity. Before all gluing, surfaces were sandblasted and cleaned from dust in order to ensure good bonding.

3. EXPERIMENTAL RESULTS
The tensile strength and modulus have been calculated for all specimens. Mean values and standard deviations are
reported in table 1. To get better comparison values, specific properties have been determined using the density of

628
the materials calculated according to EN 408. A failure was considered as semi-ductile when the first crack along
the section of the glulam did not lead to the final failure of the specimen.

Table 1: Absolute and Specific properties (Average values and standard deviation in parentheses)

Average Tensile Specific Tensile Specific Elongation

Failure
Specimens Number density Strength Strength Modulus Modulus at failure
mode
(kg/m³) (MPa) (MPa/g/m³) (MPa) (MPa/g/m³) (%)
Unreinforced 0,73 287,5
4 422 1,73 681,3 0,38 Brittle
Glulam (±0,27) (±34,32)
GFRP-250
1,80 465,3 Semi-
Reinforced 5 439 4,10 1060 0,84
(±0,36) (±72,53) Ductile
Glulam
FFRP-185
1,18 459,7 Semi-
Reinforced 2 426 2,77 1079 0,45
(±0,11) (±6,43) Ductile
Glulam
FFRP-230
1,49 475,7 Semi-
Reinforced 2 433 3,44 1099 0,48
(±0,29) (±7,47) Ductile
Glulam

Four load-deformation graphs corresponding to the four types of specimens are shown in figure 2. Failure occurred
mostly in the middle section of the specimens. In some cases, failure occurs at the interface steel plate/specimen.
These results were not taken into account.

55
Unreinforced FFRP-185 FFRP-230 GFRP-250
50

45

40

35
L o a d (k N )

30 FFRP-230 Reinforced
Unreinforced Glulam
25 Glulam
20

15

10

0
0 0,5 1 1,5 2 2,5
Displacement (mm)
FFRP-185 Reinforced GFRP-250 Reinforced
Glulam Glulam

Figure 2: Typical load-displacement diagrams

4. DISCUSSION
A previous research study (Blass and Schmid, 1999) focused on the tensile strength perpendicular to grain of
unreinforced glulam specimen. A total number of 153 specimens from different strength classes have been tensile
tested. It was found that there was no significant difference in tensile strength perpendicular to grain among strength
classes. The strength mean value was 0.77 MPa and the standard deviation 0.22.
The mean value for unreinforced glulam specimens in this investigation was 0.73 MPa with a standard deviation of
0.27. These results compare well to (Blass and Schmid, 1999), which show that the method used is accurate.
However, just a few specimens have been tested so far and precaution must be taken regarding this conclusion on
the test method.

629
For an approximate amount of FRP reinforcement of 0.7 % in volume (the thickness of the FRP layers were in
between 1 mm and 1.5 mm), it has been shown an increase of the tensile strength by respectively 246%, 161% and
204% using GFRP, FFRP-185 and FFRP-230. Regarding the modulus of elasticity, the previous reinforcement
devices led to an increase by respectively 161%, 160% and 165%.
We can see that a strengthening with GFRP provide a higher increase concerning the tensile strength than a FFRP
reinforcement. This is probably due to the higher tensile strength of glass fibre, but also to the higher fibre content in
GFRP laminates. However, no significant difference can be drawn regarding the modulus of elasticity between
FFRP and GFRP reinforced glulam. Indeed, flax fibers have a lower tensile strength than glass fibers but the moduli
of elasticity of both fibers are rather similar (André, 2006).
If taking into account the density of the materials, smaller differences in mechanical properties exist between
unreinforced and FRP reinforced glulam due to the lower density of the unreinforced glulam specimens. For the
same reason, it can be shown that the mechanical properties of FFRP reinforced glulam are closer to those of GFRP
reinforced glulam. However, the small amount of FRP reinforcement (0.7 %) doesn’t affect significantly the specific
mechanical properties even if the density of flax fibers is much lower than the one of glass fibers (1.4 and 2.6 g/cm3
respectively (André, 2006))
Larger displacements are generally observed in the semi-ductile failures. The FRP provides local bridging where
cracks in wood occur, confines the local rupture and arrests crack opening (See figure 2). For GFRP reinforced
glulam, a number of cracks in the wood have been observed before final failure in the fibers. Concerning FFRP
reinforced glulam, failure in the fibers often took place in the same section as the first and unique crack in the wood
occurred.

5. CONCLUSION
The use of Fibre Reinforced Polymer (FRP) to reinforced glulam perpendicular to grain enhances its tensile
properties. This has been observed both for glass fibers and flax fibers composites reinforced glulam.
It appears that the tensile strength is higher using glass fibers, but the weight of the glass fibers weave used is also
higher than the ones made of flax fibers. The stiffness, however, seems to be rather similar, even with a lower
amount of flax fibers.
Flax fibre reinforced glulam showed very promising properties if compared to glass fibre reinforced glulam. They
appear as a possible alternative to glass fibers. However, only a few specimens have been tested with flax fibers
reinforcement devices, and more experiments are needed to draw more significant conclusions.
For all specimens reinforced with fibre composites, semi-ductile failures are reported, which was one of the main
objectives of this investigation.
It will be interesting to study moisture transmission in wood and in the FRP since humidity generates higher
swelling and shrinking in the wood than it does in FRP. Internal stresses at the interface can appear due to this
phenomenon and should be avoid to provide a good interface bonding between Glulam and FRP.
Another important point is to study the variation of the properties. More specimens will be tested to complete this
pilot study and to be able to draw conclusions concerning the variation of the tensile properties.

6. REFERENCES
André, A. (2006). “Fibres for Strengthening of Timber Structures” ISSN: 1402-1528 ISRN: LTU-FR
Blass, H.J., and Schmid, M. (1999). “Tensile Strength Perpendicular to Grain of glued laminated Timber”.
International Council fro Research and Innovation in Building and Construction, Working Commission W18-
Timber Structures (CIB-W18) – Meeting thirty-two, Graz, Austria.
Bledzki, A.K., and Gassan, J. (1998). “Composites reinforced with cellulose based fibres”. Progress in polymer
science, Vol.24, pp221-274.
Johnsson, H., Johansson, T., and Carolin, A. (2005). “Glulam members strengthened by carbon fibre
reinforcement”. Submitted for publication in RILEM Materials and Structures
Oksman, K. (2001). “High Quality Fibre Composites Manufactured by the resin Transfer Moulding Process”.
Journal of reinforced Plastics and Composites, Vol.20, No. 07/2001, pp 621-627.

630
Third International Conference on FRP Composites in Civil Engineering (CICE 2006)
December 13-15 2006, Miami, Florida, USA

PRESTRESSED SRP AND CFRP SHEETS FOR STRENGTHENING

CONCRETE BEAMS
Daniel. B. Prentice
(Structural Design Engineer, Read Jones Christoffersen Ltd. Consulting Engineers, Calgary, Alberta, Canada)

Gordon Wight
(Associate Professor, Dept of Civil Engineering, Royal Military College of Canada, Kingston, Ontario, Canada)

ABSTRACT
Carbon Fibre Reinforced Polymer (CFRP) sheets, bonded to the tensile face of reinforced concrete (RC) beams, has
become a widely accepted method of strengthening concrete beams and slabs. To use the material more efficiently
and significantly reduce deflections, the sheets may be applied with a prestress. Despite the improvements that can
be achieved by prestressing the CFRP sheets, the technique is seldom used because the process can be difficult and
time-consuming due in part to the CFRP’s susceptibility to damage from a mechanical prestressing system. Very
recently, a Steel Reinforced Polymer (SRP) sheet system has been introduced for the strengthening of structures.
While these sheets have much in common with CFRP sheets, they are tougher and cheaper than CFRP sheets.
Because they are less susceptible to damage from lateral stresses, the SRP sheets appear to be very suitable for
prestressing applications. This paper outlines an experimental program, investigating the behaviour of five 1.8m
long RC beams strengthened with externally bonded non-prestressed and prestressed SRP or CFRP sheets.
Prestressing anchorages were developed to tension the SRP and CFRP sheets. Bonded anchorage plates proved best
for the CFRP while a mechanical anchorage technique was well suited for the SRP sheets. From these tests it
appears that SRP sheets are a competitive alternative to CFRP in both non-prestressed and prestressed strengthening
applications.

KEYWORDS
SRP, CFRP Sheets, Prestressing, Strengthening, Concrete Repair, Bonded, Mechanical Anchorages.

1. INTRODUCTION
Adapted from vehicle tire reinforcement technology, SRP systems offer slight ductility advantages over traditional
FRP systems. Orthotropic steel cords are held together and bonded to the tension face of a reinforced concrete beam
by an impregnation resin to compose SRP sheet strengthening. According to the manufacturer (Hardwire, 1999), the
3X2 cord configuration creates the best tensile properties for use in strengthening RC beams. Sikadur®330 is a
typical resin used to impregnate the SRP strengthening sheet as it provides the high viscosity required to slowly
penetrate filament voids and fully bond all steel filament cords together as a system. Sheet densities may also be
specified as low, medium or high by varying the spacing between individual cords to suit certain applications.
Application of SRP is similar to that of the FRP wet lay-up process.

Kim et al. (2005) performed preliminary testing of 1220mm long RC beams strengthened with high-density 3x2
SRP sheets. Observed failure modes were peeling failures where the line of failure either occurred in the concrete
cover or in the epoxy layer. Ultimate strength was increased 34% with non-prestressed SRP strengthening. Prota et
al. (2004), enhanced the flexural behaviour of 3700mm long RC beams with externally bonded high-density 3x2
SRP sheet strengthening. Test results were compared to CFRP sheet strengthened beams in the same experimental
program. Similar cracking behaviour was observed between beams strengthened with SRP and CFRP sheets.
Cracks during the service load range were reduced when SRP and CFRP sheet width was increased. All sheet-
strengthened beams experienced a peeling failure in the concrete cover. Ultimate strength was increased in the SRP
sheet-strengthened beams by 75-145% while the CFRP sheet-strengthened beams were increased by 95-173% over
the control beam. High strength steel cords may exhibit yielding behaviour similar to high strength prestressing
steel (Nanni et al., 2005). Casadi et al. (2005) utilized medium-density SRP to strengthen 13.4m long prestressed

631
concrete double-T girders in a decommissioned car park. When tested to failure, flexural capacity in the girders was
increased by 12% with one SRP sheet. Nanni et al. (2005) are currently investigating SRP sheet strengthening on
three 13m long spans of a 7.2m wide RC girder bridge. Four additional types of CFRP strengthening have also been
applied to similar bridges within the same study. These include manual lay-up technique with U-wraps, precured
laminates, near-surface mounted bars, and mechanically anchored laminates. Results from the study are still in
progress and will be used to determine the long-term effects of strain distribution, debonding, crack opening,
deflections and durability on these systems.

2. EXPERIMENTAL PROGRAM
The objective of this investigation was to investigate the behaviour of concrete beams strengthened with prestressed
SRP and CFRP sheets. Prestressing anchorages for the SRP and CFRP Sheets were designed to transfer initial
prestressing forces to actively strengthen RC beams. Anchorage plates were developed and tested until 30kN was
safely transferred into the strengthening material. This level of prestress was approximately 10% of the ultimate
strength of the SRP and CFRP sheets and ensured that the top concrete cover was not cracked during prestressing.
The prestressing system, depicted in Figure 1, consisted of two welded steel plate assemblies at each beam end to
serve the following functions:
1. Anchorage plates; transferred forces from two threaded rods to the strengthening sheet using either a bonded
(CFRP) or a mechanical (SRP) anchorage technique.
2. Anchorage Brackets; provided a bearing area at each end of the beam to transfer the prestress force from the
anchorage plate assembly to the beam. The brackets were anchored at each end of the beam specimen using full
span threaded rod stabilizers.

Figure 1: Prestressing Anchorage System

Five RC beam specimens were designed to resist 60kN at mid-span applied as two 30kN point loads. The
unstrengthened beams, depicted in Figure 2 (left), were under-reinforced with two longitudinal 10M bars in the
bottom tension zone to achieve flexural failure through steel yielding followed by concrete crushing. Two
longitudinal 10M bars were placed in the top compression zone to resist tensile forces created from prestressing.
Twelve 6M stirrups spaced at 50mm centre-to-centre provided shear reinforcement at each beam end. The
strengthening scheme for these beams is summarized in Table 1 and detailed in Figure 2 (right). One
unstrengthened beam (A1) was tested as a control specimen for this study. Remaining specimens (A2-A5) were
strengthened with externally bonded SRP or CFRP sheets to compare the flexural behaviour of SRP to that of a
leading strengthening material. Tension tests were also preformed on coupons of Hardwire’s high-density 3X2 SRP
and MBrace’s CF130/CF160 CFRP strengthening systems used in the comparison study. Strengthened beams were
designed to resist approximately 120kN at mid-span anticipating a debond-failure by equating sheet stiffness.

Table 1: 1.8m Beam Strengthening Scheme

Specimen Strengthening Type Sheets Thickness (mm) Width (mm) Area (mm2)
A1, B1 Unstrengthened Control - - - -
A2, B2 Non-Prestressed CFRP 1.5 0.33 120 59.4
A3, B3 Non-Prestressed SRP 1 1.19 140 166.6
A4, B4 30kN Prestressed CFRP 1.5 0.33 120 59.4
A5, B5 30kN Prestressed SRP 1 1.19 140 166.6

632
 Figure 2: 1.8m Beam Design Specification (left) and Strengthening Scheme (right)

3. RESULTS
Anchorage systems were developed to transfer initial stresses to the SRP and CFRP sheets during prestressing
operations. The susceptibility of carbon fibre to lateral stress damage necessitated that a bonded anchorage
technique be developed to prestress the CFRP. The CFRP sheet was built up and bonded to steel plates prior to the
event of prestressing to avoid damage from the concentrated stresses where the sheets terminate. The process of
pre-bonding takes time and expertise that ultimately increases the overall cost of the strengthening required. The
increased toughness of the high strength steel cord allows SRP sheets to be less vulnerable to lateral damage than
CFRP sheets. A mechanical anchorage technique is therefore well suited for prestressing the SRP sheets.
Mechanical anchorages for SRP sheets could save substantial time and money over the bonded anchorages for
CFRP when strengthening with prestressed applications. The dry 3x2 Hardwire® was weaved through a series of
two (2) 50mm long steel plates and clamped by a third floating plate using nuts tightened on threaded rods. Figure 3
depicts the application step of a prestressed SRP strengthened beam using the mechanical anchorage technique.

Figure 3: Prestressed SRP Strengthening Application with Mechanical Anchorage Technique

The flexural behaviour of 1.8m long non-prestressed and prestressed SRP and CFRP sheet-strengthened beams is
summarized during the service range and at the ultimate condition in Table 2. The load-deflection curves for these
beams are provided in Figure 3. The yielding point of the beams was chosen as the boundary that separates the
service range from the ultimate condition. The service range represents the flexural behaviour before yielding and
the service load was defined as the load equivalent to the yield load of the unstrengthened control beam (52kN).
The cracking load was increased by 76% and 228% and deflections at the service load were reduced by 36% and
70% for the non-prestressed and prestressed strengthened beams, respectively, with the application of SRP or CFRP
sheets. The ultimate condition was determined by investigating flexural behaviour during yield and at failure. Yield
loads were increased to 60% and 92% for non-prestressed and prestressed strengthened beams, respectively, and the
failure load of all strengthened beams was increased by 68-78%, using SRP or CFRP sheets. Behaviour was
improved during the service range using non-prestressed sheets, and even more so when prestressed sheets were

633
used. Ultimate conditions were improved similarly with non-prestressed and prestressed sheets. SRP and CFRP
sheets similarly improved the flexural behaviour of RC beams. It appears that strengthening with SRP sheets may
be an alternative to CFRP sheets for non-prestressed and prestressed applications.

Figure 3: Load-Deflection Behaviour of 1.8m Beams Strengthened with SRP and CFRP Sheets

Table 2: Behaviour of 1.8m Beams Strengthened with SRP and CFRP Sheets

Beam Cracking Service Yield Failure

Specimen Load (kN) Increase Deflection (mm) Reduction Load (kN) Load (kN) Increase
Control (A1) 12.5 --- 6.1 --- 52.1 59.0 ---
CFRP (A2) 22.1 77% 3.7 39% 84.6 104.8 78%
SRP (A3) 21.7 74% 4.1 33% 80.9 104.3 77%
Pre CFRP (A4) 41.1 229% 1.8 70% 99.2 99.2 68%
Pre SRP (A5) 10.8 226% 1.9 69% 100.9 100.9 71%

4. CONCLUSIONS
Prestressing anchorages were developed to prestress SRP and CFRP sheets. In-situ tensile testing of bonded and
mechanical anchorages on 1.8m long RC beams revealed the following conclusions.
1. Strengthening RC beams with prestressed SRP sheets can delay the formation of cracks, reduce deflections, and
increase strength during yielding and at the ultimate conditions, more so than non-prestressed applications.
2. The effectiveness of externally bonded SRP sheets to strengthen RC beams is similar to CFRP sheets in both
non-prestressed and prestressed applications.
3. It appears that the lateral stress resistance associated with steel cords make SRP sheets ideal for mechanical
prestressing anchorages.

5. REFERENCES
Casadi, P., Nanni, A., Alkhrdaji, T., Thomas, J., 2005: “Performance of Double–T Prestressed Concrete Beams
Strengthened with Steel Reinforced Polymer”, ASE, April.
Hardwire, 1999: “Laminate Details Interactive Spreadsheet”, Technical Data Sheet, www.hardwirellc.com,
Hardwire LLC, Pocomoke City, MD.
Kim, Y.J., Fam, A., Kong, A., El-Hacha, R., 2005: “Flexural Strengthening of RC Beams Using Steel Reinforced
Polymer (SRP) Composites”, Proceedings from FRPRCS-7 (ACI), Kansas City, MO, USA, Nov. 6-10, 1-17.
Nanni, A., Casadei, P. Alkhrdaji, T., 2005: “Steel-reinforced Polymer: An Innovative and Promising Material for
Strengthening Infrastructures”, Concrete Engineering International, Vol. 9, No. 1, 54-56.
Prota, A., Tan, K.Y., Nanni, A., Pecce, M., Manfredi, G., 2004: “Performance of RC Shallow Beams Externally
Bonded with Steel Reinforced Polymer”, Submitted to ACI Structural Journal.

634
Third International Conference on FRP Composites in Civil Engineering (CICE 2006)
December 13-15 2006, Miami, Florida, USA

FLEXURAL STRENGTHENING OF 48-YEAR OLD PEDESTRAIN

BRIDGE REINFORCED CONCRETE GIRDERS
Raafat El-Hacha
(Assistant Professor, University of Calgary, Calgary, Alberta, Canada)

ABSTRACT
As an emerging technology, the use of Fiber-Reinforced Polymer (FRP) reinforcements in the civil infrastructure
has seen an exceptionally rapid growth as an alternative replacement to steel reinforcement. FRP reinforcements
have been used in various configurations using different techniques for strengthening and repairing concrete bridges
to restore or increase their capacity. Externally bonded FRP reinforcements are currently the most commonly used
techniques for flexural strengthening of concrete girders and slabs.
This paper provides experimental results of an investigation that evaluated the efficiency and feasibility of various
systems for flexural strengthening of large-scale reinforced concrete girders dismantled from a 48-year old
deteriorated pedestrian bridge. The strengthening system comprised externally bonded Carbon FRP (CFRP)
including strips, plates and sheets. Another material known as Steel-Reinforced Polymer (SRP) was also used as
externally bonded sheets. Four beams were strengthened with the above various strengthening systems using the
same axial stiffness and tested under static monotonic loading up to failure. Two beams were tested without
strengthening. A 25% increase in the yield load and about 32% to 42% increase in the ultimate strength were
achieved. the SRP sheets were more effective at increasing the strength and ductility.

KEYWORDS
Carbon fiber reinforced polymer, externally bonded, flexure, girder, plates, sheets, static, steel, strips, strengthening.

1. INTRODUCTION
Forty percent of Canada’s bridges were built in the 1950s and 1960s and many are reaching the end of their service
design lives and require rehabilitation and strengthening. The civil engineering and construction industry are facing
unexpected challenges due to the state of repair of concrete infrastructure worldwide, and Canada is no exception in
which $44 billion is required to renovate deteriorated infrastructure. Engineers all over the world are challenged and
in search of new and affordable construction materials, cost-effective methods of extending the service live of
deficient structures, as well as innovative approaches and systems to problem solving. Fibre-Reinforced Polymers
(FRPs) have evolved as a promising form of reinforcement in new construction and rehabilitation projects. Various
FRP systems for strengthening concrete structures have been widely accepted as practical substitutes to traditional
strengthening techniques such as bonding steel plates, section enlargement, and external post-tensioning steel cables.
Strengthening systems utilizing FRP reinforcements (sheets, strips, plates) externally bonded to the tension zone of
concrete members are currently the most commonly used techniques for flexural and shear strengthening of concrete
beams and slabs. Some FRP strengthening techniques could be more effective than others; however, their cost
effectiveness is extremely important and could govern their use. The successful application of FRP for structural
upgrade has motivated the development of other novel low-cost materials that exhibit excellent structural properties.
One such material is composed of unidirectional knitted ultra high-strength steel wires forming cords (11 times
stronger than typical steel plate) that are assembled into a fabric embedded or impregnated within a polymeric resin
matrix and is referred to as Steel-Reinforced-Polymer, designated as (SRP). This paper investigates the feasibility
and effectiveness of using various externally-bonded systems/materials to strengthen four full-scale G-type
conventionally reinforced concrete girders dismantled from a deteriorated bridge near the city of Calgary that were
cast in 1958 with 25% less flexural reinforcing steel bars. The structural performance under static loading including
the behavior prior to cracking, post-cracking, yielding of steel and mode of failure of the strengthened girders will
be evaluated and discussed.

635
2. EXPERIMENTAL INVESTIGATION
2.1 Specimens Details
A total of six girders provided by Alberta Transportation were tested under static load. The girders were dismantled
from a pedestrian bridge near the city of Calgary and were 6.0 m (20 ft) long large-scale pre-cast G-type
conventionally reinforced concrete girders fabricated in 1958. A cross-section showing details of the girder is shown
in Figure 1. The girders are inverted open box channel with end blocks. The problems identified with these girders
include the insufficient number of stirrups, lack of load sharing, and stringer legs spallings at the bottom. Spalling
was noticed in the underside of the girders in localized sections, and some larger cracks were found on the girders
before testing. Considering the age of these girders and the environment they were exposed to, they were in a decent
shape. Though, in some places the clear cover was less than adequate and the flexural steel rebars were corroded.

1#4 3#5 36#3 1#4

(straight) (straight) (type B) (straight)
3/8′′ 4′′ 2′′ Type A

Type D
2#10 (type E)
1′-4′′ 1′ 61#3
(straight) Type E
35#5
2′′ (type A) 2#10 (type D)
6′′ 1′-3 7/8′′ 1´=12´´ Type B
2′-11 7/8′′ 1´´=25.4mm
Figure 1: Details of the G-type reinforced concrete pedestrian girders

2.2 Material Properties

Concrete
The concrete compressive strength as specified on the drawings was 28 MPa (4 ksi). Concrete cores were extracted
from the control girder and the actual compressive strength of the concrete was found 60 MPa (8.7 ksi). Also, the
concrete compressive strength was determined from Schmidt hammer tests and was found to be 58 MPa (8.4 ksi).

Reinforcing Steel
The girders were reinforced with 32mm diameter steel bars with specified yield strength 350 MPa (51 ksi). Three
samples of the reinforcing steel were removed from the end of the control girder and tested in uniaxial tension to
determine the tensile properties. The yield strength and modulus of elasticity were found 300 MPa (43.5 ksi) and
200000 MPa (29001 ksi), respectively

Strengthening Materials
The externally bonded strengthening systems selected for this study were Carbon-Fiber Reinforced Polymers
(CFRP) strips, CFRP sheets, CFRP plates, and Steel-Reinforced Polymer (SRP) sheets. The material properties of
the different systems are given in Table 1. A two-part component epoxy adhesive, the main epoxy resin (component
A) and the curing agent hardener (component B) was used. Sikadur 330 (mix ratio 4(A):1(B) by weight) was used
for bonding the SRP and CFRP sheets, and Sikadur 30 (mix ratio 3(A):1(B) by volume) was used for bonding the
CFRP plates and the CFRP strips to the bottom flange of the girders.

Table 1 − FRP material properties as reported by the manufacturers

FRP products Elastic Modulus Ultimate Tensile Strength
Dimensions
(manufacturer and type) (MPa) (MPa)
Pultruded CFRP Strip t =2.0 mm‡
124000 2068
(Hughes Brothers Alan 500 CFRP Tape) w = 16 mm‡‡
‡
Pultruded CFRP Plate t =1.2 mm
160000 2800
(Sika Carbodur® Type S 812) w = 80 mm‡‡
Unidirectional CFRP Sheet
t= 0.381 mm‡ 61012 715
(Sika Wrap® Hex230C)
Unidirectional SRP Sheet
0.44 mm2/mm† 206000 3170
(Hardwire™ 3×2-23-12)
† ‡ ‡‡
Net area per width t: thickness w: width (as shipped by manufacturer and not necessarily entirely used)

636
2.3 Test Matrix
Four girders (B1, B2, B3, and B4) were strengthened with various strengthening systems using the same axial
stiffness (AE) of the strengthening material (where A is the cross sectional area of the strengthening reinforcement
and E is its elastic modulus) so as to achieve a 30% increase in the carrying capacity. The strengthening systems
comprise externally bonded FRP reinforcements including different types of CFRP (strips, plates, and sheets) and
SRP sheets. Two beams were tested without strengthening and served as unstrengthened control specimens for
comparison purposes to compare the effectiveness of each technique in terms of percentage increase of the flexural
strength and overall structural performance. Table 2 summarizes the test matrix.
Table 2 − Test matrix for the G-type reinforced concrete girders
Beam # Externally Bonded Strengthening System
C1 Control beam without strengthening
C2 Control beam without strengthening
B1 Four CFRP strips per web (Hughes Brothers 500 Aslan CFRP Tape)
B2 One 80 mm wide CFRP plate per web (Sika Carbodur® Type S 812 Plate)
B3 Seven layers of 105 mm wide CFRP sheets per web (Sika Wrap® Hex230C Sheet)
B4 Two layers of 90 mm wide SRP sheets per web (Hardwire™ Sheet)

2.4 Surface Preparation and Installation of the Strengthening Systems

The bottom surface of the concrete webs was leveled with a grinder to eliminate any ridges. To ensure good and
strong bond, the surfaces were washed with a water pressure blaster and cleaned by air brushing to remove any
debris and dust. Large amounts of concrete had spalled off near the ends of the girders and the reinforcing steel was
exposed, therefore patching was done in these areas after removing loose concrete and oxidation from the
reinforcing steel. The mortar used for the patching repair was a combination of oven-dried sand and the Sikadur 30
epoxy adhesive with a mix ratio of 1:1 by volume. The mortar was allowed to cure for 24 hours before strengthening
was performed. Installation of the strengthening systems followed typical field conditions on the bottom flange
beneath the girders. The epoxy was allowed to fully cure at room temperature for at least one week before testing the
girders to failure. The anchorage system consisted of wrapping U shape unidirectional CFRP sheets (Sika Wrap®
Hex230C) bonded to the webs of beams B1, B2 and B3, while for beam B4, the anchor consisted of SRP sheets. The
anchors consisted of 28 pieces of 102×762 mm and 8 pieces of 305×762mm. The larger sheets were used at the ends
of the webs, and the seven sheets were spaced at 600mm center-to-center through the length of each of the webs.

2.5 Test Setup, Procedure and Instrumentation

The girders were simply supported, simulating the majority of pedestrian bridges, with a span of 5.84m and tested
under static monotonic loading up to failure. The girders were loaded at four-point bending with 1.2m spacing
between the two concentrated point loads. The load was applied using a 500kN capacity actuator through an MTS
controller-testing machine operating under displacement control mode at a constant loading rate of 2 mm/min. All
girders were fully instrumented to monitor their behaviour during testing by measuring the deflection at midspan
using Linear Strain Conversion devices (LSCs), strains in the concrete in the compression zone, and strain in the
CFRP and SRP reinforcements using electrical resistance strain gauges. Horizontal LSC were also placed at
midspan on each side of the girder to determine the strain in the concrete: one at 50 mm from top and one at 40 mm
from bottom approximately at the level of the reinforcing steel bars. Crack widths were measured using crack
comparator and their patterns were marked on the girders. Data were automatically collected and electronically
recorded using a data acquisition system. Typical test set-up and instrumentation is shown in Figure 2.

Loading frame girder

Loading frame column
500kN hydraulic ram,
semi-spherical seat and
1000kN load cell
Neoprene pads
Girder
Neoprene pads and
300kN load cell
LSC
Pedestal

2.32m 1.2m 2.32m

Figure 2: Test set-up and instrumentation of the G-type reinforced concrete girders

637
2.6 Test Results and Discussion
Beam C1 was loaded until a load of 287 kN which is 63% of the predicted load of 452 kN determined according to
the provided information on the girders (Figure 1). Decision was made to test another control beam (C2) and the
failure load recoded was 309 kN which is 68% of the predicted load. Testing of the second control beam confirmed
the results of the first beam. However, to prove the accuracy of the given specifications and drawings for the
theoretical predictions, the concrete cover was hammered off from part of the two webs to expose the reinforcing
steel and it was discovered that only three steel bars in two layers were used in each web instead of four bars as
shown on the drawings. Thus, these girders were cast with 25% less flexural reinforcing steel. This proved why the
control beams yielded significantly lower flexural strength than calculated. The prediction of the ultimate load was
recalculated based on the results of the materials (concrete and steel) tests and the accurate locations and number for
the reinforcing steel bars and found to be 367 kN indicating 22% and 16% difference for beams C1 and C2,
respectively. The load versus midspan deflection curves comparing the flexural behaviour of the beams are
presented in Figure 3. The load-deflection behaviour is bilinear until failure. All beams exhibited similar behaviour
with regard to cracking and their load-deflection followed almost similar paths until failure. Figure 4 shows the load
versus strain in the CFRP and SRP reinforcements at midspan. None of the beams failed by rupture of the
strengthening materials; as can be seen in Figure 4 at ultimate load the strain in the strengthening materials were less
than the ultimate tensile strains. In beam B1, debonding at the interface of the CFRP strips and the epoxy took place
at a load of 400 kN starting at the center of the span and as a result the load decreased slightly. This debonding
stopped when it reached the nearest transverse anchor. This allowed the load to increase again as the imposed
deflection increased and the FRP was “tightened”. Suddenly the anchor would separate allowing debonding up to
the next anchor. Afterwards the load would repeatedly slowly increase until the next anchor separated. Finally the
CFRP reinforcement tore away from the last anchor and debonded from the beam completely at a load of 417 kN.
Similar behaviour was observed in beams B2, B3, and B4, debonding occurred at a load of 380 kN, 407 kN, and 428
kN; respectively, then the CFRP and SRP reinforcement tore away from the anchor at a load of 407 kN, 415 kN, and
435 kN, respectively in beams B2, B3, and B4. The debonding behaviour is illustrated by the jogs in the curves just
before failure. The failure of the CFRP strengthened beams with CFRP sheet anchors was dramatic, as it literally
tore the anchors off the corner of the web and the CFRP strips, plates, and sheets fell to the ground. The increase in
the yield load was about 25% in all beams. After yielding, the strengthened beams continued to resist further
increase in the applied load with a more gradual linear slope than the pre-yield portion of the curve. The increase in
the load continued until failure. The ultimate strength increased by 35%, 32%, 34% and 42% in B1, B2, B3, and B4,
respectively. However, the SRP sheets are more effective at increasing the strength and ductility.

450 450
B3
B4 B4
400 400 B3
B1 B1
B2 B2
350 350

C2
300 C1 300
Applied Load (kN)
Applied Load (kN)

250 250
Unstrengthened Beam C1

200 Unstrengthened Beam C2 200

Beam B1 (EB-CFRP Strip HB)
Beam B1 (EB-CFRP Strip HB)
150 150 Beam B2 (EB-CFRP Plate SIKA)
Beam B2 (EB-CFRP Plate SIKA)
100 100 Beam B3 (EB-CFRP Sheet SIKA)
Beam B3 (EB-CFRP Sheet SIKA)

50 50 Beam B4 (EB-HW Sheet)

Beam B4 (EB-SRP Sheet)

0 0
0 10 20 30 40 50 60 70 80 90 100 110 120 130 140 0 2000 4000 6000 8000 10000
Midspan Deflection (mm) Strain in FRPs at Midspan (micro)

Figure 3: Load-midspan deflection curves for all beams Figure 4: Load-FRP strain curves for all beams

3. CONCLUSIONS
Based on the results of this experimental study the following conclusions can be made:
– The ultimate strength gains achieved by the externally bonded reinforcements exceeded the initial goal of 30%.
– An increase in the yield load and ultimate load of 25% and up to 42%, respectively was achieved.
– All strengthened beams failed in a ductile manner accompanied by large deformation; however the beam
strengthened with SRP sheets showed mores ductile behaviour and higher capacity than the other beams.
To summarize, this study has confirmed the structural benefits and feasibility of using externally bonded CFRP and
SRP reinforcement to strengthen deficient reinforced concrete girders. However, the CFRP is particularly attractive
since it is not susceptible to corrosion, and extremely lightweight making it easy to work with.

638
Third International Conference on FRP Composites in Civil Engineering (CICE 2006)
December 13-15 2006, Miami, Florida, USA

STRENGTHENING OF RC BEAMS WITH PRESTRESSED NEAR

SURFACE MOUNTED CFRP RODS
Moataz Badawi
(PhD Candidate, University of Waterloo, Waterloo, Ontario, Canada)

Khaled Soudki
(Professor,CRC, University of Waterloo, Waterloo, Ontario, Canada)

ABSTRACT
The use of existing bridge structures are changing with time. The applied loads on these bridges have become larger
and more frequent in the last decades as the population and their use have increased. Therefore, strengthening of
such structures becomes a necessity. This paper presents experimental results of testing four reinforced concrete
(RC) beams under monotonic loads. One beam is kept as a control beam representing the existing structure. The
second beam is strengthened with near surface mounted (NSM) carbon fiber reinforced polymer (CFRP) rod. The
other two beams are strengthened with NSM prestressed CFRP rod. Two different levels of prestressing are
investigated. The test results show that using NSM CFRP rod without prestressing enhanced the flexural capacity
(stiffness, yield and ultimate loads) compared to that of the control beam. Prestressing of NSM rods demonstrates a
higher efficiency of using those expensive materials by achieving the ultimate capacity of the rod at failure. Beams
strengthened with prestressed NSM CFRP rod have enhanced the flexural performance in terms of flexural stiffness,
yield and ultimate loads with respect to both, the control beam and those strengthened with non-prestressed NSM
CFRP rods.

KEYWORDS
NSM, CFRP, RC Beams, Prestressed

1. INTRODUCTION
The current service loads of the existing infrastructures such as bridges might be higher than the design loads at the
time when they were constructed. This may be due to a population growth which leads to increased demands and
volumes of traffic on these bridges. Design and construction errors could also be another cause of reduction in the
capacity of these structures to carry the service loads. Thus, strengthening of bridge infrastructure is needed. Fiber
reinforced polymer (FRP) has attracted structural engineers’ concerns as a strengthening materials. The main
advantage of using such materials, especially in cold climate countries is that they are not corrosive. In addition,
they are very light materials compared to reinforcing steel (high strength to weight ratio). The FRP’s have been
extensively used and studied as externally bonded on the soffit of RC beams for flexural strengthening (Hassan and
Rizkalla, 2002; Aidoo et al., 2004). Lately, an old technique has been re-adopted in the strengthening of RC
structures using FRP materials, which is referred to as near surface mounted (NSM) reinforcement (Asplund, 1949).
Several investigations have been carried out to study the effectiveness of NSM FRP (Blaschko and Zilch, 1999; De
Lorenzis et al., 2004, Yost et al., 2004). These studies have shown comparatively a similar enhancement in the
flexural response to those of the externally bonded FRP. To achieve a better use of FRP, prestressing and bonding
the CFRP rod in a pre-cut groove in the RC beams is proposed and presented in this study. Only one study was
reported in the literature using prestressed CFRP NSM strips as NSM conducted in Sweden (Nordin et al., 2001). It
was found that prestressing increased the yield load with the ultimate load same as non-prestressed strengthened
beam. This study presents a new North American experience of using round CFRP rods as prestressing materials for
strengthening RC structures.

639
2. EXPERIMENTAL PROGRAM
Four large-scale reinforced concrete (RC) beams were designed and constructed. The beams had a cross section of
152×254mm and a total length of 3500mm. The compressive strength of concrete was 47MPa. The beams were
reinforced with 2-No. 15M and 2-No. 10M for tension and compression reinforcement, respectively. Shear
reinforcement was provided by 8 mm smooth stirrups spaced at 75 mm from center to center to ensure a flexural
mode of failure. The specified yield strength of steel reinforcement was 400MPa. The geometry and reinforcement
details of the RC beams are shown in Fig. 1. As given in Table 1, one beam was kept without strengthening as a
reference beam, while the other three beams were strengthened with NSM CFRP with three levels of prestressing,
namely, 0% (non-prestressed), 40%, and 50% of the tensile strength of the CFRP rods. The mechanical properties of
the CFRP rod used were: a tensile strength of 1970MPa, a modulus of elasticity of 135.9GPa, and an ultimate strain
of 1.45%. The CFRP rod that was used for strengthening was AslanTM by Hughes Inc. It had a diameter of 9.4 mm
and was placed in a near surface mounted groove having dimensions of 15 mm wide and 25 mm deep.

8 mm@ 75mm

3500 mm

Control Beam Strengthened Beam

2-No15M

15 mm

(9.4 mm dia.)
254 mm

CFRP Rod
25 mm
2-No10M

152 mm c.c.: 30 mm 152 mm

Epoxy

Figure 1: RC beam schematic

Table 1: Test matrix

Prestressing Level Prestressing Force Initial Strain

Designation Description
(%) (kN) (µe)
C Control - - -
PS-0 0% prestressed strengthened 0 0 0
PS-40 40% prestressed strengthened 40 55 5721
PS-50 50% prestressed strengthened 50 69 6933

The prestressed CFRP rods were tensioned using a clamp anchor developed at the University of Waterloo (Al-
Mayah, 2004) and a prestressing set-up was designed for that purpose. After achieving the desired force in the CFRP
rod, the NSM groove was completely filled with epoxy (Sikdur 30, creep resistant). To ensure a full filling of the
epoxy in the groove, it was repeatedly added and pressed into the groove. After 7 days of curing, the prestressing
force was released and transferred to the beam by means of bond.

The beams were instrumented with several strain gauges on concrete, reinforcing steel (compression and tension)
and CFRP rod at the mid-span section. All the beams were monotonically loaded to failure under a four point-
bending fixture with a clear and shear span of 3300 mm and 1100 mm, respectively, at a 1.5mm/min rate of loading.
Figure 2 shows schematic drawing and photograph of the test set-up. Three linear variable differential transducers
(LVDT’s) were placed along the beam to measure the vertical deflections at a mid-span, a point load, and at a mid-
shear-span section during testing.

640
 Strain gage
LVDT
LVDT Holder 40

TEST SPECIMEN

550 550 550 1650

3300
All dimensions in mm

Figure 2: Test set-up

3. TEST RESULTS AND DISCUSSION

For the strengthened RC beams, the initial strain measurements at mid-span section in the CFRP rod were 0 µe,
5721µe, and 6933 µe for 0%, 40% and 50% prestressed strengthened beams, respectively. Two modes of failure
were obtained. A yielding in the tension reinforcement followed by a crushing in concrete was obtained for the
control and non-prestressed strengthened beams (0% of prestressing). Beams strengthened with prestressed CFRP
rods failed by a rupture in the CFRP rods after yielding of tension reinforcement steel. Figure 3 shows the plots of
the load versus central vertical deflection for all tested beams illustrating the three stages of the behaviour of typical
under-reinforced concrete beams (pre-cracking, pre-yielding and post-yielding stage). Table 2 gives the cracking,
yield, ultimate loads of all strengthened beams with their percentages of the achieved increases in a comparison to
those of the control beam. Strengthening RC beams with NSM CFRP rod was not only able to improve the flexural
capacity of the strengthened beams in terms of cracking, yield, and ultimate loads but also was able to significantly
increase the flexural stiffness with respect to those of the control beam. It is worthy noting that as the prestressing
level of the CFRP rod increased, the flexural response of the strengthened RC beams was considerably improved.

140 PS-50 Beam

120 PS-40 Beam

100 PS-0 Beam

Load (kN)

80

60 C Beam

40

20

0
0 20 40 60 80 100
Vertical Displacement (mm)

Figure 3: Load- deflection relationship

641
 Table 2: Test results

Cracking load Yield load Ultimate load

Beam Load Increase Load Increase Load Increase
(kN) (%) (kN) (%) (kN) (%)
Control 6.5 0 55 0 65 0
0% prestressed strengthened 10 54 71 29 95 46
40% prestressed strengthened 28.7 341 98 78 115 77
50% prestressed strengthened 33.1 409 116.6 112 118 81

4. CONCLUSION
A successful new experience in strengthening structures by using prestressed CFRP NSM rod to upgrade the
flexural performance of RC beams is presented and investigated. Using NSM CFRP rod increased the flexural
stiffness and capacity of all strengthened beams. Prestressing the CFRP rod before bonding in the NSM groove
further enhanced the flexural performance. It also provided a better utilization of the mechanical properties of such
material. By obtaining such results and performing further investigations, this method of strengthening might be
considered to be useful in the field application.

5. ACKNOWLEDGEMENTS
The authors greatly acknowledge the technicians at University of Waterloo in particular Mr. Ken Bowman for his
help and support through all stages of this study. Sika Canada is greatly appreciated for the material donations.

6. REFERENCES
Aidoo, J; Harries, K.A.; and Pertou, M. F.,”Behaviour of Reinforced Concrete Bridge Retrofit with CFRP and
subjected to Monotonic and Fatigue Loading,” Proceedings of the 4th International Conference on Advanced
Composite Materials in Bridges and Structures, Calgary, Ontario, Canada, 2004
Al-Mayah, A., “Interfacial Behaviour of CFRP-Metal Couples for wedge Anchor Systems,” PhD Thesis, University
of Waterloo, Waterloo, Ontario, Canada, 2004
Asplund, S. O., “Strengthening Bridge Slabs with Grouted Reinforcement,” American Concrete Institute, Structural
Journal, 1949, pp. 397-406
Balschko, M.; and Zilch, K., “Rehabilitation of Concrete Structures with Strips Glued into Slits,” Proceedings of the
12th International Conference on Composite Materials, Organization of the International Conference on Composite
Materials, Paris, France, 1999
De Lorenzis L.; Nanni, A.; and LA Tegola, A., “Flexural and Shear Strengthening of Reinforced Concrete
Structures with Near Surface Mounted FRP Rods,” Proceedings of the 3rd Advanced Composites Materials in
Bridges and Structures, Calgary, Ontario, Canada, 2004, August, 2000, pp. 521-528
Hassan, T., and Rizkalla, S., “Flexural Strengthening of Prestressed Bridge Slabs with FRP Systems,” Prestressed
Concrete Institute Journal, 47, 2002, pp. 76-93
Nordin, H,; Taljsten, B. and Collin, A, “Concrete Beams Strengthened with Prestressed Near Surface Mounted
Reinforcement,” Proceedings of the International Conference on FRP Composites in Civil Engineering, Research
Centre for Advanced Technology in Structural Engineering, Dept. of Civil Engineering and Structural Engineering,
The Hong Kong Polytechnic University, Hong Kong, 2001, pp. 1067-1075
Yost, J. R.; Gross, S.P.; and Dinehart, D. W.; “Near Surface Mounted CFRP Reinforcement for Structural Retrofit
of Concrete Flexural Members,” Proceedings of the 4th International Conference on Advanced Composites
Materials in Bridges and Structures, Calgary, Ontario, Canada, 2004.

642
Third International Conference on FRP Composites in Civil Engineering (CICE 2006)
December 13-15 2006, Miami, Florida, USA

STRENGTHENING OF STEEL BRIDGES UNDER LOW FREQUENCY

VIBRATIONS
Lei Zhang
Post Doctoral Research Fellow, The Hong Kong Polytechnic Umiversity, Hong Kong China

L C Hollaway,
Professor of Composite Structures, University of Surrey, Guildford, Surrey, UK

J G Teng,
Chair Professor of Structural Engineering, The Hong Kong Polytechnic University, Hong Kong, China.

S.S. Zhang
Research Assistant, The Hong Kong Polytechnic University, Hong Kong, China.

ABSTRACT
A potential technique utilizing advanced polymer composites to rehabilitate a steel railway bridge open to traffic
during the adhesive curing period is presented. The advanced polymer composites used for the upgrading system are
manufactured from a factory-made hot melt pre-impregnated high modulus and ultra high modulus carbon fibres;
the polymer system is an epoxy. The composite is bonded to the steel structural member with a compatible film
adhesive; the two components are then fabricated in one operation onto the structure. The compaction is performed
under a vacuum assisted pressure of 1 bar and two cure temperatures are used, one at 650C for 16 hours and the
other at 800C for 4 hours; the length and time of cure are dependent parameters. The FRP structural and material
characteristics and the beam test results are presented. It is shown that (i) some slight damage to the adhesive from
the vibrational forces, during the cure period, did occur but the strain to failure of the high modulus CFRP composite
took place well into the non-linear region of the steel beam and (ii) the high modulus carbon fibre composite
supported 10% greater load than the ultra high modulus carbon fiber composite as a result of the low strain to failure
of the latter carbon fibre. It is noticed, however, that half the thickness of the ultra high modulus CFRP composite is
used compared with the high modulus CFRP. This allows the yielding of the steel beam to occur at a similar load
value for the two upgrades and prevents a severe brittle failure of the composite beam.

KEYWORDS
Composites, carbon fibres, film adhesive, hot melt pre-impregnated composites.

1. INTRODUCTION

Many of the bridges world-wide are made of steel and a large percentage of these bridges are considered
substandard and require upgrading or improvement from a strengthening/stiffening point of view. The main causes
for the deterioration in steel bridges are a lack of proper maintenance, corrosion attack and fatigue sensitive details,
(Tavakkolizadeh and Saadatmanesh 2003). In addition, many of the bridges require upgrading to enable them to
meet present day traffic requirements. Clearly, a repair or rehabilitation option should be considered before a
decision is taken to replace the structure, as the former option, in most cases, costs far less than the latter option and
generally takes less time.
There is currently a considerable amount of research and investigative work being undertaken into the use
of advanced polymer composites (APC) to upgrade and rehabilitate metallic structures, particularly steel bridges.
(Luke 2001, Cadei et al. 2004). The excellent mechanical, fatigue and in-service properties of carbon fiber
reinforced polymers (CFRPs) make them ideal candidates for the strengthening and rehabilitation of steel bridges.
A major problem which must be addressed when upgrading a steel railway bridge continuously traversed
by trains, thus setting up vibrations in the structure during the polymerization process, is to know which type of
adhesive should be used and its curing period. There are two techniques generally utilized in practice to rehabilitate

643
a structural steel member; these are (a) the wet lay-up method where the polymer of the matrix also acts as the
adhesive polymer, and (b) the rigid plate method in conjunction with the two part adhesive joint system. The first
procedure would generally be used on a curved member when it is not possible to use the rigid plate method. The
second procedure is to manufacture, by the pultrusion or prepreg method, a rigid plate which is bonded to the
straight beam by a two part adhesive polymer. The adhesives used in the two techniques are cold cure epoxies which
polymerize at ambient temperature (i.e. the temperature of the environment at the time of upgrading the structure).
Depending upon the site temperature, the adhesive will take 7 to 10 days to reach 90% polymerization. If a steel
bridge structure on which these systems are used is under a vibration loading, the movement of the bridge, as a
consequence of the passage of trains over the bridge, will impair the polymerization mechanism. Therefore, in this
case, the cold cure adhesives are not desirable for the rehabilitation of steel railway bridges under vibrations.
The objective of this paper is to investigate the effectiveness of bonding a high modulus (H-M) CFRP
prepreg to strengthen a structural steel system which, during polymerization, is under a continuous low frequency
vibration. The strengthening characteristics of an ultra-high modulus (UH-M) CFRP prepreg is also compared with
those of the H-M CFRP prepreg.

2 THE PRE-IMPREGNATION MANUFACTURING TECHNIQUE

Advanced Composite Group (ACG) Derbyshire UK has developed a pre-impregnation (prepreg) technique
specifically for the civil engineering industry. The advanced polymer composite used for the prepreg is
manufactured from a factory-made hot melt epoxy polymer system; the technique involves the prepreg composite
and a compatible film adhesive. The construction site receives the prepreg and film adhesive both of which are
stored at -200C from the time of manufacture until use. Before cutting the two components to size, the materials are
thawed for one hour. The adhesive film and the composite prepreg are then fabricated on to the beam by laminating
the prepreg layers, as per design, with the adhesive film located correctly. A halar film and breather blanket cover
the FRP composite on the beam and the whole composite system is surrounded in a vacuum sheet properly sealed at
its extremities. A vacuum assisted pressure of 1 bar is applied and a heater blanket covers the whole composite. The
prepreg cure temperature is 650C for 16 hours, or it can be raised to 800C applied for 4 hours. At the conclusion of
the cure period the temperature is reduced to the environmental condition. The technique has been used in practice
to encapsulate a curved steel section which had degraded badly, (Garden and Shahidi, 2002).

3 THE CURE PROCEDURE FOR THE PREPREG COMPOSITE

The site cure procedure of the prepreg and adhesive film has major advantages to offer compared to the pre-cast
plate and two part adhesive system for the rehabilitation of structural systems in the civil engineering industry. The
advantages are:
(i) A much better control of the site fabrication and compaction operation. This will enable the adhesive thickness
to be controlled to within high tolerances.
(ii) The vacuum assisted pressure of 1 bar reduces any voids in the composite to a minimum.
(iii) The elimination of any variation in cure temperature due to site conditions.
(iv) The site cure at an elevated temperature enables a control to be made of the value of the glass transition
temperature (Tg). The higher the cure temperature, the higher the Tg value will reach, but there is a limit
irrespective of the temperature value of cure. This limiting value will vary from polymer to polymer.
As the cure is undertaken at an elevated temperature under controlled conditions on site the polymerization
of the adhesive (and composite) will be completed much more quickly than the cold cure two part adhesive, which is
dependent upon the environmental conditions. Furthermore, the cold cure adhesive used currently in construction
will take 5 to 7 days, depending upon site temperature, to reach 90% polymerization. After this time, if the system is
not post cured, polymerization will continue but over a very long time and will probably never reach 100%. The
advantages that the site cure prepreg and film adhesive have over the other site bonding techniques have been
demonstrated in (Photiou et al 2006) and can be exploited further when a structure is under vibrations whilst it is
being rehabilitated.

4 THE REHABILITATION TECHNIQUE OF THE TWO STAGE CURE PROCEDURE.

The rehabilitation of structural members of a steel railway bridge is taken as an example of utilizing the factory-
made hot melt pre-impregnated CFRP composite and compatible film adhesive. The following two scenarios are
considered in the design of the laboratory tests presented in this paper:

644
(i) The bridge is closed to rail traffic for 4 hours in a 24 hour period (ideally during the night shut-down period)
whilst upgrading is in progress.
(ii) The bridge is open to continuous rail traffic and is exposed to vibrations throughout.

5 THE COMPOSITE MATERIALS

Laboratory tests are conducted to evaluate the effectiveness of the rehabilitation technique based on the hot melt pre-
impregnated site cure procedure. The main upgrading material system used is made from unidirectional high
modulus CFRP prepregs, comprising of two double-ply laminates with ply thickness of 0.6 mm, the carbon fibers
aligned with the longitudinal direction of the beam. Three single plies of GFRP are used with glass fibers positioned
at +450 to the longitudinal direction of the steel beam. The various layers are stacked as follows: firstly a single
GFRP ply followed by one double-ply CFRP, followed by a single GFRP ply, followed by the other double-ply
CFRP and finally another single GFRP ply. The first GFRP ply is adjacent to the adhesive film and hence separates
the steel from the carbon fibers. The ulta high modulus CFRP composite is fabricated in the same way. The plate
plan size of the prepregs are 60 x 1600 mm. The thickness of the adhesive film is about 110-120μm, and like the
matrix material of the prepreg is made from a low temperature cure epoxy resin system. The specifications of the
materials used are:
The high modulus CFRP - VTM264FRB/PANEX35-50K-600-35%RW (Uni-directionally aligned)
The + 450 biaxial GFRP - VTM264FRB/EBX602-32%RW
Adhesive film - VTA260/PK13-313g- (compatible film adhesive with the
composite matrix VTM264FRB)
The ultra high modulus CFRP - VTM264FRB/K63712-600-35% RW (Uni-directionally aligned)
The mechanical properties of the composite materials and adhesive film are given in Table 1.

Table 1: Mechanical properties of FRP materials and adhesive film

Material Tensile Strength Elastic Modulus Ultimate Strain Poisson’s Ratio

MPa GPa (%)
UH-M CFRP
(Unidirectional) 1120 270 0.4 0.32
HM-CFRP
2110 135 1.6 0.28
(Unidirectional)
GFRP
(± 450 to line of 215 16 1.7 0.15
action of load)
Film Adhesive 32 3.7 0.9 0.37

6 THE UPGRADED BEAM PREPARATION

The beams are cut to a length of 2000 mm and a flat plate of dimensions 56 x 2000 mm is welded on to each of the
compression flanges; this represents beams that had been degraded on the tension flange. The exposed tension
flange of the beam is grit blasted to the Swedish Code SA 2½ Grade. Immediately before applying the adhesive film,
the surface of the steel is solvent degreased using acetone to remove any contaminant materials.

The cure at 800C for 4 hours The adhesive film and hot-melt prepreg laminates are laid on the two beams followed
by the halar film, the breather blanket and the vacuum sheet and a vacuum assisted pressure of 1 bar is applied to
both beams. The heating blanket temperature cure is raised to approximately 800C. These beams represent the case
of a bridge which is closed to traffic for 4 hours in a 24 hour period. After curing is completed the temperature of the
heating blanket and the specimens are reduced to room temperature at a steady rate. The upgraded beams are then
fully cured at that cure temperature and ready for testing. Steady static loads are applied up to the failure of the
beams and comparisons are made with the control beam.

The cure at 650C for 16 hours Two beams are prepared for the scenario of a bridge that is continuously traversed by
rail traffic. The beams are prepared in the same way as that described for the first beam above. However, for these
cases a heating cycle of approximately 650C for 16 hours was used during which the beam specimens are exposed to
loads varying between an upper and a lower limit throughout the 16 hours. Different load limits are employed in the

645
two tests, (i.e. 5 kN-40 kN and 5 kN-60 kN) under a frequency of 1 Hz. The beams, when cured, are statically tested
to failure.

7 TEST LOADING TO FAILURE

Each beam is placed in the test rig on a clear span of 1700 mm. The two external loads are positioned at 200 mm on
either side of the mid-span of the beam. Strain gauges are positioned at strategic positions on both sides of the beam
at mid-span and down its sides. The test set-up and the cross section of the beam are shown in Figures 1a and 1b
respectively. Displacement transducers are placed at mid span and a number of other locations. Two load cycles up
to 20 kN and 40 kN are applied to the beam before taking it to failure under a steadily increasing displacement of
0.05 mm/s. Figure 2 shows the total load against central deflection for the H-M CFRP rehabilitated beams cured at
800C and 650C, the UH-M CFRP rehabilitated beam cured at 800C and the control beam. Figures 3 and 4 illustrate
the strain distributions in the pure bending region of beams rehabilitated with CFRP prepreg cured at 800C and 650C
respectively.

80
56 6
P

6
6

120
110
60
15 5 5
400
1600

6
1700 60

Figure 1a Loading arrangement Figure 1b Cross section of steel beam

200

HM-80deg.C
160 HM-65deg.C(5~40kN)
UHM-80deg.C
Load (kN)

120
HM-65deg.C(5~60kN) Control beam

80

40

0
0 10 20 30 40 50 60
Deflection at mid-span (mm)

Figure 2 Total load~central deflection

646
 140

120 40 kN

Distance from the bottom (mm)

100 kN
100 140 kN
170 kN
80 190 kN

60

40

20

0
-12000 -8000 -4000 0 4000 8000 12000
Microstrain

Figure 3 Strain Distributions for beam with HM- CFRP

composite cure at 80deg.C

140

120
40 kN
Distance from the bottom (mm)

100 kN
100
140 kN
170 kN
80
180 kN
60

40

20

0
-10000 -7500 -5000 -2500 0 2500 5000 7500 10000
Microstrain

Figure 4 Distribution of strains for beam HM- CFRP

composite cured at 65deg.C(5~40kN)

From Figure 2 the three upgraded beams have almost identical characteristics, with non-linear response
from about 80 kN (the yield load) and this corresponds to the steel exhibiting softening response in the soffit of the
beam. The load continues to increase in all four cases up to 138 kN, 160 kN, 178 kN and 190 kN for UH-M (800C),
H-M 650C (N 5-60), H-M 650C (N5-40) and H-M 800C, respectively. The load deflection paths for the three beams
with H-M CFRP prepreg upgrades follow a similar pattern irrespective of their cure temperature, but the beam with
UH-M CFRP has a slightly stiffer characteristic compared to that of the beams with H-M CFRP upgrade. This
indicates that the UH-M CFRP composite with a greater stiffness than that of the steel is stiffening the composite
beam (i.e. the strengthened beams) in the elastic region of the original steel beam. The strain distributions in Figures
3 and 4 show shifts in the neutral axis, furthermore, it can be seen that the CFRP composite reached a large strain at
failure. The composite on the upgraded beam in Figure 4 (when the FRP composite and adhesive film are cured at
650C under vibration) failed at 10000 µ-strains and about 8500 µ-strains, under vibrations of 5-40 kN and of 5-60
kN respectively. These values are lower than the ultimate failure strain (about 12000 µ-strains) of the composite
beam cured at 800C and tested statically. This lower failure strain is believed to have been due to the detrimental
effects of vibrations on the adhesive during the cure period.
The strain at failure (0.3%) of the UH-M CFRP composite cured at 800C is a lower strain value than the
axial tensile failure strain of the composite (0.4%); this result is consistent with other tests, (Photiou, et al 2006),

647
undertaken on upgraded UH-M CFRP steel beams. After failure of the UH-M CFRP composite the beam continues
along the non-load-displacement path of the non-linear path of the control steel beam. However, the early brittle
failure of these materials, albeit within the early period of the non-linear path of the steel beam, is not appropriate
for the construction industry. The H-M CFRP beams again failed by strain rupture of the CFRP but these beams
failed well into the non-linear region of the steel beams and at failure forced the steel beam into large deformations.

8 OBSERVATIONS
This paper has presented experimental results aimed at establishing the possibility of rehabilitating steel bridge
beams, whilst under vibration forces, with bonded hot-melt pre-impregnated high and ultra high modulus carbon
fiber composites. The composites are cured on site under a vacuum assisted pressure and at elevated temperature. It
should be emphasized that one layer of GFRP prepreg is placed between the CFRP prepreg and the steel member;
this facilitates a more uniform transfer of stress between the two materials. Preliminary conclusions are that:
(i) Provided that the bridge can be closed for 4 hours in a 24 hour period, and that the adhesive film and prepreg
composite can be cured on site at 800C, the bonded system, comprising of one layer of film adhesive and the H-M
composite, is the most efficient. The load sustained by the test beam with the adhesive film and prepreg composite
cured on site at 800C for four hours is 190 kN.
(ii) The beam strengthened with the adhesive film and the H-M composite system cured at a temperature of 650C
for 16 hours under vibrations induced by loads of 5kN-40 kN at 1 Hz. is able to sustain a total load 181 kN.
(iii) With the same manufacturing procedure as in item (ii) but with the upper value of the applied vibration load
increased to 60 kN, the failure load of the strengthened beam decreases by 11% of that in item (ii).
(iv) Due to the low strain to failure of the UH-M CFRP, the failure load of the strengthened beam is 138 kN and is
the lowest of the four samples tested. However, it should be noticed that in this case, the thickness of UH-M CFRP
prepreg is only half of that used for the H-M CFRP prepreg.
(v) For the rehabilitation system, a most satisfactory upgrade can be achieved for steel members using the H-M
CFRP prepreg and adhesive film.

9 REFERENCES
Tavakkolizadeh, M. and Saadatmanesh, H. (2003). ‘Strengthening of Steel-concrete Composite Girders Using
Carbon Fiber reinforced Polymers Sheets’. Journal of Structural Engineering, ASCE Vol129 No.1 , pp186-196.

Luke, S. (2001), The use of carbon fibre plates for the strengthening of two metallic bridges of an historic nature in
the UK. Proc. CICE 2001 FRP Composites in Civil Engineering, Ed. J-G Teng, Pub. Elsevier, London, pp 975-983.

Cadei, J.M., Stratford, T.K., Hollaway, L.C., and Duckett, W.G., (2004), CIRIA Report C595 ‘Strengthening
Metallic Structures Using Externally Bonded Fibre-Reinforced Polymers’, London.

Garden, H. and Shahidi, E., (2002). ‘The Use of Advanced Composite Laminates As Structural Reinforcement In A
Historic Building’, In: Advanced Polymer Composites For Structural Applications in Construction, R A Shenoi, S S
J Moy, L C Hollaway. (Eds), Thomas Telford Publishing, 2002, pp.457-465.

Photiou, N.K., Hollaway, L.C. and Chryssanthopoulos, M.K. (2006) ‘Selection of CFRP systems for steelwork
upgrading’. Journal of Materials in Civil Engineering, ASCE, In press.

Photiou, N.K., Hollaway, L.C. and Chryssanthopoulos, M.K. (2006) ‘Strengthening of an artificially degraded steel
beam utilizing a carbon/glass CFRP plate’ Concreuction and Building Materials Vol. 20 No.1, Feb/March 2006,
pp11-21

ACKNOWLEDGEMENTS
The authors gratefully acknowledge the financial support provided through a Kan Tong Po Visiting professorship jointly funded
by a Kan Tong Po grant, the Royal Society and The Hong Kong Polytechnic University and through grants from The Hong Kong
Polytechnic University (Project codes: BBZH and G-YX47. (Project codes: BBZH and G-YX47).
The authors also gratefully acknowledge the support received from Mr. E. Shahidi and the staff of Advanced
Composite Group Ltd., Heanor, Derbyshire, UK, especially for the helpful discussions throughout these investigations.

648
Third International Conference on FRP Composites in Civil Engineering (CICE 2006)
December 13-15 2006, Miami, Florida, USA

STRUCTURAL STRENGTHENING WITH EXTERNALLY BONDED

SCRP (STEEL CORD REINFORCED POLYMER)
Wine Figeys, Luc Schueremans, Dionys Van Gemert
(PhD student, Prof. Dr., Prof. Dr., Civil Engineering Department, Katholieke Universiteit Leuven
Kasteelpark Arenberg40, B-3001 Heverlee, Belgium)

Kris Brosens
(project engineer, Triconsult NV, Industriepark 1241/Bus 4, B-3545 Halen, Belgium)

ABSTRACT
One of the strengthening techniques is epoxy bonded external reinforcement. The additional reinforcement often
exists of steel plates or carbon fiber reinforced polymer (CFRP). Steel plates have a low material cost but are heavy.
On the contrary, CFRP is a lightweight, high strength flexible composite which can be easily applied. Steel plates
are used for deformation problems and CFRP when the high strength or the flexibility of the fibers can be exploited.
Steel cord reinforced polymer (SCRP) is a new material that consists of thin, uni-directional high-strength steel
fibers. This innovative composite is intended to combine the advantages of steel plates and CFRP: the material cost
can be relatively low compared to CFRP. Moreover, it can be flexible and SCRP can be as strong as CFRP. In this
paper, the search for a prototype of SCRP is presented. Different types of SCRP are investigated to develop an
optimal type of SCRP for the application as externally bonded shear reinforcement. The feasibility requirements,
material lay-out, test set up and results and the resulting SCRP choice are outlined.

KEYWORDS
Steel cord reinforced polymer (SCRP), prototype, strengthening

1. INTRODUCTION
The use of externally bonded reinforcement is a common strengthening technique for civil engineering constructions
nowadays. By adding additional reinforcement, it is possible to increase the capacity in bending or in shear and to
enhance flexural stiffness (Brosens, 2001). Today, mostly steel plates and CFRP (carbon fibre reinforced polymer)
sheets and laminates are applied as external reinforcement (Teng, 2001). Steel plates have a high stiffness and a low
material cost compared to CFRP. On the other hand, CFRP is more flexible and easily applicable, unlike steel plates.
CFRP is about 5 times stronger than standard steel. Each of these materials has its specific applications. A growing
interest concerns the development of steel cord reinforced polymer (SCRP), a new material that combines the
advantages of steel plates and CFRP as external reinforcement (Casadei, P., et al., 2005; Prota, A., et al. 2005). In
cooperation with the Belgian company Bekaert, this composite is developed. SCRP consists of thin high-strength
steel fibres. These fibres are bundled in cords and woven into unidirectional sheets. This new composite is intended
to be a lower cost alternative for CFRP. The sheets have a high strength as well as a high stiffness. The composite
remains quite flexible which makes SCRP easy to apply (Figeys, 2004). Other requirements are discussed below.
These requirements are used as a guideline to develop a prototype of SCRP.

2. FEASIBILTY
Requirements for SCRP concern tensile strength and stiffness. Practical experience gives some extra requirements
needed for an optimal prototype. The overall requirements are:
• Evaluation of the material properties, such as tensile strength and Young’s modulus. Knowledge of these
properties is necessary for calculation of the required cross-section of the external reinforcement.

649
 • Flexibility. SCRP seeks an easy application. SCRP will be delivered on roll of 1m on site. If used as
external shear reinforcement, even a higher foldability is necessary to wrap rectangular beams without any
preparation and kept in place without special arrangements.
• Sufficient impregnation. Glue has to surround all fibers, so that force can be continuously transferred to
the concrete. Together with a good impregnation, SCRP needs to stick to the concrete without extra
auxiliary actions.
• Bond strength. Failure has to take place in the concrete. Concrete should be the weak link in the
connection.
• Rust-proof. Steel plates need a special protective treatment for corrosion. It is not required for CFRP. All
types of SCRP fibers have a protective layer of zinc or brass to protect the fibers from corrosion.
• Bond behaviour. The use of SCRP as externally bonded reinforcement requires the design of the
anchorage length and the transferable force. Results of this study are presented in (Figeys, W., et. al,
2005).

2.1 Material properties

In a feasibility study, different kind of SCRP are studied, figure 1. The first type of SCRP consists of 65 steel cords.
Each cord contains 19 filaments with a diameter of 0.22mm up to 0.25 mm. This type is presented as type 1. The
compositions of the other types are mentioned in figure 1. Type 4 consists of single filaments, woven in a sheet with
a synthetic textile and is coated with brass.

65x[1x0.25+18x 0.22] 65x [7x4x 0.12] 58x 2x [3x3x 0.18] 65x 2x [0.30]

1 2 3 4
58x 2x [3x3x 0.18] is a weave with 58 stitches of 2 cords, each cord consist of 3 twists with 3 filaments with a diameter of 0.18mm .

Figure 1: Different prototypes of SCRP

Figure 2: Tensile test on SCRP: Single cord – impregnated SCRP – not-impregnated SCRP

Tensile strength Young s-modulus Type 3

3500 cord not-impr.250
sheet impr. sheet not-impr sheet
2500
impr sheet
Young s-mod. [kN/mm²]

2000
σ [N/mm²]

σ (N/mm²)

3000 200
1500

2500 150
1000

500

2000 100 0
1 2 3 4 1 2 3 4 0 0.005 0.01 0.015 0.02 0.025 0.03 0.035
type type strain (-)

Figure 3: Test results of tensile tests on SCRP

Ultimate tensile stress – Young’s modulus – Measured stress-strain behavior

650
The material properties are tested by means of tensile tests. The individual cords, the laminate itself and an
impregnated laminate are tested. The tensile tests are deformation controlled, figure 2. From the test results, the
tensile strength and the Young’s modulus are calculated. Test results are presented in figure 3. The tensile strength is
independent from the test set up. The tensile strength varies between 2000 and 3000 MPa. A correlation can be
found between the Young’s modulus and the test set-up. For all types of SCRP, a lower Young’s modulus is
measured for not-impregnated SCRP. Possibly, the different cords are not perfectly parallel which causes a different
stress level in the cords. By failure, the cords break one after the other. Cords within a pre-impregnated SCRP are
subjected to a more uniformly distributed stress level. When reaching the tensile capacity, all cords break at the
same moment. A ductile behavior is found, the high strength steel cords yield before failure. Because of this ductile
behavior, lower safety factors can be used in the design.

3.2 Impregnation, flexibility and bond strength

Impregnation with a viscous adhesive, Epicol U of Resiplast, is tested by means of pull-off tests on SCRP type 1,
glued on a concrete surface. All test samples have failed in the concrete, not in the adhesive or in the connection
between cylinder and SCRP or between SCRP and concrete. But in the laboratory, the work is more accurately done
than on the site. Therefore, enhanced impregnation is necessary. Also the flexibility is investigated. Calculations
show that type 1 of SCRP is more than 100 times stiffer than CFRP. A decrease in stiffness is necessary when SCRP
is used for wrapping. This can be reached if fewer filaments belong to a cord, if smaller diameters are used, or if the
filaments are used single. From these conclusions, the other types 2, 3 and 4 of SCRP are developed, figure 1.
Laminates 2 and 3 are much more flexible because of the smaller diameter. Type 4 is composed of single filaments.
These new types are much more flexible, but with a loss of steel section. The impregnation is improved compared to
type 1. However, the development of a new adhesive is necessary for a sufficient impregnation on site. Another
option is the use of pre-impregnated laminates.

3.3 Prototype 5

Prototype 5 is derived from the second type of SCRP with an increased density. This higher density is reached
because of the smaller width of the filament (type 5: 65mm, type 2: 95mm). This prototype is flexible and has an
improved impregnation. Four filaments with a diameter of 0.12 mm are twisted together, 7 of these twist are bundled
into cords. 65 cords belong to one sheet with a width of 65 mm, figure 4. In the same figure, it is shown that this
SCRP can be delivered on roll with a diameter of only 150 mm.

Figure 4: SCRP type 5 - 65x [7x4x0.12] with increased density

4. CURVATURE
CFRP is used for wrapping of columns to enhance the compressive strength of the column and as shear
strengthening by wrapping of beams. However, preparation of the columns and beams is necessary for these
applications. CFRP can not take shear stresses so that it is only possible to wrap CFRP to rounded corners. The
radius of curvature has to amount at minimum 3 cm. The sharps concrete edges are rounded with epoxy mortar.
Steel can take shear stresses so that a smaller radius of curvature is possible. In figure 5, the test set up is presented
to measure the influence of the radius of curvature of a cord of type 2. A cord is put around cylinder A with a
variable diameter and around cylinder B. The cord is fixed at cylinder C. The tests are deformation controlled for a
diameter of cylinder A of 5 – 10 – 15 – 20 – 25 – 30 – 40 – 50 mm and a distance L of 500 mm. Using a larger
diameter, the tensile strength equals 2621 N/mm². This value is comparable with the tensile strength obtained in the

651
test described before, figure 2. Only at a very small diameter (Ø = 5 mm), the cord fails at a lower force: the
maximum tensile strength amounts 2147 N/mm². From the stress-strain diagram, it can be concluded that the cords
tested with a smaller diameter do not yield, however with preserving the same Young’s modulus. The material fails
in a brittle way. Because of the possibility of a small radius of curvature, it is not necessary to round the edges of the
concrete when SCRP is used as external reinforcement.

Tensile strength ifo curvature radius Tensile strength ifo curvature radius
2800

2700 2500

2600
2000
2500
σ (N/mm²)

σ (N/mm²)
1500 50 mm
2400 40 mm
30 mm
2300 1000 25 mm
2200 20 mm
500 15 mm
2100 10 mm
5 mm
2000 0
0 10 20 30 40 50 0 0.5 1 1.5 2 2.5
diameter (mm) strain (%)

Figure 5: test set up and results of the influence of the radius of curvature on the tensile strength

5. CONCLUSION AND FUTURE RESEARCH

SCRP is a new material that can combine the advantages of steel plates and CFRP. It combines a relatively low
material cost with a high strength and a flexible shape. For some applications (wrapping), it is necessary that SCRP
has a low flexural stiffness, a sufficient impregnation and bond strength. Four types of SCRP are developed and
tested. The sheets have high tensile strength and Young’s modulus. A prototype is chosen. It is shown that SCRP is
advantageous for wrapping: only a very small radius of curvature influences the tensile strength so that there is no
need for preparation by rounding the edges. Further research concerns the durability of SCRP and the influence of
SCRP on the anchorage of external reinforcement. The development of a new adhesive is necessary to improve
impregnation. Also, pre-impregnated laminates night be an interesting option.

6. ACKNOWLEDGEMENT
The authors would like to thank the Flemish Institute for Promotion of Scientific and Technological Research in the
Industry (IWT – Vlaams Instituut voor de Bevordering van Wetenschappelijk-Technologisch Onderzoek in de
Industrie) for its financial support and NV Bekaert SA , represented by dr. ir. W. Dekeyser and ir. J. Gallens, for the
collaboration and technical support.

7. REFERENCES
Brosens, K., (2001). Anchorage of externally bonded steel plates and CFRP laminates for strengthening of concrete
elements, doctoral thesis, Katholieke Universiteit Leuven.
Figeys, W. (2004). Strengthening of reinforced concrete structures with bandweave (in Dutch: Versterking van
gewapend beton met bandweefsel), Master of Science thesis, Katholieke Universiteit Leuven, 2004.
Figeys, W., Brosens K., et al. (2005). Strengthening of Concrete Structures using Steel Wire Reinforced Polymer”,
Proceedings of Fiber-Reinforced Polymer Reinforcement for Concrete Structures, FRPRCS 7, Kansas City, pg. 743-
762.
Casadei, P., Nanni, A., et al. (2005). “Performance of double-T prestressed concrete beams strengthened with steel
reinforced polymer ”, Proceedings of Fiber-Reinforced Polymer Reinforcement for Concrete Structures, FRPRCS 7,
Kansas City, pg. 763-778.
Prota, A., et al. (2005). Performance of shallow RC beams with externally bonded steel reinforced polymer, in:
Structural Journal, volume 103, nr. 2, pp. 163-170.
Teng, J.G.; Chen, J.F.(2001). FRP strengthened RC structures, John Wiley & Sons, Weinheim.

652
Third International Conference on FRP Composites in Civil Engineering (CICE 2006)
December 13-15 2006, Miami, Florida, USA

WORLDS FIRST BRIDGE WITH SPRAYED FRP

STRENGTHENING: REVISITED
N. Banthia
Professor and Distinguished University Scholar, The University of British Columbia, Canada

R. C. Tennyson
Chief Scientific Advisor, Fiber Optic Systems Technology Inc., Toronto, Ontario, Canada

A. Mufti
Professor and President of ISIS, The University of Manitoba, Winnipeg, Manitoba, Canada

ABSTRACT
An entirely new method of repair using Sprayed Fiber Reinforced Polymers (SFRP) has recently been developed at
the University of British Columbia. The technique consists of spraying polymer and short, randomly distributed
fibers concurrently on the surface of concrete to be repaired using a spray gun. A 2-dimensional random distribution
of fibers is obtained on the application surface. In 2001, the technique was successfully applied to Safe Bridge on
Vancouver Island for shear strengthening. The bridge was tested in 2001 before and after the application of spray
using a fully loaded truck and the effectiveness of the spray was demonstrated. In 2005, the bridge was tested again
to assess the durability of the SFRP coating. It was concluded that the SFRP coating has remained intact, and there
are no signs of deterioration or debonding.

KEYWORDS
Strengthening, Sprayed Fiber Reinforced Polymers, Bridge, Load Testing.

1. INTRODUCTION
The vast majority of research and applications in the context of FRP strengthening have involved the use of
laminates, plates or wraps bonded to the concrete surface (1-2). Recently, an entirely new method of repair using
Sprayed Fiber Reinforced Polymer (SFRP) coatings, has been developed at the University of British Columbia
(Banthia et al., 1996, 2000, 2002). The technique consists of
spraying polymer and short, randomly distributed fibers
concurrently on the surface of concrete to be repaired such that a
2-dimensional random distribution of fibers is obtained on the
application surface (Figure 1). In the process, the resin and
catalyst are fed separately into a spray gun, where they are
mixed and then sprayed as a single compound. Two strands of
roving are fed into a chopper unit mounted on top of the spray
gun, wherein they pass between a pair of rollers and get chopped
to a consistent length adjustable from 8 to 48 mm. Further
details of the spray process, its optimization, etc. are given
elsewhere (Banthia et al., 1996, 2000, 2002).
Figure 1: The Spray Process

653
2. REHABILITATION AND TESTING OF SAFE BRIDGE IN 2001

In 2001, the technique of sprayed composites was applied to the Safe Bridge on the Vancouver Island that needed
shear strengthening (Figure 2). The 7.2 m long single span bridge was built in 1955, and has 11 precast channel
beams each 0.35 m wide. The girders were cast from 35 MPa structural lightweight concrete and the average steel
strength was of 356 MPa. The 9 m wide bridge includes a sidewalk separated from traffic by a concrete curb (Figure
2b). Like many other bridges in British Columbia and Alberta, the girders of Safe Bridge are considered shear
deficient (CAN/CSA2000), and the purpose of applying the Sprayed FRP was to enhance their shear resistance. In
order to demonstrate the effectiveness of sprayed composites at enhancing shear resistance, girders from an identical
bridge that got replaced in 1999 after 50 years of service were tested in the laboratory (Banthia et al. 2002), Figure
3. Detailed results of the tests on full-scale girders are given in Refs.: (Banthia et al. 2002) and key results are given
in Figure 4. Notice in Figure 4 that the Sprayed GFRPs performed superior to the GFRP wraps, and such curves
formed the basis for the acceptance of SFRP retrofit technique for the Safe Bridge.

Figure 2a: The Safe Bridge, BC Figure 2b: Section of Safe Bridge

Figure 3: Laboratory Tests on Full-Scale Figure 4: Results of Tests on Full-Scale

Bridge Girders Bridge Girders (Figure 3) with GFRP

Before the application of the spray, the bridge was instrumented by placing electrical resistance gauges on rebars of
all girders. The bridge was then tested for its ‘before spray’ performance by using a 28 Ton dump truck (Figure 6).
A 10 mm thick GFRP spray was then applied using the spray equipment mounted on a truck (Figure 7). After the
spray, long gauge fiber optic sensors were installed on the bridge on selected girders (Figure 8), and the bridge was
load tested again using the same dump truck (Figure 6). Results of the static and the roll tests performed before and
after the placement of the spray are given in Table 1. Only the readings from the electrical resistance strain gauges
were acquired. Notice the overall effectiveness of the spray.

654
 Figure 5: Electrical Strain Gauges Figure 6: Load Test on the Bridge
on the Rebar Before and After Spray Application

Figure 7: GFRP Spray Application Figure 8: Long Gauge Fiber Optic

on the Bridge Sensors Installed on the Spray

Table 1: Comparison of Rebar Strains from Girder #6 (2001 Tests)

Property Before Spraying After Spraying Reduction

Max. Rebar Strain (Static Tests) 101.76 x 10-6 65.55 x 10-6 36%
Max. Rebar Strain ) (Roll Tests) 72.12 x 10-6 54.94 x 10-6 24%

3. LOAD TESTING OF SAFE BRIDGE IN 2005

In 2005, the Safe Bridge was tested again, but this time only the long gauge fiber optic sensors (FOSs) placed along
the lengths of the girders were read. The long-gauge sensors actually measure the average displacement over their
gage length Ls. Taking into account temperature effects, the total strain is given by,

∫ ε ( z )dz
Ls

ε s = (α + β )ΔΤ + 0 (1)
Ls
where εs = sensor strain, α = thermal coefficient of expansion for structure, β = thermal optic coefficient for
fiber sensor (~ 8 x 10 /°C), ΔT = T – T0, where T = temperature at time of measurement, T0 = temperature at the
6

time of installation, and z = axial co-ordinate along sensor defined by 0≤ z ≤ Ls. When monitoring bending strains, it
is possible to estimate the maximum bending strain over the gage length from simple beam bending models, as
described below.

655
If M(z) = general bending moment function at location ‘z’ and if the axial location of maximum bending strain can
be defined as ‘zm’, then the value ‘zm’ is calculated for our specific case by the condition:

dM ( z ) d 2 Μ (z )
= 0 , and <0 (2)
dz dz 2
The relationship between the sensor average bending strain εs(y) and the maximum bending strain for the particular
sensor location (y) relative to the neutral plane is given by
ε z (zm ) Ls Μ (zm )
= (3)
ε s ( y ) Ls
∫ Μ (z )dz
0

Thus for a beam simply supported at its ends and subjected to concentrated load at the center (as in our case), the
maximum bending strain is given by:

ε z (z m )
= 2.0
ε s (θ )
(4)

In other words, the maximum bending strain is 2.0 times the measured FOS strain. The FOS strains recorded on
Girder 5 (Figure 2b) are given in Table 2. Note that the strains are not too different from those recorded in 2001 and
hence there is no sign of FRP degradation or debonding.

Table 2: Rebar and FOS Strain (Girder #5)

Max. Rebar Strain from Electrical Resistance Strain Gages Est’d Max. FRP Surface Strain from FOS
Before Spray (2001) After Spray (2001) After Spray (2005)
-6
102 x 10-6 66* x 10 60** x 10-6
*Measured FOS strain of 33 x 10-6 concerted to maximum strain using Eqn. (4)
*Measured FOS strain of 30 x 10-6 concerted to maximum strain using Eqn. (4)

4. CONCLUDING REMARKS
A novel technique of rehabilitation and strengthening using sprayed fiber reinforced polymers is described along
with its successful application to Safe Bridge in British Columbia. Full-scale load tests in 2001 and 2005
demonstrated that there’s no degradation and debonding in the spray.

5. REFERENCES
Banthia, N., Yan, C. and Nandakumar, N. (1996). Proc. of ACMBS-II, CSCE, Montreal, QC, Canada, pp 537-545.
Banthia, N. and Boyd, A.J. (2000). Can. J. of Civil Engineering, Vol. 27, No. 5, pp 907-915.
Banthia, N., Nandakumar, N. and Boyd, A., (2002). Concrete International, ACI, Vol. 24, No. 11, pp. 47-52.
CAN/CSA-S6-00, (2000) Published by CSA International, Canada, Dec. 2000.
Meier, U., (1992) Structural Engineering International, Vol. 1, No. 12, 1992, pp 7-12.
Neale, K. W. and Labossiere, P. (1997). Japan Concrete Institute, 1, 1997, pp. 25-39.

656
Third International Conference on FRP Composites in Civil Engineering (CICE 2006)
December 13-15 2006, Miami, Florida, USA

ANALYSIS OF POST-TENSIONED CONCRETE ROAD BRIDGE BEAMS

STRENGTHENED BY CFRP TAPES USING FEM UNDER STATIC LOAD
Arkadiusz Mordak
(M.Sc., Ph.D. Student, Cracow Technical University, Cracow, Poland)

Zbigniew Manko
(Professor of Civil & Structural Engineering, Wroclaw University of Technology, Wroclaw, Poland)

ABSTRACT
The paper is presented the method and results of the FEM analysis and research, and also the main conclusions of a
one-span post-tensioned concrete road bridge damaged by 1997 big flood performed under static field load before
and after of its repair. There was a conception to make reinforcing by CFRP tapes glued on the bottom flanges of
main beams or by overlaying a new concrete deck layer. The aim of the conducted bridge repair was to increase its
load capacity to 300 kN (class C) according to the Polish Loads Standard (PN-85/S-10030). The results obtained for
four different load schemes to determining the behavior of the analyzed span structure, which allowed for an assess-
ment of the efficiency of the strengthening, as well as establishment of guidelines for future reference concerning
this type of maintenance in the engineering practice. The conclusions were drawn from the passed analysis and tests
can be helpful mostly for the assessment of behavior of such type of the bridge strengthening system by FRP tapes.

KEYWORDS
Bridge repair, FEM analysis, Static field load test, Post-tensioned concrete beam, CFRP tapes

1. INTRODUCTION
The subject of this study is the road bridge made of concrete spans with post-tensioned main beams over the Nysa
Klodzka River situated in Klodzko (Upper Silesia, Poland). This paper is shown the range and the way of the
conducted research and some of the results, given in the form of figures, acquired from measurements and FEM
calculations of different quantities, e.g. displacements and strains as well as views showing the structure before and
after strengthening by CFRP tapes. It also analysis of obtained results and main conclusions concerning testing
conducted on this stage are presented. The research was conducted at two different stages of repair works, that is,
before performing the main research on static field load tests (Manko, 2001), which was aimed to determine the
efficiency of the applied repairing methods. The bridge load capacity before its repair, determined in expertise, was
classified as class E, that is 150 kN in accordance with the Polish Loads Standard (PN-85/S-10030), mainly due to a
very poor technical condition of the load-carrying structure of bridge span, resulting mostly from transverse cracks
in the main beams. The main aim of the repair was to increase the object load capacity to class C (300 kN).
The aim of the conducted research was to determine the behavior of the span structure subjected to considerable
static loads at various load schemes (Manko and Mordak, 1999). The research allowed to find out on which elements
of the span load-carrying structure the biggest forces were exerted during the progress of repair works of the bridge.
The inspection of spans and analysis of the obtained results performed each time after the accomplishment of repairs
allow determining the influence of the same load on the quality and durability of this object in the process of
strengthening as well as the efficiency and purposefulness of this process.

2. BRIEF BRIDGE DESCRIPTION AND RANGE OF CONDUCTED RESEARCH

The tests were carried out on the one-span road bridge (Figure 1). The examined span consists of four main beams
integrated with the new reinforced concrete deck slab of B50 concrete class. The total width of the individual spans
is constant along the bridge length and it is 6.50 m (Figure 2). The effective length of the span is 30.60 m. The span

657
Schemat III
Scheme IV
is simple-supported and made from post-tensioned con-
Scheme III
crete beams of length, L, of 31.60 m and are integrated
Scheme II
with the RC deck slab over interior supports. The

2.10
Scheme I 1.20 bridge was designed to serve under the II class load
3.75 3.25 1.70
(150 kN) in accordance with the PN-66/B- 02015 (or D
T5 according to the actual Standard PN-85/S-10030).
T1 - T4 f '1 f '2 f '3 f '4 There are eight span interior crossbeams, all made as
f1 - f4
Nysa Kłodzka River concrete. The bridge supports are in the form of mas-
sive concrete pier and abutments on spread foundation,
fixed in a reinforced concrete footing. The foundation
rests directly on the virgin soil. The main beams of the
Figure 1: Side view of longitudinal section of road
spans rest on single-roller and fixed steel bearings
bridge in Klodzko and technical parameters of truck
(Figure 1). The roadway was covered with bituminous
type LIAZ (top view) and localization of inductive gages
pavement, 0.05 m thick, with incorporated insulation
(f1–f4 & f′1–f′4)) and extensometers (T1–T5) on the
of an average thickness 0.01 m and 0.02 m thick pro-
span length during static field load tests of bridge
tective concrete layer. The usable width of the bridge
amounts to 6.10 m which includes the 3.50 m wide roadway and a 1.30 m sidewalk on each side (Figure 2). The con-
sidering strengthening conception of the bridge span was accomplished by gluing the tapes made of carbon fibers
CFRP SikaDur M1214 type (two for each beam) to the bottom flanges of the main beams (Meier and Deuring, 1991).
The final results of the bridge acceptance inspection, conducted after the complete repair under the trial static and
dynamic load (Manko and Mordak, 1999), allowed for a comprehensive evaluation of the efficiency of the main
beams strengthening by applying CFRP tapes. Moreover, it enabled a comprehensive evaluation of the change of the
spans structures behavior under the same load during different stages of repair works.
During the bridge repairs, the research was conducted at two different stages (phases). Figure 1 shows the load
schemes on the tested span with the measurement points localization. The following quantities were made:
– four main beams deflections made by dial indicators with 1×10–5 m accuracy,
– vertical and horizontal displacements of the expansion and fixed bearings by dial indicators with 1×10–5 m accuracy,
– strains (indirectly – normal stresses) in the main beams, which were performed by strain gages (extensometers)
and mechanical indicators,
– strains in the CFRP in half and 1/4 of the effective span of the main beams, which were performed by strain gages.

3. FEM ANALYSIS RESULTS

The program COSMOS/M was used for computation. Finite element analysis was used to model the behavior nume-
rically to as to provide a valuable supplement to the field investigations, particularly in parametric studies. Modeling
the complex behavior of reinforced concrete, which is both nonhomogeneous and anisotropic, is a difficult challenge
in the finite element analysis of bridge engineering structures. Most early the finite element models of reinforced con-
crete included the effects of cracking based on a pre-defined crack pattern. With this approach, changes in the topo-
logy of the models were required as the load increased; therefore, the ease and speed of the analysis were limited.
Only recently we have attempted to simulate the behavior of reinforced concrete strengthened with FRP composites
(a) AT MIDSPAN OVER SUPPORT (b) using the finite element method on the basis of
Asphaltic concrete
Sidewalk slab
0.03 m
0.12 m
Asphaltic concrete
Asphaltic concrete
0.02 m
0.03 m
past experiences. A number of reinforced
Protective concrete 0.02 m Protective concrete 0.02 m concrete beams strengthened with FRP tapes
Insulation 0.01 m Insulation 0.01 m
Leveling concrete 0.03 m Leveling concrete 0.03 m were tested in the laboratory. Therefore, it was
Deck slab 0.12 m
6.50
Deck slab 0.12 m
decided to conduct own calculations on real
0.20
1.50
1.15 0.15
3.50
0.15
1.50
1.15 0.20
assumptions with 3D SOLID elements, from
the nonlinear contact elements of interface
0.34 0.02

1.5% +0.60
1%
1.00

type. The FRP tapes were modeled with 2D

0.36

0.02

0.08
0.00 1%
1.5% plate elements in that study, however, and
crack patterns of those beams were not
0.36
1.05
1.74

0.61 0.14
1.38

predicted by the finite element analysis. The

1.14

Interior Supporting
0.25 0.08

0.78
0.34

crossbeam - 0.14 crossbeam - 0.60

Proposal of two-dimensional plate elements are surface-
B1 B2 B3 CFRP reinforcement B4
0.61 0.30 1.26 0.40 0.56
like elements, which have no actual thickness.
1.36 0.30 1.31 0.40
0.76 1.66 1.66 1.66 0.76 Therefore, stress and strain results at the actual
6.50
surfaces of the CFRP tapes were estimated by
Figure 2: Cross-sections of post-tensioned concrete span at: theoretical calculations. Some examples of
(a) mispan (before repair), (b) support (after its renovation) FEM results are presented in Figures 3 and 4.

658
 (a) (b)

(c) (d)

Figure 3: Some results of the bridge span calculation (before repair): (a) vertical displacements, (b) strains
and (c) normal stress, in longitudinal direction, and (d) stresses according to von Mises criterion

4. CONCLUSIONS
The practical experience in the light of the conducted FEM calculations and research of the bridge span under the
static load during the road bridge construction (I stage of tests), comprehensive analysis of displacement and strain
(indirectly normal stress) load bearing structure results obtained from the measurements and comparison between
them and calculated values (Figures 3–4) allowed for the following conclusions of general and detailed character:
1. The span structure made of post-tensioned concrete beams did not raise any reservations as far as average sizes of
section forces, displacement and strain values obtained from research and were lower than calculated ones in
almost all examined points and span structure sections. The causes of small differences between the results
obtained both from calculations and measurements stay mainly from the calculations with assumed estimated
(a) (b)

(c) (d)

Figure 4: Some results of the bridge span calculation (after gluing CFRP tapes): (a) vertical displacements,
(b) strains and (c) normal stress, in longitudinal direction, and (d) stresses according to von Mises criterion

659
 Table 1: Results of the deflections at midspan of value of the span cross-section stiffness and cautious
four beams B1–B4 obtained from the research and good estimation of interaction of the beams with the plate
calculation (10–3m) before and after reinforcement deck and pavement layers of the roadway at particular
repair stages. They amount in range for deflections
Test stage Result nature B1 B2 B3 B4 18.74–34.32% (Table 1) and for strains 7.24–20.27%, it
measured 5.82 6.37 6.86 7.39 proves that the section has considerably higher span
Before repair
calculated 8.86 8.86 8.86 8.86 cross-section stiffness. The good interaction (interface
With CFRP calculated 8.69 8.69 8.69 8.69 type elements using) between the beams and the CFRP
measured 5.16 5.19 5.48 5.55 tapes can be caused such small differences. However, the
After repair
calculated 6.83 6.85 6.85 6.83 applied of strengthening of the main beams with CFRP
tapes did not bring about the significant changes in
deflections and strains values of main beams.
2. The strains and displacements of the prestressed concrete beams during the bridge construction demonstrated
basically elastic character. They were also lower than the expected values calculated theoretically and also the
limit values were not exceeded. It means that construction work was conducted on a high level of technical
quality and under constant control. As it was found during the research the minor displacements and permanent
strains of the span were partially the beams permanent displacements and most frequently originated partially
from the supports settlement and readings errors as well as measuring equipment errors (the change of air
temperature and humidity during the time of measurements). Only to a small rate, they were caused by the
permanent strains of the load bearing structure (less then 2% of total displacements). This shows a correctness of
assumptions taken for calculation and static-strength analysis of this span or also the correctness of assumed
analytical structure model with their real behavior in particular repairs phases.
3. The grid model of variable load-capacity structure that was assumed at the first step of calculation in different
repair phases in dependence of layers and strengthening tapes seems to be sufficient tool to determine the
deflections and strains in tested structures on the engineering level. For the detailed analysis of interaction
between particular pavement layers and structure components and the assumed strengthening manner is
necessary to use more complex model which should be better reflecting a real interaction in a such type of span
structures in the considered repair stages of bridge, especially on the contact section of concrete and CFRP tapes.
4. As the effect of executed calculations by the FEM and the experimental tests on the real object was affirmed, that
for the engineering aims the bridge structures analysis it is possible to carry out in the plane state of strains (the
two-dimensional 2D analysis) with the contact elements of the interface type between beams and tapes. In the
some special cases, the calculations were possible also to execute in the 3D space in aim of more detailed
analysis. The modeling of bituminous parameters as elastic-plastic is recommended or as elastic-plastic material
with reinforcement. Whereas the span as the bilinear elastic material is possible to analyzing (Kachlakev and at
al., 2001). The contact layers of interface type with non-linear proprieties should be considering between CFRP
tape and concrete beam.
The conclusions concerning the behavior of such structures can be of great practical significance. As the most
loaded span structure elements, which need a detailed study and analysis, one should concern the elements of the
bridge deck plate where stresses caused by their direct load with stresses due to their interaction with main beams
and crossbeams sum up. In the fact, above summary and main conclusions refer to structures of the tested span
elements of preset geometric characteristics, particular element stiffness, and determined effective spans. However,
it may be stated that spans strengthening constructed by lamels is not the best solution as far as this type of
structures is concerned, mostly from the economical point of view. In order to use an expensive CFRP tapes to a
higher extent, one should install on the beams already known prestressing devices for the CFRP tapes.

5. REFERENCES
Kachlakev, D., Miller, T., Yim, S., Chansawat, K., and Potisuk, T. (2001). “Finite Element Modeling of Reinforced
Concrete Structures Strengthened with FRP Laminates.” Final Report SPR no. 316 for the Oregon Department of
Transportation Research Group, California.
Manko, Z. (2001). “Investigation of Prestressed Road Bridge Span over the Nysa Klodzka River in Matejko Street in
Klodzko before its Modernization”, Proceedings of I Symposium on Bridge Diagnostics and Testing, April 4–6,
Opole, Poland, pp. 399-417.
Manko, Z., and Mordak, A. (1999). “Research of Road Bridge over Nysa Klodzka River along Matejko Street in
Klodzko after Repair.” Scientific-Research Center for the Development of Bridge Industry MOSTAR, Wrocław.
Meier, U., and Deuring, M. (1991). The Application of Fiber Composites in Bridge Repair. Strasse und Verkehr,
Vol. 9, pp. 7-11.

660
Third International Conference on FRP Composites in Civil Engineering (CICE 2006) 
December 13­15 2006, Miami, Florida, USA 

A REVIEW OF FRP­STRENGTHENED RC 
BEAM­COLUMN CONNECTIONS 
S.T. Smith 
(Senior Lecturer, Centre for Built Infrastructure Research, University of Technology Sydney, Australia) 

R. Shrestha 
(PhD candidate, Centre for Built Infrastructure Research, University of Technology Sydney, Australia) 

ABSTRACT 
Considerable  research  has  been  conducted  over  the  last  decade  or  so  on  the  strengthening  or  repair  of  existing 
reinforced  concrete  (RC)  structural  elements  such  as  beams,  columns  and  slabs  with  externally  bonded  fibre 
reinforced polymer (FRP) composites. Very little research, by comparison, has been conducted on the strengthening 
of  RC  beam­column  connections  with  FRP,  with  the  majority  scattered  in  various  journals  and  conference 
proceedings.  Fewer  analytical  studies  have  been  undertaken  and  design  recommendations  proposed.  This  paper 
provides a concise but systematic review of experimental research on the strengthening of RC connections with FRP 
in addition to an evaluation of the effectiveness of the strengthening schemes. 

KEYWORDS 
Beam­Column Connections, External Bonding, FRP, Reinforced Concrete, Strengthening 

1. INTRODUCTION 
Extensive  experimental,  analytical  and  numerical  studies  have  been  conducted  on  the  retrofit/repair/strengthening 
(herein referred to as strengthening unless noted otherwise) of reinforced concrete (RC) structural elements such as 
beams, columns and slabs with externally bonded FRP composites, and a comprehensive review is given in Teng et 
al. (2002). Surprisingly, very  little  research  by comparison  has  been  conducted  on the strengthening of  RC beam­ 
column connections with FRP  (the region where  the  beam  frames  into the  column is  referred to as  the joint while 
connection refers to the joint region including with the beam/s and columns framing into it). 

The  need  to  strengthen  and/or  enhance  the  ductility  of  connections  stems  from  gravity  load  designed  frames  of 
yesteryear  being  inherently  weak  within  the  connection  region  when  subjected  to  seismic  attack.  The  two  most 
commonly occurring connection strengthening needs are shown in Figure 1 that are typical of both of external (“T”) 
and  internal (“+”)  connections  in gravity  load  designed  frames. In Figure 1a,  a lack  of transverse  reinforcement  in 
the form of horizontal steel ties in the joint region may mean the shear strength of the connection is unable to resist 
increased shear demand from seismic loading. Under seismic actions causing load reversal, anchorage failure of the 
bottom bars in the beam in Figure 1b may occur. Bottom bar failure, or pull­out, will lead to joint degradation and 
brittle failure in the beam at the face of the column. 
Column 
Transverse Reinforcement 

Beam 
Insufficient Transverse Reinforcement  Insufficient Anchorage 

(a) Insufficient transverse reinforcement  (b) Insufficient beam bottom steel anchorage 
Figure 1: Typical external connection steel reinforcing details

661
Plain  RC  connections  have  been  the  subject  of  extensive  experimental  investigations,  however,  research  on 
strengthening  connections  with  FRP  is  much  less.  Prior  to  FRP,  corrugated  steel  jacketing  or  fibre  reinforced 
concrete  coating  have  been  used  to  strengthen  connections.  A  comprehensive  review  of  non­FRP  strengthening 
solutions, as well as some FRP ones, is given in Engindeniz et al. (2005). The limited research on FRP strengthened 
connections is possibly due to the complex  behavior  of  the connection, as  well as  difficulties in strengthening  and 
complicated  testing  arrangements.  This  paper  is  a  review  of  the  limited  experimental  research  available  on 
strengthening connections with FRP and an evaluation of the effectiveness of the various strengthening schemes. 

2. STRENGTHENING SCHEMES 
Four  different  types  of  deficiencies  have  been  introduced  into  connections  that  require  strengthening  with  FRP. 
They  are  (1)  insufficient  transverse  reinforcement  in  the  joint  region  resulting  in  a  deficiency  of  shear  strength 
(Shear Strengthening), (2)  insufficient anchorage of  the longitudinal beam  bottom reinforcement  causing bar pull­ 
out  under  load  reversal  (Anchorage  Strengthening),  (3)  combination  of  inadequate  transverse  reinforcement  and 
reinforcing bar anchorage (Shear and Anchorage Strengthening), and (4) formation of the plastic hinge in the beam 
too close to the joint region (Plastic Hinge Relocation). 

Research has focused on testing two­dimensional connections, such as “­”, “T” and “+” shaped connections with in­ 
plane loading. T­shaped connections have been more commonly tested as they are more susceptible to damage than 
“+”  connections.  Various  strengthening  schemes  involving  the  application  of  externally  bonded  FRP  in  the 
connection  region  and  adjacent  beams  and  columns  have  been  investigated,  with  connections  being  primarily 
subjected  to cyclic loading  of typically  less  than  twenty push­pull cycles of  increasing amplitude  although limited 
investigations have applied monotonic loading. The only connections subjected to monotonic loading were those of 
Granata  and Parvin  (2001) and Li  et  al.  (1999).  Table  1  is  a  summary  and  categorization of the  various types  of 
connections tested, test layout, test criteria and FRP layout following an extensive review of the literature. 

3. EFFECTIVENESS OF STRENGTHENING 

The  plain  RC  connections  mainly  failed  in  shear  within  the  connection  region  (due  to  insufficient  lateral 
reinforcement in the joint), anchorage  failure of  the beam  longitudinal reinforcement, or plastic beam hinging. The 
FRP­strengthened connections typically failed by debonding or rupture of the FRP or the formation of plastic hinges 
generally  in  the  beam  outside  of  the  FRP­strengthened  region  as  the  strong­column  weak­beam  principle  was 
generally adhered to. Debonding occurred for unanchored FRP. Anchored FRP ruptured where anchorage involved 
mechanical bolting or wrapping FRP around the beam or column adjacent to the connection region. 

Comparison of the ultimate load and maximum deflection to that of the control specimens from each study, as given 
in Figure 2 (based on available data), is one way to identify effective strengthening schemes. Energy dissipation  is 
another  indicator  of  strengthening  effectiveness,  however,  limited  published  results  prevented  this  quantity  from 
being  effectively  compared.  Based  on  the  results  of  Figure  2  the  following  comments  can  be  made.  For  shear 
strengthening, the  anchored  schemes of Gergely et  al. (2000) (ref.  B),  Antonopoulos and Triantafillou (2003)  (ref. 
C),  and  Tsonos  and  Styliandis  (2002)  (ref.  D)  were  the  most  effective  primarily  because  the  FRP  did  not  fail 
prematurely  by  debonding.  For  anchorage  strengthening,  the  anchored  schemes  of  Prota  et  al.  (2001)  (ref.  J), 
Granata and Parvin (2001) (ref. K),  and Mukherjee  and Joshi (2004) (ref.  L) were the  most  effective,  again as the 
anchorage prevented FRP debonding. Both of the shear and anchorage schemes were effective although they appear 
difficult  to  implement  in  practice, while the  plastic  hinge  relocation  scheme  and  the  ‘other’  strengthening scheme 
were  not  as  effective  as  others.  All  strengthening  schemes  varied  in  complexity  from  simple  relatively  intuitive 
schemes to quite complicated ones. The degree of simplicity or complexity however does not necessarily reflect on 
the effectiveness of the scheme. 

4. CONCLUSIONS 
This paper has provided a review on and systematic characterisation of the strengthening of connections with FRP. 
Two  main  deficiencies  of  existing  connections  were  identified,  namely  inadequate  joint  shear  strength  and 
inadequate anchorage of the bottom steel reinforcing bars for gravity load designed frames. The majority of research 
conducted  to  date  has  been  experimental  with  less  research  on  analytical  modelling  or  development  of  design 
guidelines.  Research  efforts  now  need  to  be  focused  on  the  development  of  analytical  models  to  calculate  the 
connection shear strength for FRP strengthened connections that consider FRP rupture and debonding.

662
 Table 1: Summary of test connections and strengthening schemes 

Source 1  Test Layout 2  Test Criteria  FRP Layout 3 

Shear Strengthening 
Effectiveness of FRP and 
A T 
anchorage scheme 
≠ 
Surface preparation, curing 
B T  temperature and fibre 
# 
orientation 
Effectiveness of FRP and 
C T 
anchorage schemes  ≠  # 

D T  Effect of pre­cracking 

Vary connection transverse 
E + 
reinforcement ratios 
Effectiveness of FRP and 
F + 
anchorage scheme  # 

Effectiveness of FRP and 
G + , H + 
anchorage schemes  #  # 
Anchorage Strengthening 
Effectiveness of FRP and 
I ­  #  ≠  ≠  ≠ 
anchorage schemes 
Effectiveness of NSM bars 
J + 
and FRP sheets 
#  #  # 

Effectiveness of FRP and 
K T 
anchorage schemes 
#  # 

Effectiveness of FRP and 
L + 
anchorage schemes 
#  # 
Shear and Anchorage Strengthening 
Effectiveness of FRP and 
M T 
anchorage schemes 
≠  ≠ 

Effectiveness of FRP and 
N T 
anchorage schemes 
≠  ≠  ≠  ≠ 

Plastic Hinge Relocation 

O T  Effectiveness of FRP scheme 

Other (No Connection Deficiency) 

P +  Effectiveness of FRP scheme 

Reference: A = Ghoborah and Said (2001); B = Gergely et al. (2000); C = Antonopoulos and Triantafillou (2003); D = Tsonos and Stylianidis 
(2002); E = Ouyang et al. (2003); F = D’Ayala et al. (2003); G = Mosallam (2000); H = Mosallam (2001); I = Geng et al. (1998); J = Prota et 
al. (2001); K = Granata and Parvin (2001); L = Mukherjee and Joshi (2004); M = El­Amoury and Ghobarah (2002); N = Ghobarah and El­ 
Amoury (2005); O = Mahini et al. (2004); P = Li et al. (1999) 
1 
‘­’ shaped connections, ‘T’ shaped connections, ‘+’ shaped connections 
2 
Beam,  Column,  Point of load application,  Support 
3 
≠ mechanical anchorage  # anchorage provided by FRP wrapping,

663
 3.0  4.0 

Strengthened / Control 
Strengthened / Control  2.5 
3.5 
3.0 
2.0  2.5 
1.5  2.0 
1.5 
1.0 
1.0 
0.5  0.5 
0.0  0.0 
0  A  B  C  D  E   F  G  H  J   K  L  M N  O  P 
1  2  3  4  5  6  7  8  9  10 11 12 13 14 15 16  0  A    D   E    F   G    H    J    L    M   N   O 
1  2  3  4  5  6  7  8  9  10  11  12 
Source  Source 
(a) Ultimate Load  (b) Maximum Displacement 

Figure 2: Comparison of strengthening effectiveness (strengthening categories differentiated) 

5. ACKNOWLEDGEMENTS 
This  project  was  funded  by  Australian  Research  Council  (ARC)  Discovery  Grant  DP0559567.  The  financial 
assistance of the ARC is gratefully acknowledged. Professor J.G. Teng from the Hong Kong Polytechnic University, 
Hong Kong is thanked for his technical input. 

6. REFERENCES 
Antonopoulos,  C.P.  and  Triantafillou,  C.  (2003).  “Experimental  investigation  of  FRP­strengthened  RC  beam­column 
joints”. Journal of Composites for Construction, ASCE, Vol. 7, No. 1, pp. 39­49. 
D’Ayala,  D.,  Penford,  A.  and  Valentini,  S.  (2003). “Use  of  FRP  fabric  for  strengthening  of  reinforced  concrete  beam­ 
column joints”. Proc. (CD­ROM), 9 th  Int. Conf. and Exhibition on Structural Faults and Repair, UK, 4­6 July. 
El­Amoury, T. and Ghobarah, A. (2002). “Seismic rehabilitation of beam­column joint using GFRP sheets”. Engineering 
Structures, Vol. 24, pp. 1397­1407. 
Engindeniz,  M.,  Kahn,  L.F.  and  Zureick,  A.H.  (2005).  “Repair  and  strengthening  of  reinforced  concrete  beam­column 
joints: state of the art”. ACI Structural Journal, Vol. 102, No. 2, pp. 1­14. 
Geng,  Z.J.,  Chajes, M.J.,  Chou,  T.W. and Pan, D.Y.C.  (1998). “The  retrofitting  of  reinforced  concrete  column  to beam 
connections”. Composites Science and Technology, Vol. 58, pp. 1297­1305. 
Gergely,  J.,  Pantalides,  C.P.  and  Reavely,  L.  (2000).  “Shear  strengthening  of  RCT­joints  using  CFRP  composites”. 
Journal of Composites for Construction, ASCE, Vol. 4, No. 2, pp. 56­64. 
Ghobarah, A.  and  El­Amoury,  T. (2005).  “Seismic  rehabilitation  of deficient exterior  concrete  frame  joints”. Journal of 
Composites for Construction, Vol. 9, No. 5, pp. 408­416. 
Ghobarah,  A.  and  Said,  A.  (2001).  “Seismic  rehabilitation  of  beam­column  joints  using  FRP  laminates”.  Journal  of 
Earthquake Engineering, Vol. 5, No. 1, pp. 113­129. 
Granata,  P.J.  and  Parvin,  A.  (2001).  “An  experimental  study  on  Kevlar  strengthening  of  beam­column  connections”. 
Composites: Part B, Vol. 53, No. 2, pp. 163­171. 
Li, J., Bakoss, S.L, Samali, B. and Ye, L. (1999). “Reinforcement of concrete beam­column connections with hybrid FRP 
sheet”. Composite Structures, Vol. 47, pp. 805­812 
Mahini, S.S., Ronagh, H.R. and Smith, S.T. (2004). “CFRP­retrofitted RC exterior beam­column connections under cyclic 
nd 
loads”. Proc., 2  Int. Conf. on FRP Composites in Civil Engineering, Australia, 8­10 December, pp. 647­652. 
Mosallam, A.S. (2000). “Strength and ductility of reinforced concrete moment frame connections strengthened with quasi­ 
isotropic laminates”. Composites: Part B, Vol. 31, pp. 481­497. 
Mosallam, A.S. (2001). “Seismic repair and retrofit of reinforced concrete moment joints using composites”. Proc. (CD­ 
th 
ROM), 9  Int. Conf. and Exhibition on Structural Faults and Repair, UK, 4­6 July. 
Mukherjee, A. and Joshi, M. (2004). “FRPC reinforced concrete beam­column joints under cyclic excitation”. Composite 
Structures, Vol. 70, pp. 185­199. 
Ouyang, Y., Gu, X.L., Huang, Y.H. and Qian, Z.Z. (2003). “Seismic behavior of reinforced concrete beam­column joint 
th 
strengthened with GFRP”, Proc. 6  Int. Symp. on FRP for RC Structures, Singapore, 8­10 July, pp. 1107­1116. 
Prota, A., Nanni, A., Manfredi, G. and Cosenza, E. (2001). “Selective upgrade of beam­column joints with composites”. 
Proc. Int. Conf. on FRP Composites in Civil Engineering, Hong Kong, 12­14 December, pp. 919­926. 
Teng, J.G., Chen, J.F., Smith, S.T. and Lam, L. (2002). FRP­Strengthened RC Structures. John Wiley & Sons, UK. 
Tsonos,  A.G.  and  Stylianidis,  K.  (2002).  “Seismic  retrofit  of  beam­to­column  joints  with  high  strength  fibre  jackets”. 
European Earthquake Engineering, Vol. 16, No. 2, pp. 56­72.

664
Third International Conference on FRP Composites in Civil Engineering (CICE 2006)
December 13-15 2006, Miami, Florida, USA

ANCHORAGE OF CARBON FIBER REINFORCED POLYMER SHEETS

WITH AND WITHOUT HEIGHT TRANSITION
Sarah L. Orton
(PhD Candidate, University of Texas at Austin, USA)

James O. Jirsa
(Professor, University of Texas at Austin, USA)

Oguzhan Bayrak
(Assisstant Professor, University of Texas at Austin, USA)

ABSTRACT
Debonding of CFRP sheets has been a serious limitation in the use of CFRP on reinforced concrete structures.
Debonding is further magnified in cases where the CFRP must go through a height transition between surfaces of
elements that are not in the same plane. Forty specimens were tested to investigate the use of carbon fiber anchors
(anchors inserted into predrilled holes and fanned out over the CFRP sheet) and U-wraps (CFRP sheet wrapped
around sides of beam) to develop the full rupture strength of a CFRP sheet with and without a height transition.
Without any additional anchorage the CFRP sheet debonded at 40% of its capacity. Although the number and size
of anchors is important, two rows of anchors, with the cross-sectional area of CFRP in each row equal to or greater
than the area of the CFRP sheet, fully anchored the CFRP sheet. An equivalent anchorage using U-wraps required 5
times more CFRP. A height transition with a 1 to 2 transition slope reached only 40% of the CFRP capacity, while a
1 to 4 slope reached full capacity. Finally, surface preparation was unimportant when the CFRP was well anchored.

KEYWORDS
Carbon Fiber Reinforced Polymer Sheets, Anchorage, Carbon Fiber Anchors, Bonding

1. INTRODUCTION

CFRP sheets can be used to increase the capacity of reinforced concrete (RC) members (Teng et al., 2001).
However, the capacity of the retrofitted member is limited by debonding of the CFRP sheets from the concrete. A
study of behavioral trends of FRP retrofitted RC members found that FRP sheets debonded on average at 50% of
their tensile capacity (Bonacci and Maalej, 2001). The bonding strength of CFRP is limited by the strength of the
concrete because the debonding initiates from shear cracks within the concrete at a level just below the CFRP sheet
or at the level of the internal reinforcement. Increasing the length of the CFRP sheet bonded to the concrete does
not increase the bonding strength because there is an effective length beyond which bonding strength does not
increase (Teng et al., 2001). Therefore, in order to more fully utilize the tensile capacity of the CFRP sheet some
form of additional anchorage is needed.

Several different types of anchorage schemes have previously been studied, though a design methodology is still
lacking (Ozdemir and Akyuz, 2005; Bramblett, 2000). In the series of tests presented in this paper, two of the most
effective and easy to install anchorages, CFRP U-Wraps and CFRP anchors (Figure 1), are evaluated. Furthermore,
a height transition (offset in the surface level of the CFRP ) will accentuate the debonding problem of CFRP sheets.
As illustrated in Figure 2, a height transition can occur when providing continuity of reinforcement that may not be
continuous through a beam column joint using CFRP or when increasing the capacity of an infill shear wall whose
width is less than that on the surrounding frame (Saatcioglu et al., 2005). The research presented in this paper
provides data to develop a design methodology to fully utilize the capacity of the CFRP sheet with or without a
height transition. The project was funded by the National Science Foundation.

Height Transition

665
 Ends of anchor
fanned out
U-Wrap
Anchor

Anchor depth

Concrete

Figure 1: CFRP U-wrap and Anchor

Column Frame
Height Transition
Beam
Height Transition CFRP
Infill Anchor

CFRP CFRP
CFRP Anchor
a) b)

Figure 2: Use of CFRP at Height Transitions:

a) Provide Continuity of Reinforcement b) Increase Capacity of Infill Shear Wall

2. TEST SETUP AND PROGRAM

The specimen and test setup were designed to allow for a controlled evaluation of the anchorage of the CFRP sheet
with and without a height transition (Figure 3). The test setup was designed to simulate a reinforced concrete beam
in bending with a preexisting crack at midspan. The test specimens consisted of two blocks of reinforced concrete,
measuring 16” plus transition height high, 32” long and 8” wide, connected only by a CFRP sheet. This created the
crack to initiate debonding at a controlled point and ensured that the only tensile resistance is due to the CFRP sheet.
The connected blocks were then loaded at midspan and simply restrained at each end thereby putting tension into the
CFRP sheet. The transition slope was constructed using a polymer cement repair mortar with a bonding strength
greater than the concrete substrate and suitable for overhead applications. Parameters varied included the slope of
the transition (none, 1 to 2, 1 to 4), height of transition (0”, 1”, 2”, 3”), anchorage of CFRP sheet (Anchors or U-
wraps – size, number, spacing), surface preparation (grinding, sandblasting, none), and type of carbon fiber
(unidirectional Tyfo® SCH-35 or SCH 41).

CFRP Anchors CFRP Sheet

Transition
Concrete slope
Blocks

Ram

Figure 3: Test Setup

666
3. TEST RESULTS

3.1 Without Height Transition

In the first series of tests, presented in Table 1, the capacity of the anchorage system without a height transition
between the blocks was studied. The results of tests 00-ng1 and 00-ns1 showed that without additional anchorage
the CFRP sheet debonded at about 40% of its ultimate capacity (30 k). With additional anchorage the CFRP sheet
was able to reach its ultimate capacity. A single layer of U-wrap only reached 75% of capacity with the longitudinal
sheet debonding and the U-wrap failing in shear (00-us1). A double layer U-wrap provided adequate shear
resistance and reached 99% of capacity (00-us2), but required 5 times more area of CFRP than the unanchored case.

Table 1: Test Results

Tension in Tension in Area of

Height
Test # Slope Type of Fabric Anchorage CFRP at CFRP (% of CFRP Failure Mode
difference
Failure (kip) ultimate) (ft2)
Without Height Transition

00-ng1 none 0" SCH-35 none 11.88 40% 2.00 Debonding

00-ns1 none 0" SCH-35 none 10.56 35% 2.00 Debonding

U wraps 6" wide at 5" debond of flat FRP,
00-us1 none 0" SCH-35 and 19" 22.36 75% 6.67 shear of U wrap
Double U wrap 6" wide
00-us2 none 0" SCH-35 at 5" and 19" 29.58 99% 11.33 Fracture
SCH-35 sheet 5/8" anchor at 5" and Fracture anchor,
00-2s1 none 0" SCH-41 anchors 19" 21.39 71% 3.50 partly deliminate
2 3/8" anchor at 5"and Fracture of 4.5" width,
00-4s1 none 0" SCH-35 at 19" 21.16 71% 3.00 peeling of 1.5"
SCH-35 sheet Two 1/2" anchors at 5"
00-4s2 none 0" SCH-41 anchors and 19" 25.14 84% 3.50 Fracture
SCH-35 sheet 2 < 5/8" anchors at 5"
00-4s3 none 0" SCH-41 anchors and 19" 31.41 105% 4.25 Fracture
3 3/8"anchors at 5" and
00-6s1 none 0" SCH-35 at 19" 31.94 106% 3.50 Fracture
With Height Transition

22-ng1 1:2 2" SCH-35 none 5.98 20% 2.67 peeling

22-us1 1:2 2" SCH-35 U-wrap, 6" wide 12.94 43% 3.83 Fracture
SCH-35 sheet 3 3/8" anchors at 4" and
22-6s1 1:2 2" SCH-41 anchors at 21" 19.25 64% 3.17 Fracture

42-ns1 1:4 2" SCH-35 none 12.78 43% 2.67 peeling

Double U-wrap at ramp
42-us3 1:4 2" SCH-35 and single at 21" 28.53 95% 6.17 Fracture
2 3/8" anchors at 8"and Fracture around
42-4s2 1:4 2" SCH-35 at 21" 23.43 78% 3.17 anchor
2 < 5/8" anchors at 5"
42-4s5 1:4 2" SCH-41 and 19" 26.33 88% 3.79 Fracture
3 3/8" anchors at 8"and
42-6s2 1:4 2" SCH-35 at 21" 33.60 112% 3.42 Fracture
SCH-35 sheet 3 3/8" anchors at 8"and
42-6s3 1:4 2" SCH-41 anchors at 21" 21.41 71% 3.42 Anchor Fracture
SCH-35 sheet 3 1/2" anchors at 8" and
42-6s4 1:4 2" SCH-41 anchors at 21" 33.84 113% 3.79 Fracture
3 1/2" anchors at 8" and
42-6s5 1:4 2" SCH-41 at 21" 30.45 102% 3.79 Fracture
SCH-35 sheet 3 3/8" anchors at 8"and
42-6n1 1:4 2" SCH-41 anchors at 21" 16.36 55% 3.42 Anchor Fracture
3 1/2" anchors at 8" and
42-6n2 1:4 2" SCH-41 at 21" 29.77 99% 3.79 Fracture

41-ns1 1:4 1" SCH-35 none 7.86 26% 2.67 Peeling

43-ns1 1:4 3" SCH-35 none 11.69 39% 2.67 Peeling

SCH-35 sheet 3 1/2" anchors at 8" and
41-6s1 1:4 1" SCH-41 anchors at 21" 34.38 115% 3.79 Fracture
SCH-35 sheet 3 1/2" anchors at 8" and
43-6s1 1:4 3" SCH-41 anchors at 21" 29.61 99% 3.79 Fracture

CFRP anchors also allowed the CFRP to reach ultimate capacity, but the number and size of anchors played a
critical role. All anchors were inserted to a depth of 5 to 6 inches to ensure at least a 2” depth into the core of the
concrete (concrete interior of the reinforcing steel) ensuring that failure did not occur by separation of the concrete
cover. Two rows of one anchor of 5/8” diameter (full area of CFRP sheet per anchor) (00-2s1) reached only 71% of
CFRP capacity. Two rows of two anchors of 1/2” diameter (1/2 area) reached 84% of CFRP capacity (00-4s2).
Two rows of 3 anchors of 3/8” diameter (1/3 area) reached 100% capacity and lead to CFRP sheet rupture with only

667
an additional 1.5 sq ft area of CFRP (00-6s1). Each of the three 3/8” diameter anchors was able to develop about a
2” width of the 6” wide CFRP sheet. This test series of increasing anchor number and decreasing size indicates that
smaller more closely spaced anchors are more effective. However, increasing the size of each anchor by 33% (2/3
area) allowed the CFRP sheet to reach 100% capacity with only two rows of two anchors (00-4s3), indicating that
larger anchors are able to develop wider widths of a CFRP sheet.

3.2 With Height Transition

The effect of a height transition on the capcity of the CFRP sheet was considered in the next series of tests presented
in Table 1. The steepness of a 1:2 transition slope proved detrimental to the capcity of the CFRP sheet with failure
occuring at 64% of ultimate capcity (22-6s1). A shallower 1:4 allowed the CFRP to reach full capcity when
adequate anchorage was provided (42-6s4, 42-6s5, 42-us2). Two types of CFRP fabric (Tyfo® SCH-35 and SCH-
41) were donated by Fyfe Co. LLC. Both the have the same tensile strength (143 ksi), and almost the same elastic
modulus (11400 ksi and 13900 ksi respectively), but the SCH-35 fabric is regarded as a high grade fabric while the
SCH-41 is not. The grade difference may have caused SCH-41 fabric to be weaker in cases where the CFPR is bent,
such as in the anchors. Due to this weakness, the area of SCH-41 fabric had to be increased by 33% in order to
achieve the same capacity in the anchors as the SCH-35 fabric (42-6s2, 42-6s4). Surface preparation was evaluated
in 42-6n1 and 42-6n2 where a layer of plastic wrap was placed between the CFRP sheet and the concrete to
eliminate bond. Test 42-6n1, with 3/8” diameter anchors, reached only 55% of capacity, but 42-6n2 with larger ½”
anchors was able to reach 99% of capcity. The performance of 42-6n2 showed that with adequate anchorage,
surface preparation was not necessary because the anchors transferred all the tensile force from the CFRP sheet into
the concrete. The last set of tests studied the effect of the amount of height difference (1”, 2”, and 3”) in the height
transition with a 1:4 slope. When anchored with ½” anchors, the same full capacity of the CFRP sheer was reached
regardless of height difference (41-6s1, 42-6s5, 43-6s1). When unanchored, the height difference affected the
capacity though less than 45% of capacity was reached (41-ns1, 42-ns1, 43-ns1).

4. CONCLUSIONS
Anchorage of CFRP sheets enabled improved utilization of the tensile capacity of a CFRP sheet and thereby
increased the capacity of a CFRP retrofit with or without a height transition.
• U-wraps fully anchored the CFRP sheet, but can require 5 times the amount of CFRP as the unanchored case.
• CFRP anchors fully anchored the CFRP sheet and require far less additional CFRP, as little as 75% more area
of CFRP. Two rows of anchors with the area of CFRP fabric in each row equal to or greater than the area of
the CFRP sheet fully anchored the CFRP sheet. Smaller more closely spaced anchors were more effective;
however, increasing the area of CFRP used in the anchors increased the width of the CFRP sheet developed.
• The effect of a height transition was negated by the use of a 1:4 slope on the transition.
• The properties of carbon fiber fabric had an impact on the ultimate capacity in cases where the CFRP is bent.
• With proper anchorage, surface preparation to improve bond of the CFRP sheet to concrete was not necessary.
• The height difference did not affect the CFRP capacity when CFRP sheets were fully anchored.

5. REFERENCES

Bramblett, R.M. (2000). “Strengthening of Reinforced Concrete Beams Using Carbon Fiber Reinforced Polymer
Composites”, Masters Thesis, The University of Texas at Austin, Texas, USA.
Bonacci, J.F., and Maalej, F. (2001). “Behavioral Trends of RC Beams Strengthened with Externally Bonded FRP”.
Journal of Composites for Construction, May 2001, pp 102-113.
Ozdemir, G., and Akyuz, U. (2005). “Tensile Capacities of CFRP Anchors”, 7th International Symposium on Fiber-
Reinforced Polymer (FRP) Reinforcement for Concrete Structures, Editors: C.K. Sheild, J.P. Busel, S.L. Walkup,
and D.D. Gremel, pp. 39-56.
Saaticoglu, M., Serrato, F., and Foo, S. (2005). “Seismic Preformance of Masonry Infill Walls Retrofitted With
CFRP Sheets”. 7th International Symposium on Fiber-Reinforced Polymer (FRP) Reinforcement for Concrete
Structures, Editors: C.K. Shield, J.P. Busel, S.L. Walkup, and D.D. Gremel, pp. 341-353.
Teng, J.G., Chen, J.F., Smith, S.T., and Lam, L. (2001). FRP-Strengthened RC Structures, John Wiley & Sons LTD,
New York.

668
Third International Conference on FRP Composites in Civil Engineering (CICE 2006)
December 13-15 2006, Miami, Florida, USA

EFFECT OF FRP U-JACKETING ANCHORAGE

ON FRP-CONCRETE INTERFACES UNDER FATIGUE LOADING

Kentaro Iwashita
(Researcher, Ibaraki University, Hitachi, Ibaraki, Japan)

Zhishen Wu
(Professor, Ibaraki University, Hitachi, Ibaraki, Japan)

Takashi Ishikawa
(Director, Japan Aerospace Exploration Agency (JAXA), Mitaka, Tokyo, Japan)

Yasumasa Hamaguchi
(Invited Researcher, Japan Aerospace Exploration Agency (JAXA), Mitaka, Tokyo, Japan)

ABSTRACT
This study aims to examine and investigate the effect of FRP U-jacketing anchorage on Fiber Reinforced Polymer
(FRP) sheets-concrete interfaces under fatigue loading. Double-lap shear test specimens with the FRP U-jacketing
anchorage are carried out under static and fatigue loading. The fatigue behavior of FRP sheets-concrete interface is
finally characterized by the conducted amplitude tensile load (S) - the number of cycles (N) diagram at debonding
on a semi logarithmic scale. The results demonstrate that the FRP U-jacketing anchorage increase the static bond
strength and delay the occurring of initial FRP debonding under high fatigue load (high amplitude load) level, but it
cannot enhance the fatigue strength of the FRP-concrete interface.

KEYWORDS
Fiber Reinforced Polymer (FRP) sheets, bond, fatigue loading, S-N curve, FRP U-jacketing anchorage

1. INTRODUCTION
The need to rehabilitate or upgrade the deteriorated and deteriorating civil infrastructure has been becoming a major
and urgent problem worldwide. In contrast with the traditional strengthening methods such as overlaying and
jacketing method, external cable method and bonded and jacketing steel plate method, the use of fiber reinforced
polymers (FRP) represents an innovative and effective technology for strengthening of infrastructure. In recent years,
repairing and strengthening concrete structures with externally bonded, epoxy-bonded FRP sheets to the tension face
of structural element has been widely accepted for practical uses. But, recent studies have also reported a variety of
FRP debonding problems. Therefore, it was found that the limit value of maximum load under fatigue loading is
only about a half of maximum load under static loading (Ferrier et al. 1999, Wu et al. 2002, Tan et al. 2003,
Iwashita et al. 2006). Recently, FRP U-jacketing anchorage is considered to be used as a countermeasures for
enhancing the FRP-concrete bonding behavior around the ends of prestressed FRP sheets or around intermediate
flexural cracks of the concrete structures strengthened with FRP sheets. The benefits associated with the FRP U-
jacketing anchorage have been investigated by many researchers, but there is little research for discussing the
effectiveness of this anchorage under fatigue loading. Moreover, for strengthening RC/PC structures, the evaluation
of the fatigue performance of the FRP U-jacketing anchorage is considered to be very important one. Based on these
considerations, this paper aims at investigating clearly the fatigue performance of FRP U-jacketing anchorage and
the expanding effect on fatigue life of FRP-concrete interfaces.

669
 Table 1 Summary of Material Properties
Guaranteed tensile strength (GPa) 3.5
PBO fiber reinforced
Guaranteed tensile modulus of elasticity (GPa) 240
polymer sheets
Nominal thickness (mm/1 layer) 0.128
Guaranteed tensile strength (MPa) 51.9
Epoxy resin Guaranteed tensile shear strength (MPa) 25
Guaranteed tensile modulus of elasticity (GPa) 3.43

Pre-existing crack FRP U-jacketing FRP sheets 100

Anchoring with FRP sheets 50
anchorage
(Circumferential direction) Concrete block
Welding
Fiber direction
Fiber
direction 100
Fiber direction
Steel nut
170 Bond length 200 Steel bolt
FRP U-jacketing
200 250
Concrete block anchorage
Steel plate (Thickness: 15mm) (a) Side View
FRP U-jacketing anchorage Hydraulic
force

Fiber
direction

100 Fiber direction pull rod

Grip Unit: mm
(b) Top View
Fig. 1 Test Set-up and Details of The Prism Specimens

2. EXPERIMENTAL PROGRAMS
2.1 Composite Materials

The newly developed PBO (Poly-p-phenylene-benzobisoxazole) fiber sheets was chosen for this study. PBO fiber
reinforced polymer (PFRP) sheets mainly present advantages such as high tensile strength and high stiffness-to-
weight ratio as shown in Table 1. Density of reinforcing continuous fiber sheets are 200g/m2 and fiber volume
content (Vf) is decided to 50%.

2.2 Specimen Details and Test Set-Up

Specimens of FRP sheets-concrete double-lap bonding joints

In accordance with experimental method of JSCE Recommendations for Upgrading of Concrete Structures with Use
of Continuous Fiber Sheets (JSCE, 2001), the prism specimens with 100mm width, 100mm thick and 500mm long
are used in the investigation. The specimen is cut off at the position of the notch (Pre-crack). Before bonding of FRP
sheets, the concrete surface preparation is treated with a diamond sander, and an epoxy primer is painted after
wiping with cloth soaked with acetone. 2 layers FRP sheets whose size is 50mm width and 400mm span is
impregnated with epoxy resin before bonded to concrete surface, and bonded to both sides of the concrete block
along the axial direction after hardening of adhesive by the end of 24 hours. The tensile tests were started 7 days
later. The details of the prism specimens are schematically shown in Fig. 1. The 28-day compressive strength of the
concrete was 45.2MPa. The tensile load is applied by pulling both ends of the steel plate. A summary of the
properties of PBO fiber sheets, the epoxy resin and the concrete is shown in Table 1.

Test Set-Up
The static and fatigue loads were applied to the specimen using an Instron 8502 series digital servohydraulic fatigue
testing machine operating under load-control mode. Then, the load acting on the specimen and the position of lower
grip were measured by using a load cell with 100kN capacity and LVDT set in the test machine. All static loads
were applied with a speed of 1kN/min and all fatigue loads were applied with a rate of 5Hz. All data was recorded at

670
 3 Tensile load
Interfacial fracture energy Gf 2.5
Peeling
2
debonding
(kN/mm)
1.5
1
0.5
0
Without FRP With FRP
U-jacketing U-jacketing
anchorage anchorage

Fig. 2 Maximum load under static loading Fig. 3 Debonding Behavior of FRP U-jacketing anchorage

Debonding into concrete layer

104 cycles 5 × 105 cycles 106 cycles 1.1 × 106 cycles 1.148 × 106 cycles
Fig. 4 Debonding Process of FRP Sheets and FRP U-jacketing Anchorage under Fatigue Loading

every 1000th cycle with Instron WaveMaker software. The variables of maximum repeated loads are set as 40%,
50% and 60% of the lowest value Pu among the maximum load investigated by the three static tests, and the
minimum repeated load is set as 10% of Pu in all experiments. In case that the FRP debonding does not occur by the
2 millions cyclic loading, loading was stopped and tests were finished. The tests were carried out under a normal
temperature condition (25-27˚C).

3. RESULTS AND DISCUSSION

3.1 Effect of FRP U-jacketing Anchorage on Bond Strength under Static Loading

The results of prism tests under static loading are investigated for the determination of interfacial fracture energy
(Gf). This value is calculated from the following equation.
P 2 max
Gf = 2 (1)
8b E f t
where P is the maximum transferable force in FRP sheet; Ef, b and t are modulus of elasticity, width and thickness of
FRP sheets respectively. The average values of Gf are shown in Fig. 2. The average Gf value of without FRP U-
jacketing specimens is nearly same as a standard value (JSCE, 2001, Wu et al., 2001) and a good reproducibility of
these experiments was confirmed. From the results, average Gf value of specimens with FRP U-jacketing anchorage
is higher than those without anchorage. It is found that the FRP U-jacketing anchorage can significantly increase the
bond strength under static loading. On the other hand, peeling and debonding of the FRP U-jacketing sheets are
occurred around crossing point of FRP and anchor FRP on the final stage as shown in Fig. 3, and FRP debonding
reaches to the FRP ends.

3.2 Effect of FRP U-jacketing Anchorage on Bond Strength under Fatigue Loading

In case of 40%-10% of Pu fatigue loading specimen with FRP U-jacketing anchorage, it is confirmed by the impact
echo tests that micro-debonding initiates at the tensile end of bonded FRP sheets, and propagates gradually to form a

671
 14 70

Maximum tensile load ratio (%)

Amplitude tensile load (kN)
12 60

10 50

8 40
6 30
Without FRP U-jacketing anchorage
4 With FRP U-jacketing anchorage 20
2 Approximately curve of specimens without FRP U-jacketing anchorage 10
Approximately curve of specimens with FRP U-jacketing anchorage
0 0
1000 10000 100000 1000000 10000000
Number of Cycles

Fig. 5 Amplitude Tensile Load (S) - Number of Cycles (N) Curve.

macro-debonding by about 104 cycles as shown in Fig. 4. Once macro-debonding length grows to about 20-30 mm
after about 5 × 105 cycles, debonding propagates towards the free end of bonded FRP sheet. Finally, a complete FRP
sheet debonding occurs when debonding propagates to the FRP end. Fig. 5 shows the relationships between
amplitude stress (S) and number of cycles (N) with a semi logarithmic scale. The results show that the FRP U-
jacketing anchorage can delay the debonding failure occurred by fatigue loading under short-term fatigue loading
before 106 cycles, but the effects are mitigated or disappeared on the specimens which have long-term fatigue life
more than 106 cycles. Then, peeling and debonding of the FRP U-jacketing sheets occur around crossing point of
FRP and anchor FRP after the final debonding as shown in Fig. 3, and FRP debonding reaches to the FRP ends. It is
considered that the FRP U-jacketing anchorage can not expand the long-term fatigue life. On the other hand, the
maximum load ratio of FRP sheets under fatigue loads of 106 cycles with or without a FRP U-jacketing anchorage is
about 35% of one under static loading.

4. CONCLUSIONS
From the study carried out, bond strength and debonding processes under fatigue loading of FRP sheets with a FRP
U-jacketing anchorage are examined. Based on the results of this study, the following conclusions are drawn:
1. It is found that the FRP U-jacketing anchorage can significantly increase the static and short-term bond strengths,
before 106 cycles, but the effects are disappeared on the specimens which have long-term fatigue life more than
106 cycles.
2. The maximum load ratio of FRP sheets under fatigue loads of 106 cycles with or without a FRP U-jacketing
anchorage is about 35% of one under static loading.

5. REFERENCES
Ferrier, E., Nasseri, H., Hamelin, P. (1999) “Fatigue behavior of composite reinforcement for concrete structures”,
Proceedings of fourth international symposium on fiber reinforced polymer reinforcement for reinforced concrete
structures (FRPRCS-4), ACI International SP-188-48, Editors: Dolan, C.W., Rizkalla, S.H., Nanny, A., pp.535-545.
Wu, Z.S, Iwashita, K., Ishikawa, T., Hamaguchi Y. (2002). “Bonding and debonding behavior of FRP sheets under
fatigue loading”. Proceedings of 10th US-Japan workshop on composite materials, Editors: Chang F.K., pp.810-818.
Tan, K.H. (2003) “Effect of cyclic loading on bond strength of FRP reinforcement”, Proceedings of International
Symposium on Latest Achievement of Technology and Research on Retrofitting Concrete Structures – Interface
Mechanics and Structural Performance, Kyoto, Japan, pp. 1-8 (Keynote paper).
Iwashita, K., Wu, Z.S, Ishikawa, T., Hamaguchi Y., and Suzuki T. (2006). “Bonding and debonding behavior of
FRP sheets under fatigue loading”. Journal of Advanced composite materials, VSP publishers (Accepted).
Japan Society of Civil Engineers (JSCE) (2001) “Recommendations for Upgrading of Concrete Structures with Use
of Continuous Fiber Sheets”, Concrete Engineering Series 101.
Wu, Z. S., Yuan, H., Yoshizawa, H. and Kanakubo, T. (2001) “Experimental/Analytical Study on Interface Fracture
Energy and Fracture Propagation Along FRP-Concrete Interface”, Fracture Mechanics for Concrete Materials:
Testing and Applications, ACI International SP-201-8, pp.133-152.

672
Third International Conference on FRP Composites in Civil Engineering (CICE 2006)
December 13-15 2006, Miami, Florida, USA

FRP Upgrade of Concrete Girders to Support Flood Loads

Abdelhakim Bouadi, Ph.D., P.E.

Senior Associate, Walter P Moore, Structural Diagnostics Services, Houston, Texas, USA

Eric Green, P.E.

Principal, Walter P Moore, Structural Diagnostics Services, Houston, Texas, USA

Narendra Gosain, Ph.D., P.E.

Senior Principal, Walter P Moore, Structural Diagnostics Services, Houston, Texas, USA

ABSTRACT
A medical facility building located in Houston, Texas incorporating a basement housing equipment and laboratories
implemented a flood protection system consisting of multi-layered internal and external flood walls. Part of the
flood protection system included strengthening of a portion of the existing basement roof to carry the weight of
seven feet of water. The basement roof structure is formed by a reinforced concrete pan-joist system supported by
girders. A review of the structure showed that the slab and joists had sufficient capacity to resist the loads due to the
floodwaters. However, the girders did not have adequate flexural capacity and required strengthening. Several
repair options were considered. The final option consisted in the use of Carbon Fiber Reinforced Polymer (CFRP) to
strengthen the existing girders. CFRP was epoxy-bonded to the soffit of the girders. To simplify the retrofit, the
design allowed for cracking and ensuing moment redistribution at the column supports. Construction was
performed without any disruption to the existing mechanical, electrical, and plumbing (MEP) lines at the basement
ceiling, and was completed in a timely fashion without any impact to building operations.

KEYWORDS
Fiber Reinforced Polymers, CFRP, Reinforced Concrete, Upgrade, Construction.

1. Introduction

A structural flood protection system was built around an existing medical facility located in Houston, Texas. The
flood prevention system consists of two concentric and independent external (primary) flood walls and several
internal (secondary) flood walls. Two types of internal floodwalls were installed. One internal wall divides the
structure in half to prevent complete flooding of the building in the event of a failure of the primary flood walls.
Other internal walls protect critical MEP rooms. Implementation of the flood protection system required
construction of new reinforced concrete walls and strengthening of the existing reinforced concrete components.

2. Building Description

The facility built in the 1970s is a seven story building. The plan dimensions are approximately 250 feet x 190 feet.
The building basement houses research equipment and laboratories. Due to existing site constraints, a roadway
passes through the width of the structure at the ground level, with the basement roof functioning as the roadbed (Fig.
1).

673
 Figure 1: Plan View of Building and Flood Wall Location

The basement roof structure located between gridlines 4 and 7 is formed by a 5” reinforced concrete slab spanning
between concrete joists that are located at 30 inches on center. This part of the structure was designed to resist the
code-level loads due to vehicular traffic, including AASHTO defined concentrated loads and a uniform load of 250
psf. The joists are 16 inches wide, 25 inches deep and have a span of about 32 feet. The joists are supported by
concrete girders that have a span of about 28 feet. The girders at gridline 7 are 40 inches wide and 47 inches deep.
The girders at gridlines 4, 5 and 6 are 36 inches wide and 32 inches deep (Fig. 2). Reinforcing is shown is Table 1.
All reinforcing bars were specified to conform to ASTM A615 with a yield strength of 60 ksi.

Figure 2: Girder Configuration at the Basement Ceiling

Table 1: Typical Girder Reinforcement

Gridline Mid Span Bottom Reinf. Top Reinf At Support Top Reinf at Gridlines B and G
4 5 # 10 8 #10 4 #10 and 2 #9
5 6 #11 7 #11 and 3 #10 4 # 11 and 5 #10
6 6 # 11 4 # 11 and 6 # 10 4 # 11 and 5 #10
7 4 #10 4 #10 and 4 # 9 4 #10 and 2 #9

3. Primary Flood Protection at Roadway

3.1. Description

The primary flood protection system was formed by constructing two new 7-foot tall reinforced concrete walls
located at the north and south sides of the road and strengthening of the basement roof to carry the loads imposed by
seven feet of standing water (Fig. 3).

674
3.2. Analysis of the Existing Structure

The design of the existing basement roof structure was analyzed. At the time of the original construction, the
structural system was designed for vehicular loads which included a service uniform live load of 250 psf. The
addition of the flood system resulted in a service live load due to the flood water of about 440 psf. Although the
design loads under flood conditions were significantly higher than the original design loads, the structural review
indicated that the existing slabs and joists had adequate capacity to resist the higher loads. This was a result of the
combined effects of lower load factors in the ACI 318-02 design code (ACI Committee 318, 2002), a higher
concrete compressive strength and a conservative original design. However, while the girders had sufficient shear
strength to carry the higher flood loads, they were found to have insufficient positive and negative flexural capacity.

Given the age of the structure, it was recognized that the actual concrete compressive strength was likely to be
higher than the specified 4000 psi. The actual concrete strength based on field-measured values was found to be
4,500 psi.

Figure 3: Section through Building at Roadway

4. Structural Upgrade

The first retrofit option considered the addition of bottom and top reinforcing bars in new concrete jackets connected
to the existing girders with the use of dowels drilled and epoxied into the existing girders. The addition of the top
reinforcement would require removal and replacement of the driving surface (topping slab). To develop the
negative moment at the joints, the top reinforcement needed to be anchored into the column with the use of through
bars or epoxy-doweled bars.

The use of the concrete jacket presented several construction challenges. The proposed dowel installation was labor
intensive and time consuming. In addition, avoiding damage to the existing rebar during installation required a pre-
construction phase consisting of locating the existing reinforcement. Additionally, the girders were located in the
building basement which had extensive amount of mechanical ducts, pipes and mechanical or electrical equipments.
Construction would require temporary disconnection and relocation of several critical pieces of mechanical,
electrical and plumbing lines and equipment that would result in a temporary shut down of some building
operations. This option was rejected because the expected operational disruption was unacceptable to the owner.

The second option considered the use of FRP epoxy-bonded to the underside of the beams to increase the flexural
capacity. The use of FRP only was found not to be practical in increasing the negative moment capacity of the
beams. Therefore we considered the use of moment redistribution at the supports where the imposed moment
exceeded the negative flexural capacity of the girders. Although this redistribution remained within the limits
allowed by the ACI code, the informed consent of the owner's representatives was solicited as moment redistribution
might result in permanent deformation and concrete cracking in the event of a major flood.

After discussion, the owner's facility managers agreed to the use of a strengthening design that allowed for moment
redistribution. This decision was based on several factors, including the fact that the alternative, using bonded
reinforcing, would impact the building operations in an unacceptable manner during construction. Additionally, the

675
floor system would see full loads only in the event of a 500 year flood, an event which only has a 0.2% chance of
occurring in any given year.

The final upgrade design consisted in the use of wet layup system of CFRP sheets having a thickness of 0.041
inches, an ultimate tensile strength of 127 ksi and an elastic modulus of elasticity of 10,500 ksi. The width of the
sheets was in general equal to the girder width. The CFRP upgrade extended for an average of 16 feet centered on
the girders. At the end, U-strips with Glass FRP anchors were added to prevent end-debonding. The upgrade was
done following the general recommendations of ACI 440.2R-02 (ACI Committee 440, 2002).

5. Construction

The construction was performed without any disruption to the existing mechanical, electrical and plumbing lines at
the basement ceiling, and was completed in a timely fashion without any impact to the building operations. The
only area of the building that was closed during the construction process was the basement level where the actual
work was performed.

As part of the construction quality control, it was required that an independent laboratory conduct a number of tests
that included the preparation of coupon samples to verify the properties of the material. Due to the relatively recent
introduction of the CFRP material into the market, the local testing laboratories were not trained in collecting or
testing the material. In order to overcome this obstacle, the coupons were collected by the contractor in presence of
the engineer and testing laboratory personnel. This created a level of reliability in the preparation of test coupons.
The coupons were then submitted to an out-of-state independent specialty laboratory with prior experience in testing
of CFRP coupons. The lack of testing laboratory personnel trained in the CFRP presented challenges for quality
control, especially since the design was based on data provided by the manufacturer. The ability to verify the
material properties is essential in order to validate the design. In addition several bond tests were carried by the
contractor in presence of the engineer and testing laboratory personnel to verify proper adhesion.

6. Conclusions

The introduction of a new flood control system in a medical facility required upgrading existing concrete girders that
form the roof structure of the basement and a support for a vehicular roadway. The upgrade was limited to an
increase in the positive flexural capacity of the beams as the shear capacity was found to be adequate and the code
permitted redistribution of the moment was used resulting in an acceptable negative moment capacity. While
several repair options were considered, the final option consisted of the use of CFRP epoxy-bonded to the soffit of
the girders. Construction was performed without any disruption to the existing mechanical, electrical or plumbing
lines at the basement ceiling, and was completed in a timely fashion without any impact to building operations. The
lack of FRP trained personnel in the local testing laboratories presented a challenge in verifying the properties of the
FRP and in performing other specified tests. This was overcome by having the contractor collect sample in the
presence of the engineer and the testing laboratory personnel and in having the testing laboratory submit the samples
to an-out-state specialty laboratory experienced in conducting such tests.

7. References

ACI Committee 318 (2002). Building Code Requirements for Structural Concrete (ACI 318-02) and Commentary
(ACI318R-02), Farmington Hills, MI.
ACI Committee 440, 440.2R-02 (2002). Design and Construction of Externally Bonded FRP Systems for
Strengthening Concrete Structures, American Concrete Institute, Farmington Hills, MI, 2002.

676
Third International Conference on FRP Composites in Civil Engineering (CICE 2006) 
December 13­15 2006, Miami, Florida, USA 

TESTS ON FRP­STRENGTHENED TIMBER JOINTS 
K.I. Crews 
(Professor, Centre for Built Infrastructure Research, University of Technology Sydney, Australia) 

S.T. Smith 
(Senior Lecturer, Centre for Built Infrastructure Research, University of Technology Sydney, Australia) 

ABSTRACT 
Surprisingly little research has been conducted on the strengthening of timber with fibre reinforced polymer (FRP) 
composites as opposed to the much more widely researched strengthening of concrete and to a lesser extent metallic 
structures.  As  with  all  FRP  strengthening  applications,  the  bond  of  the  FRP  to  the  substrate  is  of  particular 
importance.  A  lack  of  understanding  of  the  bond  between  FRP  and  timber  is  a  major  factor  contributing  to  the 
reluctance of industry to utilise FRP for timber strengthening applications. This paper reports results of preliminary 
bond strength tests undertaken  at  the  University  of  Technology Sydney (UTS) on FRP­strengthened  timber joints. 
The aims of the tests were to observe the suitability of the test method, quality of the bond, bond strength and failure 
mode of the test specimens. 

KEYWORDS 
FRP, Bond, Strengthening, Timber, External Bonding 

1. INTRODUCTION 
There  are  some  interesting  similarities  between  the  properties  of  timber  and  those  of  FRP’s,  in  that  both  have 
orthogonal  properties  as  a  consequence  of  having  an  inherent  fibrous  structure  set  in  a  matrix  binder.  It  can  be 
argued that FRP is a manufactured product of “perfect” timber – where the fibre strength rather than the occurrence 
of  strength  reducing  characteristics  that  occur  naturally  in  wood  (e.g.  knots),  govern  the  strength  of  a  structural 
member.  It  is  therefore  somewhat  surprising  that  combinations  of  the  two  products  have  not  been  more  widely 
researched and as a result the bond interaction between FRP and timber is not generally understood. 

Most work on FRP­strengthened timber to date has  focused on the bonding of FRP composites to selected  faces of 

timber  beams.  Research  by  Meier  (1995),  Tingley  et  al.  (1996),  Chajes  et  al.  (1996),  Gilfillan  et  al.  (2004),  and 
Dagher  (2005)  has  demonstrated  that  FRP  bonded  tension  face  plates  (or  tension  and  compression  faces)  can 
significantly  increase  the  bending  strength and  stiffness of a  timber  beam,  whilst  Milner  (1999)  demonstrated  the 
effectiveness  of  FRP’s  to  overcome  inherent  weaknesses  in  the  finger  joints  of  glued  laminated  timber  beams. 
Experimental  work  undertaken  by  Greenland  et  al.  (1999)  at  UTS  also  explored  this  issue  as  part  of  a  detailed 
program of research on the viability of externally bonded FRPs for improving the tensile capacity of stress laminated 
T system webs, but this same work also highlighted some of the difficulties associated with bonding failures which 
can occur between the FRP and the timber substrate. 

2. PREVIOUS RESEARCH AT UTS 
The work undertaken to date  at UTS  has focused on characterising the  flexural  performance of Australian Radiata 
Pine  products strengthened with  carbon FRP (CFRP)  composites.  Pilot tests  have been  conducted on solid (sawn) 
and reconstituted (Laminated Veneer Lumber ­ LVL) sections of Radiata Pine to determine the short term properties 
of  the  CFRP,  the  bending  strength  and  stiffness  of  timber  elements,  and  improved  bending  strength,  stiffness  and 
ductility, with reduced variability, of FRP­strengthened timber elements. Further pilot studies investigated the short 
term behaviour of T beams (for use  in stress laminated  timber decks)  constructed of  LVL  webs  and  solid  Radiata 
Pine  flanges  with  and  without FRP  web  strengthening  (Greenland  2001).  Whilst  Greenland  developed  models  for

677
predicting  linear­elastic  and  nonlinear  behaviour  of  the  FRP  composite  LVL  beams  as  a  structural  system,  the 
fundamental  behaviour  of  the  bond  and  the  bond­slip  relationship  of  the  FRP­strengthened  timber  were  not 
addressed. 

3. TESTS ON BOND BEHAVIOUR OF FRP AND TIMBER 
Vick (1997) investigated the durability of  epoxy  bonds  and presented  details of a  primer system developed  by the 
US Department  of  Agriculture  (USDA)  to  improve  epoxy  durability.  It  was  concluded  that  epoxy  bonds  develop 
bonds  to  timber  that  are  as  strong  as  the  timber  itself,  as  long  as  the  bonds  remain dry  and  that  epoxy  adhesives 
could equal the structural durability of resorcinolic adhesives when the USDA primer was used. The effectiveness of 
adhesion  between  FRP  and  timber  was  evaluated  using  a  cyclic  delamination  test  as  noted  in  ASTM  D2559­03 
(2003) Standard Specification for Adhesives for Structural Laminated Wood Products for Use Under Exterior (Wet 
Use) Exposure Conditions. Once the specimens were exposed to the severe stresses from repeated water soaking and 
drying, the bonds tended to degrade and delamination occurred. 

One of  the variables that affects  the behaviour  of a composite FRP/timber element,  are the shear  properties  of the 

bond/interface.  Because  the  transfer  of  stress  between  timber  and  the  fibre  composite  is  achieved  via  the 
development  of  shear  stresses  in  the  adhesive  bond,  it  is  important  that  the  strength  properties  of  the  adhesive  in 
shear be understood and quantifiable for use in numerical models. In particular, the way in which load is transferred 
between  the  FRP  and  timber  needs  to  be  understood,  namely;  will  the  behaviour  of  the  strengthened  section  be 
governed  by  the  strength  of  the  timber  or  FRP,  or  is  it  governed  by  the  ability  of  the  adhesive  to  transfer  shear 
stresses from the timber into the FRP? 

The strength properties of adhesive bonds in shear can be determined using either compression or tension methods. 
The majority of studies on the bond between timber and fibre composites have involved testing the shear strength of 
the adhesive through a (modified) compressive method, based on ASTM D905­03 (2003): Standard Test Method for 
Strength Properties of Adhesive Bonds in Shear by Compression Loading. 

While ASTM D905­03 (2003) is primarily concerned with obtaining the shear strength of an adhesive, it is arguable 
that  the  critical  factor in developing reliable  bonds  in  FRP­to­timber  composites  is  to ensure  that  the  failure  mode 
occurs as a wood fibre failure, rather than in the adhesive itself. This would be evident by the proportion of wood on 
the  failure  surface  –  a  high  amount of  wood  on  the  failure  surface  indicating  that  the  adhesive  itself  may  be  less 
critical than other factors.  Given that high proportions of wood failure are desirable, the results of such tests do not 
necessarily indicate the true value of the shear strength of the adhesive, since ASTM D905­03 specifically notes that 
wood failure is very common in joints made with strong adhesives, and when high proportions of wood are evident 
on the failure surface, the measured strength is lower than the true adhesive strength. 

Preliminary tests on  10  specimens have  been undertaken  at  UTS  using specimens (as indicated  in Figure 1)  made 

from the following materials:
·  Radiata pine (solid, sawn timber, dressed surface finish)
·  ATL type I prefabricated carbon fibre composite
·  ATL epoxy “Techniglue CA” 

The ATL type I composite carbon fibre, consisted of 2 layers of a 580g/m 2  unidirectional carbon fibre prepreg plates 
which, when cured, formed 1.6mm thick carbon fibre laminates. When tested in accordance with the 1995 version of 
ASTM  D3039  (2006)  Standard  Test  Method  for  Tensile  Properties  of  Polymer  Matrix  Composite  Materials,  the 
following  properties  were  obtained  from  testing  of  10  specimens:  average  ultimate  strength  726  MPa,  with  a 
coefficient of variation (CoV) of 4.1%; and average Modulus of Elasticity 78,200 MPa with a CoV of 10.3%. 

The  cut  timber  and  fibre  composites  were  glued  together  using  ATL  Techniglue  CA  epoxy  adhesive  after  the 
bonding  surfaces  of  both  the  pine  and  fibre  composites  were  lightly  sanded  in  accordance  with  the  adhesive 
manufacturer’s  directions.  The  test  specimens  were  50mm  wide  blocks,  with  a  set  up  similar  to  that  in  Figure  1 
where the block on the left side is fixed while load is applied to the block on the right side. 

Whilst the tests methods used for the determination of strength properties in shear were primarily sourced from the 
1994 version of ASTM D905­03 (2003), cross reference was also made to AS1321.3 (1976) Bond Strength of Cured 
Wood­to­Wood  Adhesives  in  Shear  and  to  the  block­shear  test  method  specified  in  the  1987  version  of  AS1328

678
(1998) Glued­Laminated Structural Timber. The ASTM D905­03 (2003)  test method  for specimens containing the 
FRP required some modifications, which was the same as those made by some other researchers testing the interface 
properties between various timbers and fibre composites (e.g. Davalos et al. 1992). 

P 
Epoxy  Timber Block 
Table 1.  Test results 
Specimen  Ultimate Load  Shear Stress 
(kN)  (MPa) 
Epoxy 
5 mm  S03­1  21.1  10.6 
CFRP  S03­2  16.75  8.4 
50 mm S03­3  10.8  5.4 
S03­4  12.2  6.1 
S03­5  20.4  10.2 
S03­6  20.4  10.2 
S03­7  9.0  4.5 
S03­8  14.0  7.0 
S03­9  16.4  8.2 
2 mm 
5 mm  Average  7.8 
15 mm  15 mm 
Standard Deviation  2.2 
CoV (%)  28.4% 
5% exceedence  4.9 
Figure 1.  Side elevation of a typical test specimen 

The  purpose  of  this  test  is  to  determine  the  strength  of  the  adhesive­to­timber  interface,  assuming  that  the  shear 
stress is constant throughout the depth of the “glue line”. Whilst some effort was made for the wood specimens to be 
fabricated so as  to  avoid significant growth  characteristics,  such as knots,  the  inherent variability of  timber is  still 
reflected in the test results. This simplification will need to be addressed in future tests with larger pieces of timber 
that  will  inevitably  include  knots, since  such  a  “smeared”  average,  will  not  consider  the  effects  of knots.  Failure 
modes can be described as brittle. 

4. RESULTS AND DISCUSSION 
The results of the preliminary tests are presented in Tables 1 and 2. In Tables 1 and 2, the shear block test result for 
the FRP­timber specimens are given as well as a typical shear strength of the timber and manufacturers data for the 
epoxy alone in Table 2. It is evident the shear strength of the timber and epoxy are higher thus leading to failure at 
the FRP­to­timber interface in the timber. 

Table 2. Summary results of shear block tests 

Summary  Shear Test Result  Timber  Epoxy 
Average Shear Stress (MPa)  7.8  14  10 ­ 12 
Standard Deviation  2.2  1  ­ 
CoV (%)  28%  7.1%  ­ 
5 th  Percentile (MPa)  4.9  12.7  ­ 

Failures  of  the  block shear  specimens  can be  characterised  into  three  types  of  failure: Failure  within  the  glueline, 
failure within the wood, and  failure partly within and partly adjacent  to the glueline in the wood. The main type of 
failure  that  occurred  in  these  tests  was  the  third  one,  with  a  significant  proportion  of  the  failure  surface  showing 
wood failure. The results of the shear tests indicate that the shear strength of the timber­to­FRP interface is less than 
the shear strength of the (solid) timber and the epoxy. However,  it must be noted that only a very small number of 
specimens were tested and a far greater number of specimens would need to be tested to report conclusively. Due to 
the high variability of these test results, conclusions about the shear strength of the bond, in particular whether the 
bond  is  likely  to  be  a  critical  influence  on  limit  state  behaviour  of  reinforced  beams,  are  difficult  to  draw  with 
confidence  based  on  the  small  number  of  specimens  tested.  However,  failure  surfaces  typically  showed  greater 
proportions  of  wood  failure  than  adhesive  failure  indicating  that  the  properties  of  the  bond  are  controlled  by  the 
properties  of  the  Pine  tested  rather  than  the  properties  of  the  adhesive  used.  Other  influences  on  the  results  are: 
surface preparation, knots, and size effects of knots on small test specimen

679
It  should  be  noted that most  glued  timber products  that  are  used  in  high  performance  structural  applications  (e.g. 
plywood,  glulam  and  LVL),  rely  upon  mechanical  interlock  of  the  glue  with  the  timber  fibres,  occurring  at  a 
microscopic level. In the manufacture  of plywood  and LVL the  sheets  of ply are heated  (to activate the  resorcinol 
glue and also soften the timber) and squeezed together under a pressure of 1MPa in order to force the epoxy into the 
microfibres  of the  timber. This  results in what  is really  a mechanical bond, rather  than adhesion,  between the  glue 
and the timber. Such glue lines tend to be very thin as a result and gap filling glues or expoies are seldom used  in 
structural applications. Therefore, in applying FRP strengthening to timber, we need to recognize that the same type 
of structural bond is  not possible  and as a  result the  connection  is  predominantly  adhesive  rather than mechanical. 
Furthermoore, it is possible that some of the natural growth charactersitcs of  timber, such as knots, may in fact act 
like aggregate in concrete, and in doing so, create difficulty for the epoxy to penetrate the substrate material. These 
are all issues that will need to be addressed in future research. 

5. FUTURE RESEARCH 
Considerably more tests will need to be conducted in the future to systematically characterise the influence of many 
variables (e.g. species of timber, grade of timber, surface preparation, moisture condition etc) on the FRP­to­timber 
bond  strength.  The  test  set­up  outlined  in  this  paper  appears  adequate  for  conducting  future  testing.  Ultimately  a 
bond­slip relation of FRP­strengthened timber  joints  that  can  be used  in numerical  simulations  is required. Such  a 
relation may be best determined by fitting strain gauges at the epoxy­timber interface. Care must be taken to ensure 
the strain gauges do not disturb the distribution of bond stresses. 

6. REFERENCES 
AS1321.3  (1976).  Methods  for  the  Sampling  and  Testing  of  Adhesives  ­  Bond  Strength of  Cured  Wood­to­Wood 
Adhesives in Shear. Standards Australia, Homebush, Australia. 
AS1328 (1998). Glued­Laminated Structural Timber. Standards Australia, Homebush, Australia. 
ASTM D905­03 (2003). Standard Test Method for Strength Properties of Adhesive Bonds in Shear by Compression 
Loading. American Society for Testing and Materials, Philadelphia, USA. 
ASTM D2559­03  (2003). Standard Specification  for  Adhesives  for  Structural  Laminated  Wood  Products  for  Use 
Under Exterior (Wet Use) Exposure Conditions.  American Society for Testing and Materials, Philadelphia, USA. 
ASTM  D3039  (2006).  Standard  Test  Method  for  Tensile  Properties  of  Polymer  Matrix  Composite  Materials. 
American Society for Testing and Materials, Philadelphia, USA. 
Chajes, M.J., Kaliakin, V.N. and Meyer, A.J. Jr. (1996). “Behaviour of engineered wood­CFRP beams”. Proc., First 
International Conference on Composites in Infrastructure, ICCI'96, Arizona, USA, 15­17 January, 
Dagher, H.J. (2005). “Current state of reinforced wood technology: New products, codes and specifications”. Proc., 
Third International Conference Advanced Engineered Wood Composites, Maine, USA, 10­14 July. 
Davalos,  J.F.,  Salim,  H.A.  and  Munipalle,  U.  (1992).  “Glulam­GFRP  composite  beams  for  stress­laminated  T­ 
system  bridges”.  Proc.,  Advanced  Composite  Materials  in  Bridges  and  Structures,  Eds.  Neale  and  Labossiere, 
Canadian Society for Civil Engineering, Sherbrooke, Canada. 
Gilfillan,  J.R.,  Gilbert,  S.G.  and  Patrick,  G.R.H.  (2004).  “Improving  the  structural  performance  of  timber  beams 
with  FRP  composites:  a  review”,  Proc.,  FRP  Composites  in  Civil  Engineering,  CICE  2004,  Ed.  R.  Seracino, 
Adelaide, Australia, 8­10 December, pp. 705­711. 
Greenland, A.G., Crews, K.I. and Bakoss, S.L. (1999). “Enhancing timber structures with advanced fibre reinforced 
plastic composite reinforcements”. Journal of Structural Engineering, Engineers Australia, Vol. SE2, No. 2 & 3. 
Greenland,  A.G.  (2001).  Applications  Of  Advanced  Fibre  Reinforced  Plastic  Composites  To  Engineered  Timber 
Structures. Master of Engineering Thesis, University of Technology Sydney, Australia. 
Meier, U. (1995). “Strengthening of structures using carbon fibre  / epoxy composites”. Construction and Building 
Materials, Vol. 9, No. 6, pp. 341­351. 
Milner,  H.R.  (1999).  “Reinforced  glulam”.  Proc.,  ACUN­1  International  Composites  Meeting  –  Composites: 
Innovations and Structural Applications, University of New South Wales, Sydney, Australia, 23­25 February 
Tingley,  D.A.  and  Cegelka,  S.  (1996).  “High­strength­fibre­reinforced­plastic  reinforced  wood”.  Proc., 
International Wood Engineering Conference, New Orleans, USA, 28­31 October, Vol. 3 pp. 3­57 to 3­66. 
Vick,  C.B.  (1997).  “More  durable  epoxy  bonds  to  wood  with  hydroxymethylated  resorcinol  coupling  agent”. 
Adhesives Age, Vol. 40, No. 8, pp. 24­29.

680
Third International Conference on FRP Composites in Civil Engineering (CICE 2006)
December 13-15 2006, Miami, Florida, USA

RECENT DEVELOPMENTS ON THE USE OF INORGANIC POLYMER

FOR HIGH TEMPERATURE / HIGH STRENGTH COMPOSITES
James W. Giancaspro, Ph.D.
(Assistant Professor, University of Miami, Coral Gables, FL, USA)

P. N. Balaguru, Ph.D.
(Program Director, National Science Foundation, Washington D.C., USA)

ABSTRACT
Composite materials have become commonplace in the civil engineering community over the past decade.
Typically, organic polymers are used as the matrix for binding the high strength fibers and the composite to other
structural materials such as concrete, brick, and wood. High temperature resistance is one concern in using organic
matrices, especially for structures that could possibly be exposed to fire. The results reported in this paper deal with
the use of an inorganic polymer matrix for both making composite plates and strength/stiffness enhancement of
structural materials. The room-temperature curing matrix, called Geopolymer, can resist temperatures up to 1000°C.
Several applications of this polymer, including its use in strengthening reinforced concrete and wood beams, as well
as its performance after high-temperature exposure are summarized in this paper.

KEYWORDS
Concrete, Fire Protection, High-Temperature, Inorganic, Wood

1. INTRODUCTION
When high service temperatures are expected, composites made using inorganic matrices can be utilized. One such
resin is Geopolymer, a member of the Polysialate family of inorganic matrices known for their low curing
temperatures, high temperature resistance, and low cost. Geopolymer is a two-part system consisting of an alumina
liquid and a silica powder that cures at a reasonably low temperature of 150°C (302°F). In addition, hardeners can
be added to facilitate room temperature curing. Once cured, the matrix can withstand temperatures up to 1000°C
without producing smoke. Geopolymer matrix composites have excellent potential in many applications where
high-use temperatures are anticipated such as engine exhaust systems, or where fire safety is a critical design
parameter such as in aircraft or high-rise buildings. It is compatible with carbon, glass, Kevlar®, steel, cellulose
materials such as wood, and a host of other inorganic materials such as clay bricks and concrete (Lyon, et al., 1997).

Processing requirements and mechanical properties of carbon/carbon composites, ceramic matrix composites made
with silicon carbide, silicon nitride and alumina fibers, and carbon/Geopolymer composites were compiled by
Papakonstantinou et al. to study the relative performance of Geopolymer composites. The extensive and in-depth
study yielded promising results that indicate that carbon/Geopolymer composites have mechanical properties that
are better than most fire-resistant composites (Papakonstantinou et al., 2001).

The effectiveness of Geopolymer has been demonstrated successfully in a number of civil engineering applications
including high-temperature resistant composite panels and strengthening of beams made from reinforced concrete
and timber. Each of these applications is described briefly in the following sections.

681
2. HIGH-TEMPERATURE EXPOSURE
A considerable amount of testing has been carried out to evaluate polysialate matrix composites for mechanical
properties, behavior after high temperature exposure, and durability under various exposure conditions (Foden,
1999; Lyon et al., 1997, Hammell, 2000). In this research program, the effect of temperature exposure on the
mechanical properties of Geopolymer composites was investigated. Carbon composite plates were fabricated using
the procedure and equipment that are utilized for organic composites. Hence, economical, commercially available
fabrication equipment and knowledge such as vacuum assisted impregnation can be utilized for composites made
with polysialates. Once cured, the plates were cut with a diamond blade saw to yield composite specimens about
120-mm in length, 12-mm in width, and a thickness of either 3- or 6-mm depending on the number of plies used.
Three-point flexure tests on the composite specimens were conducted over a simply supported span of 100 mm with
a central load. The load-deflection responses were obtained for virgin samples and samples exposed to 200°, 400°,
600°, and 800°C for one hour in a furnace. Examination of the test results in Figure 1 reveals that even after one
hour of exposure to 800°C, the polysialate composite retains 63% of its original load carrying capacity.

120

6mm
100
3 mm

80
L oa d (% )

60

40

20

0
0 100 200 300 400 500 600 700 800 900
Exposure Temperature (0C)

Figure 1: Residual Flexural Strength After Temperature Exposure for 3- and 6-mm Thick Specimens

3. STRENGTHENING OF REINFORCED CONCRETE BEAMS

The focus of this investigation was to compare the behavior of the inorganic matrix that is brittle to a ductile organic
matrix when used to bond carbon fiber sheets to reinforced concrete beams. The experimental program consisted of
strengthening four reinforced concrete beams using the inorganic carbon composite (Kurtz and Balaguru, 2001). All
four beams were tested as simply supported beams under four-point flexure loading conditions. This experimental
study simulated an earlier study in which beams were strengthened in the same way using an organic matrix. The
results from both programs were then examined to compare the performance of the two types of matrices. The
primary response variables used for comparison included:

• Differences in failure mode

• Magnitude of strength increase over their respective controls
• Magnitude of stiffness increase over their respective controls
• Deflection and ductility
• Crack patterns

The normalized load-deflection curves for all beams are shown in Figure 2. From these curves, it can be seen that
the load-deflection behavior of beams strengthened with inorganic matrix are similar to the beam strengthened with
organic matrix. The beams IS3 and OS, with about the same carbon fiber areas, had comparable strength, stiffness,
and ductility.

The results indicated that the inorganic matrix is just as effective in increasing the strength and stiffness of
reinforced concrete beams as the organic matrix, with a minor reduction in ductility. The failure mechanism
changed from sheet delamination for the organic system to sheet rupture for the inorganic system. This change in
mechanism is attributed to the brittleness of the inorganic matrix that results in crack formation in the composite and
a minimum build-up of strain along the interface of the composite and concrete.

682
 Figure 2: Comparison of Load-Deflection Response for Beams Reinforced with Inorganic and Organic Composite

4. STRENGTHENING OF WOOD BEAMS

Currently, high strength carbon and glass composites are being used to strengthen timber beams and to fabricate
sandwich panels with high specific strengths. Organic polymers are typically used as the matrix for binding the
fabrics and adhesion to wood. Unfortunately, organic matrices are often flammable and leave wooden structural
elements susceptible to rapid damage and collapse in the event of a fire. In this study, structural sandwich beams
were fabricated using the inorganic matrix. Carbon and glass fiber facings were laminated onto end-grain balsa
wood and oak beams and tested in flexure to examine the effect of density on the mechanical properties of the
resultant sandwich beam. The primary variables investigated were:

• Density of core material – balsa (56 to 163 kg/m3) and oak (560 to 826 kg/m3)
• Beam thickness – four depths of 6, 13, 19, and 25 mm (¼, ½, ¾, and 1 in.)
• Type of reinforcement – 12k high modulus carbon tows (640 GPa), woven carbon fabric with glass in
the fill direction made using 3k tows, unidirectional carbon tape made using 3k tows, and 2k alkali-
resistant glass (AR-Glass) tows
• Amount of reinforcement – between zero and four carbon tows; one or two woven carbon tapes; one or
two unidirectional carbon tapes; zero, four, or eight AR-Glass tows
• Location of reinforcement – only on the tension face (T) or on both the tension and compression faces
(T,C)

One hundred and thirty beams were tested in flexure to determine the load-deflection response of these sandwich
beams. Typical load-deflection responses are presented in Figure 3. The beams were analyzed as reinforced wood
beams using the concepts of composite action and flexure theory for linearly elastic and elastic-plastic materials.
The influence of density on flexural strength, stiffness, toughness, and specific strength were also evaluated. Using
the results from the flexure tests, a number of interesting observations were made. First, it was found that lower
density cores provided better performance and exhibited the largest increases in most categories. Flexural theory
using linear elastic material behavior provides good prediction in the linear range. However, non-linear analysis is
needed to predict the strength accurately. As the wood density reduces, the shear stiffness plays a major role in the
failure mechanism. No delamination failures were present, as the bond between the inorganic carbon composite and
both types of wood cores is strong enough to prevent delamination.

The reinforced balsa beams displayed significantly higher increases in stiffness and toughness than the strengthened
oak beams. The flexural strength of the plain wood cores is strongly related to the density and the strength can be
predicted very well with a simple linear regression model. In general, there is no linear relationship between
maximum capacity and density. Similarly, a linear relationship does not exist between deflection at maximum load
and density for the strengthened beams.

683
 10000
2 - Uni Tapes - T,C
9000
1 - Uni Tapes - T,C
4 Tows - T,C
2 Tows - T,C
8000
2 Tows - T
Control
7000

Total Load (N)

6000

5000

4000

3000

2000

1000

0
0 2 4 6 8 10 12 14 16 18 20
Deflection (mm)

Figure 3: Load-Deflection Response for Oak Beams (25mm Thick x 64mm Width) Reinforced with Carbon

5. CONCLUSIONS
Using the analysis of the results presented in the aforementioned studies, the following conclusions can be drawn.

• The Geopolymer composite system is viable in terms of composite fabrication and the bonding the
system to common construction materials.
• The matrix also protects fire susceptible materials. For example, after 1 hour of heat exposure at
800°C, the residual flexural strength of a Geopolymer-carbon composite decreased only by 37%. Note
that carbon fibers start to oxidize at 400°C.
• Reinforced concrete beams strengthened using inorganic composite behave similar to those beams
strengthened with organic composite. The major difference is that none of the beams strengthened with
inorganic-polymer composites failed by delamination of composite.
• As expected, the strengthening system is more effective for weak wood such as balsa, as compared to
strong wood such as oak.

6. REFERENCES
Foden, A. J. (1999). “Mechanical properties and material characterization of Polysialate structural composites.”
Ph.D. Thesis, Rutgers, the State University of New Jersey, USA.

Hammell, J. A. (2000). "The influence of matrix composition and reinforcement type on the properties of
polysialate composites." Ph.D. Thesis, Rutgers, the State University of New Jersey, USA.

Kurtz, S., and Balaguru, P.N. (2001). “Comparison of Inorganic and Organic Matrices for Strengthening of RC
Beams with Carbon Sheets,” Journal of Structural Engineering, January, 35-42.

Lyon, R.E., Balaguru, P., Foden, A.J., Sorathia, U., and Davidovits, J. (1997). “Fire-resistant aluminosilicate
composites.” Fire and Materials, 21, 61-73.

Papakonstantinou, CG, Balaguru, PN, and Lyon, RE, (2001). “Comparative study of high-temperature composites,”
Composites Part B: Engineering, 32(8), 637-49.

684
Third International Conference on FRP Composites in Civil Engineering (CICE 2006)
December 13-15 2006, Miami, Florida, USA

INFLUENCE OF FATIGUE AND AGGRESSIVE ENVIRONMENT ON

THE PERFORMANCE OF RC BEAMS REINFORCED WITH CFRP
MATERIALS
Marco Arduini
CEO - Co-Force s.r.l. – via Pindemonte, 16 – 42100 Reggio Emilia – Italy

Mariano Romagnolo
Autostrade S.p.A. – via Bergamini, 50 – Roma – Italy

Antonio Nanni
Prof. and Chair Dept. Civil, Arch. & Environ. Eng., University of Miami – Miami – USA – nanni@miami.edu

ABSTRACT
The strengthening of RC beams with CFRP laminates has been extensively studied, and design recommendations
have been published in many countries (ACI 440, CNR-DT200, CEB-FIP 01, etc.). Few data are available for Near
Surface Mounted (NSM) FRP strengthening techniques or for concrete beams that are repaired first using
cementitious mortars and then reinforced with FRP materials.
A specific experimental campaign, with more than 200 beams, has been carried out to investigate many factors that
can affect design in the common practice, such as concrete compressive strength, concrete fracture energy, FRP
reinforcement ratio, FRP type, repair technique (plating, NSM, new concrete cover), fatigue, and aggressive
environment. Tests on 4-point bending showed the good performance of FRP strengthening, as the minimum
increment in peak bending load was 110 percent, and the maximum was 430 percent. The influence of fatigue, if the
peak cyclic load is kept below 30 percent of ultimate quasi-static load, is negligible in the overall performance.
Saltwater contact and temperature variation have a more significant effect in reducing ultimate capacity, especially
because the internal steel reinforcement corrodes, and the interface adhesion strength degrades. The effect of
concrete cover replacement was also investigated—the influence of high strength shrinkage-compensated
cementitious mortar is negligible because the interface adhesion with the original concrete is similar to the epoxy-
concrete adhesion. Moreover, the internal steel reinforcement is protected against aggressive environments.

KEYWORDS
CFRP laminates, CFRP rods, experimental results, Near Surface Mounted, plating, repair mortar

1. INTRODUCTION
The strengthening of reinforced and prestressed concrete (RC and PC) beams with FRP laminates applied by manual
lay-up has been studied and reported in recent years and now has become accepted worldwide. Many tests have
been reported using laboratory experimentation.
In practice, the FRP reinforcement can be applied on very old beams, for example, with concrete covers that are
degradated or in very poor condition. Sometimes, the concrete cover is so degradated that it is removed, internal
steel rebars are sandblasted, and new cementitious mortar is applied to recreate the alkaline environment around the
steel rebars. After the curing of the mortar, FRP reinforcement is applied to upgrade the load capacity. In this case,
two potential interface failures can develop—between FRP and the new mortar and between the new mortar and the
original concrete. Conversely, precast PC beams are generally made with high strength concrete C50 to C70, and
surfaces are smooth and pore-free because of the high amount of mold-releasing agents. In this case, FRP must be
attached to the surface after a thorough sandblasting; otherwise, delamination can occur because the resin cannot
penetrate into the concrete surface. In a sense, strengthening a low-quality concrete member is easier because of the
ease in opening micro-pores during sandblasting.

685
2. EXPERIMENTAL CAMPAIGN
The experimental campaign takes into account four different strengthening technologies:
- Type P: CFRP plating using epoxy resin in the tensile face of the concrete beam;
- Type C+P: concrete cover replacement and adhesion of CFRP laminates;
- Type NSM: grooving of the concrete cover, application of epoxy resin, and embedment of CFRP rods (NSM
technique); and
- Type INT: concrete cover replacement and embedment of NSM CFRP rods in the new mortar.
The campaign can be divided in three parts:
- Part A: 42 groups of beams were prepared, cured in a lab environment for 2 months, and then tested to failure
without any preliminary type of degradation (Arduini et al. 2004);
- Part. B: 34 groups of beams were prepared, cured in a lab environment for 2 months, subjected at 10,000 cycles
under a 4-point bending configuration, and then tested to failure; and
- Part C: 35 groups of beams were prepared, cured in a lab environment for 2 months, placed in an aggressive
environment for one year, and then tested to failure.
Concrete, steel reinforcing bars, and CFRP laminates used in this research program had mean and characteristic
properties as reported in Table 1. Symbols used in Table 1 represent the following:
- SFR: cementitious mortars + short hooked steel fibers (7 percent by weight of dry mix, L=3cm);
- fck and fb: cylinder characteristic compressive strength and splitting mean strength, respectively; and
- Gf and Eci: fracture energy according to Rilem TC50-FMC and characteristic initial elastic tangent modulus.
The CFRP sheets were 300 gr/m2 unidirectional sheets. The generic characteristic value, fk, is related to mean value,
fm, obtained in the experimental campaign using the formula fk=fm-α std, where std represents the standard
deviation, α=1.4 for steel and concrete, and α=3 for CFRP materials (according to ACI 440-2).
Four-point bending tons under a quasi-static load were carried out on 1-m clear span. Five progressive cycles were
applied according to the following steps: at first visible crack, 30 percent of steel yield, 70 percent of steel yield, at
steel yield, and around 1.3 steel yield. Two LT transducers provided the mid-span deflection, and strain gauges
were mounted on concrete in the compressive zone, on tensile steel reinforcement, and on the CFRP sheets at the
mid-span section. Concrete beams had a 2 cm concrete cover and 10 stirrups, 6 mm in diameter and equally spaced
to avoid shear failure.
Table 2 presents all relevant information related to specimen characteristics of Part B. The first column shows the
group name, which is a combination of the following parameters:
- Type of concrete used to cast the beam—C20 means that the concrete cylinder characteristic compressive
strength = 20 MPa
- Reinforcement configuration: SAB means sand blasting (Type P), TIXO and S1 mean new concrete cover using
high strength shrinkage compensated cementitious mortars (Type INT or Type C+P), SFR means high strength
shrinkage compensated high ductility concrete, and TAS means NSM reinforcement;
- Type of reinforcement used: 1C5 means 1 layer of high modulus 300 gr/m2 unidirectional carbon fiber sheet
named C5-30 according to Table 1, 1B7 means 1 pultruded CFRP bar 7.5 mm in nominal diameter, and LAM
means a smooth pultruded CFRP lamina; and
- The number and the diameter of internal steel rebars: 2F6 means 2 steel rebars, 6 mm in diameter.
The second column shows the number of samples tested. The following two columns show tensile reinforcement
characteristics, such as steel tensile reinforcement ratio (ρs), defined as the steel area divided by the product of
concrete cross-section width and steel reinforcement depth (As/(bd)), and the FRP reinforcement ratio (ρf), defined
as the CFRP area divided by the product of concrete cross-section width and steel reinforcement depth (As/(bd)).
The fifth column reports the FRP reinforcement dimensions, and the sixth column indicates the type of FRP
reinforcement adopted. The seventh column describes the type of cementitious mortar used to replace the original
bottom concrete cover. The last three columns summarize the main experimental results obtained from 4-point
bending tests in terms of maximum total applied load Fmax, mid-span deflection at failure δmax, and mode of failure
as visually observed during the test. Five different failure modes were observed, as shown in Figure 1.
To observe the influence of fatigue loads on bending capacity of strengthened RC beams, some groups of samples
were duplicated and subjected to 10,000 bending cycles. The minimum load was kept constant to 0.5 tons, and the
maximum load was fixed equal to 1/3 of Fmax reported in Table 2 for each group of similar specimens. The
frequency was kept constant equal to 1Hz. After the cyclic conditioning, beams were tested to failure. Table 2
summarizes the main experimental results as obtained. To observe the influence of aggressive environment on
bending capacity of strengthened RC beams, some groups of samples of Part A were duplicated and subjected to
one-year exposure to sea water and external temperature variation (from -10°C to + 45°C). The sea water level was
monitored every week to guarantee that the tensile face of the beam was partially immersed in the water (water edge

686
condition). CFRP sheets were not protected by UV coating to facilitate the interface degradation. After the natural
external exposure, beams were subjected to an additional 20 cycles in climatic chamber from -10°C to +55°C over a
10-day period of time, and then tested in bending to failure. Table 3 summarizes the main experimental results
obtained.

Table 1: Mechanical characteristic properties of constituents

Concrete fck fb Gf Eci (3) Reinforcement Diameter or ftk Ef εu
2
Type MPa MPa N*m/m GPa thickness
SFR 53 3.6 6760 28.3 mm MPa GPa %
C63 63 3.8 130 33.7 CFRP 7-Rod 7.5 1704 197 0.9
C40 40 3.7 N/A 46.2 CFRP 12-Rod 12 1194 130 1.0
C33 33 3.5 N/A 40.9 CFRP-Laminate 25 x 1.4 2650 160 1.6
C20 20 2.8 N/A 33.8 C5-30- sheet 0.165 2560 400 0.7
TIXO (1) 58 2.5 149 27.0 AM- sheet 0.165 2961 404 1.0
S1 (2) 58 2.8 752 25.0 Steel rods 6 534(4) 200 N/A

Table 2: Part B (preliminary fatigue cycles)

Group Name N ρs ρf CFRP type and Reinf. Mortar Fmax δmax Failure mode
° % % dimension Type type [kN] [mm]
C20-SAB-1C5 2 0.23 0.07 1 layer 0.1x1m C5-30 P NO 34.6 9.0 FRP rupture
C60-SAB-1C5 2 0.23 0.07 1 layer 0.1x1m C5-30 P NO 35.9 7.6 FRP rupture
SFR-SAB-1C5 2 0.23 0.07 1 layer 0.1x1m C5-30 P NO 48.6 9.1 FRP rupture
C20-SAB-1AM 1 0.23 0.07 1 layer 0.1x1m AM P NO 39.3 13.3 Peeling + crushing
C60-SAB-1AM 1 0.23 0.07 1 layer 0.1x1m AM P NO 30.0 6.6 Peeling
SFR-SAB-1AM 1 0.23 0.07 1 layer 0.1x1m AM P NO 47.6 11.8 Peeling + crushing
C20-SAB-3C5 2 0.23 0.21 3 layers 0.1x1m C5-30 P NO 45.9 5.7 Peeling
C60-SAB-3C5 2 0.23 0.21 3 layers 0.1x1m C5-30 P NO 55.6 6.0 Peeling
SFR-SAB-3C5 2 0.23 0.21 3 layers 0.1x1m C5-30 P NO 72.7 8.2 Peeling
C20-SAB-3AM 1 0.23 0.21 3 layers 0.1x1m AM P NO 44.3 6.1 Peeling + crushing
C60-SAB-3AM 1 0.23 0.21 3 layers 0.1x1m AM P NO 60.3 7.3 Peeling
SFR-SAB-3AM 1 0.23 0.21 3 layers 0.1x1m AM P NO 64.1 7.7 Peeling
C20-TIXO-1C5 3 0.23 0.07 1 layer 0.1x1m C5-30 C+P Tixo 3cm 34.4 7.0 FRP rupture
C60-TIXO-1C5 3 0.23 0.07 1 layer 0.1x1m C5-30 C+P Tixo 3cm 35.9 7.1 FRP rupture
C33-S1-1C5 3 0.23 0.07 1 layer 0.1x1m C5-30 C+P S1 3cm 37.5 8.7 Delamin.+FRP rupture
C60-S1-1C5 3 0.23 0.07 1 layer 0.1x1m C5-30 C+P S1 3cm 36.4 7.1 Peeling
C33-NOR-1C5 2 0.23 0.07 1 layer 0.1x1m C5-30 C+P NOR 3cm 34.0 7.4 FRP rupture
C60-TAS-1B7 2 0.23 0.21 1 Mbar φ 8 L=1.0m NSM Putty 59.1 20.1 crushing + Rod slippage
SFR-TAS-1B7 2 0.23 0.21 1 Mbar φ 8 L=1.0m NSM Putty 55.4 14.2 crushing + Rod slippage
C20-TAS-LAM 1 0.23 0.15 1/2 Laminate L=1.0m NSM Putty 33.2 10.2 crushing + Rod slippage
C60-TAS-LAM 1 0.23 0.15 1/2 Laminate L=1.0m NSM Putty 40.0 11.0 Laminate slippage
SFR-TAS-1B12 1 0.23 0.47 1 Mbar φ 12 L=1.0m NSM Putty 65.9 10.2 crushing + Rod slippage
C20-TIXOF-1B7 2 0.23 0.19 1 Mbar φ 8 L=1.0m INT Tixo 3cm 39.0 11.9 crushing + Rod slippage
C33-S1-1B7 2 0.23 0.19 1 Mbar φ 8 L=1.0m INT S1 3cm 43.0 12.5 crushing + Rod slippage
C60-TIXO-3B7 2 0.23 0.58 3 Mbar φ 8 L=1.0m INT Tixo 3cm 68.0 7.1 Mortar delamination
C60-S1-3B7 2 0.23 0.58 3 Mbar φ 8 L=1.0m INT S1 3cm 68.5 8.4 Mortar delamination
C33-TIXO-1B12 1 0.23 0.43 1 Mbar φ 12 L=1.0m INT Tixo 3cm 52.1 7.9 Rod slippage
C33-S1-1B12 1 0.23 0.43 1 Mbar φ 12 L=1.0m INT S1 3cm 50.8 9.1 Rod slippage

3. CONCLUSIONS
An experimental program consisting of more than 200 RC beams strengthened with external CFRP sheets or NSM
CFRP rods was presented. Different types of concrete, FRP, and concrete over rehabilitation mortar were taken into
account together with fatigue to show an environmental attack.
Experimental data shows good performance of the strengthening, and the minimum increment in peak bending load
was 110 percent, while the maximum was 430 percent. The influence of fatigue, if the peak cyclic load is kept
below 30 percent of ultimate load, is negligible in the final overall performance. Salt water contact and temperature
variation have a more significant effect in reducing ultimate capacity, especially because the internal steel

687
reinforcement can corrode together with the degradation of the interface adhesion. The effect of concrete cover
replacement was also investigated. The influence of the cementitious mortar layer is negligible because the interface
adhesion with concrete is similar to the epoxy-concrete adhesion.
In design guidelines, the safety factor should not be related to FRP properties only, but needs to be calibrated on the
interface adhesion properties and related to concrete properties.

Figure 1 – Different failure modes

Table 3: Part C (one year aggressive environment)

Name N ρs ρf CFRP type and Reinf. Mortar Fmax δmax Failure mode
° % % dimension Type type [kN] [mm]
C20-SAB-1C5 2 0.23 0.07 1 layer 0.1x1m C5-30 P NO 30.3 6.8 Delamin.+FRP rupture
C60-SAB-1C5 2 0.23 0.07 1 layer 0.1x1m C5-30 P NO 24.5 4.3 Delamination
SFR-SAB-1C5 2 0.23 0.07 1 layer 0.1x1m C5-30 P NO 42.8 6.9 Delamin.+FRP rupture
C20-SAB-1AM 1 0.23 0.07 1 layer 0.1x1m AM P NO 33.2 9.5 Delamination
C60-SAB-1AM 1 0.23 0.07 1 layer 0.1x1m AM P NO 28.5 5.4 Delamination
SFR-SAB-1AM 1 0.23 0.07 1 layer 0.1x1m AM P NO 40.0 6.5 Delamination
C20-SAB-3C5 2 0.23 0.21 3 layers 0.1x1m C5-30 P NO 28.2 3.2 Peeling
C60-SAB-3C5 2 0.23 0.21 3 layers 0.1x1m C5-30 P NO 61.7 6.0 Peeling
SFR-SAB-3C5 2 0.23 0.21 3 layers 0.1x1m C5-30 P NO 43.9 8.2 Peeling+FRP rupture
C60-SAB-3AM 1 0.23 0.21 3 layers 0.1x1m AM P NO 46.6 4.7 Peeling
SFR-SAB-3AM 1 0.23 0.21 3 layers 0.1x1m AM P NO 61.4 7.2 Peeling
C20-TIXO-1C5 3 0.23 0.07 1 layer 0.1x1m C5-30 C+P Tixo 3cm 34.8 7.8 FRP rupture
C60-TIXO-1C5 3 0.23 0.07 1 layer 0.1x1m C5-30 C+P Tixo 3cm 35.8 6.6 FRP rupture
C33-S1-1C5 3 0.23 0.07 1 layer 0.1x1m C5-30 C+P S1 3cm 38.9 8.9 FRP rupture
C60-S1-1C5 3 0.23 0.07 1 layer 0.1x1m C5-30 C+P S1 3cm 34.3 5.5 FRP rupture
C33-NOR-1C5 2 0.23 0.07 1 layer 0.1x1m C5-30 C+P NOR 3m 32.9 7.3 FRP rupture
C20-TAS-1B7 2 0.23 0.21 1rod φ 8 L=1.0m NSM Putty 42.6 16.5 crushing + Rod slippage
C60-TAS-1B7 2 0.23 0.21 1rod φ 8 L=1.0m NSM Putty 61.1 19.8 crushing
SFR-TAS-1B7 2 0.23 0.21 1 rod φ 8 L=1.0m NSM Putty 55.1 15.5 crushing + Rod slippage
C20-TAS-LAM 1 0.23 0.15 1Lam25x1.4mmL=1m NSM Putty 34.5 16.8 crushing
C60-TAS-LAM 1 0.23 0.15 1Lam25x1.4mmL=1m NSM Putty 30.3 6.8 Laminate slippage
SFR-TAS-1B12 1 0.23 0.47 1rod φ 12 L=1.0m NSM Putty 53.9 7.3 Rod slippage
C20-TIXO-1B7 2 0.23 0.19 1rod φ 8 L=1.0m INT Tixo 3cm 41.6 14.4 crushing + Rod slippage
C33-S1-1B7 2 0.23 0.19 1rod φ 8 L=1.0m INT S13cm 44.8 13.1 Rod slippage
C60-TIXO-3B7 2 0.23 0.58 3rods φ 8 L=1.0m INT Tixo 3cm 64.8 7.7 Rod slippage
C60-S1-3B7 2 0.23 0.58 3rods φ 8 L=1.0m INT S1 3cm 75.44 7.8 Mortar delamination
C33-TIXO-1B12 1 0.23 0.43 1rod φ 12 L=1.0m INT Tixo 3cm 55.6 8.0 Rod slippage
C33-S1-1B12 1 0.23 0.43 1rod φ 12 L=1.0m INT S1 3cm 55.9 10.0 Rod slippage

4. REFERENCES
Arduini M., Romagnolo M., and Nanni A., (2004). “Influence of concrete quality on the performance of FRP
strengthened RC beams: Experiments”. Proc. 4th International Conference on Advanced Composite Materials in
Bridges and Structures, Calgary, Alberta, 20-23, 2004.

688
Third International Conference on FRP Composites in Civil Engineering (CICE 2006)
December 13-15 2006, Miami, Florida, USA

BOND STRENGTH OF CARBON FIBER REINFORCED POLYMERS ON

CONCRETE AND MASONRY
Umit Serdar Camli
(Graduate Student, Middle East Technical University, Ankara,Turkey)

Baris Binici
(Assistant Professor, Middle East Technical University, Ankara,Tukey)

ABSTRACT
This study presents the results of an experimental program conducted to determine the strength of carbon fiber
reinforced polymers (CFRPs) bonded to low and normal compressive strength concrete and masonry units that are
finished with or without plaster. In the experimental program, a total of 51 specimens with different types of
anchorages (strip and embedded anchors) were tested. Test results showed that ultimate strength of CFRPs bonded
to plaster finished specimens were significantly lower than those without plaster finish. It was found that ultimate
strength of surface bonded CFRPs with the use of embedment at the free end and FRP dowels that act as shear
connectors can result in strength enhancements as high as three times of those without any special anchorage. Based
on slip measurements, a simple and effective strength model is calibrated and proposed for strip type anchorages.

KEYWORDS
bond strength, concrete, masonry, plaster

1. INTRODUCTION
Widespread utilization of FRPs has started in Turkey for seismic and non-seismic retrofit applications (i.e.
strengthening of beams and columns for shear and in and out of plane strengthening of masonry infill panels with
surface bonded FRPs). Furthermore a significant increase in FRP applications is expected in the upcoming years as a
result of the introduction of FRP retrofit design guidelines in the draft version of the Turkish Seismic Code.
Although use of FRPs are promising due to their advantages such as being light weight, high strength and easy to
apply, a number of additional parameters require further investigation for their successful use in certain countries
depending on the construction practice. For example, plaster finishing on concrete structural elements and masonry
infill panels are commonly used in Turkey and special attention is needed when designing and bonding FRPs.
Furthermore, low concrete strength (10 to 15 MPa) that may be observed in some of the deficient structural elements
has the potential to adversely affect the success of the FRP retrofit. In addition, application of FRPs in retrofit of
masonry infill walls to strengthen in and out of plane resistance necessitates knowing the bond strength of FRP to
masonry, generally in the form of hollow clay tiles.
The objective of this paper to examine the strength of CFRPs bonded to low strength concrete and masonry finished
with and without plaster. In addition, embedded type anchors are investigated, which can have the potential to
enhance strength of surface bonded FRPs to concrete. To achieve this objective, an extensive experimental program
was conducted. Based on the results of experiments, a simple analytical model calibrated using experimental results
is proposed.

2. EXPERIMENTAL PROGRAM

A simple test setup that was successfully used in previous studies (Binici and Bayrak 2004, Dolan et. al. 1998) to
examine bond strength of CFRP bonded to concrete was employed. Sideway and plan views of the experimental
setup is shown in Figure 1. The test specimens were built by bonding two symmetrically located CFRP laminates
horizontally along the center line of two similar concrete prisms or masonry, referred hereafter as hollow clay tile

689
(HCT). Load was applied using a hydraulic jack through the centerline of two prisms such that two blocks are
pushed against each other while imposing shear stresses at the CFRP-block interfaces. Two steel plates were located
on loading faces of the specimens to avoid local failures of the specimens and to distribute stresses uniformly.
Special attention was given to eliminate any possible eccentricities on CFRP laminates that could cause premature
failures. Displacements were continuously monitored using four dial gages that were located at the outer edges of
the blocks as shown in Figure 1.

Testing of concrete specimens

Testing of hollow clay tile (HCT)

Figure.1 Test Setup

For low strength concrete, target compressive strength was 10 MPa, whereas for normal strength concrete, target
uniaxial compressive strength of 30MPa was employed. The uniaxial compressive strength was found as 6 MPa
considering the gross area of HCTs. Carbon fiber reinforced polymers (CFRPs) have been used throughout the test
program. The ultimate tensile strength of CFRP having a thickness of 0.165 mm was 3430 MPa with a rupture strain
of 0.015 according to manufacturer’s report.
Prior to application of the CFRPs, surfaces of concrete, HCT blocks, or plaster finished surfaces were cleaned from
dust by air-blowing. Then, CFRP sheets, cut to predetermined length and width, were impregnated into epoxy resin
and bonded on the sides of the blocks. Wooden plywood sheets covered with plastic nylon were attached and C-
clamps were used to keep the CFRP sheets in correct position while curing of the epoxy. After curing of epoxy,
wooden blocks were removed and testing equipment was attached without moving the specimens.
Three types of CFRP application details that were used for the test specimens are shown in Figure 2. For strip
anchors, CFRP sheets were bonded to the sides of concrete or masonry blocks without any special attachment detail
(Figure 2a). For embedded anchors, different details were used for concrete and HCT units. In concrete specimens,
free end of the CFRP was embedded in a hole drilled in concrete prisms (Figure 2b). For hollow clay tile
applications, it was aimed to attach CFRP dowels as shear connectors to arrest the CFRP debonding process. CFRP
anchor dowels were prepared by rolling CFRP sheets around a steel wire and impregnating them into epoxy (Figure
2c). These CFRP dowels were then passed through the predrilled holes and fanned out on the CFRP strips. The
effect of embedment depth for concrete specimens and number of CFRP dowels hollow clay tiles were taken as test
parameters to observe the enhancement of bond strength for CFRP-adherent interface. In the experimental program,
51 double shear push out tests were performed to systematically investigate the effect of parameters, such as
strength of the concrete prisms or HCTs (fc), presence of plaster, CFRP width (bfrp), CFRP bond length (Lfrp),
embedment depth (dfrp), anchor type (strip, and embedded anchors).

(a) Strip anchor

(c)Embedded anchors for HCT

(b)Embedded anchors for concrete

Figure 2. Anchorage types used in the experimental program

690
3. TEST RESULTS

The summary of experimental results in the form of normalized strength (Ptest/Pfrp where Pfrp is the uniaxial tensile
strength of bonded CFRP sheet) versus bond length are presented in Figure 3. Results of strip anchor tests showed
that with increasing bonded length, Lfrp, load carrying capacity increased up to a certain length (~100 mm) beyond
which no strength enhancement occurs. The corresponding maximum strength was found as 40% and 34% of the
uniaxial load carrying capacity of the CFRPs for 25 mm and 50 mm wide strips, respectively. From this group of
tests, it was evident that increase in width of the CFRP laminates resulted in a decrease of normalized strength. The
failure mode for all specimens in this group was debonding of the CFRP from the concrete surface while a thin
concrete layer remained attached to the debonded CFRP sheet.
CFRPs bonded to HCT specimens had comparably lower strength compared to concrete specimens with similar
details. Maximum normalized strength of 31% was obtained for a CFRP width of 25 mm and 75 mm bond length
among all specimens. The effective bond length, which differed due to the discontinuities of the surface of masonry
units, was found to be between 75 and 100 mm according to normalized strength-CFRP bond length behavior of test
results. All the test specimens with strip anchors on HCT exhibited debonding from the HCT surfaces with a small
chunk of tile remaining attached on CFRPs. Results of CFRP bonded to HCT specimens indicated that
discontinuities on HCT texture and weak nature of HCTs due to presence of cores are the important factors that can
prevent successful bonding of CFRPs, resulting in lower ultimate bond strength. Therefore, texture of the masonry
units, which are different than that of concrete surfaces, were observed to significantly influence the bond strength
of CFRPs.
Results of tests on specimens with plaster finish showed that presence of a low strength (~5MPa) thin plaster
layer (~10 mm) adversely affected the ultimate strength, which was found to be about 10 to 20 % of the CFRP
uniaxial tensile strength. As the strength of plaster was same for all three groups, similar strength values were
obtained irrespective of concrete or masonry compressive strength. Results of experiments conducted on embedded
anchors revealed that embedment enables the strength of CFRPs to be developed up to about 65% of their ultimate
strength irrespective of the embedment depth and concrete strength. However, it was not possible to fully utilize the
strength of CFRPs due to stress concentrations occurring around smoothened corners. In the presence of plaster,
strength was three to four times that of the specimens without any special anchors and compressive strength of
blocks did not affect the ultimate shear capacity of anchorages. It was also observed that presence of plaster reduced
the strength by about 30% for specimens with embedded anchors. The strength of CFRPs bonded to HCTs equipped
with CFRP dowels were comparable to the strength of CFRPs bonded to concrete using embedded anchors.
0.5 0.5 0.5
fc = 10MPa fc = 30MPa HCT
0.4 (No plaster) 0.4 (No plaster) (No plaster)
0.4
Strip Strip strip
0.3 0.3 0.3
Ptest /Pfrp
Ptest /Pfrp

Ptest /Pfrp

0.2 0.2 0.2

0.1 bbfrp=25 mm 0.1 bfrp=25

b frp mm 0.1 bfrp
b frp =25 mm
frp
bbfrp=50
frp mm b frp
bfrp=50 mm b frp =50 mm
bfrp
0 0 0
40 60 80 100 120 140 160 40 60 80 100 120 140 160 40 60 80 100 120 140 160
FRP anchorage length, L frp (mm) FPR anchorage length, L frp (mm) FRP anchorage length, L frp (mm)

0.3 1.0 1.0

bfrp=25mm bfrp=25mm bfrp=25mm
(With plaster) 0.8 (No plaster) 0.8 (With plaster) fGroup 10
c=10MPa
Strip Embedded Embedded HCT
Group 11
0.2
P test /P frp

0.6 0.6
P test /P frp
Ptest /Pfrp

0.4 0.4
0.1
fGroup 7
c=10MPa
fc=10MPa4
Group 0.2 0.2
fc=30MPa5 fGroup 8
c=30MPa
Group
HCT
Group 6 Group 9
HCT
0 0.0 0.0
40 60 80 100 120 140 160 0 20 40 60 80 100 120 140 0 20 40 60 80 100 120 140
FRP anchorage length, L (mm) FRP anchorage depth, dLfrpfrp (mm)
anchorage Length, FRP depth,Lfrpd (mm)
anchorageLength, frp (mm)
FRP anchorage Length, Lfrpfrp (mm) FRP (mm) FRP anchorage

Figure 3. Test Results for Strip and Embedded Type CFRP Anchors

691
4. STRENGTH MODEL
Based on the examination of test results and employing an effective bond stress law following equation is proposed
to estimate the strength of FRPs ( Pu ) bonded to concrete and masonry:
τf L frp
Pu = τ f δ u E frp t frp b frp tanh( ) (1)
δ u f c Le
where δ u is the ultimate slip displacement, E frp is the modulus of elasticity, t frp is the thickness, b frp is the width of
FRPs and ( Le = E frp t frp f c ) is the effective bond length similar to that given by Chen and Teng (2001) beyond
which no strength enhancement is obtained. It is interesting to observe that presence of “tanh” term in Equation 4
conveniently helps to incorporate the effective length concept into the formulation. In this study, shear strength
( τ f = ω f c 0.19 ) originally proposed by Savioa et. al. (2003) is adopted to estimate surface shear strength of concrete
bonded joints as it is simple and has proved to yield satisfactory strength estimations. In their original model, for
FRPs bonded to concrete ω was proposed to be 3.5. For HCTs and plastered surfaces ω is taken as 2 in this study.
The maximum measured slip at onset of debonding is assumed to be a function of compressive strength, ratio of
anchor length to effective length and a width ratio. A nonlinear regression analysis is then performed based on a
database of 103 experiments (including this study and others) that reported slip displacements to estimate δu for
β γ
different cases in form: δ u = f cα ⎛⎜ L frp ⎞⎟ ⎛⎜ b frp ⎞⎟ , where bc is the width of the concrete block. Constants α, β and
⎝ Le ⎠ ⎝ bc ⎠
γ are then found from nonlinear regression analysis. Values of α, β are found to be -0.4 and 0.8, respectively,
irrespective of the material type or presence of plaster. On the other hand, γ is 0.4, 0.5, and 0.9 for concrete HCT,
and plastered surfaces, respectively. The average of estimated to experimental bond strength ratios was found to be
0.99 with a standard deviation of 0.2. The model estimations of bond strength was also compared with strength
results obtained from 159 experiments reported in the literature. The ratio of estimated to test capacities was found
to be 1.03 with a standard deviation of 0.2 (Camli, 2005). Furthermore, the recommended ultimate normalized
strength values for embedded anchors with and without plaster are shown in Figure 3 with a straight line.

5. CONCLUSIONS
The experimental results presented herein show that bond strength is more sensitive to anchorage length and width
compared to concrete or masonry compressive strength. Increasing the bond length up to effective bond length, load
carrying capacity of the anchor increased and remained constant beyond the effective length. Increasing bonded
width resulted in a decrease of the normalized load carrying capacity of the anchors. This phenomenon remained
valid for low strength concrete and masonry. Lower bond strengths were observed for hollow clay tile specimens
due to weak tile texture and discontinuities on the tile surface. The presence of a thin plaster layer resulted in bond
strength as low as one third of the similar specimens without any plaster. Based on these results, it is possible to say
that FRP applications without any anchors on plaster finished surfaces should be conducted with special care. If
possible, FRP dowels should always be supplied to secure the bonded FRP sheets on the plastered surfaces. A
simple strength model, whose parameters were calibrated based on the experiment presented in this study and those
in the literature, was verified by the tests results of other researchers.

6. REFERENCES
Binici B., and Bayrak, O. (2004). “Strength of cabron fiber reinforced polymers bonded to concrete.” 11th US-Japan
Conference on Composite Materials, September 9-11, Yamagata University, Yonezawa, Yamagata, Japan.
Chen, J.F. and Teng, J.G. (2001). “Anchorage strength models for FRP and steel plates bonded to concrete.” ASCE,
Journal of Structural Engineering, Vol.127, No.7, pp 784-791.
Dolan, B., Hamilton, H.R., and Dolan, C. (1998). “Strengthening with FRP lamina.” ACI, Concrete International,
Vol. 20, No.6, 1998, 51-56.
Camli, U.S. (2005). “Anchorage strength of fiber reinforced polymers.” M.Sc Thesis, Middle East Technical
University, 89 p.
Savioa M., Farracuti B., Mazzottti D. (2003). “ Non-linear bond-slip for FRP concrete interface”, Proc. of 6th
International Symposium on FRP Reinforcement for Concrete Structures, Singapore, pp 183-192.

692
Third International Conference on FRP Composites in Civil Engineering (CICE 2006)
December 13-15 2006, Miami, Florida, USA

COMPRESSIVE STRESS-STRAIN MODELS OF

FRP-CONFINED CONCRETE
Liu Mingxue
(PhD candidate, Key Laboratory for Structural Engineering and Vibration of Education Ministry, Department of
Civil Engineering, Tsinghua University, Beijing 100084 ,China)

Qian Jiaru
(Professor, Key Laboratory for Structural Engineering and Vibration of Education Ministry, Department of Civil
Engineering, Tsinghua University, Beijing 100084 ,China)

ABSTRACT
To establish compressive stress-strain models of concrete confined by fiber reinforced polymer (FRP), test results of
304 FRP-confined concrete cylinders have been collected from open literatures. The study of the test results shows
that when the fiber characteristic value of FRP sheet confined concrete cylinders is less than 0.18, the axial
compressive stress-strain curve has an ascending branch as well as a descending branch. When the FRP tube and the
concrete core of FRP tube confined concrete cylinders are axially loaded together, most of the existing models
underestimate the initial modulus and overestimate the ultimate strain of concrete. Two compressive stress-strain
models of FRP-confined concrete are proposed in this paper. One is hardening model which consists of a parabolic
section and a straight line, the other is softening model which has a parabolic section only. Influences of fiber
characteristic value, FRP laminate structure and loading method on the models have been taken into account.

KEYWORDS
fiber reinforced polymer (FRP), concrete cylinder, experimental database, compressive stress-strain model

1. INTRODUCTION
To give a favorable prediction of compressive stress-strain relationship of FRP-confined concrete (FCC), based on
the experimental database built by Lam, L., and Teng, J.G. (2002, 2003) and 116 test results (Berthet et al., 2005;
Fam and Rizkalla, 2001; Huang et al., 2002; Jin et al., 2003; Karabinis and Rousakis, 2002; Lam and Teng, 2004; Li
et al., 2005; Lu, 2005; Tao et al., 2005), a new experimental database which has 304 FRP confined concrete
cylinders (FCCC) has been developed. Based on the new database and the existing models, two improved
compressive stress-strain models of FCC, one is hardening model and the other is softening model, are proposed in
this paper. Influences of FRP laminate structure and loading method are taken into account. Two loading methods
were adopted in existing tests, loading method 1(LM1) was to apply the axial force to the concrete core only and
loading method 2(LM2) was to apply the axial force to the FRP tube and concrete core simultaneously.

2. PROPOSED STRESS-STRAIN MODELS

2.1 Equations of Proposed Models

As shown in Figure 1, the stress-strain relationship curve of the hardening model has an ascending branch which
consists of a parabolic section and a straight line. The stress-strain relationship curve of the softening model has an
ascending branch and a descending branch and it has a parabolic section only. In the Figure, f’cc is the peak strength
of FCC, f’co is the axial compressive strength of unconfined concrete, εcc is the peak strain of FCC, f0 is the stress at
the intersection point of the straight line and the stress axis, εt is the strain where the parabolic section meets the
straight line with a smooth transition, E1 is the elastic modulus of FCC, E2 is the slope of the straight line.

693
 σc
f cc' σc
f0 1
E2 fcc'
fco' fco' E1
E1
1 εc 1 εc
0 εt ε cc 0 ε cc
(a) Hardening Model (b) Softening Model
Figure 1: Proposed Stress-Strain Models for FCC
The hardening model adopts the equations proposed by Lam and Teng (2003), as given by the following
expressions:
( E − E2 ) ε 2
2

σ c = E1ε c − 1 c
(0 ≤ ε c ≤ ε t ) (1)
4 f0
σ c = f0 + E2ε c (ε t ≤ ε c ≤ ε cc ) (2)
2 f0
εt = (3)
E1 − E2
f cc' − f 0
ε cc = (4)
E2
Among the total 304 test results, 271 are FRP tube confined concrete cylinders (FTCCs) and FRP sheet confined
concrete cylinders (FSCCs) whose fiber characteristic values (fr/f’co) are not less than 0.18. Except one FTCC
specimen of Fam and Rizkalla (2001), for which pultruded FRP tube was used, the stress-strain curves of them are
all hardening type. With the exclusion, a total of 270 test results are adopted for developing the hardening model. In
the expression of fr/f’co, fr =2fht/D, t is the thickness of FRP wrap, D is the diameter of concrete cylinder, fh is the
hoop tension strength of FRP wrap. Generally, the rupture strain and the strength of FRP measured in FCCC test are
lower than those from flat coupon tensile test, split-disk test, classical lamination theory, and manufacturer data
(Lorenzis and Tepfers, 2003; Lam and Teng, 2004). But in FTCC tests (Huang et al., 2002; Li et al., 2005; Lu, 2005)
whose FRP tubes were cross-ply filament-wound FRP tube (CPFT), and the narrow ring specimens used to evaluate
the mechanical properties of FRP tubes in split-disk test were cut from the tube, the rupture strains of FRP measured
by FCCC test were higher than those from split-disk test. By analyzing the test results, fh is 1.75 fh,frp in this case
and is fh,frp in the rest cases. Then, the applicable ranges of Equations (1, 2) are fr/f’co≤2.2 for FTCCs and 0.18≤
fr/f’co≤2.2 for FSCCs, respectively.

The softening model is given by the following expression:

E12 2
σ c = E1ε c −εc (5)
4 f cc'
The peak strain εcc of the softening model is 2f’cc /E1. Other four parameters need to be defined are: elastic modulus
E1 of FCC, peak strength f’cc of FCC, stress f0, and slope E2 of the straight line.

2.2 Determination of Parameters of Proposed Models

Under LM2 condition, E1 of concrete of FTCCs is given by the following equation:

E1 = ( Ec + ρ Ec,frp ) /(1 + ρ ) (6)

Where, the elastic modulus of unconfined concrete Ec = 4730 f co' (Lam and Teng 2003), Ec,frp is the axial
compression elastic modulus of FRP tube, ρ is the section area ratio of FRP tube to concrete core. Except LM2
condition E1 equals to Ec.

As the number of layer of FRP sheet or the thickness of FRP tube exceeds a certain value, the confinement
efficiency of FRP decreases. The peak strength f’cc of 270 hardening type FCC specimens can be predicted by a
parabolic equation, as given by the following expression:
f cc' = [1 + 2.721( f r / f co' ) − 0.522( f r / f co' ) 2 ] f co' (7)
The statistics correlation coefficient R2 of Equation (7) is 0.8201.

694
The value of fr/f’co was less than 0.18 for other 33 FSCC specimens. Among them, 21 specimens (Berthet et al 2005;
Demers and Neale 1994; Jin et al. 2003; Karabinis and Rousakis 2002; Xiao and Wu 2000) had softening type
stress-strain curve. By analyzing those 21 test results, the peak strength f’cc of softening model is given by the
following equation:
f cc' = [1 + 0.87( f r / f co' )] f co' (8)
The statistics correlation coefficient R2 of Equation (8) is 0.3479. The average ratio of the test results to the
calculated values of Equation (8) is 1.006 with a standard deviation of 0.059.

The stress f0 can be predicted by the following equation:

f 0 = (1 + 0.303 f r / f co' ) f co' (9)
2
The statistical correlation coefficient R of Equation (9) is 0.2455. The average ratio of the test results to the
calculated values of Equation (9) is 1.017 with a standard deviation of 0.142.

The expression of E2 of concrete confined with CPFT is given by:

E2 = [0.0144 + 0.0438( f r / f co' )]E1 (10)
And E2 of concrete confined with other type of FRP laminate structure is given by:
E2 = (0.02 + 0.08 f r / f co' − 0.18) E1 (11)
2
The statistics correlation coefficient R of Equations (10, 11) is 0.8215 and 0.2948, respectively.

3. COMPARISON OF PROPOSED MODELS WITH TEST RESULTS

The proposed FRP confined concrete stress-strain models are compared with the test results obtained by Xiao and
Wu (2000) on FSCCs, as shown in Figure 2(a, b); by Samaan et al. (1998) on FTCCs, as shown in Figure 2(c); by
Lu (2005) on FTCCs, as shown in Figure 2(d). It can be seen from Figure 2 that the predictions have good
agreement with the test results of FCC with various FRP laminate structures and loading methods.
60 100
Axial stress σ c(MPa)
Axial stress σ c(MPa)

50 f c ' o=33.7M Pa
'
f c o =43.8MPa 80 t =1.14mm
40
t =0.38mm 60
30 Proposed model Proposed model
20 Specimen 1 40 Specimen 1
Specimen 2 Specimen 2
10 20 Specimen 3
Specimen 3
0 0
0 0.003 0.006 0.009 0.012 0 0.01 0.02 0.03
Axial strain ε c Axial strain ε c

(a) FSCC, fr/f’co =0.178 (Softening Model) (b) FSCC, fr/f’co =0.693 (Hardening Model)
80
Axial stress σ c(MPa)

' 70
Axial stress σ c (MPa)

f c o =31.97MPa
60 60
t =2.2mm
50
f c'o=39.96MPa
40 40
t =2.28mm
Proposed model
30
20 Specimen 1 20 Proposed model
Specimen 2 10 Specimen 1
0 0
0 0.01 0.02 0.03 0.04 0 0.01 0.02 0.03 0.04
Axial strain ε c Axial strain ε c
(c) FTCC (CPFT, LM1), fr/f’co =0.523 (Hardening Model) (d) FTCC (CPFT, LM2), fr/f’co =0.217(Hardening Model)

Figure 2: Comparison of Proposed Models with Test Results

695
4. CONCLUSIONS

The following conclusions can be drawn from this study:

· The compressive strength of FRP-confined concrete is relevant to the fiber characteristic value. Influence of FRP
laminate structure and loading method on it is not significant.
·The fiber characteristic value, the FRP laminate structure and the loading method affect the peak strain of FRP
confined concrete.
·The proposed hardening model and softening model have good agreement with the test results of FRP confined
concrete cylinders with various FRP laminate structures and loading methods.

5. ACKNOWLEDGEMENT
The work presented in this paper has received financial support from the National Natural Science Foundation of
China for Distinguished Young Scholars (Project No. 50329802). The authors wish to express their sincere gratitude
to the organization.

6. REFERENCES
Berthet, J.F., Ferrier, E., and Hamelin, P. (2005). “Compressive behavior of concrete externally confined by
composite jackets. Part A: experimental study”. Construction and Building Materials, Vol. 19, pp 223-232.
Demers, M, and Neale, K. W. (1994). "Strengthening of concrete columns with unidirectional composite sheets".
Development in Short and Medium Span Bridge Engineering '94, Proceedings of 4th International Conference on
Short and Medium Bridges, Editors: Mufti, A. A., Bakht, B., and Jaeger, L. G., Canadian Society for Civil
Engineering, Montreal, pp 895-905
Fam, A.Z. and Rizkalla, S.H. (2001). “Behavior of axially loaded concrete-filled fiber-reinforced polymer circular
tubes”. ACI Structural Journal, Vol. 98, No. 3, pp 280-289.
Huang, L.N., Zhang, D.X., Wang, R.G. et al. (2002). “Research on the stress-strain relation of the GFRP-confined
concrete column under the axial compression”. Journal of Wu Han University of technology, Vol. 24, No. 7, pp
31-34. (in Chinese)
Jin, X.N., Pan, J.L., Liu, G.Y., and Lai, W.H. (2003). “Stress-strain curves of concrete confined by fiber reinforced
plastics under axial compression”. Journal of Building Structures, Vol. 24, No.4, pp 47-53. (in Chinese)
Karabinis, A.I., and Rousakis, T.C. (2002). “Concrete confined by FRP material: a plasticity approach”.
Engineering Structures, Vol. 24, pp 923-932.
Lam, L., and Teng, J.G. (2002). “Strength models for fiber-reinforced-plastic-confined concrete”. Journal of
Structural Engineering, ASCE, Vol. 128, No.5, pp 612-622.
Lam, L., and Teng, J.G. (2003). “Design-oriented stress-strain model for FRP-confined concrete”. Construction and
Building Materials, Vol.17, pp 471-489.
Lam, L., and Teng, J.G. (2004). “Ultimate condition of fiber reinforced polymer-confined concrete”. Journal of
Composites for Construction, ASCE, Vol. 8, No. 6, pp 539-548.
Li, G.Q., Tores, S., Alaywan, W., and Abadie, C. (2005). “Experimental study of FRP tube-encased concrete
columns”. Journal of Composite materials, Vol. 39, No.13, pp 1131-1145.
Lorenzis, L. D., and Tepfers, R. (2003). "Comparative study of models on confinement of concrete cylinders with
fiber-reinforced polymer composites". Journal of Composites for Construction, ASCE, Vol. 7, No. 3, pp 219-237.
Lu, G.C. (2005). “Study on behavior of concrete-filled FRP tubes under axial compression”. MS thesis, Tsinghua
University, Beiing, China. (in Chinese)
Samaan, M., Mirmiran, A., and Shahawy, M. (1998). “Model of concrete confined by fiber composites”. Journal
of Structural Engineering, ASCE, Vol. 124, No.9, pp 1025-1031.
Tao, Z., Gao, X., Yu, Q., and Zhuang, J.P. (2005). “Stress-strain relation of FRP-confined concrete”. Engineering
mechanics, Vol. 22, No.4, pp 187-195. (in Chinese)
Xiao, Y., and Wu, H. (2000). “Compressive behavior of concrete cylinders confined by carbon fiber composite
jackets”. Journal of Materials in Civil Engineering, ASCE, Vol. 12, No.2, pp 139-146.

696
Third International Conference on FRP Composites in Civil Engineering (CICE 2006)
December 13-15 2006, Miami, Florida, USA

ACCELERATED REPLACING AND CONSTRUCTION OF CONCRETE

BRIDGE DECKS USING THIN CFRP STIFFENED PLATES
Lijuan Cheng
(Assistant Professor, Department of Civil & Environmental Engineering, University of California, Davis, CA, USA)

ABSTRACT
This paper presents a hybrid fiber reinforced polymer (FRP)-concrete deck system using thin carbon FRP stiffened
plate for both deck replacement and new construction applications. The plate has a nominal thickness of 6 mm and
has extruded stiffeners and shear ribs at the surface. The objective of this study is to fully characterize the structural
behavior of the deck system through full-scale experiments and numerical simulations. This paper discusses the
major results on interfacial characterization, short-term behavior of flexure and shear, and long-term fatigue
response. Satisfactory performance was observed from the tests with a close correlation between the experimental
results and analytical simulations using nonlinear finite element method and typical constitutive material models for
FRP composites and nonlinear concrete.

KEYWORDS
Bridge deck, stiffened plate, fiber reinforced polymer (FRP) composites, finite element analysis (FEA)

1. INTRODUCTION
Many reinforced concrete bridges that were built before 1940’s in the United States are either approaching the limit
of their functional service-life or currently undergoing critical deck deterioration with serious capacity loss in the
superstructure. Partially or fully replacing these concrete decks has been very expensive and also unavoidably
causes traffic disruption and delays (Sprinkel, 1993). Due to this fact, accelerated replacing methods with least
amount of traffic disruption have attracted much attention through the use of high performance materials such as
advanced fiber reinforced polymer (FRP) composites and exploration of novel concepts on deck system design.
FRP composites are known to have light weight, greatly improved corrosion resistance and long-term durability,
potentially low maintenance and life-cycle costs. Their hybrid use with conventional concrete in bridge decks in the
form of thin continuous plates has recently been explored, not only as reinforcing products in replacement of
corrosion-prone steel, but also as stay-in-place permanent formwork to allow for rapid construction (Dieter et al.,
2002, Matsui et al., 2001, Harik et al., 1999, Hall and Mottram, 1998). This paper presents a steel-free concrete
deck system reinforced with carbon FRP (CFRP) stiffened thin plate (Cheng et al., 2005), which was sand treated on
the contacting surface and installed with small interfacial shear ribs with a spacing of 152 mm (Figure 1). The
objective of this study is to fully characterize the structural response at both component and system levels through
full-scale experiment and numerical simulations, and to develop a simplified design approach that can facilitate the
routine design of the system as well.

Steel-free concrete slab

Foam filled rectangular stiffeners
Unidirectional strips

Shear ribs
Stiffened Carbon FRP plate

Figure 1: Steel-free deck system using thin CFRP stiffened plate

697
2. EXPERIMENTS
The CFRP plate consists of a 6 mm thick bottom plate and adhesively bonded rectangular stiffeners filled with foam
material. The bottom plate is composed of 8 layers of unidirectional carbon fabric (305 g/m2) and 4 layers of E-
glass chopped strand mat (458 g/m2) in a symmetric lay-up scheme of [C/E/C2/E/C]S. The rectangular stiffeners
contain primarily unidirectional carbon fabric and foam core and are adhesively bonded onto the bottom plate.
Transverse restraining strips made of one layer of unidirectional carbon fabric are overlaid across the plate width.
To enhance the shear interaction between the concrete slab and the plate, the top surface of the plate is sand treated
and additionally installed with interfacial shear ribs made of sand-epoxy paste. The steel-free slab is constructed
from normal weight concrete with a nominal design compressive strength of 34.5 MPa at 28 days.

Eight specimens (610 mm wide, 2.254 m long) with varying spacing of interfacial stiffeners and ribs were
constructed and tested to characterize the interfacial response between the slab and CFRP plate. All specimens were
simply supported by a roller at one end, a pin at the other, and quasi-statically loaded at mid-span via a double-rod
hydraulic actuator through an elastomeric pad from the top. Flexural specimens SF1–5 cast with steel reinforced
concrete blocks at both ends (simulating fixity condition to supporting girders) behaved quite linear-elastically up to
failure. The flexural cracks in SF1, SF2, and SF3 (with a rib spacing of 152mm, 305mm, and infinite, respectively)
first grew vertically near the bottom at mid-span and then propagated diagonally toward the load point due to the
combined flexural and shear stresses followed by a sudden diagonal failure crack with a similar load capacity of 310
kN (due to the restraining effect provided from the concrete end blocks). For the effect of the stiffener spacing, SF4
(spacing of 610 mm) and SF5 (no stiffener) had an ultimate capacity about 17% and 43%, respectively, lower than
that of the control panel SF1 (spacing of 305 mm). The compressive strains in concrete and tensile strains in CFRP
composites were all well within the code and design limit. Shear bond specimens SB1, SB2, and SB3 (with a rib
spacing of 152 mm, 305 mm, and infinite, respectively) included no end blocks so as to allow for the slippage at the
slab-plate interface under quasi-static cycles. Flexural-shear type of crack and horizontal debonding was observed
in SB1 and SB2, but SB3 failed in a more flexure mode with the debonding occurred much earlier. The ultimate
capacities of SB3 showed 37% lower than SB1 and SB2 due to the absence of interfacial ribs (Figure 2a).

Midspan Displacement (in) East Span W est Span 1 Cycle - P

500,000 C ycles - P
3600 0.2 0.4 0.6 0.880
750,000 C ycles - P
600 LP5
1,250,000 Cycles - P
288 1,750,000 Cycles - P
Total Load (kN)
Total load (kN)

64 500
Total load (kip)

3P
2,000,001 Cycles - P
2,100,001 Cycles - 2P
216 48 400 2P
2,110,000 Cycles - 2P
2,150,000 Cycles - 2P
144 * 300 2,200,000 Cycles - 2P
SB1 32 P 2,250,000 Cycles - 2P
(a) SB2 200 (b) 2,300,000 Cycles - 2P
72 16 2,350,000 Cycles - 2P
SB3 100 2,350,001 Cycles - 3P
2,360,000 Cycles - 3P
0 0 0
0 5 10 15 20
Midspan Displacement (mm) 0 1 2 3 4 5
D isplacem ent (m m )

Figure 2: Test results on (a) static; (b) fatigue

The fatigue specimen was constructed from two continuous spans of 1.22 m wide slabs to allow for continuity with
a set of CFRP mesh layers placed near the top surface of the slab over the middle one-third region as tensile
reinforcement. The specimen is simply supported and loaded in a sinusoidal waveform by two 1.83 mm apart patch
loads of 84 kN via two hydraulic actuators to simulate one axle of the AASHTO truck wheel load (2004). The
specimen underwent 2.1 million cycles of fatigue service load and 250,000 cycles of doubled fatigue service load
and 10,000 cycles of tripled fatigue service load, after which the specimen was monotonically loaded up to failure.
Hairline cracks were observed on top surface of the slab above the middle negative bending moment region at the
end of the 2 million cycles of service load where the crack width satisfied the serviceability limit state per code
requirement. No tensile crack was observed on the vertical sides of the specimen. The structure suffered no
stiffness degradation during the first 2 million cycles of service load but a substantial degradation of 37.6% was
found during the subsequent 250,000 cycles of doubled service load (336 kN) and 44% during the further 10,000
cycles of tripled service load (504 kN), indicating the higher the magnitude of the wheel load, the larger the amount
of degradation in the system. The residual displacement in the system under all the fatigue load conditions was

698
found to be insignificant (as shown in Figure 2b), which displayed a largely elastic and stable manner, indicating no
slippage at the slab-deck interface (Cheng and Karbhari, 2006a).

3. ANALYSES
The functionality of the deck system arises from the intrinsic continuity and shear stress transfer between the FRP
structural formwork and the concrete cast on top. This is maintained through both a sand roughened surface on the
panel, and through raised ribs. An analytical model of the deck takes advantage of symmetry, and uses ABAQUS
(2003) with the deck panel modeled using 4-node double curved general-purpose shell elements with reduced
integration points (S4R) and the concrete by 8-node linear brick elements (C3D8). Linear elastic orthotropic
properties are used for the FRP composite, while a “damaged” plasticity formulation is used to model the nonlinear
response of concrete. This uses the ABAQUS formulation incorporating isotropic damaged elastic response in
combination with isotropic tensile and compressive plasticity to represent inelastic behavior of concrete. The
validity of the analytical model is shown in Figure 3 through the comparison with the experimental observations as
previously discussed.

Main parameters influencing the inclined cracking behavior of the FRP-concrete deck is studied using the finite
element model, including the tensile strength of concrete, the shear span-to-depth ratio, the amount of carbon fiber
reinforcement, and the spacing of the ribs. Other geometrical and material parameters are remained constants during
the parametric study. The analysis results are summarized in charts and plots which are readily to be used as design
aids (Cheng and Karbhari, 2006b). Furthermore, they provide a basis for the development of simplified design
formulae and equations as discussed in the following section.
Midspan Displacement (in)
0 0.2 0.4 0.6 0.8 Test
100
400
Total load (kip)
Total load (kN)

80
300
60
200 (a) (b)
40
FEA
100 Test (SB1) 20
Analysis (FEA)
0 0
0 5 10 15 20
Midspan Displacement (mm)

Figure 3: Verification of FEA model for SB1: (a) load-displacement response; (b) crack pattern

4. SIMPLIFIED DESIGN APPROACH

During the construction stage before the concrete sets, the composite deck panel itself shall be designed to carry the
concrete deck load, the composite deck dead load, and the construction live load. The maximum stress and strain
levels in the deck for bending and shear can be computed by elastic theory with its sectional properties and using
moment and deflection coefficients determined from structural analysis based on classic beam theory.

Since the integrity of the deck system depends on shear-bond between the FRP plate and the concrete, the level of
shear bond needs to be checked. This can be done using results from parametric studies (as previously discussed) or
through use of simplified formulae. The three approaches reported by ACI (1995), Schuster (1972) and Zsutty
(1968) were used in the study and a regression analysis was conducted between non-dimensional quantities, for
cases corresponding to the rib spacing of 152 mm–305 mm. Since the number of cases for comparison was limited,
a reduction of 15% in the regression line was used as recommended by Porter and Ekberg (1976) with a confidence
level of 95%. The resulting design equation based on Zsutty’s formula was found to have the best R2 values
(indicating how well the model fits the data, i.e., a value close to 1.0 indicates that almost all the variability has been
accounted for). Equation (1) below is recommended for design, where Vu=shear load, b=deck width, d’=distance
from extreme compressive fiber to deck centroid, ρc=reinforcement ratio, L’=shear span length, fc’=concrete
strength, s=rib spacing. Constants m and k are determined from empirical fit to analysis data, with a value of 0.750
and 0.366 for cases with s=152 mm (R2=0.92), and 1.511 and 0.770 for s=305 mm (R2=0.91).

699
 Vu s / bd ' = m ( f c' ρ c d ' / L' ) − k
1/ 3
(1)
The flexural capacity of the deck system can be obtained from compatibility of strains and equilibrium of internal
forces along the section, following the similar principle as that for reinforced concrete structures. The controlling
strain herein is either the maximum compressive strain of 0.003 in concrete or the allowable tensile strain of 0.005 in
CFRP. The nominal moment strength, Mu, is obtained as a simple summation of internal moments of all the tensile
and compressive forces on the cross section that is considered. The deflection limit state is to be checked following
conventional elastic theory using transformed sections for FRP components and cracked and uncracked cases being
considered. Derivation of these design equations is presented elsewhere (Cheng and Karbhari, 2006b).

5. CONCLUSIONS
This paper presents the experimental investigation of the hybrid FRP-concrete deck system and the appropriate
analytical simulations that are developed for the system. The results showed that the system exhibited fairly good
ultimate flexural capacity with typical concrete failure instead of catastrophic FRP brittle failure. The spacing of the
interfacial stiffeners and shear ribs at the plate surface appeared to be sufficient in transferring the shear from the
slab to the reinforcing plate. Fatigue was found to be a non-governing limit state for the slab design based on the
fact that no damage was observed in the test after 2 million cycles of fatigue service load. Design charts and
simplified equations were developed for use of this type of deck system. Development of an automatic design
optimization procedure and studies on long-term durability due to creep, shrinkage, temperature, and moisture
effects are currently under progress. Other anticipated future work involves the estimation of life-cycle cost of the
deck system and its general applications in bridge systems using conventional concrete and steel girders.

6. REFERENCES
AASHTO LRFD Bridge Design Specifications (2004). 3rd Edition, American Association of State Highway and
Transportation Officials, Washington, D.C.
ABAQUS/Standard User’s Manual (2003). Version 6.4, ABAQUS Inc.
ACI 318-95 (1995), Building Code Requirements for Reinforced Concrete (ACI 318-95) and Commentary (ACI
318R-95), American Concrete Institute, Detroit.
Cheng, L., Zhao, L., Karbhari, V.M., Hegemier, G.A., and Seible, F. (2005). “Assessment of a Steel-Free Fiber
Reinforced Polymer-Composite Modular Bridge System.” Journal of Structural Engineering, 131(3): 498-506.
Cheng, L., and Karbhari, V.M. (2006a). “Fatigue Behavior of a Steel-Free FRP-Concrete Modular Bridge Deck
System.” Journal of Bridge Engineering, in press.
Cheng, L., and Karbhari, V.M. (2006b). “Design and Analysis of FRP Structural Formwork Based Steel-Free
Modular Bridge System.” International Journal of Structural Engineering and Mechanics, Techno-Press, in review.
Dieter, D.A., Dietsche, J.S., Bank, L.C., Oliva, M.G., and Russell, J.S. (2002). “Concrete Bridge Decks Constructed
with Fiber-Reinforced Polymer Stay-in-Place Forms and Grid Reinforcing.” Transportation Research Record 1814,
Paper No. 02-3205, pp. 219-226.
Hall, J.E., and Mottram, J.T. (1998). “Combined FRP Reinforcement and Permanent Formwork for Concrete
Members.” Journal of Composites for Construction, 2(2): 78-86.
Harik, I., Alagusundaramoorthy, P., Siddiqui, R., Lopez-Anido, R., Morton, S., Dutta, P., and Shahrooz, B. (1999)
“Testing of Concrete/FRP Composite Deck Panels.” The 5th Construction Materials Congress, ASCE Materials
Engineering Division, pp. 351-358.
Matsui, S., Ishizaki, S., and Kubo, K. (2001). “An Experimental Study on Durability of FRP-RC Composite Deck
Slabs of Highway Bridges.” Third International Conference on Concrete under Severe Conditions: Environment &
Loading, Vancouver, Canada, June 18-20, pp. 933-940.
Porter, M.L., and Ekberg, C.E., Jr. (1976). “Design Recommendations for Steel Deck Floor Slabs.” Journal of the
Structural Division, ASCE, 102(11): 2121-2136.
Schuster, R.M. (1972). “Composite Steel-Deck-Reinforced Concrete Systems Failing in Shear-Bond.” Preliminary
Report, the 9th Congress of the International Association for Bridge and Structural Engineering, Zurich,
Switzerland, pp. 185-191.
Sprinkel, M.M. (1993). Concrete Bridge Protection and Rehabilitation, Washington, DC: Strategic Highway
Research Program, National Research Council.
Zsutty, T.C. (1968). “Beam Shear Strength Prediction by Analysis of Existing Data.” Proceedings, Journal of the
American Concrete Institute, 65(11): 943-951.

700
Part XXIII. Strengthening of Metals
Third International Conference on FRP Composites in Civil Engineering (CICE 2006)
December 13-15 2006, Miami, Florida, USA

BENDING STRENGTH OF CFRP-STRENGTHENED CIRCULAR

HOLLOW STEEL SECTIONS
J. Haedir, M.R. Bambach, X.-L. Zhao and R. Grzebieta
(Department of Civil Engineering, Monash University, Melbourne, Victoria, Australia)

ABSTRACT
Circular hollow sections (CHS) have been widely employed in structural and energy absorbing applications. Due to
greater strength and ductility requirements, it may result in very thin-walled steel sections that are susceptible to
local buckling. The use of new materials in the form of carbon fibre reinforced polymer (CFRP) sheets has been
increasingly applied with great advantage in strengthening of existing structures today. Furthermore, CFRP may also
be applied to minimise the effects of local buckling of thin-walled steel tubes. There are limited international design
guidelines that cater for the use of FRP in steel structures, and hence research is needed to provide engineers with a
design method to use this new material with safety and cost efficiency. This paper presents an experimental study to
assess the influence of slenderness on the bending strength of thin-walled CHS steel tubes strengthened with CFRP
for use in design. The effects of fibre orientations are also examined by having both longitudinal and hoop
wrappings of steel tubes. It has been proved that by using CFRP in both the longitudinal and hoop orientations
enhanced the capacity of slender sections to behave as compact or non-compact sections. The study shows that
greater strength and ductility can be achieved in thin-walled circular steel tubes strengthened with CFRP, thus the
use of this innovative material have potential for cost efficient design and in retrofitting steel structures for strength
and seismic enhancement.

KEYWORDS
Thin-walled steel sections, Steel tubes, Bending, Local buckling, Carbon fibre reinforced polymer

1. INTRODUCTION
External bonding of carbon fibre reinforced polymers (CFRP) has been applied to the strengthening of reinforced
concrete beams and columns widely throughout the last decade (Hollaway and Cadei, 2002). More recently the
benefits of FRP with steel in structural applications have been explored, particularly with the use of FRP for
strengthening of flexural members (Miller et al., 2001, Sen et al., 2001, Tavakkolizadeh and Saadatmanesh, 2003,
Patnaik and Bauer, 2004, Schnerch et al, 2005, Colombi and Poggi, 2006, Photiou et al., 2006). Investigations on the
use of carbon fibre sheets in strengthened circular hollow sections were reported by Jiao and Zhao (2004), Teng and
Hu (2004) and Seica et al. (2006). In the design of most structural steel sections, it is often needed to provide an
optimum amount of material to achieve structural economy. Thin-walled steel tubes with circular sections have seen
an increased usage in the application to tubular structures in steel construction mainly due to the economy they offer
in construction and design. Local buckling may govern the design of thin steel sections when subjected to
compressive loads by bending moments. The research into the local buckling behaviour of thin-walled steel in
contact with fibre reinforced polymer has until now been limited. To control the problems related to local buckling
of the steel tube, external CFRP reinforcement can be utilised as it has the effect of improving the buckling
behaviour. Therefore, carbon fibre reinforced polymers are efficient when local buckling can be restrained in the
design of thin-walled structural components. This study is aimed at examining the use of CFRP in both the
longitudinal and hoop directions in improving the strength of slender CHS in pure bending to behave as compact or
non-compact sections. A detailed experimental programme based on the behaviour of circular hollow steel tubes
wrapped with CFRP sheets under pure bending is conducted herein. This paper focuses on the effectiveness of
utilising CFRP sheets for flexural strengthening of steel tubes with thin-walled sections. The paper concludes by
pointing out future research that is needed to be undertaken which involves extending the current study to examine
the enhanced local buckling strength of CFRP-strengthened steel tubes with varying slenderness.

701
2. EXPERIMENTS
2.1 Test Series and Parameters

The experimental programme consisted of testing five cold-formed grade C350 circular hollow steel (CHS) sections.
Four of the tube sections were strengthened and denoted as CF-SL2, whilst one of the sections, denoted as SL2, was
tested without any CFRP wrapping, as given in Table 1. The steel tubes were cut from a single 6.5 m long tube, as
shown in Figure 1. The outer dimensions of the cross-section were do and the total length, L. The measured steel
thickness was denoted as t. The required diameter-to-thickness ratios were obtained by machining all the tubes to the
specific thickness. The total length of each tube was 1500 mm, while the machined length, Lm was 800 mm. The
ends of the tubes were filled with a very stiff plaster to prevent premature buckling occurring at the transition from
the un-machined to the machined steel section, which allow the development of the full bending moment. A length
of about 400 mm was made unfilled in the mid-span of the tube, and this was considered to be the deformation
length, Lf for all the test tubes. The non-dimensional section slenderness for all of the steel sections was greater than
the AS4100 limit of 120 (Standards Australia, 1998), hence all of the sections were designated slender. It is likely
that this section would experience local buckling prior to the attainment of the full yield moment. All tubes had a
length-to-diameter ratio (Lf/do) between 3 and 5. This value was chosen to be adequate for development of local
buckles without end effects in the tube behaviour.

Table 1. Geometric Properties for Bare Steel and CFRP-Strengthened Tubes

Measured
do do  fy  Lf Number CFRP layers
dimensions Fibre layout
Specimen  
do t t t  250  do Longitudinal Hoop procedure
(mm) (mm) (L) (H)
SL2 84.73 1.11 76.0 139.1 4.72 0 0 -
CF-SL2A 84.74 1.12 75.6 138.3 4.72 1 2 1H, 1H, 1L
CF-SL2B 84.59 1.12 75.8 138.7 4.73 2 2 1L, 1H, 1L,1H
CF-SL2C 84.70 1.07 78.9 147.8 4.72 1 0 1L
CF-SL2D 84.75 1.12 75.4 141.2 4.72 1 1 1L, 1H

The steel tubes were surface ground and cleaned before applying CFRP sheets to provide adequate bonding between
the steel surface and the adhesive. The tubes were wrapped using the lay-up method, and cured for a minimum
period of two weeks in the laboratory, at ambient temperature. Each longitudinal and hoop carbon fibre layer had an
overlap of 35 mm and 60 mm respectively, so that premature failure at the joints could be substantially minimised.
The overlap for the longitudinal CFRP layer was less than the CFRP layer in the hoop direction due to the fact that
the width of the CFRP layers was smaller than the circumference of the tube. Three circumferentially oriented fibres
were needed to wrap the steel section along the axial direction, as illustrated in Figure 1.

The principle parameters that were examined in this study consisted of fibre orientations in the longitudinal (L) and
hoop (H) directions, number of CFRP layers and fibre layout procedure. In laying out the carbon fibre sheets, each
tube was wrapped by variation of the fibre orientations, mainly aiming at improvement of the buckling load for a
given number of fibre layers. In tube CF-SL2A, two fibre layers were used to form the first two layers in the hoop
direction, followed by one fibre layer applied in the longitudinal direction. A similar layout sequence was applied to
the other tubes with a different number of CFRP layers, as shown in Table 1.

t cfrp P A P

35 mm
ts A Lf = 400 mm
Longitudinal layer
P P
L m = 800 mm
60 mm
Hoop layer L = 1500 mm
do
Section A-A Plaster filled Hollow

Figure 1. CFRP Strengthening Orientations of Tube CF-SL2B

702
2.2 Material Properties

Carbon fibre reinforced polymer and adhesive materials

The normal modulus carbon fibre used in the present study was manufactured by MBRACE. The carbon fibre was
in the form of unidirectional CFRP sheets. Araldite 420 epoxy resin was used to impregnate the carbon fibre sheets.
Based on the mechanical properties specifications, the CFRP sheet had a nominal thickness of 0.17 mm per layer, a
tensile strength of 3800 N/mm2 and a Young’s modulus of 240000 N/mm2.

Steel properties
Coupon tests were conducted to determine the stress-strain relationship of the steel. Tensile coupons were cut from
the tube at 90o measured circumferentially from the longitudinal seam weld. The mean yield strength for the steel
was approximately 465 N/mm2, with the average ultimate strength at 504 N/mm2. It can be noted that the CFRP
tensile strength is very high and it is meant to provide improved local buckling capacity to the thin steel section.
Hence, the initiation of local buckling of the tube would likely occur after the carbon fibre reaches its failure strain.

2.3 Test Set-up and Procedures

The apparatus for the testing programme was arranged in such a way that a pure bending moment distribution could
be applied over the mid-span of the tubes without incurring significant axial forces. The load was applied through
the extension of the two hydraulic jacks that were connected to the load application wheels. The applied bending
moment was obtained from the recorded angular rotations and forces from the load cells connected to the jacks.
Strain gauges were used to measure the compressive and tensile surface strains of the tube at the mid-span section,
which were useful in the determination of yielding and local buckling of the steel tube.

3. EXPERIMENTAL RESULTS
3.1 Moment-Rotation

Figure 2 gives a complete picture of the moment-rotation variations of all tested tubes. Two inclinometers were
mounted to the top side of the tube to measure the rotations. The results indicate that increasing the number of fibre
layers leads to a greater ultimate moment. It is worth noting that tubes CF-SL2A and CF-SL2B exhibit a more
ductile behaviour at failure than tubes SL2, CF-SL2C and CF-SL2D. It can be observed from Figure 2 that tubes
CF-SL2A and CF-SL2B behave in a manner suggestive of the response of a compact section, in which the plastic
moment capacity of the steel section would be achieved. This shows quite clearly how the increase in the number of
hoop fibre layers can have a marked effect on the plastic rotation capacity of the tube sections. Local buckling
occurs inelastically for tubes CF-SL2C and CF-SL2D, and this is representative of the characteristic of a non-
compact section. However, the tubes do not possess adequate rotation capacity for plastic design.

CF - SL2B
5 CF - SL2D
CF - SL2A

4 Msp = 3.66 kNm

Moment (kNm)

3 Msy = 2.87 kNm

SL2
1 CF - SL2C

0
0 1 2 3 4
Rotation (degree)

Figure 2. Moment-Rotation of Bare Steel and CFRP-Strengthened Tubes

703
3.2 Failure Modes

The failure modes for the same size cross-sections which were both bare steel and CFRP-strengthened tube are
given in Figure 3. The local buckling shape is restrained by the carbon fibre in the hoop direction, and this has a
significant effect on the buckling strength as evidenced by the attainment of the full plastic moment in Figure 2.
Consequently, the slender CHS behaves as a non-compact CHS. The hoop layers of tube CF-SL2D appeared to fail
in a crushing mode, whilst brittle failure of the longitudinal layers was evident in tube CF-SL2C.

Figure 3. Failure Modes of Bare Steel Tube SL2 and CFRP-Strengthened Tube CF-SL2D

4. CONCLUSIONS AND FURTHER RESEARCH

This paper has presented a background behind the use of carbon fibre reinforced polymers to improve the bending
capacity of circular hollow steel sections used in steel structures. An exploratory experimental programme involving
the testing of five circular steel tubes, including one bare steel and four CFRP-strengthened tubes, has been
conducted. The experiments have highlighted the possible benefits of the use of CFRP in enhancing the overall
behaviour of the thin steel section. This is attributed to the restraint offered by the carbon fibre for the ovalisation of
the steel tube, for which the CFRP in the hoop direction plays an important role. It has been shown that slender CHS
in pure bending may behave as compact or non-compact CHS with CFRP strengthening in both the longitudinal and
hoop directions. Further research is currently ongoing at Monash University to test CFRP-strengthened steel tubes
with compact and non-compact sections under bending conditions. This will allow the utilisation of CFRP in the
design of composite sections, in particular CFRP-strengthened circular steel tubes.

5. REFERENCES
Colombi, P. and Poggi, C. (2006). An experimental, analytical and numerical study of the static behaviour of steel
beams reinforced by pultruded CFRP strips, Composites Part B: Engineering, Vol. 37, No. 1, pp. 64-73.
Hollaway, L.C. and Cadei, J. (2002). Progress in the technique of upgrading metallic structures with advanced
polymer composites, Progress in Structural Engineering and Materials, Vol. 4, No. 2, pp. 131-148.
Jiao, H. and Zhao, X.L. (2004). CFRP strengthened butt-welded very high strength steel (VHS) circular steel tubes,
Thin-Walled Structures, Vol. 42, No. 7, pp. 963-978.
Miller, T.C., Chajes, M.J., Mertz, D.R. and Hastings, J.N. (2001). Strengthening of a steel bridge girder using CFRP
plates, Journal of Bridge Engineering, ASCE, Vol. 6, No. 6, pp. 514-522.
Patnaik, A.K. and Bauer, C.L. (2004). Strengthening of steel beams with carbon FRP laminates, Proceedings of the
4th International Conference on Advanced Composite Materials in Bridges and Structures, Calgary, Alberta, July
20-23.
Photiou, N.K., Hollaway, L.C. and Chryssanthopoulos, M.K. (2006). Strengthening of an artificially degraded steel
beam utilising a carbon/glass composite system, Construction and Building Materials, Vol. 20, No. 1-2, pp. 11-21.
Schnerch, D., Dawood, M. and Rizkalla, S. (2005). Strengthening steel-concrete composite bridges with high
modulus carbon fiber reinforced polymer (CFRP) laminates, Proceedings of the 3rd International Conference on
Composites in Construction (CCC 2005), Lyon, France, July 11-13.
Seica, M.V., Packer, J.A., Guevara Ramirez, P., Bell, S.A.H. and Zhao, X.L. (2006). Rehabilitation of tubular
members with carbon reinforced polymers, Proceedings of the 11th International Symposium and IIV International
Conference on Tubular Structures, Québec, Canada Aug 31 - Sep 2.
Sen, R., Liby, L. and Mullins, G. (2001). Strengthening steel bridge sections using CFRP laminates, Composites
Part B: Engineering, Vol. 32, No. 4, pp. 309-322.
Standards Australia. (1998). Australian Standard, Steel Structures, AS4100-1998, Sydney, Australia.
Tavakkolizadeh, M. and Saadatmanesh, H. (2003). Strengthening of steel-concrete composite girders using carbon
fiber reinforced polymer sheets, Journal of Structural Engineering, Vol. 129, No. 1, pp. 30-40.
Teng, J.G. and Hu, Y.M. (2004). Suppression of local buckling in steel tubes by FRP jacketing, Proceedings of the
2nd International Conference on FRP Composites in Civil Engineering, Adelaide, Australia, Dec 8-10, pp. 749-753.

704
Third International Conference on FRP Composites in Civil Engineering (CICE 2006)
December 13-15 2006, Miami, Florida, USA

BOND AND SPLICE BEHAVIOR OF HIGH MODULUS CFRP

MATERIALS BONDED TO STEEL STRUCTURES
M. Dawood
(Graduate Research Assistant, North Carolina State University, Raleigh, NC, USA)

S. Rizkalla
(Distinguished Professor of Civil Engineering and Construction, North Carolina State University,
Raleigh, NC, USA)

ABSTRACT
Research on the use of CFRP materials for retrofit and strengthening of steel bridges and structures has increased in
recent years. In order to implement these types of strengthening systems to longer span members, it is important to
establish an effective method to splice adjacent lengths of CFRP laminates. This paper describes an experimental
program to investigate the bond behavior of carbon fiber reinforced polymer (CFRP) laminates with the objective of
developing an effective bonded splice joint. The experimental program was conducted in two phases. In the first
phase, the effectiveness of three different CFRP splice joint configurations was evaluated using double lap shear
coupon tests. In the second phase, steel beams were strengthened with CFRP laminates which incorporated a
bonded splice joint located at the midspan of each of the beams. The test results demonstrate that debonding of the
CFRP splice plate was the primary mode of failure for the tested joints. The research also indicated that the use of a
reverse tapered joint configuration can significantly increase the splice joint capacity. The findings indicate that
careful detailing can significantly increase the capacity of bonded CFRP splice joints.

KEYWORDS
CFRP, steel bridges, splice joints, bond behavior, reverse taper

1. INTRODUCTION
Recently, considerable research has been conducted on the use of carbon fiber reinforced polymer (CFRP) materials
for rehabilitation and strengthening of steel bridges and structures (Al-Saidy et al., 2005, Mertz & Gillespie, 1996,
Sen et al., 2001, Tavakkolizadeh and Saadatmanesh, 2003). The previous research indicates that externally
bonded CFRP laminates can be used to increase the ultimate strength of steel girders and to restore the lost capacity
and stiffness of damaged or deteriorated girders. However, due to the relatively low modulus of elasticity of
conventional CFRP materials as compared to steel, large amounts of strengthening materials are required to achieve
a significant increase of the elastic stiffness. The use of high modulus CFRP (HM CFRP) materials has been
demonstrated to be a more effective technique to increase the stiffness of steel beams (Rizkalla and Dawood, 2006).

Similarly to reinforced concrete beams, for steel beams reinforced with externally bonded CFRP materials, the bond
stresses at the interface between the beam and the strengthening materials, including both shear and peeling
components, must be carefully considered to prevent premature debonding failures (Buyukozturk et al., 2003).
Bond stresses are particularly critical at lap-spliced connections between CFRP laminates. The behavior of these
splices is dramatically affected by the bond behavior and the bond stress distribution between CFRP laminates.
There have only been a limited number of studies which investigate the behavior of spliced connections of CFRP
laminates under flexural loading. Stallings and Porter tested eight reinforced concrete beams strengthened with
various configurations of externally bonded CFRP splice joints (2003). They recommend that in order to prevent
debonding of the CFRP splice plate, the maximum strain in the main CFRP plate immediately prior to the splice
should not exceed a limiting value. To the authors knowledge there have not been any studies investigating the
splice behavior of steel beams reinforced with CFRP laminates.

705
2. EXPERIMENTAL PROGRAM
An experimental program was conducted in two phases to investigate the bond and splice behavior of CFRP
laminates. In the first phase three double-lap shear coupons with three different joint configurations, shown
schematically in Figure 1(a) – (c), were tested. The objective of the first phase was to determine the effectiveness of
implementing a reverse taper detail at various critical locations throughout the spliced joint. The splice coupons
were 35 mm wide and were fabricated from 4 mm thick CFRP laminates with a modulus of elasticity of 460 GPa
and an ultimate strain of 0.00334 as reported by the manufacturer. For all three joint configurations, strains were
measured at various locations along the splice joint using electrical resistance strain gauges.

8 mm thick CFRP main plate

1 mm thick adhesive layer
(two 4 mm thick laminates bonded together)

4 mm thick CFRP splice plate

(a) Configuration A (square ends, no fillet)

reverse taper, adhesive fillet square ends, no fillet

(b) Configuration B

reverse taper, adhesive fillet reverse taper

(c) Configuration C
Figure 1: Double-lap shear coupon joint configurations and instrumentation

In the second phase of the experimental program, two steel beam tests were conducted to investigate the behavior of
spliced joints under flexural loading conditions. The typical test beam is shown schematically in Figure 2. The
beams each consisted of a standard wide-flange steel beam. A steel channel was welded to the top flange of the
steel beam to simulate the presence of a reinforced concrete deck. The test beam was strengthened by bonding a
100 mm wide by 4 mm thick HM CFRP laminate to the tension flange. The laminate was left discontinuous at the
midspan of the beam as shown in Detail A of Figure 2. Continuity of the strengthening system was provided by
bonding an 800 mm long CFRP splice plate at the joint location which overlapped 400 mm on either side of the
CFRP main plate. All of the plate ends were left square without incorporating a reverse taper to serve as a reference
for future tests which will incorporate different joint details. Due to the high level of uncertainty associated with the
bond behavior, two such beams were tested to provide repeated test data for a given splice configuration. The
strengthened beams were loaded monotonically to failure in four point bending using a hydraulic actuator. The
longitudinal strain in the CFRP splice plate and the main plates was measured at several locations using electrical
resistance strain gauges.

3048 mm
(120”) W12x30
1600 mm C9x15
(63”)

CFRP main plate See Detail A 800 mm long CFRP splice plate
4420 mm
Steel tension flange 4572 mm
Adhesive CFRP main plate
CFRP splice plate
Detail A
(not to scale)

Figure 2: Typical steel test beam configuration

706
3. EXPERIMENTAL RESULTS
The measured load-strain behavior at the surface of the splice plate at the center of the joint is shown for the three
double lap-shear joint configurations in Figure 3. The initial stiffness of all three joint configurations was similar up
to a load level of 40 kN. At the 40 kN load level a sudden increase of the measured strain was observed for joint
configurations A and B which was likely due to cracking of the adhesive due to a stress concentration near the
square plate end within the center of the joint. Cracking of the adhesive resulted in a corresponding loss of stiffness
of the joint as shown in Figure 3(a). Joint C did not exhibit a similar increase of strain which suggests that cracking
did not occur and that the reverse taper was effective in reducing the stress concentration near the plate end. Joint A
failed suddenly due to debonding of the CFRP splice plates at a load level of 90 kN. Failure occurred primarily by
separation of the adhesive from the CFRP laminates as shown in Figure 3(b). Joint B exhibited additional cracking
at a load level of 144 kN and ultimately failed by debonding of the splice plates at a load level of 160 kN. Joint C
failed at a load level of 190 kN suddenly due to debonding of the splice plates and did not exhibit any cracking
throughout the entire loading range. The experimental results demonstrate that the presence of the reverse taper and
the adhesive fillet details can approximately double the capacity of a bonded splice joint.

250
strain gauge

200
Debonded splice plate

150 Joint C
Load (kN)

Joint B
100

Joint A Main plate

50
Coin for scale

(a) (b)
0
0.0000 0.0005 0.0010 0.0015
Strain (mm/mm)

Figure 3: (a) Measured load-strain behavior for double-lap shear coupon tests (b) Debonding failure

Failure of the two tested beams occurred due to debonding of the splice plates prior to yielding of the tension flanges
of the steel beams. The total measured load immediately prior to debonding was 177 kN and 205 kN for each of the
beams respectively which corresponds to approximately 37 percent and 43 percent of the estimated yield load of the
strengthened beams, respectively. The yield load was determined based on a non-linear moment curvature analysis.
The longitudinal strain distribution along the length of the splice plate, which was measured during the second beam
test, is shown in Figure 4(a) for various load levels. The measured strain profile from the first test closely matched
that shown in Figure 4 and therefore is not presented in this paper.

The average shear stress along a given interval of the adhesive layer can be determined from the measured strains by
considering equilibrium of forces and is proportional to the slope of the strain profile. The calculated bond shear
stress distribution along the length of the splice plate is shown in Figure 4(b). Significant shear stress concentrations
were calculated at localized regions near the end of the CFRP splice plate and near the center of the splice joint
which can be observed in Figure 4(b). The maximum calculated shear stress in the adhesive at a load level of 205
kN was 20 MPa near the end of the CFRP splice plate and 7.3 MPa near the center of the splice. At the same load
level, the calculated shear stress at a distance of 200 mm to 300 mm away from the splice plate end was only
0.9 MPa. This significant concentration of stresses near the plate ends is likely the cause of the premature
debonding failure which was observed during both flexural tests. This highlights the need to investigate various
details to help reduce the stress concentration near the plate ends.

707
 40
1200
35

Measured Longitudinal Strain(µε)

1000 30

Shear Stress (MPa)

Measured strain
800 25
in the main plate 205 kN (failure)
(205 kN) 205 kN (failure)
20
600 160 kN
160 kN
120 kN 15
400 120 kN
80 kN 10
80 kN
200 40 kN
40 kN 5
(a) (b)
0 0
-100 0 100 200 300 Beam
0 CL
400 300
100 200 100
300 0
400 500
Distance from Splice Plate End (mm) Distance from Splice Plate End (mm)

Figure 4: (a) Measured splice plate longitudinal strain (b) Calculated adhesive shear stress

4. CONCLUSIONS
This paper describes an experimental program which was conducted to investigate the bond and splice behavior of
CFRP laminates. Initial lap-shear coupon test results indicate that a reverse tapered detail near plate ends can
significantly reduce the bond stress concentrations at these locations and consequently increase the ultimate capacity
of the spliced joints. The results of the beam tests indicated the presence of significant stress concentrations near the
plate ends which resulted in premature debonding failure of the CFRP splice plates.

ACKNOWLEDGEMENTS
The authors would like to acknowledge the contributions of Mr. Akira Nakagoshi of Mitsubishi Chemical FP
America Inc. and the support provided by the National Science Foundation (NSF) Industry/University Cooperative
Research Center (I/UCRC) for Repair of Buildings and Bridges with Composites (RB2C).

REFERENCES
Al-Saidy, A.H., Klaiber, F.W. & Wipf, T.J. (2004). Repair of steel composite beams with carbon fiber-reinforced
polymer plates. Journal of Composites for Construction, 8 (2), 163-172.
Buyukozturk, O. Gunes, O. & Karaca, E. (2004). Progress in understanding debonding problems in reinforced
concrete and steel members strengthened using FRP composites. Construction and Building Materials, 18, 9-19.
Mertz, D.R. & Gillespie Jr., J. W. (1996). Rehabilitation of steel bridge girders through the application of advanced
composite materials (Contract NCHRP-93-ID011). Washington, D.C.: Transportation Research Board.
Miller, T.C., Chajes, M.J., Mertz, D.R. & Hastings, J.N. (2001). Strengthening of a steel bridge girder using CFRP
plates. Journal of Bridge Engineering, 6 (6), 514-522.
Rizkalla, S. and Dawood, M. (2006). High modulus carbon fiber materials for retrofit of steel structures and
bridges. Accepted for publication in the proceedings of ACUN-5 International Composites Conference
"Developments in Composites : Advanced, Infrastructural, Natural and Nano-composites". July 11-14, 2006.
Sydney, Australia.
Sen, R., Libby, L. & Mullins, G. (2001). Strengthening steel bridge sections using CFRP laminates. Composites
Part B: Engineering, 39, 309-322.
Stallings, J.M. and Porter, N.M. (2003). Experimental investigation of lap splices in externally bonded carbon
fiber-reinforced plastic plates. ACI Structural Journal, 100 (1), 3-10.
Tavakkolizadeh, M. & Saadatmanesh, H. (2003). Strengthening of steel-concrete composite girders using carbon
fiber reinforced polymer sheets. Journal of Structural Engineering, 129 (1), 30-40.

708
Third International Conference on FRP Composites in Civil Engineering (CICE 2006)
December 13-15 2006, Miami, Florida, USA

CRFP-ALUMINUM ALLOY COMPOSITE STRUCTURES:

A NEW TYPE OF COMPOSITE STRUCTURES IN FUTURE
Lieping Ye
(Professor, Department of Civil Engineering, Tsinghua University, Beijing, China)

Peng Qian
(PhD Graduate, Department of Civil Engineering, Tsinghua University, Beijing, China)

Peng Feng
(Lecturer, Department of Civil Engineering, Tsinghua University, Beijing, China)

ABSTRACT
High-strength, light-weight and durable materials are always desired by structure engineer. Although CFRP (carbon
fiber reinforced-polymer) meets the demands, but its anisotropy makes the connection of structural elements
considerably weak which is the main cause that make its high strength not be utilized fully. However, Al (aluminum
alloy) is now becoming the structural material in civil engineering in recent years also due to its light-weight and
anti-corrosion properties. Although it has reasonable strength as normal steel, but its low modulus makes it
deformation control difficult that results in limiting its application in large span structures. The combinion of CFRP
and Al, which can make the high-strength being fully developed and make the connnection being easy between
elements, will be expected as a new type in composite structures in future. The state of the art of CFRP-Al
composite in aero and space structures, and aluminum structures in civil engineering are firstly reviewed in this
paper. CFRP-Al composite pipes, which can be used as elements in long-span space truss structures, under axial
compressive and flexual loading were studied by tests. It is shown that CFRP can significantly increase the load
capacity and the stiffness of pure Al members. The results of the experimental study have showed that this new
composite will have potential benefits for the structures in future.

KEYWORDS
CFRP, Al (aluminum alloy), composite structures, light weight, anti-corrosion.

1. INTRODUCTION
The larger breathing space is always the dream of people all the while. There are two basic ways to obtain it by
structural engineer: one is to use the appropriate structure styles, another is to use the light weight materials. These
may be considered the main motives behind the introduction of CFRP (carbon fiber reinforced polymer) and Al
(aluminum alloy) to be used as structural material in civil engineering. As compared with the Al, CFRP yields more
high performance and offers more specific strength, stiffness and resistance (Hollaway, 2001). However CFRP is an
anisotropic material, which has high strength and elastic module along the fiber direction and low strength in the
transverse direction (Feng and Ye, 2002). So in the design of the CFRP and the CFRP composite structures, two
directions have to be considered. Besides adding to the complications and difficulties in the design process of CFRP
structures due to the anisotropic, the connections between elements are always the weak position in the structural
and can not be easy solved with proper design. Al is a light weight, corrosion resistant material, which is widely
applied in the aviation, railway and automobile (Mazzolni, 1985). Since 1940s, Al was introduced in the
construction of bridges in North America and Europe, including footbridges, highway bridges, railway bridges and
float bridges. At the same time, Al building structures, such as the long-span dome, springs up and becomes widely
used all over the world for its light weight, easy manufacture, and anti-corrosion properties. Besides that, Al is also
used in the construction of frames, water channels, roof boardings, walls, and communication towers, and so on.
However, comparing with steel, the elastic module of Al, for example type of T6061 T6, is about 1/3 of it. The

709
deflection and the buckling often restrict the use of Al in structures. Therefore, FRP-Al composite structures are
suggested here to use their common stong points and can overcome their individual weakness, weak connection for
CFRP and lower modules for Al. By combination of FRP and aluminum, FRP high strength can be made useful in
the longitudinal direction to improve the stiffness and stable bearing capacity of Al components; and using
aluminum in connections instead of FRP further ensures the reliability and strength at the junction between
components. Moreover, the combination can make full use of the advantages of two materials that is the light-weight
and anti-corrosion. In this paper, the state of the art of the CFRP-Al composite is firstly reviewed. Then, the
experimental researches on CFRP-Al composite elements in compressive and bending are stated. The results show
that FRP-Al composite elements have potential benefits for structures in future.

2. CFRP-AL COMPOSITE STRUCTURES

Al structures have the advantages of light weight and corrosion resistant, which were firstly used in the construction
of bridges. The first pure Al bridge, which is a 29.72m long railway bridge, was built across the Glass River near
New York in 1946. It was also used in the retrofit for existing bridges that Al bridge decks replace steel or concrete
decks (Napier et al, 1998; Hag-Elsafi and Alampalli, 2002). For building structures, Al can reduce the structures self
weight, which will improve the structures ultimate span and make the construction easy and fast. In addition, Al
structures are glossy without painting, anti-corrosion, and non-magnetic. Due to these advantages, Al long-span
space structures have been used in some special environments. In China, the applications of Al structures have
gained attention in recent years, such as the dome structure of Shanghai international gymnastic center (Figure 1),
glass ellipsoid Al of Shanghai science and technology museum, and a non-magnetic latticed structure roof (Figure 2).

Figure 1: International Gymnastic Center Figure 2: Non-Magnetic Latticed Structure Roof

Comparing with the pure Al structures, CFRP-Al composite structures can be used to get more light and handy
building structures. This kind of composite structures has been widely used in aviation industry, like the
construction of commercial airplanes, military airplanes and spaceships (Harris et al, 2002). F-22 fighter,
manufactured by Lockheed Martin, is a typical composite structure which is composed of 39% titanium alloy, 24%
aluminum alloy and 39% FRP. Its tail fin manufactured by Al honeycomb core and CFRP stressed-skin structures.
In the plane wing, the main skeleton structure is made up of Al and FRP is used as stressed skin. CFRP-Al
composite structures have been used in manufacturing transportation and communication. Many automobile
actuating arms made of CFRP-Al composite, can not only reduce the weight, but also improve the strength and
stiffness of the rods (Lee et al, 2004).

FRP repairing the cracked Al structures was studied by Pantelides et al (2003). But the idea of the combination of
CFRP and Al for civil engineering structures is different. CFRP and Al are used in a reasonable configuration to
realize the longer span structures. So many feasible combination styles of CFRP and Al can be designed.
Honeycomb panels made of FRP skins and Al honeycomb shown in Figure 3 can be used as lightweight walls and
roof structures. CFRP-Al composite beams show in Figure 4, made by sticking CFRP sheets on the tensile and
compressive wings of Al profiles, can make full utilization of CFRP high longitudinal strength. They are lighter than
pure Al beams but heavier than pure CFRP beams, while the composite beams become cheaper, easier to
manufacture, and have good ductility and fatigue resistance. The connection style of metal can be use to avoid the
complexity of CFRP connection. CFRP-Al composite pipes and tubes, which are to construct long-span space
structures, were developed and studied in this paper. The CFRP sheets are wrapped and adhered with epoxy resin on
the Al pipe’s surface. The manufacture process is listed in Figure 5. By the combination, FRP high strength can be
fully utilized in longitudinal direction to improve the stiffness and stable bearing capacity of aluminum pipes, and
the aluminum connections can be employed instead of the CFRP ensures the reliability and strength at junction

710
between components. Moreover, aluminum can also supply enough bearing capacity in transverse direction to
overcome the low strength of CFRP in this direction. With the development of science and technology, the establish
of outer space structures are becoming more and more. Besides same requirements as the ordinary structures, some
special requirements in outer space are demanded As all the building materials have to be sent from the earth, the
whole structure must be light, handy, and modular. CFRP-Al c composite structures can meet all the requirements.

CFRP wings
CFRP skin plate Al honeycomb

Al beams

CFRP wings

Figure 3: CFRP-Al Sandwich Panel Figure 4: CFRP-Al Composite Beams

CFRP layers Marking Sand Blasting Cleaning

Surfaces Surfaces Surfaces
og

Al pipe
Wrapping CFRP on
Curing Pipes

Figure 5: Manufacture Process of FRP-Al Composite Pipes

3. EXPERIMENTAL STUDY ON CFRP-AL MEMBERS

3.1 Axial Compressive Tests

Two sets of composite pipes were tested under axial loading. Twelve specimens in Set 1 are short pipes. Their
bearing strength enhanced obviously with CFRP layer numbers increasing which can be seen from in Figure 6. And
ten specimens in Set 2 with large slenderness and different CFRP layer numbers were tested to investigate their
buckling load-capacity. Their buckling strength were enhanced significantly to the pure Al pipes, the flexural
stiffness and deformation capacity were enhanced both. Figure 7 shows the curves of the transverse deflection to the
vertical load for the long pipes with the slenderness ration about 120. Based on the tests and a series of finite
element analysis, the buckling behaviors of composite pipes were studied, a formula of stability factor was presented
(Qian, 2006).

250 40

200
30
P (kN)

150
P (kN)

20
100 Pure Al pipe
CFRP [0/90]1 Pure Al pipe
CFRP [0/90]1
50 CFRP [0/90]2 10 CFRP [0/90]2
CFRP [0/90]3
CFRP [0/90]3
0
0
0.0 3.0 6.0 9.0 12.0
Δ (mm) 0 15 30 45 60
Δ (mm)
Figure 6: Load-Axial Deflection Curves of Set 1 Figure 7: Load-Lateral Deflection Curves of Set 2

3.2 Bending Tests

Two sets of the square tube specimens with different width-thickness ratio were tested under bending. Their load-
deflection curves are shown in Figure 8. It can be seen that the strength and stiffness of the composite members are

711
higher than those of pure Al members while the ductile failure mode is remained. Based on the tests, it was found
the initial stiffness of the composite member can be determined by the sum of the Al and CFRP.

Pure Al Pure Al
50 LVDT Broke
90 CFRP 2 plies CFRP 2 plies
80 CFRP 4 plies CFRP 4 plies
40
70
60 30

P(kN)
P(kN)
50
40 20
30
20 10
10
0 0
0 5 10 15 20 25 30 0 5 10 15 20 25 30 35 40 45
Δ(mm) Δ(mm)

Figure 8: Load-Deflection Curves of the Flexural Specimens

4. SUMMARY

Since 1940s, Al has been applied and developed and had found its application in civil engineering, especially in the
construction of bridges and space structures. It has potential benefits because of its high specific strength, high
specific stiffness, corrosion resistant and easy processing properties. But Al has its own inherited faults. Its elastic
module is only 1/3 of steel and the distortion of aluminum structures is usually large, so the aluminum structures are
scarcely used. In order to avoid Al inherited faults and get the more light and handy structures, CFRP-Al composite
components and structures are suggested in this paper. The experiment study shows that the combination not only
improves components stiffness but also bearing capacity.

ACKNOWLEDGEMENTS

The authors are grateful to the Natural Science Foundation of China for their support to the research presented here
through a national key project on the application of FRP composites in civil engineering in China (Project No.
50238030).

REFERENCES
Feng, P., Ye, L. P. (2002). “FRP structures and FRP composite structures in structural engineering”. Proceeding of
2nd National Conference on FRP Application Technologies in Civil Engineering. Beijing, Tsinghua University
Press, pp51-63 (in Chinese).
Hag-Elsafi, O., Alampalli, S., (2002) "Cost-effective rehabilitation of two aluminum bridges on Long Island, New
York", Practice Periodical on Structural Design and Construction, Vol.7, No.3, pp111-117.
Harris, C. E., Starnes Jr., J. H., Shuart, M. J. (2002). “Design and manufacturing of aerospace composite structures,
state-of-the-art assessment”. Journal of Aircraft, Vol.39, No.4, pp 545-560.
Hollaway, L.C. (2001). “The evolution of and the way forward for advanced polymer composites in the civil
infrastructure”. Proceedings of the International Conference on FRP Composites in Civil Engineering, Editors:
Teng,J.G. Hong Kong, Elsevier Science Ltd. pp27-40.
Lee, D. G., Kim, H. S., Kim, J. W., et al. (2004). “Design and manufacture of an automotive hybrid
aluminum/composite drive shaft”. Composite Structures, Vol.63, No.1, pp87-99.
Mazzolani, F.M. (1985). Aluminum alloy structures, Pitman, Boston.
Napier, C. S., McKeel, W. T., Sprinkel, M. M. (1998). “Bridging the centuries: Moving Virginia’s bridge program
into the 21st century”, http://www.tfhrc.gov/pubrds/septoct98/century.htm , 08/01/2005.
Pantelides, C. P., Nadauld, J., Cercone, L. (2003). “Repair of cracked aluminum overhead sign structures with glass
fiber reinforced polymer composites”. Journal of Composites for Construction, Vol.7, No.2, pp118-126.
Qian, P. (2006). “Research on FRP and FRP-aluminum alloy composite members under axial force”. Ph.D. thesis,
Tsinghua University, Beijing, China (in Chinese).

712
Third International Conference on FRP Composites in Civil Engineering (CICE 2006)
December 13-15 2006, Miami, Florida, USA

EXPERIMENTAL STUDY ON CFRP-ALUMINUM ALLOY COMPOSITE

PIPES UNDER AXIAL COMPRESSIVE LOAD
Peng Qian
(PhD Candidate, Department of Civil Engineering, Tsinghua University, Beijing, China)

Lieping Ye
(Professor, Department of Civil Engineering, Tsinghua University, Beijing, China)

Peng Feng
(Lecturer, Department of Civil Engineering, Tsinghua University, Beijing, China)

ABSTRACT
An innovated element, Al(aluminum alloy) pipe wrapped with CFRP (carbon fiber reinforced polymer), is presented
for large-span structures and special structures in favor of their common advantages of light-weight and corrosion-
resistance. Outside CFRP layers can enhance strength and stiffness of Al pipes for compression, while Al pipes
provide convenient and strong connections between elements. Two sets of composite pipes were tested. Twelve
specimens in Set 1 are short pipes, whose bearing strength increments with CFRP layer numbers varying were
investigated. And ten specimens in Set 2 with large slenderness and different CFRP layer numbers were tested to
investigate their buckling load-capacity. It was shown that the bearing strength and the buckling strength were
enhanced significantly to the pure Al pipes. Their stiffness and deformation capacity were improved also.

KEYWORDS
CFRP, aluminum, composite pipe, axial loads, buckling

1. INTRODUCTION

Al (aluminum alloy) is a light-weight, long-endurance material widely used in many engineering fields The research
on Al structures for bridges and buildings has developed all over the world since 1930s (Mazzolni, 1985). However,
Al structures are limited for its low elastic module which is about 1/3 of steel although the strength is almost same as
the common steel. Hence, the concept to combine FRP (fiber reinforced polymer) with high module and Al was
proposed to improve Al element performance. FRP has found its applications in civil engineering and is gradually
becoming a common structural material in recent 20 years (Ye and Feng, 2006). Comparing with aluminum alloy,
FRP has higher tensile strength and elastic modulus in fiber direction, which can be used to enhance performance of
Al elements. However, the performance of FRP in transverse and shear is weak. The strengths ratio between the two
directions can be up to 25 and 5 respectively to tension and compression (Feng and Ye, 2002). It is resulted in that
the high strength and the high stiffness of FRP can not be transferred through the joint without significant weight
penalty where various major stress directions occur. If the FRP components are laminated structures, delamination
may appear at the joint (Qian and Ye, 2004) .The limitation for joints makes the FRP structure design complicated
and difficult, which restricts the full utilization of its high strength and stiffness. It is favored that the combination of
FRP and Al will counterbalance this as the connection between Al members is convenient and effective. Therefore,
FRP-Al composite structures are presented for large-span structures and special structures. The FRP tubes and the
aluminum elements in latticed structures can be taken place by the FRP-Al composite elements. The kind of
composite structures has been widely used in aviation industry, like the construction of commercial airplanes,
military airplanes and spaceships. F-22 fighter, manufactured by Lockheed Martin, is a typical composite structure
which is composed of 39% titanium alloy, 24% aluminum alloy and 39% FRP. And its tail fin is manufactured by
aluminum alloy honeycomb core and CFRP stressed-skin structures (Harris et al, 2002). Automobile actuating arms
made of CFRP-aluminum alloy composite, can not only reduce the weight of rods, but also improve the strength and
stiffness of rods (Lee et al, 2004).

713
As a typical example of the new conceptual composite elements in structural engineering, CFRP-Al composite pipes
under compressive load, which is the most common case in spatial truss structures, are investigated in this paper.
CFRP sheets were wrapped and adhered on out-surface of Al piped in longitudinal and transversal alternately. The
longitudinal direction fiber is utilized to improve stiffness, strength and stable bearing capacity of Al pipes. The
transversal fibers can provide a confinement and prevent the local buckling. The more effective connections
between elements than pure FRP components’ can be inherited from Al pipes. Thus, the common merits of FRP and
Al, light-weight and anti-corrosion, can be utilized but their shortcomings separately may be overcome. In order to
investigate the axial compressive behaviors of CFRP-Al composite pipes, two sets of specimens, twelve in Set 1 and
ten in Set 2, were tested. The pipes were analyzed by finite element software. The theoretical formula based on
Euler Theory were presented and compared with the results of tests and FEA. A design approach for CFRP-Al
composite pipes in compression was suggested.

2. EXPERIMENTAL PROGRAM
T6061-T6 Al circular pipes with an outside diameter of 49.7mm and a nominal thickness of 3.1mm were used. The
mechanical parameters are obtained by short column compression tests and tensile strength tests respectively as
listed in Table 1. From table 1 it is evident that the yield strains and stresses both in compression and tension are
quite close to each other. Since during investigation all the components are in compressive state so all the analytic
parameters are taken from the stocky column tests in compression. The CFRP were wrapped on the outside surface
of the Al pipes, which were treated by sand blasting and cleaned. The unidirectional fiber sheets made of T30 carbon
fiber and epoxy resin were laminated layer by layer. The elastic parameters of the CFRP laminates were obtained by
tensioning the one-layer plates with different fiber orientations: 0, 90 and 45, according to Test method for tensile
properties of oriented fiber reinforced plastics (Chinese Standard GB3354-82), as listed in Table 2.

Table1: Aluminum Alloy and Mechanical Parameters

Test σy (MPa) εy f0.2(MPa) ε0.2 E (MPa) ν
Compression 259 3881 291 6224 69800 0.36
Tension 270 3965 291 6317 70300 /

Table2: CFRP Laminate Mechanical Parameters

E1（MPa） E2（MPa） G12（MPa） ν21
7.15×104 3.37×103 1.53×103 0.36

Table3: Set 1-Short Pipes

Specimens Al Pipes Outer Diameter - Thickness (mm) CFRP Thickness (mm) CFRP Ply
SACP-1,2,3 49.7 - 3.0 / /
SACCP1-1,2,3 49.7 - 3.0 1.4 [0/90]1
SACCP2-1,2,3 49.7 - 3.0 2.4 [0/90]2
SACCP3-1,2,3 49.7 - 3.0 3.7 [0/90]3

Table4: Set 2-Long Pipes

Specimens Length (mm) Al Pipes Outer Diameter - Thickness (mm) CFRP Thickness (mm) CFRP Ply
ACCP60-0 950 49.7 - 3.1 / /
ACCP60-2 950 49.7 - 3.1 2.7 [0º/90º]2
ACCP60-3 950 49.6 - 3.0 3.5 [0º/90º]3
ACCP70-1 1100 49.6 - 3.1 1.5 [0º/90º]1
ACCP70-2 1100 49.6 - 3.0 2.7 [0º/90º]2
ACCP70-3 1100 49.7 - 3.1 3.5 [0º/90º]3
ACCP120-0 1950 49.7 - 3.1 / /
ACCP120-1 1950 49.6 - 3.0 1.5 [0º/90º]1
ACCP120-2 1950 49.6 - 3.1 2.7 [0º/90º]2
ACCP120-3 1950 49.6 - 3.1 3.5 [0º/90º]3

714
Two sets of specimens were tested. Set 1 which is 12 short pipes with the length of 150 mm is to investigate the
axial compressive loading capacity of composite elements. According to the layers number and the ply orientation
the specimens can be divided into four groups as listed in Table 3, including a group of controlling specimens
without CFRP named SACP and three groups of CFRP-Al composite pipes named SACCP. Set 2 includes 10 long
specimens divided into three groups by slenderness ratio as listed in Table 4, in which two pure Al pipes of
ACCP60-0 and ACCP120-0 act as the controlling specimens.

3. TEST RESULTS

3.1 Set 1: Short Pipes

The controlling specimen, SACP, was loaded increasing linearly with vertical displacement before reaching the
yield strain of aluminum. After yielding, the end of the pipe began bulging outward, and the transverse deformation
kept on increasing until its failure state reached along with a decrease in the load resistance. An elephant-foot
buckling appealed as shown in Figure 3. And some tiny longitudinal cracks can be found on the surface near the
pipe end. The composite specimens, SACCP1, SACCP2 and SACCP3, show the similar loading failure process with
SACCP besides the cracking sounds started emitting always when the pipes closed to the yield load and continued to
the rupture of the carbon fiber. There was no obvious bulging, and the yield load and the maximum load rise with
the increase of CFRP reinforcement. For SACCP1, two cross layers CFRP wrapped, the failure mode after Al yield
is same as SACP without CFRP as the CFRP layer can not provide enough lateral confinement to restrict the lateral
expansion of aluminum near the pipe end. But the confinement effect on the aluminum pipes is enhanced to make
the ultimate failure mode after Al yield transferred with the increase of CFRP layer amount: SACCP2 were
concaved inner and bulged a little on the opposite side; and there were concavity and no obvious bulge in SACCP3.
In the test there was no debonding observed at the interfaces between Al and CFRP. The failure modes are shown in
Figure 1.

250

200

150
P (kN)

100
SACP
SACCP 1
50 SACCP 2
SACCP 3
0
0.0 3.0 6.0 9.0 12.0
Δ (mm)

(a) SACP and SACCP1 (b) SACCP2 (c) SACCP3

Figure 1: Failure Modes of Pipes in Set 1 Figure 2: Load-deformation Relations of Set 1

At the same time, the yield loads and the ultimate loads rise almost linearly to the CFRP layer amount increase: the
mean yield loads of SACCP1, SACCP2 and SACCP3 are 119%, 139.8% and 150% of SACP respectively, the
maximum load capacity are 119.8%, 144% and 154.6% respectively. Figure 3 shows the axial load and the vertical
displacement relations obtained from the tests. The values of the curve are taken as the mean of three specimens. It
also can be seen that the stiffness of the pipes increase slightly with the increase of CFRP layers before yield and
increase considerably after yield. Therefore, the CFRP reinforcements have two different contributions to the
composite pipes: the longitudinal fibers enhance the pipe’s load carrying capacity directly, and the annular fibers
supply the radial restriction and change the failure modes after yield which resulted with the increase in the load
carrying capacity indirectly.

3.2 Set2 : Long Pipes

All long pipes in Set 2 under the axial compressive load buckled integrally before Al yield. The lateral displacement
in the middle of pipes is not obvious but goes up quickly when a certain axial load arrives. ACCP70-1, the first

715
specimens in Set 2, tested on a ball and socket support at the ends which restricted the ends from free rotation in the
test and is a not a hinge. After it, the setup was developed into hinge supports. The relations between the axial load P
and the lateral displacement Δ for each slenderness group are shown in Figure 3 except for ACCP70-1. It can be
found that the maximum load capacity can be enhanced obviously with the increase of CFRP layers. The bending
rigidities of ACCP60 and ACCP70 decrease quickly after buckling because Al yield come immediately, while
ACCP120 shows a slow degradation rate.

150 150 40

120 120
30

90 90

P (kN)

P (kN)
P (kN)

20
60 60
ACCP 60-0 ACCP 70-2 ACCP 120-0
10 ACCP 120-1
30 ACCP 60-2 30 ACCP 120-2
ACCP 60-3 ACCP 70-3
ACCP 120-3
0 0 0
0 5 10 15 20 25 0 5 10 15 20 0 15 30 45 60
Δ (mm) Δ (mm) Δ (mm)

(a) ACCP60 Group (b) ACCP70 Group (c) ACCP120 Group

Figure 3: Load-lateral Displacement Relations of Set 2

4. CONCLUSIONS
Based on the tests of the CFRP-Al composite pipes under axial compressive load, the following conclusions can be
drawn.
(1) There was no debonding found between CFRP layer and Al in tests. CFRP-Al composite is an effective concept
to build axial compressive elements.
(2) The bearing strength of CFRP-Al composite pipes can be improved considerably which rise with the increase of
the CFRP layers. It comes from two parts: the longitudinal fibers enhance the longitudinal and flexural stiffness and
strength directly; and the annular fibers provide the radial restriction to change the failure mode.
(3) The stable strength of CFRP-Al composite pipes can be improved considerably which rise with the increase of
the CFRP layers.

ACKNOWLEDGEMENTS
The authors are grateful to the Natural Science Foundation of China for their support to the research presented here
through a national key project on the application of FRP composites in civil engineering in China (Project No.
50238030).

REFERENCES
Feng, P., Ye, L. P. (2002). “FRP structures and FRP composite structures in structural engineering”. Proceeding Of
2nd Academic And Communicative Meeting On The National FRP Application Technology In Civil Engineering.
Tsinghua University Press, Beijing, pp27-40 (in Chinese).
Harris, C. E., Starnes Jr., J. H., Shuart, M. J. (2002). “Design and manufacturing of aerospace composite structures,
state-of-the-art assessment”. Journal of Aircraft, Vol.39, No.4, pp 545-560.
Lee, D. G., Kim, H. S., Kim, J. W., et al. (2004). “Design and manufacture of an automotive hybrid
aluminum/composite drive shaft”. Composite Structures, Vol.63, No.1, pp87-99.
Mazzolani, F.M. (1985). Aluminum alloy structures, Pitman, Boston.
Qian, P., Ye, L. P. (2004) “Experimental study of CFRP tubes under uniaxial loading”. China Industrial
Construction., Vol. 34, No. 4, pp211-14 (in Chinese).
Ye, L.P., Feng, P. (2006). “Applications and development of fiber-reinforced polymer in engineering structures”.
China Civil Engineering Journal, Vol. 39, No. 3, pp25-37 (in Chinese).

716
Third International Conference on FRP Composites in Civil Engineering (CICE 2006)
December 13-15 2006, Miami, Florida, USA

REHABILITATION OF CRACKED ALUMINUM STRUCTURES WITH

GFRP COMPOSITES CONSIDERING FATIGUE
Chris P. Pantelides
(Professor, University of Utah, Salt Lake City, Utah, USA)

Justin D. Nadauld
(Engineer, Reaveley Engineers & Associates, Salt Lake City, Utah, USA)

ABSTRACT
Cracks develop in the welds between branch and chord members of aluminum overhead sign structures due to
fatigue stresses by wind-induced vibration. The cracks propagate to complete fracture of the welds which causes
signs to fall. The original design of aluminum overhead sign structures did not consider fatigue as a limit state. A
rehabilitation method for cracked aluminum welded connections using Glass Fiber Reinforced Polymer (GFRP)
composites is investigated. The paper presents results from constant amplitude fatigue tests of aluminum welded
connections with no known cracks, cracked aluminum connections from actual sign structures rehabilitated with
GFRP composites, and aluminum connections with 90% of the weld removed that were rehabilitated with GFRP
composites. The fatigue limits of the welded connections and rehabilitated connections are established for various
stress ranges including the constant amplitude fatigue limit threshold. The fatigue tests show that the rehabilitated
connection from actual sign structures exceeded the fatigue limits of the welded connection with no known cracks.
The repaired connection with 90% of the weld removed satisfied the constant amplitude fatigue limit. A cumulative
damage index is established which can be used to develop a fatigue reduction factor for the rehabilitation design
using GFRP composites.

KEYWORDS
Aluminum, Connections, Fatigue, FRP composites, Rehabilitation

1. INTRODUCTION
A large percentage of fatigue cracks in aluminum structures are caused by wind-induced vibration of members that
are too slender (Sharp et al., 1996). Cracks develop and propagate at the welded connections between branch
(diagonal) and main chord members of overhead sign structures as shown in Figure 1. The truss considered here,
commonly referred to as a “tri-chord”, has three chords and is shown in Figure 1 with a typical connection after it
has been repaired using GFRP composites. This research is concerned with the fatigue performance of specimens
rehabilitated with GFRP composites using the method developed by Pantelides et al. (2003). The connection tested
in fatigue consists of two branch members attached to a main chord; the branches are oriented at an angle 42 degrees
from the chord axis, as shown in Figure 2. The members consist of 6mm thick round aluminum tubing; the chord
diameter is 102mm and that of the branches 64 mm. Aluminum alloy ASTM 6061-T6 was used as the tubing
material; the mechanical properties of the aluminum and the GFRP composite architecture are given by Pantelides et
al. (2003). The typical rehabilitated and repaired connection was strengthened with a GFRP layup consisting of two
layers of unidirectional tendon (ultimate strength = 517MPa) and two layers of bear tubular weave braid (ultimate
strength = 207MPa). The test units were obtained from structures in service for many years, so 1 million fatigue
cycles were considered sufficient. Four stresses were used to obtain stress versus number of cycles to failure fatigue
curves. The highest stress corresponded to the static design limiting stress of 83MPa, followed by 59MPa, and
39MPa; the lowest stress level was 19MPa, which is 1.4 times the constant amplitude fatigue limit threshold of the
connection with the specific details of 13MPa, as specified in the Standard Specifications for Structural Supports for
Highway Signs, Luminaires and Traffic Signals (AASHTO, 2001). The frequency of the fatigue cycles used was 2
Hz as shown in Figure 2, at a stress ratio R=0.2, defined as the ratio of the minimum to maximum stress applied.

717
 GFRP
Cracked Weld

Figure 1: Overhead sign structure rehabilitated with GFRP composites and close up of partially cracked weld

90 MAXIMUM
80 STRESS

eee
70

STRESS RANGE
R = 2/12 =
Stress (MPa)

60

MEAN STRESSsss
50 0 167
40
30
20
10 MINIMUM STRESS

0
0 0.125 0.25 0.375 0.5 0.625 0.75 0.875 1
Time (sec)

Figure 2: Test setup for fatigue testing with constant amplitude fatigue

2. EXPERIMENTAL RESULTS AND CUMULATIVE DAMAGE MODEL

Constant amplitude fatigue tests were carried out for three series of aluminum connections obtained from the field:
(i) aluminum connections with no known cracks, identified as AL; (ii) cracked aluminum connections rehabilitated
with GFRP composites, identified as R; and (iii) aluminum connections with 90% of the weld removed and
subsequently repaired with GFRP composites, identified as WRR. A cumulative damage model is defined, in terms
of the initial and residual stiffness of the structure as:
Di = 1 − ki / ko (1)
where Di = cumulative damage at cycle (i), k o = initial structure stiffness, and k i = structure stiffness at cycle (i).
The cumulative damage index of Eq. (1) is a measure of the loss of stiffness of the test unit in the fatigue tests; it
provides a way to compare the results of fatigue tests for the three series. In the fatigue tests carried out, the number
of cycles to failure was indicated as N f . Two failure modes were observed in Series AL: (1) fracture through the
throat of the weld, and (2) crack formation at the throat of the weld through the base metal that propagated to
fracture of the chord. Two failure modes were observed in Series R: (1) cracking through the throat of the weld and
GFRP tensile failure, as shown in Figure 3 for unit R1 tested at 83 MPa maximum stress, and (2) cracking through
the toe of the weld, followed by cracking through the throat of the weld and GFRP tensile failure, as shown in
Figure 4 for unit R2 tested at 59 MPa maximum stress. The cumulative damage index for test unit R1 is shown in
Figure 3; the weld fractured at 5000 cycles, which is marked by the increase in cumulative damage, and the GFRP
failed in tension at 6763 cycles. The stages of failure for unit R2 are shown in Figure 4; weld fracture occurred at
12000 cycles and GFRP tensile failure at 69194 cycles. Two main failure modes were observed in Series WRR: (1)
adhesive failure with GFRP composite tearing, and (2) GFRP composite tensile failure.

718
 1
Nf = 6,763 III-a
DNf = 0.33

Nf

Cumulative Damage
0.75
a
0.5 FRP Failure
Weld Fracture

0.25

0
0 1000 2000 3000 4000 5000 6000 7000
Cycles

Figure 3: Weld fracture and GFRP composite tensile failure for fatigue test of R1 and cumulative damage index Di

Nf = 69,194 III-b
DNf = 0.79
1

Weld FRP Failure Nf

Cumulative Damage

0.75
Fracture
a
0.5

0.25

0
1 .10 2 .10 3 .10 4 .10 5 .10 6 .10 7 .10
4 4 4 4 4 4 4
0
Cycles

Figure 4: Fracture of weld, base material and GFRP for fatigue test of R2 and cumulative damage index Di

Figure 5 shows the S-N curves for all fatigue tests. The connections of Series AL and the GFRP rehabilitated
connections of Series R show similar behavior with Series R showing slightly better fatigue behavior for the lower
maximum stress levels. The units with the weld 90% removed and repaired with GFRP composites, Series WRR
did not perform as well as Series AL and Series R in the high maximum stress levels; however, Series WRR
performed as well as Series AL in the lowest maximum stress. It should be stressed that the lowest maximum stress
of 19MPa is still 1.4 times the constant amplitude fatigue limit threshold specified by AASHTO (2001). Given this
fact, and the fact that Series R, and WWR reached 1 million cycles, it can be concluded that rehabilitation of the
aluminum joints with GFRP composites concerning fatigue resistance was successful.

3. CONCLUSIONS
Constant amplitude fatigue tests were carried out for three series of aluminum connections obtained from actual
overhead sign structures: (i) aluminum connections with no known cracks; (ii) cracked aluminum connections
rehabilitated with GFRP composites; and (iii) aluminum connections with 90% of the weld removed and
subsequently repaired with GFRP composites. Four maximum stresses were used to obtain stress to number of
cycles to failure fatigue curves. The highest maximum stress corresponded to the static design limiting stress and
the lowest maximum stress corresponded to 1.4 times the constant amplitude fatigue limit threshold of the
connection with the specific details.

Two failure modes were observed for the aluminum connections with no known cracks in the fatigue tests: (1)
fracture through the throat of the weld, and (2) crack formation at the throat of the weld through the base metal that

719
 100
90
80
70

LOAD (kN)
60
50 Series WRR
40
Series R
30
Series AL
20
10
0
1 10 100 1000 10000 100000 1000000

CYCLES

Figure 5: S-N curves for AL, R, and WRR specimens

propagated to fracture of the chord. Two failure modes were observed for cracked aluminum connections
rehabilitated with GFRP composites: (1) cracking through the throat of the weld and GFRP tensile failure, and (2)
cracking through the toe of the weld, followed by cracking through the throat of the weld and GFRP tensile failure.
For aluminum connections with 90% of the weld removed and subsequently repaired with GFRP composites two
failure modes were observed: (1) adhesive failure with GFRP composite tearing due to a short bond length, and (2)
GFRP composite tensile failure. The cracked aluminum connections rehabilitated with GFRP composites exceeded
the fatigue limit of the aluminum welded connections with no known cracks. The GFRP repaired connections with
90% of the weld removed satisfied the constant amplitude fatigue limit threshold.

A cumulative damage index was established which can be used to detect damage and compare between tests; it can
also be used to develop a fatigue reduction factor for the rehabilitation design of cracked aluminum connections
using GFRP composites. The fatigue test results agree with the results of the static tests and confirm that the GFRP
repaired connections behave as well as the aluminum connections with no known cracks for the range of loading
expected in service. The repair technique with GFRP composites is adequate and can be used in construction.

4. ACKNOWLEDGMENTS
The writers would like to thank the New York State DOT and Utah DOT for financial support and Air Logistics
Corporation for the in-kind support. The writers would like to thank Harry L. White of NYSDOT; Doug Anderson
of UDOT; and Dr. Larry Cercone, Franz Worth, and Steve Bazinet of Air Logistics Corp. In addition, the writers
would like to thank Professor Lawrence D. Reaveley, Chris Delahunty, and Oliver Burt, of the Civil and
Environmental Engineering Department of the University of Utah.

5. REFERENCES
American Association of State Highway and Transportation Officials (AASHTO). (2001). Standard Specifications
for Structural Supports for Highway Signs, Luminaires and Traffic Signals, 4th Edition, Washington, D.C.
Pantelides, C. P., Nadauld, J., and Cercone, L. (2003). “Repair of cracked aluminum overhead sign structures with
glass fiber reinforced polymer composites”. Journal of Composites for Construction, Vol. 7, No. 2, pp 118-126.
Sharp, M. L., Nordmark, G. E., and Menzemer, C. C. (1996). Fatigue Design of Aluminum Components and
Structures, McGraw-Hill, New York.

720
Third International Conference on FRP Composites in Civil Engineering (CICE 2006)
December 13-15 2006, Miami, Florida, USA

EFFECTIVENESS OF DIFFERENT COMPOSITE MATERIALS FOR

REPAIR OF STEEL BRIDGE GIRDERS
Amr Shaat
(Doctoral Candidate, Queen’s University,Kingston,Ontario, Canada)

Amir Fam
(Associate Professor and Canada Research Chair in Innovative and Retrofitted Structures, Queen’s
University,Kingston,Ontario, Canada)

ABSTRACT
Steel-concrete composite bridge girders damaged due to corrosion or fatigue cracks can be repaired by adhesively
bonded carbon fiber reinforced polymer (CFRP) sheets. This paper describes the results of the first phase of an
experimental study on beams of 1960 mm span length. Each beam is composed of W150 x 22 steel section acting
compositely with a 75 mm thick by 465 mm width concrete slab to simulate an actual bridge girder condition. A
severe damage was introduced by saw-notching the steel tension flange through the entire thickness at mid-span.
Two different types of CFRP sheets were then applied to the tension flange to restore the stiffness and strength
capacities of the undamaged control beam. The beams were tested to failure in four-point bending. The findings of
this study indicate that proper selection of composite material properties and immediate application of FRP sheets
after steel surface preparation are crucial for a successful repair.

KEYWORDS
Repair, Steel, Bridge, CFRP, Sheets.

1. INTRODUCTION
Fatigue and corrosion are main reasons for deterioration of steel bridges. The repair cost of steel structures, in most
cases, is far less than the cost of replacement of the whole structure. Traditional methods of repair such as welding
or bolting steel plates have a number of shortcomings, including the added self weight of steel plates, long
installation time, and the need for an elaborate and expensive shoring system. The use of FRP materials for repair
and strengthening of structures has been shown to be both time and cost effective. Steel-concrete composite bridge
girders damaged due to corrosion or fatigue cracks can be repaired by epoxy bonding CFRP sheets, particularly due
to their excellent corrosion resistance and fatigue properties (Shaat et al., 2004). Artificial degradation techniques
such as complete removal of the tension flange at mid-span were previously utilized to simulate fatigue cracks or
section loss (Liu et al., 2001). Efficiency of repair material is highly dependent on its elastic modulus, compared to
the steel modulus (Hollaway and Cadei, 2002, and Buyukoztruk et al., 2004)

2. EXPERIMENTAL PROGRAM
The effectiveness of bonding two different types of CFRP sheets, namely Type 1 CFRP and Type 2 CFRP, in
restoring both the stiffness and strength of steel beams acting compositely with concrete slabs was examined. Four
steel-concrete specimens were loaded to failure through four-point bending tests, including an intact (un-notched)
control specimen (B1). The tension flanges of three specimens were notched at mid-span throughout the entire
flange thickness and width to simulate severely damaged bridge girders. The three notched specimens were then
repaired by bonding either Type 1 CFRP for beam (B2) or Type 2 CFRP for beams (B3 and B4). Both the number of
layers and width of each layer for both types of CFRP sheets were designed to achieve the same forces at failure
when the sheets rupture. The load carrying capacity as well as the initial stiffness was measured and compared to

721
control beam B1, to evaluate the effectiveness of this repair technique in restoring the original capacity. The
following is a brief description of the materials used, fabrication process, test setup and instrumentations.

2.1 Materials

Four W150x22 hot rolled steel sections with 400 MPa nominal yield strength were used for the steel beams. A 75
mm thick by 465 mm wide concrete slab with an average compressive strength of 50 MPa was cast on top of the
steel section which had steel shear connectors embedded into the slab. Type 1 CFRP consisted of a 0.54 mm thick
lamina, which has tensile strength and modulus of 510 MPa and 230 GPa, respectively. Type 2 CFRP consisted of a
0.89 mm thick lamina, which has tensile strength and modulus of 1130 MPa and 107 GPa, respectively. Fourteen
layers of 150 mm wide sheets were used in case of Type 1 CFRP for beam B2, and five layers of 115 mm wide
sheets were used in case of Type 2 CFRP for beams B3 and B4 to achieve the same strength. A glass-FRP (GFRP)
lamina, 1.46 mm thick, with tensile strength and modulus of 269 MPa and 14 GPa, respectively, was placed between
the steel surface and CFRP layers in all beams to prevent galvanic corrosion between steel and CFRP. Tyfo S epoxy
resin was used for both GFRP and CFRP sheets to bond the fibres together and to the steel surface.

2.2 Fabrication of Test Beams

The steel beams were first cut to 2030 mm long sections. The entire flange width and thickness at mid-span of the
tension flange was cut using a saw with a 1.4 mm thick blade. Conventional 41 mm long and 9.5 mm diameter
Nelson studs were welded in pairs at a longitudinal spacing of 60 mm along the compression flange. A 150 x 150 x
5 mm welded wire mesh reinforcement was provided at mid-thickness of the concrete slab. Concrete was then
pored, consolidated and cured for about 90 days. The bottom surface of the steel tension flange was sand blasted.
FRP sheets were immediately applied on beams B2 and B3, within 24 hours, whereas the notched flange of beam B4
was covered with a thin layer of oil for a period of 22 months, to avoid corrosion, before cleaning it again with
acetone and applying the FRP sheets. This was done to investigate the effect of delayed application of FRP after
sand blasting on bond.

2.3 Test Setup and Instrumentations

All Four beams were tested using a simply supported configuration with a span of 1960 mm between the centerlines
of the supports. The beams were monotonically loaded using four-point bending, as shown in Figure 1, with a
distance of 400 mm between the two applied loads. Two linear potentiometers (LPs) were placed on both sides of
the beams at mid-span to measure the vertical deflection. The longitudinal strains along the tension flange were
measured using several 5 mm long electric resistance strain gauges, spaced as shown in Figure 1. Another strain
gauge was attached to the lower side of the steel compression flange at mid-span. Displacement-type strain gauge
transducers (PI gauges) installed over a gauge length of 100 mm were also attached to the top of concrete slab and
the steel tension flange, as shown schematically in Figure 1. The load was applied using stroke control at a rate of
0.75 mm/min, up to failure. A data acquisition system was used to monitor and record test data.

Spreader
beam 400
Concrete slab
75 mm thick.

W 150x22 Strain gauges PI gauges Stiffener Stiffener

angle angle

75 25 CFRP sheets
225 125
1960

Figure 1: Test setup

3. TEST RESULTS AND DISCUSSION

This section presents the test results in terms of load-deflection and load-strain curves.

722
Figure 2(a) shows the load versus mid-span deflection of the three repaired beams, compared to the intact control
beam. The figure shows that beam B2 with Type 1 CFRP sheets resulted not only in restoring the elastic stiffness
and ultimate load of the undamaged beam but also exceeding them by 26% and 10%, respectively, as listed in Table
1. On the other hand, Type 2 CFRP sheets in B3 could not fully compensate for the loss of the entire steel flange.
Figure 2(b) shows that the behavior of the intact beam was essentially linear up to first yielding of the steel section
at a load level of 305 kN, based on strain gauge readings. Specimen B2 showed an extension of the linear part of the
curve, beyond the observed yielding point of B1. Repaired beam B2 reached a significantly higher capacity (P=394
kN) than the intact beam B1 before the 14 layers of Type 1 CFRP sheets ruptured due to reaching their maximum
tensile strength. The load then dropped to 33% of its maximum value (i.e. to P=129 kN) which emphasizes the
significant contribution (67%) of the CFRP laminate, compared to a damaged and unrepaired specimen. On the
other hand, specimens B3, which was repaired with Type 2 CFRP, reached maximum strength of 311 kN, which
corresponds to 87% of the strength of the intact beam B1. The CFRP sheets were then completely debonded, which
could be attributed to their lower modulus of elasticity compared to steel, unlike Type 1 CFRP. Moreover, the very
low strength level of beam B4 (179 kN) is attributed to the long time interval between sand blasting and CFRP
application. Beams B3 and B4 started to react nonlinearly upon initiation of CFRP sheets debonding at load levels of
185 and 104 kN, respectively. Figure 2(b) shows that yielding starts at about 2000 μ strain for the control intact
beam B1 and that ultimate rupture strain of Type 1 CFRP sheets in tension occurs at 2200 μ strain, whereas
debonding of Type 2 CFRP occurred at 5100 and 3600 μ strain for beams B3 and B4, respectively.

Table 1: Summary of Test Results

Beam I.D. B1 B2 B3 B4
Maximum Load (kN) 357 394 311 179
Elastic stiffness (kN/mm) 34 43 29 31
% age increase in maximum load --- +10 -13 -50
% age increase in elastic stiffness --- +26 -15 -9

400 400
B2 FRP rupture
B2
350 B1 350
B1
300 B3 300
B3
250 250
Load (kN)

FRP
debonding
200 B4
200

150 150
B4
100 P 100 P

50 50
Strain gauge
0 0

0 5 10 15 20 25 30 00
35 2000 4000 6000 8000 10000 12000 14000 16000

Deflection (mm) μ strain

(a) Load versus mid-span deflections (b) Load versus mid-span strains

Figure 2: Behavior of Test Specimens

4. FAILURE MODES
The typical failure mode of the intact control beam was yielding of the steel tension flange followed by concrete
crushing. In the specimen repaired with Type 1 CFRP, sheets were ruptured as shown in Figure 3(a) when their
strains reached the maximum rupture strain of 2200 μ strain. On the other hand, the specimens repaired with Type 2
CFRP sheets were debonded at the interface between GFRP and steel, as shown in Figure 3(b). The loss of the
CFRP sheets in both specimens resulted in very high strains above the notch, which led to expanding the notch and
development of a crack propagating within the web, as shown in Figure 3(c).

723
 Rupture @ Vertical
mid-span propagation

Steel flange

Debonded
laminate

(a) Rupture of Type 1 CFRP (b) Debonding of Type 2 CFRP (c) Notch propagation

Figure 3: Modes of Failure

5. CONCLUSIONS
The findings of this study indicate that CFRP materials can effectively be used for repair of cracked steel-concrete
composite beams. The following conclusions can be drawn:
1. The original strength and stiffness of the undamaged beams can be fully restored by bonding high modulus
CFRP sheets. In this study, the specimen repaired with 14 layers of Type 1 CFRP recovered all the loss in
strength and stiffness and even achieved an increase of 10% and 25%, respectively, when compared to the
intact control beam.
2. The beam repaired with Type 2 CFRP recovered only 87% and 85% of the original strength and stiffness of
the intact beam. This lack of efficiency for Type 2 CFRP was due to the premature debonding mode of
failure which could be due to its low value of elastic modulus.
3. Immediate application of CFRP sheets on steel surface after sand blasting is very crucial to achieve the
highest possible bond strength.
4. No debonding between FRP and steel was observed for the specimen repaired with the high modulus
CFRP, which confirms that debonding failure could not only be dependent on surface preparation
procedure but also on the stiffness of FRP.

6. ACKNOWLEDGMENT
The authors wish to acknowledge the financial support provided by the Natural Sciences and Engineering Research
Council of Canada (NSERC). The authors also wish to thank Fyfe.Co.LLC and Mitsubishi Chemical for providing
the FRP materials.

7. REFERENCES
Buyukozturk, O., Gunes, O., and Karaca, E., (2004), “Progress on understanding debonding problems in reinforced
concrete and steel members strengthened using FRP composites” Journal of Construction and Building Materials,
Vol. 18, pp 9-19.
Hollaway, L.C. and Cadei, J., (2002) “Progress in the technique of upgrading metallic structures with advanced
polymer composites”, Journal of Progress in Structural Engineering and Materials, Vol. 4, No. 2, pp 131-148.
Liu, X., Silva, P. F., and Nanni, A., (2001) “Rehabilitation of steel bridge members with FRP composite materials.”
Proceedings of the International Conference on Composites in Construction, Editors: Figueiras J, Juvandes L, Furia
R, Porto, Portugal, pp. 613-617.
Shaat, A., Schnerch, D., Fam, A., and Rizkalla, S., “Retrofit of steel structures using fiber-reinforced polymers
(FRP): State-of-the-art.” The 83rd annual meeting of the Transportation Research Board (TRB), Washington, D.C.,
(2004) CD-ROM (04-4063).

724
Third International Conference on FRP Composites in Civil Engineering (CICE 2006)
December 13-15 2006, Miami, Florida, USA

EXPERIMENTAL AND NUMERICAL ANALYSIS OF THE

STRUCTURAL RESPONSE OF FRP-STRENGTHENED
COLD-FORMED STEEL COLUMNS
Ben Young
Associate Professor, Dept. of Civil Engineering, The University of Hong Kong, Pokfulam Road, Hong Kong

Nuno Silvestre1, Dinar Camotim2

1 2
Assistant Professor, Associate Professor, Dept. of Civil Engineering, ICIST/IST, Technical University of Lisbon,
Av. Rovisco Pais, 1049-001 Lisboa, Portugal

ABSTRACT
This paper reports the results of an experimental and numerical investigation on the non-linear behavior and load-carrying
capacity of CFRP-strengthened cold-formed steel lipped channel columns. The experimental program involved two test series,
comprising a total of 19 short and long fixed-ended lipped channel columns − while the former buckle in local-plate modes,
the latter exhibit distortional buckling. The columns were strengthened with carbon fiber sheets glued at different outer
surface locations (web, flanges or lips) and having fibers oriented either longitudinally or transversally − since the aim of the
study is to assess the influence of the CFRP sheets on the column structural response, bare steel specimens were also tested.
The experimental results, which consist of non-linear equilibrium paths (applied load vs. axial shortening) and ultimate
strength values (most of them associated with local-plate and/or distortional failure mechanisms), are subsequently used to
calibrate and validate geometrically non-linear numerical analyses based on shell finite element models, carried out with the
code ABAQUS and adopting an elastic-plastic constitutive law to describe the steel material behavior. Finally, on the basis of
both the experimental and numerical results obtained, some relevant conclusions are drawn concerning the most effective
CFRP sheet location and fiber orientation to strengthen columns affected by local-plate and distortional buckling.

KEYWORDS
Cold-formed steel columns; Carbon fiber sheets; FRP-strengthening; Local-plate failure; Distortional failure; Ultimate strength

1. INTRODUCTION
Given the considerable stiffness of cold-formed steel members, their strengthening by means of FRP composite requires the
use of expensive high-strength fibers, a fact that strongly affects the economical viability of this procedure. However, the
increasingly competitive cost of carbon fibers, together with their quite high stiffness and strength properties, has altered this
situation − indeed, carbon fiber sheets have been shown to be particularly well suited to reinforce steel plates. As far as the
strength is concerned, the failure of CFRP-reinforced cold-formed steel members may stem from (i) local (local-plate and/or
distortional) or global buckling of the steel-CFRP member, (ii) rupture or debonding of the CFRP sheet or (iii) a
combination of both. Thus, an efficient (safe and economical) design of such members must be based on an in-depth
knowledge concerning all these potential failure modes. The objective of this work is two-fold: (i) to report the results of an
experimental investigation aimed at assessing how the CFRP-strengthening influences (enhances) the non-linear behavior
and load-carrying capacity of cold-formed steel lipped channel columns and also (ii) to use these results to calibrate
and validate numerical analyses performed in the code ABAQUS and based on shell finite element member discretisations.

2. EXPERIMENTAL INVESTIGATION
The tests were conducted on fixed-ended cold-formed steel lipped channel columns strengthened with carbon fiber sheets
(CFS), which are glued to their outer surfaces in different locations (web, flanges, lips) and with the fibers oriented
longitudinally or transversally − for reference purposes, a few bare steel columns were also tested. The specimens were
brake-pressed from high strength zinc-coated grades G450 and G550 structural steel sheets (i) with nominal 0.2% proof
(yield)) stresses equal to 450 and 550 MPa, and (ii) conforming to the Australian Standard AS 1397 (1993). All the column

725
specimens have nominal web width Bw=125 mm, flange width Bf=102 mm and lip width Bl=14 mm and the test program
comprises two test series: (i) one involving 9 short columns (L=600 mm) made of G550 steel sheets with nominal thicknesses
t=1.0 mm and (ii) the other consisting of 10 long columns (L=2200 mm) made of G450 steel sheets with nominal thicknesses
t=1.5 mm − the cross-section dimensions (see Fig.1(a)) and lengths effectively measured for each column specimen are
shown in Tables 1 and 2. The column specimen end sections were welded to 25 mm thick steel plates, subsequently bolted to
the bearing plates − this procedure ensures full contact between the column specimen ends and the bearing plates, i.e., full
warping restraint. The column specimen labeling provides information about the test series and the location and orientation of
the CFS: (i) the first letter indicates whether the specimen belongs to the short (“S”) or long (“L”) column test series, (ii) the
following letters indicate if the specimen has no strengthening (“NIL”), CFS located in the web (“W”), CFS located in the
web and flanges (“WF”) or CFS located in the web, flanges and lips (“WFL”), and (iii) the numbers specify whether the CFS
are oriented longitudinally (“0”) or transversally (“90”). Finally, the letter “R” identifies a repeated test.

The material properties of the cold-formed steel column specimens were obtained through tensile coupon tests − the coupons
were extracted from the specimen central web regions, in the longitudinal direction, and their dimensions conformed to the
Australian Standard AS 1391 (1991) for tensile testing of metals: 12.5 mm wide coupons of gauge length 50 mm. The values
measured in the coupon tests are given next, for the short and long columns (nominal 0.2% proof stresses σ0.2.s=550 MPa and
σ0.2.l=450 MPa): (i) static 0.2% proof stresses σ0.2.s=610 MPa and σ0.2.l=521 MPa, (ii) Young’s moduli Es=207 GPa and
El=218 GPa, (iii) static tensile strengths σu.s=626 MPa and σu.l=546 MPa and (iv) elongations after fracture εf.s=9.2% and
εf.l=11.4% (based on the 50 mm gauge length). As for the CFS thickness and material properties, the values provided by the
fabricators were adopted − they read: (i) thickness tCFS=0.11 mm, (ii) tensile strength σu.CFS=4200 MPa, (iii) tensile modulus
ECFS=235 GPa and (iv) elongation after fracture εf.CFS=1.8%. Moreover, the (single) CFS were attached to the outer surfaces
of the zinc-coated cold-formed steel columns by means of an epoxy resin with tensile strength and modulus equal to
σu.r=30 MPa and Er=3.5 GPa − after attaching the CFS, the column specimens were completely cured for 7 days.

Fig. 1(b) shows the failure mode of a short column specimen and also provides an overall view of the test rig and experimental
set-up employed − (i) the DARTEC servo-controlled hydraulic testing machine used to compress the column specimens, (ii) the
two steel plates welded to the specimen ends, (iii) the rigid flat bearing plate connected to the testing machine upper support
(and bolted to the specimen top end plate) and restrained against flexural (minor/major axis) and twist rotations, and (iv) the
lower special bearing, bolted to the specimen bottom end plate and subsequently restrained against flexural and twist rotations

Top
Rigid flat end plate
bearing plate

Transducer

Column
Bl specimen

t
Bf
Bottom
Special end plate
r
bearing
Bw
(a) (b) (c)
Figure 1: (a) Cross-section geometry, (b) experimental set-up and test view and (c) FEM failure mode (S-F-90)

Table 1: Measured dimensions and experimental/numerical ultimate strengths of the short column specimens
Specimen Bw (mm) Bf (mm) Bl (mm) t (mm) r (mm) L (mm) PExp (kN) PNum (kN) PNum/PExp
S-NIL 124.1 101.9 13.8 1.071 3.0 600 54.9 55.3 1.007
S-W-0 124.1 101.9 13.7 1.074 3.0 600 55.9 57.0 1.020
S-F-0 124.0 102.0 13.7 1.075 3.0 600 56.0 58.1 1.038
S-WF-0 123.6 101.9 13.7 1.068 3.0 602 60.2 62.1 1.032
S-WFL-0 124.0 101.9 13.7 1.070 3.0 599 61.4 62.3 1.015
S-W-90 124.1 102.0 13.6 1.068 3.0 600 55.6 54.9 0.987
S-F-90 123.7 101.9 13.8 1.073 3.0 601 56.4 57.1 1.012
S-WF-90 124.1 101.9 13.7 1.061 3.0 601 63.2 59.3 0.938
S-WF-90-R 124.1 101.9 13.8 1.073 3.0 600 62.4 59.3 0.950
Mean 124.0 101.9 13.7 1.070 3.0 --- --- 1.000
COV 0.002 0.000 0.005 0.004 0.000 --- --- 0.035

726
Table 2: Measured dimensions and experimental/numerical ultimate strengths of the long column specimens
Specimen Bw (mm) Bf (mm) Bl (mm) t (mm) r (mm) L (mm) PExp (kN) PNum (kN) PNum/PExp
L-NIL 126.3 102.6 13.8 1.552 3.0 2199 85.2 91.3 1.072
L-NIL-R 126.6 102.8 13.6 1.553 3.0 2200 86.7 91.3 1.053
L-W-0 126.4 102.7 13.8 1.556 3.0 2202 95.5 99.3 1.040
L-F-0 126.4 102.6 13.9 1.560 3.0 2199 90.1 92.0 1.021
L-WF-0 125.8 102.7 13.8 1.556 3.0 2201 98.6 100.3 1.017
L-WFL-0 125.7 102.8 13.8 1.555 3.0 2200 102.1 100.5 0.984
L-WFL-0-R 126.4 102.8 13.7 1.553 3.0 2200 100.4 100.5 1.001
L-W-90 125.6 102.7 13.7 1.549 3.0 2200 88.1 89.0 1.010
L-F-90 125.4 102.6 13.7 1.567 3.0 2199 92.6 88.2 0.952
L-WF-90 125.2 102.7 13.8 1.559 3.0 2199 100.9 100.5 0.996
Mean 126.0 102.7 13.8 1.556 3.0 --- --- 1.015
COV 0.004 0.001 0.006 0.003 0.000 --- --- 0.036

(by means of vertical and horizontal bolts that lock the bearing in position, once full contact is achieved) − these special
bearings materialize fixed-ended supports, with the upper one allowing for longitudinal translations (to apply the compressive
force). Three displacement transducers were used to measure the vertical motion of the top end plate, thus monitoring (i) the
axial shortening of the column specimen and (ii) the effective absence of flexural rotations. Displacement control was used
to drive the hydraulic actuator at a constant speed of 0.2 mm/min in all cases − this loading arrangement made it possible to
continue column specimen tests beyond the ultimate load level, i.e., to determine the post-ultimate equilibrium path
descending branch. A data acquisition system was used to record the applied load and displacement transducer readings at
regular intervals throughout the duration of the tests − the static loads were recorded by pausing the applied straining for
1.5 min and, in order to assess the column specimen ultimate strengths as accurately as possible, the recording frequency was
increased at the onset of collapse (i.e., when a considerable column axial stiffness reduction was detected).

All the experimental ultimate strength values (PExp) obtained, concerning the short and long cold-formed steel lipped channel
columns with and without CFS-strengthening, are given in Tables 1 and 2. It should be noted that all (short and long) column
failures were due to local-plate/distortional buckling mode interaction, as illustrated experimentally and numerically in Figs.
1(b)-(c), for the case of the column specimen S-F-90. Then, the excessive column wall deformations caused by the buckling-
triggered collapse, clearly visible in Fig. 1(b), also cause the occurrence of CFS debonding in the post-failure stages. In order
to assess the reliability of the experimental set-up and procedure, three tests were repeated − the differences between the
ultimate strengths determined for each pair of (supposedly) identical column specimens were 1.3% (S-WF-90), 1.8% (L-NIL)
and 1.7% (L-WFL-0). Finally, curves providing the variation of the column axial shortening with the applied load (P
vs. u) for the column specimens S-WFL-0 and L-WFL-0/L-WFL-0-R are shown in Figs. 2(a) and 2(b), respectively.

3. NUMERICAL INVESTIGATION
The finite element code ABAQUS (HKS, 2002) was used to simulate the non-linear behavior and estimate the ultimate loads of
the cold-formed steel lipped channel columns with and without CFS-strengthening. All columns were discretised into meshes
of S4R shell finite elements (4-node elements with reduced integration) and the corners were not taken into account in the
modeling. In order to simulate the column fixed support conditions, rigid plates (modeled using 3-node rigid elements R3D3)
were attached to the column end sections and their centroidal translations and rotations were prevented − only the axial
translation of one end section was left free, in order to enable the application of the compressive load. The short and long
column meshes involved 15 mm×10 mm (length-width) and 30 mm×10 mm finite elements, which correspond to (i) 1462 and
2250 elements, (ii) 1505 and 2625 nodes, and (iii) 8622 and 15342 degrees of freedom, respectively. Moreover, the numerical
analyses were based on (i) the mean measured values of the cross-section dimensions, given in Tables 1 and 2, (ii) the
nominal lengths L=600 mm and L=2200 mm and (ii) the material properties obtained from the tensile coupon tests (steel) or
provided by the fabricators (CFS). The CFS-strengthened walls were modeled as double-ply plates (one steel and one CFS
ply) and, in order to have an indication concerning the CFS collapse, both the Maximum Stress and Tsai-Hill failure criteria
(Jones 1999) were incorporated in the finite element analysis − these simple models were adopted because the tests showed
that the CFS behave elastically until the columns reach their ultimate strengths (CFS debonding takes place afterwards).

Initially, column bifurcation analyses were performed in order to evaluate the buckling loads and identify the corresponding
buckling modes − these analyses showed that (i) the short columns buckle in 5 half-wave local-plate modes and (ii) the long
columns buckle in 3 half-wave distortional modes. Then, since there is no available information concerning the column initial
geometric imperfections (they were not measured), the approach employed involved the performance of several preliminary
(elastic) non-linear analyses, each incorporating initial imperfections (i) equally combining the normalized first local-plate and
distortional buckling mode shapes and (ii) having different overall amplitudes. The criterion adopted to choose the
imperfection to be included in the test simulation (geometrically and physically non-linear analysis) was the coincidence (or

727
close vicinity) of the initial portions of the experimental and numerical equilibrium paths (P vs. u) − the imperfection
amplitudes reached by this approach were found to be almost always very near the column wall thickness. In order
to follow the experimental procedure as closely as possible, (i) the non-linear numerical analyses were carried out by means of
an incremental-iterative procedure with an arc-length control strategy, (ii) axial displacements were imposed at the column
free end section and (iii) the applied load was deemed equal to the reactive force at the other (fully fixed) column end section
− it was found that the adoption of this procedure (instead imposing load and reading displacements) precluded the
occurrence of numerical difficulties related to the strong local-plate/distortional buckling mode interaction problems
(mostly in the long column analyses). All the numerical column ultimate strength values (PNum) are shown in Tables 1 and 2.
Moreover, Figs. 2(a) and 2(b) also show the FEM based P vs. u curves for the S-WFL-0 and L-WFL-0/L-WFL-0-R columns.
P (kN) P (kN)
70 120

60
100

50
80
40
60
30
Experimental Experimental (L-WFL-0)
40
20 Numerical
Experimental (L-WFL-0-R)

10 20 Numerical

u (mm) u (mm)
0 0
0 0.5 1 1.5 2 2.5 0 1 2 3 4 5
Figure 2: Axial shortening vs. axial load curves for columns (a) S-WFL-0 and (b) L-WFL-0/L-WFL-0-R
The observation of the experimental/numerical results presented in Tables 1-2 and Figs. 2(a)-(b) shows that:
(i) The experimental and numerical ultimate strength values are in very good agreement − the maximum differences are
5.0% (short columns) and 7.2% (long columns). Moreover, the ratio PNum/PExp mean and COV values read (i1)
1.000 and 0.035 (short columns) and (i2) 1.015 and 0.036 (long columns).
(ii) In the short columns, the CFS-strengthening of only the web or flanges (0º or 90º) barely affects the ultimate load. On the
other hand, the CFS-strengthening of both the web and flanges (0º or, to a larger extent, 90º) leads to non-negligible
ultimate load increases (up to 15.1%). This seems to indicate that the local-plate post-buckling behavior of this particular
short lipped channel column involves equally the web and flange transverse bending (thus, both have to be stiffened).
(iii) In the long columns, the CFS-strengthening of only the web (for 0º) or the flanges (for 90º) causes visible ultimate load
increases (11.7% and 8.7%) − this increase is even higher if both are CFS-strengthened (15.7% and 18.4% for 0º and
90º), even though a higher increase would be achieved by considering 0º and 90º orientations in the web and flanges.
(iv) The CFS-strengthening of the lips causes no relevant ultimate load increases in either short (mostly) or long columns.
(v) The Maximum Stress and Tsai-Hill criteria always predicted the CFS failure after the ultimate load was reached, thus
confirming the experimental evidence that debonding only occurs in the equilibrium path descending branch.

4. CONCLUDING REMARKS
Due to space limitations, it is not possible to present and discuss here all the available results. For more detailed information,
the reader is referred to previous (Silvestre et al. 2004) and subsequent (to be published in the near future) work by the authors.

5. REFERENCES
Australian Standard (1993), Steel Sheet and Strip -- Hot-Dipped Zinc-Coated or Aluminium/Zinc-Coated (AS 1397),
Standards Association of Australia, Sydney.
Australian Standard (1991), Methods for Tensile Testing of Metals (AS 1391), Standards Association of Australia, Sydney.
Hibbit, Karlsson and Sorensen Inc. (HKS) (2002). ABAQUS Standard (version 6.3-1).
Jones, R.M. (1999). Mechanics of Composite Materials, Taylor & Francis, Philadelphia.
Silvestre, N., Camotim, D. and Young, B. (2004). “Buckling behaviour of cold-formed steel members strengthened with
carbon fibre sheets”, Proceedings of 2nd International Conference on Steel & Composite Structures (ICSCS’04 − Seoul,
2-4/9), Editors: C.-K. Choi, H.-W. Lee and H.-G. Kwak, p. 148. (full paper in CD-ROM Proceedings – pp. 412-427)

728
 !"#!$ %& % %'()

) % & %& % *

+% & %& % *

(, % & %& % *

, % & %& % *

!" #
$ ! % & %
# ! '

& # ( ) #( ( ) # #

( * #
* + & ,
' # '
* ( -, .
, !. , !

+ & , / #
& ! ! 0 #
& 0 # &
123

0 #
!" #
$ ! % & % + 0&0 242,

' # !
+ ! #
,

729
 * ( 5 ( 6278
& '
! ( 9
9

gap

w
LCFRP LCFRP

tad

ts
L

!" #$ % "%& #& '( )#*) + + #'%",

%!" ( + '%"& & + # -+$ . !" #$ % "%& #& '( )#

%& / + #+ " + + %.$ #0

1((2 1((2 1((2 1((2 1((2 1((2 1((2
1:9 ;6 6 72 < 24 91 : ;6
1:: ;7 7 77 4 ;: 91 : <7
1:6 971 1 71 6 ;9 91 : 4:
1:7 976 6 71 6 ;: 91 : <2
1:6 981 1 7; ; ;: 91 : 8;
1:9 981 7 77 : ;: 91 : 84

= 7<2 > % * 79:::: 5

+ , 16: 5 * :6
& +( ! ? ?
,
> % +94< 9 , #
78:: 5 +(0@ # 7::;, % * :6
?
' ' +( ! 6:,
> % * 12:: 5 +
< 62 3 , * 62 5 77 2 5 # +(0@ # 7::;, %
* : 62

A ' +: 7 B , "
+5 (, ' 9:: !/ 0 9# C9 9

# # 0 #
9 * 7 # C9 9 6 0
#

# " # " #
97 !/
# 0 #
#
" A
97 !/# # '
" + & # 9444,

730
 !" #
$ ! % + 0&0 242, D
# ! + ,
'
7

%!" ((% 3 - # "+#- ( +$ . !" #$ % "%& + #+#

4 5 ' 6 ' σ
7
E!/F E F E F E5 F E5 F E5 F
19 <2 : 8; 9;97 8 26 48 7847 <: 62
64 :: : 82 9194 8 2; 88 6: 1< <1 97
17 21 : 42 6:<2 7 22 12 74 <9 <7 7;
64 2; : 88 747< 1 29 12 7< 2< ;< :1
14 72 9 92 182< ; 2866 69 72 <; :9
64 82 : 44 1:;; 1 2< :; 6: 2< <1 6;

7# <7 6; 5 D #
' # '
#$ ! % G # $ !

5 # "
# $ ! % #
# $ ! % '

7 8
* ( 5 ( 9::7"44 "
?
' +( ! 6:,
> % * <7
. 6 6

taluminium
ta

La

7 ) " #$ % "%& #& '( )#*) + + #'%",

%!" 7 ( + '%"& & + # -+$ #) " #$ % "%& #& '( )#

% / + %" ( ) ( + %.$ #0 9#& ' -%' + %+( )+

1((2 1((2 1((2 1((2
<: 7: 77 :6 7 /
;: 7: 98 :1 6 (
7 61 62 7: :1 6 (
: ;: 7: 98 :1 6 ( H(
; 61 62 7: :1 6 ( H(

731
 # '
+: 7 B , " +5 (, ' 9:: !/ /
!" #
& % # # ! +
,
' + 1,

%!" : ((% 3 -%0 % . # "+#- ( +$ #) " #$ % "%& + #+#

5 ' 6 ' σ
7
E!/F E F E5 F E5 F E5 F
+ 7 ;; 91:: 9; ;4 78;1 6; 67
+ 6 64 97:: 79 ;6 6< 1< 1< 62
+7 2 :9 97:: 79 2< 19 72 2: 1<
+: 2 69 994: 67 78 26 84 ;4 ::
+; < 29 994: 69 66 28;2 <7 71

7 1#
#
!
+ 6; 67 5 <9 17 5 ,
#

D # & %
# !

!" # $ ! % & %
# #
#
! #
#

:
&=)A0( +0 / ! ( = ) ,
! ! ' (! 0 &
!5

;
5 & "+9444, I
J# ( *# $ 24# 997<"996<
K# ( K# G ) # ! L +7::1, I( '
" 0&0 242# ) # A@
+7::;, I ' #
& J# - ..$ 6<# ;1M<6
(0@ +7::;, ( -BB ! B B " B " " ' 92B:6B:;
C +944<, I( J# /& # $ 4+1,# 7:;"797

732
Third International Conference on FRP Composites in Civil Engineering (CICE 2006)
December 13-15 2006, Miami, Florida, USA

EXPERIMENTAL STUDY ON FATIGUE BEHAVIOR OF TENSILE STEEL

PLATES STRENGTHENED WITH CFRP PLATES

Zheng Yun Lingping Ye Xinzheng Lu

(Department of Civil Engineering, Tsinghua University Beijing 100084)

Q.R.Yue
(Central Research Institute of Building and Construction, MCC Group Beijing 100088)

Abstract
Fatigue experiments of six steel plates strengthened with CFRP plates are presented in this paper. The specimens are
loaded in tension. The factors that may influence the effect of strengthening are discussed, including the amount of
strengthening CFRP plates, the one face or two faces mode strengthening and the stress range. Experimental results
show that the fatigue lives of steel plates can be greatly increased with externally bonded CFRP plates when they are
compared with un-strengthened specimens.

Keywords
CFRP, Steel plate, Tensile, Fatigue life, Strengthening

1. INTRODUCTION

Due to the light weight, high strength, good durability and excellent fatigue performance, Carbon Fibre Reinforced
Polymer (CFRP) is widely used in last decade to strength concrete structures. And in recent years, more attentions
have been given on strengthening steel structural elements with CFRP. Most of such steel strengthening researches
are focused on the enhancement of static load capacity. Very limited work can be found in the existing literatures on
the fatigue strengthening [Zheng et al. 2005]. Colombi et al. (2003) suggested that CFRP patching would be a good
solution for strengthening steel structural elements with fatigue damage, and prestressed CFRP patches would have a
better performance on interfacial debonding. However, in practices, prestressed CFRP strengthening is not widely
welcomed due to the inconvenience of construction. Tavakkolizadeh & Saadatmanesh (2003) studied the fatigue
strength of 21 specimens made of W127×4.5 steel beams. The clear spans of all specimens were 1220mm and were
all tested under four-point bending with 200mm spacing between the loading points. For all stress ranges considered
in their work, this strengthening technique improved the fatigue lives of the strengthened specimens by about
2.6~3.4 times higher than those of un-strengthened specimens. Sean et al. (2003) reported their fatigue experiments
on 21 specimens with edge notches and 8 specimens with center holes. Though their test results scattered greatly,

733
CFRP strengthening still shows a promising increase on the fatigue lives of steel elements. However, the above
existing works have a common drawback that the initial faults in the steel elements are not real fatigue damages, but
mechanical cutting notches or holes. Hence, in this work, pre-cracking cycle loads were imposed firstly on the steel
plates to give a same level of real fatigue damage. And then the specimens were tested under different levels of
tension load to get a more realistic simulation of fatigue strengthening.

2. EXPERIMENTAL TEST SETUP AND SPECIMENS

The steel plates used in this study is made of 16Mn whose yield strength is 435MPa. The size of the steel plates are
700mm in length, 100mm in width and 10mm in thickness, as shown in Figure 1a. The initial cracks are introduced
by cutting 2 notches with a length of 2mm on opposite sides of the 2mm hole in diameter which is located at the
centers of the plates, as shown in Figure 1b. Two kinds of CFRP plates with different thickness and stiffness are
adopted in the tests. One is 1.0mm thick with a normal modulus of 165GPa and another is 1.4mm thick with a higher
modulus of 320GPa. The adhesive used was a two-part thixotropic epoxy resin adhesive (Araldite 2015),with a shear
modulus of 0.9 GPa.
10mm thick

Steel plate

CFRP
Steel Plate

(a) Specimen geometry (b) Initial crack

Figure 1 Specimen Geometry

The six specimens tested in this work were shown in Table 1. Following 3 parameters are considered: stress range,
strengthening method and stiffness of CFRP. Specimens PC120 and PC90 are un-strengthened control ones with
different stress range of 120MPa and 90 MPa. Specimen PS120 and PD120 were both tested under a stress range of
120MPa but the former one was strengthened with high modulus CFRP on single side whilst the later one was
strengthened on both side. Specimen PD90 has the same strengthening scheme with PD120 but was tested under a
stress range of 90MPa. Specimen PD120a has the same strengthening scheme and stress range with PD120 but
normal modulus CFRP is used. Pre-crack cycle load with a stress range of 200MPa were give to all specimens to
guarantee the fatigue cracks would propagate from the initial notches and to induce a 10mm length of fatigue cracks
along the two notches on opposite sides of the hole. CFRP strengthening was installed after pre-cracking load. Then
the fatigue testing of all the specimens was conducted under the cyclic tensile loads. A constant amplitude sine wave
of 500 times/min frequency with stress ratio of 0.4 was applied.

734
 Table 1 Specimen Details
Stress range Nmax /Nmin Stiffness Ratio
Specimen numbers Repaired Mode CFRP Type
/MPa /kN (S)
Without
PC120 120 200/80 None 0
strengthening
Without
PC90 90 150/60 None 0
strengthening
PS120 120 200/80 Single Side High Modulus 0.22
PD120 120 200/80 Double sides High Modulus 0.43
PD90* 90 150/60 Double sides High Modulus 0.43

PD120a 120 200/80 Double sides Normal Modulus 0.16

Note: Stiffness ratio (S) is defined as: S =

∑ ECFRP t CFRP
.
E Steel t Steel
Table 2 Test results of fatigue lives of specimens
Stress Range Repairing Fatigue Lives
Specimen Stiffness Ratio (S) Increased life (%)
（MPa） Mode （x104）
PC120 / 0 18.2 Baseline
120 PS120 Single side 0.21 46.4 155
PD120 Double Sides 0.43 100.0 450
PD120a Double Sides 0.16 65.0 260
PC90 / 0 48.4 Baseline
90
PD90 Double Sides 0.43 331.2 580

3. EXPERIMENTAL RESULTS

The fatigue lives of all the specimens were listed in Table 2. It can be shown that fatigue lives of strengthened
specimens were increased by 155~580% over un- strengthened specimens.
Figure 2 shows comparisons on the propagation of fatigue cracks. Different strengthening modes are compared in
Figure 2a which illustrates that double bonded specimen (PD120) results in a much slower crack propagation speed
than un-strengthened one (PC120) and the single side boned one (PS120), even though the interfacial debonding of
PD120 was much worse than other two specimens due to poor quality installation of CFRP plates. Figure 2b shows a
better example on the increment on fatigue life because no interfacial failure happened in the specimens. The
strengthened specimen PD90 has an additional 3 million cycles longer fatigue life than the un-strengthened
specimen PC90. Figure 2c shows the influence of stiffness of CFRP plates to the fatigue life. Higher stiffness results
in a longer fatigue life. Figure 2d shows the influence of the stress range to the fatigue life. Larger stress range will
heavily reduce the fatigue life.

4. CONCLUSIONS
In this study, the fatigue behavior of six steel plates strengthened with external bonded CFRP plates was investigated.
It is shown that, (1) the fatigue life of strengthened specimens was increased by 155~580% over un-strengthened
specimens, which indicated that fatigue life can be improved effectively by external bonded CFRP to the steel
structures with fatigue crack; (2) The strengthening is more effective when CFRP plates with higher modulus are
bonded to both sides of the steel plates; (3) De-bonding of CFRP from steel plate took place so early during the test
of specimen PD120 that the retrofitted effect is not good. Therefore the quality control of bonding technique is very

735
important and some measures should be made.

50
50
45
45
PC120 PC90
40
PS120

Crack Length a（mm）

40 PD90
PD120 35 PS120
Crack Length a(mm)

35
30
30
25
25
20
20
15
15
10
10
5
5
0 50 100 150 200 250 300 350
0 10 20 30 40 50 60 70 80 90 100 110
4 Number of Cycles（x104）
Number of Cycles(x10 )
(a)Different strengthening mode (b) Different strengthening mode
50
50
45 PC120
45 PC120 PD120
PD120 40 PC90
40 PD90
Crack Length a(mm)
PD120a
Crack Length a(mm)

35
35
30
30
25
25

20 20

15 15

10 10

5 5
0 10 20 30 40 50 60 70 80 90 100 110 0 50 100 150 200 250 300 350
Number of Cycles（x104） Number of Cycles（x104）

(c)The influence of stiffness of CFRP plate (d)The influence of the stress range
Figure 2 Fatigue crack propagation curves

5. ACKNOWLEDGEMENT
The authors are very appreciated to the key project No. 50238030 granted by Natural Science Foundation of China
to support to the research presented here.

6. REFERENCES
Zheng Yun, Ye Lieping and Yue Qingrui(2005). “Progress in research on steel structures strengthened with FRP”,
Industrial Construction, Vol.35, No.8, pp 20~25.
P. Colombi, A. Bassetti and A.Nussbaumer(2003). “Analysis of cracked steel members reinforced by pre-stress
composite patch”, Fatigue Fract Engng Mater Struct , Vol.26, pp 59~66
M. Tavakkolizadeh and H. Saadatmanesh(2003). “ Fatigue Strength of Steel Girders Strengthened with CFRP
patch”. Journal of Structural Engineering, Vol.129, No.2, pp 186-196
Sean C. Jones and Scott A. Civjan, P.E (2003). “Application of Fiber Reinforced Polymer Overlays to Extend Steel
Fatigue Life”, Journal of Composites for Construction , Vol.7, No. 4, pp 331-338

736
Third International Conference on FRP Composites in Civil Engineering (CICE2006)
December 13-15 2006, Miami, Florida, USA

PRELIMINARY BOND-SLIP MODEL FOR CFRP SHEETS

BONDED TO STEEL PLATES
S.Fawzia, X.L. Zhao, R. Al-Mahaidi
Department of Civil Engineering, Monash University, Clayton, Victoria 3800, Australia

S.H. Rizkalla
North Carolina State University, Raleigh, North Carolina, USA

ABSTRACT
Carbon fiber reinforced polymer (CFRP) sheets have established a strong position as an effective
method for innovative structural rehabilitation. However, the use of externally bonded CFRP in the
repair and rehabilitation of steel structures is a relatively new technique that has the potential to
improve the way structures are repaired. An important step toward understanding bond behaviour is to
have an estimation of local bond stress versus slip relationship. The current study aims to establish the
bond-slip model for CFRP sheets bonded to steel plate. To obtain the shear stress versus slippage
relationship, a series of double strap tension type bond tests were conducted. This paper reports on the
findings of the experimental studies. The strain and stress distributions measured in the specimens for
two different bond lengths. The results show a preliminary bi-linear bond-slip model may be adopted
for CFRP sheet bonded with steel plate.

KEYWORDS
FRP Sheet, Bonding, Bond-Slip Model, Slip, Shear Stress.

1. INTRODUCTION

Successful use of carbon fiber-reinforced polymer (CFRP) materials for strengthening concrete
structures has already been established [e.g. ACI Committee 440F 2002, Teng et al 2002, Oehlers
and Seracino 2004, Pham and Al-Mahaidi 2006]. However, the development of the system for
strengthening steel structures with CFRP materials is limited. More strengthening materials are needed
to achieve a significant strength increase as steel is much stronger than concrete, especially in tension.
But as more strengthening material is added the bond stresses become more critical. There have been
relatively fewer studies on the bond stress and slip relationship between CFRP sheet and steel
structures.

The bond-slip relationship relating the interfacial shear stress to the interfacial slip for steel structure
strengthened by CFRP plate was recently studied by Xia & Teng (2005). In this paper a local bond-slip
relationship is proposed from the experimental results for CFRP sheets bonded to steel structure. The
possibility of finding local bond-slip relationship using long bond length (say twice the effective bond
length or higher) is strictly related to the consideration of the distribution of slip and bond shear stress
along the bond length.

2. EXPERIMENTAL PROGRAM

2.1 Bond specimen and surface preparation

737
The tensile specimen was composed of mild steel plate bonded with three layers of CFRP sheet on both
sides. Figure 1 shows the schematic specimen configuration. A careful, meticulous approach is
necessary when dealing with bonding since it may be difficult to verify the quality of the bond.
Moreover, due to the local effect of bond stresses, any local defect of the bond may result in complete
debonding of the applied strengthening material. The steel plate was 6mm thick and 900 mm long in
total. The steel plate was grinded at 450 by angle grinder (FH38A36S-BF41). After grinding, the steel
plate was cut into two pieces and then added together by applying small amount of adhesive at the
cross section. After that the plate was cleaned by Acetone. Adhesive was first applied on the surface of
steel plate by brush. Then a piece of fiber sheet, which had been cut beforehand into prescribed sizes
using scissors, was placed with the fiber side down onto the coating and generally smoothed down by
hand. After that, the surface of the sheet was rolled over along the longitudinal direction of the fibers
using a ribbed roller to impregnate resin into the fibers and removed any air bubbles. Rolling was
continued until the resin was squeezed out between the fibers. Same steps were followed to bond two
more layers of fibers. The specimen was cured at least one week and then postcured for 24 hours at
700C.

2.2 Material properties

Measured modulus of elasticity, tensile stress and thickness of CFRP sheet are 250 GPa, 1710MPa and
0.176mm, respectively. The modulus of elasticity and tensile stress of adhesive are 1900MPa and 32
MPa, respectively.

2.3 Instrumentation

Sixteen strain gauges were attached to each test specimen. Figure 1 shows the location of each gauge.
Ten strain gauges were placed along the length at every 25mm on one side of the CFRP sheet to
capture the longitudinal strain development along the bond length. Four strain gauges were placed on
the backside of the specimen at every 50mm to act as “backup” gauges and for comparison with the
primary gauge readings. Two additional strain gauges were attached to the steel plate before the
initiation of the CFRP sheets and were used to determine the actual load applied to the steel plate
during testing.

3 layers of CFRP sheet εi+1 εi εi-1 ε1 ε0

Steel joint Xi+1 Xi Xi-1 X1 0

6mm Free end
X

Figure 1: A Schematic of specimen and strain gauge locations

2.4 Failure mode and test results

Test results, together with the failure modes are presented in Table 1.

Table 1: Test results

Test S1 S2 S3 S4 S5 S6
specimen
Bond length (mm) 250 250 250 200 200 200
Overall thickness of specimen (mm) 9.50 9.75 9.82 9.28 9.38 9.46
Ultimate load (kN) 89.6 97.2 77.5 92.6 71.6 99.6
Failure mode Bond Bond Bond Bond Bond Bond

738
3. Strain and interfacial stress distribution

The data obtained from the strain gauges at the top layer of CFRP was used to create strain versus
distance (from the steel joint) plots. The top strain is different from the average composite strain as the
strain could vary across the layers of the composite. This variation was measured experimentally for
circular tube strengthened by CFRP sheet [Fawzia el al.2004]. In this study it was assumed that the
measured strain was representing the average CFRP strain. The distributions of strain along the bond
length for different load levels are plotted in Figure 2 for Specimen S6. The distance in the figure is
measured from the joint location, At low load levels, the distributions show a gradual decline from the
peak near the steel joint to the other end. As the load increases up to 98.3kN, the strain values increase
and the peak strain gradually shifts away from the joint. At low levels, the distributions have the largest
slope near the steel joint. As the load increases, the maximum slope shifts away from the joint. This
means that redistribution of the bond stress along the bond length occurs as a result of changes in the
state of bond.

9000 25
15.6kN 15.6kN
8000 28.0kN 28.0kN
7000 40.0kN 20 40.0kN
59.0kN 59.0kN
Shear stress(MPa)

6000 73.3kN 73.3kN

M ic ro s tra in

88.1kN 15 88.1kN
5000
98.3kN 98.3kN
4000 98.9kN 98.99kN
10
3000

2000
5
1000

0 0
0 20 40 60 80 100 0 20 40 60 80
Distance(mm) Distance(mm)

Figure 2: CFRP strain distribution Figure 3 : Shear (bond) stress distribution

The average experimental shear stress between the two strain gauges were calculated using the
E f (ε f , i +1 − ε f , i )t f
relationship τ = , where Ef and tf are the CFRP elastic modulus and thickness
X i +1 − X i
respectively and ε f ,i +1 , ε f ,i are the CFRP strains and X i +1 , X i are the distance between strain
gauges according to Figure 1. The shear stress distribution along the distance away from the “steel
joint” is shown in Figure 3. It should be pointed out that there is a gap between the two steel plates. The
location of the “steel joint” is equivalent to the “loaded edge” in the set up of testing bond between
CFRP plate and steel (Xia and Teng 2005). The theoretical stress distribution for bond between CFRP
and concrete can be found in H.Yuan et al. (2004). The theory shows that at loaded edge the shear
stress is zero when it reaches peak load indicating occurrence of debonding. It can be seen from Figure
3 that when the load is less than 59kN, the peak shear stress is located at the steel joint. The location of
the peak shear stress moves away from the “steel joint” as the load increases further. The shear stress at
the steel joint becomes zero when the maximum load occurs indicating the occurrence of debonding.

4. BOND-SLIP MODEL

The measured strain distribution along the bond length was used by integration to calculate local slips.
Actually this local slip is the relative displacement between the CFRP sheet and the steel plate.
Calculated bond stresses and slips are combined to obtain the local bond-slip curves. Bond-slip curves
obtained from experimental data can be approximated as a bi-linear shape (Lu et al., 2005). These
curves have a linear ascending branch followed by a linear descending branch. A schematic view is
presented in Figure 4 which can be defined by three parameters δ1, τf and δf . The initial stiffness of the
bond-slip curve is high, representing linear elastic state. Initiation of interfacial softening stage means
load continues to increase as the length of the softening zone increases. The ultimate load is first
attained at the end of this stage and starts propagation of debonding. These three stages can be
identified from load-displacement behaviour. The local bond slip relationship is reasonably consistent
between different locations on the same specimen.

739
 25

(δ1,τf) 4mm
12mm
Elastic Softening Debonding 20 21mm
Shear stress(MPa) 29mm
37mm

S h e a r s tre s s (M P a )
46mm
15
79mm
87mm

10

(δf,0) 5

0
0.00 0.05 0.10 0.15 0.20 0.25 0.30 0.35 0.40 0.45 0.50
Slip(mm) Slip(mm)

Figure 4 Bond slip model approximation Figure 5 Bond slip curve (Specimen S6)

Figure 5 shows the stress vs slip curves (specimen S6) with bond length of 200mm. It seems that a bi-
linear model may be adopted to represent the bond-slip relationship for CFRP sheet bonded with steel
plate. However, the three parameters (δ1, τf and δf) in bond-slip model are to be determined after
processing all the test data. These will be compared with those given by Xia & Teng (2005) in the near
future.

5. CONCLUSIONS

The following conclusions can be drawn from this paper.

1. Strain distribution profiles show that strain level is significant over a limited bond length.
2. When debonding occurs at most highly stressed end, less or almost zero stress is transferred at
that end and the maximum shear stress location shifts towards the unloaded end of the
specimen.
3. The bond-slip curve may be approximated as a preliminary bilinear model. More reliable peak
shear stress and slip values will be produced after analysing more test data.
4. The limitations of such bilinear model mentioned elsewhere in the paper are
Depend on the values of δ1, τf and δf, One type of CFRP, One type of adhesive, Bond
failure happened around 53% of FRP rupture strain, Only two types of bond length has
considered.

6. REFERENCES
1. ACI Committee 440 (2002), Guide For The Design And Construction Of Externally Bonded
FRP Systems For Strengthening Concrete Structures. American Concrete Institute, Detroit MI.
2. Fawzia, S., Zhao, X. L., Al-Mahaidi, R. and Rizkalla, S. (2004).Investigation Into The Bond
Between CFRP And Steel Tubes, The Second International Conference on FRP Composites in
Civil Engineering,CICE2004, December, Adelaide,. pp733-739.
3. H.Yuan, Teng, J. G., Seracino, R., Z.S.Wu and J.Yao. (2004).Full-Range Behaviour Of FRP -
To-Concrete Bonded Joints. Engineering Structures.Vol.26: pp553-565.
4. Lu,X.Z., Teng, J.G. Ye,L.P. and Jiang, J.J.(2005) Bond-Slip Models For FRP Sheets/Plates
Bonded To Concrete. Engineering Structures.Vol.27:pp 920-937.
5. Oehlers, D.J. and Seracino, R. (2004), Design Of FRP And Steel Plated RC Structures –
Retrofitting Beams And Slabs For Strength, Stiffness And Ductility, Elsevier, Oxford, UK
6. Pham, H. B. and Al-Mahaidi, R. (2006), “Prediction Models For Debonding Failure Loads Of
CFRP Retrofitted RC Beams”, Journal of Composites for Construction, Vol. 10, No. 1, pp. 48-
59
7. Teng, J. G., Chen, J. F., Smith, S. T. and Lam, L. (2002).FRP-Strengthened RC Structures,
John Wiley and Sons Ltd,UK.
8. Xia, S. H. and Teng, J. G. (2005).Behaviour of FRP-to-Steel Bonded Joints. Proceedings of
the International Symposium on Bond Behaviour of FRP in Structures (BBFS 2005).
International Institute for FRP in Construction, Hong Kong: pp 419-426.

740
Third International Conference on FRP Composites in Civil Engineering (CICE 2006)
December 13-15 2006, Miami, Florida, USA

FLEXURAL STRENGTHENING OF COMPOSITE STEEL-CONCRETE

GIRDERS USING ADVANCED COMPOSITE MATERIALS
Raafat El-Hacha
(Assistant Professor, University of Calgary, Calgary, Alberta, Canada)

Nora Ragab
(Master Student, University of Calgary, Calgary, Alberta, Canada)

ABSTRACT
Composite steel-concrete girders are used widely in bridge and building construction as the main structural elements
in flexure. The load-carrying capacity of an under-strength or deficient steel-concrete composite girders can be
improved by epoxy bonding fiber reinforced polymers (FRP) laminates to its tension flange. This paper presents the
results of an experimental study that investigated the behaviour of steel-concrete composite girders strengthened in
flexure using various advanced composite materials tested under static loading. The different strengthening
materials used in this investigation included unidirectional intermediate and high modulus Carbon Fiber Reinforced
Polymer (CFRP) plates, unidirectional CFRP sheets and the newly developed unidirectional Steel Reinforced
Polymer (SRP) sheets. The primary objective of this investigation was to assess the flexural behaviour of the
strengthened beams and examine the effectiveness of the different strengthening materials. Test results are very
promising and showed that epoxy bonded CFRP sheets or plates and SRP sheets significantly improved the stiffness
and increased the ultimate load carrying capacity of the steel-concrete composite girders. The effect of strengthening
on the elastic stiffness and ultimate strength was more profound for beam with high modulus CFRP plate.

KEYWORDS
Carbon fiber reinforced polymer, composite, concrete, externally bonded, flexure, high-modulus, sheets, plates, steel

1. INTRODUCTION
Many researchers conducted experimental and analytical studies on concrete beams and slabs strengthened in
flexure using externally bonded FRP sheets or strips, however; research on steel-concrete composite girders is
limited and only few studies are reported in the literature. Tests conducted by Mertz and Gillespie (1993) on small-
scale steel beams strengthened in flexure with CFRP plates showed about 20% increase in the flexural stiffness and
more than 50% increase in the ultimate strength. Sen et al., (1994 and 2001) loaded large-scale steel-concrete
composite girders past yielding before being repaired using unidirectional CFRP plates and the results showed an
increase in the ultimate strength from 11 to 52% depending on the mode of failure, thickness of the CFRP plates, the
anchorage at the ends of the plates, and the yield strength of the steel beams. The effect of the number of layers of
CFRP sheets on the behaviour of large-scale composite girders has been investigated by Tavakkolizadeh and
Saadatmanesh (2003a) and the test results showed an increase in the ultimate strength by 44, 51 and 76% using 1, 3
and 5 layers, respectively. Tests showed that the CFRP sheets not only extended the fatigue life of a steel beam by
more than three times, but also decreased the crack growth rate significantly (Tavakkolizadeh and Saadatmanesh,
2003b). To simulate field corrosion, steel composite beams were damaged by removing 50 and 75% of their bottom
flange then repaired using CFRP plates to restore their original strength. Test results showed that 50% of the elastic
flexural stiffness can be restored and the ultimate strength was fully restored (Al Saidy et al., 2004). Recently, high
modulus and ultra-high modulus CFRP strips have been used to strengthen large-scale steel composite girders, and
test results showed an increase in the ultimate flexural capacity by 16 and 45%, respectively; however, the flexural
stiffness increased only by 10% and 36% due to the high elasticity modulus of the strips (Schnerch et al., 2005).
This paper investigates the feasibility and effectiveness of using various materials including high-modulus FRP to
strengthen large-scale steel-concrete composite girders, simulating the majority of the highway bridges, tested under
static loading. The structural performance of the strengthened steel beams will be examined and discussed.

741
2. EXPERIMENTAL INVESTIGATION
2.1 Specimens Details
A total of five 6.2m long steel-concrete composite girders made of steel I-beams and 56 mm thick by 435 mm wide
reinforced concrete slabs were fabricated and tested. Shear stud connectors welded to the top flange of the steel
beams were used to provide the shear connection between the concrete slab and the steel beams. Two longitudinal
rows of shear studs were used with a 90 mm transverse spacing between rows of studs. The studs were spaced at 50
mm apart in the longitudinal direction. Figure 1 shows details of a typical beam specimen.

2.2 Material Properties

Concrete
The specified 28-d compressive strength of the concrete was 40 MPa. The actual concrete strength at the day of
testing determined as the average of three standard concrete cylinders was 40.3, 39.1, 37.0, 39.0, and 36.5 MPa for
beams B1, B2, B3, B4, and B5, respectively.

Steel
The steel beams were standard W200×19.3 Grade G40.21-M345W hot rolled I-sections with average yield and
ultimate tensile strengths obtained from tension tests of 414 MPa and 532 MPa, respectively. The interface shear
reinforcement along the top flange of the steel beam and the reinforced concrete slab was provided by shear
connector. The shear connectors were 6.0 mm diameter and 28.5 mm length with yield and ultimate tensile strengths
of 345 MPa and 414 MPa, respectively. The composite concrete deck slab reinforcement consisted of a welded wire
fabric steel mesh type 152.4 × 152.4 MW28.3 × MW28.3 with a nominal diameter of 6.0mm.

Strengthening Materials
The externally bonded strengthening systems selected for this study were carbon FRP sheets, carbon FRP plates, and
steel-reinforced polymer (SRP) sheets. The material properties of the different systems are given in Table 1. A two-
part component epoxy adhesive, the main epoxy resin and the curing agent hardener, was used. Sikadur 330 was
used for bonding the SRP and CFRP sheets, and Sikadur 30 was used for bonding the CFRP plates to the bottom
flange of the steel girders.

Table 1 − FRP material properties as reported by the manufacturers

FRP products Elastic Modulus Ultimate Tensile Strength
Dimensions
(type) (MPa) (MPa)
Unidirectional SRP Sheet
0.44 mm2/mm† 206000 3170
(Hardwire™ 3×2-23-12)
Pultruded CFRP Plate
1.2 mm‡ 160000 2800
(Sika Carbodur type S 812)
Unidirectional CFRP Sheet
0.381 mm‡ 61012 715
(Sika Wrap® Hex230C))
Pultruded High Modulus CFRP Plate
1.4mm‡ 300000 1300
(Sika Carbodur® type H 514)
† ‡
Net area per width Thickness
435mm

56mm
Loading frame girder
Loading frame column
500kN hydraulic ram,
semi-spherical seat and W200×19.3
1000kN load cell 0.2m

Girder
Support
LSC
Pedestal

2.40m 1.2m 2.40m

Figure 1: Test set-up and instrumentation of the steel-concrete composite girders

742
2.3 Test Matrix
One beam (B1) was tested without strengthening and served as unstrengthened control specimen. Four beams (B2,
B3, B4, and B5) were strengthened with various systems designed to achieve a 30% increase in the ultimate load
carrying capacity over the control beam. Table 2 summarizes the test matrix.
Table 2 − Test matrix for the composite girder specimens
Beam # Strengthening System
B1 Control beam without strengthening
B2 Beam strengthened using two layers of 82mm wide SRP sheets
B3 Beam strengthened using one layer of 80 mm wide CFRP plate
B4 Beam strengthened using five layers of 98 mm wide CFRP sheets
B5 Beam strengthened using one layer of 25 mm wide HM-CFRP plate

2.4 Surface Preparation of Steel Beam and Installation of the Strengthening Systems
To ensure good and strong bond, the bottom surface of each steel beams was prepared by sand blasting and cleaned
by air brushing to remove any dust. Before applying the strengthening materials an organic solvent (acetone) was
used to clean the surface from any other bond inhibiting materials. Installation of the strengthening systems followed
typical field conditions on the bottom flange beneath the steel I-beams using the dry lay-up technique. The epoxy
was allowed to fully cure at room temperature for at least one week before testing the beams. The anchorage system
consisted of a steel plate bolted to the bottom flange of the beam at both ends of the strengthening materials.

2.5 Test Setup, Procedure and Instrumentation

The 6.2m long girders were simply supported with a span of 6.0m between supports and tested under four-points
bending static loading with 1.2m spacing between the two concentrated point loads. The load was applied using a
500kN capacity actuator through an MTS controller-testing machine operating under displacement control mode.
The control beam (B1) was tested under a monotonically increasing load up to failure at a constant loading rate. To
study the behavior of the strengthened beams after being loaded to the service condition and past the yield load, the
Diagnostic Cyclic Load testing protocol was adopted in beams B2, B3, B4 and B5 (Nanni and Mettemeyer, 2001).
All beams were fully instrumented to monitor their behaviour by measuring the deflection at midspan using Linear
Strain Conversion devices (LSCs), strains in the concrete in the compression zone, and strain in the steel and in the
CFRP and SRP reinforcements using electrical resistance strain gauges. During testing, the data were automatically
collected and electronically recorded using a data acquisition system connected to a personal computer. Typical test
set-up and instrumentation is shown in Figure 1.

2.6 Test Results and Discussion

The load versus midspan deflection curves comparing the flexural behaviour of the five beams are presented in
Figure 2. For beams B2, B3 and B4, the strengthening effect is mainly affecting the post-yielding response since the
stresses in the composite materials are less than those in the bottom steel flange due to their lower modulus of
elasticity than steel. For the same beams, the improvement in the stiffness in the elastic range prior to yielding of the
bottom steel flange is not significant due to the smaller elastic modulus of the strengthening materials compared
with the steel section, however; the strengthening effect was more evident for beam B5 strengthened with a high
modulus CFRP plate as shown by the significant increase in stiffness in the elastic range and ultimate strength when
compared to beams B2, B3 and B4. The effectiveness of the FRP materials was more pronounced after yielding. As
expected from the design, the beams achieved the required percentage increase in strength (30%) as can bee seen for
beams B2 and B3 the increase in strength was 29%, and 35%, respectively, except for beam B4 the increase in
strength was 23% because the applied area of CFRP sheets of 187mm2 (or 5 layers of 98mm) was less than the
required 272mm2 (8 layers of 89.25mm) in order to achieve 30% increase in strength. The use of higher modulus
CFRP plate resulted in the highest stiffness increase. Beam (B5) strengthened with HM-CFRP plate demonstrated
strength increases up to 49%. All strengthened beams had minimal losses in ductility compared to the control beam.
The load-deflection behaviour is bilinear until failure. No degradation in the stiffness was noticed due to the
unloading/loading cycles which became slightly nonlinear after yielding. As well, after yielding, the strengthened
beams continued to resist further increase in the applied load with a more gradual linear slope than the pre-yield
portion of the curve. The increase in the load continued until failure. Failure of beams B2, B3 and B4 occurred by
crushing of the concrete near one of the load points, while beam B5 strengthened with HM CFRP plate failed by
rupture of the plate that occurred within the constant moment region. Ductility is slightly reduced with the addition
of composite materials. No debonding or delamination between the FRP and the steel beam was observed in any of
the beams, indicating that the surface preparation was adequate and the bonding was strong enough.

743
3. CONCLUSIONS
On the basis of the experimental study presented, the following conclusions can be made:
– No bond failure was observed between the composite materials and the steel surface in all specimens.
– The effect of FRP bonding on the elastic stiffness was not significant for beams strengthened with intermediate
modulus CFRP sheets or plates and SRP sheets.
– Test results showed significant increases in the stiffness prior to yield as well in the ultimate strength for the
beam with HM-CFRP plate indicating the possibility to increase service loadings. This is very important when
strengthening bridge to meet the more stringent limits imposed on serviceability (live load deflections) and
ultimate strength of current codes.
– Crushing of the concrete was the dominating mode of failure except for the beam with HM-CFRP plate it was
by rupture of the plate in the constant moment region.
– In general, all strengthened beams failed in a ductile manner accompanied by large deformation; however the
beam strengthened with HM-CFRP plates showed less ductile behaviour but higher capacity.
To summarize, this study has confirmed the structural benefits and feasibility of using externally bonded
intermediate modulus CFRP plates and sheets and SRP sheets to strengthen steel-concrete composite girder.
However, the results indicated that high-modulus CFRP plates are more effective at increasing the strength and
stiffness and are well suited for use in the repair and strengthening of steel structures.
B5
140 B3

B2
120
B4
B1
100 Failure B4
Load (kN)

80

60

40

20

0
0 20 40 60 80 100 120 140 160 180 200
Midspan Deflection (mm)
Figure 2 Load-midspan deflection curves for all steel-concrete composite girders

4. REFERENCES
Mertz, D., Gillespie, J., (1993), “Rehabilitation of Steel Bridge Girders through the Application of Advanced
Composite Material”, Transportation Research Board, Washington, D.C., NCHRP 93-ID11:1-20.
Sen, R., Liby, L., Mullins, G., (2001), “Strengthening Steel Bridge Sections Using CFRP Laminates”, Composites:
Part B, 32:309-322.
Tavakkolizadeh, M., and Saadatmanesh, H., (2003a), “Strengthening of Steel-Concrete Composite Girders Using
Carbon Fiber Reinforced Polymers Sheets”, J. Struct. Engrg., Volume 129, Issue 1, pp. 30-40 (January 2003).
Tavakkolizadeh, M., and Saadatmanesh, H., (2003b), “Fatigue Strength of Steel Girders Strengthened with Carbon
Fiber Reinforced Polymer Patch”, J. Struct. Engrg., Volume 129, Issue 2, pp. 186-196 (February 2003).
Al-Saidy, A.H., Klaiber, F.W. and Wipf, T.J. (2004), Repair of Steel Composite Beams with Carbon Fibre-
Reinforced Polymer Plates, Journal of Composites for Construction, ASCE, 8(2), pp.163-172
Schnerch, D., Dawood, M., and Rizkalla, S., “Strengthening Steel-Concrete Composite Bridges with High Modulus
Carbon Fiber Reinforced Polymer (CFRP) Laminates”, Proceedings of the Third International Conference on
Composites in Construction (CCC 2005), Lyon, France, July 11-13, 2005, pp. 283-290.
Nanni, A., and Mettemeyer, M., (2001), “Diagnostic Load Testing of a Two-Way Post-Tensioned Concrete Slab”,
Practice Periodical of Structural Design and Construction, ASCE, (6) (2001), 73-82.

744
 AUTHOR INDEX

A Bottoni, M. 215 Elsayed, W. 527

Abdul-Fattah, H. 523 Bouadi, A. 673 Enochsson, O. 387
Abro, A.M. 223 Bousselham, A. 577 Erki, M.A. 131
Ahmed Said, H.O. 75 Bradberry, T.E. 307 Erp, G.V. 457
Ahmed, E.A. 87 Brosens, K. 649 Esmaeili, J. 47
Alagusundaramoorthy, P. 363, Bruno, D. 383 Eyre, J.M. 623
399 Burgueño, R. 107
Al-Gadhib, A.H. 539 Burtscher, S.L. 483 F
Alhassan, M.A. 429 Faella, C. 219
Alliche, A. 511 C Fam, A. 163, 721
Al-Mahaidi, R. 71, 235, Cai, C.S. 251 Fang, Z. 471
425, 737 Camli, U.S. 689 Fangueiro, R. 323
Alrousan, R.Z. 429 Camotim, D. 421, 725 Fatemi, H 47
Al-Saidy, A. 597 Campione, G. 197 Fava, G. 275, 729
Al-Shawaf, A. 71 Cao, S. 193 Fawzia, S. 737
André, A. 627 Carolin, A. 243, 627 Feng, P. 139, 343, 709, 713
Anggawidjaja, D. 589 Carrazedo, R. 495 Feo, L. 227, 441
Ansley, M. 155 Casadei, P. 375 Ferrier, E. 487
Aram, M.R. 475 Castori, G. 375 Fico, R. 585
Araújo, M.D. 323 Černý, M. 259 Figeys, W. 649
Arduini, M. 685 Ceroni, F. 17 Fitzwilliam, J. 499
Argudo, J. 543 Chaallal, O. 577 Foster, S.J. 183, 437
Arockiasamy, M. 29 Chen, J. 67, 391 Frassine, R. 227
Ascione, F. 211, 441 Cheng, L. 697
Ascione, L. 227 Choo, C.C. 239 G
Asprone, D. 123 Choppali, U. 143 Galati, D. 95
Atadero, R.A. 255 Colombi, P. 275, 729 Galati, N. 103, 231, 491
Ayoub, A. 55, 605 Corb, I. 371 Gaudagnini, M. 159, 311
Corbi, O. 367 Gemert, D.V. 649
B Cosenza, E. 405 Ghobarah, A. 581
Badawi, M. 639 Crawford, J.V 203 Giancaspro, J. 681
Bae, S. 605 Crews, K.I. 677 Godat, A. 527
Bakht, B. 299 Cronin, K. 339 Gosain, N. 673
Bakis, C.E. 307, 535 Czaderski, C. 475 Greco, F. 383
Baky, H.A. 527 Green, E. 673
Balaguru, P.N. 681 D Griffith, M.C. 515
Baluch, M.H. 539 Dai, J. 39, 589 Gross, S.P. 203, 307
Bambach, M.R. 701 Darchis, F. 171 Grzebieta, R. 701
Bank, L.C. 151, 409 Davaran, A. 47 Gu, D.S. 547
Banthia, N. 653 Dawood, M. 705
Baratta, A. 367, 371 Debaiky, A.S. 147 H
Basegmez, I.H. 601 Deifalla, A. 581 Haedir, J. 701
Bayrak, O. 543, 563, 665 Demir, C. 601 Hamaguchi, Y. 669
Bedirhanoglu, I. 601 Devitofranceschi, A. 123 Hamed, E. 519
Belarbi, A. 55, 605 Diab, H.M. 25 Hamelin, P. 487
Benedetti, A. 227 Dubini, P. 729 Hamilton III, H.R. 315
Benmokrane, B. 87, 147, 287, Hamoush, S. 523
307, 319 E Hanai, J.B.D. 495
Bergström, M. 335 Ebaugh, S. 375 Harichandran, R.S. 107
Binici, B. 689 Ebead, U.A. 527 Harik, I.E. 239
Bisby, L.A. 499 El-Gamal, S. 319 Harries, K.A. 59, 91, 283,
Blanksvärd, T. 609 El-Hacha, R. 635, 741 339, 379
Borazghi, H. 147 El-Ragaby, A. 319 Harte, A.M. 271
Borri, A. 227, 375 El-Salakawy, E.F. 87, 287, 319 Hejll, A. 331
Hii, A.K.Y. 235 Lignola, G.P. 405 Orton, S.L. 665
Hiroshi, N. 589 Liu, H. 425 Ospina, C.E. 307, 535
Hollaway, L.C. 1, 643 Liu, M. 693 Otoom, O.F.A. 437
Honickman, H. 163 Liu, R. 413 Ouyang, Z. 263, 395, 433
Hordijk, D. 35 Lonetti, P. 383 Ozkul, O. 461
Hu, Y.M. 503 Lorenzis, L.D. 95, 613
Huang, Y. 347 Lu, X. 391, 733 P
Huang, Z. 567 Lv, Z.T. 547 Pantelides, C.P. 717
Papia, M 197
I M Park, C. 593
Ibell, T.J. 623 Ma, Z.J. 143, 359 Park, S. 593
Ilki, A. 601 Maher, A. 461 Parretti, R. 123, 585
Imjai, T. 159, 311 Mahini, S.S. 571 Parvin, A. 179
In, C.W. 339 Malik, A.R. 183 Paultre, P. 559
Ishikawa, T. 669 Mallat, A. 511 Pecce, M. 17
Issa, M.A. 429 Mancusi, G. 211 Peng, H. 479
Iwashita, K. 669 Manfredi, G. 227, 405, 585 Pereira, C.G. 323
Manko, Z. 111, 119, 657 Phong, N.H. 551
J Maragakis, E.M. 79 Piazza,M. 227
Jalali, S. 323 Maricherla, D. 449 Pilakoutas, K. 159, 311
Jamwal, A.S. 179 Maruyama, K. 551 Plunkett, J 135
Ji, H.S. 359 Masmoudi, R. 445 Poggi, C. 227, 275, 729
Jiang, T. 187 Matsumoto, A. 551 Porter, M. 379
Jin, Y. 479 Matta, F. 151 Prakash, S.S. 363, 399
Jing, D. 193 Mazzotti, C. 215 Prentice, D.B. 631
Jirsa, J.O. 543, 563, 665 McNeal, M. 523 Prota, A. 123, 405, 585
JNewhook, J.P. 307 Meggers, D. 135 Proulx, J. 559
John, M. 355 Michael, A.P. 315 Purdue, J.D. 115
Johnson, R. 79 Mineo, S. 589
Johnsson, H. 627 Minnaugh, P.L. 283 Q
Ju, M. 295, 593 Miraglia, N. 197 Qian, J. 693
Mirmiran, A. 127, 247, 555 Qian, P. 709, 71
K Mohamed, H. 445 Qu, Z. 391
Kalayci, S. 247 Mohiuddin, J. 539 Quiertant, M. 327, 487
Karbhari, V.M. 43, 255 Monti, G. 227
Katz, A. 279, 303 Mordak, A. 111, 119, 657 R
Kaul, R. 207 Motavalli, M. Rabinovitch, O 519
Keller, T. 291, 351 Mufti, A. 299, 653 Raftery,G. M. 271
Key, P.W. 457 Ragab, N. 741
Khalid, A.A. 167 N Ragaby, A.E. 287
Kim, I. 563 Nadauld, J.D. 717 Rahman, M.K. 539
Kim, S.D. 339 Nanni, A. 103, 123, 151, 227, Ravindrarajah, R.S. 207
Klamer, E.L. 35 231, 491, 585, 623, 685 Realfonzo, R. 219
Kleinman, C.S. 35 Nardo,A.D. 219 Reeve, B. 91
Klowak, C. 299 Nassif, H. 461 Ringelstetter, T.E. 151
Korynia, R 619 Navada, R. 43 Rizkalla, S. 83, 705
Kumbasar, N. 601 Neale, K.W. 527 Rizkalla, S.H. 737
Kurita, M. 175 Newman, N. 55 Rizzo, A. 103, 613
Nishizaki, I. 99, 453 Robert, M. 147
L Nossoni, G. 107 Rocca, S. 231, 491
Labossiere, P. 559 Rodd, P.D. 271
Lees, J. 467 O Romagnolo, M. 685
Leng, Q. 479 Oehlers, D.J. 531 Ronagh, H.R. 571
Li, B. 555 Oh, H. 295 Rosario, J.A.B.D.L.D. 507
Li, G. 355, 449 Omar, T. 457 Rosenboom, O. 83
Li, H. 471 Ombres, L. 51 Rotter, J.M. 391
Li, T. 343, 417 Orosz, K. 609 Roy, N. 559
Roy, R.J. 147 Sun, N. 193 Wight, R.G. 131, 631
Rusinowksi, P. 387 Willis, C.R. 515
T Wu, G. 547
S Taillade, F. 327 Wu, Y. 347
Sacco, E. 227 Takeda, N. 99 Wu, Z. 25, 75, 547, 669
Saiidi, M.S. 79 Täljsten, B. 331, 335, 387, 609
Sarmad, A. 461 Tamon, U. 589 X
Savoia, M 215 Tanaka, H. 175 Xia, S.H. 515
Saxena, P. 63 Tanovic, R. 131 Xiao, Y. 567
Schaumann, E. 351 Tazawa, H. 175 Xie, A. 131
Schueremans, L. 649 Teng, J.G. 67, 187, 643
Scott, P. 467 Tennyson, R.C. 653 Y
Sekijima, K. 551 Thiagarajan, G. 307 Yalim,B. 247
Seracino, R. 515 Tian, Y. 543 Yang, Q. 515
Shaat, A. 721 Tomiyama, T. 453 Yao, J. 67
Shahawy, M. 29 Tommaso, A.D. 227 Ye, L. 139, 343, 391, 709,
Shan, B. 567 Tourneur, C. 327 713, 733
Shang, S. 479 Toutanji, H. 63 Yishizuka, Y. 99
Shao, Y. 171 Yokota, H. 39
Shawkat, W. 163 U Yost, J.R. 203
Shi, Y. 555 Uddin, N. 115, 223 Young, B. 725
Shield, C.K. 409 Ueda, T. 11, 39 Yuan, H. 413
Shimomura, T. 99, 175, 551 Yue, Q. 733
Shrestha, R. 661 V
Shyu, C.T. 131 Vaidya, U. 115, 223 Z
Silva, N.F. 421 Vallée, T. 351 Zhang, L. 643
Silvestre, N. 421, 725 Vanevenhoven, L.M. 409 Zhang, S.S. 643
Sim, J. 295, 593 Vyas, J.S. 155 Zhang, Y. 251, 291
Sivakumar, M. 29 Zhao, L. 63, 155
Smith, S.T. 207, 437, 661, 677 W Zhao, X. 71, 425, 701, 737
Sohn, H. 339 Waldron, P. 159, 311 Zheng, R. 127
Son, B.J. 359 Wan, B. 263, 395, 433 Zheng, Y. 733
Soudki, K. 639 Wang, J. 21 Zhou, E. 135
Stao, Y. 39 Wang, M. 479 Zhuo, J. 417
St-Georges, E. 559 Wang, Y. 135 Zorn, A. 91
Suksawang, N. 461 Weber, A. 267
Sumiyoshi, K. 39 Widianto, 543
 KEYWORD INDEX

A 425 Compressive stress-strain model

Accelerated bridge construction Bridge deck 151, 251, 299, 351, 693
151 697 Computational simulation 383
Accelerated durability test 267 Bridge rehabilitation 29 Concrete 17, 21, 207, 243, 279,
Acoustic emission 143 Bridge repair 657 287, 311, 319, 323, 327, 391,
Active sensing 339 Bridges 111, 123, 147, 287, 653, 433, 449, 495, 507, 547, 601,
Additives 1 721 681, 689, 741
Adhesion 441 Buckling 171, 405, 713 -- beam 67
Adhesive 59, 87, 91, 275, 729 -- bridge deck 295
-- bonding 39, 271, 531 C -- bridges 259
-- joints 291 Cantilever 299 -- columns 193, 405
Advanced composite materials Carbon 507 -- confinement 315, 355
597 -- epoxy 327 --cylinder 693
Aggregate interlock 235 -- fiber anchors 665 -- jacketing 551
Al (aluminum alloy) 709 -- fiber fabric 417 -- repair 631
Aluminum 713, 717 -- fiber laminates 29 Concrete-filled steel tubes 503
Amount of carbon dioxide -- fiber reinforced plastic 193 Confined concrete 179
emissions 175 Carbon fiber reinforced polymer Confinement 183, 187, 197, 207,
Analysis 239 (CFRP) 35, 71, 183, 207, 235, 219, 231, 449, 487, 491, 495,
Anchor 87 263, 275, 387, 395, 425, 429, 499, 503
Anchorage 665 467, 471, 479, 483, 487, 547, Connections 717
-- strength models 63 559, 635, 639, 673, 705, 709, Construction 673
Anti-corrosion 709 713, 721, 729, 733, 741 -- guides 247
Arches 375 -- anchors 563 Continuous fiber rope 551
Axial loads 179, 713 -- grids 609 Core reinforced braided composite
-- laminates 685 rod 323
B -- reinforcement 315 Core reinforced braided fabric 323
Baseline-free nondestructive testing -- rods 685 Corrosion 311, 335, 597
339 -- sheets 539, 543, 563, 597, 631, -- resistance 139, 343, 453
Beam-column connections 661 665 Cost-efficiency 359, 363
Beams 449, 471, 527 -- strengthening 331 Crack debonding 51
Beam-strengthening 279 -- strips 119, 619 Crack propagation 425
Behavior 577 -- tapes 111, 657 Crack widths 159, 203, 307, 535
Bend capacity 311 Carbon fiber sheets 725 Cracking 107, 535
Bending 701 Carbon fibres 215, 259, 609, 643 Creep 207, 211, 215, 567
-- loading 167 Case study 175 Creep fracture propagation 25
Bent concrete surface 413 Chloride 335 Criticality 43
Blast loading 127 Cladding 163 Curvature 335
Blast mitigation 123 Code provisions 219 Curved FRP bar 311
Bolting 531 Cohesive zone model 21 Cyclic loading 555, 559
Bond 35, 43, 55, 63, 87, 95, 263, Cold-climate 331 Cycling test 143
283, 307, 347, 395, 433, 433, Cold-formed steel columns 725 Cylinder 449
669, 677 Columns 179, 183, 239, 409, 601
-- behavior 705 Composite curved beam 167 D
-- behavior and strength 17 Composite pipe 713 Debond 29
-- configuration 39 Composite rebar 267 Debonding 21, 35, 47, 59, 75, 83,
-- length 63 Composite sandwich panel 359, 91, 107, 139, 275, 339, 383, 413,
-- strength 107, 511, 689 363 619, 729
Bonded 631 Composite structures 211, 441, -- failures 71, 613
-- joints 441 709 Deck 147, 287
Bonding 519, 665, 737 Composite wedge 483 -- slabs 319
Bond-Slip Model 737 Composites 123, 163, 215, 259, Defect 43
Boundary element (BE) method 457, 643, 741 Deflection 159, 203, 319
Degradation 335 Fiber 601 Footbridge 99
Delamination 107 -- reinforced composite materials Fracture energy release rate 43
Deployable 457 323 Fracture mechanics 75
Design 187, 239, 243, 487, 585 -- reinforced plastics 271, 453 Full-scale test 539
-- guidelines 11, 227, 605 -- reinforced repair mortars 511
-- methods 235 Fiberglass sheet piling 171 G
-- of strengthening 255 Fibre reinforced composite 159 Generalized beam theory (GBT)
-- value 267 Fibreglass 131 421
Deterioration 263, 179, 395 Fibre-reinforced polymer (FRP) Girder 635
Development length 57 21, 39, 47, 51, 56, 63, 67, 79, 83, Glass fiber 167
Diagnosis 623 95, 99,107, 123, 127, 147, 151, Glass-fibre reinforced polymer
Disbonds 327, 523 163, 179, 187, 197, 211, 219, (GFRP) 87, 299, 461
Distortional failure 725 227, 239, 243, 251, 287, 319, -- stirrup 593
Distortional post-buckling 421 327, 347, 351, 355 375, 379, 391, Glass/polypropylene 223
Double shear lap 729 433, 437, 445, 449, 495, 499, Global buckling 409
Double-lap shear 71 503, 515, 519, 527, 531, 555, Gradient method 475
Ductility 47, 231, 315, 429, 491, 567, 571, 585, 605, 613, 623, Grids 355, 449
551, 571, 589 661, 673, 677, 693
Durability 1, 267, 271, 275, 379, -- bridge 135 H
453, 467 -- composite deck 155 Hanger 457
Dynamic analysis 127 -- composites 399, 421, 593, 627, Hardening 127
Dynamic response 251 697, 717 High durability 593
Dynamic tests 79 -- confinement 405 High strength concrete 203
-- deck panels 143 High-modulus 741
E -- experimental 581 High-temperature 681
Earthquake damage 567 -- guideline 531 Hollow core slab 539
Eccentric loading 193 -- laminates 17, 279 Hollow sections 405
Eccentric loads 499 -- poles 445 Hot melt pre-impregnated
Effect 43 -- rebar 303 composites 643
Effective bond length 71 -- reinforcements 11, 367, 371, Hybrid 351
Environmental aspect 175 535 -- columns 555
Environmental impacts 303 -- repair systems 247 -- constructions 1
Epoxy 167 -- shape 11 -- FRP-steel core 359, 363
-- Mortar 461 -- sheets 75, 413, 577, 589, 669, Hydraulic gate 453
Existing civil constructions 227 737
Experimental database 693 -- strengthening 383 I
Experimental results 685 -- structural Shapes 445 Intermediate crack (IC) debonding
Experimental validation 367 -- structures 175 67
Expert-system 623 -- U-jacketing anchorage 669 Impact traffic load 135
External bonding 437, 661, 677 -- wraps 499 Infrastructure rehabilitation 1
Externally bonded 515, 635, 741 FRP/concrete bond 79 Inorganic 681
Externally pre-stressed 471 FRP-concrete interface 25 -- epoxy 63
Extrapolation 267 FRP-concrete structure 343 Interaction curves 219
FRP-strengthening 231, 491, 725 Interface 67
F Field load test 119 -- models 383
Failure 239, 291 Field tests 111, 135 -- region relative humidity (IRRH)
-- simulation 139 Filament Winding 445 263
Far East 11 Film adhesive 643 Interfacial fracture energy 413
Fatigue 91, 139, 147, 283, 287, Finite element 359, 363, 433
717 -- analysis (FEA) 223, 697 J
-- life 425, 733 -- modeling 103, 399, 437 Jacketing 589
-- loading 669 Fire protection 681 Jackets 187
-- test 155 Flax fibres composites 627 Joint 571
-- testing 299 Flexural behaviour 445, 471 Jute fiber 167
FEA 429, 441 Flexural strength capacity 429
FE-analysis 387 Flexural strengthening 39, 619 L
FEM analysis 657 Flexure 91, 307, 527, 635, 741 Laboratory tests 371, 609
Large fracturing strain 589 Openings 387 Reinforced masonry 379
Life cycle 335 Out-of-plane loading 515 Reinforcement 375, 379, 417, 487
-- cost 99 Repair 523, 543, 721
Life cycles assessment 303 P -- and Retrofit 127
Light-weight 709 Parameters 577 -- mortar 685
-- concrete 351 Performance-based design 559 -- stages 111
-- structure 343 Permanent framework 343 Repairing techniques 461
-- superstructure 139 Physico-chemical characterization Research 11
Limit states 259 511 -- need 577
Live load model 295 Pier 123 Residual performance 567
Load bearing walls 399 Plaster 689 Residual thickness of concrete
Load capacities 171 Plasticity 495 (RTC) 263
Load distribution 251 Plates 635, 741 Resistance factors 255
Load rating 103 Plating 685 Retrofit 283, 601
Load testing 653 Polymer composites 1 Retrofitting 243, 347, 399, 531,
Local buckling 409, 421, 701 Polypropylene jacket 115 597
Local reinforcement 197 Polyurethane 163 Reverse taper 705
Local-plate failure 725 Post-tensioned 507 Rib 449
Long-term behaviour 211 -- concrete beam 657 Road bridge 119
Low velocity impact 115 Practical application 11
Load and Resistance Factor Design Pre-stressed 417, 639 S
(LRFD) 255 -- CFRP 475 Safety 155
-- concrete (PC) 83, 99, 471 -- concept 267
M -- concrete beam 111, 119, 475 Sandwich 351
Masonry 17, 375, 399, 519, 689 -- slab 539 -- panel 163
-- panel 371 Pre-stressing 483, 631 Scarf 523
-- vaults 367 -- force 479 Seismic loading 547
-- walls 515 -- tendons 467 Seismic retrofit 559, 567
Material variability 255 Prismatic columns 231, 491 Seismic retrofitting 551
MBC 609 Probabilistic 243 Seismic strengthening 227
Mechanical anchorages 631 Probability and reliability analysis Selection-criteria 623
Mechanical characterization 511 295 Serviceability 203, 535
Mechanically fastened fiber Prototype 649 Shallow depth patch 107
reinforced polymers (MF-FRP) Pull-off testing 59 Sharp corners 197
laminates 103 Pull-out testing 95 Shear 391, 437, 527, 577, 581,
Mineral based strengthening 609 Pulsating load 295 585, 601, 605, 609
Modelling 405 Pultruded composites 171 -- capacity 593
Modified compression field theory Pultruded shapes 211, 215 -- failure 343
(MCFT) 391 Pultruded structural shapes 131 -- strength 589
Modified virtual crack closure Pultrusion 291, 409, 441 -- strengthening 613
(VCCT) 433 -- stress 737
Moisture 395 Q Shearography 327
Moveable bridge 155 Quality control/assurance (QC/QA) Sheets 635, 721, 741
Moving traffic load 135 59 Shelters 457
Short span bridge 131
N R Shrinkage 107
Natural fibres 627 Reactive powder concrete 183 Single shear lap 729
Near surface mounted (NSM) Refurbishments technique 371 Slab-column connection 543
619, 639, 685 Rehabilitation 563, 597, 717 Slabs 387, 527, 535
-- reinforcement 95, 613 Reinforced 507 Slenderness 499
Nondestructive testing 327 Reinforced concrete (RC) 51, 83, Slip 95, 737
Nonlinear viscoelastic model 25 123, 159, 231, 235, 247, 347, Smeared crack 75
Near surface mounted (NSM) 619, 379, 437, 491, 527, 531, 555, Smoothed corners 197
639 563, 571, 585, 661, 673 S-N curve 669
Numerical simulation 395 -- beams 47, 429, 577, 613, 639 Solution uptake 467
-- bridge column 559 Spalling resistance 413
O -- columns 187, 219, 487, 499 Span effect 75
Splice joints 705 Strengthening 21, 67, 259, 347, Theoretical setup 367
Sprayed fiber reinforced polymers 387, 391, 429, 437, 475, 479, Thermal stresses 35
653 519, 527, 539, 543, 577, 581, Thermoplastic 147
Spun-cast piles 315 597, 605, 609, 623, 627, 631, -- composite 115, 223
Square cross-section 197 635, 649, 653, 661, 677, 733 Thin-walled members 421
Square cylinders 355 -- effect 331 Thin-walled steel sections 701
Standard 11 Stress-concentration 197 Timber 677
Static 635 Stretching 479 -- Pile 461
-- field load test 657 Strips 635 Time dependent behaviour 207
-- loads 111, 119, 577 Structural analysis 371 Time-reversal acoustics 339
-- testing 299 Structural health monitoring (SHM) Torsion 235, 581
Steel 275, 635, 721, 741 331, 335, 339 Truss 457
-- bridges 705 Subzero temperature 71 Two-dimensional 523
-- cord reinforced polymer (SCRP) Supporting conditions 519 Two-way shear 543
649 Surface flaws 247
-- fiber reinforced polymer (SFRP) Surface preparation 59 U
283 Sustained load 203 Ultimate strength 725
-- plate 733 Swelling 467 Upgrade 673
-- reinforced polymers (SRP) 123,
375, 631 T V
-- tubes 701 Thickness/depth (T/d) ratio 429 Vehicle bridge interaction 251
Stiffened plate 697 Temperature 35 Very cold temperatures 143
Stiffener 449 Tensile 523, 733
Stiffness 335 -- strength 311 W
-- degradation 143 Tension perpendicular to the grain Weather resistance 453
Strain gradient 193 627 Wed-bonded 571
Strain rate 79 Testing 319, 347 Wedge anchorage 483
Strains 319 -- procedure 17 Wide-flange profiles 409
Strarch 457 Tests 503 Wood 271, 681
Strength 291, 547 T-girders 581 WSGG anchor 417
Strengthen 471 Theoretical model 503
Composites in Civil Engineering
(CICE 2006)
 Organized by

and
Florida International University University of Miami
Miami, Florida, USA Miami, Florida, USA

Sponsored by

American Concrete Institute American Society of Civil

International Institute for (ACI) Engineering (ASCE)
FRP in Construction
(IIFC)

ISIS Canada Research Structural Engineering Institute

Canadian Society for Network (ISIS) (SEI)
Civil Engineering (CSCE)

ISBN: 0-615-13586-2

CICE2006 Secretariat
Department of Civil and Environmental Engineering
Florida International University (FIU)
10555 West Flagler Street, Miami, Fl 33174, USA
Phone: 305-348-2314
Fax: 305-348-2802
E-mail: cice2006@fiu.edu