Flexíveis

Deepwater Flexible
Risers and Pipelines

Scrivener Publishing
100 Cummings Center, Suite 541J
Beverly, MA 01915-6106
Publishers at Scrivener
Martin Scrivener (martin@scrivenerpublishing.com)
Phillip Carmical (pcarmical@scrivenerpublishing.com)
Deepwater Flexible
Risers and Pipelines
Yong Bai
This edition first published 2021 by John Wiley & Sons, Inc., 111 River Street, Hoboken, NJ 07030, USA
and Scrivener Publishing LLC, 100 Cummings Center, Suite 541J, Beverly, MA 01915, USA
© 2021 Scrivener Publishing LLC
For more information about Scrivener publications please visit www.scrivenerpublishing.com.
All rights reserved. No part of this publication may be reproduced, stored in a retrieval system, or
transmitted, in any form or by any means, electronic, mechanical, photocopying, recording, or other-
wise, except as permitted by law. Advice on how to obtain permission to reuse material from this title
is available at http://www.wiley.com/go/permissions.
Wiley Global Headquarters

111 River Street, Hoboken, NJ 07030, USA
For details of our global editorial offices, customer services, and more information about Wiley prod-
ucts visit us at www.wiley.com.
Limit of Liability/Disclaimer of Warranty

While the publisher and authors have used their best efforts in preparing this work, they make no rep
resentations or warranties with respect to the accuracy or completeness of the contents of this work and
specifically disclaim all warranties, including without limitation any implied warranties of merchant-
ability or fitness for a particular purpose. No warranty may be created or extended by sales representa
tives, written sales materials, or promotional statements for this work. The fact that an organization,
website, or product is referred to in this work as a citation and/or potential source of further informa
tion does not mean that the publisher and authors endorse the information or services the organiza
tion, website, or product may provide or recommendations it may make. This work is sold with the
understanding that the publisher is not engaged in rendering professional services. The advice and
strategies contained herein may not be suitable for your situation. You should consult with a specialist
where appropriate. Neither the publisher nor authors shall be liable for any loss of profit or any other
commercial damages, including but not limited to special, incidental, consequential, or other damages.
Further, readers should be aware that websites listed in this work may have changed or disappeared
between when this work was written and when it is read.
Library of Congress Cataloging-in-Publication Data
ISBN 9781119322726
Cover design by Kris Hackerott
Set in size of 11pt and Minion Pro by Manila Typesetting Company, Makati, Philippines
Printed in the USA
10 9 8 7 6 5 4 3 2 1
Contents
Preface xix
Acknowledgment xxi
About the Author xxiii
Part 1: Local Analysis 1
1 Introduction 3
1.1 Flexible Pipelines Overview 3
1.2 Environmental Conditions 4
1.3 Flexible Pipeline Geometry 7
1.4 Base Case-Failure Modes and Design Criteria 9
1.5 Reinforcements 10
1.6 Project and Objectives 12
References 12
2 Structural Design of Flexible Pipes in Different Water Depth 15
2.1 Introduction 15
2.2 Theoretical Models 15
2.3 Comparison and Discussion 24
2.4 Conclusions 34
References 34
3 Structural Design of High Pressure Flexible Pipes
of Different Internal Diameter 35
3.1 Introduction References 35
3.2 Analytical Models 37
3.2.1 Cylindrical Layers 37
3.2.2 Helix Layers 39
3.2.3 The Stiffness Matrix of Pipe as a Whole Helix Layers 40
3.2.4 Blasting Failure Criterion 41
3.3 FEA Modeling Description 42
3.4 Result and Discussion 46
v
vi Contents
3.5 Design 50
3.6 Conclusions 54
References 55
4 Tensile Behavior of Flexible Pipes 57
4.1 Introduction 57
4.2 Theoretical Models 58
4.2.1 Mechanical Model of Pressure Armor Layer 58
4.2.2 Mechanical Behavior of Tensile Armor Layer 61
4.2.3 Overall Mechanical Behavior 63
4.3 Numerical Model 64
4.3.1 Pressure Armor Stiffness 64
4.3.2 Full Pipe 69
4.5 Parametric Study 77
4.6 Conclusions 79
References 80
5 Design Case Study for Deep Water Risers 83
5.1 Abstract 83
5.2 Introduction 83
5.3 Cross-Sectional Design 85
5.4 Case Study 87
5.5 Design Result 94
5.6 Finite Elements Analysis 97
5.7 Conclusion 100
References 101
6 Unbonded Flexible Pipe Under Bending 103
6.1 Introduction 103
6.2 Helical Layer Within No-Slip Range 104
6.2.1 Geometry of Helical Layer 104
6.2.2 Bending Stiffness of Helical Layer 108
6.3 Helical Layer Within Slip Range 109
6.3.1 Critical Curvature 109
6.3.2 Axial Force in Helical Wire Within Slip Range 111
6.3.3 Axial Force in Helical Wire Within No-Slip Range 112
6.3.4 Bending Stiffness of Helical Layer 114
References 116
Contents vii
7 Coiling of Flexible Pipes 117

7.2 Local Analysis 120
7.2.1 Dimensions and Material Characteristics 120
7.2.2 Tension Test 120
7.2.3 Bending Test 123
7.2.4 Summary 124
7.3 Global Analysis 126
7.3.1 Modeling 126
7.3.2 Interaction and Mesh 127
7.3.3 Load and Boundary Conditions 128
7.3.4 Discussion of the Results 128
7.4.1 Diameter of the Coiling Drum 134
7.4.2 Sinking Distance of the Coiling Drum 135
7.4.3 Reeling Length 138
7.4.4 The Location of the Bearing Plate 139
7.5 Conclusions 142
References 143
Part 2: Riser Engineering 145

8 Flexible Risers and Flowlines 147
8.2 Flexible Pipe Cross-Section 147
8.2.1 Carcass 149
8.2.2 Internal Polymer Sheath 150
8.2.3 Pressure Armor 150
8.2.4 Tensile Armor 151
8.2.5 External Polymer Sheath 151
8.2.6 Other Layers and Configurations 152
8.3 End Fitting and Annulus Venting Design 152
8.3.1 End Fitting Design and Top Stiffener (or Bellmouth) 152
8.3.2 Annulus Venting System 153
8.4 Flexible Riser Design 154
8.4.1 Design Analysis 154
8.4.2 Riser System Interface Design 155
8.4.3 Current Design Limitations 156
References 158
viii Contents
9 Lazy-Wave Static Analysis 159

9.2 Fundamental Assumptions 162
9.3 Configuration Calculation 162
9.3.1 Cable Segment 163
9.3.1.1 Hang-Off Section 163
9.3.1.2 Buoyancy Section 166
9.3.1.3 Decline Section 166
9.3.2 Boundary-Layer Segment 167
9.3.3 Touchdown Segment 168
9.3.4 Boundary Conditions 170
9.4 Numerical Solution 171
9.5 Finite Element Model 174
9.5.1 Environment 175
9.5.2 Riser 175
9.7 Parameter Analysis 180
9.7.1 Effect of Seabed Stiffness 180
9.7.2 Effect of Hang-Off Inclination Angle 182
9.7.3 Effect of Buoyancy Section Length 185
9.8 Conclusions 187
References 188
10 Steep-Wave Static Configuration 189
10.2 Configuration Calculation 190
10.2.1 Touch-Down Segment 191
10.2.2 Buoyancy Segment 194
10.2.3 Hang-Off Segment 195
10.3 Numerical Solution 196
10.5 Parametric Analysis 203
10.5.1 Effect of Buoyancy Segment’s Equivalent
Outer Diameter 203
10.5.2 Effect of Buoyancy Segment Length 205
10.5.3 Effect of Buoyancy Segment Location 207
10.5.4 Effect of Current Velocity 209
10.6 Conclusions 212
References 212
Contents ix
11 3D Rod Theory for Static and Dynamic Analysis 213

11.2 Nomenclature 215
11.3 Mathematical Model 216
11.3.1 Governing Equations 216
11.3.2 Bending Hysteretic Behavior 220
11.3.3 Bend Stiffener Constraint 222
11.3.4 Pipe-Soil Interaction 224
11.4 Case Study 225
11.5 Results and Discussion 227
11.5.1 Static Analysis 227
11.5.2 Dynamic Analysis 231
11.5.2.1 Top-End Region 231
11.5.2.2 Touchdown Zone 233
11.5.3 Effect of Bend Stiffener Constraint 236
11.5.4 Effect of Bending Hysteretic Behavior 238
11.5.5 Effect of Top Angle Constraint 240
References 243
12 Dynamic Analysis of the Cable-Body of the Deep Underwater
Towed System 247
12.2 Establishment of Towed System Dynamic Model 248
12.3 Numerical Simulation and Analysis of Calculation Results 251
12.3.1 The Effect of Different Turning Radius 252
12.3.2 The Effect of Different Turning Speeds 253
12.3.3 Dynamic Analysis of the Towed System
with the Change of the Parameters of the Cable 254
12.3.4 The Effect of the Diameters of the Towed Cable 257
12.3.5 The Effect of the Drag Coefficients
of the Towed Cable 257
12.3.6 The Effect of the Added Mass Coefficient
of the Towed Cable 261
Acknowledgments 264
References 264
13 Dynamic Analysis of Umbilical Cable Under Interference 267
13.2 Dynamic Model of Umbilical Cable 269
x Contents
13.2.1 Establishment of Mathematical Model 269

13.2.2 The Discrete Numerical Method for Solving
the Lumped Mass Method 271
13.2.3 Calculation of the Clashing Force
of Umbilical Cable 277
13.3 The Establishment of Dynamic Simulation
Model in OrcaFlex 279
13.3.1 The Equivalent Calculation of the Stiffness
of the Umbilical Cable 279
13.3.2 RAO of the Platform 281
13.3.3 The Choice of Wave Theory 281
13.3.4 Establishment of Model in OrcaFlex 282
13.4 The Calculation Results 283
13.4.1 The Clashing Force of Interference 283
13.4.2 The Variation of the Effective Tension
Under Interference 285
13.4.3 The Variation of Bending Under Interference 287
13.5 Conclusion 291
References 294
14 Fatigue Analysis of Flexible Riser 295
14.2 Fatigue Failure Mode of Flexible Riser 296
14.3 Global Model of Flexible Risers 297
14.3.1 Pipe Element 297
14.3.2 Bending Stiffener 298
14.3.3 Sea Condition 299
14.3.4 Platform Motion Response 300
14.3.5 Time Domain Simulation Analysis 301
14.4 Failure Mode and Design Criteria 302
14.4.1 Axisymmetric Load Model 302
14.4.2 Bending Load Model 303
14.5 Calculation Method of Fatigue Life
of Flexible Riser 305
14.5.1 Rainflow Counting Method 305
14.5.2 S-N Curve 305
14.5.3 Miner’s Linear Cumulative Damage Theory 307
14.5.4 Modification of Average Stress on Fatigue Damage 308
14.6 Example of Fatigue Life Analysis of Flexible Riser 309
References 314
Contents xi
15 Steel Tube Umbilical and Control Systems 317

15.1.1 General 317
15.1.2 Feasibility Study 318
15.1.3 Detailed Design and Installation 319
15.1.4 Qualification Tests 320
15.2 Control Systems 320
15.2.1 General 320
15.2.2 Control Systems 321
15.2.3 Elements of Control System 322
15.2.4 Umbilical Technological Challenges and Solutions 323
15.3 Cross-Sectional Design of the Umbilical 326
15.4 Steel Tube Design Capacity Verification 327
15.4.1 Pressure Containment 328
15.4.2 Allowable Bending Radius 328
15.5 Extreme Wave Analysis 329
15.6 Manufacturing Fatigue Analysis 330
15.6.1 Accumulated Plastic Strain 330
15.6.2 Low Cycle Fatigue 331
15.7 In-Place Fatigue Analysis 331
15.7.1 Selection of Sea State Data From Wave Scatter
Diagram 332
15.7.2 Analysis of Finite Element Static Model 332
15.8 Installation Analysis 332
15.9 Required On-Seabed Length for Stability 333
References 334
16 Stress and Fatigue of Umbilicals 337
16.2 STU Fatigue Models 338
16.2.1 Simplified Model 339
16.2.1.1 Axial and Bending Stresses 339
16.2.1.2 Friction Stress 340
16.2.1.3 Simplified Approach: Combining
Stresses 342
16.2.1.4 Simplified (Combining Stresses)
Fatigue Damage 342
16.2.1.5 Simplified Model Assumptions 343
16.2.2 Enhanced Non-Linear Time Domain
Fatigue Model 343
xii Contents
16.2.2.1 Friction Stresses 344

16.2.2.2 Effect of Multiple Tube Layers 344
16.2.2.3 Combined Friction Stresses 345
16.2.2.4 Axial and Bending Stresses 345
16.2.2.5 Combining Stresses 346
16.2.2.6 Fatigue Life 346
16.2.2.7 Benefits of Enhanced Non-Linear Time
Domain Fatigue Model 347
16.3 Worked Example 348
16.3.1 Time Domain vs. Simplified Approaches 350
16.3.2 Effect of Friction on STU Fatigue 351
16.3.2.1 Influence of High Tube Friction on
Umbilical Fatigue 352
16.3.2.2 Influence of Low Tube Friction on
Umbilical Fatigue 352
16.3.2.3 Influence of Metal-to-Metal Friction vs.
Metal-to-Plastic Contact on Umbilical
Fatigue 352
16.3.3 Effect of Increasing Water Depth 353
16.3.4 Effect of Increasing the Tube Layer Radius 354
16.5 Recommendations 356
References 357
17 Cross-Sectional Stiffness for Umbilicals 359
17.2 Theoretical Model of Umbilicals 361
17.3 Bending Stiffness of Umbilicals 362
17.4 Tensile Stiffness of Umbilicals 366
17.5 Torsional Stiffness of Umbilicals 368
17.6 Ultimate Capacity of Umbilicals 368
17.6.1 Minimum Bending Curvature 368
17.6.2 Minimum Tensile Load 369
17.6.3 Tensile Capacity Curve 369
References 372
18 Umbilical Cross-Section Design 375
18.1.1 General 375
18.1.2 Sectional Composition of the Umbilical Cable 375
18.1.3 Umbilical Cable Structure Features 376
Contents xiii
18.2 Umbilicals Cross-Section Design Overview 377

18.2.1 Umbilical Cross-Section Design Flowchart 377
18.2.2 Load Analysis 378
18.3 Umbilical Cable Cross-Section Design 380
18.3.1 Umbilical Cable Cross-Section Layout Design 380
18.3.2 Tensile Performance Design 381
18.3.3 Bending Performance Design 382
References 384
Part 3: Fiber Glass Reinforced Deep

Water Risers 385
19 Collapse Strength of Fiber Glass Reinforced Riser 387
19.2 External Pressure Test 388
19.2.1 Testing Specimen 388
19.2.2 Testing System 389
19.2.3 Testing Results 389
19.3 Theoretical Analysis 390
19.3.1 Fundamental Assumptions 390
19.3.2 Constitutive Model of Materials 391
19.3.3 Establish the Equations of Motion 393
19.3.4 Establish Virtual Work Equations 394
19.4 Numerical Analysis 394
19.5 Finite Element Analysis 395
19.5.1 Establish the Finite Element Model 396
19.5.2 The Results of the Finite Element Analysis 397
19.6 Conclusion 401
References 402
20 Burst Strength of Fiber Glass Reinforced Riser 405

20.2 Experiment 406
20.2.1 Dimensions and Material Properties of FGRFP 406
20.2.2 Experiment Device 407
20.2.3 Experiment Results 407
20.3 Numerical Simulations 407
20.3.1 Mesh and Interaction 407
20.3.2 Load and Boundary Conditions 408
20.3.3 Numerical Results 409
xiv Contents
20.4 Analytical Solution 409

20.4.1 Basic Assumptions 409
20.4.2 Stress Analysis 411
20.4.3 Boundary Condition 414
20.5 Results and Discussion 416
20.6.1 Winding Angle of Fiber Glass 417
20.6.2 Diameter-Thickness Ratio 418
References 419
21 Structural Analysis of Fiberglass Reinforced Bonded Flexible

Pipe Subjected to Tension 421
21.2 Experiment 423
21.2.2 Material Characteristics 425
21.2.3 Experimental Results 426
21.3 Theoretical Solution 427
21.3.2 Cross-Section Simplification 429
21.3.3 Fiber Deformation 430
21.3.4 Cross-Section Deformation 431
21.3.5 Equilibrium Equations 434
21.4 Finite Element Model 434
21.5.1 Tension-Extension Relation 436
21.5.2 Cross-Section Deformation 437
21.5.3 Fiberglass Stress 439
21.5.4 Contribution of Each Material 439
21.5.5 Summary 440
21.6.1 Winding Angle 442
21.6.2 Fiberglass Amount 443
21.6.3 Diameter-Thickness Ratio 444
Acknowledgement 446
References 446
22 Fiberglass Reinforced Flexible Pipes Under Bending 449

22.2 Experiment 451
Contents xv
22.2.1 Experimental Facility 451

22.2.2 Specimen 453
22.2.3 Experiment Process 453
22.2.4 Experimental Results 455
22.3.1 Fundamental Assumption 457
22.3.2 Kinematic Equation 457
22.3.3 Material Simplification 459
22.3.4 Constitutive Model 462
22.3.5 Principle of Virtual Work 464
22.3.6 Algorithm of Analytical Solutions 464
22.4 Finite Element Method 465
22.5 Result and Conclusion 466
22.6.1 D/t Ratio 469
22.6.2 Initial Ovality 470
References 473
23 Fiberglass Reinforced Flexible Pipes Under Torsion 475
23.2 Experiments 477
23.3 Experimental Results 478
23.4.1 Coordinate Systems 481
23.4.2 Elastic Constants of Reinforced Layers
(k = 2, 3 … (n − 1)) 483
23.4.3 Reinforced Layers Stiffness Matrix k = 2, 3...(n – 1) 484
23.4.4 Inner Layer and Outer Layer Stiffness Matrix
(k = 1, n) 486
23.4.5 Stress and Deformation Analysis 487
23.4.7 Interface Conditions 492
23.4.8 Geometric Nonlinearity 493
23.5 Numerical Simulations 494
23.6 Results and Discussions 496
23.7.1 Effect of Winding Angle 498
23.7.2 Effect of Thickness of Reinforced Layers 498
Acknowledgments 500
References 501
xvi Contents
24 Cross-Section Design of Fiberglass Reinforced Riser 503

24.2 Nomenclature 503
24.3 Basic Structure of Pipe 505
24.3.1 Overall Structure 505
24.3.2 Material 506
24.4 Strength Failure Design Criteria 506
24.4.1 Burst Pressure 506
24.4.2 Burst Pressure Under Internal Pressure Bending
Moment 508
24.4.3 Yield Tension 508
24.5 Failure Criteria for Instability Design 510
24.5.1 Minimum Bending Radius 510
24.5.2 External Pressure Instability Pressure 510
24.6 Design Criteria for Leakage Failure 511
References 511
25 Fatigue Life Assessment of Fiberglass Reinforced
Flexible Pipes 513
25.2 Global Analysis 515
25.3 Rain Flow Method 517
25.4 Local Analysis 519
25.5 Modeling 519
25.6 Result Discussion 520
25.7 Sensitivity Analysis 524
25.8 Fatigue Life Assessment 527
25.9 Conclusion 528
References 529
Part 4: Ancillary Equipments for Flexibles

and Umbilicals 531
26 Typical Connector Design for Risers 533
26.2 Carcass 534
26.3 Typical Connector 535
26.4 Seal System 536
26.5 Termination of the Carcass 537
26.6 Smooth Bore Pipe 539
26.7 Rough Bore Pipe 540
26.8 Discussion 542
Contents xvii

References 545
27 Bend Stiffener and Restrictor Design 547
27.2 Response Model 548
27.3 Extreme Load Description 549
27.4 General Optimization Scheme 550
27.5 Application Example 552
27.6 Non-Dimensional Bend Stiffener Design 553
27.7 Alternative Non-Dimensional Parameters 556
References 558
28 End Termination Design for Umbilicals 561
28.2 Umbilical Termination Assembly 561
28.2.1 General 561
28.2.2 UTA Design 562
28.2.3 UTA Structural Design Basis 565
28.3 Subsea Termination Interface 566
References 568
29 Mechanical Properties of Glass Fibre Reinforced
Pipeline During the Laying Process 569
29.2 Theoretical Analysis 570
29.2.1 Wave Load 570
29.2.2 Motion Response of the Vessel 572
29.2.3 Dynamic Numerical Solution 573
29.3 Static Analysis 575
29.4 Dynamic Characteristic Analysis 579
29.4.1 Influence of the Wave Direction 579
29.4.2 Influencing of Different Lay Angle 582
29.4.3 Influencing Submerged Weight 584
References 586
Index 589
Preface
Deepwater flexible risers and pipelines are challenges for deepwater oil
and gas productions. This is particularly important when we use FPSO and
semi-submersible solutions. In the Gulf of Mexico, deepwater flexible ris-
ers and pipelines are also widely used for tie-in of new production wells
and manifolds to existing production systems such as TLPs and Spars, etc.
The author has been fortunately involved with deepwater flexible risers
and pipelines projects when he was working in the USA and Norway. In
the past 15 years, he has been engaged in research and consulting of rel-
evant subjects. This book summarizes his work in four different subject
areas:
Part I deals with mechanics of deepwater risers where our focus is cross-
section design of the risers under combined loads. We derived formu-
lation for strength analysis of deepwater flexible pipes under internal
pressure, external pressure, tension, torsion, and bending curvatures. We
also address strength of the pipe during coiling, installation, and possible
failure during the process.
Part II addresses global response of the riser systems in deepwater. We

derived formulations for static configuration design and dynamic response.
This is particularly important for global configuration design in static and
dynamic environment.
Part III is devoted to a new kind of flexible pipe that is made of fiberglass
material. We derived formations for pipes under internal pressure, exter-
nal pressure, tension, bending, and torsion.
Part IV summarizes experience we have for ancillary equipment design.

This is mainly for riser hang-off locations and touch down areas. We
address the design of bending stiffeners, bend restrictors and connectors.
xix
xx Preface
Riser connectors are perhaps the single most critical element in the
safety and integrity of the riser and pipeline systems.
We wish that this book will be a useful reference source of flexible risers
and pipelines design and analysis for subsea engineers. This book mainly
considers deepwater water applications. For shallow applications, we
recommend our book “Flexible Pipelines and Power Cables”. For cross-
sectional design, we recommend our book entitled “Flexible Pipes”. All
three books are published by Scrivener Publishing and Wiley.
Acknowledgment
The authors would like to appreciate Prof. Yong Bai’s graduate students and
post-doctoral fellows at Zhejiang University who provided the initial tech-
nical writing for Chapters 2, 3, and 4 (Fancesco Cornacchia, Qiangqiang
Shao, and Dr. Ting Liu), Chapter 6 (Dr. Yutian Lu), Chapter 7 (Dr. Yuxin
Xu and Dr. Pan Fang). Chapters 9 to 11 (Dr. Weidong Ruan), Chapters 12
and 13 (Dr. DaPeng Zhang), Chapters 14 (Dr. KA Jiamg), Chapters 17
and 18 (Dr. Wei Qin), Chapter 19 (Dr. XiaoJie Zhang), Chapter 20 (Dr.
Shanying Lin), Chapter 21 (Dr. Yuxin Xu), Chapter 22 (Dr. Yifan Gao),
Chapter 23 (Ms. Xinyu Sun), Chapter 24 (Dr. Mohsen Saneian), Chapters
25 to 26 (Dr. Wei Qin) Chapter 29 (Dr. Zhao Wamg). Thanks to all persons
involved in reviewing this book. The authors also would like to thank the
flexible pipe manufacturing company OPR Inc. for their support for pub-
lishing this book.
The author is grateful to Ms. Xin Zhou who provided editorial assistance.
I am thankful to Martin Scrivener and Phillip Carmical of the Scrivener
Publishing and Wiley.
Prof. Yong Bai

June 01, 2020
xxi
About the Author
Professor Yong Bai is the president of Offshore Pipelines and Risers Inc. and
also the director of the Offshore Engineering Research Center at Zhejiang
University. He has previously taught at the University of Stavanger in
Norway where he was a professor of offshore structures. He has also
worked with ABS as manager of the Offshore Technology Department and
DNV as a JIP project manager.
Professor Yong Bai has also worked for Shell International E&P as a
staff engineer. Through working at JP Kenny as manager of advanced
engineering and at MCS as vice president of engineering, he has contrib-
uted to the advancement of methods and tools for the design and analysis
of subsea pipelines and risers.
Professor Bai is the author of approximately 10 books such as “Marine
Structural Design” and “Subsea Pipelines and Risers”. He authored more
than 100 SCI and EI papers on the design and installation of subsea pipe-
lines and risers.
xxiii
Part 1
LOCAL ANALYSIS
1
Introduction
1.1 Flexible Pipelines Overview

The oil and gas industry proclaims pipelines as the most economical device
of large scale overland conveyance for crude oil and natural gas, compared
to truck and rail transportation. These can continuously carry large quan-
tities of fluids and they are considered by far to be the most reliable.
The rapid development of the petroleum industry leads to high finan-
cial investments in engineering research, which has quickly conducted
to achievement of remarkable effects, so that recently steel homoge-
neous pipes have been replaced by flexible pipelines, widely used by oil
and gas industry for both onshore and offshore purposes. Mostly, they
are employed offshore as transportation of well products such as oil, gas,
and condensate, well control lines, injection of water and gases, and export
of processed product, but also as flowlines connecting subsea wells, well-
head platforms, templates or loading terminals, processing platforms, and
jumper lines connecting fixed platforms to support vessels.
Flexible pipelines are relatively new kinds of profiles which have devel-
oped since the late 1970s; initially, they were used in few offshore areas
until being employed in many projects, thanks to the applicability in
water depths up to 8,000 ft, pressures up to 10,000 psi, and high tempera-
tures up to 150°, beside the high adaptability to different environmental
conditions and large vessels motion [1].
Nowadays, even if many researchers are employed to enhance knowl-
edge about flexible pipelines, there are no books available that systemati-
cally introduce the design procedures and analysis criteria that are always
valid for the wide spectrum of these structures. On the other hand, it is
possible to find many reference sources which can streamline the issues.
Moreover, they are widely used also because of their easy and cheap
transport and installation, due to the possibility of being prefabricated
onshore in long lengths and stored in limited size on reels; in fact, as most
relevant structural property, these pipes show very low bending stiffness
Yong Bai. Deepwater Flexible Risers and Pipelines, (3–14) © 2021 Scrivener Publishing LLC
3
4 Deepwater Flexible Risers and Pipelines
in comparison to axial tensile stiffness. Besides, the economic benefits,

they are considered to be technically proficient due to their easy and fast
laying procedure, durability, and recoverability. In the light of the above,
the petroleum industry turns into flexible structures, allowing for per-
manent connection between the subsea system and any facility at the water
surface with large motions.
Being flexible pipes, crucial elements for the right operational oil and
gas spill in terms of both performance and pollution, at the same time, a
deep finite element model (FEM) is highly recommended in order to ver-
ify the reliability of the design. This method called DTA (Design Through
Analysis) involves both the abovementioned procedures and the two-step
process is used in complementary way in order to reach less conserva-
tive outcomes in designing, thus minimizing the project CAPEX (Capital
Expenditure) and OPEX (Operating Expenditure). In some cases, it shows
that codes and regulations are over conservative, and the real behavior
can be captured through a FE simulation when the input parameters are
well defined, if not data are statistically modeled in order to produce a reli-
able distribution for a range of loads and effects.
1.2 Environmental Conditions

The wide expansion of these structures during the last decades has been
made possible thanks to the costs-design optimization. It considers a
deep understanding of the environmental conditions in which the pipe-
line will be installed and operated. Some of the drivers are: water depth
and oceanographic data, chemical composition and flow phase of the
extracted fluid, quantity of salt in the surrounding water, and operating
internal and external temperatures and pressure. In addition, transport
and installation circumstances must be considered due to the fact that in
some cases, extreme loading conditions are shown during these phases.
Most of the time, combination of loads needs to be taken into account.
The analysis of the surrounding conditions, for example, temperature and
corrosion, cannot be under estimation because of limiting capacity of the
structures.
Temperature can affect the correct purpose of the pipe, in some cases,
very high or low temperature of the fluids leads to the need of extra thermal
insulation design, such as pipe-in-pipe or wet insulation, which considers
the different thermal responses of materials. Polymeric materials exhibit
lower thermal conductivity and higher thermal expansion coefficients
Introduction 5
compared to steel, so that the plastic layers govern the temperature profile
through the pipe wall. Operating temperature is the foremost principle
for selecting the polymer material in order to ensure the correct mechan-
ical behavior of the pipe. In fact, for a given material, as the temperature
rises the magnitude of the yield stress and Young modulus decreases, so
the ultimate strain increases and vice versa, leading to various mechanical
response properties during the service life of the structures. Generally,
temperature profile impacts most of the design parameters, it is also
influenced by the water depth outwardly and by the reservoir formation
internally; thus, it must be carefully considered for both service limit state
and ultimate limit state.
Cross-sectional integrity needs to be mentioned for a remarkable pipe-
line design, which is of relevant importance especially for deep water.
In fact, here hydro-static pressure is high, and the cross-section must be
dimensioned against local buckling failure which is extremely influenced
by imperfections. Not only, any other steel components provide structural
support against axial, bending, and torsional loads, and their integrity
is essential. Corrosion and cracks damages are hard to detect, being the
structural profile made by different layers. Considering the conservative
but actual hypothesis of failure of polymeric materials, steel corrosion is
caused by interaction with salt water, air, or internal acid fluids or combi-
nation of them, which chemically alter materials. Besides the mechanical
properties and price, the main driver for the selection of steel materials
is the corrosion resistance to operating environment. The annulus condi-
tions are continuously tested and monitored, and it is common practice
to include deterioration protections such as coating, corrosion inhibitors,
application of special materials and cathodic protection, in addition to
lubrication oil distributed along the pipe during the manufacturing.
During the years, the availability of reservoirs onshore and in shal-
low water has decreased, and the need of petroleum pushed the industry
to open new challenging offshore campaigns. The employment of flexi-
ble pipes in subsea brings researchers to focus on the estimation of the
structural behavior in deep waters. Here, the environmental conditions are
tougher, and reinforcements are necessary. Hydro-static pressure rises as
water depth increases, which leads to considerable hoop stresses and buck-
ling issues. For the whole pipe, this problem is managed making use
of devices that enhance the strength in this sense, but at the same time,
they bring additional structural weight and increasing gravity loads.
Now, it can be deduced that as the water column grows, the magni-
tude of the tensile load rises. Dynamic tension is amplified by existing drag
forces if considering the substantial pipe length in deep water, which is

directly reflected on the end fitting, with consequently fatigue damages at
this point due to constraint effects that may take place.
What being said, the challenge of the engineering is trying to make the
best design, keeping the costs as low as possible, moving from local to
global analysis. Any structure needs a particular design for the specific
field and for precise environmental conditions, at the same time taking
advantages from the experiences about other projects. The selection of the
most suitable setup for any flexible structure is founded on practice and
engineering judgment, in Figure 1.1, some examples of possible riser con-
figurations using flexible pipes for which the choice is based on the above-
mentioned principles [2–4].
(a) Free-hanging Catenary
(b) Steep-S (c) Lazy-S
(d) Steep Wave (e) Lazy Wave
Figure 1.1 Employment of flexible pipelines as risers.

Introduction 7
1.3 Flexible Pipeline Geometry

Even if flexible pipelines have many applications, the common factor is
to provide large flexural deformations while they are subjected to other
loading forces, strictly necessary during installation and transport. Mostly,
this property depends on their geometrical configuration, which is made
by many layers. Generally, flexible pipes are compound by many main
layers of different materials and functions:
• Internal polymeric sealing layer provides insulation from

internal fluids, prevents corrosion and leakage due to extract
materials,
• Helical armoring layers provide the required strength
against different loading conditions, the number of lay-
ers and wires per each layer is variable and depends on the
design;
• Outer polymeric layer prevents seawater from interacting
with the armor layers.
They can be characterized in two main groups: the metal-based flexible

pipes designed to withstand high loads see Figure 1.2 and the composite-
based flexible pipes (FCP), which are much simpler and employed for
lower functional requirements. These can be further divided in two more
groups: bonded and unbonded structures.
In the bonded structures, all the elements are fused together in the sur-
rounding matrix through a vulcanization process.
Figure 1.2 Unbonded composite-based flexible pipe.

This work is focused on unbonded type, i.e., when layers are not
connected into a single structure, but each component makes up a cylin-
drical layer that is able to slide relatively to the other layers. Composite-
based unbonded flexible pipes are defined as the base-case for this work
and, basically, are composed by an innermost thermoplastic layer called
liner made of polyester material that surrounds the collected material. Two
(or four) layers of carbon steel material strip reinforcements, adjacently,
spirally wounded around the outer surface of the innermost cylinder at an
opposite lay angle of about 55°, one by one (or two by two) to withstand
torque loading. This profile must provide mainly support against tensile
loads. Finally, an outermost polymeric sheath to prevent contamination
from the external environment.
On the other hand, metal-based unbonded flexible pipes are suitable for
static and dynamic applications with length of several hundred meters;
they are compound by nine layers with different functions. The inter-
locked metal carcass layer and the pressure armor layer are added in the
profile with respect to the unbonded FCP. The first is needed to prevent the
collapse due to high hydro static pressure or sudden depressurization of
internal fluids and also to avoid erosion from the extracted materials; the
second provides strength against high hoop stresses due to internal and
external pressure. Also, the tensile strength is improved, substituting strips
with wires, which show wider cross-sectional dimensions.
A further division among unbonded pipelines is made considering the
presence of the steel carcass supporting the inner liner: if it is included in
the design, the pipe can be named rough bore; if not, it is a smooth bore
structure see Figure 1.3.
4 6
4
4 5
3 5
2
1
Figure 1.3 Unbonded metal-based flexible pipe.

Introduction 9
Other elements, such as anti-wear and bird-caging tapes, can be con-

sidered as non-structural, but they are very relevant in designing because
they rule the flow and the contacts, which can modify the properties of the
structural elements.
1.4 Base Case-Failure Modes and Design Criteria

API Recommended Practice 17B is the guideline to follow in order to have
a proficient design. Any pipelines must satisfy code requirements under
the actual environmental conditions, beyond which it is not suitable any-
more for its purpose. In order to do so, different types of load and failure
modes must be considered.
Failure is not necessary considered as a structural limit, but it can be
easily defined as a modification of the flow conditions within the pipe, in
fact leakage and reduction of internal cross-section undermine the main
achievement of the structure.
Unbonded composite-based are weaker than metal-based flexible pipes
and, in many cases, are not the best choice for deep water. However, a list
of all the possible pipe failure modes and mechanisms from the code, valid
for both cases, is delivered below:
• Collapse
• Burst
• Tensile failure
• Compressive failure
• Overbending
• Torsional failure
• Fatigue failure
• Erosion
• Corrosion
All the components are subjected to external loads and they have dif-
ferent reactions and strength, and in general, the effects increase as
the water depth grows. Some failure mechanisms for specific elements
appear as results of the failure mode due to loads for which they are not
designed for. For composite unbonded pipelines, it is common that steel
strips, which are designed for withstanding axial loads, may buckle under
excessive hydro-static pressure or tensile loads, as well as during the trans-
port phase.
If unbonded flexible pipes are subjected to axisymmetric loads, the stiff-

ness has the same order of magnitude of steel pipes and they have linear
behavior for tension, pressure, and torsion. Actually, hypothesis needs to
be introduced to simulate the real response of the structure trying to keep
the abovementioned configuration, for example, for local buckling imper-
fection are introduced symmetrically distributed along the cross-section.
It needs to be underlined that for weak pipes, such as FCP, the strength
provided by internal and external plastic layers cannot be neglected.
A series of design criteria needs to be satisfied through strength reduc-
tion factors. When dealing with oil and gas problems, in particular off-
shore, safety factors are very severe. For flexible pipes also, due to the very
high uncertainties due to the complexity related to the helical shape of the
steel layers, interactions, materials, and residual stresses, besides environ-
mental forces, the safe margin must be considered wide enough. Design
criteria express the safety for all the components; they are expressed in
terms of allowable quantities which are specified by regulations and man-
ufacturing, for:
• Strain
• Creep
• Stress
• Hydrostatic collapse
• Mechanical collapse
• Torsion
• Crushing collapse and ovalization
• Compression
• Service life factors
Details of design criterion and analysis methods may be founded in Ref [5–7].
1.5 Reinforcements
In order to improve the performance of flexible pipelines, especially for
deep water cases, reinforcements must be included in the cross-section
design, depending on the environmental conditions.
The interlocked carcass is the innermost layer, and it is the only com-
ponent in direct contact to the fluid in the bore. It is a corrugated metallic
tube, fabricated spirally winding a corrugated profile, which is shown in
Figure 1.4, around a mandrel at an angle close to 90°, so that each sec-
tion can slide on the adjacent, releasing the required bending stiffness.
Introduction 11
Figure 1.4 Interlocked carcass.
The structure is not fluid tight, and its steel material needs to be chemically
resistant to the conveyed liquids and gases.
The main reason of considering the carcass in the design is to provide
enough radial resistance against external pressure. Moreover, it delivers
the required strength in case of build-up (rapid decompression caused by
low pressure shut-down in the annulus) but also for crushing loads during
installation operations, handling of the pipe, and pippin tools.
Issues about internal pressure, which may cause high hoop stresses,
can be avoided or limited designing a pressure armor. It is a corrugate
profile which shows notable strength in the hoop direction, where the
cross-sections are interlocked and free to slide one to each other, in order
to confer low bending stiffness at the same time. The metallic cross-section
can have different forms; the most used Z-shaped and C-shaped are illus-
trate in Figure 1.5.
When dealing with deep waters, withstanding high tension is one of
the main challenges. Strips are too thin to carry substantial axial loads
and they must be increased in size, for rectangular or close to rectangular
shape. Thus, a proper design provides the employment of wires, in two or
four layers winded at an angle with the longitudinal axis within about 35°
and 55°. The winding angle is opposite one by one, or two by two, layers to
provide torsional resistance for the two directions.
Figure 1.5 Pressure armor cross-section profiles.

1.6 Project and Objectives

This part aims to understand the need of supplementary elements such us
interlocked carcass and tensile armor wires to support the design in case of
extreme environmental conditions. To satisfy safe and proficient operation
of flexible pipelines in deep water conditions, numerical and theoretical
models are developed, which must lead to an accurate prediction of the
strength with respect to the critical failure modes. Comparisons among
results are made and enhanced cross-sectional profiles are carried out for
the case of metallic strip flexible pipe and MSFP.
In order to do so, the base cases have been deeply studied, and already
existing both theoretical and numerical models applied. According to
API recommended practice 17B, innovative cross-sections for both cases
are established, combining different hypothesis and theories and verify-
ing results with the commercial finite element software ABAQUS, which
hopefully provide both useful information and understanding.
Critical conditions for two different load cases are studied, such exter-
nal pressure and tensile load, which as previously said, and are strictly
connected for deep water conditions. Against high external pressure, the
interlocked carcass is essential in order to prevent collapse. Being of intri-
cate shape, theoretical models which turn into equivalent reproductions
are used. In this work, wide spectrum of FEM for different geometries is
analyzed and an empirical formulation is carried out for the minimum
requirement of ovality, which suggests a modified theoretical model closer
to the more realistic behavior of the numerical simulation.
Strength against wide tensile loads is provided by the tensile armor lay-
ers. A profile also including pressure armor is considered; again, an equiv-
alent theoretical model is employed to catch its particular behavior due
to its shape. A simplified theoretical model is obtained which is con-
firmed by a numerical simulation. The validated model is then extended
to understand the performance when the pipe is subjected to combined
tensile force and external pressure.
References
1. N. Ismail, R. Nielsen, and M. Kanarellis, Design Considerations for Selection
of Flexible Riser Configuration, PD-Vol. 42, Offshore and Arctic Operations,
ASME, 1992.
2. Ruan W, Bai Y, Cheng P. Static analysis of deepwater lazy-wave umbilical on
elastic seabed[J]. Ocean Engineering, 2014, 91: 73–83.
Introduction 13
3. Garrett D L. Dynamic analysis of slender rods[J]. J. Energy Resour.

Technol.;(United States), 1982, 104(4).
4. Croll J G A. Bending boundary layers in tensioned cables and rods[J].
Applied ocean research, 2000, 22(4): 241–253.
5. Bai Y, and Bai Q. Subsea pipelines and risers. Elsevier, 2005.
6. Palmer A C, King R A. Submarine Pipeline Engineering[M]. PennWell
Books, 2004.
7. Orcina Ltd., Visual ORCAFLEX User Manual, Ulverston, Cumbria, UK,
2000.
2
Structural Design of Flexible Pipes
in Different Water Depth
2.1 Introduction
Flexible pipelines must be designed considering the extreme case in which
a leakage through the layers can happen. The external pressure would be
applied directly on the innermost PE layer, which provides only hydraulic
insulation thus to the carcass surface. Mostly, the issue is relevant for deep
water depth because of high hydrostatic pressures and tools to design the
interlocked carcass under code requirements are needed. This work aims
to give an estimation of the critical buckling load for interlocked carcass
accounting for imperfections, which is valid when the profile can be
considered as thin wall. The collapse behavior is influenced by the initial
geometrical imperfections which can be depicted solving the problem on
an ovalized deformed shape. The numerical simulation aims to repro-
duce the actual behavior of the structure, validating the theoretical model.
The analyzed case has been also compared with two corresponding steel
strips reinforced thermoplastic pipes (SSRTP), in order to evaluate when
the support of the interlocked carcass is needed for a real design and envi-
ronmental conditions case. The results obtained from the theoretical and
the numerical simulations lead to a remarkable confidence in the models
thanks to a relatively small difference achieved between the outcomes.
This chapter is quoted from Ref. [1].
2.2 Theoretical Models

The well-known differential equation by Timoshenko et al. [2] for the
helical shape of a thin bar is used in this work. The following approach
considers the bending moment on a steel ring having an original imper-
fection. Here, the tangential displacements can be neglected being inter-
ested in the radial problem:
15
d 2uR MR 2 (2.1)
+ u R = −
dϑ 2 EI
where radial displacements uR are considered to be small, is the circumfer-

ential reference of the helical element, M is the bending moment acting on
each cross section, E is the modulus of Young of the material, and I and R
are the moment of inertia of the cross-section and the internal radius of the
ring respectively, as can be seen in Figure 2.1.
The assumption of initial imperfection leads to an extension of Eq.
(2.1), considering that, under this hypothesis, the bending moment in
each cross section can be computed as
M = pR(uR + uR1 cos 2ϑ) (2.2)
for which, uR1 is the initial radial displacement, and p is the uniform pres-
sure applied on the external surface of the bar. The initial displacement is
function of the circumferential reference coordinate (R) and also of the
initial ovality1 so that
uR1 = δ1R (2.3)
The initial ovality is caused by manufacturing tolerances or bend-

ing loads, it must be considered at least equal to 0.002 and this can be
expressed by
D0max − D0min
δ1 = (2.4)
D0max + D0min
uR1
R
y
uR1 x D0max
ϑ
D0min
Figure 2.1 Schematic representation reference systems, initial radial displacements and
initial diameters.
Structural Design of Flexible Pipes 17
where D0max and D0min are the initial maximum and minimum diameters,
respectively, which account for initial displacement as it can be seen in
Figure 2.3.
By replacing Eq. (2.2) into Eq. (2.1), it is possible to get
d 2uR p[(uR + uR1 cos(2ϑ )]R 3

+ u R = − (2.5)
dθ 2 EI
for which the solution results in
uR1 p cos(2ϑ )
uR = (2.6)
pcr − p
where pcr is the critical buckling pressure for a perfect ring.

The ovality is computed considering it as function of the maximum
displacement uRmax. It is expressed in absolute value for cos(2) = 1; thus
Eq. (2.6) assumes the following form:
uR1 p
uRmax = (2.7)
pcr − p
Ovality shows the variation of the minor and major axes. It is com-
puted step by step for each load increment by adding and subtracting,
respectively, the magnitude of the displacements for the corresponding
step from the diameters. Limit value is considered, conservatively, as 20
times the initial one and it is equal to L = 0.04. Besides this value,
the pipe can be considered not suitable anymore for its purpose, being
rough liner crucial elements and easily affected by turbulence due to
internal flow. As it will be shown for both theoretical and numerical sim-
ulation, the pipe can be considered collapsed at L, it already exhibits large
development of ovalization for almost stable pressure value.
If the pressure armor layer is not taken into account in the pipe con-
figuration, as discussed in [1], the critical load for a perfect ring can be
expressed as
3EI eq
pcr = (2.8)
R3
where, EIeq is the equivalent ring bending stiffness of each layer per unit
length of the pipe. For the interlocked carcass it is equal to
EI 2
EI eq = Kn (2.9)
Lp
for which, n is the number of tendons in the layer, Lp is the pitch length,
K is a factor that is function of the lay angle and of the moment of inertia
of the section (for massive cross-section as carcass K = 1) and I2’ is the
smallest moment of inertia of the cross- section which can be computed,
referring to Figure 2.2, as follows:
I3 + I 2 ( I 3 − I 2 )2 + 4 I 32
2
I 2′ = − (2.10)
2 2
To better control the model and to compare results, the actual load is
normalized with respect to the theoretical buckling load that accounts for
imperfections at its threshold L, it is equal to
p
µ= (2.11)
pcr
Finally, the need of a reliable theoretical model suitable for practical

application is used in this work, in order to evaluate the need of the inter-
locked carcass. As it was demonstrated in Bai et al. [3] through experi-
mental and numerical simulations, the critical buckling load is estimated
by summing up the contribution of each layer. Thus, the collapse loads for
both the cross-section geometries of SSRTP considered here are computed
by using the following formulation:
X3' X3
X2'
X2
Figure 2.2 Carcass profile-principal outline.

Ni Nj
pcr = ∑pi =1
i
cr ,steel + ∑p j =1
j
cr ,PE
(2.12)
for which, i and j are the number of steel and PE layers respectively. The
two terms of Eq. (2.12) are derived by Eq. (2.8) for both steel and poly-
meric materials, as follows:
KnEibh3
pcri ,steel = (2.13)
4 L p ,i Ri3
j 3E jf,t I jf
p cr ,PE = (2.14)
R 3j
for which, n is the number of tendons in the layer, and the other
parameters of Eq. (2.13) are previously mentioned. Steel strips are treated
as elastic, while inner and outer PE layers are considered in plastic field.
Physical non-linearities of the plastic material are expressed considering
for each incremental step the update tangential modulus Ej, i, equivalent
moment of inertia Ij and mean radius Rj.
In this section, the behavior of the stainless-steel carcass under exter-
nal pressure is simulated using the commercial finite element software
ABAQUS [4]. Finite element method (FEM)simulation is required to con-
firm theoretical results for the collapse behavior, thus for predicting radial
displacements for each load step when the pipe is affected by hydro-static
pressure. The established model is based on the pessimistic hypothesis
that all the outer sheaths are damaged, and the external pressure acts
directly on the interlocked carcass; thus, the latter must be designed to
carry the full load. The model developed is a 3D ring model which assumes
that the lay angle can be neglected. The simpler 3D ring simulation shows
good agreement comparing outcomes with full 3D pipe model when the
purpose is the computation of the collapse pressure for the carcass layer
[5]. At the same time, this assumption reduces significantly the compu-
tational time. The latter is further reduced if the initial imperfections are
considered symmetrically distributed on the cross-section plane. The
symmetry along the longitudinal direction of the pipe can be taken into
account in order to further reduce the number of operations, so that it is
useful to consider half of the ring, as shown in Figure 2.5.
The chosen geometry is as shown in Figure 2.3, the imported profile

validates the real one since only two of the actual pitches are involved. In
fact, a full corrugate cross-section and two adjacent halves are considered,
as shown in Figure 2.4. The chosen dimensions are listed in Table 2.1
relatively to the cross-section geometry. Material properties and other
parameters needed to the computation are listed in Table 2.2.
The inertia matrix for the abovementioned cross-section dimension is
computed, showing the following results:
L5 φ1
Rtip L1
φ3 L4
L6 R2 R1
L1 L2
L7
φ2 R3 Rtip
φ1
L3
Figure 2.3 Parameterized carcass profile.
Figure 2.4 Full carcass cross section imported.
Table 2.1 Interlocked carcass cross-section parameters.

L1 [mm] 8. 00 R1 [mm] 1.00
L2 [mm] 3. 00 R2 [mm] 1.00
L3 [mm] 9. 00 R3 [mm] 3.00
L4 [mm] 4. 50 Rtip [mm] 0.50
L5 [mm] 10. 00 f1 [°] 60
L6 [mm] 3. 00 f2 [°] 45
L7 [mm] 2. 00 f3 [°] 90
Table 2.2 Interlocked carcass parameters.

n 1
K 1
E [MPa] 200,000
sp [MPa] 600
n 0.3
Lp [mm] 16.00
Rinn [mm] 76.20
t [mm] 6.40
 I11 I12 I13   0 0 0 

   
I =  I 21 I 22 I 23  =  0 44566.60 −370.14  (2.15)
 I I 32 I 33   0 −370.14 217.49 
 31
The initial imperfections are given using a helical path that can simulate
the initial displacements along both x and y directions using sweep com-
mand for two different initial radii that account for ovality equal to L = 0.04.
The cross-section is imported in the xz plane and follows half of the ellip-
tical path. The greater diameter is considered along the y direction, instead
the other lays on the x direction, as it is shown in Figure 2.5.
Z Y
X
Figure 2.5 Carcass model geometry.

Due to the complex shape, proper of the geometry of the carcass and
many possibilities of contact, “General contact” is chosen to simulate the
interactions among the three parts, until the buckling collapse is reached.
For this unbonded condition, “Frictionless” tangential behavior and “Hard
contact” normal behavior with “Allow separation after contact” are cho-
sen. The latter is defined by (p-h)model, which relates p: contact pressure
among surfaces and h: overclosure between contact surfaces respectively.
When h < 0, it means no contact pressure, while for any positive contact h
is set equal to zero [6].
Also, the plastic behavior of the material is taken into account to verify
if the collapse happens in elastic or plastic field and if the hypothesis of
considering only elastic behavior in the theoretical model makes sense.
As it is shown in Figure 2.6, the material properties for the carcass layer
consider a linear elastic behavior, following the Hooke’s law during the first
stage then again, a linear behavior due to the plastic tangent modulus for
simulating high strain against low stress increments in the plastic field,
which accounts for an isotropic hardening law [7].
The external pressure is considered as constant along the width, in z
direction and applied directly on the external surface. The kinematic is
governed by the boundary conditions that must mainly avoid rigid body
displacements. The possibility of assigning symmetric boundary condition
with respect to the xz plane is exploited in order to allow displacement in x
direction (U1) at the bases of the ring, which also fix the pitch as constant
during the simulation. For simulating the occurrence of the local bucking
of the ring, the only allowed displacements in the middle of the ring are in
y direction (U2), as depicted in Figure 2.7.
800
700
600
500
400
σ
300
200
100
0
0 0.01 0.02 0.03 0.04 0.05 0.06
ε
Figure 2.6 Steel strain-stress relationship.

Y
Z
X
Figure 2.7 Load and boundary conditions.
A dynamic implicit analysis is employed because of the necessity of

capture the variation of the stiffness changes step by step in case of non-
linearity due to frictionless contact. Non-linear geometry is also taken
into account in order to simulate large deformations proper of collapse
issue. The dynamic analysis developed, is justified by the fact that the ratio
between kinetic energy (ALLKE) and strain energy (ALLSE) for the whole
model is kept below 0.1 until the collapse point, in Figure 2.8 kinetic and
strain energies are plotted against the dimensionless load.
In this model, C3D8R (eight-node continuum linear brick elements
with reduced integration and hourglass control) is used for the three parts
7000
Kinetic energy
6000
Strain energy
5000
4000
Energy
3000
2000
1000
0
0 0.2 0.4 0.6 0.8
µ
Figure 2.8 ALLKE/ALLSE versus dimensionless load.

Z Y
X
Z X
Figure 2.9 Carcass model mesh details, cross-section, and surfaces.
as can be seen in Figure 2.9. This type of components matches very good
the purpose of the model, in fact these can be used for linear and complex
non-linear analysis producing high accuracy results when contacts, plas-
ticity and non-linear geometry are considered [8].
2.3 Comparison and Discussion

The input data for the design case of a flexible pipeline is the internal
diameter which is the same for both steel strip reinforced thermoplastic
pipes and the case improved against external pressure with stainless steel
carcass. For both numerical and numerical models, the chosen reference
surface is the outermost and both displacement along x and y directions
are analyzed. For the numerical simulation, the central body is chosen as
reference, being the only full profile in the model. For the numerical case,
displacements could differ one to the other because it simulates the actual
geometry as it is shown in Figures 2.10 and 2.11 While for some theoret-
ical limits, they will be coincident in both directions. By simulating the
mechanical behavior of the structure, it is possible to catch the buckling
pressure.
Firstly, using the parameters listed in Eq. (2.14) for Eq. (2.10), it is
possible to carry out the smallest moment of inertia of the cross-section:
I2’ = 185. 41 mm4. Then, from Eq. (2.9), the equivalent stiffness per unit
length is calculated EIeq = 2,317,671.93 MPa mm3 if the mean radius is
computed as follows: R = Rinn + t = 79.40 mm. Finally, the critical buckling

pressure for a ring without ovality is obtained and it is equal to: pcr =
13.89 MPa. This result is normalized by the critical load at L = 0.04
as shown in Eq. (2.11), and it is plotted in Figure 2.14, which shows the
sudden collapse when the structure reaches the collapse pressure.
When the imperfection is considered, the initial value of ovalization
taken into account is1 = 0. 002, as minimum requirement, so that the ini-
tial displacement can be computed using Eq. (2.3), and it is equal to uR1 =
0.1652, referring the computation to the external radius. The outcomes
of the theoretical model can be finally plotted, as shown in Figure 2.14,
against the dimensionless load and for the ovality computed for each load
increment as follows:
D max − D min (2.16)

δ = max
D + D min
where, Dmax and Dmin are the maximum and minimum diameters uploaded
step by step from the results of the of the displacements, along x and y
directions, respectively.
In the same way, outcomes for finite element model are computed
extracting U1 and U2 and they are plotted in Figure 2.12. As previ-
ously said, it can be seen that the external surface displays different
magnitude of displacements along x and y directions, as shown in Figures
2.10 and 2.11.
As it was shown by Gay [7], the pre-buckling behaviors match one to
each other with very low error. In Figure 2.14, it can be observed that for
small value of external pressure the ovality increases linearly. In this phase,
the very small difference between theoretical and numerical models can be
caused by the fact that the theoretical model regards as an equivalent ring.
Vice versa, the actual geometry shows gaps among the different parts, so
it is obvious that results show a slightly different about stiffness. When
the pressure increases, a non-linear trend is illustrated for both models.
Numerical results show wider ovality for the same load, compared to ana-
lytical outcomes because of theoretical limits that consider as critical
load the one for the case with no imperfections. As it was expected,
the comparison between results makes sense in order to understand the
pre-buckling behavior in terms of displacements and not in terms of col-
lapse load. In fact, from Figure 2.12, the relevant output that comes out, in
terms of pressure, is an error equal to 36% between theoretical and numer-
ical outcomes when dealing with imperfections Table 2.3.
U, U1
+4.023e+00
+3.352e+00
+2.682e+00
+2.011e+00
+1.341e+00
+6.705e+04
Y +1.149e+04
−6.703e–01
−1.341e+00
−2.011e+00
−2.682e+00
Z X −3.352e+00
−4.022e+00
Figure 2.10 U1 displacements.
U, U2
+3.515e+00
+3.202e+00
+2.888e+00
+2.575e+00
+2.262e+00
+1.948e+00
+1.635e+00
Y +1.321e+00
+1.008e+00
+6.944e+00
+3.810e–01
+6.758e–02
–2.459e–01
Z X
Figure 2.11 U2 displacements.

0.040
Theoretical
0.035 FEM
No imperfections
0.030
0.025
δ
0.020
0.015
0.010
0.005
0.000
0 0.2 0.4 0.6 0.8 1 1.2
μ
Figure 2.12 Ovality versus dimensionless load.
Table 2.3 Collapse pressures.

Model Collapse pressure [MPa]
No imperfections 13.89
Theoretical 13.19
FEM 8.43
Moreover, here, it can be confirmed that, for a value of ovality

equal to 4%, the external pressure no longer increases, while the ovality
keep to sharply rise, so that it can be confirmed that L can be regarded as
point of instability.
Eq. (2.6) in its original form, considers critical load for a ring without
imperfection (pcr), in fact, the collapse is reached asymptotically for
the latter value. In order to give reason to the similitude in terms of
buckling pressure, the comparison of a series of both numerical and
theoretical simulations is needed, keeping the bending stiffness of the
cross-section constant and varying the radius of the pipe. The resolution
of new pcr exhibits a valid formulation which guarantees theoretical results
closer to the actuals and so to the FEM outcomes, which works in terms of
both displacements and collapse pressure.
The available theoretical formulation which is based on thin-wall

hypothesis gives results with lower error for high values of D/t. On the
contrary, the gap between theoretical and numerical outcomes rises as
the ratio under analysis decreases, which is captured in Figure 2.13. Thus,
from this behavior, the consequence of considering zero error for high D/t
ratios and justify the following assumption for the theoretical model.
An overall of eight models are analyzed for the same carcass cross-
section and material, while different D/t ratios. In Figure 2.14, the collapse
loads are plotted against different geometries when they reach the ovality
limit. As it was expected, the critical buckling loads decreases as the tube
diameters increase. Moreover, it is possible to see the asymptotic behavior
between numerical and theoretical results for growing D/t ratios.
With that being said, the hypothesis of considering zero error for D/t
= 30, which is the widest geometry considered. The choice to stop the
computation at that value is made because it is rare to contemplate higher
diameter configurations for deep water environments. Moreover, even if
the error goes to zero asymptotically, as shown in Figure 2.15, this
assumption wants to be a conservative suggestion since considering it
for higher D/t standards will lead to a higher collapse pressure for the case
under analysis.
Once the error is eliminated, the actual collapse loads accounting for
imperfections can be obtained, and their behavior is shown in Figure 2.16.
0.040
4in
0.035 4in
4.5in
0.030 4.5in
5in
0.025 5in
5.5in
Ovality
0.020 5.5in
6.5in
0.015 6.5in
7in
0.010 7in
7.5in
0.005 7.5
0.000
0 20 40 60
External pressure
Figure 2.13 Ovality versus load for different geometries, where dashed lines stand for
numerical results and continuous line stand for theoretical results.
50
45 Theoretical
FEM
40
No imperfections
35
Collapse load
30
25
20
15
10
5
0
15 20 25 30
D/t
Figure 2.14 Critical loads versus dimensionless diameters.
0.70
Theoretical
0.60
Theoretical modified
Theoretical-Trend Line
0.50
0.40
Error
0.30
0.20
0.10
0.00
15 20 25 30
D/t
Figure 2.15 Error trend.
The extracted polynomial trend line depicts the guideline for the
new theoretical model, and has the following formulation:
2
 D D
pcr = 0.0512   − 3.6206 + 66.3 (2.17)
 t t
Eq. (2.17) is an estimation of the collapse pressure valid for the initial
imperfection considered. It shows conservative results and it obviates the
50
Theoretical
45
FEM
40
No imperfections
Collapse load 35
New theoretical
30 New theoretical-Trend line
25
20
15
10
5
0
15 20 25 30
D/t
Figure 2.16 Critical load comparison for all the model established versus dimensionless
diameters.
employment of the buckling load for the perfect ring, in the theoretical
model, which leads to underestimation of collapse conditions.
In order to get a behavior closer to the actual, Eq. (2.7) is developed
again exploiting Eq. (2.16) as theoretical derivation to get the critical buck-
ling load, which, for the case with D = 6 in, produces pcr = 9.12 MPa. As
previously done, dimensionless load and ovality are plotted for both theo-
retical and numerical models, as shown in Figure 2.17.
The modified theoretical model exhibits an acceptable error equal to e =
2.58%, and it obviates to the previous discrepancy simulating the behavior
in terms of load in addition to displacements.
Finally, the theoretical prediction of local stresses is precise if consider-
ing pipes or rings geometry but for the stainless-steel interlocked carcass,
due to the intricate profile of the cross section, while affordable results
could be obtained from FEM analysis. The actual stress magnitude is not
constant along the structure; in fact, dissimilar values are reached in two
different points under analysis. In Figure 2.18, it is reported the relation-
ship between stress and strain for the critical point between the two, at
limit value of ovality. It is important to focus the attention on the highest
stress value reached which is equal to about 400 MPa, which is far below
the proportional limit stress equal to 600 MPa. Thus, it is possible to delib-
erate on the right assumption done for the theoretical model of consider-
ing elastic material properties only.
In this section, the results about the collapse study for the interlocked
carcass are compared with the one for a steel strip reinforced thermoplastic
0.040
New theoretical
0.035
FEM
0.030
0.025
δ
0.020
0.015
0.010
0.005
0.000
0 0.2 0.4 0.6 0.8
μ
Figure 2.17 Results comparison for FEM and modified theoretical model.
700
600
True stress
500 Proportional limit state
400
σ
300
200
100
0
0 0.0002 0.0004 0.0006 0.0008 0.001
ε
Figure 2.18 Stress-strain relationship.
pipes, in order to understand until which water depth there is no need

of further reinforcement. The SSRTP overall design is made by a variable
number of layers of thin steel strips wrapped two by two at opposite lay
angles, surrounded by an inner and outer HDPE layer.
25
20
15
σ
10
0
0 0.05 0.1 0.15
ε
Figure 2.19 HDPE stress-strain relationship.
Steel strips are considered to have an elastic behavior with a modulus of

Young equal to 206,000 MPa. While HDPE layers properties account for
an elastic-plastic behavior, where the modulus of Young is equal to 930
MPa and the yielding stress equal to 6.52 MPa, as shown in Figure 2.19.
The first case under analysis, named SSRTP-1, presents internal and
external HDPE layers and in between four layers of six steel strips each.
Geometrical parameters are listed in Table 2.4, where t are the thicknesses
of the plasic layers, a is the winding angle between the strip and the pipe
axes, and b and h are the strip cross-section dimensions.
The second case proposed, named SSRTP-2, is an improved configura-
tion to withstand higher tensional loads and external pressure. It accounts
for two more steel strip layers with higher lay angle in order to provide
higher strength in the radial direction. The overall design shows the
innermost and outermost HDPE layers, and six steel strip layers, num-
bered in outward direction, with different thicknesses and cross-section
Table 2.4 SSRTP-1 parameters.

tin [mm] 6.00
tout [mm] 4.00
a [°] 54.7
h [mm] 0.50
b [mm] 48.00
dimensions. In Table 2.5, all the needed geometrical parameters are listed,
named as previously done.
The solution derived from the already verified theoretical model in
Bai, Yong, et al. [3] is applied for both the design cases and results are
reported in Table 2 . 6. SSRTP-2 shows collapse pressure highly improved
compared to SSRT-1, while, as it was expected, it is still much lower than
the one that considers the interlocked carcass in the design. Being the
strips weak elements and mostly suitable for withstanding tensile and tor-
sional loads, it is necessary to underline that the contribution of the
plastic layers cannot be negligible, being equal to the 71% and 32% of the
overall resistance, for SSRT-1 and SSRT-2, respectively.
Considering that for each meter of water depth, the corresponding
external pressure acting is equal to 0.01MPa; it is possible to deliberate
that a 6-in inner diameter flexible pipe with interlocked carcass is suitable
above 900-m depth. On the other hand, very low strength against exter-
nal pressure for steel strip reinforced thermoplastic pipes is revealed
from the analysis compared to the reinforced cross-section, which means
that they are mostly suitable for cases of very shallow water, being the
cross-section not massive. In particular, SSRTP-1 and SSRTP-2 are appro-
priate until 20 and 37 m of water depth. Outcomes do not consider safety
factor which must be taken into account considering the environmental
conditions at the site.
Table 2.5 SSRTP-2 parameters.

Rin [mm] 76.20 Rout [mm] 90.10
tin = tout [mm] 5.00 b [mm] 48.00
a1 = a2 [°] 73 a5 = a6 = a3 = a4 [°] ±54.7
h1 = h2 [mm] 0.45 h3 = h4 = h5 = h6 [mm] 0.75
n1 = n2 3 n3 = n4 = n5 = n6 6
Table 2.6 Collapse pressures.

Model Collapse pressure [MPa]
SSRTP-1 0.20
SSRTP-2 0.37
Reinforced 9.12
2.4 Conclusions
Throughout this report, the collapse behavior of an interlocked carcass
is simulated, and results are investigated theoretically and numerically.
A variation of a primary adopted theoretical model is proposed which
may be of interest for practical applications. In order to do so, a series of
numerical models for the calibration were needed, leading to an empirical
formulation. The latter is valid in terms of both pre-buckling and collapse
conditions for a defined value of initial ovalzation, so that the critical pres-
sure is computed neglecting the friction between layers for both models.
Outcomes are compared with a steel strip reinforced thermoplastic pipes
under the same requirements, in order to understand when the reinforce-
ment is required. This study can provide support for factory engineers
due to the accurate and reliable results that show very small difference
between numerical and theoretical models.
References
1. Fergestad, D, Lotveit, S.A., ‘Handbook on Design and Operation of
Flexible Pipes[Z], NTNU, 4Subsea and MARINTEK, 2014.
2. Timoshenko, S. P., and Gere, J. M., 1961, Theory of Elastic Stability,
McGrawHill International Book Company, Inc., New York.
3. Bai, Yong, et al. “Buckling stability of steel strip reinforced thermoplas-
tic pipe subjected to external pressure.” Composite Structures 152 (2016):
528-537.
4. ABAQUS. 2014. User’s and theory manual version[Z]. 2014.
5. American Petroleum Institute, 2002, API recommended Practice 17B,
Information Handling Services, API, Washington D.C.
6. An C, Duan M, Toledo Filho RD, et al. Collapse of sandwich pipes with
PVA fiber reinforced cementitious composites core under external pres-
sure. Ocean Eng 2014;82:1–13.
7. Neto, Alfredo Gay, and Clóvis de Arruda Martins. “A comparative wet
collapse buckling study for the carcass layer of flexible pipes.” Journal of
Offshore Mechanics and Arctic Engineering 134.3 (2012): 031701.
8. Kim T S, Kuwamura H. Finite element modeling of bolted connections
in thinwalled stainless steel plates under static shear[J]. Thin-Walled
Structures, 2007, 45(4): 407-421.
3
Structural Design of High
Pressure Flexible Pipes of
Different Internal Diameter
3.1 Introduction References

Accurate prediction of the pipe’s burst pressure is an important subject
in the integrity assessment of oil and gas transmission application. As we
know, flexible pipes are manufactured with several layers; a typical struc-
ture is shown in Figure 3.1. From the inside to the outside, it consists of
the following:
Carcass: an inner carcass with S profile providing resistance to collapse.
Internal pressure sheath: an internal plastic sheath providing fluid leak
proofness. Pressure armor: a Zeta-shaped metallic layer wound at short
pitch providing support for the internal plastic sheath and part of the resis-
tance to the radial forces.
Anti-wear layer: Prevent contact and friction between metal layers
and can’t resist loads.
Tensile armor: an even number of metallic layers wound at long pitch
providing resistance to the axial forces and also partly to the radial forces.
Outer sheath: an external plastic sheath for protection against corrosion.
Due to the function of each layer is independent, the factory makes
different layers combination to adapt to its own serving condition.
To the date, there have been published some works focused in burst analy-
sis of flexible pipes, both numerical and analytical models are studied. For
the study of burst pressure, Fernando et al. [1] studied the pressure layer
subjected to internal pressure using two models: a 2D axisymmetric one,
including only the internal pressure loading, and a three-dimensional (3D)
model, including pressure layer manufacturing simulation and also the
internal pressure layer. Neto et al. [2] proposed a linear analytical formu-
lation assuming that the pressure armor behaves exactly as a thin- walled
cylinder considering the equivalent thickness of the profile of the pressure
35
Figure 3.1 Structure of flexible pipe.
armor. A most simplified analytical model was presented by Oliviera et al.

[3] to estimate the burst strength considering its axial and circumferential
deformation. Zhu et al. [4] proposed an average shear stress yield criterion,
and compared the predicted values with the experimental results. It was
found that the predicted values of Mises criterion were in good agreement
with the actual values of the materials. Chen [5] analyzed the burst failure
mechanism of flexible pipe by two-step method, first step is to study the
pressure armor’s behavior and second step is to study the tensile armor
after the pressure armor fails, and obtained a simple burst pressure formula
by using ring compression theory. Furthermore, Fergestad [6] used the
curved beam theory to consider the resistance of armor layers subjected
to internal pressure, and used the ultimate strength as the failure criterion.
Both of them use a simple radial and axial equilibrium relationship and
only consider the resistance of the axial stress of the helix to internal pres-
sure. On the basis of researches, Kebadze [7] deduced and summarized the
theoretical solution models of different kinds of layers and combined them
into various solutions to carry out the theoretical solution of the model.
Concerning the structural analysis of flexible pipes subjected to inter-
nal pressure, several researchers made their contribution to this field. In
Structural Design of High Pressure Flexible Pipes 37
this chapter, according to the Kebadze’s research, the theoretical analysis

considers the material nonlinearity of armor layers and geometrical non-
linearity in the prediction of burst. Verifying the correction of the theoret-
ical model by Abaqus, it is promising to design cross-section of pipes by
mathematical software conveniently.
Industry practice requires several types of flexible riser configurations
typically used in conjunction with floating production/loading systems.
Figure 3.1 illustrates these six main types of flexible riser configurations.
The feasible configurations differ in the use of buoyancy modules and the
methods of anchoring to the seafloor. Configuration design considerations
include several factors such as water depth, host vessel access/hang-off
location, field layout such as the quantity and types of risers and mooring
layout, environmental data, and host vessel motion characteristics.
3.2 Analytical Models

Before the theoretical deduction, several simplified hypotheses are made
ahead:
1) All materials are homogeneous and isotropic and have linear

elastic behavior.
2) The geometrical deformations are small.
3) Thickness variations in the layers are assumed to be uniform
in each layer.
4) There is no contact between adjacent tendons in the same
helical layer.
5) The layers remain in contact.
6) No gap between adjacent layers is allowed in the unstressed
(initial) state.
7) All layers at all cross-sections present the same twist per unit
of pipe length and the same elongation.
3.2.1 Cylindrical Layers

Cylindrical components in flexible pipes are mainly used as barriers to
liquid. Internal tubes transport the hydrocarbons while as external sheaths
protect the steel from sea water and corrosion. Sometimes, they are used
at an intermediate position between steel layers in order to reduce wear
and friction. The thin tube theory with the consideration of radial strain is
applied to the cylindrical layers.
The internal strain energy is
1
U=
2 ∫ ∫ ∫ (ε σ + ε σ
1 1 2 2 + ε 3σ 3 + τ 12γ 12 )dv (3.1)
where ε1, ε2, ε3 and γ12 are strain components, σ1, σ2, σ3, and τ12 are corre-
sponding stress components. The related parameters definition is shown
in Figure 3.2.
Considering the global elongation, torsion, and volume change of the
cylindrical components under pressure, the work of external forces is
expressed by
W = PiΔVi – PoΔVo (3.2)
where Pi, Po, ΔVi, ΔVo denote internal pressure and external pressure and
the changes in internal volume and in external volume, respectively.
Using the principle of virtual work, which implies equating the variation
of the external work to the variation of the internal energy, the equilibrium
equations are derived.
y' y
L ∆U
x'
∆φ x
R+∆R
γ12 ε2
ε3
ε1 z
R
t+∆t
t
Undeformed tube
Deformed tube
σ2
σ3
σ2
σ1
Figure 3.2 Cylindrical mathematical parameters definition.

δU = δW (3.3)
To calculate this equation, the matrix form of stiffness equations for
the thin tube theory is derived:
 uL 
 
 L  
 (2G + λ )A λA λA  2 2 
0  ∆ϕ   N + π Pi Ri − π Po Ro 
   
 0 GJ 0 0  L T 
 = 
 λA 0 (2G + λ )A λA  ur   2 Piπ Ri R − 2 Poπ Ro R
    −Pπ R t − P π R t 
 λA 0 λA (2G + λ )A   R 
   i i o o
 ut 
 t 
(3.4)
3.2.2 Helix Layers

The helix component is main force resistant structure of flexible pipe,
which has an important influence on the mechanical properties of pipe.
For pressure armor, it is necessary to simplify the section. In order to keep
the radial continuity, the same thickness of the original section is used. In
addition, the axial stiffness of the helix is mainly considered, so keep the
area same as the original section. The exact formula for axial strain of
a helical element was derived by Knapp [8]. In order to obtain a linear
equilibrium equation of helical tendons, the exact formula is simplified as
uL u ∆ϕ
ε1 = cos 2 α + r sin 2 α + R sin α cosα (3.5)
L R L
The radial strain is derived as:
ut
ε2 = (3.6)
t
Therefore, the internal work can be expressed as these two parts above
adding together:
nEA
U=
2(1 − v 2 )cos a ∫ (ε 2
1 )
+ 2vε1ε 2 + ε 22 dz (3.7)
Similarly, the stiffness matrix of helix layers can be derived:
 u 
L
 
L  
 k11 k12 k13 k14  2 2 
  ∆ϕ   N + π Pi Ri − π Po Ro 
 k21 k22 k23 k24  L   T 
  u  = 

k31 k32 k33 k34
 R   2 Piπ Ri R − 2 Poπ Ro R 
 k41 k42 k43 k444   R   − Piπ Rit − Poπ Rot 
 u 
 t 
 t 
(3.8)
3.2.3 The Stiffness Matrix of Pipe as a Whole Helix Layers

Subjected to the axisymmetric load, all layers at all cross-sections present the
same twist per unit of pipe length and the same elongation see Figure 3.3.
R
Helical Strip α
L
S
S' α’
Pipe Centreline θ·R

UL
θ · (R+Ur) + R · ∆φ
(a) (b)
Figure 3.3 Helix mathematical parameters definition.

It will be assumed that all layers of the pipe are numbered consecutively
from 1 (the innermost layer) to N (the outermost layer), where N is the total
number of layers of the pipe. Combine the stiffness matrix together, Eqs.
(3.9) to (3.12) are derived:
N N N N
 (uR ) j (ut ) j 
∑j =1
(k11 ) j
uL
L
+ ∑
j =1
(k12 ) j
∆ϕ
L
+ ∑
j =1
(k13 ) j
 R j
+ (k14 ) j
t j 
=N + ∑ (π P R − π P R )
j =1
i i
2
o
2
o
(3.9)
N N N
 (uR ) j (ut ) j 
∑j =1
u
(k21 ) j L +
L ∑ j =1
(k22 ) j
∆ϕ
L
+ ∑ (k
j =1
)
23 j
Rj
+ (k24 ) j
tj 
=T
(3.10)
uL ∆ϕ (uR ) j (ut ) j
(k31 ) j + (k32 ) j + (k33 ) j + (k34 ) j = 2π ( Pi ) j Ri R − 2π ( Po ) j Ro R
L L Rj tj
(3.11)
uL ∆ϕ (uR ) j (ut ) j
(k41 ) j + (k42 ) j + (k43 ) j + (k44 ) j = −π ( Pi ) j ( Ri ) j t j − π ( Po ) j ( Ro ) j t j
L L Rj tj
(3.12)
Still, lacking of equations for solving all unknown variables, the consis-
tent in the radial direction is added to solve the problem:
uR,j – uR,j+1 + (ut,j + ut,j+1)/2 = 0, (j = 1,2,…,Nt − 1) (3.13)
where j denotes the layer number. For some applications, the calculated
contact pressure between layers may be found negative. Treating such a case
requires a change of unknowns. The contact pressure is known and equal
to zero and the gap between the two layers becomes the new unknown.
3.2.4 Blasting Failure Criterion

In order to use the above method to predict the burst pressure of flexi-
ble pipe, it is necessary to introduce the failure criterion into the program.
API17J [9] stipulates that the bearing capacity of the structure can be
determined by the yield strength of the material or 0. 9 times of the ulti-
mate tensile strength. In order to simulate the properties of steel used in
practice, the Ramberg-Osgood model expressed in Eq. (3.14) is used to
describe the non-linear stress-strain relationship of the armor layers. For
this model, the pressure armor’s ultimate strength is 630 MPa.
σ 3 σ
n−1

ε = 1 +  (3.14)
E  7 σ y 
Where σy = 600MPa, E = 207GPa, n = 13.

Specifically, the total load is decomposed into a limited increment for
loading. Attention should be paid to the step size of the incremental step
is small enough to ensure the convergence of the results. In this chapter,
Mises stress is used as failure criterion. When Mises stress reaches ultimate
strength during loading, it is considered that the pressure armor failures.
3.3 FEA Modeling Description

In this chapter, ABAQUS model is used to verify the correction of theoret-
ical model. The finite element model consists of a Z-type self-locking pres-
sure armor layer, two layers of tension armor layer with opposite winding
angles and other polymer layers. Specific size information is shown below
in Table 3.1, Table 3.2.
As the structure of the model is quite complex and there are a lot
of contact between layers, implicit static analysis is not only time-
consuming but also brings a lot of convergence problems. Therefore, this
chapter uses dynamic explicit method to carry out the analysis. Due to the
non-bonding between the layers of flexible pipe, contact and slip might
occur between the layers during the loading process and self-contact will
occur in the pressure armor layer, which makes the contact situation more
complex. Therefore, the “All with itself ” algorithm is used to simulate
contact in the ABAQUS model. This algorithm can automatically identify
contact pairs and can consider the possible layer separation. The contact
Table 3.1 Geometric and material parameters of FEM.

Layer name Parameters
1. Pressure sheath Internal radius: 76 mm
Thickness: 6 mm
Young modulus: 1,040 Mpa
Poisson ratio: 0.45
Thickness: 10 mm
Number of wires: 1
2. Pressure armor Profile: shown in Figure 3.4
Winding angle: +88.8°
Poisson ratio: 0.3
Thickness: 1.5 mm
3. Anti-wear tape Young modulus: 301 Mpa
Poisson ratio: 0.45
Thickness: 5 mm
Number of wires: 46
Profile: 5 mm × 11 mm
4. Tensile armor Winding angle: +54.7°
Poisson ratio: 0.3
Thickness: 1.5 mm
5. Anti-wear tape Young modulus: 301 Mpa
Poisson ratio: 0.45
Thickness: 5 mm
Number of wires: 47
Profile: 5 mm × 11 mm
6. Tensile armor Winding angle: −54.7°
Poisson ratio: 0.3
Thickness: 4 mm
7. Outer plastic Young modulus: 1,040 Mpa
Poisson ratio: 0.45
normal direction is set by hard contact, and the tangential direction is set
by non-friction.
In order to control the boundary conditions of the model conve-
niently, two reference point (RP-1 and RP-2) are set at the center of the
Table 3.2 Pressure armor layer geometrical properties.

Internal diameter 164 mm
Cross section area (profile) 102.53 mm2
Minimum moment of inertia (profile) 79.67 mm4
Pitch 11.87 mm
Thickness of the profile 10 mm
Length of the profile 18.50 mm
Figure 3.4 Pressure armor cross section profile.
RP-2
Figure 3.5 Kinematic Coupling at RP-2.
cross-section at both ends of the model. Kinematic coupling which is

shown in Figure 3.5 is set up between all degrees of freedom of each layer
in the cross-section with the reference point. The constraints on the end
face are imposed at two reference points (RP-1 and RP-2). All degrees of
freedom on RP-1 are constrained, and the torsional direction of RP-2
along z axis is also constrained.
In order to accurately obtain the thickness change of each layer and

the stress change along the thickness direction, solid elements are used
to simulate each layer of the model. C3D8I element is used in cylindrical
layers; this kind of element is a non- conforming mode element. It can
overcome the self-locking of shear force in the first- order element with
complete integration. It can obtain high precision results by using fewer
elements, but it is sensitive to the distortion of the element. Therefore, all
the spiral armor layers use C3D8R cell. More refined mesh partition can
avoid hourglass problems that may exist in this kind of cell. The detailed
meshing of the model is shown in Figure 3.6.
The model used in this chapter is compared with the experimental data
of public tables to verify the feasibility of the FEA model. The test was
recorded in a paper published by J. A. Witz [10]. The most common way to
evaluate whether the simulation produces the correct result is to study the
energy in the model. ALLKE is the kinetic energy of whole model, while
ALLIE is a summation of all internal energy quantities. As a general rule,
the kinetic energy of the deforming material should not exceed a small
fraction (typically 10%) of its internal energy throughout most of the
process. As shown in Figure 3.7, it can be seen that during the simula-
tion process, the ratio of kinetic energy to internal energy is less than 10%
in the general. This shows the correctness of the simulation. In addition,
artificial strain energy associated with constraints used to remove singular
modes (such as hourglass control), and with constraints used to make the
drill rotation follow the in-plane rotation of the shell elements. The artifi-
cial strain energy is below approximately 10% of the total internal energy,
indicating that hourglass is not a problem. In order to reduce the influence
Z X
Figure 3.6 Meshing of the flexible pipe model.

Kinetic energy/Internal energy

0.12
0.1
0.08
0.06
0.04
0.02
0
0 20 40 60 80 100
Internal pressure (MPa)
Figure 3.7 Kinetic energy/internal energy curve.
Artificial strain energy/Strain energy

0.12
0.1
0.08
0.06
0.04
0.02
0
0 20 40 60 80 100
Figure 3.8 Artificial strain energy/strain energy curve.
of the hourglass energy, the ratio of artificial strain energy to strain energy
is extracted as shown in Figure 3.8. With the loading process, it can be seen
that the ratio is smaller than 5%, which indicates the hourglass energy
is quite small and the result of the simulation is reliable. But, when the
pressure occurs to 80 MPa approximately, the curve rises sharply which
indicates that the pressure might reach to the ultimate strength and the
model appears the large deformation. From the result of Figures 3.7 and
3.8, it can be regarded that the FEA model is correct.
3.4 Result and Discussion

It should be noted that this chapter is limited to prediction of pressure
armor stresses during pipe operation only, and that the residual wires
X
Y
Z
Figure 3.9 Model of pressure armor layer.
stresses from manufacturing are disregarded and not taken in consider-

ation. Analytical models follow a series of restrictive assumptions while
finite element models are not limited to them. Thus, there will be a certain
error between these two analysis methods. Theoretical models and finite
element models are carried out respectively to exam the behavior of a flex-
ible pipe under internal pressure. The numerical results are selected from
the middle span of the flexible pipes since the stress near the reference
points is inapplicable due to boundary conditions.
Pressure armor layer which shown in Figure 3.9 is the main force-
resistant structure of pipe under internal pressure. Figure 3.10 shows the
change of Mises stress of pressure armor in this section with the increase
of internal pressure. In order to express the stress changes of its Z-section
clearly, the range is fixed at 600 MPa, and red markers are used when it
is higher than this value. It can be seen from the figure that the stress of
Z-section decreases gradually from bottom to top, which accords with the
general law of stress variation along the thickness direction of the section
under internal pressure. Because of the shape of Z-section, the maximum
stress points often occur at the left ends. With the increase of internal pres-
sure, the inner part of the section yields first and then expands gradually
to the outside. When the internal pressure is between 70 and 80 MPa,
the whole section yields, the result is similar to the conclusion of the
energy analysis.
Figure 3.11 shows the relationship between maximum von Mises
stress of pressure armor and the internal pressure. As can be seen, before
the yielding of the pressure armor, the results of the two methods are in
good agreement. After the yielding occurs, the errors of the two methods
10MPa 20MPa
30MPa 40MPa
50MPa 60MPa
70MPa 80MPa
Figure 3.10 Mises stress of Z-shaped section.
begin to appear. The growth rate of the Mises stress obtained from FEM is
slower than that from the theory. The main reason is that the self-locking
of the pressure armor in the finite element method will cause the redistri-
bution of stress in the section, which makes it difficult to predict the stress.
While the theoretical method doesn’t take this change into consideration.
Abaqus Matlab
900
Maximum von Mises stress 800
700
600
500
(MPa)
400
300
200
100
0
0 20 40 60 80 100
Figure 3.11 Pressure-maximum von Mises stress curve.
Abaqus Matlab
0 20 40 60 80 100
0
Axial displacement (mm)
−1
−2
−3
−4
−5
−6
−7
−8
Figure 3.12 Pressure-axial displacement curve.
After yielding, the maximum von Mises stress of two models keep
increasing slowly and the Matlab’s reaches to the ultimate strength firstly.
Generally speaking, the difference between the two methods is not very
big. The analytical result is 71 MPa while numerical result is 74 MPa when
the maximum von Mises stress reaches to the yield strength. As for getting
to the ultimate strength, the analytical result is 75 MPa while the numer-
ical result is 80 MPa.
If the deformation of pipe is not restricted, the pipe cannot function
normally. Figures 3.12 and 3.13 show the relationship between the axial
displacement and the internal pressure and between the radial displace-
ment and the internal pressure, respectively. The axial displacement takes
the value of the coupling point RP2, and the radial displacement takes the
Abaqus Matlab
0.7
Radical displacement (mm)
0.6
0.5
0.4
0.3
0.2
0.1
0
0 20 40 60 80 100
Figure 3.13 Pressure-radial displacement curve.
value of the middle node of the Z-section selected. It can be found that
the theoretical curve is in same trend with the finite element curve.
While, FEM always lags behind the theoretical curve. This is because
the deformation of the pressure armor will be limited by its self-
locking structure, but this effect is not considered in the theoretical
model. Therefore, in the subsequent deformation development, the
finite element method needs more internal pressure to obtain the same
deformation as the theoretical model. When the internal pressure contin-
ues to increase and reaches to the ultimate strength, the axial displacement
and radial displacement begin to increase sharply. This also indicates that
the pressure armor is the main internal pressure resistant structure. When
it fails, it is considered that the pipe will soon fail.
3.5 Design
It can be found that the results in two models are in good agreement and
the theoretical model has high accuracy in predicting the burst pressure of
pipe. The results may be of interest to the manufacture factory engineers.
It is convenient to design the structure of pipes by using the theoretical
model under different internal pressure and different given radius. Other
most simplify formulas are proposed in Handbook [6] which are shown
in the below. The contribution of tensile armor to burst pressure resistance
is expressed by
ttot
ph = Ff σ u sin 2 a (3.15)
R
where ttot is the total thickness of the double tensile armor layers, R is the
mean radius of layer, a is the winding angle, and σu is the ultimate tensile
strength of the layer. The contribution of the tensile armor to end cap pres-
sure resistance is expressed by
R
pa = 2 t F σ cos 2 a
2 tot f u (3.16)
Rint
where Rint is the inner radius of the layer. The contribution of the pres-
sure armor to burst pressure resistance is expressed by
NP
tj
pp = ∑RF σ
j =1
fj uj (3.17)
where tj denotes the thickness of pressure spiral with layer number j and
R is the mean radius of the Np pressure layers, respectively. The fill factor
Ffj for pressure spiral wire layer j. The total hoop pressure resistance is the
obtained by summing the contribution from each layer as
phoop = pp + ph (3.18)
The burst pressure is then given by the smallest of phoop and pa:
pb = min(phoop, pa) (3.19)
It can be seen that R, t tot, and a take most significant roles in

influencing the burst pressure. In the design procedure, we can adjust
these parameters to meet the design requirement. In shallow water,
steel strip reinforced thermoplastic pipe (related materials are shown in
Table 3.3) is often used. Bai [11] has done some researches on the mechan-
ical responses of SSRTP under pressure loads. In order to design structure
of pipes economically and safely, this chapter uses two theoretical models
to predict burst pressure in more models (shown in Table 3.4) to illustrate
the procedure of designing under different serve conditions.
Table 3.3 Steel strip geometrical properties.

Models 207 GPa
Ultimate stress 960 MPa
Poisson radio 0.3
Profile 0.5 mm × 52 mm
Winding angle 54.7°
Table 3.4 Models with different inner radius.

Model Inner radius Layers
A1 25 mm Internal sheath + four layers steel strips + out sheath
A2 25 mm Internal sheath + six layers steel strips + out sheath
B1 50 mm Internal sheath + pressure armor + two tensile

armors = out sheath
B2 100 mm Internal sheath + pressure armor + two tensile

armors = out sheath
Table 3.5 Prediction by two theoretical models.

Mode1 Theoretical mode 1 Theoretical mode 2
A1 43 MPa 38 MPa
A2 55 MPa 53 MPa
B1 134 MPa 99 MPa
B2 38 MPa 30 MPa
In Table 3.5, it shows the prediction of theoretical model 2. It is always

smaller than theoretical model 1’s, the main reason is that the theoretical
model 2 doesn’t take the contribution of the cylindrical layers into con-
sideration, so it is underestimated in Tables 3.4 and 3.5, the procedure of
design can be concluded as below.
By given small radius and pressure, take 25 mm and 30 MPa, for
example, the four layers SSRTP could meet this requirement enough.
While the internal pressure increases to about 50 MPa, it’s necessary
to adjust the winding angle or add more layers to satisfy the require-
ment. When the given radius is more than 50 mm or pressure is more than
60 MPa, the SSRTP might not satisfy in this condition. In addition, the
pressure armor and tensile armor are often used in big radius pipe to sub-
jected to high pressure. If the requirement of radius or pressure keeps
increasing, adjust the winding angle or thickness of pressure armor and
tensile armor. Based on these conclusions, it is easily to design a software
to design structure of pipes when the internal pressure increases. The flow-
chart is shown in Figure 3.14.
Give the inner radius and

internal pressure
Experimental
Try steel strip reinforced
investigated
thermoplastic pipe
Check whether the strength Yes Section design

requirement is satisfied parameters
Not satisfied
Yes
Adjust the parameters Numerical
investigated
No satisfied
Yes
Add the pressure armor
and tensile armor
Figure 3.14 Computer design flowchart of pipe section.

3.6 Conclusions
Within this chapter, the burst behavior of the flexible pipe was investi-
gated by both theoretical model and numerical simulation. The accuracy
and reliability of the theoretical model is verified by the good agree-
ment between the two sets of results. But it is worth noting that this
chapter is limited to prediction of pressure armor stresses during pipe
operation only, and that the residual wires stresses from manufacturing
are disregarded and not taken in consideration. Also, a simplified software
to design structure section with the given radius and internal pressure is
presented, which can provide some references for the factory engineers.
From the research, we can learn that:
(1) The results of the theoretical model show good linearity of

flexible pipes under internal pressure in the elastic phase.
Theoretical model adopted here is basically valid in cal-
culating the physical quantities stated above according to
the comparison with the FEM. But the theoretical model
doesn’t take the self-locking of pressure armor into condi-
tion, so there are certain errors in the comparison.
(2) The result of theoretical model and numerical model shows
when the pressure armor yields, the axial displacement and
radial displacement begin to increase sharply, which indi-
cates that the pressure armor is the main internal pressure
resistant structure. For safety, it is acceptable to consider
when pressure armor fails, the pipeline will soon fail.
(3) In the process of internal pressure loading, the stress of
Z-shaped section increases gradually from inside to outside,
and the inner part near the end yield first.
(4) This chapter uses two theoretical models to predict the burst
pressure of pipe and put up a software to design structure of
pipes with given radius and pressure. When the radius and
pressure are small, the steel strip reinforced thermoplastic
pipe is useful to resistant certain pressure. While the radius
or pressure is big, it is necessary to add pressure armor and
tensile armor to subject to pressure.
(5) As in future work, a validation against test data is recom-
mended for both the ABAQUS and Analytical models.
References
1. Fernando, U. S., Sheldrake, T., Tan, Z., and Clements, R., 2004, “The Stress
Analysis and Residual Stress Evaluation of Pressure Armor Layers in
Flexible Pipes Using 3D Finite Element Models,” Proceedings of ASME 23rd
International Conference on Offshore Mechanics and Arctic Engineering.
2. Neto, A. G., de Arruda Martins, C., Pesce, C. P., Meirelles, C. O. C., Malta,
E. R., Neto, T. F. B., and Godinho, C. A. F., 2013, “Prediction of Burst in
Flexible Pipes”, Journal of Offshore Mechanics and Arctic Engineering, 135,
1, 011401.
3. De Oliveira, J. G., Goto, Y., and Okamoto, T., 1985, “Theoretical and method-
ological approaches to flexible pipe design and application”, In Proceedings.
Offshore Technology Conference. New York NY[PROC. OFFSHORE
TECHNOL. CONF.]. Vol. 3, pp. 517.
4. Zhu X K, Leis B N. Average shear stress yield criterion and its application
to plastic collapse analysis of pipelines[J]. International Journal of Pressure
Vessels and Piping, 2006, 83(9): 663–671.
5. Chen B, Nielsen R, Colquhoun R S. Theoretical models for prediction of burst
and collapse and their verification by testing[C]. Flexible Pipe Technology-
International Seminar on Recent Research and Development, Norway. 1992.
6. D. Fergestad and S. A. Løtveit, Handbook on design and operation of flexible
pipes[R]. 2014.
7. Kebadze E. Theoretical modelling of unbonded flexible pipe cross-
sections[D]. South Bank University, 2000.
8. Knapp R H. Derivation of a new stiffness matrix for helically armoured cables
considering tension and torsion[J]. International Journal for Numerical
Methods in Engineering, 1979, 14(4): 515–529.
9. API 17J. Specification for unbonded flexible pipe[S]. Washington, DC:
American Petroleum Institute, 2014.
10. Witz, J. A. (1996). A case study in the cross-section analysis of flexible risers.
Marine Structures, 9(9), 885–904.
11. Bai Y, Chen W, Xiong H, et al. Analysis of steel strip reinforced thermoplastic
pipe under internal pressure. Ships Offshore Struct 2015:1–8.
4
Tensile Behavior of Flexible Pipes
4.1 Introduction
Flexible pipelines must be designed considering the extreme situation for
which additional components are needed against severe loading and environ-
mental conditions. Tensile armor is employed in case of high tensile forces,
which increase with water depth; and massive components used against high
internal and external pressures. In this chapter, the mechanical behavior of the
pipe under pure tension is investigated by both theoretical and numerical anal-
ysis, and then, the contribution of the external pressure to the axial problem
is examined using an analytical analysis. As the deformations along the lon-
gitudinal direction of the studied pipe highly depend on the radial stiffness
of the cross section, the influence of the pressure armor is carefully evaluated
in the inception phase, and its radial stiffness is verified by its own numerical
results. The plastic behavior for both steel and polymer layers is taken into
account by using the secant modulus method. Moreover, this work answers a
question raised by producers that need to know when tensile armor is required
to reinforce Metallic Strip Flexible Pipes (MSFP). The theoretical model is
employed to carry out the comparison between pipes with different configu-
rations, in order to investigate the influencing parameters of the tensile stiff-
ness. The results obtained from the theoretical and the numerical simulations
lead to a remarkable confidence in the analytical solution thanks to a relatively
small difference between the outcomes. This chapter is quoted from Ref. [1].
A more detailed configuration of the strip-wound flexible tubing can be found
in the relevant literature [2, 3]. Knapp et al. [4, 5] used the energy method to
study the stiffness matrix of the spiral reinforcement layer under tensile and
torsional loads, and derived some classic formulas. Feret et al. [6] proposed a
simplified formula to calculate different stresses and contact pressures under
axisymmetric loads. Ramos et al. [7, 8] made some additional contributions to
the pipeline response. Saevik[9–11] has developed a model to predict the stress
of axisymmetric effects Dong [12], Guo [13], and Neto [20] made further
57
HDPE
z layers
x
y
Pressure Tensile
armor armor
Figure 4.1 Linear-longitudinal profile.
research on the mechanical model of flexible pipe from the perspective of vir-
tual work principle.
4.2 Theoretical Models

4.2.1 Mechanical Model of Pressure Armor Layer
The benchmark analyzed in this chapter focuses on a composite pipe
composed of one pressure armor layer and two conversely winded tensile
armor layers embedded into two HDPE layers, as in Figure 4.1.
The theoretical model is based on two main hypotheses.
The contribution of the pressure armor due to the axial strength can be
neglected in terms of tensile resistance as the winding angle is close to 90°,
as discussed by De Sousa [14]. While, it confers most of the radial stiffness,
which highly influences the tensile capacity of the pipe.
The contribution of the HDPE layers to the radial stiffness is neglected.
This is possible due to the presence of the pressure armor layer, which gives
the main contribution in terms of radial stiffness.
Therefore, the computation in radial direction is reduced into two com-
ponents: pressure armor layer and tensile armor layers. No initial gaps
between different layers are assumed. Winding angle and thickness varia-
tions, as well as friction are neglected [15].
The tensile response of the pipe is estimated considering tensile force
and external pressure at the same time. The external pressure Pext can be
applied directly on the external surface of the outermost tensile armor layer
due to the weakness of the surrounding HDPE coat. The tensile contribu-
tion of HDPE cylinders is included in calculating the total tensile resistance.
Tensile Behavior of Flexible Pipes 59
Rm2
Rm1
R2
R1
X
PW
Pext PC y
Figure 4.2 Contact pressures between layers mechanical model of pressure armor layer.
According to what has been already discussed in the paper by Yue

et al. [15], the authors in the present contribution are considering the
same assumptions for the present study. Such simplification does not affect
the results which are in very good agreement with the numerical model.
Pressures can lead to radial contraction of the layers, which can reduce
the tension stiffness of the flexible pipe. It is possible to study tensile and
pressure armors separately, in order to get the radial displacements of each
layer. Let PC be the contact pressure between the pressure armor and the
innermost tensile armor layer, while PW represents the contact pressure
between the tensile armor layers, as shown in Figure 4.2, where Rm1 and Rm2
are the mean radii of the tensile armor, R1 and R2 are inner and outer radii
of the pressure armor, respectively.
When subjected to tensile loading, the resistance of the flexible pipe is
influenced by its radial stiffness due to the helical structure of the layers.
The first effort of this chapter is to investigate the radial stiffness of the
pressure armor in order to obtain relatively accurate results for this axial
problem. Due to its complex shape, which shows different properties in
longitudinal and transverse directions, an equivalent theoretical model is
used to simplify the computation. The use of negligible elastic modulus
in the longitudinal direction is a common practice in the study of equiva-
lent helical structures of flexible pipelines with winding angle close to 90°
using analytical approaches. Such null modulus is not considered in the FE
analysis (which has actual mechanical properties of the materials used). In
fact, the pressure armor layer can be seen as an orthotropic cylinder, with
Young’s modulus equal to zero along the longitudinal direction, while in

the radial direction its equivalent thickness heq and Young’s modulus Eeq are
computed as previously done by De Sousa [14]:
12 I eq (4.1)
heq =
A′
nA′
Eeq = E (4.2)
heq L p
for which, Ieq is the equivalent moment of inertia per unit length, A′ is
the cross- sectional area of the pressure armor according to API 17B [17],
n is the number of tendons per each layer, Lp is the pitch length, and E is the
Young’s modulus of the material. Ieq is computed as in Ref. [1]:
knI 2′
I eq = (4.3)
Lp
where k is depending on both lay angle and moment of inertia of the cross
section, and I2′ is the smallest moment of inertia, which can be computed,
referring to Figure 4.3, as follows:
I3 + I 2 ( I 3 − I 2 )2 + 4 I 32
2
I 2′ = − (4.4)
2 2
Then, as discussed by Yue et al. [15], if the orthotropic cylinder is loaded

by radial confining pressure PC induced by the adjacent tensile armor
X3
X3'
X2'
X2
Figure 4.3 Pressure armor profile-principal outline.

Pc
eq
Rm1
eq
Rm2
Figure 4.4 Contact pressure and equivalent radii.
layers, it can be simplified into a planar ring under the same pressure. The
equivalent Young’s modulus Eeq and thickness heq for pressure armor can be
used in the calculation, where the mean radius for the equivalent cylinder
is kept the same as its actual one, as can be seen in Figure 4.4.
Keeping the main radius as reference, it is possible to figure out the
radial stiffness of the pressure armor, defined as done by Lu [18]:
PC
K= (4.5)
∆RC
where, DRC is the radial displacement of the external surface of the cyl-
inder due to PC.
Radial stiffness, according to the elastic theory for a thin-walled tube, as
shown in [18], can be expressed as follows:
(ν + 1) (1 − 2ν )R eq 2 eq 2
 eq
2 + R1  R 2
K= (4.6)
(
Eeq R eq 2 eq 2
2 − R1 )
where ν is the Poisson ratio of the material.
4.2.2 Mechanical Behavior of Tensile Armor Layer

Due to the helical shape of the tensile armor wires, once they are subjected to
tensile load, they exhibit elongation strain along the wire axis εi (with i = 1, 2
stands for the inner and outer armor layers, respectively). It can be expressed
as a combination of DL (displacements along the longitudinal direction)
and DRW(displacements in the radial direction), as shown in Figure 4.5,
where s and s′ represent the undeformed and deformed length of the wire,
ΔL ΔL
s'
L L
s
2πΔRw
R 2πR
ΔRw
Figure 4.5 Contraction and elongation for a representative pitch length of tensile wire.
respectively. The mathematical model adopted here was quoted from Knapp
[4]. Ignoring the rotation term, the expression can be written as
∆L 2 ∆R
ε i = ε l ,i + ε r ,i = cos α − W sin 2α (4.7)
L Rm ,i
where α is the winding angle of the wires, Rm,i is the mean radius, and L
is length of the pipe.
The tensile force along the axial direction of the helix can be divided into
two components: the hoop direction and the axial directions of the pipe.
The hoop stresses per each wire can be expressed as
σh,i = Esεi cos2 α (4.8)
where Es is the secant Young’s modulus of the constituent material.

It should be pointed out that, in the incremental process, Es changes in
every step in order to take the plasticity of the material into account. As Es
used in the current step is from the previous one, whose value is actually
larger, the total tensile force obtained might be greater than its real situ-
ation. However, if the increments are small enough, this error would be
controlled in the tolerable range.
The hoop stress of the tensile armor layer results in confining or extrud-
ing pressure to its adjacent layer. Due to the gaps between the wires in the
same layer, the filling factor βi is introduced, which exhibits the relation-
ship between the area filled by wires and gap. The equilibrium state of the
Pi+l
Pi
σh, i σh, i
Figure 4.6 Radial loading condition of tensile armor layer.
tensile armor’s cylinder can be seen in Figure 4.6, and the contact pressures
can be derived as
σ h ,i hβi
Pi − Pi+1 = (4.9)
Rm ,i
where h is the thickness of the wire.

Taking the inner tensile armor layer as example, by substituting (4.7)
into (4.8) and then into (4.9), and considering the effects of the confined
pressure produced by the outer tensile armor layer, one can get the contact
pressure PC as
Pc =Ω1ΔL−Ω2RW+Pext (4.10)
 1 1 1
Ω1 = hβ Es sin 2α cos 2 α  +
 Rm1 Rm 2  L (4.11)
 1 1 
Ω2 = hβ Es sin 4α  2 + 2  (4.12)
 Rm1 Rm 2 
4.2.3 Overall Mechanical Behavior

In the hypothesis of no separation between layers, the radial displacement
can be considered the same for each layer: DRC = DRW.
By solving Eqs. (4.5) and (4.10) simultaneously, the two unknowns, i.e.,
radial deformation DRW and contact pressure PC of the problem can be
obtained. Once these two outputs are known, it is possible to compute the
strain for each wire using Eq. (4.7). The total tensile strength of the pipe F
can be obtained by summing up the tensile resistance from the force of each
wire, as well as the contributions of the internal and external HDPE layers:
2 2
F= ∑n ε E A cos α + ∑ A
i =1
i i s
2
j =1
Pj
∆L
L
EP (4.13)
where A is the cross-sectional area of a single wire, APj, with j = 1, 2, stands

for inner and outer HDPE layers, is the area of the cross-sections, and EP is
the secant Young’s modulus of HDPE material.
4.3 Numerical Model

In this section, the tensile behavior of the flexible pipe is simulated using
the finite element software ABAQUS [19] in order to verify the reliability
and accuracy of the theoretical model.
4.3.1 Pressure Armor Stiffness

Firstly, the validity of the theoretical formulation for the pressure armor
radial stiffness is verified as the accuracy of the theoretical value K will
directly affect the final outcomes.
The imported profile of the pressure armor for the simulation is chosen
based on API 17B [17], and its relative dimensions are shown in Figure 4.7.
Two pitch lengths of the corrugate section are used to validate the real one,
whose longitudinal section is also shown in Figure 4.7.
The developed simulation considers a 3D ring model where the lay
angle is neglected. As discuss by Neto et al. [20], this simulation illustrates
good agreement comparing results to full 3D pipe model when solving the
radial problem, even if initial imperfections are not considered, as shown
in Figure 4.8. As the ring is symmetric to xy plane, it is possible to run the
simulation for half of the 3D ring, which can further improve the speed
of the simulation. It is noteworthy that, no initial imperfection or ovality
is introduced in this simulation as its fundamental purpose is to obtain
the relationship between the ring’s radial displacement with the external
pressure.
The matrix of the inertia for the cross-section of pressure armor can be
obtained from AutoCAD referring to Figure 4.3, and the corresponding
results for Figure 4.7 are shown in Eq. (4.14), which could be used in Eq.
(4.4) to calculate its smallest moment of inertia
4.20 95
85
0.65
0.74 0.85 3.00
0.65
1.50
85
85
85
85 3.00
10.5
4.10
3.00
3.00
4.10
10.50 3.00 85
95
2.40
85 85
5.00
0.85 0.65
0.74
3.00 0.65
95 4.20 85
Figure 4.7 Pressure armor-parameterized cross-section (lengths in millimeters and

angles in degrees) and profile used for FEM.
X Y
X Y
Figure 4.8 Pressure armor’s load and boundary conditions.

 I11 I12 I13   0 0 0 

   
I =  I 21 I 22 I 23  =  0 10000 1148  (4.14)
 I I 32 I 33   0 1148 3694 
 31
Material properties and other parameters needed for the computation

are listed in Table 4.1. The pressure armor is considered as linear elastic.
Based on the formulations illustrated in Section 2.1, the radial stiffness
could be calculated as 253.08 MPa/mm.
Due to the intricate shape of the imported cross-section and potential
contacts a contact of type “General contact” is employed to simulate the
interactions between the two parts. In ABAQUS environment such contact
typology is not related to any specific configuration but is able to relate
two surfaces of general shape (even of complex shape as in the present case).
In order to assume that surfaces in contact slide freely without friction
“Frictionless” tangential behavior is selected, while in order to carry out the
contact pressure analysis “Hard contact” normal behavior with “Allow sep-
aration after contact” is chosen. This means that ABAQUS is able to put in
contact two surfaces by indicating “Hard contact” option and the two sur-
faces during simulation might not be in contact according to the algorithm
“Allow separation after contact” as discussed in [16]. The latter is defined by
(p-h) model, which relates the contact pressure p among surfaces and the
overclosure h between contact surfaces. When h < 0, it means no contact
pressure, while for any positive contact h is set equal to zero, as discussed
in [21].
Table 4.1 Parameters for pressure armor layer.

Parameters Value
Number of tendons n 1
Coefficient factor K 1
Young’s Modulus E [MPa] 200,000
Poisson ration ν 0.3
Pitch length Lp [mm] 14.86
Inner radius Rinn [mm] 76.20
Thickness t [mm] 9.84
The external pressure supposed as constant along the width of the ring
and applied directly on the external surface. Kinematic is governed by the
boundary conditions, and symmetric boundary condition with respect to
the xy plane is exploited. The pitch length is kept unchanged during the sim-
ulation. U1 displacements are allowed at the bases of the ring (z = 0), while
U2 displacements are allowed in the middle surface of the ring (x = 0), as
shown in Figure 4.8.
C3D8R element (eight-node continuum linear brick elements with
reduced integration and hourglass control) is used for the mesh of the two
parts, as shown in Figure 4.9. These elements can be used for linear and com-
plex non-linear analysis producing high accuracy results when contacts and
non-linear geometry are considered, as discussed by Kim et al. [22].
Due to the frictionless contacts of this case, “Dynamic implicit” simu-
lation is activated to obtain the radial displacements, thanks to the possi-
bility of applying larger increments and achieve higher accuracy compared
to a Dynamic explicit analysis. Large strains are activated automatically by
ABAQUS environment and implementation within its FE analysis because
the flag “Nonlinear geometry” has been activated during the ABAQUS sim-
ulation. Moreover, material is considered as elastic in this case, with prop-
erties shown in Table 4.1. In order to achieve relatively accurate quasi-static
analysis results from the dynamic analysis, the ratio between kinetic energy
(ALLKE) and strain energy (ALLSE) for the whole model should be low
Z X
Y
Z X
Figure 4.9 Pressure armor mesh.

enough during its whole process, and their corresponding energies are illus-
trated in Figure 4.10, which shows that the numerical results are reliable.
The external surface of the ring’s displacements in radial direction
against external pressure is extracted out from 12 points, located along the
external surface. The behaviors for two representative points are shown in
Figure 4.11; they exhibit likely linear relationship, slightly influenced by
non-linearities which can be attributed to the presence of gaps between the
cross-sections.
Each curve is then linearized, and the corresponding radial stiffness can
be obtained. For the pressure armor, the radial stiffness is acquired as the
average among the results for the 12 points and it results in 256.22 MPa/
mm. Comparing the value with the theoretical results 253.08 MPa/mm,
the percentage error is equal to 1.23%.
Once the numerical radial stiffness of the pressure armor is verified, the
comparisons for the whole model are treated.
40
35 ALLKE
30 ALLSE
Energy (J)
25
20
15
10
5
0
0 10 20 30 40 50
External pressure (MPa)
Figure 4.10 Strain and kinetic energies against load.
0.25
Radial displacement (mm)
K = 239.69 MPa
0.2
K = 268.8 MPa
0.15
0.1
0.05
0
0 20 40 60
External pressure (MPa)
Figure 4.11 Pressure against radial displacements for two representative points.
4.3.2 Full Pipe

The profile of the pressure armor used in this full pipe FEM is exactly the
same as the one in Figure 4.8. More than 33 of its pitches are involved, with
the total length to be around 500 mm. Geometrical parameters for the sim-
ulated pipe are listed in Table 4.2.
Elastic-plastic behavior is considered for both steel and HDPE, and their
corresponding properties are shown in Figures 4.12 and 4.13, respectively,
which are delivered from the experimental results. It is supposed that pres-
sure and tensile armor layers are made of the same material with a yielding
stress equal to 578 MPa.
An “Explicit dynamic” analysis is developed, which allows for the defini-
tion of general contact conditions and it is suitable to perform quasi-static
Table 4.2 Pipe’s parameters.

Parameters Value
Wire thickness h (mm) 5.00
Wire width b (mm) 17.50
Winding angle α (°) 54.7
Inner layer’s wire number n1 19
Outer layer’s wire’s number n2 20
Inner cylinder’s thickness t1 (mm) 6
Outer cylinder’s thickness t2 (mm) 4
1000
800
Stress (MPa)
600
400
200
0
0.000 0.004 0.008 0.012 0.016
Strain
Figure 4.12 Steel stress-strain relationship.

25
20
Stress (MPa)
15
10
0
0 0.05 0.1 0.15
Strain
Figure 4.13 HDPE stress-strain relationship.
analyses with complicated contact conditions, as explained in [19]. The

selection of an explicit instead of an implicit dynamic analysis is justified
by the fact that accurate results will be achieved if the increments are small
enough, guaranteed by the time equal to two; otherwise, the solution will
diverge because the equilibrium is not strictly enforced.
For this case, in order to easily apply the load and boundary conditions
to the model, two reference points for all the degrees of freedom are neces-
sary at the two end surfaces of the pipe, as shown in Figure 4.14. The link
type used in the present simulations is “coupling” which represent kine-
matic type coupling of all the degrees of freedom are constrained between
the reference point and the other surfaces on the modeled pipeline.
Displacement along the z direction is applied at RP-2 to simulate the
tensile load. Symmetric boundary conditions are set at RP-1 for which
Y
RP-1
Z X
Y RP-2
Z X
Figure 4.14 Reference point at the end surfaces.

× RP-2
Y
Z X
Figure 4.15 Interlayered structure mesh.
9000
8000 ALLKE
7000 ALLSE
6000
Energy (J)
5000
4000
3000
2000
1000
0
0 5 10 15
Longitudinal displacement (mm)
Figure 4.16 Strain and kinetic energies.
U3 = UR1 = UR2 = 0. C3D8R is also used in this model, and their complex
geometry and structure is shown in Figure 4.15.
The obtained kinetic and strain energy are shown in Figure 4.16, Figure
4.17, which shows relatively good results when the elongation reaches
about 12 mm.

The comparison between the curves of the tensile forces versus elongation
strain of the pipe obtained from both theoretical and numerical results
are conducted. These two curves are in good agreement in terms of both
elastic and plastic behavior, revealing a maximum percentage difference of
4.50% when the strain is equal to 2.4%. The relatively rough surfaces might
2000
Tensile force (kN)

1500
1000
FEM
500
Theory
0
0 0.005 0.01 0.015 0.02 0.025
Elongation strain
Figure 4.17 Tensile force comparison.
yield to an uneven distribution of the confining pressure and in the stress

components too. Consequently, in some regions also, the pressure armor
may reach to the plastic stage in the numerical simulation, which is not
considered in the theoretical model.
In order to avoid end effects, the stresses located in the mid-span of the
pressure armor are extracted out from the numerical simulation. The external
surface of the layers is subjected to much more severe loading conditions, two
regions along one pitch length are selected due to the uneven distributions
of the stresses, and their stress variations are shown in Figure 4.18, points
A and B are two groups of points selected not according to different geo-
metric conditions but due to different state conditions. Points A represent
the region with the most rigorous stress conditions while Points B shows the
ones with less harsh conditions, both selected on the external surface of the
pressure armor along the pitch length as represented in Figure 4.19. From this
graph, it can be observed that, the stress for points A have already exceeded
800
Point A-1
700
Point A-2
Mises stress (MPa)
Proportional limit stress

600 Point A-3
500 Point A-4
Point A-5
400 Point B-1
300 Point B-2
200 Point B-3
Point B-4
100 Point B-5
0
0 0.005 0.01 0.015 0.02 0.025
Elongation strain
Figure 4.18 Mises stress of pressure armor from FEM.

Figure 4.19 Points selected on the pressure armor external surface for contact pressure
analysis.
the material’s proportional limit stress. This might result in a smaller tensile
force in the numerical model, especially in the later stage when comparing to
the theoretical results which regards the pressure armor as elastic.
From the theoretical model, it can be found that, the contribution of the
plastic layers corresponds to 5.74% of the total tensile strength, as shown in
Figure 4.20, where W stands for tensile wire strength, while WP accounts
also for HDPE layers. This result demonstrates the neglection of their influ-
ence is reasonable for further studies in order to get conservative results.
2000
1500
Tensile force (kN)
1000
W
500
WP
0
0.000 0.005 0.010 0.015 0.020 0.025
Elongation strain

The comparison in terms of radial displacement does not provide reli-

able results if the interlocked pressure armor is included in the pipe profile.
Several reasons could be attributed to this deficiency. The assumption of
equivalent pressure armor as an orthotropic cylinder leads to the hypoth-
esis of continuous geometry in the theoretical model, while gaps, rough
surfaces and winding angle appear in the FEM simulation. Besides, they
could also be affected by the thickness variation in different layers, which
is not considered in the analytical model.
In order to get acceptable results from FEM, points which are not in
correspondence of the gaps between wires are chosen to investigate the
contact pressure, and the selected points are shown in Figure 4.19.
The average results among the outcomes are compared to the theoretical
ones, as shown in Figures 4.21 and 4.22. As they show good accordance,
60
50
Contact pressure (kN)
40
30
20 FEM
10 Theory
0
0 500 1000 1500 2000
Tensile force (kN)
Figure 4.21 Contact pressure between pressure armor and inner tensile armor layer.
30
Contact pressure (MPa)
25
20
15
10
FEM
5 Theory
0
0 500 1000 1500 2000
Tensile force (kN)
Figure 4.22 Contact pressure between tensile armor layers.

Eq. (4.9) could be used as a rough estimation to carry out the contact pres-
sure between the two different layers.
The evaluation in terms of stresses regards the tensile armor is con-
ducted, being the one that provides most of the strength in the longitu-
dinal direction. The theoretical model assumes that strain and stress is
axisymmetric, i.e., the stress for all the wires at the same pipe cross sec-
tion is assumed to be the same. Von Mises stress contour plot of the outer
tensile armor layer is shown in Figure 4.23, with its active view cut in z
plane. Selecting the points from all the wires located at the mid-span and
comparing their stresses and strain with the theoretical results as shown
in Figure 4.24, it can be observed that, even if there are some fluctuations,
their amplitude is still not high.
A random wire from each layer was chosen to conduct the comparison
between the Mises stress and the elongation, as shown in Figures 4.25 and
4.26. The comparison between the trends leads to a remarkable confidence
in terms of stress behavior.
Once the validity of the theoretical model is proved, it is extended in
order to consider the contribution of the external pressure. Longitudinal
displacements are applied at a constant rate until 20 mm, while constant
pressure equal to 20 MPa is applied on the outer surface of the tensile
S, Mises
(Avg: 75%)
+9.284e+02
+8.673e+02
+8.062e+02
+7.451e+02
+6.839e+02
+6.228e+02
+5.617e+02
+5.005e+02
+4.394e+02
+3.783e+02
+3.172e+02
+2.560e+02
+1.949e+02
Z X
Figure 4.23 Mises stress—outer layer of wires.

0.006 600
550
0.005 500
Stress (MPa)
Theory-Strain 450
Strain
0.004 FEM-Strain 400
Theory-Stress
350
FEM-Stress
0.003 300
250
0.002 200
0 10 20
Wire number
Figure 4.24 Strain and stress distribution for outer tensile armor layer.
1000
800
Stress (MPa)
600
400
FEM
200
Theory
0
0 0.005 0.01 0.015 0.02 0.025
Elongation strain
Figure 4.25 Mises stress comparison for inner tensile armor layer.
900
800
700
600
Stress (MPa)
500
400
300
FEM
200 Theory
100
0
0 0.005 0.01 0.015 0.02 0.025
Elongation strain
Figure 4.26 Mises stress comparison for outer tensile armor layer.
2000
Tensile force (kN)

1500
1000
500 T
TP
0
0 0.01 0.02 0.03 0.04
Elongation strain
Figure 4.27 Tensile strength comparison for the pipe subjected to pure tension and
combined external and tensile loads.
armor layer. Only the contribution of the tensile armor is considered as

previously explained in Figure 4.20. As it was expected, the presence of the
pressure armor in the profile of the pipe makes the structure stiff enough
in radial direction so that the hydrostatic pressure does not affect the ten-
sile capacity of the pipe in a notable way, which is just slightly reduced, as
shown in Figure 4.27, where T stands for pure tensile load and TP stands
for combined tensile and external pressure loads.
4.5 Parametric Study

In this section, the reference case of “metallic strip flexible pipes” subjected
to pure tensile load is included, which is regarded as Case 1. The study has
already been developed in [16] for a range of MSFP which does not include
the pressure armor layer in the profile. It has been observed that, found out
that, for this kind of pipe, the tensile force is mainly due to the PE layers.
Based on MSFP’s structure, the extra pressure armor layer is added
in its inner reinforcement cross-section design, regarded as Case 2.
Figure 4.28 MSFP-based reinforced longitudinal profile.

As previously shown, the contribution to the overall radial stiffness induced

by the pressure armor allows the possibility of neglecting the innermost
and outermost HDPE layers. The tensile armor layers, existing in Case 3,
are substituted by four steel strip reinforced layers accounting for thickness
h = 0.5 mm. The detailed longitudinal profile of Case 2 is illustrated in
Figure 4.28. For this numerical campaign, the inner diameter and the load-
ing condition are the same as the previous case. The theoretical model is
extended to the new geometry, for which the coefficients expressed in Eqs.
(4.11) and (4.12) should be modified by considering the variation in terms
of number of layers and thickness, which can be expressed as
sin 2α cos 2 α  1 1 1 1 
Ω1 = hβ Es  + + + (4.15)
L  Rm1 Rm 2 Rm3 Rm 4 
 1 1 1 1 
Ω2 = hβ ES sin 4α  2 + 2 + 2 + 2  (4.16)
 Rm1 Rm 2 Rm3 Rm 4 
Tensile forces comparison for those pipes with three different config-
urations is shown in Figure 4.29. As expected, the strength provided by
MSFP is the lowest. In fact, for the same inner diameter, the MSFP shows a
decrement in axial resistance equal to 81.07% compared to the case which
includes two tensile armor layers. At the same time, it is possible to see
the improvement in terms of tensile capacity when the pressure armor
2000
Tensile force (kN)
1600
1200 Case 1
Case 2
800 Case 3
400
0
0 0.01 0.02 0.03 0.04
Elongation strain

is included in the MSFP design equal to 81.53%, which results from the
comparison between Case 1 and Case 2. When both the pressure and ten-
sile armor layers are included in the design, the high strength in radial
direction is not only provided by the pressure armor but also by the ten-
sile armor due to its relevant thickness. Thus, reducing the thickness of
the wires not only affects the tensile capacity itself but also the loss of the
external capacity.
Being the elongation of the pipe strictly related to its weight, which
mostly depends on the amount of the reinforcement needed as well as the
water depth, it is possible to assert that the steel strip reinforcements in
terms of axial strength of the pipe are suitable for shallow waters. On the
other hand, the contribution of thick steel wires is suitable for extreme
loading conditions such as for deep water design. It is noteworthythat for
both Cases 2 and 3, the contribution of the radial stiffness induced by pres-
sure armor plays an important role in terms of axial capacity of the pipe.
4.6 Conclusions
In this chapter, an easy theoretical method for estimating the tensile stiff-
ness of the unbonded flexible pipe is verified by numerical simulations.
Secant modulus is employed in order to carry out the plastic behavior of
the material, and this theoretical model is suitable for high loading con-
ditions, which can provide relatively accurate tensile strength for pipeline
engineers. The following conclusions could be drawn.
When considering both pressure and tensile armor layers in the pipe’s
profile, the external pressure will not have very big impact on its tensile
capacity as the radial stiffness of the pressure armor are large enough to
resist the radial deformation induced by external pressure.
MSFP is only suitable for shallow water application. Adding a certain
profile of pressure armor into MSFP leads to a significant increase in terms
of resistance capacity (about eight times). In order to avoid material waste,
the profile of the pressure armor could be adjusted according to the water
depth, and this can make MSFP exploitable for deeper water depth.
Tangent modulus should be used for next works in order to obtain more
accurate results. The contribution of the interlocked carcass should also
be taken into account in future works, to verify whether its radial stiffness
leads to a remarkable increasing tensile capacity of the pipe.
References
1. Fergestad, D., Løtveit, S. A., ‘Handbook on Design and Operation of Flexible
Pipes’, NTNU, 4Subsea and MARINTEK, 2014.
2. Bai, Liu T, et al. Buckling stability of steel strip reinforced thermoplastic pipe
subjected to external pressure[J]. Composite Structures 152(2016)528–537.
3. Bai Y, Liu T, et al. Mechanical behavior of metallic strip flexible pipe sub-
jected to tension[J]. Composite Structures, 170(2017)1–10.
4. Knapp, R. H. (1975, September). Nonlinear analysis of a helically armored
cable with nonuniform mechanical properties in tension and torsion. In
OCEAN 75 Conference (pp. 155–164). IEEE.
5. Knapp, R. H. “Derivation of a new stiffness matrix for helically armoured
cables considering tension and torsion.” International Journal for Numerical
Methods in Engineering 14. 4(1979): 515–529.
6. Feret, J. J., and C. L. Bournazel. “Calculation of stresses and slip in structural
layers of unbonded flexible pipes.” Journal of Offshore Mechanics and Arctic
Engineering 109. 3(1987): 263–269.
7. Ramos, R., Martins, C. A., Pesce, C. P., etc. Some further studies on the axial–
torsional behavior of flexible risers[J]. Journal of Offshore Mechanics and
Arctic Engineering, 2014, 136(1): 1–11.
8. Ramos, R., Kawano, A. Local structural analysis of flexible pipes subjected
to traction, torsion and pressure loads[J]. Marine Structures, 2015, 42(1):
95–114.
9. Sævik, S., Bruaseth, S. Theoretical and experimental studies of the axisym-
metric behaviour of complex umbilical cross-sections[J]. Applied Ocean
Research, 2005, 27(2): 97–106.
10. Sævik, S. Theoretical and experimental studies of stresses in flexible pipes[J].
Computers & Structures, 2011, 89(23): 2273-2291.
11. Sævik, S., Gjøsteen, J. Strength analysis modelling of flexible umbilical
members for marine structures[J]. Journal of Applied Mathematics, 2012,
2012(1): 1-18.
12. Dong L, Zhang Q, Huang Y. Energy approaches based axisymmetric analysis
of unbonded flexible risers[J]. Huazhong Keji Daxue Xuebao (Ziran Kexue
Ban)/J Huazhong Univ Sci Technol (Nature Science Edition)2013;41(5).
13. Guo Y, Chen X, Fu S, Wang D. Mechanical Behavior Analysis for Unbonded
Umbilical under Axial Loads[J]. Journal of Ship Mechanics, 1007-7294(2017)
06-0739-11.
14. de Sousa, JoséRenato M., et al. “Structural response of a flexible pipe with
damaged tensile armor wires under pure tension.” Marine Structures
39(2014): 1–38.
15. Yue, Qianjin, et al. “Tension behavior prediction of flexible pipelines in shal-
low water.” Ocean Engineering 58(2013): 201–207.
16. Jiang, K., Liu, T., Yuan, S., & Bai, Y. (2018, June). Mechanical Behaviors of
Metallic Strip Flexible Pipe Under Axisymmetric Loads. In ASME2018 37th
International Conference on Ocean, Offshore and Arctic Engineering (pp.

005T04A010-V005T04A010).
17. American Petroleum Institute, 2002, API recommended Practice 17B,
Information Handling Services, API, Washington D. C.
18. Lu, M. W., Luo, X. F., 2001. Basic Elastic Theory, second ed. Bei Jing. (in
Chinese).
19. ABAQUS. 2014. User’s and theory manual version.
20. Gay Neto, A., and Martins, C. A., 2010, “Burst Prediction of Flexible Pipes,”
Proceedings of the 29th International Conference on Offshore Mechanics
and Arctic Engineering, 2010.
21. An C, Duan M, Toledo Filho RD, et al. Collapse of sandwich pipes with
PVA fiber reinforced cementitious composites core under external pressure.
Ocean Eng 2014;82: 1–13.
22. Kim T S, Kuwamura H. Finite element modeling of bolted connections
in thin-walled stainless steel plates under static shear[J]. Thin-Walled
Structures, 2007, 45(4): 407–421.
5
Design Case Study for Deep Water Risers
5.1 Abstract
Flexible pipe is compounded by many concentric layers, with different
structural and operational functions. These layers are usually made of dis-
tinct materials, including metal and plastic. To accurately meet the work-
ing requirements of the pipe and reduce its production cost, the primary
purpose of this chapter is to present the cross-sectional design procedure
and the case study for a specific unbonded flexible pipe is also illustrated.
In this chapter, the mathematical analysis and finite element analy-
sis are employed to study the properties of pipe under different working
conditions. A theoretical model for stresses and deformations of the pipe
has been studied, and the obtained results have been compared with the
ones from the FEM which is used to simulate the pipe under different
working conditions. Additionally, the several models will be developed to
study mechanical responses of pipes subjected to several loads. The results
and FEA models can be useful for the designing structure of flexible pipes.
5.2 Introduction
The increasingly employment of unbonded flexible pipe in the oil/gas
industry has brought significant challenges in the design process. The
design of flexible pipe includes cross-sectional design and accessories
design (bend stiffeners, etc.). The primary purpose of this chapter is to
present the cross-sectional design procedure and the case study for a spe-
cific unbonded flexible pipe is also illustrated. It is widely recognized that
the special construction of layers contributes to the extensively application
of unbonded flexible pipe in the deep water.
The geometrical parameters of each layer need to be identified in the
basic design phase. The structure of flexible pipe can be categorized into
three different components: (1) Metallic helix layer (carcass and pres-
sure armor) with large winding angle and irregular profile. Generally, the
83
winding angle of carcass and pressure armor is between 85° and 90°,
which makes the interlocked structure. (2) Metallic helix layer (tensile
armor) with small winding angle and rectangular profile. This component
is used to sustain longitudinal load. The profile, winding angle and num-
ber of helix can be modified to accommodate different working environ-
ments. The winding angle is usually between 35° and 60° considering the
mechanical properties and manufacture condition. (3) Cylindrical layer,
which mainly works as a sealing or anti- friction/anti-corrosion compo-
nent, and thus, it is usually made of polymers such as HDPE and PA, etc.
The criterion and requirements on unbonded flexible pipe are generally
from API. API 17B [1] gives detailed potential failure mechanisms of dif-
ferent failure modes which include collapse, burst and compressive failure,
etc. But, it does not provide the exact formula or methodology to predict
these failure modes. Understanding the mechanical responses of flexible
pipes subjected to different load conditions is significant for predicting
ultimate strength, so it is vital for the cross-sectional design.
The basic design idea can be concluded as follows: (1) Obtain the
different load conditions and load value that the flexible pipe may sustain
in the installation and operation process by global analysis. (2) Specify
the geometrical and material parameters of each layer based on the load
conditions and “engineer judgment”. (3) Verify the ultimate strength of
flexible pipe under different load conditions and make sure the designed
flexible pipe meet the corresponding standard requirements. In this chap-
ter, the analytical method is used to give an estimation of the mechanical
properties and the strength capacities for some specified cases which are
part of the design requirement prescribed in API 17J [2]. There are lots of
achievements in designing process in the literature for decades. Feret and
Bournazel [3] proposed the equation of equilibrium between stresses of
each layer and pressures in the radial direction, their method can be used
to estimate the mechanical responses of flexible pipe under axisymmetric
loads. Witz [4] first presented a case study in the structural analysis of an
unbonded flexible riser cross-section under tension, torsion and bending.
Handbook of Design and Operation of flexible pipes [5] described some
simple method to estimate mechanical responses of flexible pipes under
different failure modes, but it may be too conservative. Chen [6] consid-
ered the helix structure of flexible pipes and the contributions of burst
capacity provided by tensile armor. Corresponding experiments were car-
ried out to validate his theory. Wellstream Corporation [7] used Barlow’s
formula for thin-walled cylinders and the correction factor called “filling
fraction” to describe pressure armor. Neto et al. [8] simplified the pressure
armor as ring to predict burst pressure, an FE model was also established
Design Case Study for Deep Water Risers 85
with specific profile and winding angle of pressure armor to verify the
theory. But, they ignored the contribution of tensile armor layers. de Sousa
et al. [9] considered the carcass and pressure armor as orthotropic shells,
the equivalent thickness was applied to simulate these layers.
5.3 Cross-Sectional Design

The requirements and criteria on flexible pipe are mainly based on API 17J
[2], and the Handbook on Design and Operation of Flexible Pipes [5] also
provides some crucial background and guidance to the methods given in
API. With reference to API 17B [1], the design procedures of flexible pipe
can also be carried out in the following stages:
Stage 1. Material selection.

Stage 2. Cross-section configuration design.
Stage 3. System configuration design.
Stage 4. Dynamic analysis and design.
Stage 5. Detailed and service life design.
Stage 6. Installation design.
In the design process, all associated factors should be taken into con-
sideration to optimize and ensure the reliability and safety of the designed
pipelines. The correlative factors include the transported product, the
functional requirement, the surrounding infrastructure and environment
of the pipelines in service, and the operation and installation conditions.
In view of all those factors, the designed flexible pipe should satisfy the
corresponding standard requirements.
Understanding the mechanical responses and failure modes of the flex-
ible pipe when subjected to different load conditions is of key importance
for obtaining a relatively reliable designed cross-section. The ultimate
strength of the cross-section is governed by the geometrical configuration
of the linear, outer sheath of the pipe, the tendon and the winding angles of
the reinforcement layers, and the cross-section design analysis is normally
based on the pressure rating and use of analytical models considering
axisymmetric alone [5]. In this part, the analytical method is used to give
an estimation of the mechanical properties and the strength capacities for
some specified cases which are part of the design requirement prescribed
in API 17J [2]. Those issues address here include the burst pressure of the
metallic strip flexible pipe subjected to internal pressure, tension, the col-
lapse pressure due to external pressure can endure. The below formulas
can be used under the specific condition and in a simplified way to provide
some references for the early stages of the design procedures.
In the design process, all associated factors should be taken into con-
sideration to optimize and ensure the reliability and safety of the designed
pipelines. The correlative factors include the transported product, the
functional requirement, the surrounding infrastructure and environment
of the pipelines in service, and the operation and installation conditions.
In view of all those factors, the designed flexible pipe should satisfy the
corresponding standard requirements.
The cross-section configuration and dimensions are selected based on
the pipe’s functional requirements. Pipe dimensions are often specified by
the purchaser. Cross-section design calculations and checks are typically
carried out by the manufacturer using proprietary software. In order for
the manufacturer to supply pipes according to API 17J (2008) or API 17K
(2005), the design methodology including software validation shall be ver-
ified by an independent third party [5].
Carcass is a corrugated metallic tube fabricated by forming the strip
and spirally winding it around a mandrel. The edges of each winding are
formed so that they interlock with the adjacent winding. The carcass layer
prevents collapse under external hydrostatic pressure and in cases of rapid
decompression. The carcass structure is not fluid-tight. Its material has
been selected to be chemically resistant to the conveyed fluid.
Pressure sheath is a polymer layer extruded over the carcass to form
a boundary for the conveyed. For each pipe, the material and thickness are
selected to assure sufficient integrity against leakage over the service life.
Materials are selected that retain high strain at break relative to allowable
values when subjected to the expected thermal and chemical degradation
based on the production fluid conditions.
Pressure armor, as shown in Figure 5.2, is an armor layer consisting of
a wound profiled strip which resists internal and external pressure in the
hoop direction and provides a guard against creep extrusion of the pressure
sheath. The strip profile interlocks the edges as they are formed around
the pipe. The considerations for the selection of material of pressure
armor are strength, resistance to wear and compatibility with the chemi-
cals in the annulus environment see Figure 5.1.
Figure 5.1 Typical carcass profile.

Figure 5.2 Typical pressure armor.
The tensile armor layers consist of carbon steel wires formed into helix,
installed in contra-wound pairs of layers. The layers are preformed and
wound over the underlying layer and secured by tape. Tensile armor wires
are essentially rectangular in cross-section and are laid with small gaps
between adjacent wires. Tensile armor layers provide both hoop and axial
strength, the component of each governed by the lay angle. The con-
siderations for the selection of material of tensile armor wires are strength,
resistance to fatigue, and compatibility with the chemicals in the annulus
environment.
The tape layer prevents metal to metal contact during repetitive bend-
ing, which is used on all flexible pipe in dynamic service. The tape elimi-
nates metal wear at this interface during manufacturing, installation, and
operation. The effectiveness of the anti-wear layers in preventing wear
of the steel reinforcement is documented by several full scale tests,
References [10] and [11] (Design Report for Flexible Dynamic Riser).
The shield is an external polymer barrier extrusion applied to resist
mechanical damage and intrusion of sea water. For each pipe design, the
material and thickness is selected to assure sufficient integrity against
intrusion of seawater, mechanical abrasion and expected ultraviolet expo-
sure over the service life. Pipe bending, axial elongation and compres-
sion, external and internal pressure, installation, abrasion, local loads, and
manufacturing tolerances are considered in specifying the material and
thickness.
The flow chart of design procedure is summarized in Figure 5.3.
5.4 Case Study

The shield is an external polymer barrier extrusion applied to resist
mechanical damage and intrusion of sea water. For each pipe, the material
and thickness is selected to assure sufficient integrity against intrusion of
seawater, mechanical abrasion, and expected ultraviolet exposure over
the service life. Pipe bending, axial elongation and compression, external
Start
Design
requirement
dimension
Identify loads and

load conditions
Assume a designed cross-

sectional configuration
(Geometrical and Material
parameters of each layer)
Use theoretical method to

calculate the most dangerous
Yes
conditions of different layers
Dissatisfied
The stress/strain exceed the

allowable stress/strain?
No
Establish FEM
modle
Verify the capacity

of designed pipe
Satisfied
end
Figure 5.3 Flow chart of design procedure.
and internal pressure, installation, abrasion, local loads, and manufactur-

ing tolerances are considered in specifying the material and thickness.
The requirement is based on the client’s requirements. See Table 5.1 for
detailed information.
In terms of the pipe with one sealed end, the ultimate strength under
internal pressure can be calculated with three standards. One is the stress
in the pressure armor ultimately reach the yield point and another is after
yielding of the pressure armor, the additional internal pressure resisted by
Table 5.1 Design requirements.

Dimension (mm) 203.2
Design Internal Pressure (MPa) 27.73
Water Depth (m) 100
Design External Pressure (MPa) 1.2
Design Tension (kN) 1,709
the tensile armor layers. The other is the axial load on the tensile armor
layers could be much greater than the hoop load on the tensile armor
layers, which may fail in axial tension without the first standard reached.
In terms of the pipe with one free end, the ultimate strength under
internal pressure can be calculated with two standards. One is the stress
in the pressure armor ultimately reach the yield point, and another is after
yielding of the pressure armor, the additional internal pressure is resisted
by the tensile armor.
In this study, the standards of the pipe with one sealed end and free end
are compared.
Theoretical Calculate
In order to simulate the mechanical behavior of the flexible pipes through
two kinds of layers, several simplifying hypotheses are made ahead as
follows:
(1) All materials are homogeneous and isotropic and have lin-
ear elastic behavior.
(2) The strains, which occur in any part of the pipe, must be
small enough.
(3) Mean radius and thickness variations of the layers are
assumed to be uniform in each layer.
(4) Pipe ovalization due to the applied loads can be neglected.
(5) Gap between adjacent layers is not allowed in the unstressed
initial state.
(6) Friction is neglected in the whole analytical process.
(7) There is no contact between adjacent tendons of the
same layer, whatever be the conﬁguration.
Considering the contribution of all the layers to the internal pressure

resistance, the approaches available were derived based on the principle of
virtual work by Kebadze [12].
As stated before, the thin wall theory is chosen for the cylindrical com-
ponents in the chapter. The strain is assumed constant throughout the
thickness while considering the uniform change in wall thickness. Thus, the
internal strain energy of the cylindrical components can be expressed by
1
U=
2 ∫ ∫ ∫ (ε σ + ε σ
1 1 2 2 + ε 3σ 3 + τ 12γ 12 )dv (5.1)
where, ε1, ε2, ε3, and γ12 are strain components, and σ1, σ2, σ3, and τ12 are
corresponding s tress components. The related parameter definition is
The potential of external forces is expressed by
W = NpuL + TpΔφ + (PiΔVi − PoΔVo) (5.2)
εh
ε1
ε3
L
ε2
γ12
Ri
Ro
Figure 5.4 Mathematical parameters definition.

where ΔVo denote internal pressure and external pressure, the changes in
internal volume and in external volume, respectively. According to the
theory of energy conservation, the matrix form of stiffness equations for
the thin tube theory is derived:
 uL 
 
 L   
 (2G + λ )A 0 λA λA  Np
 ∆ϕ   
   
 0 GJ 0 0  L Tp 
 = 
 λA 0 (2G + λ )A λA  ur   2 Piπ Ri R − 2 Poπ Ro R 
   
 λA 0 λA (2G + λ )A   R 
   − Piπ Rit − Poπ Rot 
 ut 
 t  (5.3)

where A = π ( Ro2 − Ri2 ) , J = π ( Ro4 − Ri4 ) , G is shear modulus, ν is poison

1
2
ratio, and ur and ut are the changes in radius and thickness, respectively.
The helical layers are main structures of unbonded flexible pipes, which
have a significant influence on the mechanical behavior of pipe. The exact
formula for axial strain of a helical element was derived by Knapp. In order
to obtain a linear equilibrium equation of helical tendons, the exact for-
mula is simplified as
uL u ϕ
εh = cos 2 α + r sin 2 α + R sin α cosα (5.4)
L R L
Since the stiffness of helical layers is much higher compared to that of

cylindrical components due to their material’s characteristics, the radial
strain of the helical tendons should not be neglected. Therefore, the
internal work can be expressed as these two parts above adding together:
1
∑ EAi
∫ (ε )
n
2
U= h + 2νε h ε 3 + ε 32 dz (5.5)
2 i =1 (1 − ν 2 )cos α
where εh and ε3 denote the axial strain and radial strain, respectively.
Similarly, differential internal work is equal to differential external work,
which assembles the equations into a specific matrix as
 uL 
 cos3 α i R sin α i cos2 α i sin 2 α i cosα i v cosα i  
  L  
 R sin α i cos 2 α i R 2 sin 2 α i cosα i R sin3 α i vR sin α i  Nh 
∆ϕ   
   
nEAi  sin 4 α i v sin 2 α i  L Th 
2  sin 2 α i cosα i R sin3 α i  = 
1−ν cosα i cosα i  ur   2Piπ Ri R − 2Poπ Ro R
    
 v sin 2 α i 1  R   − Piπ Ri R − Poπ Ro R 
 v cosα i vR sin α i  ut 
 cosα i cosα i  
 t 

(5.6)
where A is the area of a helical tendon, Nh and Th denotes the longitudinal

tension and torsion sustained by a specific helix layer, respectively, and R
is the mean radius.
The equilibrium equations developed for each individual layer are then
assembled to formulate a total stiffness matrix, according to the specific
arrangements of cylindrical layers and helical layers. Still, the total matrix
contents have certain unknown values. Thus, the consistent in the radial
direction is employed to solve the problem:
ur,i − ur,i+1 + (ut,i + ut,i+1)/2 = 0 (5.7)
where i denotes the layer number. Therefore, the mechanical behavior of

flexible pipes can be obtained. Then, the contact pressure between each
adjacent layer can be examined. If the contact pressure is negative, it
means these two layers are separated. Then, the contact pressure should
be taken as zero and one corresponding consistent equation should be
removed to solve the whole equilibrium equations again to obtain the
available results.
Collapse Design
Resistance to collapse is one of the design requirements; it is necessary
to check the ability of pipeline to bear external pressure. According to
document S A. [5], the critical pressure used to predict crushing can be
calculated as follows.
Np
3EI eqi
pcr = ∑ i =1
R3
(5.8)
Et 3
EI eq = (5.9)
12(1 − ν 2 )
EI 2′
EI eq = Kn (5.10)
Lp
Tension Design
For the situation of a riser, the catenary theory is introduced to validate the
design results from the previous parts under tension. Based on the config-
uration as shown in Figure 5.5, The specific calculation is shown as follows:
TV = TT sin θ = ωl (5.11)
TO = TT cos θ = ωl cot θ (5.12)
 N 


∑ π ∆ ( Rn2 ) ρn + π Ri2 ρi 
 (5.13)
ω = n=1
− ρw  g π Ro2
 π Ro2 
2a
l = h 1+ (5.14)
h
Tv TT
θ
A To
h
col
To X
O X
Figure 5.5 Catenary configuration for the tension design.

TO
a= (5.15)
ω
θ ranges from 75° ~85° based on the engineering practice. ρi is given by

the design requirement. Then, TT can be obtained putting all the related
parameters from the design results into the Eqs. (5.11)~(5.15), and it can
be seen that the design tension is less than TT considering the safety factor.
Throughout this report, the collapse behavior of an interlocked carcass
is simulated and results are investigated theoretically and numerically.
A variation of a primary adopted theoretical model is proposed which
may be of interest for practical applications. In order to do so, a series of
numerical models for the calibration were needed, leading to an empirical
formulation. The latter is valid in terms of both pre-buckling and collapse
conditions for a defined value of initial ovalization, so that the critical pres-
sure is computed neglecting the friction between layers for both models.
Outcomes are compared with a steel strip reinforced thermoplastic pipes
under the same requirements, in order to understand when the reinforce-
ment is required. This study can provide support for factory engineers
due to the accurate and reliable results that show very small difference
between numerical and theoretical models.
5.5 Design Result

The geometrical parameters designed are displayed in Table 5.2, and the
material parameter for the design requirement based on the Ref. [8] is dis-
played in Table 5.3.
Load Analysis
The allowable stress of metallic layers can be refer to API 17J, in which
the material utilization factor is 0.67 for carcass and 0.85 for both pressure
armor layer and tensile armor layer in extreme operation situation. The
allowable strain of polymeric layer is selected to be 7.7%. The detail infor-
mation can be seen in Table 5.4, σy denotes the yield stress.
The axisymmetric loads can be divided into tension, internal pressure
and external pressure. The different load cases are essential to evaluate
the maximum stress or strain that each layer may sustain. Table 5.5 shows
the load cases the pipe might subjected to during operation.
Table 5.2 Geometrical parameters.

Dimensions
Description (mm) Number Ri (mm) Ro (mm) Alpha
Carcass 48 × 1.2 1 101.5 107 −88.8
Pressure – 1 106.5 113 –
Sheath
Zeta Layer 11.94 × 8 1 113 123 −88.26
Fabric Tape 76 × 1.5 1 123 125 −80.83
First Tensile 11.3 × 4.5 52 124.5 129 −35
Armor
Fabric Tape 76 × 1.5 1 129 130.5 −81.15
Second Tensile 11.3 × 4.5 55 130.5 135 35
Armor
External – 1 135 140 –
Sheath
Table 5.3 Material parameters.

Yield strength Module Possion
Description Material (MPa) (MPa) ratio
Carcass AISI316L 310 200,000 0.3
Pressure Sheath PVDF 40 284 0.45
Zeta Layer – 700 200,000 0.3
Fabric Tape PVDF 40 301 0.45
First Tensile Armor – 700 200,000 0.3
Fabric Tape PVDF 40 301 0.45
Second Tensile Armor – 700 200,000 0.3
External Sheath HDPE 40 600 0.45
Table 5.4 Utilization factors for flexible pipe.

Allowable stress Allowable strain
Carcass Pressure armor Tensile armor Polymeric layer
0.67 × σy 0.85 × σy 0.85 × σy 7.7%
Table 5.5 Load cases.

Internal pressure External pressure
Local case (MPa) (MPa) Tension (kN)
A 27.73 × 1.5 0 0
B 27.73 × 1.25 0 1,139 × 1.5
C 0 1.48 × 1.5 0
D 0 0 1,139 × 1.5
The Von mises stress in metallic layers and the equivalent strain in
polymeric layers are calculated separately using the theoretical method
mentioned above four load cases. Then, the maximum stress in the metal-
lic layers and the maximum strain in the polymeric layers under differ-
ent load cases are summarized for the pipe design. Table 5.6 displays the
results of the analysis, which suggest that the pipe designs are suitable for
Table 5.6 Maximum stresses and strains summary by layers.

Vonmises stress
Layers (Equivalent strain) σ/σa(ε/εa) Load case
Carcass 350.030MPa 0.746 D
Pressure Sheath 0.0283 0.368 A
Zeta Layer 517.416MPa 0.870 A
Fabric Tape 0.0062 0.081 B
First tensile armor 276.472MPa 0.465 B
Fabric Tape 0.0048 0.062 B
Second tensile armor 276.467MPa 0.465 B
External Sheath 0.0030 0.039 B
the intended service conditions. σa and εa denote the allowable stress and
allowable strain, respectively.
The σ/σa(ε/εa) can be regarded as utilization ratio, which represents the
percentage of strength usage in layers. It can be observed that all the uti-
lization ratio are below 1.0, which suggests that the designed pipe can
satisfy the requirements and work safely under designed situations. The
wet collapse pressure is calculated using Eqs. (15.1)~(5.15). The collapse
pressure is 7.8 MPa.
5.6 Finite Elements Analysis

As a whole, the utilization of the main components in the pipe demon-
strates the design results available and the finite elements method is
adopted in the following to validate them in another way.
The geometrical and material parameters of FE model can be referred
to Tables 5.2 and 5.3, respectively. It is noteworthy that the pressure
armor layer which is Zeta shape is adopted as equivalent rectangle to
save computational resources, and the polymeric layers are all considered
as cylindrical components since they have little influence on the analyti-
cal process. Besides, Carcass is considered as cylindrical component either
using equivalent method for it makes little contributes to the internal
pressure resistance. Figure 5.6 shows the detailed geometry of the 8-layer
model based on the proper cut of each layer.
Z X
Figure 5.6 Detailed geometry of FE model.

In order to efficiently control the boundary conditions, two reference

points (RPs) are created at the center of both end cross- sections. The end
cross-section nodes are all included in kinematic coupling with the related
reference points. The right reference point is ENCASTRE while the left
reference point is free but may be applied with tension when necessary.
The internal pressure is applied on the inner circumferential surface of the
pressure sheath since Carcass is not sealed and the fluid/gas can penetrate
carcass layer.
The model was simulated by solid elements (C3D8I for cylindrical com-
ponents and C3D8R for helical components) which can reflect the vari-
ation of wall thickness and stress distribution. The ABAQUS software is
selected to simulate the possible behaviors for its specialty in contacts and
nonlinear analysis. All with itself algorithm that ABAQUS automatically
identifies all possible contact pairs in the model is used. The quasi-static
analysis is adopted against the difficulties of convergence.
The failure criterion is chosen as the theoretical method where one of
layers reaches its allowable stress or strain. Figure 5.7 shows the failure
moment when the pipe is subjected to internal pressure (pressure armor
failure). Figure 5.8 shows the failure moment when the pipe is subjected to
tension only (tensile armor failure).
The collapse FE model is established to verify the wet collapse pres-
sure of the pipe. This model is a ring model with a winding helix outside
and inside. The inner helix represents the carcass and the outside helix
S, Mises
(Avg: 75%)
+2.916e+03
+2.673e+03
+2.430e+03
+2.187e+03
+1.944e+03
+1.701e+03
+1.458e+03
+1.215e+03
+9.723e+02
+7.293e+02
+4.864e+02
+2.435e+02
+5.866e-01
Y ODB: complete-copy-2-30htansu-new045-1200Z.odb Abaqus/Explicit 6.14-1 Mon Mar 21 14:20:06 GMT+08:00 2016

Step: Step-1
Increment 15423559: Step Time = 0.1395
Z X Primary Var: S, Mises
Deformed Var: U Deformation Scale Factor: +1.000e+00
Figure 5.7 Burst failure mode of flexible pipe.

U, U3
+1.784e+00
+1.465e+00
+1.146e+00
+8.263e+01
+5.070e-01
+1.876e-01
-1.318e-01
-4.511e-01
-7.705e-01
-1.090e+00
-1.409e+00
-1.729e+00
-2.048e+00
Y ODB: complete-combine-TIP-30h-T-IPd-allcon.odb Abaqus/Explicit 6.14.1 Sat Mar 12 10:04:34 GMT+08:00 2016
Step: Step-2
Z X Primary Var: U, U3
Deformed Var: U Deformation Scale Factor: +1.000e+00
Figure 5.8 Tension failure mode of flexible pipe.
simulates the remained seven layers with equivalent stiffness. The collapse
moment of the model is shown in Figure 5.9.
The theoretical results and FEM results are contrasted in Table 5.7. The
theoretical results have favorable agreement with the FEM results, which
shows that the two methods are valid to predict the mechanical behavior
of flexible pipes.
U, U2
+3.116e+01
+2.595e+01
+2.074e+01
+1.553e+01
+1.032e+01
+5.114e+00
−9.566e−02
−5.305e+00
−1.051e+01
−1.572e+01
−2.093e+01
−2.614e+01
−3.135e+01
Y ODB: Riks-ring-3h-02-final.odb Abaqus/Standard 6.114-1 Fri Jul 31 09:59:14 GMT+08:00 2015

Step: Step-2
Increment 197: Arc Length = 3.229
Z X Primary Var: U, U2
Deformed Var: U Deformation Scale factor: +1.000e+00
Figure 5.9 Collapse failure moment.

Table 5.7 Comparison between theoretical and FEM results.

Failure mode Theoretical results FEM results differential
Burst 57.2 MPa 53.5 MPa 6.92%
Tension Failure 2521 kN 2,611 kN 3.45%
Wet collapse 7.8 MPa 8.2 MPa 4.88%
It is worth noting that the Burst/Design ratio and the Collapse/Design ratio is 2.17 and 3.28,
respectively, which suggests that the pipe can successfully meet the design requirements.
5.7 Conclusion
In this chapter, a design procedure and a case study for a typical unbounded
flexible pipe is illustrated. Based on the design requirements, firstly the
cross-sectional configuration of the pipe was proposed. Then the possible
load conditions were presented and the theoretical method was used to
calculate the mechanical behaviors of this pipe under those load condi-
tions. The possible stresses/strains of each layer are then collected together
to verify the most dangerous load conditions for each layer, and the
utilization ratio of each layer which should be less than 1.0 are obtained.
Finally, the FEM model was established to verify the accuracy of the
designed flexible pipe.
It turns out that the theoretical and FEM method exhibit good agree-
ment. The basic idea of theoretical method is to use the principle of virtue
work to get the equivalent equilibrium of different layers and then assem-
bly them in the layers order. The helical construction of tape layer was
neglected for its low contributions to the strength capacity and thus it was
considered as cylindrical components.
In the FE model, the equivalent thickness and stiffness was introduced
to simulate the complicated profile of carcass and pressure armor layer to
save computational resources. And the results validated this simplification.
Further work includes the torsional and bending analysis. And the
bending may combine with other loads to form different load conditions.
The corresponding experiments are also necessary to validate the designed
cross-section.
References
1. API, Recommended Practice for Flexible Pipe, API RP 17B, Fifth Edition,
American Petroleum Institute, 2014.
2. API, Specification for Unbonded Flexible Pipe, API Specification 17J,
Furth Edition, American Petroleum Institute, 2014.
3. Feret, J. J., and C. L. Bournazel. “Calculation of stresses and slip in structural
layers of unbonded flexible pipes. “Journal of Offshore Mechanics and Arctic
Engineering 109. 3(1987): 263–269.
4. Witz, J. A. “A case study in the cross-section analysis of flexible risers.
“Marine Structures 9. 9(1996): 885–904.
5. S. Berge and A. Olufsen, Handbook on Design and Operation of Flexible
Pipes, SINTEF Report STF70, A92006, 1992.
6. Chen, B. and Nielsen, R., 1992, “Theoretical models for prediction of burst
& collapse and their verification by testing”, International seminar on recent
research and development within flexible pipe technology.
7. Colquhoun, R. S., R. T. Hill, and R. Nielsen. “Design and materials con-
siderations for high pressure flexible flowlines. “Advances in Subsea Pipeline
Engineering and Technology. Springer Netherlands, 1990. 145–178.
8. Neto, Alfredo Gay, et al. “Burst Prediction of Flexible Pipes. “ASME 2010
29th International Conference on Ocean, Offshore and Arctic Engineering.
American Society of Mechanical Engineers, 2010.
9. de Sousa, JoséRenato M., et al. “A study on the response of a flexible pipe to
combined axisymmetric loads. “ASME 2013 32nd International Conference
on Ocean, Offshore and Arctic Engineering. American Society of Mechanical
Engineers, 2013.
10. Design report for flexible dynamic riser, B725ENG006, Wellstream
Corporation.
11. Kebadze, Elizbar, 2000, “Theoretical modelling of unbonded flexible pipe
cross-sections”. South Bank University.
12. Knapp, R. H., 1979, “Derivation of a new stiffness matrix for helically
armoured cables considering tension and torsion”, International Journal for
Numerical Methods in Engineering, 14. 4, pp. 515–529.
6
Unbonded Flexible Pipe Under Bending
6.1 Introduction
Unbonded flexible pipes are helically reinforced structures with high axial
stiffness and low bending stiffness. The pipe consists of several layers, and
each layer has its own particular function. One layer is tensile armor made
of helical wires with a rectangular cross-section to carry either axial or
bending loads. The helical and armor layers represent the main character-
istic components of the helical structure. The behavior of the helical wire
is simple when the pipe is only exposed to axisymmetric loads because it
is assumed not to have relative sliding between the helical wire and under-
lying surface. However, pipe bending is complex and is caused not just by
slip mechanism but also by the non-axisymmetric status of the pipe.
Many studies on this issue have been completed. Love [1] developed
a set of nonlinear differential equations that describes the equilibrium of
thin rods under both bending and twist. Costello [2] assumed the helical
wire as a helical spring and applied Love’s equations to predict the bend-
ing stiffness of a helical spring during bending. Lutchansky [3] directly
derived the axial stress of the helical wire by assuming the slip is along
the original path. Knapp [4] assumed the helical wire follows the initial
line of contact between the wire and underlying cylindrical surface during
bending and investigated the axial strain of a helical wire with and without
friction. Feret and Bournazel [5] rigorously analyzed the behavior of an
unbonded flexible pipe and presented the wire stresses by assuming the
helical wire follows the geodesic path once slippage takes place. Raoof [6]
proposed that the wire bending stress is composed of two components: the
first component is the axial stress generated in the wires due to interwire/
interlayer shear interactions between the wires in a bent cable, and the
second component is associated with the wires bending about their own
axes. Witz and Tan [7] developed a general analytical model to predict the
relationship between curvature and bending moment of a flexural struc-
ture. It assumed that at a certain curvature slippage would take place in the
103
top region and the strain would distribute into a non-zero constant strain
state. Kraincanic and Kebadze [8] gave an analytical formula to determine
the bending moment–curvature relationship. As the curvature reaches the
minimum critical curvature, some parts of the helical wire will slip along
their own axes; however, the rest will not slip. Zhang and Qiu [9] developed
a numerical model to simulate the tensile behavior of a pipe bending. It
was assumed that all slippage along a wire is proportional to the geodesic.
In other words, the wire configuration during slipping as the pipe bends
can be completely represented by a single proportional factor. Savik [10]
proposed a numerical, finite element solution for the response of the heli-
cal wires of a flexible pipe under bending that included friction restraining
slippage. Corre and Probyn [11] focused on the validation of the 3-D FEM
for cyclic bending under a tension load case. The hysteretic curves of the
axial strain in helical wires subjected to bending were derived.
The mechanism of the unbonded flexible pipe under bending is com-
plicated because the helical wire can slip relatively to the underlying layer.
The slippage will dramatically change the strain of the helical wire which,
in turn, induces a large response in the pipe.
The bending analysis of an unbonded flexible pipe is based on Kebadze’s
theory model, including the changes of the interlayer pressure between
layers during the process. The theory model can get the axial force in heli-
cal wires and the bending stiffness of unbonded flexible pipes. The main
assumptions used throughout the analysis are follows:
1. Tendons can slide only along their own axes.

2. The pipe is subjected to a uniform bending which takes place
far from the terminations so that the end restraint effect can
be neglected.
3. Static and dynamic friction coefficients are equal.
6.2 Helical Layer Within No-Slip Range

During installation and operating condition, flexible pipes will have inter-
layer pressure between the layers. When the curvature of the flexible pipe
is small, the helical wire can’t overcome the friction between the layers and
relative slip will not occur.
6.2.1 Geometry of Helical Layer

When a cylinder is subjected to a pure bending, it is formed into a toroid
with a constant curvature, as indicated in Figure 6.1.
Unbonded Flexible Pipe Under Bending 105
e2
Oc e3
aw
e1
ac E2 ρ=1/κ
E1
E3
Oρ
Figure 6.1 Reference systems for toroid surface.
In the figure, the helical layer contains two basic components: the cylin-
der and helical wire.
To describe the geometry of the wire in the deformed cable, a referential
coordinate system with origin is defined, the right-handed orthonormal
set of basis vectors, is shown in Figure 6.1. The vectors and define the ver-
tical plane in which the centerline of the cable bends. A local right-handed
orthonormal coordinate system is attached to the cross-section of the cable
at its center, such that is tangent to the cable centerline and is parallel to.
The orthonormal local basis can then be expressed in terms of components
on the reference basis as
e1 = (cosφ, −sinφ, 0)
e2 = (sinφ, cosφ, 0) (6.1)
e3 = (0,0,1)
With respect to the reference coordinate system, the center of the wire
in the cable cross-section is located at
a = ac + aw (6.2)
where is the vector from to the center of the cable cross-section as shown
in Figure 6.2.
e2
Oc e3 (Neutral Axis)
aw
θi
Helcial wire i
Figure 6.2 Cross-section of helical layer.
At this point, it is assumed that the cable is deformed into a curve with
constant curvature. Letting denote the constant radius of curvature mea-
sured from along. Furthermore, using a polar coordinate system, where R
is the distance from the center of the cable to the center of the wire and is
constant for a given layer, is the polar angle from to, and both are defined
on the cable cross-section as shown in Figure 6.2.
Substituting these relations into Eq. (6.3), one obtains
 ( ρ − Rcosθ )sinϕ 
 
a =  ( ρ − Rcosθ )cosϕ  (6.3)
 − Rsinθ 
 
For a curve on a toroid as shown in Figure 6.3, a curvilinear coordinate

triad constituted by a tangent t, a normal n and a binormal vector b is given
by
da
t=
ds
dt/ds (6.4)
n=
dt/ds
b = t ×n
b n
Oc
Figure 6.3 Darboux frame.
In the differential geometry of surfaces, a Darboux frame is a natural

moving frame constructed on a surface
 dt 
 
 ds   0 κ n −κ g   t 
 dn    
 ds  =  −κ n 0 τ  n  (6.5)
   κ g −τ 0   b 
 db   
 ds 
 
where is the geodesic curvature of the curve, is the normal curvature of the
curve, and is the relative torsion (also called geodesic torsion) of the curve.
The following definitions of the curvature components are shown:
dt dn
κn = n⋅ = −t ⋅
ds ds
db dt (6.6)
κ g = t ⋅ = −b ⋅
ds ds
dn db
τ = b⋅ = n⋅
ds ds
Along the helical wire, the arc length coordinate is expressed as

cosα
ϕ= s (6.7)
ρ
sinα
θ= s (6.8)
R
where is the lay angle of the helical wire and is the arc length of the helical
wire.
The longitudinal component of the Green strain tensor along the rod
center line is further determined by considering the displacement relative
to the undeformed reference system
1 1
E11 = (t ⋅ t − t 0 ⋅ t 0 ) = − Rcos 2α cosθκ + ( Rcosα cosθκ )2 (6.9)
2 2
6.2.2 Bending Stiffness of Helical Layer

The axial strain of the helical wire under bending deformation is shown in
Eq. (6.10) by ignoring the higher order small amount
εib = −Rcos2αcosθiκ (6.10)
where is a different helical wire in helical layer. Substituting Eq. (6.4) in

Eq. (6.6),
sin 2α (6.11)
κn = − cos 4α cosθκ
R
κg = −cosα(1 + sin2α)sinθκ (6.12)
sinα cosα
τ= + sinα cos 3α cosθκ (6.13)
R
The strain energy of the helical wire is obtained, as shown below.
∫ ( EAε )
s
1 2
Ui = ib + EIb∆κ n2 + EIn∆κ g2 + GJ∆τ 2 ds (6.14)
2 0
As the flexible pipe is subjected to pure bending, the potential of exter-

nal force is
W = MκL (6.15)
where is the bending moment at the end of the flexible pipe and is the
length of the flexible pipe.
For a static system,
Π=U–W (6.16)
According to the principle of stationary potential energy, bending stiff-

ness is achieved
δΠ = δU – δW = 0 (6.17)
1
Mns = nEAR 2cos 3ακ (6.18)
2
1
EIns = nEAR 2cos 3α (6.19)
2
From Eqs. (6.18) and (6.19), the bending stiffness of the helical layer is
linear when no relative slip occurs.
6.3 Helical Layer Within Slip Range

When an unbonded flexible pipe undergoes bending deformation and the
curvature is greater than a certain critical value, the helical wire can over-
come the friction between layers. By the increasing of the curvature, the
sliding area will be larger and the axial stress of the helical wire will be
reduced due to slippage, reducing the bending stiffness of the helical layer.
6.3.1 Critical Curvature

Figure 6.4 shows the axial force in helical wire. As the bending curvature
reaches a certain critical value called the critical curvature, the helical wire
begins to slip. When slippage occurs
Pt+dPt
Pt
fds
ds Pt+dPt Pt
ds
dz
Figure 6.4 Micro-section of helical wire.
dPt
≥f (6.20)
ds
where is tension in the helical wire and is the friction of the helical wire.
The tension in the helical wire contains two parts:
Pt = Pta + Ptb (6.21)
where is the tension induced by the axisymmetric load and is the tension
induced by the bending.
The uniform radial load produced by the helical wire is shown below:
pn = −κnPt (6.22)
Considering the external pressure load, the friction of the helical wire is
f = μi(κnPt + Pow) + μoPow (6.23)
where is the friction coefficient between the helical wire and inner tube
and is the friction coefficient between the helical wire and external tube.
Substituting Eq. (6.21) into Eq. (6.23)
f = f0 – μiκnEARcos2αcosθκ (6.24)
f0 = μi(κnPta + Pow) + μoPow (6.25)

The curvature of the helical wire along the arc length is variable under
bending deformation, but compared to the initial value, the change part is
small. Eq. (6.20) can be simplified as
f0 ≤ κEA sinα cos2α(μisinα cosθ + sinθ) (6.26)
By Eq. (6.26), the critical curvature is derived as
f0
κ cr = (6.27)
EAsinα cos 2α 1 + µi2 sin 2α
6.3.2 Axial Force in Helical Wire Within Slip Range

When the curvature exceeds the critical curvature, slip occurs and the axial
balance is satisfied,
dPst
=f (6.28)
ds
According to the Eq. (6.8),
dPst sinα
⋅ = µi (κ n Pt + Pow ) + µo Pow (6.29)
dθ R
Under bending, deformation is very small, so it is reasonable to ignore

the influence of
dPst R
− µi sinα Pt − ( µi + µo )Pow = 0 (6.30)
dθ sinα
Solving the differential equation
R
Pst = D1e µiθ sinα + ( µi + µo )Pow (6.31)
µi sin 2α
where.
R
Pta ( µi + µo )Pow
µi sin 2 α
D1 =
e µi (π /2 )sinα
6.3.3 Axial Force in Helical Wire Within No-Slip Range

When the bending curvature is greater than the critical curvature, the
helical wire will slip. The helical wire slides by gradually increasing the
curvature and spreading the slippage area. The sliding mechanism can be
validated through a simplified model as shown in Figure 6.5. A beam of
rectangular cross-section is set on the ground and the ends of the beam are
fixed. In the first stage, the beam is subjected to a horizontal concentrated
force at the middle point and there is no friction. In the second stage, the
uniformly distributed load and friction are applied. In the third stage, an
opposite direction concentrated force is applied at the middle point of the
beam.
Finite element software ABAQUS was used to simulate the process and
obtain the axial displacement of the beam, as shown in Figure 6.6. When
there is no friction, all points of the beam slide. When friction is applied,
the sliding part is divided into three regions: the slip area, transition area,
and no-slip area. The slide area is near the midpoint of beam and the size of
the sliding zone is related to the friction and concentration force.
Figure 6.7 describes the change in axial force in the beam. In slip area
C, the axial force distribution of the beam is a straight line and the slope of
the curve is the friction. In the transition area B, the axial force of the slope
becomes smaller as it approaches 0.
According to the Eq. (6.10), the axial force of helical wire in the no-slip
stage is shown in Figure 6.8.
F F' F
(a) (b)
Figure 6.5 The sliding mechanism model.

0.07
First stage
Third stage
The axial displacement of beam, [mm]
0.06
0.05
0.04
No-slip area Slip area No-slip area
0.03 A C A
Transition area Transition area
0.02 B B
0.01
0
0 20 40 60 80 100
The length of beam, [mm]
Figure 6.6 The axial displacement of beam.
Considering symmetry, when the curvature is greater than the critical

curvature, and are the critical points. When is between the helical wire will
slip. In Figure 6.7, the axial force is continuous at the critical point, so the
axial force of sliding part is as follows [12]:
Pnst(θ) = Pst(θ1) – EARcos2α(cosθ – cosθ1)κ (6.32)
30
First stage
Third stage
20
The axial force of beam, [mm]
10
−10
−20
−30
0 20 40 60 80 100
The length of beam, [mm]
Figure 6.7 The axial force of beam.

Pt
0 θ1 θ2 π 2π θ
Figure 6.8 The axial force of helical wire.
Pnst(θ) = Pst(θ2) – EARcos2α(cosθ – cosθ2)κ
The first derivative of the axial force at the critical point has to be con-
tinuous as well:
dPnst R
− µi sinα Pnst − ( µi + µo )Pow = 0 (6.33)
dθ sinα
Combining Eqs. (6.31) and (6.32), the axial force of the helical wire can
be obtained.
6.3.4 Bending Stiffness of Helical Layer

When the bending curvature is less than the critical curvature of the
unbonded flexible pipe, the helical wire will not slip. The bending stiffness
of the helical layer can be obtained with Eq. (6.19). When the bending cur-
vature is greater than the critical curvature, slippage occurs. The bending
moment mainly comes from the axial force of the helical wires. The helical
layer is equivalent to a tube, as shown in Figure 6.9.
With Figure 6.9, the bending moment of the helical layer can be obtained:
 θ1 θ2 π 
M = −2 fc 
 ∫
0
σ ns +
∫
θ1
σs +
∫
θ1
σ ns  Rcosθ cos 2α Rtdθ (6.34)

e2
σ
σcosα
Rdθcosα
Rdθ
α
2πR
Oc e3
R
Rcosθ
θ
w
t dz
Figure 6.9 The axial force of helical wire.
where is the axial stress in no-slip part and is the axial stress in the slipping
part.
In order to ensure the helical layer and equivalent tube have the same
area, a filling coefficient is introduced.
nw
fc = (6.35)
2π Rcosα
Substituting Eqs. (6.31) and (6.32) into Eq. (6.34),
nEAR 2cos 3ακ nµ sinα cosα RD1

M= f1 (θ1 ,θ 2 ) + i f 2 (θ1 ,θ 2 )
2π (
π 1 + µi2 sin 2α )
1
f1 (θ1 ,θ 2 ) = (sin2θ 2 − sin2θ1 ) − (θ 2 − θ1 ) + π
2
f 2 (θ1 ,θ 2 ) = e µiθ2 sinα ( µi sinα sinθ 2 − cosθ 2 ) − e µiθ1sinα ( µi sinα sinθ1 − cosθ1 )
(6.36)
The bending stiffness is also derived.
nEAR 2cos 3α
EI s = f1 (θ1 ,θ 2 ) (6.37)
2π
References
1. E. H. Love. A treatise on the mathematical theory of elasticity: Cambridge
University Press;1944.
2. G. A. Costello. Theory of wire rope: Springer;1990.
3. M. Lutchansky. Axial stresses in armor wires of bent submarine cables.
Journal of Manufacturing Science and Engineering. 1969;91: 687–91.
4. R. H. Knapp. Helical wire stresses in bent cables. Journal of Offshore
Mechanics and Arctic Engineering. 1988;110: 55–61.
5. J. Feret, C. Bournazel. Calculation of stresses and slip in structural lay-
ers of unbonded flexible pipes. Journal of Offshore Mechanics and Arctic
Engineering. 1987;109: 263–9.
6. M. Raoof, Y. Huang. Wire stress calculations in helical strands undergoing
bending. Journal of Offshore Mechanics and Arctic Engineering. 1992;114:
212–9.
7. J. Witz, Z. Tan. On the flexural structural behaviour of flexible pipes, umbil-
icals and marine cables. Marine structures. 1992;5: 229–49.
8. I.Kraincanic, E. Kebadze. Slip initiation and progression in helical armouring
layers of unbonded flexible pipes and its effect on pipe bending behaviour.
The Journal of Strain Analysis for Engineering Design. 2001;36: 265–75.
9. Y. Zhang, L. Qiu. Numerical model to simulate tensile wire behavior in
unbonded flexible pipe during bending. ASME 2007 26th International
Conference on Offshore Mechanics and Arctic Engineering: American
Society of Mechanical Engineers;2007. p. 17–29.
10. S. Sævik, J. Gjøsteen. Strength Analysis Modelling of Flexible Umbilical
Members for Marine Structures. Journal of Applied Mathematics. 2012;2012.
11. V. L. Corre, I. Probyn. Validation of a 3-Dimensional Finite Element Analysis
Model of Deep Water Steel Tube Umbilical in Combined Tension and Cyclic
Bending. In: Proceeding of OMAE: ASME;2009.
12. L. Dong, Y. Huang, G. Dong, Q. Zhang, G. Liu. Behavior of tensile wires in
unbonded flexible pipe under compression and design optimization for pre-
vention. Marine structures. 2015;42: 154–174.
7
Coiling of Flexible Pipes
7.1 Introduction
Composite flexible pipes used in ocean engineering have become more
and more popular due to their special properties, such as corrosion resis-
tance and strong flexibility, which makes this type of pipes easy to trans-
port, install, and operate. As a relatively new type of composite flexible
pipe, metallic strip flexible pipe (MSFP) is composed of two HDPE lay-
ers separated by four steel strips reinforcement annular. This pipe with
a sandwich structure is a particular kind of composite characterized by
the combination of two different materials, contributing with their single
properties to the global structure performance. Compared to other types
of composite pipe, MSFP has advantages in relatively cheap reinforcement
material and a simpler manufacturing process, which results in low pro-
duction costs. Thus, they are regarded as a popular choice for submarine
pipelines transporting oil and gas.
During the practical manufacturing process, ready-made flexible pipe-
lines are usually twined around a coiling drum, as shown in Figure 7.1, so
that much more space can be saved. It is clear that the smaller the drum’s
radius is, the less space it will take. Thus, more pipes can be transported at
one time and the cost will be reduced. However, the lack of an interlocked
carcass layer in MFSP may lead to a weak radial stiffness. For example, the
bending-induced radial pressure may provoke buckling failures (Figure
7.2) with possible radial formations as the reeling process is usually con-
ducted with low tension and bending. Pipe mechanics analysis in reeling
operation is illustrated in Figure 7.3.
With considerable attention has been paid to the security and reliability
of flexible pipes in the laying process, literature on its mechanical behavior
during reeling operation has emerged relatively slowly and in a more
scattered way. Szczoka [1] adopted the rigid finite element method to
present the static and dynamic response of an offshore pipe reel-laying
process. Ruan et al. [2] presented a safe assessment of a floating platform
117
Figure 7.1 Pipe coiled in the reel drum.
Figure 7.2 Buckling failure of MSFP.
during pipelaying installation based on numerical and experimental meth-

ods, in which tension and bending behavior of flexible pipe are considered.
These researches focused on the mechanical performance on the intact
pipeline but ignored the defect caused before pipelaying process, which
would result in inaccurate analysis. Longva and Sævik [3] proposed a new
Lagrangian-Eulerian formulation for the reeling analysis, claiming that
the torsion is the reason causing spiraling or various forms of damage to
the tensile armor. Their following research [4] stated that torsional fail-
ures occurred to pipes transferred along a vertical-axis turn-tables placed
in confined cargo holds. Four possible failure modes of pipes under
Coiling of Flexible Pipes 119
FT2
MSFP
FR1
FT1
FR2 M
Coiling drum
FR2 FT1 M
FR1
M
FT2 MSFP
qR
FT1
Coiling drum
FT1
M
Figure 7.3 Pipe mechanics analysis in reeling.
severe curvature deformations and large rotations are given based on

finite element simulation. Maincon [5] pointed out that the flexible pipe
may start to roll along its axis under some conditions and result in the “pig
tailing” failure that makes the flexible pipe take a spiral shape. Therefore,
engineers should consider not only the mechanical behavior during the
laying process but also the possible buckling failure and other imperfect
resulted from reeling operation.
Over the past decades, a lot of effort has been made in the field of com-
posite flexible pipes under tension and bending. Xu et al. [6] investigated
the mechanical behavior of fiberglass reinforced bonded flexible pipe sub-
jected to tension by experimental, theoretical, and finite element method,
proving that the outer HDPE layer provides tension resistance during the
loading process. Bahtui et al. [7] used finite element analysis software
ABAQUS to simulate a flexible pipe subjected to different loads. In this
model, contact elements are defined between each layer while no inter-
action is considered between tendons of helical armor layers. Bai and his
cooperators have also completed a series of experiments and theoretical
analysis about metallic pipes under various loadings, such as external
pressure [8], torsion [9], and combined loadings [10]. However, these
researches focused on the ultimate strength of the estimated flexible pipe,
the mechanical responses of MSFP during reeling operation are still quite
complicated and not fully understood. The pipeline coiled on the reeling
drum is in the form of space spiral. Besides, the friction and complex
interactions between multiple layers in unbonded pipes greatly increase
the difficulty of mathematical derivation and finite element analysis. The
layered combination of two materials also contributes to the nonlinear
behavior of MSFP. Therefore, experimental methods would be a favorable
way to explore the mechanical behavior of MSFP.
The objective of this chapter is to predict the global mechanical behavior
of MSFP during the reeling operation before transportation. Laboratorial
tests of MSFP under tension and bending are conducted to acquire its non-
linear tension-strain and moment-curvature relation, which would be
introduced into the global model. A finite element model with MSFP,
bearing plate and coiling drum is built to obtain the global mechanical
responses during the reeling operation. Additional parameter analyses are
done in ABAQUS to acquire the influence factors. The given conclusions
in this chapter may provide a comprehensive concept for understanding
the mechanical behavior of MSFP during reeling operation and a valuable
reference for optimizing its design.
7.2 Local Analysis

7.2.1 Dimensions and Material Characteristics
The MSFP analyzed in this chapter is produced through the helical
tape wrapping method. After four reinforcement layers (two steel strips
in each layer) are wound helically on the inner HDPE layer, the papery
polyester tapes (PETs) are then wrapped over the steel strips to prevent
them from curling up, otherwise, it may lead to the piercing of the outer
HDPE layer surrounded outside. The inner two reinforcement layers are
wound at 54.7° in clockwise direction while the other two reinforcement
layers are wound at 54.7° in anticlockwise direction. Figures 7.4 and 7.5
show the cross-section and the outermost reinforcement layer of MSFP.
The mechanical characteristic of HDPE and steel strips used for MSFP are
measured by uniaxial tensile tests and the stress-strain curves are shown in
Figures 7.6 and 7.7.
7.2.2 Tension Test

As shown in Figure 7.8, the tensile test is conducted by a 3,000 kN
electromagnetic servo-controlled testing machine. The specimens are
6 mm 4*0.5 mm 4 mm Steel strip reinforcement layers
Inner HDPE layer Outer HDPE sheath
Inter diameter
50 mm
Outer diameter 74 mm
Figure 7. 4 Cross-section of MSFP.
Figure 7.5 MSFP with partly outer sheath peeling off.
25
20
15
Stress (MPa)
10
0
0 0.05 0.1 0.15 0.2 0.25
Strain
Figure 7.6 Stress-strain curves of HDPE.

1000
800
Stress (MPa) 600
400
200
0
0 0.005 0.01 0.015 0.02
Strain
Figure 7.7 Stress-strain curves of steel strip.
End fitting
340 mm
Test section
1100 mm
End fitting
340 mm
Figure 7.8 Tensile test of MSFP.
connected to the machine by two flanges to avoid the test failure caused
by specimens slipping during the test. The axis of specimens is aligned
with fixtures of the testing machine to make sure that uniformly load is
applied. According to (ASTM) D2105-2001 [11], the loading speed is set
as 1 mm/s. The tensile tests were performed on two sets of MSFP speci-
mens, and the corresponding tension-extension curves for two specimens
are illustrated in Figure 7.9.
80
Specimen 1
Specimen 2
60
Load (kN)
40
20
0
0 50 100 150 200 250 300
Extension (mm)
Figure 7.9 Tension-extension curves of two specimens.
The trend of tension-extension relation is similar to the HDPE stress-

strain relation when the elongation is below 10%. The yield stress of
HDPE is calculated as 21 MPa, which is a bit lower than that of the test
specimens. After the elongation exceeds a certain value, the outer HDPE
cover cracks and the ensuing descending trajectory in those two curves
could be observed, albeit at different extensions.
7.2.3 Bending Test

As shown in Figure 7.10, the bending test is conducted by a four-point
bending machine. The test was carried out in the horizontal plane to elimi-
nate the influence of the specimen’s deadweight. Because the part between
the support roller and loading roller is a rigid section, the test section can
be considered as being subjected to pure bending. The test condition is
limited by manual loading, so the loading process must be slow, stable,
and constant by keeping the speed of the loading beam at about 0.6 mm/s.
Figure 7.11 gives the deformation of MFSP specimen during bending
test. The time history of displacement and load are recorded by a displace-
ment gauge and load sensor, respectively. The relationship between the
bending moment and curvature can be deduced by data collected from
the tests and geometry properties of the loading machine. The moment-
curvature curves of two sets of MSFP specimens are shown in Figure 7.12.
The bending moment rises as the curvature increases and the two test curves
are close to each other. Although the test curves have a slight fluctuation
2F
displacement gauge
Loading beam
Loading roller
Rigid region
r
Δ Test section
θ
r
Support
roller
Test section
300 mm 600 mm
800 mm
1400 mm 300 mm
l L1 l
L2
Figure 7.10 Diagrammatic sketch of bending machine.
Figure 7.11 Moment-curvature curves of two specimens.
due to the man-made operation, the linear relationship between bending

moment and curvature can still be observed.
7.2.4 Summary
In this section, tensile and bending tests of MSFP are conducted to
acquire its tension-extension and moment-curvature relationship. The
experimental results are input into the following global model established
by the finite element method and compared with the simulation results to
reveal the interaction between the combined loads. Besides, the obtained
2000 Specimen 1
Specimen 2
Bending moment (N•M)
1500
1000
500
0
0.0 0.5 1.0 1.5 2.0 2.5 3.0
Curvature (rad/m)
Figure 7.12 MFSP specimen before and after bending.
mechanical properties can also be used as a criterion for determining the

failure of the pipeline in engineering applications.
Due to the uncontrollable factors during the test process, the obtained
curves show certain volatility. Therefore, a fitting program is employed to
obtain the smooth tension-extension curve before the pipe out of efficiency,
as shown in Figure 7.13. The elastic modulus is calculated as the secant
70
Specimen 1
Specimen 2
60 Fitting curve
50
15
40
Load (kN)
12
30
9
20
10 6
0.5 1.0 1.5 2.0 2.5 3.0
0
0 20 40 60 80 100 120 140 160 180
Extension (mm)
Figure 7.13 Fitting tension-extension curve.

1500
Specimen 1
Specimen 2
Bending moment (N.M) Fitting curve
1000
500
0
0.0 0.5 1.0 1.5 2.0
Curvature (rad/m)
Figure 7.14 Fitting bending-curvature curve.
modulus when the strain is 0.05% and 0.25% based on the standard
ISO527-2012 [12], and the value is taken as 978 MPa. Moreover, the bend-
ing behavior of MSFP is presented by a typical bi-linear bending model
[13], as shown in Figure 7.14.
The ready-made flexible pipes are usually tens of meters or even hun-
dreds of meters, so it is impossible to simulate reeling operation without
any simplification of the structure considering the limitation of compu-
tational costs. In this chapter, the reeling pipeline, therefore, is simplified
as a long beam owing properties akin to original MSFP in the following
global analysis.
7.3 Global Analysis

7.3.1 Modeling
The global model, shown in Figure 7.15, consists of a coiling drum, a
simplified pipeline, and a bearing plate. The coiling drum and bearing
plate employ the discrete rigid element so that no deformation will hap-
pen in the two parts. The pipeline, however, employs deformable beam
elements. The local direction of the beam element is defined as shown in
Figure 7.16.
The radius of the coiling drum should be no more than 2 m for conve-
niently transporting. The coiling drum will sink slowly in the y direction
when the pipe is rolled up so that the pipe can spirally attach on the drum.
The length of the pipeline is set as 50 m, which is able to wind two
circles around the coiling drum. The bearing plate (length, 2 m; width,
Z X
Figure 7.15 The global model of reeling operation.
Figure 7.16 The local direction of the beam element.
1 m) is employed to restrict the horizontal and vertical movement of the

pipeline during the reeling operation.
7.3.2 Interaction and Mesh

The interactions between the pipeline and the bearing plate as well as the
pipeline and the coiling drum are set as surface-to-surface contact. In
this scheme, because the coiling drum and bearing plate are rigid bodies,
they are naturally set as master surfaces whose meshes should be coarser
than those of the slave surface. Penalty function is taken as the friction
formulation for tangential. The friction coefficient between the coiling
drum and the pipeline is set as 0.2 while the contact condition between the
bearing plate and the pipeline is frictionless.
Z X
Figure 7.17 Mesh condition of the global model.
Three-dimensional linear interpolation beam (B31) is selected for the

pipeline because it is suitable for analyzing contact problems and also can
take shear deformation into consideration, while R3D3 element is used for
coiling drum and bearing plate. The mesh condition of the global model is
7.3.3 Load and Boundary Conditions

The coiling drum should sink down in y direction to make sure that the
pipeline can spirally wind in an expected way. The rotation angle is 12.6
rad and the sinking distance is 0.2 m, which means the drum sinks down
0.1 m when the pipeline coils around once. It should be noted that there
is a 20,000 N pulling force applied at the end of the pipeline to keep it
straight, which is the same as the practical reeling operation. At the other
end of the pipe, the start point of the pipe is coupled with the coiling drum
so that they can move together. The load and boundary conditions are
7.3.4 Discussion of the Results

The pipeline to be reeled keep a straight line due to the tensile force on the
end and the restriction of the bearing plate when the coiling drum whirls
as well as sinks down. The result (Figure 7.19) shows that almost two cir-
cles of MSFP adhere to the surface of the coiling drum.
During the whole process, tension (SF1) along the pipeline is quite sim-
ilar everywhere, almost equaling to the pulling force applied at the end.
The magnitude of transverse shear force in local 2 direction (SF2) and local
Z X
Figure 7.18 Load and boundary condition of the global model.
Figure 7.19 The final deformation of the global model.
1-2 plane (SF3), however, are relatively small. Figures 7.20, 7.21, and 7.22
present the contour plots of forces in three directions when the reeling
is finished. It can be observed that these forces are quite average in the
straight section but differ in coiled section.
In order to obtain the stress along the pipeline after winding, a path is
established along the axial direction of the pipeline, as shown in Figure 7.23.
The nodes are allocated on the pipe every other 0.1 m. The endpoint tension
applied at is Node 1 and the endpoint adhered to the coiling drum surface
is Node 501. Whereas the true distance along the path is defined from Node
501, which means the true distance at Node 501 is 0 m and the true distance
at Node 1 is 50 m. Figure 7.24 shows the variation of tension along the
true distance. It can be observed that SF1 has fluctuations, but the varia-
tion is quite small. The difference between the minimum value (18,544 N)
and the maximum value (20,000 N) is 7.3%. Besides, the axial force of
SF, SF1
(Avg: 75%)
+2.007e+04
+1.993e+04
+1.979e+04
+1.965e+04
+1.951e+04
+1.936e+04
+1.922e+04
+1.908e+04
+1.894e+04
+1.880e+04
+1.866e+04
+1.852e+04
+1.837e+04
Figure 7.20 SF1 of the pipeline.
SF, SF2
(Avg: 75%)
+8.205e+01
+6.467e+01
+4.728e+01
+2.990e+01
+1.251e+01
-4.874e+00
-2.226e+01
-3.965e+01
-5.703e+01
-7.442e+01
-9.180e+01
-1.092e+02
-1.266e+02
SF, SF3
(Avg: 75%)
+5.703e+02
+3.699e+02
+1.696e+02
-3.077e+01
-2.311e+02
-4.315e+02
-6.318e+02
-8.322e+02
-1.033e+03
-1.233e+03
-1.433e+03
-1.634e+03
-1.834e+03

End: 1
6 11
16 21
2631
3641
4651
56
6166
7176
8186
91
96
101
106
111116
121126
131
136141
146
151156
161166
171176
181186
191196
201
206
211216
221226
231
23
Figure 7.23 A picked path for the pipeline.
20000
19500
SF1 (N)
19000
18500
0 5 10 15 20 25 30 35 40 45 50 55
True distance along the path (m)
Figure 7.24 SF1 along the path.
the pipeline coiled on the drum is smaller than the unwinding part, which
means bending behavior reduces the effect of axial force to some extent.
Another load drawing more attention is the bending moment in the
reeling plane, especially when the radius of the coiling drum is small.
Excessive curvature resulted in the small radius would probably cause
the buckling failure of the pipeline. The twisting moment is so small that
can be neglected. Figure 7.25 shows the contour plots of bending moment
(SM2) around local 1 direction when the reeling is finished, and Figure
7.26 gives the relationship of corresponding bending moment and true
SM, SM2
(Avg: 75%)
+6.565e+02
+6.011e+02
+5.457e+02
+4.904e+02
+4.350e+02
+3.796e+02
+3.242e+02
+2.688e+02
+2.135e+02
+1.581e+02
+1.027e+02
+4.732e+01
-8.055e+00
Figure 7.25 Contour plot of SM2 along the path.
700
600
500
400
SM2(N·m)
300
200
100
-100
0 5 10 15 20 25 30 35 40 45 50 55
Figure 7.26 SM2 along the path.
distance along the path. It can be observed that the distribution of bend-
ing moment along the reeled pipeline is even, slightly fluctuating around
540 N⋅m. Meanwhile, the unrolled pipeline does not show the existence
of the bending moment. According to the MSFP bending test (Figure
7.14), the corresponding bending moment obtained by the curvature of
the coiling drum is 401 N⋅m, which is smaller than the simulating result.
The difference occurred possibly because the reeled pipeline is no lon-
ger a standard circle resulted from the downward movement of the coiling
drum. Besides, the existence of the tension in the pipeline also causes the
rise of the bending moment.
Figure 7.27 presents the contour plots of SM3 distributed along the
pipe when the reeling is finished. The bending moment around local-3
direction (SM3) may be caused by the deadweight of the pipeline and the
component force induced by space helical structure. It can be observed
that the bending moment in this direction is not so large. SM3 distributed
along the above path is shown in Figure 7.28. The bending moment SM3
of the start point of the pipeline is 0 N⋅m and the magnitude of bend-
ing moment increases with the true distance raising. The largest bending
SM, SM3
(Avg: 75%)
+1.542e+01
+1.016e+01
+4.908e+00
-3.466e-01
-5.601e+00
-1.086e+01
-1.611e+01
-2.137e+01
-2.662e+01
-3.188e+01
-3.713e+01
-4.238e+01
-4.764e+01
Figure 7.27 Contour plot of SM3 along the path.
10
-10
SM3 (N·m)
-20
-30
-40
-50
0 10 20 30 40 50
Figure 7.28 SM3 along the path.

moment is 45.28 N⋅m, located on the attached point between the pipe-
line and the bearing plate. The magnitude of SM3 then drops to 0 N⋅m.
Therefore, engineers should pay more attention to the end of the reeled
pipeline near the bearing plate in the practical reeling operation.

In this section, factors influencing the mechanical behaviors of MSFP
during reeling operations are investigated by the finite element method.
Other parameters should stay unchanged while only one of them is changed.
7.4.1 Diameter of the Coiling Drum

In the practical reeling process, the mechanical behavior of the pipeline
can be changed by adjusting the diameter of the coiling drum, so that
local buckling failure or other forms of damage can be avoided in coiling,
loading, and transporting. The diameter of the coiling drum is limited to
4 m due to convenient transportation and space limitation, so the curva-
ture shall not be too large. Thus, the diameter is restricted from 3 to 4 m.
Therefore, 3 m (Case 1), 3.2 m (Case 2), 3.4 m (Case 3), and 4 m (Case 4)
are selected for the coiling drum’s diameter in the parameter analysis,
respectively.
The variables that should be paid more attention to are SF1, SM2, and
SM3. The comparison results of them in each case are shown in Figures
7.29, 7.30, and 7.31. The variations of SF1 in four analyses are not so
obvious. It can be observed that SF1 in four conditions increases to the
applied axial force of 20,000 N after the pipeline leaving the bearing
plate. The bending moment SM3 increases linearly first and then suddenly
decreases to nearly zero. In addition, the larger the diameter of the coiling
drum is, the smaller the maximum bending moment SM3 is, and the slope
of the corresponding curve is also smaller. The coiling loop of pipeline
is changed by the diameter of the drum, which results in different pipe
weight and spatial spiral shape of the pipeline. Table 7.1 shows that SM3
is sensitive to the enlarge of drum diameter. The difference between the
maximum and minimum value is 46%.
Similarly, the larger the outside diameter of the drum is, the smaller the
maximum bending moment SM2 is. The difference between the maximum
and minimum value is 15%.
Therefore, the diameter of the coiling drum has a significant influence
on the mechanical behavior of MSFP, especially on the combined bending
20100
19800
19500
19200
SF1 (N)
18900
Case1
18600 Case2
Case3
Case4
18300
18000
0 5 10 15 20 25 30 35 40 45 50 55
Figure 7.29 SF1 along the path in different coiling drum diameter.
20
-20
SM3 (N·m)
-40
Case 1
-60 Case 2
Case 3
Case 4
-80
-100
0 5 10 15 20 25 30 35 40 45 50 55
Figure 7.30 SM3 along the path in different coiling drum diameter.
moment. In the actual reeling process, the bending moment will decrease
by adding the diameter of the coiling drum.
7.4.2 Sinking Distance of the Coiling Drum

In this analysis, the sinking distances are 0.15 m (Case 1), 0.2 m (Case 2),
and 0.3 m (Case 3). The winding angle of MSFP on the drum can be
altered in this way. The variations of SF1, SM3, and SM2 along the path of
800
600
SM2 (N·m)
Case1
400
Case2
Case3
Case4
200
0 5 10 15 20 25 30 35 40 45 50 55
Figure 7.31 SM2 along the path in different coiling drum diameter.
Table 7.1 The maximum SM3 and SM2 in

different coiling drum diameter.
SM3 (N·m) SM2 (N·m)
Case 1 84.23 785.20
Case 2 60.73 639.23
Case 3 58.97 598.99
Case 4 45.28 589.04
20100
19800
19500
19200
SF1 (N)
18900
18600 Case1
Case2
18300 Case3
Case4
18000
0 5 10 15 20 25 30 35 40 45 50 55
True distance along path (m)
Figure 7.32 SF1 along the path in different sinking distance.

each case are shown in Figures 7.32, 7.33, and 7.34, respectively. It can be
found that the axial force SF1 changes slightly in each case and there is
no big difference with each other. Bending moment SM3 has a linear
increasing trend. The greater the sinking distance is, the larger the winding
angle of MSFP will be achieved. The maximum SM3 in each case is shown
in Table 7.2.
20
-20
SM3 (N·m)
Case 1
-40 Case 2
Case 3
Case 4
-60
0 5 10 15 20 25 30 35 40 45 50 55
Figure 7.33 SM3 along the path in different sinking distance.
600
400
SM2 (N·m)
Case1
Case2
200 Case3
Case4
0 5 10 15 20 25 30 35 40 45 50 55
Figure 7.34 SM2 along the path in different sinking distance.

Table 7.2 The maximum SM3

in different sinking distance.
SM3 (N·m)
Case 1 34.17
Case 2 45.28
Case 3 56.69
Case 4 67.76
The difference between the maximum value and the minimum value
is 50%, indicating that the sinking distance has a great impact on SM3.
SM2 for each case, however, is almost identical. This is because the outer
diameters of the coiling drums remain the same so that the curvatures of
the reeled pipes also remain quite similar.
In a word, sinking distance of the drum has a significant influence on
SM3 but the slight effect on SM2. The sinking distance should be con-
trol to a certain value during the reeling process.
7.4.3 Reeling Length

In the actual reeling progress, the winding pipeline on the coiling drum
is often multiple loops, which can ensure the integrity and uniformity
of the pipeline. The winding pipeline can wind up to 10 circles. In this
analysis, however, the numbers of winding circles are set to 1 (Case 1),
1.5 (Case 2), 2 (Case 3), and 2.5 (Case 4). The variations of SF1, SM3, and
SM2 along the path of each case are shown in Figures 7.35, 7.36, and 7.37,
respectively.
It can be found that the longer the winding length of the pipeline is, the
shorter the pipeline length at which the maximum axial force 20,000 N is
reached, and the larger the fluctuation area is. For the bending moment
SM3, the longer the winding length is, the larger the maximum bend-
ing moment will be. In other words, the larger the reeling length is, the
larger SM3 will be. The section between the bearing plate and the coiling
drum should be given more attention. The values of the maximum bend-
ing moment in each case are shown in Table 7.3. The difference between
the maximum and minimum values is 23%. In practical reeling operation,
the winding length of the pipeline can be used for controlling the maxi-
mum value of SM3. Moreover, the reeling length has no influence on the
maximum bending moment SM2.
20100
19800
19500
19200
SF1 (N)
18900
Case1
18600 Case2
Case3
18300 Case4
18000
0 5 10 15 20 25 30 35 40 45 50 55
Figure 7.35 SF1 along the path in different reeling length.
20
0
SM3 (N·m)
-20
Case 1
Case 2
Case 3
-40 Case 4
0 5 10 15 20 25 30 35 40 45 50 55
Figure 7.36 SM3 along the path in different reeling length.
7.4.4 The Location of the Bearing Plate

In reeling operation, the mechanical behavior of the pipeline can eas-
ily be changed by just moving the bearing plate. In this analysis, the dis-
tances between the bearing plate and the coiling drum (Node 1 and Node
2 shown in Figure 7.38) are changed as 2.5 m (Case 1), 3 m (Case 2), 3.5 m
(Case 3), and 4 m (Case 4). The variation of SF1, SM3 and SM2 along the
path of each case is shown in Figures 7.39, 7.40, and 7.41.
600
400
SM2 (N·m)
Case1
Case2
200 Case3
Case4
0 5 10 15 20 25 30 35 40 45 50 55
Figure 7.37 SM2 along the path in different reeling length.

in different reeling length.
SM3 (N·m)
Case 1 37.46
Case 2 44.84
Case 3 45.28
Case 4 48.58
Node 1
Node 2
Y
Z X
Figure 7.38 The defined distance.

20100
19800
19500
19200
SF1 (N)
18900
Case1
18600 Case2
Case3
18300 Case4
18000
0 5 10 15 20 25 30 35 40 45 50 55
Figure 7.39 SF1 along the path in different location of the bearing plate.
15
0
SM3 (N·m)
-15
Case 1
Case 2
Case 3
-30 Case 4
-45
0 5 10 15 20 25 30 35 40 45 50 55
Figure 7.40 SM3 along the path in different location of the bearing plate.
Axial forces SF1 reach the maximum value 20,000 N after a certain fluc-
tuation in all cases. It can be found that the value of bending moment SM3
is smaller as the distance between the bearing plate and the drum becomes
larger. The value of maximum SM3 in each case is shown in Table 7.4. The
difference between the maximum and minimum value is 8%. In addition,
there is also little difference between SM2 in each case.
600
400 Case1
SM2 (N·m)
Case2
Case3
Case4
200
0 5 10 15 20 25 30 35 40 45 50 55
Figure 7.41 SM2 along the path in different location of the bearing plate.

of each case.
SM3 (N·m)
Case 1 46.45
Case 2 45.28
Case 3 44.20
Case 4 42.93
7.5 Conclusions
In this chapter, the mechanical behavior of MSFP during reeling operation
is investigated through local and global analysis. Tensile and bending tests
are conducted to get the tension-extension and moment-curvature rela-
tion. The mechanical properties of MSFP are then invoked into the global
analysis model in which deformation and forces are predicted and pre-
sented. After that, an extensive parametric analysis using FEM is carried
out to study the influencing mechanisms on MSFP. Beneficial conclusions
can be drawn as follows:
• The tensile force in the pipe distributes quite uniformly

without too much variation, but it has a negative influence
on the moment capacity of the pipeline.
• The bending moment in the reeling plane and vertical reel-

ing plane should be paid more attention in reeling opera-
tion, especially the section between the plate bearing and
the coiling drum because the combined bending moment
is the largest.
• The spatial spiral shape and deadweight of the pipeline
coiled on the drum has significant influence on the bending
moment, which can be altered by the diameter of the coiling
drum, the sinking distance, the reeling length, and the dis-
tance between bearing plate and the coiling drum.
References
1. Szczotka, M. (2011). Dynamic analysis of an offshore pipe laying opera-
tion using the reel method. Acta Mechanica Sinica, 27(1), 44–55.
2. Ruan, W., Bai, Y., Zhang, T., Cao, Y., & Liu, D. (2018). Safety assessment
study of a planned offshore floating platform pipelaying test. Ships
and Offshore Structures, 13(sup1), 202–213.
3. Longva, V., & Sævik, S. (2015). A Lagrangian–Eulerian formulation for
reeling analysis of history-dependent multilayered beams. Computers
& structures, 146, 44–58.
4. Longva, V., & Sævik, S. (2016). On prediction of torque in flexible
pipe reeling operations using a Lagrangian–Eulerian FE framework.
Marine Structures, 46, 229–254.
5. Maincon, P. E. (2017). Torsion in Flexible Pipes, Umbilicals and Cables
Under Loadout to Installation Vessels. In ASME 2017 36th International
Conference on Ocean, Offshore and Arctic Engineering-Volume 3B:
Structures, Safety and Reliability.
6. Xu, Y., Bai, Y., Fang, P., Yuan, S., & Liu, C. (2019). Structural analysis of fibre-
glass reinforced bonded flexible pipe subjected to tension. Ships and
Offshore Structures, 14(7), 777–787
7. Bahtui, A., Bahai, H., & Alfano, G. (2009). Numerical and analytical model-
ing of unbonded flexible risers. Journal of offshore mechanics and
Arctic engineering, 131(2)
8. Bai, Y., Liu, T., Cheng, P., Yuan, S., Yao, D., & Tang, G. (2016). Buckling
stability of steel strip reinforced thermoplastic pipe subjected to external
pressure. Composite Structures, 152, 528–537.
9. Fang, P., Yuan, S., Cheng, P., Bai, Y., & Xu, Y. (2019). Mechanical responses of
metallic strip flexible pipes subjected to pure torsion. Applied Ocean
Research, 86, 13–27.
10. Bai, Y., Han, P., Liu, T., Yuan, S., & Tang, G. (2018). Mechanical responses
of metallic strip flexible pipe subjected to combined bending and external
pressure. Ships and Offshore Structures, 13(3), 320–329.
11. ASTM D2015-01. (2014). Standard test method for longitudinal tensile
properties of “fiberglass” pipe and tube.
12. ISO 527-2012. Plastics-determination of Tensile Properties, 1–11.12.
13. Fergestad, D., & Løtveit, S. A. (2014). Handbook on design and operation of
flexible pipes. NTNU, 4Subsea and MARINTEK, 1.
Part 2
RISER ENGINEERING
8
Flexible Risers and Flowlines
8.1 Introduction
Flexible risers and flowlines (referred to as flexible pipes in the remainder
of this chapter) trace their origins to pioneering work carried out in the
late 1970s. Initially, flexible pipes were used in relatively benign weather
environments such as offshore Brazil, the Far East, and the Mediterranean.
However, since then, flexible pipe technology has advanced rapidly, and
today, flexible pipes are used in various fields in the North Sea and are
also gaining popularity in the Gulf of Mexico. Flexible pipe applications
include water depths own to 8,000 ft, high pressure up to 10,000 psi, and
high temperatures above 150°F, as well as the ability to withstand large
vessel motions in adverse weather conditions.
The main characteristic of a flexible pipe is its low relative bending to
axial stiffness. This characteristic is achieved through the use of a number
of layers of different material in the pipe wall fabrication. These layers are
able to slip past each other when under the influence of external and inter-
nal loads, and hence, this characteristic gives a flexible pipe its property of a
low bending stiffness. The flexible pipe composite structure combines steel
armor layers with high stiffness to provide strength, and polymer sealing
layers with low stiffness to provide fluid integrity. This construction gives
flexible pipes a number of advantages over other types of pipelines and ris-
ers such as steel catenary risers. These advantages include prefabrication,
storage in long lengths on reels, reduced transport and installation costs,
and suitability for use with compliant structures.
8.2 Flexible Pipe Cross-Section

There are two types of flexible pipes in use: bonded and unbonded flexible
pipes. In bonded pipes, different layers of fabric, elastomer, and steel are
bonded together through a vulcanization process. Bonded pipes are only
147
used in short sections such as jumpers. Figure 8.1 shows a bonded flexible
pipe cross-section [1, 3, 4].
Figure 8.2 shows a picture from the Heidrun TLP where 5½ “produc-
tion and 2” gas lift hoses are connected to production tree.
On the other hand, unbonded flexible pipes can be manufactured
for dynamic applications in lengths of several hundred meters. Unless
otherwise stated, the rest of this Chapter shall deal with unbonded flexible
pipes. Figure 8.3 shows a typical cross-section of an unbonded flexible
pipe. This figure clearly identifies the five main components of the flexi-
ble pipe cross-section. The space between the internal polymer sheath and
Figure 8.1 Bonded flexible pipe (Antal et al., 2003) [1].
Figure 8.2 Production and gas lift hoses on the Heidrun TLP (Antal et al., 2003) [1].
Flexible Risers and Flowlines 149
8
7
6
5
4
3
2
1
1. Carcass
2. Internal Fluid Barrier
3. Pressure Armor
4. Anti-Wear Layer
5. Tensile Armor
6. Anti-Wear Layer
7. Armor Wire
8. Anti-Birdcaging Layer
9. External Fluid Barrier
Figure 8.3 Typical cross-section of an unbonded flexible pipe (Zhang et al., 2003) [8].
the external polymer sheath is known as the pipe annulus. The five main
components of the flexible pipe wall shall be dealt with in the following
sections [2, 4−6].
8.2.1 Carcass
The carcass forms the innermost layer of the flexible pipe cross-section. It
is commonly made of a stainless steel flat strip that is formed into an inter-
locking profile as seen in Figure 8.2. Different steel grades can be used to
form the carcass, and the choice of material usually depends on the inter-
nal fluid characteristics. The most common grades used to manufacture
the carcass are AISI grades 304 and 316, and Duplex. The inner bore fluid
is free to flow through the carcass profile and therefore the carcass material
needs to be corrosion-resistant to the bore fluid.
The main function of the carcass is to prevent pipe collapse due to
hydrostatic pressure or build-up of gases in the annulus. The build-up of
gases in the annulus could be a potential failure mode for the pipe and
occurs in hydrocarbon-carrying pipes when gases from the inner pipe bore
diffuse through the internal polymer sheath into the annulus. In the case of
a well shut-down and subsequent depressurization and evacuation of the

inner bore, the annulus gas pressure could cause pipe collapse. Thus, the
steel carcass is designed to withstand this collapse pressure. Pipes that do
not carry any hydrocarbon fluid (e.g., water injection pipes) can be con-
structed without a carcass if there is no potential for gas build-up in the
annulus to cause pipe collapse.
8.2.2 Internal Polymer Sheath

The internal polymer sheath provides a barrier to maintain the bore fluid
integrity. Exposure concentrations and fluid temperature are key design
drivers for the internal sheath. Common materials used for the internal
sheath include Polyamide-11 (commercially known as Rilsan®), high den-
sity polyethylene (HDPE), cross-linked polyethylene (XLPE), and PVDF.
Polyamide-11 and HDPE are the two materials most commonly used. They
both can withstand temperatures up to about 149°F (65°C), and have an
allowable strain of 7%. PVDF can be used for applications that require a
higher temperature tolerance. This material can withstand a temperature
of 266°F (130°C), but its allowable strain is only 3.5%. The polymer sheath
layer thickness is a function of various parameters such as inner bore fluid
temperature, composition, and inner bore pressure. The average size of this
sheath is about 5–8 mm, but pipes with up to 13 mm of internal polymer
sheath have also been manufactured.
8.2.3 Pressure Armor

The role of the pressure armor is to withstand the hoop stress in the pipe
wall that is caused by the inner bore fluid pressure. The pressure armor
is wound round the internal polymer sheath and is made of interlocking
wires. These wire profiles (as seen in Figure 8.4) allow bending flexibility
and control the gap between the armor wires to prevent internal sheath
extrusion through the armor layer. In order to best resist the hoop stress in
the pipe wall, the pressure armor is wound at an angle of about 89° to the
pipe longitudinal axis.
The material used for the pressure armor wire is typically high strength
carbon steel. The choice of wire typically depends on whether the pipe is
qualified for “sweet” or “sour” service. (“Sour” service is defined by NACE
MR 01-75). The highest strength steel wires used in flexible pipe applications
have an ultimate tensile strength (UTS) of 1,400 MPa (200 ksi). However,
these high-strength wires are prone to hydrogen induced cracking (HIC)
and sulfide stress cracking (SSC). Hence, it might not be possible to use such
stiffener
bell mouth
θ
Τ
θ Τ
Figure 8.4 General arrangement for local curvature analysis at the bellmouth or stiffener
(Zhang et al., 2003).
high strength steel wires for “sour” pipe applications. The alternative would
be to use additional steel layers with a UTS as low as 750 MPa (105 ksi).
8.2.4 Tensile Armor

The tensile armor layers are always cross-wound in pairs. As their name
implies, these armor layers are used to resist the tensile load on the flexible
pipe. The tensile armor layers are typically made of flat rectangular wires
and laid at about 30°–55° to the longitudinal axis. A lay angle of 55° results
in a torsionally balanced pipe, and this angle is used in pipe designs where
the hoop stress is also resisted by the tensile armor layers, and no pressure
armor layer exists.
The tensile armor layers are used to support the weight of all the pipe
layers, and to transfer the load through the end fitting to the vessel struc-
ture. High tension in a deepwater riser may require the use of four tensile
armor layers, rather than just two. The tensile armor wires are made of
high strength carbon steel like the pressure armor wires.“Sweet” or “sour”
service conditions are a determining factor on the wire strength that can be
used, because high strength wire is more prone to HIC and SSC.
8.2.5 External Polymer Sheath

The external polymer sheath can be made of the same materials as the
internal polymer sheath. The main function of the external sheath is as a
barrier against seawater. It also provides a level of protection for the armor
wires against clashing with other objects and during installation.
8.2.6 Other Layers and Configurations

Besides these five main layers of the flexible pipe, there are other minor
layers which make up the pipe cross-section. These layers include anti-
friction tapes wound round the armor layers and whose purpose it is to
reduce friction and hence wear of the wire layers when they rub past each
other as the pipe flexes due to external loads. Anti-wear tapes can also be
used to make sure that the armor layers maintain their wound shape. These
tapes ensure that the wires do not twist out of their pre-set configuration, a
phenomenon called “birdcaging” which results from hydrostatic pressure
causing axial compression in the pipe.
In some flexible pipe applications, because of high tensile loads, it is
required to use high tensile wires for the tensile armor layers, and yet the
presence of a “sour” environment means that these wires would suffer
an unacceptable rate of HIC/SSC. A solution to this situation is to fab-
ricate a pipe cross-section with two distinct annuli rather than one. The
inner annulus could contain the pressure armor layer which need not
be made of very high tensile steel, and which would therefore not suffer
serious corrosion problems due to a high concentration of H2S. An extra
polymer sheath could then be laid between the pressure armor layer and
the tensile armor layers. This polymer sheath would prevent a high H2S
concentration in the outer annulus. A certain amount of H2S would still
be able to diffuse through this polymer sheath from the inner to the outer
annulus. However, the concentration of H2S in the outer annulus could
now be low enough to permit the use of high tensile wires for the tensile
armor layers.
8.3 End Fitting and Annulus Venting Design

8.3.1 End Fitting Design and Top Stiffener (or Bellmouth)
The end fitting design is a critical component of the global flexible pipe
design process. The main functions of the end fitting are to transfer the
load sustained by the flexible pipe armor layers onto the vessel structure
and to complement the sealing of the polymer fluid barrier layers.
A number of critical issues need to be considered during the design and
manufacture of end fitting arrangements. Tight manufacturing tolerances
are essential for the pressure sheath and sealing ring dimensions, pressure
armor termination, and bolt torquing to ensure the adequate transfer of
load from the steel layers of the pipe onto the vessel structure. Epoxy filling
should be carried out using the appropriate techniques to ensure no air
gaps are produced. The correct positioning and functioning of the annulus
vent ports are also important to ensure no build-up of gases in the annulus
(discussed further in Section 8.3.2).
The most severe location for fatigue damage in the risers is usually in
the top hang off region. The riser is protected from over bending in this
area by either a bend stiffener or a bellmouth. The detailed local anal-
ysis for the curvature or bellmouth is carried out using 2D finite ele-
ment model. Figure 8.4 shows the basic arrangement for the top stiffener
and bellmouth.
The bending stiffener is modeled by a 2D tapered unsymmetric beam
element and the pipe is simulated by a 2D beam element. The interface
between the pipe and the bend stiffener is represented by a 2D general
contact element. Both to non-linear stress strain curve of the bend stiffener
and the non-linear bending curvature hysteresis loop are considered in the
analysis. If a bellmouth is used, a steel 2D solid element is adopted.
Knowing the top tensions and angles for each load case from the global
dynamic analysis, the curvature distribution along the flexible riser can be
found by applying these tensions and angles at the bottom of the models as
8.3.2 Annulus Venting System

Over time the fluids that are transported in the pipe bore will diffuse
through the internal polymer sheath into the pipe annulus. These dif-
fused gases include water, CO2 and H2S, and their presence in the annulus
could have a deleterious effect on the steel layers. Water and CO2 have the
tendency to cause general or pitting corrosion in the pressure and ten-
sile armor layers. The presence of water could also have a negative effect
on the fatigue life of the steel layers. The presence of H2S causes HIC and
SSC in steel and its concentration is carefully assessed during the design
stage because a pipe qualified for “sweet” service could make use of high
strength tensile steel (that would otherwise suffer unacceptable corrosion
in a “sour” environment).
Besides the negative corrosion and fatigue effects that the presence of
these diffused gases could have on the steel layers of the pipe, the build-up
of pressure in the annulus due to the presence of the gases could also cause
collapse of the internal polymer sheath of the pipe. In case of a sudden
pressure loss in the inner bore of the pipe (say through an emergency
shutdown of the system), the pressure in the annulus due to these diffused
gases might be greater than the pressure in the inner bore. This could lead
to collapse of the internal polymer sheath, loss of fluid integrity, and failure
of the pipe. The steel carcass is designed to withstand this collapse pressure
due to gas build-up in the annulus. However, pipe designs without a steel
carcass also exist, especially for non-hydrocarbon carrying pipes, and this
failure mode has been documented to occur in flexible pipes.
In order to prevent the build-up of gases in the annulus due to diffu-
sion, a venting system is incorporated into the pipe structure to enable
the annulus gases to be vented out to atmosphere. Three vent valves are
incorporated into both end fitting arrangements of a pipe. The vent valves
are directly connected to the annulus and are designed to operate at a pre-
set pressure of about 30–45 psi. The vent valves in the end fitting arrange-
ment located subsea are sealed to prevent any ingress of seawater into
the annulus.
8.4 Flexible Riser Design

8.4.1 Design Analysis
The essential tasks for design and analysis of flexible risers are similar to
those described for other types of risers, see below.
1. Design Basis Document: The document should as mini-

mum include the following:
• host layout and subsea layout;

• wind, wave, and current data and vessel motion that are
applicable for riser analysis;
• applicable design codes and company specifications;
• applicable design criteria;
• porch and I-tube design data;
• load case matrices for static strength, fatigue, and inter-
ference analysis;
• applicable analysis methodology.
2. FE Modeling and Static Analysis: A finite element model

is built and a nonlinear static analysis is carried out assum-
ing the vessel is in NEAR, FAR, and CROSS positions.
3. Global Dynamic Analysis: A global regular wave dynamic

analysis is carried out assuming the vessel is in NEAR, FAR,
and CROSS positions. A sensitivity study is performed on
critical parameters such as wave periods, effect of marine
growth, and hydrodynamic coefficients.
4. Interference Analysis: A dynamic regular wave analysis is
carried out to check the minimum clearance between the
risers and with the vessel system along the water column, for
various predefined load cases. The interference analysis shall
confirm that selection of riser hang-off angles and departure
angles, etc.
5. Cross-sectional Model: A detailed cross-section model is
built to calculate key cross-sectional properties such as
bending stiffness, axial stiffness, etc., FAT pressure, etc.
6. Extreme and Fatigue Analysis: The wire and tube stresses
are calculated at design pressure. An extreme response
analysis is carried out using regular wave theory to estimate
tensions and cyclic angles, etc. The cross-sectional model is
then used to perform to fatigue analysis.
7. Design Review: This includes check of global configuration,
bell mouth design, interference and fatigue design etc. In
some special situations, upheaval buckling and on-bottom
stability of flexible flowlines are also checked, following
pipeline design practice.
Typically, a detailed design of flexible pipe is carried out by the sup-

plier for the flexible pipe materials. A 3rd party, normally a riser engi-
neering company, is engaged to carried out a verification of the design, as
aforementioned.
8.4.2 Riser System Interface Design

Figure 8.5 shows a flowchart for integrated mooring and riser system
design. Due to the complexity of a field development like Barracuda/
Caratinga project, the design of most topside and subsea equipment
directly affected and was affected by the riser mooring system design.
The interfaces between the mooring and risers and other components of
the field development facilities are an integral part of the overall
mooring and riser system design. For example, estimates of maxi-
mum expected riser loads are required for the riser supporting structure
Preliminary Mooring System Design

(Configuration & Chain Design)
Perform Vessel Motion

Model Test Analysis
Mooring Design
Mooring Line
RAO Offset
Loads Optimization
Process
Preliminary Riser System Design Design Criteria

(Configuration & Pipe Design) Met?
Perform Dynamic Mooring System

Riser Analysis Design Complete
Riser Config. &

Dynamic Riser Reduce Vessel
Pipe Design
Loads Offset
Optimization
Design Criteria
Met?
Riser System
Design Complete
Figure 8.5 Mooring and riser system design (Seymour et al., 2003) [7].
design. This impacts the modifications required to the hull structure to

support the external riser porches.
Figure 8.6 shows riser system interface design for flexible risers. The
overall riser configuration impacts the design of the bell-mouths at the base
of the individual riser I-tubes, and also the design of the bend-stiffeners
required to provide bending resistance at the vessel interface. Additional
riser global configuration considerations include the on-bottom stability
of the risers and flowlines in order to mitigate against possible interference
between the large number of individual lines during both installation and
in-service conditions.
8.4.3 Current Design Limitations

Figure 8.7 gives an indication of the current design limitations for flexible
pipes. The existing water injection lines and the newly qualified pipe are
collected by Remery et al. (2004) [6].
Riser Mooring System Solution
Dynamic
Riser Loads
Riser Pipe Design Interface

Confirmed Design
Vessel Topside:
Manufacturing Qualification Bellmouth, I-tube,
Testing Porch, etc.
Riser Ancillaries:
Delivery and Certification Bend Stiffener,
Installation Abrasion Protection
Subsea Structure:
Seabed Anchor
Figure 8.6 Overview of riser system interface design (Seymour et al., 2003) [7].
Water Injection Risers Delivered

80000
70000 New Qualified

Pipe
60000
PxID (psi x inch)
50000
40000
Existing WI lines
30000
20000
10000
0
0 1000 2000 3000 4000 5000 6000 7000
WD (feet)
Figure 8.7 Water injection flexible pipe technology limits (Remery et al., 2004).
References
1. Antal, S., Nagy, T., Boros, A. (2003), “Improvement of Bonded Flexible Pipe
according to New API Stnadrad 17K”, OTC 15167.
2. API Specification 17J(1999), “Specification for Unbonded Flexible Pipe”.
3. API Specification 17K(2001), “Specification for Bonded Flexible Pipe”.
4. API Recommended Practice 17B(2002), “Recommended Practice for
Flexible Pipe”.
5. NACE MR 01-75, “Sulfide Stress Cracking Resistant Metallic Materials for
Oilfield Equipment”.
6. Remery, J., Gallard R. and Balague, B. (2004), “Design and Qualification
Testing of a Flexible Riser for 10, 000 psi and 6300 ft WD for the Gulf of
Mexico”, Deep Oil Technology Conference, New Orleans.
7. Seymour, B., Zhang, H. and Wibner, C. “Integrated Riser and Morring
Design for the P-43and P-48 FPSOs”, OTC 15140.
8. Zhang, Y., Chen, B., Qiu, L., Hill, T. and Case, M. “State of the Art Analytical
Tools Improve Optimization of Unbonded Flexible Pipes for Deepwater
Environments”, OTC 15169.
9
Lazy-Wave Static Analysis
9.1 Introduction
Industry practice requires several types of flexible riser configurations
typically used in conjunction with floating production/loading systems.
Figure 9.1 illustrates these six main types of flexible riser configura-
tions. The feasible configurations differ in the use of buoyancy modules
and the methods of anchoring to the seafloor. Configuration design
considerations include several factors such as water depth, host vessel
access/hang-off location, field layout such as the quantity and types of
risers and mooring layout, environmental data, and host vessel motion
characteristics.
• Free-Hanging Catenary
This is the simplest configuration for a flexible riser. It is also the cheap-
est to install because of the minimal subsea infrastructure and easy
installation. However, a free-hanging catenary is exposed to severe
loading due to high vessel motions. The riser simply lifted off or low-
ered down on the seabed. Thus, the free-hanging catenary is likely to
suffer from compression buckling at the riser touchdown point (TDP)
and tensile armor wire “birdcaging”. The riser is appropriate for water
depth from medium to deep water in medium environmental condi-
tions. However, in deeper water, the top tension is large due to the long-
suspended riser length.
• Lazy Wave and Steep Wave

In the wave type, buoyancy and weight are added along a longer length
of the riser to decouple the vessel motions from the TDP of the riser. Lazy
waves are preferred to steep waves because the former require minimal
subsea infrastructure. However, lazy waves are prone to configuration
alterations if the internal pipe fluid density changes during the riser life-
time. On the other hand, steep wave risers require a subsea base and subsea
159
Free Hanging Lazy Wave Steep Wave

Catenary
Lazy-S Steep-S Pliant Wave®
Figure 9.1 Flexible riser configurations.[1]
bend stiffener and can maintain their configuration even if the riser fluid
density changes.
The wave type risers are appropriate for water depth from shallow to
deep water. The steep wave risers are suitable for congested seabed devel-
opments and have a good dynamic response.
• Lazy-S and Steep-S

In the lazy-S and steep-S riser configurations, there is a subsea buoy, either
a fixed buoy which is fixed to a structure at the seabed or a buoyant buoy
which is positioned by chains. The addition of the buoy removes the prob-
lem with the TDP, as described above. The subsea buoy absorbs the tension
variation induced by the floater and the TDP has only small variation in
tension if any.
Lazy-S configurations are considered only if catenary and wave config-
urations are not suitable for a field. This is primarily due to the complex
installation required. A lazy-S configuration requires a mid-water arch,
tether, and tether base, while a steep-S requires a buoy and subsea bend
stiffener. The riser response is driven by the buoy hydrodynamics and
complex modeling is required due to the large inertial forces in action.
In case of large vessel motions, a lazy-S might still result in compression
problems at the riser touchdown, leaving a steep-S as a possible alternative.
This configuration is good at shallow water and for satellite tie-backs with
several risers with a good dynamic response.
Lazy-Wave Static Analysis 161
• Pliant Wave
The pliant wave configuration is similar to the steep wave configuration
except a subsea anchor controls the TDP, i.e., the tension in the riser is
transferred to the anchor and not to the TDP. The pliant wave has the
additional benefit that it is tied back to the well located beneath the floater.
This makes well intervention possible without an additional vessel.
This configuration can accommodate a wide range of bore fluid densi-
ties and vessel motions without causing any significant change in configu-
ration and inducing high stress in the pipe structure. Due to the complex
subsea installation that is required, it would be suggested only if simple
catenary, lazy wave, or steep wave configurations are not viable. Moreover,
this configuration is appropriate for a wide range of water depths and
retains the advantages of both lazy wave and steep wave.
In this chapter, a reasonable analytical model is presented to study the
static behavior of deep water lazy-wave risers by taking the effect of envi-
ronmental loads and elastic seabed into account. Furthermore, a series of
sensitivity analyses are also presented to highlight the influencing param-
eters in lazy wave configuration, like seabed stiffness, hang-off inclination
angle, etc. The works of the theoretical analysis and FEA in the chapter are
quoted from the paper of “Static analysis of deep water lazy-wave umbilical
on elastic seabed”.[2] Lazy wave configuration is illustrated in Figure 9.2.
Buoyancy Module Section Typical Strake Section

(Not to Scale) (Not to Scale)
Riser Base
GC726
Well Approach GC683
Well Approach
Figure 9.2 Configuration of lazy-wave riser.

9.2 Fundamental Assumptions

Due to the elastic seabed, ocean current, buoyancy modules, etc., the static
nonlinear analysis of deep water lazy-wave configuration can be very
complex. To facilitate this analysis, the following assumptions are made
from the engineering application point of view:
1. A static two-dimensional problem is considered under the

action of a steady planar ocean current, which means the
pipeline is restricted to move in the vertical plane.
2. Both axial deformation and twisting behavior of the riser is
neglected here, because they have scarce effect in practical
applications.
3. The seabed is simplified as a horizontal and flat elastic
foundation.
4. The boundary-layer phenomena at the hang-off location
and two ends of the buoyancy section are neglected.
9.3 Configuration Calculation

The behavior of lazy-wave riser is described in terms of the position of
the centerline of the riser according to the classical theory of rods.[3]
Typical configuration of lazy-wave riser is shown in Figure 9.3. As illus-
trated in this figure, the analytical model of lazy-wave riser consists of
three segments:
Sea level x HOP
y
Current
Buoyancy section
Water depth WD
LP
Decline section
DP Hang-off section
Touchdown segment BLP
TDP
Seabed x1/x2 Boundary-layer segment
y1/y2
Figure 9.3 Lazy-wave riser configuration.

1. Cable segment: The segment suspended in water from the

hang-off point (HOP) to the boundary-layer point (BLP).
The suspended riser is very long in the deep water and can
be treated as a cable. Due to the addition of the buoy-
ancy modules along some length of riser section, the cable
segment can be divided into three parts: hang-off section,
buoyancy section, and decline section.
2. Boundary-layer segment: The segment from the BLP to
the TDP. Boundary-layer segment is considered as an indi-
vidual segment because it stays very close to the seabed and
exhibits boundary-layer phenomenon.[4] It deviates from the
property of the cable and behaves like a beam. The potential
failures of local bending and high compression may occur in
the zone of boundary-layer segment.
3. Touchdown segment: The segment laid on the seabed from
the TDP to the seabed infinity, which behaves like a beam.
The seabed is simplified as an elastic Winkler-like founda-
tion, which is a little different from the actual seabed con-
dition. That is to say, the effect of the seabed can be treated
as a series of elastic springs and the seabed supporting
the touchdown riser suffers from only elastic compression
deformation.
According to the configuration partition, three coordinate systems are

established in the lazy-wave riser model: one global coordinate system
(x, y) is set up by locating the origin at the HOP, and two local coordi-
nate systems (x1, y1) and (x2, y2) with the same origin at the TDP. The global
coordinate system (x, y) here is to describe the behavior of the whole lazy-
wave configuration. The two local coordinate systems (x1, y1) and (x2, y2)
are used to describe the behavior of the boundary-layer segment and
touchdown segment, respectively.
9.3.1 Cable Segment

9.3.1.1 Hang-Off Section
The hang-off section is the riser from the HOP to the lift point (LP). The
arc length of the hang-off section is L1. The riser is hung off by a ball
joint to prevent high bending moment induced by waves at the HOP. Due
to the lifting effect of buoyancy section and displacement constraint of
the vessel, the configuration looks like a “U” with a sag bend. The vertical
distance of the lowest point in the sag bend from the sea level is herein
set as ysag. For a differential element of the riser illustrated in Figure 9.4,
the equilibrium of force and bending moment can be derived as
dV = Fτds sin θ – Fnds cos θ – wds (9.1)
dH = Fnds sin θ + Fτds cos θ (9.2)
dM = Vds cos θ – Hds sin θ – w(ds)2 cos θ/2 – Fn(ds)2/2 (9.3)
Fn = 0.5ρwCdD ds|vc sin θ|vc sin θ (9.4)
Fτ = 0.5ρwCτπD ds|vc cos θ|vc cos θ (9.5)
T = V2 + H2 (9.6)
where T, V, and H represent axial tension, vertical force, and horizontal

force, respectively. dV and dH represent the increment of the vertical and
horizontal force; dM is the increment of bending moment; Fn and Fτ repre-
sent the drag force in the normal and tangential direction per unit length,
Sea level x
V
y
M H
Current F nds
M+dM
WD
H+dH
y arch
wds
Water depth
V+dV
y sag
Arch bend
Sag bend
Seabed
Figure 9.4 Force sketch of a differential element of suspended riser.

respectively. Cd and Cτ are the drag coefficient in the normal and tangen-
tial direction, respectively. ρw is sea density; vc is current velocity; D is the
outer diameter of the riser; w is the submerged weight per unit length of
the riser; ds is the length of each differential element; θ is the inclination
angle of the riser from horizontal direction to axial direction.
The effect of riser’s flexural stiffness leads to Eq. (9.3), which is too
complicated to solve directly. For simplification, the hang-off section is
considered as a cable without any flexural stiffness due to its property
of long length in deep water. Catenary theory is thus adopted to ana-
lyze the behavior of the hang-off section. The inclination angle, θ, can
be derived from Eq. (9.3) by ignoring the higher order terms and flexural
stiffness:
V
θ = arctan   (9.7)
 H
The curvature, κ, along the riser can be obtained based on the catenary
theory:
κ = dθ/ds (9.8)
As both axial deformation and shear deformation are not considered in

the analysis, the horizontal and vertical displacements in the global coor-
dinate (x, y) can be expressed as follows:
dx = ds cos θ, dy = ds sin θ (9.9)
Once the displacement, inclination angle, and curvature along the riser
are obtained, the approximation of the real bending moment, M, and shear
force, S, can be derived through the following:
dθ
M = EI κ = − EI (9.10)
ds
dM d 2θ
S=− = EI 2 (9.11)
ds ds
where EI is the flexural stiffness of the riser. Note: shear force is set to be
positive when in clockwise direction.
9.3.1.2 Buoyancy Section

The buoyancy section is the section of riser from the LP to the decline
point (DP). The arc length of the buoyancy section is L2. The buoy-
ancy modules are uniformly attached along the long length of the buoy-
ancy section, to decouple the vessel motions from the TDP of the riser.
Buoyancy modules are made of syntactic foam which has the desirable
property of low water absorption. The buoyancy modules need to be
clamped tightly to the riser to avoid any slippage which could alter the
riser configuration and induce high stress.[5] The mechanical behavior
of the buoyancy section is very different from that of the hang-off sec-
tion. To conveniently perform the analytical model, a uniformly dis-
tributed lifting force is acted on the buoyancy section, representing the
loading combination of the buoyancy force and weight. Furthermore,
the equivalent outer diameter De, equivalent submerged weight unit
length, we, and equivalent flexural stiffness, EIe, are considered in the
buoyancy section. Generally, the equivalent flexural stiffness of the
buoyancy section is very close to the flexural stiffness of the riser, due
to the small length of buoyancy modules attached, which means buoy-
ancy modules have little influence on the flexural stiffness of the buoy-
ancy section. Because of the lifting effect of the buoyancy force, the
configuration behaves like an arch bend. The vertical distance of the
highest point of the arch bend from the sea level is herein set as yarch.
The governing equations of the buoyancy section are similar with those
of the hang-off section. The main difference is that the flexural stiff-
ness, EI, submerged weight per unit length, w, and outer diameter, D,
in Eqs. (9.1) to (9.11) should be replaced by the equivalent values EIe,
we, and De, respectively.
9.3.1.3 Decline Section

The decline section is the section of riser from the DP to the BLP. The
arc length of the decline section is L3, which is unknown in the analysis.
Its configuration behaves like a free-hanging catenary. As the mechani-
cal behavior of the decline section is the same with that of the hang-off
section, the governing equations are also the same with those of the hang-
off section.
9.3.2 Boundary-Layer Segment

Many researchers have investigated the mechanical behavior of the
boundary-layer segment by simply neglecting its flexural stiffness. As a
result, bending moment near the TDP is distorted and the continuity of
shear at the TDP can’t be guaranteed. This is because the boundary-layer
segment resides very close to the seabed and behaves like a beam. This
problem has been well solved by a boundary layer technique based
on singular perturbation theory, which consists in superposing the “outer”
catenary solution to an “inner” boundary layer solution valid only in a
neighborhood of the boundary.[4]
In this section, boundary-layer segment is individually separated from
the decline section due to its boundary-layer phenomenon. Figure 9.5
shows the force sketch of a differential element of boundary-layer seg-
ment. From the figure, it can be found that the effect of ocean current is
ignored in the boundary-layer segment. As the boundary-layer segment
stands very close to the seabed, just ahead the TDP, its inclination angle
is very small. Also, the change of axial tension along boundary-layer seg-
ment can be disregarded.[4] By taking into account the flexural stiffness,
the boundary segment is therefore considered in the framework of linear
deformation theory of beam:
d 4 y1 d 2 y1
EI − T = w ( L ≤ x1 ≤ 0) (9.12)
dx14 dx12
S+dS M
T
θ+dθ
θ
T+dT M + dM S
wds
BLP
Seabed x1 TDP
y1
Figure 9.5 Force sketch of a differential element of boundary-layer segment.

where T denotes the unknown constant axial tension at the TDP; L is the
x1-coordinate at BLP in the local coordinate system (x1, y1).
The general solution of Eqs. (9.4)–(9.12) can be obtained as following:
w 2
y1 = − x1 + c1 + c2 x1 + c3 sinh(γ x1 ) + c4 cosh(γ x1 ) (9.13)
2T
where c1, c2, c3, and c4 are the unknown coefficients and
T
γ= (9.14)
EI
The associated bending moment and shear force can be obtained on

the basis of linear deformation beam theory:
d 2 y1
M1 ( x ) = − EI (9.15)
dx12
d 3 y1
S1 ( x ) = EI 3 (9.16)
dx1
9.3.3 Touchdown Segment

The touchdown segment is mainly subjected to the submerged weight,
seabed vertical resistance, and the traction at the TDP, ignoring the effect
of seabed friction (seeing Figure 9.6). The touchdown segment will embed
itself into the seabed to make the equalize the riser’s submerged weight and
the seabed resistance in the vertical direction. Embedment into the seabed
will contribute to the pipeline’s lateral resistance to being moved sideways
and walking, which is significant in the analysis of lateral buckling and
stability against hydrodynamic forces.[6] Thus, it is necessary to quantify
the embedment of the touchdown segment. For better investigating the
embedment into the seabed, the seabed is simplified as an elastic Winkler-
like foundation. It can be seen clearly that the deformation of the touch-
down segment is very small, tending to be horizontal. Based on linear
ky2ds
S+dS M T
θ+dθ
θ
T +dT M+dM S
wds
Seabed x2 TDP
y2
Figure 9.6 Force sketch of a differential element of riser laid on seabed.
deformation theory of beams, the governing equation can be derived by

taking the force equilibrium in the vertical direction into consideration:
d 4 y2 d 2 y2
EI −T + ky 2 = w (0 ≤ x 2 ≤ +∞) (9.17)
dx 24 dx 22
where T is the axial tension at the TDP, and k is the elastic stiffness of the
seabed. As the effect of seabed friction is neglected, axial tension along
the length of the touchdown segment keeps nearly constant. It can be
found that Eq. (9.17) is a four-order ordinary differential equation. For
T > 2 EIk , no real solution can be obtained. Whereas, for T ≤ 2 EIk , the
general solution can be obtained:
w
y2 = + c5 exp(−α x 2 )cos(β x 2 ) + c6 exp(−α x 2 )sin(β x 2 )
k
+ c7 exp(α x 2 )cos(β x 2 ) + c8 exp(α x 2 )sin(β x 2 )
(9.18)
where c5, c6, c7, and c8 are the unknown coefficients and
1 k T 1 k T
α= 2 + , β= 2 − (9.19)
2 EI EI 2 EI EI
When x2→∞, the elastic deformation of the seabed is only subjected to the
submerged weight and seabed resistance, i.e., y2=w/k. Therefore, c7 and c8
are both equal to zero and Eq. (9.18) then can be reduced to:
w
y2 = + c5 exp(−α x 2 )cos(β x 2 ) + c6 exp(−α x 2 )sin(β x 2 ) (9.20)
k
Bending moment and shear force of the touchdown segment can be

obtained:
d 2 y2
M 2 ( x ) = − EI (9.21)
dx 22
d 3 y2
S2 ( x ) = EI (9.22)
dx 23
9.3.4 Boundary Conditions

As the hang-off location of lazy-wave riser is hinged to the floating pro-
duction platform by a ball joint, the boundary condition of the HOP in the
global coordinate system (x, y) can be expressed as:
 y(0) = 0

HOP  θ (0) = θ 0 (9.23)

 M (0) = 0
where θ0 represents the inclination angle at HOP, which can be measured

by an instrument.
For the boundary-layer segment, the continuity conditions of the dis-
placement, inclination angle, curvature, and shear at the BLP should be
guaranteed according to the two coordinate systems (x, y) and (x1, y1):
 y1 ( L ) = y BLP

 θ1 ( L ) = θ BLP
BLP  (9.24)
 κ 1 ( L ) = κ BLP
 S1 ( L ) = SBLP

For the touchdown segment, the continuity conditions of the dis-

placement, inclination angle, bending moment, and shear, at the TDP
should also be guaranteed according to the two coordinate systems
(x1, y1) and (x2, y2):
 y1 (0) = y 2 (0) = 0

 θ1 (0) = θ 2 (0)
TDP  (9.25)
 M1 (0) = M 2 (0)
 S1 (0) = S2 (2)

9.4 Numerical Solution

In this numerical solution, the inclination angle at the HOP θ0 can be
obtained during the installation process. However, the tension at the
HOP T0 is unknown in the analysis. Once the value of T0 is determined,
the configurations of the hang-off section and buoyancy section can be
easily obtained by recursive method form HOP according to Eqs. (9.26)
to (9.33). Note, the outer diameter, D, submerged weight per unit length,
w, and flexural stiffness EI should be replaced with the equivalent values
De, we, and EIe, respectively, when calculating the behavior of the buoyancy
section.
Vi = Vi−1 + Fτds sin θi−1 – Fnds cos θi−1 – wds (9.26)
Hi = Hi−1 + Fnds sin θi−1 + Fτds cos θi−1 (9.27)
Ti = Vi 2 + Hi2 (9.28)
θi = arctan(Vi/Hi) (9.29)
xi = xi−1 + ds cos θi−1 (9.30)

yi = yi−1 + ds sin θi−1 (9.31)
dθi
Mi = − EI (9.32)
ds
dMi
Si = − (9.33)
ds
When calculating the configuration of the decline section by recursive

method, one problem occurs, which is that the arc length of the decline
section L3 is unknown and when to terminate the recursive calculation.
Then, the inclination angle at boundary-layer point θBLP is chosen to
determine the arc length of the decline section. Once the value of θBLP is
obtained, the configuration of the decline section can also be calculated
according to Eqs. (9.26) to (9.33).
For the boundary segment and touchdown segment, the analytical
expressions (9.13) and (9.20) of respective configuration can be derived
on the basis of linear deformation theory of beam. As the lazy-wave
configuration of the riser is divided into three segments, the continu-
ity conditions of the displacement, inclination angle, shear, curvature,
or bending moment at BLP and TDP should be considered according to
Eqs. (9.24) and (9.25). Equations (9.34) to (9.37) represent the continuity
conditions at BLP, and Eqs. (9.38) to (9.41) represent the continuity con-
ditions at TDP. Note, considering the diameter of the riser laid on the sea-
bed, the actual water depth at TDP should reduce the radius of the riser in
Eq. (9.34).
y1(L) = yBLP – (WD – D/2) (9.34)
y1′( L ) = tan(θ BLP) (9.35)
− ELy1′′( L ) = M BLP (T0 ,θ BLP) (9.36)
EIy1′′′( L ) = SBLP (T0 ,θ BLP) (9.37)

y1(0) = y2(0) = 0 (9.38)
y1′(0) = y 2′ (0) (9.39)
y1′′(0) = y 2′′(0) (9.40)
y1′′′(0) = y 2′′′(0) (9.41)
The calculation process developed in this chapter can be summa-

rized in the following steps. First, the hang-off tension, T0, and incli-
nation angle at the BLP, θBLP, should be estimated in advance. According
to Eqs. (9.26) to (9.33), the configuration of the riser suspended in the
Start
Input initial parameters
Assume initial inclination angle θBLP at BLP
Assume initial hang-off tension T0
Calculate the configuration of cable segment
Solve the 7×7 algebraic problem including

equations (44), (45) and (47) - (50)
Adjust the No
|WD-D/2-yBLP+y1(L)| <ε1
value of T0
Yes
No Adjust the
|S+ BLP-S- +
BLP|/S BLP<ε2 value of θBLP
Yes
Output the final configuration, inclination angle,
bending moment and shear
End
Figure 9.7 Flow chart of numerical calculation.

water can be calculated. In addition, the displacement, inclination angle,

bending moment, tension, and shear at the BLP can also be calculated,
which will be a function that depends on the two variables T0 and θBLP.
Then, considering the continuity conditions of the displacement, inclina-
tion angle, shear, curvature, or bending moment at BLP and TDP, Eqs.
(9.34) to (9.41) can be structured with seven unknowns c1, c2, c3, c4, c5, c6,
and L. Combining Eqs. (9.35), (9.36), and (9.38)–(9.41), seven unknowns
c1, c2, c3, c4, c5, c6, and L can be calculated efficiently from the seven equa-
tions. It can be found that Eqs. (9.34) and (9.37) are remaining. As the
value of T0 and θBLP are estimated and not true, T0 can be derived by Eq.
(9.34) and θBLP can be derived by Eq. (9.37). Additionally, it should be
noted that Eq. (9.34) ought to be satisfied prior to Eq. (9.37) to make
the solution more efficient. This is because the displacement continuity
should be determined prior to the shear continuity. At last, once the nine
equations above are all satisfied, the final configuration of the riser and its
mechanical behavior can be obtained. Figure 9.7 shows the main steps of
the numerical solution for predicting the behavior of deep water lazy-wave
riser. In the flow chart, ε1 and ε2 represent the specified small quantities.
9.5 Finite Element Model

A finite element model by OrcaFlex is employed to simulate the nonlinear
behavior of deep water lazy-wave riser. Figure 9.8 illustrates a schematic
z
200m
x
Figure 9.8 Sketch of OrcaFlex Model.

view of lazy-wave riser model, and the main features of this FEM are pre-
sented below.
9.5.1 Environment
A 3D space is set up to model the sea environment. The seabed model
is defined with “Flat” shape type and linear stiffness, neglecting the effect
of seabed friction. Current is assumed to be a steady flow and current
method is adopted by interpolation method. The wind load is neglected
in the model.
9.5.2 Riser
Based on the lumped mass method, the riser is modeled as a line, which
is divided into a series of line segments.[7] The riser is thus modeled by
this element with constant outer diameter, submerged weight, and flexural
stiffness, except for the buoyancy section. The buoyancy section is mod-
eled with the equivalent outer diameter, submerged weight, and flexural
stiffness, which is quite different from other parts of lazy-wave riser.

The HOP of lazy-wave riser is fixed to the sea surface, and the bottom-end
of the riser on the seabed is anchored to the seabed. The two connections
are both pin joints.

To verify the reliability of the analytical model, comparisons are per-
formed with the results obtained from the proposed method and FEM
by OrcaFlex. To perform the analysis, basic properties of the riser and
environment are shown in Tables 9.1 and 9.2, respectively. Comparisons
of lazy-wave configuration, tension, bending moment, and shear are
shown in Figures 9.9, 9.10, 9.11, and 9.12, respectively. From these figures,
the results obtained from the two methods seem in very strong qualita-
tive agreement. It can strongly verify the accuracy and reliability of the
analytical model. Differences between the analytical model and FEM by
OrcaFlex are shown in Table 9.3. Also, differences in TDP area are shown
in Table 9.4.
176
Table 9.1 Riser parameters.

Hang-off section Buoyancy section Axial
Parameter w(N/m) L1(m) D(m) EI(KN·m2) we(N/m) L2(m) De(m) EIe(KN·m2) Stiffness EA(KN)
Value 501.7 1176 0.20 4.736E2 −548.2 672 0.426 4.736E2 1.652E6
(Note: Mechanical behavior of decline section is the same with hang-off section).
Deepwater Flexible Risers and Pipelines
Table 9.2 Other environmental parameters.

Parameter WD(m) ρ(kg/m3) k(N/m2) vc(m/s) θ0(deg) Cd Cτ
Value 1,500 1,025 2E5 0.2 87 0.7 0.008
0
Analytical model
200 FEM by OrcaFlex
400
600
800
y (m)
1000
1200
1400
1600
800 700 600 500 400 300 200 100 0
x (m)
Figure 9.9 Comparison of lazy-wave configurations.
500
Analytical model
450 FEM by OrcaFlex
400
350
300
Tension (KN)
250
200
150
100
50
0
800 700 600 500 400 300 200 100 0
x (m)
Figure 9.10 Comparison of lazy-wave tensions.

10 8
Analytical model Analytical model
8 7
Bending moment (KN·m)

FEM by OrcaFlex FEM by OrcaFlex
6 6
4 5
2
4
0
3
-2
2
-4
-6 1
-8 0
-10 -1
800 700 600 500 400 300 200 100 0 760 740 720 700 680 660 640 620 600
x (m) x (m)
(a) (b)
Figure 9.11 (a) Comparison of lazy-wave bending moments. (b) Comparison of bending
moments in TDP area.
1.50
Analytical model 1.25
Analytical model
1.25 FEM by OrcaFlex FEM by OrcaFlex
1.00
1.00
0.75 0.75
Shear (KN)
Shear (KN)
0.50 0.50
0.25
0.25
0.00
-0.25 0.00
-0.50 -0.25
800 700 600 500 400 300 200 100 0 770 760 750 740 730 720 710 700
x (m) x (m)
(a) (b)
Figure 9.12 (a) Comparison of lazy-wave shears. (b) Comparison of shears in TDP.
Table 9.3 Result comparisons of analytical model and FEM by OrcaFlex.

Item Msag (KN·m) March (KN·m) THOP (KN) xTDP (m) ysag (m) yarch (m)
Analytical model 8.440 −8.897 485.9 753.85 912.5 606.7
FEM by OrcaFlex 8.406 −8.849 486.2 750.82 912.7 606.6
Difference (%) 0.40 0.54 0.06 0.40 0.02 0.02

Table 9.4 Result comparisons in TDP area.
Item emax (mm) Mmax (KN·m) MTDP (KN·m) θTDP (°) Smax (KN) STDP (KN) TTDP (KN)
Analytical model 4.481 6.869 2.605 0.3488 1.225 1.176 33.93
FEM by OrcaFlex 4.533 6.887 2.656 0.3554 1.256 1.205 32.72
Difference (%) 1.15 0.26 1.92 1.86 2.45 2.41 3.70

Note: emax represents the maximum embedment into the seabed.
Lazy-Wave Static Analysis
179
As shown in Figure 9.9, the lazy-wave configurations closely coincide

with one another. It should be noted that the thicker line in the figure
represents the buoyancy section. The horizontal distances from the HOP
to the TDP obtained from the analytical model and FEM by OrcaFlex are
753.85 and 750.82, respectively, and the difference is less than 1%. As illus-
trated in Figure 9.10, the tensions from the two methods also correlate
well. It can be found that three local maximum tensions occur at the HOP,
LP, and DP. Among the three local maximum tensions, the tension at the
HOP seems to be the maximum, which is far less than the hang-off tension
915.7 kN in simple free-hanging configuration under the same condition.
Figure 9.11 shows the comparison of riser’s bending moments. The
overall bending moments almost perfectly match (shown in Figure 9.11a),
and the main difference arises near the TDP (shown in Figure 9.11b). As
expected, three local maximum bending moments occur near the TDP,
the peak point in arch bend, and the lowest point in sag bend. In addition,
the values of bending moments at HOP, LP, and DP turn out to be zero.
From Figure 9.11b, we can find that the two curves of bending moments
near TDP are almost parallel, which is mainly caused by the location dif-
ference of TDP. The local maximum bending moments near the TDP from
the analytical model and FEM by OrcaFlex are 6.869 and 6.887 KN·m,
respectively, which are very close.
The comparison of riser’s shears is shown in Figure 9.12, which high-
lights the localization around the TDP (see Figure 9.12b). As shown in
Figure 9.12a, the overall shears match very well, except at the three loca-
tions of HOP, LP, and DP. The shear distortion is mainly caused by neglect-
ing the flexural stiffness of the zones near HOP, LP, and DP due to the
boundary phenomenon. Furthermore, the two curves of the shears near
the TDP are also almost parallel. The reason is the same with that of the
bending moment. The local maximum shears near the TDP from the ana-
lytical model and FEM by OrcaFlex are 1.225 and 1. 256 KN, respectively.
In addition, the curve of the shear at the BLP by the analytical model does
not seem smooth. This is because the flexural stiffness is considered in the
boundary-layer segment, but not considered in the decline section.
9.7 Parameter Analysis

9.7.1 Effect of Seabed Stiffness
Seabed stiffness plays an important role in the embedment analysis of the
pipe-soil interaction. Embedment into the seabed is one of the principal
factors that determine the riser’s lateral resistance to prevent from being
moved sideways.[8] It is, therefore, necessary to investigate the effect of seabed
stiffness. Six different kinds of seabed stiffnesses ranging from 50 to 300
kN·m2 are selected for the analyses, and the corresponding configurations
look almost the same. As shown in Figure 9.13, the local maximum bend-
ing moment in TDP area changes little as the seabed stiffness increases.
However, the bending moment at TDP surprisingly decreases. Figure 9.14
shows the results of the local maximum shear in TDP area and shear at
TDP with different seabed stiffness, which both increase accordingly as the
9
Mmax in TDP area
8 Moment at TDP
Bending moment (KN·m)
2
50 100 150 200 250 300
Seabed stiffness (KN·m2)
Figure 9.13 Bending moment with the variation os seabed.
1.30
Smax in TDP area
1.25 Shear at TDP
1.20
1.15
Shear (KN)
1.10
1.05
1.00
0.95
50 100 150 200 250 300
Figure 9.14 Shear with the variation of seabed.

variation of seabed stiffness
Inclination angle at TDP ( ° )

0.7
0.6
0.5
0.4
0.3
50 100 150 200 250 300

Seabed stiffness (KN•m2)
Figure 9.15 Inclination angle at TDP with the variation of seabed stiffness.
seabed stiffness increases. Also, the results of the inclination angle at TDP
and maximum embedment into the seabed are shown in Figures 9.15 and
9.16, respectively. It can be found that the inclination angle at TDP decreases
dramatically with the increasing of seabed stiffness. The maximum embed-
ment into the seabed also decreases, similar to the inclination angle at TDP.
When the seabed stiffness goes to infinity, the inclination angle at TDP and
maximum embedment into the seabed would tend to be zero. Thus, large
seabed stiffness will reduce the on-bottom stability of the riser. In short, it
can be concluded that seabed stiffness has scarce effect on the configura-
tion, tension, and bending moment, but leads to the increasing of the shear
in TDP area and the decreasing of embedment into the seabed.
9.7.2 Effect of Hang-Off Inclination Angle

The effect of hang-off inclination angle is very important in practical
applications. It is one of the principal factors that influence the mechanical
stiffness
14
Maximum embedment (mm)
12
10
8
6
4
2
50 100 150 200 250 300
Figure 9.16 Maximum embedment with the variation of seabed stiffness.

behavior of lazy-wave configuration. To evaluate the effect of different

hang-off inclination angles on lazy- wave configuration, the analyses are
conducted with the hang-off inclination angle ranging from 83° to 88°. The
corresponding results, such as lazy-wave configuration, tension, bending
moment, and shear, are illustrated in Figures 9.17 to 9.20, respectively. As
shown in Figure 9.17, the lazy-wave configuration becomes steeper and
steeper with the increase in hang-off inclination angle. From this figure,
the horizontal distance from HOP to TDP decreases remarkably as the
hang-off inclination angle increases. The tension at HOP declines by
3.22%, from 499.5 to 483.4 kN, while the tension at TDP declines by
60.74%, from 67.0 to 26.3 kN. This shows that hang-off inclination angle
has a more significant effect on the tension at TDP than on the tension at
0
83º
200 84º
400 85º
86º
600 87º
88º
y (m)
800
1000
1200
1400
1600
1200 1000 800 600 400 200 0
x (m)
Figure 9.17 Lazy-wave configuration with the variation of hang-off inclination angle.
500
400
Tension at HOP
Tension (KN)
300
Tension at TDP
200
100
0
83 84 85 86 87 88
Hang-off inclination angle (deg)
Figure 9.18 Tension with the variation of hang-off inclination angle.

HOP. Also, since lazy-wave configuration tends to be steeper and steeper,

the three local maximum bending moments increase as the hang-off incli-
nation angle increases (seeing Figure 9.19). Shear results due to the vari-
ation of the hang-off inclination angle are demonstrated in Figure 9.20.
From this figure, both the local maximum shear in TDP area and shear
at TDP increase with the increasing of hang-off inclination angle, and the
rate of the growth increases as the hang-off inclination angle increases. It
can be concluded that a large hang-off inclination angle will result in
the steep configuration and appropriately reduce the tension of lazy-wave
configuration, yet aggravate the bending moment and shear.
15 Mmax in sag bend

12
Mmax in arch bend
Bending moment (KN·m)
9
Mmax in TDP area
6
3
0
−3
−6
−9
−12
83 84 85 86 87 88
Figure 9.19 Bending moment with the variation of hang-off inclination angle.
1.5 Smax in TDP area

1.4 Shear at TDP
1.3
1.2
Shear (KN)
1.1
1.0
0.9
0.8
0.7
0.6
83 84 85 86 87 88
Figure 9.20 Shear with the variation of hang- off inclination angle.
9.7.3 Effect of Buoyancy Section Length

The buoyancy section is a long length of the riser attached with buoyancy
modules. Due to the lifting force of these buoyancy modules, the lazy-wave
configuration is formed. Thus, the influence of buoyancy section length is
well investigated with its length ranging from 522 to 772 m. The relevant
results are presented here. Figures 9.21, 9.22, 9.23, and 9.24 represent the
results of lazy-wave configuration, tension, bending moment, and shear,
respectively. As shown in Figure 9.21, the lazy-wave configuration tends
to be steep and the length of the suspended riser in water becomes larger
and larger with the increase of the buoyancy section length. In addition,
0
522m
200 572m
400 622m
672m
600 722m
800 772m
y (m)
1000
1200
1400
1600
800 700 600 500 400 300 200 100 0
x (m)
Figure 9.21 Lazy-wave configuration with the variation of buoyancy section length.
500
400
Tension at HOP
Tension (KN)
300
Tension at TDP
200
100
0
500 550 600 650 700 750 800
Length of buoyancy section (m)
Figure 9.22 Tension with the variation of buoyancy section length.

10.0
7.5
5.0
Bending moment (KN·m) Mmax in sag bend
2.5
Mmax in arch bend
0.0
Mmax in TDP area
−2.5
−5.0
−7.5
−10.0
500 550 600 650 700 750 800
Figure 9.23 Bending moment with the variation of buoyancy section length.
Smax in TDP area

1.23
Shear at TDP
1.22
1.21
Shear (KN)
1.20
1.19
1.18
1.17
1.16
500 550 600 650 700 750 800
Figure 9.24 Shear with the variation of buoyancy section length.
the horizontal displacement from HOP to TDP increases accordingly as

the buoyancy section length increases. As the buoyancy section length
grows from 522 to 772 m, the tension at HOP decreases by 10. 28%, from
519.4 to 466.0 kN. On the other hand, the tension at TDP changes very
little, decreasing by 2.61%, from 34.5 to 33.6 kN. It can clearly be seen that
the buoyancy section length has a more remarkable influence on the
tension at HOP than on the tension at TDP, which is different from the
effect of hang-off inclination angle. Furthermore, the three local bending
moments surprisingly rise linearly with the increasing of the buoyancy
section length. The local maximum shear in TDP area and shear at TDP
also increase nearly linearly with the increase of the buoyancy section
length. It is thus obvious that buoyancy section length plays a significant
role in the optimization of lazy-wave configuration. Large length of the
buoyancy section will also reduce the tension of lazy-wave configuration,
and aggravate the bending moment and shear, which is the same effect
with hang-off inclination angle.
9.8 Conclusions
This study presents a reasonable analytical model to study the mechanical
behavior of deep water lazy-wave riser considering the effect of environ-
mental loads and elastic seabed. Owing to the boundary-layer phenom-
enon, the analytical model is divided into three parts to analyze: cable
segment suspended in the water, boundary-layer segment near the TDP,
and touchdown segment laid on the seabed. To verify the accuracy of
the proposed model, a finite element model by OrcaFlex was employed.
The results obtained from the analytical model and FEM by OrcaFlex
closely coincide with one another, indicating the accuracy of the proposed
model in predicting the static behavior of deep water lazy-wave riser.
Furthermore, parametric studies, such as seabed stiffness, hang-off incli-
nation angle, etc., are conducted to determine the different effects of the
selected parameters on the static behavior of lazy-wave riser. The following
conclusions can be drawn:
• Seabed stiffness has minimal effect on the configuration,

tension, bending moment, and shear of the riser suspended
in the water. However, the increase of seabed stiffness will
lead to the decrease of bending moment at TDP, and the
increase of shear in TDP area. Also, the embedment into
the seabed decreases dramatically with the increasing of
eabed stiffness. Therefore, high seabed stiffness will reduce
the on-bottom stability of the riser.
• The lazy-wave configuration becomes steeper and steeper
with the increase of hang- off inclination angle. The tension
decreases linearly with the increasing of hang-off inclination
angle, especially for the tension at TDP. However, bend-
ing moment and shear increase expectedly as the hang-off
inclination angle increases, and the rate of the growth also
increases as hang-off inclination angle increases. It is thus
clearly evident that hang-off inclination angle has a signifi-

cant influence on the mechanical behavior of lazy-wave con-
figuration. Large hang-off inclination angle will result in the
steep configuration and reduce the tension of lazy-wave con-
figuration but will aggravate the bending moment and shear.
• With the increase of buoyancy section length, the lazy-wave
configuration tends to become steeper and the length of the
suspended riser becomes longer. The tension decreases lin-
early with the increase of buoyancy section length, especially
for the tension at HOP. However, bending moment and
shear increase nearly linearly as the buoyancy section length
increases. It is thus obvious that buoyancy section length
plays a significant role in the optimization of lazy-wave con-
figuration. Large length of buoyancy section will also reduce
the tension of lazy-wave configuration but increases the
bending moment and shear.
In conclusion, the comparison analyses have demonstrated the abil-

ity of the proposed analytical model in predicting the static behavior of
deep water lazy-wave riser. This chapter provides reasonable guidelines
for engineering application of riser installation. Further research needs
to be carried out on dynamic response of deep water lazy-wave riser.
References
1. N. Ismail, R. Nielsen, and M. Kanarellis, Design Considerations for
Selection of Flexible Riser Configuration, PD-Vol. 42, Offshore and Arctic
Operations, ASME, 1992.
2. Ruan W, Bai Y, Cheng P. Static analysis of deepwater lazy-wave umbilical on
elastic seabed[J]. Ocean Engineering, 2014, 91: 73–83.
3. Garrett D L. Dynamic analysis of slender rods[J]. J. Energy Resour. Technol.;
(United States), 1982, 104(4).
4. Croll J G A. Bending boundary layers in tensioned cables and rods[J].
Applied ocean research, 2000, 22(4): 241–253.
5. Bai Y, and Bai Q. Subsea pipelines and risers. Elsevier, 2005.
6. Palmer A C, King R A. Submarine Pipeline Engineering[M]. PennWell
Books, 2004.
7. Orcina Ltd., Visual ORCAFLEX User Manual, Ulverston, Cumbria, UK,
2000.
8. Palmer A. Touchdown indentation of the seabed[J]. Applied Ocean Research,
2008, 30(3): 235–238.
10
Steep-Wave Static Configuration
10.1 Introduction
To overcome harsh environments and high pressure, six main configura-
tions of steel catenary riser have been developed, such as simple catenary,
lazy wave, steep wave, lazy S, steep S, and pliant wave. Each configuration
has its advantage to make itself better appropriate for particular applica-
tions. Among these configurations, the steep wave riser is considered better
to accommodate the geometry in deep water, and it is suitable for applica-
tions where the riser terminates at a subsea base or completion. As shown
in Figure 10.1, typical steep wave riser is composed of a taut lower pipe
connected to a positively buoyant section of catenary pipeline which itself
joins the sagging part of the riser which in turn ends at the vessel[1]. It is
similar to the lazy wave configuration but has a nearly vertical connection
at the seabed where a subsea base controls the touch-down point. Steep
wave risers can accommodate large vessel offsets and effectively reduce the
potential compression buckling at TDP. It is able to maintain its configu-
ration even if the riser fluid density changes. However, it requires a subsea
base and subsea bend stiffener.
Relatively, little research has been done on steep wave riser. Pinto and
Lima[2] presented the final results of FSO steep-wave riser configuration
design in the deepwater (815 m) Roncador field and gave a complete report
of that project. Then, numerical simulations by Riflex were performed to
study the influence of the armor on the response of the steep-wave riser in
water at a depth of 300 m under irregular waves[3].
Sun and Bo[4] adopted the lumped mass method to perform the global
analysis on the steep wave flexible riser and found that the steep wave riser
is not very sensitive to the ocean current. Steep wave configuration and
lazy wave configuration used on the Auger TLP were specifically devel-
oped for application with a tanker FPSO located in a harsh environment[5].
Key mechanical design issues and material selection for different ser-
vice conditions are discussed. Budgetary costs for a range of typical field
189
Sea level
FPSO
Buoyancy modules
Subsea base
Figure 10.1 Configuration of steep wave riser.
application are also presented in this chapter. Santillan S. T.[6] put forward
an elastic approach to describe the analysis of steep-wave flexible risers by
a finite difference technique. A free-vibration analysis was also conducted
for small-amplitude oscillations.
This chapter presents a numerical method to study the static two-
dimensional behavior of steep wave riser by taking the effect of current
loading in to account. Due to the lifting effect of buoyancy modules
attached to part of the riser, the behavior of the steep wave riser is sub-
jected to large deformation and behaves nonlinearly. Therefore, based on
the large deformation beam theory and mechanics equilibrium principle,
a set of non-linear ordinary differential equations describing the motion
of the steep wave riser are developed. Then, finite difference method and
shooting method are combined effectively to solve the ordinary differen-
tial equations with zero moment boundary conditions at both the seabed
end and surface end of the steep wave riser. The resulting non-linear finite
difference formulations can be solved by the Newton-Raphson method
with the Jacobian matrix evaluated in an approximate manner. To verify
the accuracy of the proposed method, FEM by OrcaFlex is employed and
the results from the two methods show excellent consistency.
10.2 Configuration Calculation

The behavior of the steep wave riser is described in terms of the position of
the centerline of the riser according to the classical theory of rods[7]. Due to
the attachment of buoyancy modules, the configuration of steep wave riser
can be divided into three segments, as shown in Figure 10.2.
Steep-Wave Static Configuration 191
Sea level HOP
Hang-off segment
Current
Buoyancy segment
L3
Water depth WD
y L2
DP
LP
L1
TDP Touch-down segment

Seabed x
Figure 10.2 Steep wave riser configuration.
1. Touch-down segment: the riser from the touch-down point

(TDP) to the decline point (DP).
2. Buoyancy segment: the riser from the DP to the lift point
(LP).
3. Hang-off segment: the riser from the LP to the hang-off
point (HOP) at the vessel or platform.
In this figure, one global coordinate system (x, y) is established in the

steep wave riser model by locating the origin at the TDP. Ocean current is
set to be positive when flowing in the positive x direction.
10.2.1 Touch-Down Segment

The arc length of the touch-down segment is L1. It is connected vertically
to the seabed foundation by a subsea base, at the TDP. Due to the lifting
force by the buoyancy segment and the displacement constraint of the sub-
sea base, the configuration of the touch-down segment behaves like part
of a simple catenary riser. Thus, the behavior of the touch-down segment
is subject to large-angle deflections. Based on the large deformation beam
theory, the basic differential equations for the touch-down segment can
be obtained from the force system acting on a differential segment of the
touch-down segment, as illustrated in Figure 10.3[8].
y T+δT
θ+δθ
F+δF
δs
Fn
δs
Fτ
F wδs
θ
T
x
Figure 10.3 Forces acting on the touch-down segment.
Resolving forces normal to the differential segment axis leads to
(F + δF)cos δθ – (T + δT)sin δθ – F + wδs cosθ + Fnδs = 0 (10.1)
1
Fn = ρw Cd D ds vc sinθ vc sinθ (10.2)
2
F and T represent shear force and axial tension, respectively; δF and δT

represent the increment of shear force and axial tension; θ is the inclina-
tion angle of the differential segment from horizontal direction to axial
direction; δθ represents the increment of the inclination angle; w is the
submerged weight per unit length of the riser; δs is the length of the dif-
ferential segment; ρw is sea density; vc is current velocity; D is the outer
diameter of the riser; Fn represents the drag force in the normal direction
per unit length of the riser, acting perpendicular to the riser axis. The drag
force in the normal direction Fn can be obtained by Morison’s equation.
As the length of the differential segment δs is very small, the increment
variable δθ could be infinitesimal. Therefore, cos(δθ)≈1 and sin(δθ)≈δθ.
For simplification, Eq. (10.1) then leads to
dF dθ
−T + w cosθ + Fn = 0 (10.3)
ds ds
Furthermore, the shear force F herein is related to the bending moment

by F = dM/ds. Hence, Eq. (10.3) can be derived as follows:
d2M dθ
−T + w cosθ + Fn = 0 (10.4)
ds 2
ds
Based on the large deflection beam theory, the bending moment then
can be obtained:
dθ
M = EIκ = EI (10.5)
ds
where EI denotes the flexural stiffness of the riser and κ denotes the cur-
vature of the riser.
Combining Eqs. (10.4) and (10.5), the final normal mechanics equilib-
rium differential equation can be expressed as
d 3θ dθ
EI −T + w cosθ + Fn = 0 (10.6)
ds 3
ds
On the other hand, resolving forces parallel to the segment axis leads to
(F + δF)sin δθ + (T + δT)cos δθ − T – wδs sin θ + Fτδs = 0 (10.7)
1
Fτ = ρwCτ π D ds vc cosθ vc cosθ (10.8)
2
where Fτ represents the drag force in the axial direction per unit length of
the riser, and Cτ is the drag coefficient in the axial direction.
For simplification, Eq. (10.7) then can be simplified as
dθ dT
F + − w sinθ + Fτ = 0 (10.9)
ds ds
Then, the final axial mechanics equilibrium differential equation can be

expressed as
d 2θ dθ dT
EI + − w sinθ + Fτ = 0 (10.10)
ds 2 ds ds
As both the axial deformation and shear deformation are not consid-
ered in the analysis, the horizontal and vertical displacements in the global
coordinate (x, y) then can be expressed as
∫ ∫
x = cosθ ds , y = sinθ ds (10.11)
Once the displacements, inclination angle and curvature along the riser
are obtained, the bending moment M and shear force F can be derived as
follows:
dθ
M = EIκ = EI (10.12)
ds
dM d 2θ
F= = EI 2 (10.13)
ds ds
It should be noted that shear force is set to be positive when in clockwise

direction.
10.2.2 Buoyancy Segment

The arc length of the buoyancy segment is L2. To decrease the upper end
tension required at the vessel and decouple the vessel motions from the
TDP of the riser, buoyancy modules made of syntactic foam or thin-walled
Aircans are uniformly attached along the long length of the buoyancy seg-
ment. As the buoyancy force of the buoyancy modules and riser in the
buoyancy segment is larger than their weights, the configuration behaves
like an arch bend. To simplify the numerical analysis, a uniformly distrib-
uted lifting force is acting on the buoyancy segment, which is equivalent to
the loading combination of the buoyancy force and weight. The uniformly
distributed lifting force here is called “equivalent submerged weight unit
length”. Also, the equivalent outer diameter and equivalent flexural stiff-
ness are considered in the buoyancy segment. The adoption of the equiv-
alent outer diameter here is to make sure that the buoyancy force of the
equivalent buoyancy segment is equal to the buoyancy combination of
the buoyancy modules and riser per unit length. The equivalent flexural
stiffness is considered, due to some effect of these buoyancy modules on

the flexural stiffness of the buoyancy segment. However, the equivalent
flexural stiffness of the buoyancy segment is generally close to the flexural
stiffness of the riser due to the small length of buoyancy modules attached.
Thus, it is generally acknowledged that the equivalent flexural stiffness of
the buoyancy segment is equal to the flexural stiffness of the riser. The
vertical displacement of the highest point in the arch bend is herein set
as yarch. Due to its large- deflection behavior, the configuration analysis of
the buoyancy segment is also based on the large deflection beam theory.
According to the mechanics equilibrium principle, the governing differ-
ential equations for the buoyancy segment are similar with those of the
touch-down segment. The main difference is that the flexural stiffness EI,
submerged weight w, and outer diameter D in Eqs. (10.1) to (10.13) should
be replaced by the equivalent values EIe, we, and De, respectively.
10.2.3 Hang-Off Segment

The arc length of the hang-off segment is L3. It is hung off by a ball joint
to prevent high bending moment induced by waves, current, and vessel
motion at the hang-off location. However, as the study is confined to the
two-dimensional static analysis, the effects of wave and vessel motions are
not considered in this study. Due to the lifting effect of the buoyancy seg-
ment and the vessel, the configuration of the hang-off segment behaves
like sag bend. The vertical displacement of the lowest point in the sag bend
herein is set as ysag. As the mechanical behavior of the hang-off segment
is precisely the same with that of the touch-down segment, the governing
differential equations for the hang-off segment are also the same with that
of the touch-down segment.

The touch-down location of the steep wave riser is generally connected
to a subsea base, controlling the TDP. The connection between the riser
and subsea base herein is simplified as a hinged connection. Therefore, the
boundary condition of the TDP in the global coordinate system (x, y) can
be expressed as:
M (0) = 0 (10.14)
On the other hand, the hang-off location of the steep wave riser is usu-
ally hinged to the vessel or platform by a ball joint. The inclination angle
of the hang-off point can be measured by instrument. Thus, the boundary
condition of the hang-off point in the global coordinate system (x, y) can
be expressed as
y(L1 + L2 + L3) = WD (10.15)
θ(L1 + L2 + L3) = θ (10.16)
M(L1 + L2 + L3) = 0 (10.17)
where WD represents water depth.
10.3 Numerical Solution

In this study, the behavior of the steep wave riser is mainly dominated
by Eqs. (10.6) and (10.10). According to the two differential equilibrium
equations and boundary conditions, it can be found that the essence of
steep wave riser’s global analysis is a multipoint boundary value problem.
Thus, the finite difference method is applied to solve the two differential
equations numerically in terms of the coordinates of adjacent nodes. For
the case of equally spaced nodes, it can be expressed at node i, in terms of
the element length, ds, between nodes and the angles at four nodes: the
node itself, the two nodes immediately above it, and the one node imme-
diately below it, as shown in Figure 10.4. The node number here is marked
from 1 to (n + 1), where the TDP is set as node 1 and the hang-off point is
set as node (n + 1). By considering the effect of ocean current, Eqs. (10.6)
and (10.10) can be expressed numerically in finite difference form by Eqs.
(10.18) and (10.19), respectively:
θi + 2 − 3θi +1 + 3θi − θi −1 θ −θ 1
EI − Ti i +1 i + w cosθi + ρwCd D vc sinθi vc sinθi = 0
(ds ) 3
ds 2
(10.18)
i+2
i+1 ds
i ds
i-1 ds
θ
x
Figure 10.4 Forces acting on the touch-down segment.
θi +1 − 2θi + θi −1 θi +1 − θi Ti +1 − Ti 1
EI ⋅ + − w sinθi + ρwCτ π D vc cosθi vc cosθi = 0
(ds )2 ds ds 2
(10.19)
where i runs from 1 to n.

When the length of the riser reaches between L1 and (L1 + L2), that is to
say when calculating the behavior of boundary segment, the flexural stiff-
ness EI, submerged weight, w, and outer diameter, D, should be replaced
by the equivalent values EIe, we, and De, respectively. Then, a 2n × 2n matrix
equation can be formulated containing n + 3 angle unknowns (θ0, θ1, θ2,
……, θn+1 and θn+2) and n + 1 tension unknowns (T1, T2, ……, Tn and Tn+1).
According to the boundary condition (10.14) and (10.17), the value of θn+2
can be expressed in terms of θn+1 and θn, and θ0 in terms of θ1 and θ2. Also,
the value of θn+1 can be obtained from the boundary condition (10.16).
Finally, it can be found that there are still 2n + 1 unknowns (θ1, θ2, ……, θn
and T1, T2, ……, Tn+1) left in the matrix equation.
Then, the shooting method[9] is employed and the top tension Tn+1 is
assumed in advance. According to Newton-Raphson iteration method, the
2n × 2n matrix equations including 2n unknowns (θ1, θ2, ……, θn and T1,
T2, ……, Tn) can be solved effectively. As Tn+1 is assumed and not the true
value, the value of the Tn+1 should be adjusted until Eq. (10.15) is satisfied
according to the shooting method. Figure 10.5 shows the main procedures
of the numerical solution. In the flow chart, ε respects the specified small
quantity.
Start
Input basic parameters
Assume initial top tension Tn+1 at HOP
Calculate steep wave configuration by

shooting method
Adjust the No
|WD-y1|<ε
value of Tn+1
Yes
Obtain initial value of θ1, θ2,……θn, and T1,
T2, …… Tn+1, select Tn+1 as a known parameter
Solve the 2n×2n matrix equation with 2n unknowns

θ1, …… θn, and T1, ……Tn by Newton-Raphson method
Yes
No Adjust the
|WD-y1|<ε
value of Tn+1
Output the final configuration, inclination

angle, bending moment and shear
End
Figure 10.5 Flow chart of numerical calculation.

In order to verify the accuracy and reliability of the numerical method,
comparisons are performed with the results obtained from the proposed
method and FEM by OrcaFlex. To perform the analysis, basic properties
of the flexible riser and environment are shown in Tables 10.1 and 10.2,
respectively. Comparisons of steep wave configuration, tension, bending
moment, and shear force are shown in Figures 10.6 to 10.9, respectively.
From these figures, it can be seen that the results obtained from the two
methods coincide well. It can strongly verify the accuracy and reliability
of the proposed method. Differences between the numerical method and
FEM by OrcaFlex are shown in Table 10.3.
Table 10.1 Flexible riser parameters.

Items Value
L1(m) 450
L2(m) 672
L3(m) 1076
w(N/m) 501.7
we(N/m) −785.9
D(m) 0.220
De(m) 0.460
EI(kN.m2) 4.736×102
(Note: the flexural stiffness of the boundary segment is
assumed to be the same with the other two segments, that is
to say EI=EIe).
Table 10.2 Environmental parameters.

Items Value
WD(m) 1,500
3
ρ(kg/m ) 1,025
vc(m/s) 0.2
θ(°) 80
Cd 0.7
Cτ 0.008
As shown in Figure 10.6, the steep wave configurations from the two
methods seem to be consistent. The difference mainly happens at the hang-
off segment. The horizontal distances from the hang-off point to the TDP
obtained from numerical method and FEM by OrcaFlex are 1,018 and
1,012 m, respectively, and the difference is less than 3.0%. Also, the highest
vertical locations of the arch bend from the two methods are 891.6 and
893.2 m, respectively. The lowest vertical locations of the sag bend from the
two methods are 755.1 and 755.1 m, respectively.
1600
Numerical method
1400 FEM by OrcaFlex
1200
1000
y (m)
800
600
400
200
0
0 100 200 300 400 500 600 700 800 900 1000
x (m)
Figure 10.6 Comparison of steep-wave configurations.
Table 10.3 Results of numerical method vs. FEM by OrcaFlex.

Items Numerical method FEM by OrcaFlex Difference (%)
xHOP (m) 1,018 1,012 0.59
yarch (m) 891.6 893.2 1.79
ysag (m) 755.1 755.1 0.01
THOP (kN) 454.7 454.7 0.00
TTDP (kN) 225.9 226.4 0.22
March (kN·m) −4.520 −4.588 1.48
Msag (kN·m) 2.919 2.939 0.68
FDP (kN) −0.1145 −0.1187 3.54
FLP (kN) 0.8261 0.8258 0.04
The tensions illustrated in Figure 10.7 seem also in excellent agreement.

It can be found that three local maximum tensions occur at the hang-off
point, LP, and DP, respectively. Among the three local maximum tensions,
the tension at the hang-off point seems to be the maximum, arriving at
454.70 kN. While under the same condition, the upper end tension in sim-
ple catenary configuration without the addition of buoyancy modules can
reach upto 915.63 kN, which is greater than that in the steep wave con-
figuration. It can be seen that the steep wave configuration can effectively
reduce the upper end tension.
Figure 10.8 shows the comparison of riser’s bending moments, and the
overall bending moments almost perfectly match. As expected, two local
500
Numerical method
450
FEM by OrcaFlex
400
350
300
Tension (KN)
250
200
150
100
50
0
0 100 200 300 400 500 600 700 800 900 1000
x (m)
Figure 10.7 Comparison of steep-wave tensions.
5
Numerical method
4
FEM by OrcaFlex
3
2
1
0
–1
–2
–3
–4
–5
0 100 200 300 400 500 600 700 800 900 1000
x (m)
Figure 10.8 Comparison of steep-wave bending moments.

maximum bending moments occur at the peak point in the arch bend
and the lowest point in the sag bend. In addition, the values of bending
moments at the LP and DP turn out to be zero. It can be found that the
bending moments in the neighborhood of the LP and DP change very
obviously. The maximum bending moments in the steep wave configura-
tion from numerical method and FEM by OrcaFlex are −4.520 and −4.588
kN·m respectively, which are very close. The comparison of riser’s shear
forces is shown in Figure 10.9. The overall shear forces in Figure 10.9 match
very well, except at the two locations of HOP and TDP. Furthermore, two
local maximum shear forces occur at the DP and LP and the shear force in
other location is very close to zero.
0.90
Numerical method
0.75 FEM by OrcaFlex
0.60
Shear force (KN)
0.45
0.30
0.15
0.00
−0.15
0 100 200 300 400 500 600 700 800 900 1000
x (m)
Figure 10.9 Comparison of steep-wave shear forces.
0.03
0.00
Shear force (KN)
–0.03
–0.06 DP
DP
–0.09
Numerical method
–0.12 FEM by OrcaFlex
–0.15
100 105 110 115 120 125 130 135 140 145 150
x (m)
Figure 10.10 Comparison of steep-wave shears near DP.

1.0
0.8 Numerical method
FEM by OrcaFlex
0.6
Shear force (KN)
LP
0.4
LP
0.2
0.0
−0.2
450 455 460 465 470 475 480 485 490 495 500
x (m)
Figure 10.11 Comparison of steep-wave shears near LP.
Figures 10.10 and 10.11 show the shear forces in the localization around
the DP and LP, respectively. The two curves of shear forces in the neighbor-
hood of the DP are almost parallel and the two curves of shear forces in the
neighborhood of the LP are also almost parallel. It is mainly caused by the
location difference of the DP and LP obtained from the two methods. The local
maximum shear forces in the neighborhood of DP from numerical method
and FEM by OrcaFlex are −0.1145 and −0.1187 kN, respectively. The local
maximum shear forces in the neighborhood of DP from numerical method
and FEM by OrcaFlex are 0.8261 and 0. 8258 kN, respectively.
10.5 Parametric Analysis

10.5.1 Effect of Buoyancy Segment’s Equivalent Outer
Diameter
Five buoyancy segment’s equivalent outer diameters (0.450, 0.455, 0.460,
0.465, and 0.470) are chosen to study its influence on steep wave riser. The
corresponding results, such as steep wave configuration, tension, bending
moment, and shear force, are illustrated in Figures 10.12 to 10.15, respec-
tively. Steep wave riser configuration shown in Figure 10.12 become steeper
and steeper with the increasing of the buoyancy segment’s equivalent outer
diameter. As a result, the horizontal distance between TDP to HOP tends
to be shortened. While the equivalent outer diameter increases from 0.450
to 0.470 m, the tension at HOP decreases by 2.34% from 460.7 to 449.9
kN. However, the tension at TDP increases by 42.51% from 187.7 to 267.5
kN. Thus, it can be seen that the increasing equivalent outer diameter can
reduce slightly the tension at HOP and aggravate dramatically the tension
1600
0.450m
1400 0.455m
0.460m
1200 0.465m
1000 0.470m
y (m)
800
600
400
200
0
0 200 400 600 800 1000
x (m)
Figure 10.12 Steep wave configuration with the variation of buoyancy segment’s
equivalent outer diameter.
equivalent outer diameter

500
450
400
Tension (KN)
350
300
250
200
150 Tension at HOP
Tension at TDP
100
0.450 0.455 0.460 0.465 0.470
Equivalent diameter of buoyancy segment (m)
Figure 10.13 Tension with the variation of buoyancy segment’s equivalent outer diameter.
at TDP. As illustrated in Figure 10.14, the equivalent outer diameter has

little influence on the maximum bending moment in the sag bend but can
increase the maximum bending moment in the arch bend. The equivalent
outer diameter also has little effect on the maximum shear force at LP and
yet can reduce the maximum shear force at DP. Therefore, it can be seen
that the equivalent outer diameter of the buoyancy segment can affect the
behavior of the touch-down segment and buoyancy segment, but has little
influence on that of the hang-off segment.
7.5
Mmax in arch bend
5.0 Mmax in sag bend

2.5
0.0
–2.5
–5.0
–7.5
0.450 0.455 0.460 0.465 0.470
Figure 10.14 Bending moment with the variation of buoyancy segment’s equivalent outer
diameter.
1.0
0.8
0.6
Shear force(KN)
Fmax at DP
0.4
Fmax at LP
0.2
0.0
–0.2
0.450 0.455 0.460 0.465 0.470
Figure 10.15 Shear force vs. buoyancy equivalent outer diameter.
10.5.2 Effect of Buoyancy Segment Length

The influence of the buoyancy segment length is well investigated with its
length ranging from 522 to 772 m. The relevant results corresponding to
different buoyancy segment lengths are shown in Figures 10.16 to 10.19,
respectively. In Figure 10.16, steep wave configuration tends to be steep
and the vertical location of arch bend rises up accordingly as the buoy-
ancy segment length increases. However, it can be found that the buoyancy
1600
572m
1400 622m
672m
1200
722m
1000 772m
800
y (m)
600
400
200
0
0 200 400 600 800 1000
x (m)
Figure 10.16 Steep wave configuration with the variation of buoyancy segment length.
500
400
Tension (KN)
300
200
Tension at HOP
Tension at TDP
100
550 575 600 625 650 675 700 725 750 775 800
Figure 10.17 Tension with the variation of buoyancy segment length.
segment length has little effect on the horizontal distance between TDP
to HOP. In addition, the tension at HOP decreases by 8.57%, from 476.3
to 435.5 kN, while the tension at TDP increases by 60.05%, from 175.7 to
281.2 kN (shown in Figure 10.17). It shows that the buoyancy segment
length has a more prominent influence on the tension at TDP than on the
tension at HOP. Also, the two local maximum bending moments increase
linearly with the increasing of the buoyancy segment length. However, the
maximum shears at DP and LP decrease linearly. Thus, it can be found that
large length of the buoyancy segment can optimize the distribution of the
7.5
Mmax in arch bend
Bending moment (KN.m)
2.5
0.0
–2.5
–5.0
–7.5
550 575 600 625 650 675 700 725 750 775 800
Figure 10.18 Bending moment with the variation of buoyancy segment length.
1.2
Fmax at DP
1.0 Fmax at LP
0.8
Shear force (KN)
0.6
0.4
0.2
0.0
–0.2
550 575 600 625 650 675 700 725 750 775 800
Figure 10.19 Shear force with the variation of buoyancy segment length.
tension and shear force along the steep wave riser; however, it will aggra-
vate the bending moment.
10.5.3 Effect of Buoyancy Segment Location

The buoyancy segment location plays a vital role in the optimization of
steep wave riser. To study its influence, several locations of the buoyancy
segment are selected to move towards the hang-off point along the riser
length by ± 50 or ± 100 m. The effects of buoyancy segment location are
well illustrated in Figures 10.20 to 10.23. As buoyancy segment moves
closer to HOP, i.e., the hang-off segment length is shortened, the vertical
location of arch bend raises up accordingly.
As the buoyancy segment location varies from −100 to 100 m, the ten-
sion at HOP decreases from 508.1 to 402.1 kN, by 20.86%. On other hand,
the tension at TDP decreases from 278.0 to 175.1 kN, by 37.01%. Therefore,
buoyancy segment location can effectively reduce the tension at the two
1600
–100m
1400 –50m
0m
1200 50m
1000 100m
y (m)
800
600
400
200
0
0 200 400 600 800 1000
x (m)
Figure 10.20 Steep wave configuration with the variation of buoyancy segment location.
550
500
450
400
Tension (KN)
350
300
250
200
150 Tension at HOP
Tension at TDP
100
–100 –50 0 50 100
Location of buoyancy section (m)
Figure 10.21 Tension vs. buoyancy segment location.

7.5
Mmax in arch bend
2.5
0.0
–2.5
–5.0
–7.5
–100 –50 0 50 100
Location of buoyancy section (m)
Figure 10.22 Bending moment vs. buoyancy segment location.
1.0
0.8
0.6
Shear force(KN)
Fmax at DP
0.4 Fmax at LP
0.2
0.0
−0.2
−100 −50 0 50 100
Location buoyancy section (m)
Figure 10.23 Shear force vs. buoyancy segment location.
ends of steep wave riser when close to the hang-off point. In addition, it
also will result in the deterioration of maximum bending moments in arch
bend and sag bend, and the reduction of shear force along the riser.
10.5.4 Effect of Current Velocity

Current is usually assumed to be steady, and it could mainly induce the
drag force acting on the slender rod. The results of steep wave riser are
shown in Figures 10.24 to 10.27 corresponding to current velocity of −0.4,

−0.2, 0, 0.2, and 0.4 m/s. As shown in Figure 10.24, the effect of current
velocity is obvious on the steep wave configuration. Additionally, the ten-
sions at HOP and TDP vary a little because the magnitude of the drag
force acting on the riser caused by current is far less than the riser effective
1600
-0.4m/s
1400 -0.2m/s
0m/s
1200
0.2m/s
1000 0.4m/s
800
y (m)
600
400
200
0
0 200 400 600 800 1000
x (m)
Figure 10.24 Steep wave configuration vs. current velocity.
500
400
Tension (KN)
300
200
Tension at HOP
Tension at TDP
100
–0.4 –0.3 –0.2 –0.1 0.0 0.1 0.2 0.3 0.4
Current velocity (m/s)
Figure 10.25 Tension vs. current velocity.

7.5
Mmax in arch bend
2.5
0.0
–2.5
–5.0
–7.5
–0.4 –0.3 –0.2 –0.1 0.0 0.1 0.2 0.3 0.4
Figure 10.26 Bending moment vs. current velocity.
1.0
0.8
0.6
Shear force(KN)
Fmax at DP
0.4 Fmax at LP
0.2
0.0
0.2
0.4 0.3 0.2 0.1 0.0 0.1 0.2 0.3 0.4
Figure 10.27 Shear force with the variation of current velocity.
weight. The local maximum bending moments in arch bend and sag bend
descend to some extent. In Figure 10.27, as current velocity increases from
−0.4 to 0.4 m/s, the shear at HOP increases from 0.732 to 0.873 kN, by
19.26%, while the shear at TDP increases from −0.089 to −0.128 kN, by
43.82%. As expected, the change rates of tension, bending moment and
shear grow with the absolute value of current velocity increases. It can be
found that with current velocity increases, the effect on the behavior of
steep wave riser will be more and more obvious.
10.6 Conclusions
The mechanical behavior of 2D steep wave riser under current load is stud-
ied based on large deformation beam theory and mechanics equilibrium
principle. With an assumption that the riser connections to the subsea base
and to FPSO are both hinged connections, finite difference method and
shooting method are applied to solve the equations. Sensitivity of param-
eters, including buoyancy segment’s equivalent outer diameter, buoyancy
segment length, buoyancy segment location, and current velocity are stud-
ied carefully. Results from the proposed method have matched well with
the results from OrcaFlex. The results effectively demonstrate the ability
of the proposed method in predicting the non-linear static behavior of 2D
steep wave riser under current load.
References
1. Seyed F B, Patel M H. Mathematics of flexible risers including pressure and
internal flow effects. Marine structures, 1992, 5(2): 121–150.
2. Pinto P, Lima G. Installation of steep-wave flexible riser in deepwater
on roncador FSO. 20th International Conference on Offshore Mechanics and
Arctic Engineering, Brazil, June 3–8, 2001.
3. Bonnemaire B. Response of an Armoured Riser for Arctic Offshore Loading.
The Fourteenth International Offshore and Polar Engineering Conference.
International Society of Offshore and Polar Engineers, 2004.
4. Sun L, Qi B. Global analysis of a flexible riser. Marine Science and Application,
2011, 10(4): 478–484.
5. Hatton S A. Rigid Risers for Tanker FPSOs FPSO World Congress &
Technology Exhibition, Aberdeen, October. 1995: 800.
6. Santillan S. T., Virgin L. N., Plaut R. H., 2010. Static and dynamic behavior of
highly deformed risers and piplines. J Offshore Mech Arct. 132, 1–6.
7. Garrett D L. Dynamic analysis of slender rods. J. Energy Resour. Technol.;
8. Sparks C P. Fundamentals of marine riser mechanics: basic principles and
simplified analyses[M]. PennWell Books, 2007.
9. Bai Y, Yu B, Cheng P. Offshore Installation of Reinforced Thermoplastic
Pipe(RTP). The Twenty-second International Offshore and Polar Engineering
Conference. International Society of Offshore and Polar Engineers, 2012.
11
3D Rod Theory for Static
and Dynamic Analysis
11.1 Introduction
Nowadays, unbonded flexible pipe is extensively used to transport gas/
oil drilled from underneath the seabed to platforms in numerous offshore
engineering applications due to its advantages of good compliance with
platform movements, high flexibility to absorb harsh environmental load-
ings, good corrosion resistance, and low installation cost. So far, it has been
operated in water depths of more than 3,000 m. As illustrated in Figure
11.1, the typical structure of an unbonded flexible pipe is mainly made
of several steel and plastic concentric layers with low bending stiffness
but relatively high radial and longitudinal stiffness. Three types of metal-
lic layers turn out to mainly withstand the imposed structural loads
(Fergestad et al., 2014): inner carcass mainly provides strength against
external hydrostatic pressure and crushing loads during installation oper-
ations; pressure armor provides resistance against the hoop stress caused
by internal pressure; tensile armors provide strength against the axial stress
caused by internal pressure and by external loads.
As the global performance of an unbonded flexible riser in the marine
environment is crucial during the riser-configuration design phase, the rel-
evant research has drawn more and more attention. Due to the versatility
in handling complex flexible pipe profile and boundary conditions, finite
element method (FEM) turns out to be the most popular methodology to
solve nonlinear riser dynamics problems[2−7]. It could be interpreted as an
application of Galerkin’s method for a particular choice of shape func-
tions. On the basis of conservation of linear momentum and moment of
momentum, the rod theory[8−12] was put forward and improved to acquire
the riser/mooring line dynamic behavior by using Galerkin’s method. The
rod’s nodal position coordinate and its derivatives can be directly defined
in the global coordinate system. It is the author’s belief that, to date,
Texas A&M University[13−17] did the more prominent research on the rod
213
Outer sheath
Outer layer of tensile armour

Anti-wear layer
Inner layer of tensile armour
Anti-wear layer
Back-up pressure armour
Interlocked pressure armour
Internal pressure sheath
Carcass
Figure 11.1 Typical unbonded flexible pipe wall structure[1].
theory and has developed CABLE 3D numerical code to study the static/
dynamic response of the floating production system including the plat-
form, mooring lines and risers.
The current study on the riser dynamics is prevailingly aimed at the
rigid pipe. However, due to the multilayer structure and low bending stiff-
ness, the dynamic problem of the unbonded flexible pipe is considerably
different from that of rigid pipe, e.g., complex boundary condition and
bending hysteretic behavior. Traditionally, the flexible pipe global analy-
sis simply considers the top-end connection condition as hinged, and the
bend stiffener (BS) is excluded since its influence on the response is con-
trolled within a relatively small region. Only a few literature introduced
the bend restriction effect into the flexible pipe global analysis[18–19], and
numerous studies on the BS constraint mainly tended to the local analy-
sis[18, 20–25]. In addition, the bending hysteretic behavior of an unbonded
flexible pipe[26–31] has been widely investigated based on the local analy-
sis, such as FEM and experiment. To the author’s knowledge, the effect
of bending hysteretic behavior has not been adequately investigated in the
global analysis.
In this chapter, a developed finite element model based on the rod
theory is put forward to acquire the static/dynamic response of a flexible
riser subjected to hydrodynamic loadings and platform motion. The for-
mulations proposed are defined in the absolute coordinate. In this math-
ematical model, Newmark-β method combined with Newton-Raphson
iterative method is employed for the time domain integration of the riser
3D Rod Theory for Static and Dynamic Analysis 215
dynamics. The effects of the BS constraint and flexible pipe bending hyster-
etic behavior are also considered in the model to accurately simulate the
static/dynamic response of flexible pipe in the case study, especially for the
top-end region and touchdown zone (TDZ). Thus, the effects of BS con-
straint, bending hysteretic behavior and top angle constraint are also fully
investigated and some conclusions are obtained.
11.2 Nomenclature
ai Hermite cubic r, rn position vector

Ai pipe inner cross section area R pipe external radius
Ao pipe outer cross section area s undeformed length coordinate
D pipe external diameter S arc-length from the top-end
Di bend stiffener’s inner t time
diameter
Do bend stiffener’s outer Sb bend stiffener’s length
diameter
ey unit vector in y direction Tw wall tension
Eb Young modulus of bend Te effective tension
stiffener
EA pipe axial stiffness U internal flow velocity
EIb pipe stick bending stiff- uin displacement parameter

ness
EIe pipe effective bending WD water depth
stiffness
EI pipe slip bending stiffness λ, λm Lagrange multiplier
f
k seabed normal stiffness ε axial strain

L undeformed element length κ pipe local curvature
M system structural mass κcr pipe critical curvature
matrix
pi internal hydrostatic pressure ρi internal flow density
pm quadratic δκ curvature increment
po external hydrostatic pres- υ Poisson’s ratio

sure
q external forces per unit ξ a non-dimensional coordinate
length
qspring seabed vertical resistance
11.3 Mathematical Model

11.3.1 Governing Equations
As illustrated in Figure 11.2, the unbonded flexible pipe in the global
analysis model is assumed to be a slender rod and its static/dynamic
motion can be described in terms of the position of the rod centerline.
On account of its multilayer composite structure and internal slip between
each layer, the bending behavior of unbonded flexible pipes is a more
complex phenomenon to analyze than the axisymmetric load case and the
flexural response shows a pronounced hysteresis behavior (Fergestad et
al., 2014). Therefore, a bi-linear hysteresis model is generally adopted to
reflect the typical moment-curvature relation for an unbonded flexible
pipe and its bending hysteretic behavior will be discussed in detail later
in this chapter. Then, based on the original elastic rod theory (Nordgren,
1974; Garrett, 1982), the motion governing equation could be derived by
taking the internal flow into account:
y z
Wave x Platform motion
Current
Water depth
S
Bend stiffener constraint
Unbonded flexible pipe
Touchdown point
Anchoring point
Seabed
Figure 11.2 Global analysis model of unbonded flexible pipe.

Mr + 2 ρi AU
i r ′ + ( EI e r ′′ )′′ − (λ r ′ )′ = q (11.1)
2
where λ = Tw + po Ao − pi Ai − ρi AU
i − EI eκ 2 = Te − EI eκ 2 . λ is a scalar
function, also called Lagrange multiplier, since it is introduced as a result
of the assumption of extensible condition in the stretching constraint
equation below. The corresponding detailed derivation process can be
obtained from the paper by Garrett (1982). Herein, a prime denotes
differentiation with respect to arc-length s, and a superposed dot denotes
differentiation with respect to time t.
In this mathematical model, the flexible riser is considered to be
extensible and the stretch is linear and small. Therefore, the vector r must
satisfy a stretching constrain equation:
r′ · r′ = (1 + ε)2 ≈ 1 + 2ε (11.2)
Taking the combination of wall tension, internal pressure, external

pressure, and hoop stresses causing axial strains proportional to Poisson’s
ratio υ, the axial strain of the riser ε then can be expressed as
2
Tw + 2υ ( po Ao − pi Ai ) λ − (1 − 2υ )( po Ao − pi Ai ) + ρi AU (11.3)
ε= ≈ i
EA EA
where the item EIeκ2 is very small compared to the effective tension Te
and may be ignored when calculating the effective tension. Namely, λ
approximates to the effective tension Te.
Then, the constraint equation with small elongation can be derived
from Eqs. (11.2) and (11.3):
2
1 λ − (1 − 2υ )( po Ao − pi Ai ) + ρi AU (11.4)
(r ′ ⋅ r ′ − 1) − i
=0
2 EA
Based on the tensile tests of unbonded flexible pipes, Witz (1996)

found that the axial displacement versus axial load curve is nearly linear.
Therefore, the axial stiffness EA in the constraint equation is treated as
constant.
To effectively develop the element formulations, the position of the
rod r and Lagrange multiplier λ in the three-dimensional space could be
approximated as
rn (s , t ) = uin (t )ai (s ) λ (s , t ) = λm (t ) pm (s ) (11.5)
where n = 1–3 (representing x, y, and z directions, respectively), i = 1–4 and

m = 1–3. Hermite cubic ai(s) and quadratic pm(s) are interpolation func-
tions defined on the interval 0 ≤ s ≤ L. The expressions about Hermite
cubic and quadratic can be referenced in the paper by Garrett (1982).
As shown in Figure 11.3, the node variables uin and λm are the values
of the solution at the nodes of the finite element, which can be defined as
following:
u1n (t ) = rn (0, t ) u2n (t ) = Lrn′(0, t ) u3n (t ) = rn ( L , t ) u4n (t ) = Lrn′( L , t )

λ1 (t ) = λ (0, t ) λ2 (t ) = λ ( L / 2, t ) λ3 (t ) = λ ( L , t )
(11.6)
Therefore, the approximations of the other parameters, such as EIe, q, M,

and ε are similar with λ .
Since the governing equations are highly nonlinear, Galerkin’s method
is hence used to discretize the partial differential Eqs. (11.1) and (11.4)
in space, which may be reduced to a set of nonlinear ordinary differen-
tial equations in the time domain. According to Galerkin’s method, the
motion equation could be obtained by multiplying both sides of the equa-
tion with ai(s) and integrating it with respect to s from 0 to L:
∫ {Mra (s) − 2ρ AUra′(s) + EI r ′′a′′(s) + λ r ′a′(s) − qa (s)}ds

0
i i i i e i i i
{ }
L
= EI e r ′′ai′(s ) 0L + λ r ′ − ( EI e r ′′ )′ − 2 ρi AU
i r ai (s )
0
(11.7)
u22
y u23 y
-
(u11,u12,u13) u21 λ1
z z L
-
u41 λ2
(u31,u32,u33)
u43 u42 -
x λ3
x
(a) (b)
Figure 11.3 Illustration for node variables: (a) uin; (b) λ m .

The first term on the right-hand side of the equation above is related
to the moments at the element ends, and the second term is the forces at
the element ends. The right-hand terms symbol the generalized internal
forces, denoted as “fin”.
Inserting Eq. (11.5) and shape functions into Eq. (11.7), the following
set of ordinary differential equations could be obtained for small elonga-
tion element:
γ ikm Mnjmukj − 2ρi AU

i ζ iku
 kn + α ikm EImukn + βikm λmukn = µimqmn + fin
(11.8)
Similarly, the discretized form of the constraint equation (11.9) could

also be obtained by multiplying both sides of the equation with pm(s) and
integrating it with respect to s from 0 to L:
1 1 λ − (1 − 2v )( pon Ao − pin Ai ) + ρi AiU 2

βikmuinukn − τ m − ηmn n =0
2 2 EAn
(11.9)
where the tensor summation algorithm is employed, and i and k run

from 1 to 4; j, m, and n run from 1 to 3; and
L L L
α ikm =
∫ 0
ai′′(ξ )ak′′(ξ ) pm (ξ )dξ βikm =
∫0
ai′(ξ )ak′ (ξ ) pm (ξ )dξ γ ikm =
∫0
ai (ξ )ak (ξ ) pm (ξ )dξ
L L L L
µim =
∫ 0
ai (ξ ) pm (ξ )dξ τ m =
∫0
pm (ξ )dξ ηmn =
∫ 0
pm (ξ ) pn (ξ )dξ ζ ik =
∫ 0
ai′((ξ )ak (ξ )dξ
(11.10)
where ξ = s/L, is a non-dimensional coordinate.

Since the hydrodynamic loadings acting on the flexible riser depend on
the riser’s instantaneous position, Newton-Raphson iterative method com-
bined with Newmark- β method may be employed to solve this problem.
Based on Newton-Raphson iterative method, Eqs. (11.8) and (11.9) could
be reconstructed in a matrix form aδx = b for each element, where the
unknown vector δx consisting of the 15 unknowns is defined as
{ }
T
δ x = δ u11 , δ u12 , δ u13 , δ u21 , δ u22 , δ u23 , δλ1 , δλ2 , δ u31 , δ u32 , δ u33 , δ u41 , δ u42 , δ u43 , δλ3

(11.11)
As the generalized internal forces between the two neighboring ele-

ments are equal in size and in the opposite direction, the generalized
internal forces can be cancelled out. Nevertheless, it is necessary to acquire
the boundary conditions of the first and last nodes. Once the global stiff-
ness matrix system is assembled, the static/dynamic problem can be solved
efficiently. Finally, the corresponding static/dynamic response of a flexible
riser can be acquired.
11.3.2 Bending Hysteretic Behavior

In fact, the bending behavior of an unbonded flexible pipe approximately
shows hysteretic phenomenon due to the complex structure components.
This hysteretic behavior may be explained by the internal frictional slip
between each layer. As shown in Figure 11.4, the hysteretic behavior stiff-
ness curve can be decomposed into three main phases (Péronne et al.,
2015).
1) When the curvature κ of an unbonded flexible pipe is ini-

tiated and very small, the armor layers stick together and
the internal friction between each layer prevents the relative
slip, thereby obtaining a high initial stiffness, EIb. This phase
is called stick or rigid mode.
EIf
Moment
EIb
κcr Curvature
Figure 11.4 Hysteretic bending stiffness curve (Péronne et al., 2015).

2) Then, the curvature increases with the bending moment,

and the available internal friction is insufficient to prevent
parts of the tensile armors from slipping axially. The slip-
page will reduce further increases in the bending moment.
Consequently, the effective bending stiffness, i.e., the slope
of the moment-curvature curve, reduces continuously
until the internal friction is fully overcome and the tensile
armors are free to slip. This phase is called the transition
phase between the stick mode and the slip mode.
3) Finally, when the bending moment reaches up to the fric-
tion moment and internal friction forces are dissipated, the
armor layers slip, resulting in a low stiffness, noted as EIf.
This phase is called slip mode. The curvature at which the
internal friction is fully overcome is called critical curva-
ture κcr.
In this chapter, the bending hysteresis curve is simplified as a bi-linear

hysteresis model. As the hysteretic behavior is not considered in the static
analysis, the static effective bending stiffness EIe is related to the current
curvature magnitude, yet has nothing to do with the curvature loading
path. As illustrated in Figure 11.5, it can be calculated with the following
expressions:
slip mode
C
stick mode
EIf
Bending Moment
B 1
EIe
D
1
EIb
κcr
1
Curvature
A κcr
κ
Figure 11.5 Parallelogram hysteresis loop (Zhang et al., 2008).

 EI = EI for κ ≤ κ cr ;
e b
 (11.12)
 EI e = [EIb ⋅κ cr + EI f ⋅ (κ − κ cr )]/ κ for κ ≥ κ cr .
The effective bending stiffness here is defined as the slope of the

moment-curvature curve. As the stick bending stiffness EIb is much larger
than the slip bending stiffness EIf, it can be seen from the figure that the
static effective bending stiffness decreases with the curvature in the slip
phase according to Eq. (6.12).
On the other hand, the bending hysteretic behavior is taken into account
in the dynamic analysis. Thus, the dynamic effective bending stiffness EIe
is not only related to the current curvature magnitude but also related to
the curvature loading path. The effective reversal point therefore should be
recorded in the curvature loading path, and the dynamic effective bending
stiffness EIe in each step then can be obtained from the previous curvature
loading condition:
 EI
e ,n +1 =( EI e ,n ⋅κ n + EI b ⋅ δκ n +1 ) κ n +1 for κ n+1 ,κ n instick mode;

 EI e ,n +1 =( EI e ,n ⋅κ n + EI f ⋅ δκ n +1 ) κ n +1 for κ n+1 , κ n inslipmode;
 EI for κ n instick mode;κ n+1 inslipmodee.
 e ,n +1 =( EI e ,n ⋅κ n + EI b ⋅ δκ n +1, st + EI f ⋅ δκ n +1, sl ) κ n +1
(11.13)
where the subscript n or n + 1 denotes the time step; “st” and “sl” rep-
resent the stick mode and the slip mode, respectively.
11.3.3 Bend Stiffener Constraint

In offshore engineering, the top end of an unbonded flexible pipe usu-
ally hangs from the platform with a certain angle ranging between 5° and
15° from the vertical (Tanaka et al., 2009). As the unbonded flexible pipe
is usually subjected to the hydrodynamic loadings and platform move-
ments, its top-end connection may experience failure from accumulation
of fatigue damage due to cyclic excitation from wave induced platform
motions and excessive bending stress induced by large curvature due
to extreme loads (Vaz et al., 2004). In order to avoid this situation and to
ensure the pipe integrity, an ancillary component called a BS is employed
and applied at the riser top end with rigid connection attached to the plat-
form to provide a gradual physical transition between the platform and the
flexible riser to avoid over-bending and accumulation of high cyclic fatigue
damage. The BS’s root end is rigidly fixed to the platform and its tip end is
free. BSs are usually fabricated with polymeric materials, such as a poly-
ether polyurethane (Vaz et al., 2004). In this study, the top-end boundary
condition is rigidly fixed and the restriction effect of the BS is also taken
into account to reflect a more realistic condition in the top-end region.
Thus, more accurate stresses can be obtained and a better estimation
of the fatigue damage can also be achieved. The polyurethane stress-strain
constitutive model herein is assumed to be linear, that is to say, the BS may
possess an elastic Young’s modulus. Therefore, the bending stiffness of the
BS may strongly depend on the polyurethane constitutive model and geo-
metric configuration. Also, the loads acting on the BS could be described
as a function of the BS’s length and bending stiffness.
As illustrated in Figure 11.6, the BS geometry is generally given in
terms of the constant inner diameter Di and variational outer diame-
ter Do(S) as a function of the BS’s length. The effects of the radial clear-
ance and friction force between the BS and flexible pipe are ignored in this
study. Then the bending stiffness distribution along the BS’s length can be
obtained:
Ebπ 4
EIb (S ) =
64
(
Do (S ) − Di4 ) 0 ≤ S ≤ Sb (11.4)
Then, the total bending stiffness distribution along the BS’s length
could be written as the bending stiffness combination of the BS and
unbonded flexible pipe:
Rigid connection
Flexible pipe
Do(S)
Di
Sb
Radial clearance
Figure 11.6 Bend stiffener schematic diagram.

EItotal (S) = EIe (S) = EIb (S) 0 ≤ S ≤ Sb (11.15)
Considering the effect of BS constraint in the riser’s global analysis, the

total bending stiffness EItotal along the BS’s length should take the place of
the effective bending stiffness EIe of the flexible pipe in Eq. (11.1) when
calculating the riser behavior along the BS.
11.3.4 Pipe-Soil Interaction

The pipe-soil interaction is of great importance in the global dynamic anal-
ysis, and most studies generally consider the seabed as either a rigid seabed
or a Winkler foundation model for simplicity. However, You et al. (2008)[32]
found that rigid seabed assumptions result in extremely conservative con-
clusions. Also, Thethi et al. (2001)[33] pointed out that the prediction of
riser’s fatigue life depends on the seabed stiffness. Therefore, Katifeoglou
et al. (2012)[34], Elosta et al. (2013)[35], and Bai et al. (2015)[12] adopted a
nonlinear pipe-soil interaction model to obtain the more accurate dynamic
response and strength performance of the riser in the TDZ. In this chapter,
a Winkler foundation model is employed to realize the pipe-soil interaction
to improve the robustness. The nonlinear pipe-soil interaction model will
be introduced into this mathematical model to better calculate the riser’s
fatigue life. According to Winkler foundation model, the pipe- soil interac-
tion can be considered in terms of a pipe resting on the seabed with a series
of vertical linear springs (see Figure 11.7). The vertical resistance acting on
the pipe per unit length can be expressed as:
 k( R − r ⋅ e − W ) R − r ⋅ e y − WD > 0
 y D
q spring =  (11.16)
 0 R − r ⋅ e y − WD ≤ 0
Due to the seabed vertical resistance, an extra term representing the

distributed seabed vertical resistance should be added to the governing
equation of motion. Multiplying both sides of Eq. (11.16) by the shape
function ai(s) and then integrating it with respect to s from 0 to L of an
element touching the seabed:
L L L
∫ 0
q spring ai (s )ds =
∫
0
ai (s ) ds ⋅[k( R − WD )] −
∫
0
ai (s )ak (s )ds ⋅ k ⋅ uk 2
(11.17)
Self-weight penetration Maximum trench Pipe centerline

depth
Mudline
Figure 11.7 Winkler foundation model.
11.4 Case Study

In this case study, the top end of the unbonded flexible pipe is rigidly fixed
to the platform with its location (0, 0, 0) and top angle 6°. Meanwhile, the
anchoring point is hinged onto the seabed with its location (520, −600, 0).
The main parameters of an unbonded flexible pipe are summarized in
Table 11.1. The bending moment-curvature curve not only depends on
the contact pressures between the layers of the pipe but also depends on the
annular condition. Thus, the curve varies in each cross-section. However,
due to the lack of moment-curvature data, a constant moment-curvature
hysteretic loop curve is adopted for the whole pipe, as illustrated in
Table 11.1 Unbonded flexible pipe parameters.

Designation Unit Quantity
Segment length m 1,000
Outer diameter m 0.44471
Inner diameter m 0.29834
Dry weight kg/m 217.37
Axial stiffness N 5.13447 × 108
Poisson’s ratio / 0.3
Flow density kg/m3 881
Flow velocity m/s 1.0
Figure 11.8 (Hou et al., 2013), which may result in the inaccuracy of the
curvature distribution. Also, a 2-m-long BS is rigidly fixed to the platform
with a Young’s Modulus of 45 MPa. It has a root outer radius of 0.625 m and
tip outer radius of 0.450 m. It also has a constant inner radius of 0.450 m.
The profile of the BS is shown in Figure 11.9.
Table 11.2 shows the main parameters of the environmental and the
hydrodynamic coefficients. The current is not considered in this study.
Therefore, airy wave theory is adopted to simulate the wave response.
The propagation direction of the wave is set along the positive x direction.
100
50
Bending Moment (kN·m)
–50
–100
–0.1 –0.05 0 0.05 0.1
Curvature (rad/m)
Figure 11.8 Bending hysteresis loop.
0.80
External profile
0.60
Bend stiffener profile (m)
0.40
Internal profile
0.20
0
0 0.5 1.0 1.5 2.0
Curvature (rad/m)
Figure 11.9 Bend stiffener profile.

Table 11.2 Environment and hydrodynamic coefficients.

Designation Unit Quantity
Water depth m 600
Seabed stiffness N/m2 6 × 103
Sea density kg/m3 1,025
Wave period s 8
Wave height m 3
Wave phase ° 0
Normal drag / 1.2
Tangential drag / 0.008
Normal added-mass / 1.0
Tangential added-mass / 0.008
11.5 Results and Discussion
The mathematical model described in Section 11.3 has been implemented

by the computer program MATLAB for the static/dynamic analysis of a flex-
ible riser. In order to better understand its static/dynamic response under
the combined effects of hydrodynamic loadings and top harmonic exci-
tation in marine environment, the corresponding static/dynamic behavior,
i.e.,, riser configuration, effective tension, bending curvature, and declina-
tion angle, are conducted and studied in this section. Furthermore, to
verify the accuracy and reliability of the proposed mathematical model,
comparisons between the mathematical model and FEM by OrcaFlex are
presented below.
11.5.1 Static Analysis

Figures 11.10 to 11.13 show the corresponding static result comparisons of
the riser configuration, effective tension, bending curvature, and declina-
tion angle obtained from the two methods, respectively. From these figures,
it can be found that the static results obtained from the two approaches
seem in very strong qualitative agreement. It can strongly verify the
0 FEM by OrcaFlex
Mathematical model
–100
–599.2
–200
–300 –599.6
y
–400
–600.0
–500
160 170 180 190
–600
0 100 200 300 400 500

x
Figure 11.10 Unbonded flexible riser configuration.
800
90
700
600
75
Effective tension (kN)
500
400 60
600 625 650 675 700
300
200
100 FEM by OrcaFlex

Mathematical model
0
0 200 400 600 800 1000
Riser arc length (m)
Figure 11.11 Effective tension along arc length.
accuracy and reliability of the proposed mathematical model in obtaining

the static response of a flexible riser. Significant differences between the
proposed model and FEM by OrcaFlex are recorded in Table 11.3.
As illustrated in Figure 11.10, both the curves of the static global
configurations and local configurations in the TDZ approach very closely.
The static x-displacements of the touchdown point (TDP) derived from
0.020
FEM by OrcaFlex
0.015 Mathematical model
0.010
Curvature (rad/m)
0.005
0.000
0.000 0.02
–0.005
0.01
–0.005
–0.010
0.00
–0.010
–0.015 0 2 4 6 8 500 600 700
0 200 400 600 800 1000

Figure 11.12 Curvature along arc length.
180
FEM by OrcaFlex
Mathematical model
160
Declination angle (°)
140 176
175
120
174
100
173
0.0 2.5 5.0 7.5 10.0
80
0 200 400 600 800 1000
Figure 11.13 Declination angle along arc length.
FEM by OrcaFlex and mathematical model are 169.65 and 169.77 m,

respectively, which are little in difference. Also, from Table 11.3, the max-
imum seabed penetrations in the two solutions are nearly equal 0.2164 m
(0.4866D). Figure 11.11 illustrates the comparison of the effective tension
distribution curves and it also matches well. Nevertheless, the main
tension difference occurs in the TDZ with large curvature, which may
be primarily induced by the neglect of the item EIeκ2 when calculating
the effective tension. Due to the BS constraint and top rigid-connection,
Table 11.3 Static result comparisons.

FEM by Mathematical Difference
Designation Unit OrcaFlex model (%)
x displacement of m 169.6522 169.7682 0.07
TDP
Maximum seabed m 0.2164 0.2164 0.03
penetration
Top effective tension kN 766.1419 764.3202 0.24
TDP effective tension kN 62.6622 62.3531 0.49
Anchoring effective kN 62.5688 62.6513 0.13
tension
Maximum curvature rad/m 0.0166 0.0166 0.18
in TDZ
Curvature at BS’s root rad/m −0.0063 −0.0064 0.95
end
Curvature at BS’s tip rad/m −0.0062 −0.0062 0.02
end
Declination angle at ° 174. 6810 174.6991 0.01
BS’s tip end
two local maximum curvatures generate in the top-end region and TDZ,
respectively (see Figure 11.12). In the top-end region, there are typically
two critical points at the BS flange, where the curvature limit may be
exceeded, namely, the BS’s root end and tip end. In this case study, the
curvature magnitudes in the BS’s root end and tip end are very close.
The bending curvature first decreases along the constant-diameter part of
the BS, then raises up along the variational-diameter part, finally drops
beyond the scope of the BS constraint. The zone of a riser significantly
influenced by the bending stiffness EI(s) equals to about 3 EI/T from
the end terminations, according to Sparks[36]. It can be observed that the
curvature tends to be zero after 7-m riser arc length as shown in Figure
11.12, which can demonstrate the valid zone influenced by the bending
stiffness.
As shown in Figure 11.13, the declination angle first raises up in the top-
end region, then declines along the flexible riser suspended in the water
and finally keeps nearly constant along the flexible riser resting on the
seabed. The declination variation in the top-end region might be domi-

nant by the top-end constraint.
11.5.2 Dynamic Analysis

The dynamic results in the case study will be extracted on two critical
regions with large curvature distribution, which are considered as the most
dangerous for the extreme loads and fatigue damage. The first region is
the top-end region, where the high tension and large curvature may be
generated, and energy concentration may be induced. The second is the
TDZ, where large curvature may be produced, and axial compression phe-
nomenon may occur that affect armor wires. The dynamic behavior of the
proposed mathematical model under top heave harmonic excitation with
amplitude 1.0 m and period of 8 s is examined with the aid of Figures 11.14
to 11.21. These figures depict the time history of top effective tension, cur-
vatures at BS’s root end and tip end, maximum bending curvature in TDZ,
vertical displacement at static TDP, and effective tension at TDP for a 64s
simulation time. It is immediately apparent that the result comparisons
show good consistency.
11.5.2.1 Top-End Region

According to the static results above, the static maximum effective tension
occurs at the top end of the flexible riser. Hence, the top-end region may
be the most critical zone for the tension loading. Under the top heave
harmonic excitation, the top effective tension in Figure 11.14 tends to peri-
odically vary after the steady dynamic response is attained. The two ten-
sion curves obtained from the two methods match very well. After steady
dynamic response, the top effective tension ranges from 637.8 to 888.3
kN by the mathematical model; meanwhile, the value ranges from 640.3 to
889.2 kN derived from FEM by OrcaFlex. In addition, in the mathematical
model, the peak of the top effective tension increases 16.22% in compari-
son with the static value 764.32 kN, which is essentially affected by the top
heave harmonic excitation characteristics.
As mentioned above, the effect of the BS constraint is taken into account
to simulate the realistic response of a flexible riser, especially for the top-
end region. Whereas, with the BS constraint, two critical points may occur
with large curvature, i.e., BS’s root end and tip end. Figures 11.15 and
11.16 show the time history responses of the curvatures at the BS’s root
end and tip end. It can be observed that after three periods, both the cur-
vatures at the BS’s root end and tip end vary dramatically and the variation
950
Mathematical model
900 FEM by OrcaFlex
Top effective tension (kN) 850
800
750
700
650
600
0 8 16 24 32 40 48 56 64
Time (s)
Figure 11.14 Time history response of top effective tension.
−0.0045
Curvature at bend stiffener’s root end (rad/m)
Mathematical model
−0.0050 FEM by OrcaFlex
−0.0055
−0.0060
−0.0065
−0.0070
−0.0075
−0.0080
−0.0085
0 8 16 24 32 40 48 56 64
Time (s)
Figure 11.15 Time history response of curvatures at bend stiffener’s root end.
trends by the two techniques are very similar. Additionally, the dynamic
curvature results obtained from FEM by OrcaFlex are slightly more severe
than those from the mathematical model. The curvature magnitude at the
BS’s root end by the mathematical model, as illustrated in Figure 11.15,
ranges from −0.0051 to −0.0081 rad/m. In the meantime, the curvature
variation at the bend stiffener’s tip end (see Figure 11.16), is from −0.
0050 to −0.0079 rad/m. These two curvature distribution curves approach
somewhat closely.
–0.0045
Curvature at bend stiffener’s tip end (rad/m)

–0.0050
–0.0055
–0.0060
–0.0065
–0.0070
–0.0075
–0.0080
Mathematical model
–0.0085 FEM by OrcaFlex
–0.0090
0 8 16 24 32 40 48 56 64
Time (s)
Figure 11.16 Time history response of curvatures at bend stiffener’s tip end.
11.5.2.2 Touchdown Zone

Suffering from large curvature and high cumulative fatigue damage,
the TDZ is generally recognized as the most critical region in the riser
dynamic analysis. Hence, more and more attention has been paid to it.
Figures 11.17 and 11.18 show the time history responses of the maximum
curvature in the TDZ and the corresponding bending moment vs. curva-
ture relation respectively, which reflect the realistic time-domain response
between the bending moment and curvature. As shown in Figure 11.17,
0.030
Mathematical model
0.028
Maximum curvature in TDZ (rad/m)
FEM by OrcaFlex
0.026
0.024
0.022
0.020
0.018
0.016
0.014
0.012
0 8 16 24 32 40 48 56 64
Time (s)
Figure 11.17 Time history response of maximum curvature in TDZ.

50
Mathematical model
FEM by OrcaFlex
40
Bending moment (kN·m)

30 1
20
10
0
0.000 0.005 0.010 0.015 0.020 0.025 0.030
Bending curvature (rad/m)
Figure 11.18 Bending moment vs. curvature in TDZ.
the peak of the maximum bending curvature by the mathematical model

seems slightly larger than that obtained from FEM by OrcaFlex. In the
mathematical model, the maximum bending curvature in the TDZ var-
ies periodically from 0.0140 to 0.0262 m and the corresponding curvature
peak increases 57.83% in contrast with the static value 0.0166 m. Thus, the
impact of the top heave harmonic excitation on the dynamic curvature
is greatly significant in the TDZ. From Figure 11.18, it can be found that
the bending behavior of the maximum curvature point in the TDZ shows
the hysteresis loop response, which would be beneficial to calculate more
accurate stress variation and consequently the fatigue cumulative damage.
The hysteresis loop response by the mathematical model is more remark-
able than that from FEM by OrcaFlex, which can better validate the result
in Figure 11.17.
The effect of the effective tension on the effective bending stiffness varia-
tion is well investigated in Figure 11.19. After the steady dynamic response
is attained, the relation between the effective tension and effective bending
stiffness of the maximum curvature point in the TDZ shows an approxi-
mately closed loop. When the effective tension increases, the bending cur-
vature decreases. As a result, the effective bending stiffness then decreases,
such as the process ① in Figure 11.18.
The time history responses of the vertical displacement and effec-
tive tension at the static TDP are illustrated in Figures 11.20 and 11.21,
respectively. In this case study, the vertical coordinate of the mud line
is −599.7776 m. In Figure 11.20, the TDP vertical displacement above
the mud line varies more remarkably than that below the mud line.
1.8E6
Effective bending stiffness (N·m2)

1.7E6
1.6E6
1.5E6
1.4E6
1.3E6
30 40 50 60 70 80 90 100
Effective tension (kN)
Figure 11.19 Effective tension vs. effective bending stiffness in TDZ.
−599.0
Mathematical model
Vertical displacement at TDP (m)
FEM by OrcaFlex
−599.2
−599.4
−599.6
−599.8
−600.0
0 8 16 24 32 40 48 56 64
Time (s)
Figure 11.20 Time history response of vertical displacement at static TDP.
Additionally, under the top heave harmonic excitation, the TDP effective
tension derived from the mathematical model ranges from 36.4014 to
91.2413 kN, the peak of which increases 46.33% in comparison with the
static value 62.3531 kN. It follows that the top heave harmonic excitation
has a more significant enhancement on the TDP effective tension than the
top effective tension. Once the top heave harmonic excitation tends to be
more sever, effective compression phenomenon may occur in the TDZ,
which may lead to the global instability of a flexible riser and, therefore, to
the possible overbending.
110
Mathematical model
100 FEM by OrcaFlex
Effective tension at TDP (kN)

90
80
70
60
50
40
30
0 8 16 24 32 40 48 56 64
Time (s)
Figure 11.21 Time history response of effective tension at static TDP.
11.5.3 Effect of Bend Stiffener Constraint

The main difference between the flexible riser dynamics and rigid riser
dynamics is the top connection condition. In the flexible riser dynamic
analysis, the top connection is rigid connection and the BS is attached
onto the top end to limit the riser bending curvature. However, in the
rigid riser dynamic analysis, the top hinged connection is employed with
the stress joint. In this chapter, three top connection conditions are well
investigated, i.e., rigid connection with BS, rigid connection, and hinged
connection. In the static response, the main difference of the curvature
distribution appears in the top-end region, shown in Figure 11.22a. In
the rigid connection case, the curvature magnitude drops rapidly from
−0.0137 to −0.0040 rad/m in the top-end region. Meanwhile, in the
rigid connection case with bend stiffener, the curvature magnitude
then drops slowly from −0.0064 to −0.0062 rad/m in the top-end region.
Comparing the results from the two cases, it can fully verify the positive
effect of the bend stiffener that can provide a gradual physical transition
between the platform and the flexible pipe to avoid over-bending. In
the other regions, the curvature magnitudes with the three different top
connection conditions are nearly equal, especially for the TDZ. Moreover,
as illustrated in Figure 11.22b, the time-history curves of the maximum
curvature in the TDZ almost coincide with the three top connection con-
ditions. It is visible that the BS constraint has scarce effect on the static/
dynamic curvature response in the TDZ. Figure 11.22c exhibits the time
history responses of the top effective tension with three top connection
conditions, and the corresponding results also show little difference. As
0.020 0.030
0.02

0.015
0.01 0.027
0.010
0.00 0.024
Curvature (rad/m)
0.005 500 550 600 650 700

0.021
0.000
0.000
0.018
–0.005 –0.005
–0.010
–0.010 0.015
–0.015
0 2 4 6 8 10
–0.015 0.012
0 200 400 600 800 1000 0 8 16 24 32 40 48 56 64
Riser arc length (m) Time (s)
(a) Static curvature distribution (b) Maximum curvature variation in TDZ

950 0.000
Curvature at BS’s root end (rad/m)
900 –0.003
Top effective tension (kN)
850
–0.006
800
–0.009
750
–0.012
700
650 –0.015
600 –0.018
0 8 16 24 32 40 48 56 64 0 8 16 24 32 40 48 56 64
Time (s) Time (s)
(c) Top effective tension variation (d) Curvature variation at BS’s root end
Curvature at 22m arc length from top (rad/m)
0.000 0.00035
Curvature at BS’s tip end (rad/m)
–0.001 0.00030
–0.002 0.00025
–0.003 0.00020
–0.004
0.00015
-0.005
0.00010
–0.006
0.00005
–0.007
0.00000
–0.008
0 8 16 24 32 40 48 56 64 –0.00005 0 8 16 24 32 40 48 56 64
Time (s) Time (s)
(e) Curvature variation at BS’s tip end (f) Curvature at 22m arc length point from
top end
Rigid connection+BS Rigid connection Hinged connection
Figure 11.22 Static/dynamic response with different top connection conditions.

the effect of the BS constraint that influences the riser’s effective tension is
mainly limited within the top-end region, it can be assumed that the top
connection condition may have little impact on the riser’s effective tension.
Figures 11.22d and 11.22e show the time history responses of the curva-
ture at the BS’s root end and tip end, respectively. In the hinged connection
case, the curvature magnitudes at the BS’s root end and tip end slightly
fluctuate in zero curvature, which differ greatly from the other two cases.
Since the top hinged connection would lead to the curvature distortion
in the top-end region, it is not feasible to simplify the top connection as
hinged when conducting the flexible riser’s global analysis. Additionally, in
the rigid connection case, the curvature magnitude and amplitude at the
BS’s root end are much larger than these in the rigid connection with BS
case, as shown in Figure 11.22d. However, in the rigid connection case,
the curvature magnitude and amplitude at the BS’s tip end are a little
less than those in the rigid connection with BS case (see Figure 11.22e).
It can be observed that the BS not only can optimize the curvature distri-
bution but also can optimize the curvature amplitude. Furthermore, the
curvature variation curves at 22-m arc length point measured from the
top end with the three top connection conditions (shown in Figure 11.22f)
are almost coincided and, consequently, the top connection condition has
hardly any influence on the curvature behavior below a certain arc length
measured from the top end. The effective influencing zone might depend
on the effective tension, bending stiffness, and BS’s length.
11.5.4 Effect of Bending Hysteretic Behavior

It is well known that the bending behavior of an unbonded flexible pipe
is hysteretic, which indicates that the bending moment depends on the
history of curvature path applied as well as the current curvature. To
study the influence of the bending behavior in the global static/dynamic
analysis, four bending stiffness cases (hysteretic stiffness, nonlinear stiff-
ness, stick stiffness EIb, and slip stiffness EIf) are conducted and studied.
With different bending behaviors, the primary difference of the static cur-
vature distribution may occur in the two regions, i.e., top-end region
and TDZ, as shown in Figure 11.23a. In the top-end region, the curvature
magnitude in the slip stiffness case is much larger than those in the three
other cases. Though with the protection of the BS constraint, the curvature
at the bend stiffener’s tip end reaches up to −0.0102 rad/m, which is much
larger than the value −0.0067 rad/m at the BS’s root end. This shows that
the curvature distribution in the top-end region is not only related to the
BS, but also related to the flexible riser’s bending stiffness. Therefore, the
0.020 0.032
0.02

0.015
0.01 0.028
0.010
0.00
Curvature (rad/m)
0.024
0.005 500 550 600 650 700
0.020
0.000
0.000
0.016
–0.005 –0.005
–0.010 0.012
–0.010
0 2 4 6 8 10
0 200 400 600 800 1000 0 8 16 24 32 40 48 56 64
Riser arc length (m) Time (s)

950 –0.0045
Curvature at BS's root end (rad/m)
900 –0.0050
850 –0.0055
–0.0060
800
–0.0065
750
–0.0070
700
–0.0075
650 –0.0080
600 –0.0085
0 8 16 24 32 40 48 56 64 0 8 16 24 32 40 48 56 64
Time (s) Time (s)
–0.004
–0.006
Curvature at BS's tip end (rad/m)
–0.008
–0.010
–0.012
–0.014
0 8 16 24 32 40 48 56 64
Time (s) Hysteretic stiffness Nonlinear stiffness
(e) Curvature variation at BS’s tip end Stick stiffness Slip stiffness
Figure 11.23 Static/dynamic response with different bending behaviors.
design of the BS should take the mechanical characteristic of the attached

flexible riser into account to make sure of the riser’s integrity. Similarly, the
curvature in the TDZ in the slip stiffness case is slightly larger than those
in the other cases.
The time-history curves of the maximum curvatures in the TDZ are

illustrated in Figure 11.23b with the four bending behaviors. Due to the
minimum bending stiffness in the four cases, the maximum curvature
in the TDZ in the slip stiffness case is the largest, reaching up to 0.0310
rad/m. Nonetheless, the curvature variation in the hysteretic stiffness
case is the most realistic. The result in the nonlinear stiffness case is the
closest to that in the hysteretic stiffness case, which signifies that the non-
linear non-hysteretic bending model may be adopted to study the global
response if necessary. From this figure, it can also be found that in the
TDZ the impact of the bending behavior on the curvature performance is
of great significance and should not be over looked.
As shown in Figure 11.23c, the top effective tensions change little in
the four bending behavior cases and the corresponding curves are almost
identical. Therefore, the bending behavior also has little impact on the ris-
er’s effective tension, which is almost identical to the effect of the top con-
nection condition.
Figures 11.23d and 11.23e show the time history responses of the curva-
tures at the BS’s root end and tip end, respectively. From the figures, the curva-
ture in the slip stiffness case seems more severe than those in the other cases.
Since the curvature in the top region is always within the stick mode, the cur-
vature responses in the other cases agree well with each other. The effect of the
bending hysteretic behavior will be obvious once the curvature enters into the
slip mode, such as the maximum curvature variation in the TDZ; however,
this effect can be negligible if the curvature is always within the stick mode.
11.5.5 Effect of Top Angle Constraint

The top rigid angle plays a vital role in the mechanical behavior of the
flexible riser in the top-end region. An unreasonable top angle may induce
excessive bending stress due to large curvature in the top-end region, which
is serious for the integrity and security of the flexible riser. The effect of the
top angle herein is well studied in three different top angle cases, namely, 5°,
6°, and 7°. As illustrated in Figure 11.24a, the top angle constraint has little
influence on the static curvature distribution in the TDZ. However, it has sig-
nificant effect on the curvature distribution in the top-end region. In the top
hinged connection case, the top angle is equal to 5.3°. Displayed in Figure
11.24a, the top rigid angle deviates further from the top hinged angle, the static
curvature magnitude will aggravate more obviously. Consequently, the static
curvature distribution in the top-end region is very sensitive to the top angle.
Figure 11.24b shows the time history of the dynamic maximum cur-
vatures in the TDZ in the three top angle cases and the results are nearly
equal. Additionally, the top effective tension curves in the three top angle
cases (see Figure 11.24c) almost coincide. It can be seen that the effects
of the top angle on the maximum curvature in the TDZ and top effective
tension are similar with those of the BS.
0.020 0.028
0.02

0.015 0.026
0.01
0.010 0.024
0.00
Curvature (rad/m)
0.005 0.022
500 550 600 650 700
0.020
0.000
0.000
0.018
–0.005 -0.005
0.016
–0.010 -0.010
-0.015 0.014
–0.015 0 2 4 6 8 10
0.012
0 200 400 600 800 1000 0 8 16 24 32 40 48 56 64
Unbonded flexible pipe arc length (m) Time (s)
950 0.000
Curvature at BS's root end (rad/m)
900 −0.002
850 −0.004
800 −0.006
750 −0.008
700 −0.010
650 −0.012
600 −0.014
0 8 16 24 32 40 48 56 64 0 8 16 24 32 40 48 56 64
Time (s) Time (s)
0.000
Curvature at 22m arc length from top (rad/m)
0.00035
–0.002 0.00030
Curvature at BS’s tip end (rad/m)
–0.004 0.00025
–0.006 0.00020
–0.008 0.00015
–0.010 0.00010
–0.012 0.00005
0.00000
–0.014
0 8 16 24 32 40 48 56 64 0 8 16 24 32 40 48 56 64
Time (s) Time (s)
(e) Curvature variation at BS’s tip end (f) Curvature at 22m arc length point
from top end
Top angle 7° Top angle 6° Top angle 5°
Figure 11.24 Static/dynamic response with different top angles.

However, in the top-end region, the top angle has a significant influ-
ence on the curvature variation (see Figures 11.24d and 11.24e). As the
top angle 5° is the closest angle to the top hinged angle in the three cases,
the dynamic curvature magnitude in top angle 5° case is controlled within
a small value in the top-end region. The setting of top angle 5° is
therefore more reasonable than the two other top angles in this case study.
Furthermore, from Figures 11.24d and 11.24e, it seems that the curvature
variation at the BS’s root end or tip end is almost the same for all three
cases, and the main difference is found in the average curvature value.
Furthermore, the corresponding curvature amplitudes seem nearly equal
for the three cases. In short, the top angle can alter the curvature distri-
bution in the top-end region. The time history responses of the curvatures
at 22-m arc length point measured from the top end with different top
angles are illustrated in Figure 11.24f and the results show little difference.
11.6 Conclusions
This study develops a flexible riser global dynamic analysis model which
takes the BS constraint and flexible pipe bending hysteretic behavior into
account. The model can accurately predict the static/dynamic response of
the flexible riser, especially for the top-end region and TDZ. In this chap-
ter, a case study with the top heave harmonic excitation is conducted and
the static/dynamic response characteristics in the top-end region and TDZ
are studied. The effects of BS constraint, bending hysteretic behavior, and
top angle constraint are also investigated and conclusions are summarized
as follows:
(1) The effective tension in the TDZ obtained from the math-
ematical model seems a little less than that derived from
FEM by OrcaFlex, which may be primarily induced by the
neglect of the item EIeκ2 in calculating the effective tension.
(2) The amplitude effect of the top heave harmonic excitation
on the TDP effective tension is greater than that on the top
effective tension.
(3) The relation between the effective tension and effective
bending stiffness of the maximum curvature point in the
TDZ shows an approximately closed loop. When the effec-
tive tension increases, the effective bending stiffness then
decreases and vice versa.
(4) BS constraint, bending hysteretic behavior, and top angle

constraint all have a small influence on the effective tension
of a flexible riser subjected to hydrodynamic loadings and
platform motion.
(5) The BS not only can optimize the curvature distribution
but can also optimize the curvature amplitude. However,
the top angle can alter the curvature distribution in the
top-end region and have slight impact on the curvature
amplitude.
(6) The curvature response in the top-end region is very sensi-
tive to the top angle. The more the top rigid angle deviates
from the top hinged angle, the more obviously the curva-
ture magnitude will be aggravated.
(7) The effect of the bending hysteretic behavior will be obvi-
ous once the curvature enters into the slip mode, such as
the maximum curvature variation in the TDZ; however, the
effect can be negligible if the curvature is always within the
stick mode.
(8) Both the bend stiffener constraint and top angle have little
effect on the static/dynamic curvature performance in the
TDZ.
References
1. Fergestad D, Løtveit S A. Handbook on Design and Operation of Flexible
Pipes[J]. 2014.
2. Owen, D. G., Qin, K., 1986. Model Tests and Analysis of Flexible Riser
Systems Proceedings of the 5th International Conference on Offshore
Mechanics and Arctic Engineering, Tokyo 1986.
3. Vogel, H., Natvig, B. J., 1987. Dynamics of flexible hose riser systems.
Journal of Offshore Mechanics and Arctic Engineering, 109, No. 3, 244–248.
4. Seyed, F. B., Patel, M. H., Holland, A. P., 1987. Dynamic Analysis
of Flexible Catenary Risers Including Internal Flow Effects Dynamics of
Compliant Structures, Final Report, Project 12, Vol. III 1987.
5. McNamara J F, O’Brien P J, Gilroy S G. Nonlinear analysis of flexible ris-
ers using hybrid finite elements[J]. Journal of offshore mechanics and arctic
engineering, 1988, 110(3): 197–204.
6. Fylling, I., Larsen, C. M., Sodahl, N., Passano, E., Bech, A., Engseth, A. G., Lie, E. ,
Ormberg, H., 1998. Riflex user’s manual. Marintek Report, Trondheim, Norway.
7. Chai Y T, Varyani K S. An absolute coordinate formulation for three-
dimensional flexible pipe analysis[J]. Ocean Engineering, 2006, 33(1): 23–58.
8. Nordgren R P. On computation of the motion of elastic rods[J]. Journal

of Applied Mechanics, 1974, 41(3): 777–780.
9. Garrett D L. Dynamic analysis of slender rods[J]. J. Energy Resour. Technol.;
10. Paulling, J. R., Webster, W. C., 1986. Large-amplitude analysis of the coupled
response of a TLP and tendon system, Proceedings of OMAE’86, Tokyo, 86 1986.
11. Chen H, Xu S, Guo H. Nonlinear analysis of flexible and steel catenary ris-
ers with internal flow and seabed interaction effects[J]. Journal of Marine
Science and Application, 2011, 10(2): 156–162.
12. Bai X, Huang W, Vaz M A, et al. Riser-soil interaction model effects on the
dynamic behavior of a steel catenary riser[J]. Marine Structures, 2015, 41:
53–76.
13. Chen X, Zhang J, Johnson P, et al. Dynamic analysis of mooring lines with
inserted springs[J]. Applied Ocean Research, 2001, 23(5): 277–284.
14. Chen X. Studies on dynamic interaction between deep-water floating struc-
tures and their mooring/tendon systems[M]. 2002.
15. Chen X, Ding Y, Zhang J, et al. Coupled dynamic analysis of a mini TLP:
Comparison with measurements[J]. Ocean Engineering, 2006, 33(1): 93–117.
16. Tahar A, Kim M H. Hull/mooring/riser coupled dynamic analysis and sensi-
tivity study of a tanker-based FPSO[J]. Applied Ocean Research, 2003, 25(6):
367–382.
17. Tahar A, Kim M H. Coupled-dynamic analysis of floating structures with
polyester mooring lines[J]. Ocean engineering, 2008, 35(17): 1676–1685.
18. Caire M, Vaz M A. The effect of flexible pipe non-linear bending stiffness
behavior on bend stiffener analysis[C]//ASME 2007 26th International
Conference on Offshore Mechanics and Arctic Engineering. American
Society of Mechanical Engineers, 2007: 103–109.
19. Rocha G, Parrilha R, Oliveira A M. Influence of Nonlinear Mechanical
Contact on Hang-Off Location in Flexible Risers’Structural Response[C]//
ASME 2010 29th International Conference on Ocean, Offshore and Arctic
Engineering. American Society of Mechanical Engineers, 2010: 487–494.
20. Vaz M A, de Lemos C A D. Geometrical and material non-linear formu-
lation for bend stiffeners[C]//ASME 2004 23rd International Conference
on Offshore Mechanics and Arctic Engineering. American Society of
Mechanical Engineers, 2004: 201–206.
21. Smith R. Bending stiffeners for extreme and fatigue loading of unbonded
flexible risers[C]//ASME 2008 27th International Conference on Offshore
Mechanics and Arctic Engineering. American Society of Mechanical
Engineers, 2008: 405–413.
22. Tanaka R L, da Silveira L M Y, Novaes J P Z, et al. Bending stiffener design
through structural optimization[C]//ASME 2009 28th International
Conference on Ocean, Offshore and Arctic Engineering. American Society
of Mechanical Engineers, 2009: 411–418.
23. Sodahl N, Ottesen T. Bend stiffener design for umbilicals[C]//ASME 2011

30th International Conference on Ocean, Offshore and Arctic Engineering.
American Society of Mechanical Engineers, 2011: 449–460.
24. Drobyshevski Y. Investigation into non-linear bending of elastic bars with
application to design of bend stiffeners[J]. Marine Structures, 2013, 31:
102–130.
25. Hou Y, Yuan J, Zhang Y, et al. Direct Hang-Off Model to Evaluate Fatigue
Damage at Riser Hang-Off[C]//ASME 2013 32nd International Conference
on Ocean, Offshore and Arctic Engineering. American Society of Mechanical
Engineers, 2013: V04BT04A004-V04BT04A004.
26. Witz J A. A case study in the cross-section analysis of flexible risers[J].
Marine Structures, 1996, 9(9): 885–904.
27. Leroy J M, Estrier P. Calculation of stresses and slips in helical layers of
dynamically bent flexible pipes[J]. Oil&Gas Science and Technology,
2001, 56(6): 545–554.
28. Zhang J, Tan Z, Sheldrake T. Effective Bending Stiffness of an Unbonded Flexible
Riser[C]//ASME 2008 27th International Conference on Offshore Mechanics
and Arctic Engineering. American Society of Mechanical Engineers, 2008:
117–123.
29. Tan Z, Quiggin P, Sheldrake T. Time domain simulation of the 3D bend-
ing hysteresis behavior of an unbonded flexible riser[J]. Journal of Offshore
Mechanics and Arctic Engineering, 2009, 131(3): 031301.
30. Vargas-Londoño T, de Sousa J R M, Magluta C, et al. A Theoretical and
Experimental Analysis of the Bending Behavior of Unbonded Flexible
Pipes[C]//ASME 2014 33rd International Conference on Ocean, Offshore
V06BT04A027-V06BT04A027.
31. Péronne S, Izarn C, Estrier P, et al. Flexible Pipe Hysteretic Bending Behavior:
Comparison With Experimental Characterization and Finite Element
Method[C]//ASME 2015 34th International Conference on Ocean, Offshore
V05AT04A032-V05AT04A032.
32. You J, Biscontin G, Aubeny C P. Seafloor interaction with steel catenary
risers[C]//The Eighteenth International Offshore and Polar Engineering
Conference. International Society of Offshore and Polar Engineers, 2008.
33. Theti R. Soil interaction effects on simple-catenary riser response[J].
Pipes&pipelines international, 2001, 46(3): 15–24.
34. Katifeoglou S A, Chatjigeorgiou I K. Dynamic interaction of catenary
risers with the seafloor[J]. Applied Ocean Research, 2012, 38: 1–15.
35. Elosta H, Huang S, Incecik A. Dynamic response of steel catenary riser using a
seabed interaction under random loads[J]. Ocean Engineering, 2013, 69: 34–43.
36. Sparks C P. Fundamentals of marine riser mechanics: basic principles and
simplified analyses[M]. PennWell Books, 2007.
12
Dynamic Analysis of the Cable-Body
of the Deep Underwater Towed System
12.1 Introduction
Underwater towed system is a kind of underwater detecting device which
is widely used in ocean monitoring, ocean research and military affairs.
It is usually composed of marine exploration ship, guided towline, towed
cable, towed underwater vehicle. The actual marine environment disturbs
the normal operation of underwater towed systems at all times. Moreover,
there is a complex interaction between towing ships, towed cables and
towed bodies. Therefore, it is very important to obtain the motion law and
get the characteristics of the towed system during the process of moving,
more and more scholars and researchers are working on this research.
Keqiang et al. [1] established the marine cable-body system model and
simulated the motion of the system in time domain. Koh and Rong [2]
used finite difference method to analyze the cable hydrodynamic model,
in their research, the cable position at each time was taken as a variable
to approximate the cable governing equation. The model of Ablow and
Schechter [3] was based on a fully three-dimensional code and the finite
difference method has been used to solve the numerical problem. It can be
used to predict the motion of the towed cable system under most work-
ing conditions, which has been widely used in the towed cable systems. A
single node finite element method to calculate the dynamic model of the
towed cable was used by Sun [4], in which the displacement of the node
was replaced by the position of each node, which greatly simplifies the cal-
culation process of the finite element method.
At present, the underwater vehicle 6-DOF equations were firstly pro-
posed by Gertler and Hargen and then improved by Abkowitz which is
used for the simulation of the hydrodynamic characteristics of the under-
water body at the cable tail [5, 6]. When the mass difference between sur-
face ship and towed underwater vehicle is not large, it is considered that
247
the influence of towed body on towing ship cannot be neglected, Buckham

[7] studied the coupled motion response between the towing ship and
underwater towed body in this state, the effect of the towing cable on the
ship motion response was neglected in his research.
In this chapter, a three-dimensional hydrodynamic model is used to
simulate the motion of an underwater towed system under the influence
of ocean currents by OrcaFlex, in which the tension of the towed cable and
the depth of the towed body have been calculated. The dynamic character-
istics of the towed system during the 360° rotation of the towing ship have
been obtained. By analyzing the influence of towed parameters on towing
motion and towing tension, some reliable conclusions have been given,
which can be helpful for practical engineering design.
12.2 Establishment of Towed System Dynamic Model

The whole system is shown in Figure 12.1, consisting of underwater ship,
towed cable and towed body. Since the ship mass is much larger than the
towed cable and towed body, it is assumed that the motion of the ship is not
affected by the cable tension.
In order to simplify the model, the hydrodynamic forces acting on the
depicted wings of the towed body in Figure 12.1 was not taken into account
in this chapter.
In this chapter, the cable model is considered as a slender and flexible
cylindrical cable. Discrete lumped mass model is used to solve the nonlinear
boundary value problem. The basic idea of the model is to divide the towed
cable into N segments, and the mass of each element is concentrated on one
node, so that there can be N + 1 nodes. The tension T and shear force V acting
at the end of each segment can be considered as a concentrated action on one
node and any external hydrodynamic load is considered to be concentrated on
one node. The motion equation of the i-th node is
M Ai Ri = Tei − Tei −1 + FdIi + Vi − Vi−1 + wi ∆ si (12.1)
where
R represents the node position of the cable;
 π  π
M Ai = ∆ si  mi + Di2 (Can − 1) I − ∆ si Di2 (Can − 1)(τ i ⊗ τ i−1 )
 4  4
Cable-Body of the Deep Underwater Towed System 249
Figure 12.1 The constitution of the towed system.
represents the mass matrix of the node, I is 3 × 3 unit matrix;
∆s0i
Tei = EAε i = EA
∆sεi ,
is the effective tension at the node; Δs0i = L0/(N − 1) is the original length
of each cable,
Δsεi = |Ri+1 − Ri|
is the stretched length of the cable, EA is the axial stiffness of the

towed cable; FdIi represents the external hydrodynamic force that each
node is subjected, which can be calculated according to the Morrison
equation:
1 π
FdIi = ρsw Di 1 + ε i ∆ si (Cdni |vni | vni + π Cdti |vti| vti ) + Di2 ρswCani ∆ si (awi − (awi ⋅ τ i ))τ i
2 4
where
ρsw is the sea density, Di is the diameter of the towed cable, Cdni is the
normal drag coefficient, Cdti is the tangential drag coefficient, Cani is the
normal inertia coefficient;
EIi+1τ i × (τ i × τ i+1 ) EIiτ i × (τ i−1 × τ i ) Hi+1τ i × τ i+1

Vi = − +
∆sεi ∆sεi+1 ∆sε2i ∆sεi
Vi is the shear force on the i-th node, wi is the weight of the unit length
of the marine cable, and H is the torque on the cable.
During the course of the towed system, the towed cable and the towed
body interacts with each other. The towed cable provides tension to drag
the towed body forward, and in turn the drag of the towed body also affects
the motion of the towed cable, especially the original configuration and
tension distribution of the towed cable. In this chapter, the towed body is
regarded as a six-degree-of-freedom rigid body with a certain mass and
mass distribution. Its center of gravity coordinates (CG) in the local coor-
dinate system is expressed as (xG, yG, zG), (ϕ, θ, ψ) represent the Euler
angles of the local coordinate system (x, y, z) with respective to the inertial
coordinate system-roll angle, pitch angle, and yaw angle. The transforma-
tion matrix T between the inertial coordinate system and the local coordi-
nate system can be expressed as
 cosψ cosθ cosψ sinθ sin φ − sinψ cos φ cosψ sinθ cos φ + sinψ sin φ 
 
T =  sinψ cosθ sinψ sinθ sin φ + cosψ cos φ sinψ sinθ cos φ − cosψ sin φ 
 − sinθ cosθ sin φ cosθ cos φ 
 
(12.2)
According to the analysis of the drag force and its plane motion equa-
tion, the six-degree-of-freedom motion equation of the towed body is
written as
X = m[(u + qw − rv ) − xG (q 2 + r 2 ) + yG ( pq − r ) + z G ( pr + q )]
Y = m[(v + ru − pw ) − yG ( p 2 + r 2 ) + z G (qr − p ) + xG (qp + r )]
Z = m[(w + pv − qu) − z G ( p 2 + q 2 ) + xG (rp − q ) + yG (rq + p )]
K = I x p + ( I x − I y )qr − I xz (r + pq) + I yz (r 2 − q 2 ) + I xy ( pr − q ) +
m[ yG (w + pv − qu) − z G (v + ru − pw )]
M = I y q + ( I x − I z )rp − I xy ( p + qr ) + I xz ( p 2 − r 2 ) + I yz (qp − r ) +
m[z G (u + qw − rv ) − xG (w + vp − qu)]
N = I z r + ( I y − I x ) pq − I yz (q + rp) + I xy (q 2 − p 2 ) + I xz (rp − p ) +
m[xG (v + ru − pw ) − yG (u + qw − vr )]
(12.3)
In the above formula: u, v, w, p, q, and r correspond to longitudinal

velocity, lateral velocity, heave velocity, roll angular velocity, pitch angular
velocity, and yaw angular velocity, respectively; the right side of the equa-
tion is the inertia force and torque on the towed body; m is the mass of
the towed body; Ix, Iy, and Iz are mass moment of inertia for the towed
body; Ixy, Ixz, and Iyz are the cross moments of inertia of the towed body
respectively; xG, yG, and zG are the coordinates of the center of gravity in
the local coordinate system of the towed body, respectively; the left side
term F = (X, Y, Z) and M = (K, M, N) of the equation are the external forces
and external moments acting on the towed body, these external forces and
external moments include the gravity and buoyancy, the hydrodynamic
force, the tension of the towed cable, and the forces and corresponding
moments generated by the disturbance of the marine environment.
In the towed body motion governing equation, the angular velocity
transformation relation between two coordinate systems can be repre-
sented as follows:
 p  
1 0 − sinθ   φ 
    
 q  = 0 cosφ cosθ sin φ   θ  (12.4)
 r   0 − sin φ cosθ cosφ   ψ 

In the towed system, the boundary condition of the towed cable can be
divided into two parts. The first part is the first boundary condition of the
towed cable, that is, the velocity at the upper end of the cable is the same
as the speed of the towing ship; the other part is the bottom end boundary
condition, that is, the bottom end velocity of the towed cable is the same
as that of the towed body, and the initial value of the bottom end tension is
not zero. The initial condition of this chapter is that the ship travels straight
at a certain speed, so the velocity components at the upper end of the towed
cable vx, vy, and vz can be given.
12.3 Numerical Simulation and Analysis

of Calculation Results
Dynamic analysis of the towed system during the process of 360° rotary
motion Rotary motion is a common form of motion in towed systems.
Accurate prediction for the depth and stability of the towed body is of
great significance. When the towing ship rotates, different cable length and
speed have influence on the rotating radius of towed body. The rotating
radius of a towed body can be expressed as:
L ln(1 − U 0 /U )
R0 = − (12.5)
0.52
where
R0, the rotating radius of the towed body; L, the length of the towed
body; U, the direct speed of the towed body before turning; and U0, the
turning speed of the towed body.
Tables 12.1 and 12.2 show the physical parameters of the towed cable
and the towed body: L1 and L2 are the lengths of the towed cable and the
towed body, respectively; D and d are the diameters of the cable and the
towed body, respectively; m is the weight of the unit length of the marine
cable, m0 is the mass of the towed body; Cn and Ct are the normal drag
coefficient and the tangential drag coefficient, respectively; Ax and Az are
the towing areas in the horizontal and vertical directions, respectively; EI
is the bending stiffness of the towed cable; and μ is Poisson ratio. Torque is
ignored in calculation.
12.3.1 The Effect of Different Turning Radius

The whole system first takes a direct route with the speed V = 2.5 m/s for
400s, and then it carries on the 360° turning operation, the turning time
is 800 s, the turning radius R is taken as 80, 95, 120, 160, 179, and 286 m,
respectively (these radii are not from reference [8]). The rotation plan is
The simulation results are in good agreement with reference [8]: during
the whole process, there is a turning critical radius, the towed body motion
always lags behind the towing ship motion; in the course of turning, the
trajectory of the towed body is located inside the trajectory of the tow-
ing ship, the turning radius of the towed body and the time required for
steady rotation vary with the change of the turning radius of the towing
ship. Here, I have to point out that as reference [8] does not use the same
cable and towed body specifications, it is impossible to make some com-
parisons between the model in this chapter and in reference [8] see Figure
12.3. But, the conclusions between the reference [8] and Figure 12.2 are
the same. Here, I made some simulations which have the same parameters
Table 12.1 Physical parameters of the towed cable.

m0/kg L1/m d/m Cn Ct EI μ
1.1 450 0.025 1.2 0.008 0 0.5
Table 12.2 Physical parameters of the towed body.

m/kg L2/m D/m Cd Ax/m2 Az/m2
2,500 3.9 1 0.5 1 0.78 2.7
as in reference [8]. In order to reduce computation time and make the two
systems have the same radius of rotation, the towed speed of the towing
ship in my model is given a higher speed and angular velocity. As a result,
we can see that the time for the system to reach stability is different from
that in reference [8]. But, when the system is in the stable state (in refer-
ence [8], it is 2–7 h; in my simulation model, it is 500–2,000 s), the curves
of my simulation model in OrcaFlex (in Figure 12.4) are basically consis-
tent with the curves in reference [8]. This shows that the dynamic model
of reference [8] can be reproduced in OrcaFlex under certain assumptions
see Figure 12.5.
12.3.2 The Effect of Different Turning Speeds

The towing ship carries out at speeds of 2 and 3 m/s for 360° rotary motion,
with a turning radius of R = 230 m, the turn distance is 1,445 m, giving
turning time 723 and 482 s at 2 and 3 m/s speed, respectively. The turning
process is shown in Figures 12.6 to 12.8.
From Figures 12.6 to 12.8, it can be concluded that if the radius of
rotation of the towing ship remains unchanged, the towing speed has a
significant influence on the turning radius of the towed body during the
360° turning operation, and the smaller the towing speed, the larger the
turning radius of the towed body will be; the depth of the towed body
increases during the rotation and the tension of the towed end decreases
as the towed depth increases, the simulation results are in agreement with
reference [8].
200
250
Y(m)
R=80m ship
Y(m)
R/L=0.17 body R=95m ship
R/L=0.21 body
150 200
150
100
100
50 50
0
0
–50
–150 –100 –50 0 50 100 150 –200 –100 0 100 200
X(m) X(m)
(a) R=80m (b) R=95m
300 R=120m ship 400

Y(m)
Y(m)
R/L=0.26 body R=160m ship
R/L=0.35 body
200 300
200
100
100
0
0
–200 –100 0 100 200 –300 –200–100 0 100 200 300
X(m) X(m)
(c) R=120m (d) R=160m
600
Y(m)
500 R=286m ship

Y(m)
R=179m ship R/L=0.64 body

R/L=0.40 body
400
400
300
200 200
100
0 0
–100
–400 –200 0 200 400 –400 –200 0 200 400
X(m) X(m)
(e) R=179m (f) R=286m
Figure 12.2 Plan view of circle maneuver in towed system.
12.3.3 Dynamic Analysis of the Towed System

with the Change of the Parameters of the Cable
The steady towed depth and cable tension are two crucial performance
parameters in the design process, which depend on a number of fac-
tors, such as the towing speeds, drag coefficients, cable density, and cable
elasticity. Here, the motion of the towed system is studied by numerical
1000
R=
3000 m
1500 2600 m
2200 m
2000
Vehicle Depth (m)
1800 m
2500
3000 1400 m
3500 100 m
600 m
4000
0 1 2 3 4 5 6 7 8
Time (hours)
Figure 12.3 Curves of vehicle depth in reference [8].
R=600m
R=1000m
R=1400m
R=1800m
R=2200m
0 R=2600m
R=3000m
1000
Vehicle Depth/m
2000
3000
4000
0 500 1000 1500 2000

Time/s
Figure 12.4 Curves of vehicle depth of my simulation model in OrcaFlex.
simulation and the influence of various parameters is analyzed. In this sec-

tion, another simulation will be run with new properties.
The length of the towed cable is 300 m, its normal drag coefficient is
Cn = 1.44 and its tangential drag coefficient is Ct = 0.015, the mass of the
towed body is 1,000 kg. The whole system keeps straight with the speed
of V = 2.5 m/s for 1,000 s. Then, we keep the towing speed unchanged
and change the parameters of towed cable to study the influence of various
parameters on the motion of the towed cable.
Figure 12.5 The configuration of the towed cable of reference [8] in OrcaFlex.
600 ship
Y(m)
body 600 ship

body
400
400
200 200
0 0
–400 –400 0 200 400 –400 –200 0 200 400

X(m) X(m)
(a) V=2m/s (b) V=3m/s
Figure 12.6 Plan view of circle maneuver.
depth 200
Z(m)
Z(m)
360 depth
250
380
300
400
350
420 400
440 450
0 300 600 900 1200 1500 0 300 600 900 1200
T(s) T(s)
(a) V=2m/s (b) V=3m/s
Figure 12.7 Depth variation of the towed body.

F(KN)
22
30
F(KN)
20 27
24
18 21
18
16
15
200 400 600 800 1000 1200 1400 0 200 400 600 800 1000 1200
T(s) T(s)
(a) V=2m/s (b) V=3m/s
Figure 12.8 Tension variation of the towed end.
12.3.4 The Effect of the Diameters of the Towed Cable

The mass of the towed cable is constant, and the diameter d is taken as
0.025, 0.028, 0.030, and 0.035 m, respectively. Figures 12.8 and 12.9 show
the effect of the change of the diameter of the towed cable on the towed
depth and the maximum tension at the upper and bottom ends.
As can be seen from Figure 12.9 and 12.10: as the diameter of the cable
increases, the stable towed depth of the towed body decreases; the larger
the diameter of the towed cable, the smaller the maximum tension at its
upper end, this is because the increase in the diameter of the cable causes
the buoyancy increase and the upper end tension decrease. However, the
maximum tension of the bottom end which is connected with the towed
body show less change with the increase of the diameter of the towed cable.
As is known to all, if there is no towed body at the bottom end (it means
there is no towed body in the system), the tension at the bottom end is zero
and independent of the change of the diameter of the towed cable.
12.3.5 The Effect of the Drag Coefficients of the Towed Cable

The towing drag force consists of two parts, the tangential and the normal
drag force, which determine the tension and the towed depth of the towed
body to a great extent. The drag coefficient of the towed cable is affected
by its surface roughness, Reynolds number, Froude number, and current
velocity. The analysis of the influence of the drag coefficient on the drag
50
100
Z(m)
150
d=0.025m
200 d=0.028m
d=0.030m
250 d=0.035m
300
0 200 400 600 800 1000 1200

T(s)
Figure 12.9 Depth variation of the towed body with the change of the diameter of the
cable.
8.0
Fmax(KN)
d=0.025m
d=0.028m
7.5 d=0.030m
d=0.035m
7.0
6.5
6.0
5.5
5.0
0 50 100 150 200 250 300
L(m)
Figure 12.10 Maximum tension variation of the towed body with the change of the
diameter of the cable.
system is of great importance in the design process. The magnitude of the

normal and tangential drag coefficient values for smooth, rigid cylinders
that are only subjected to purely normal or pure axial flow has been univer-
sally confirmed; however, the above coefficients are not consistent at any
angle between the water flow and the towed cable, which is mainly because
of the changes of Reynolds number. The Reynolds number can be defined
by the diameter or length, expressed as follows:
Ud
Red = ≈ Ud × 106 ,
ν
Red is the Reynolds number which is related to the diameter;
Ul
Rel = ≈ Ul × 106 ,
ν
Rel is the Reynolds number which is related to the length. For the towed
cable normal resistance, the viscous pressure resistance is the main com-
ponent. When the Reynolds number range is 100 < Red < 105, the normal
drag coefficient is Cn ≈ 1.2; in the Reynolds number 105 < Red < 106, the
normal drag coefficient will decrease rapidly, which may reach about 0.3.
But, in fact, as the cable swings in the water, it is difficult to determine
the drag coefficient of the cable in the water, and due to variation in angle
between water flow and cable tangent, the normal drag coefficient may
reach 1.4 to 1.7 or even higher.
The tangential drag force of the cable is closely related to the Reynolds
number, the angle of attack, the surface roughness and the vibration of
the cable. Approximation functions for Ct can be found in the relevant
research [9, 10]. The tangential drag coefficient is difficult to determine. In
general, the tangential drag coefficient differs from the normal drag coeffi-
cient by two orders of magnitude. The relationship can be written as: Ct =
γCn, γ is a constant, generally, 0.01 ≤ γ ≤ 0.03.
In this section, the influence of the tangential and normal drag coef-
ficients on the steady-state motion of the towed system is analyzed. The
following four values are calculated, respectively, and the rest of the param-
eters remain unchanged.
Ct = 0.01, 0.015, 0.020, 0.025
Cn = 1.2, 1.44, 1.53, 1.84
The results are shown in Figures 12.11 to 12.14.

The results of the simulation are in agreement with reference [11]: when
the normal drag coefficient is constant, the tangential drag coefficient
Ct has a great effect on the maximum tension of the towed cable. As Ct
increases, the maximum tension of the towed cable increases. When the
tangential drag coefficient is constant, the normal drag coefficient Cn has a
great influence on the stability of the towed body, the larger the Cn is, the
smaller the towed depth is, but the maximum tension of the towed cable
shows less change with the increase of Cn. It can be seen that when the
F(KN)
Ct=0.01
Ct=0.015
7
Ct=0.02
Ct=0.025
6
3
0 200 400 600 800 1000 1200
T(s)
Figure 12.11 Maximum tension variation of Ct (Cn = 1.44).
8
F(FN)
Cn=1.2
7 Cn=1.44
Cn=1.53
Cn=1.84
6
3
0 200 400 600 800 1000 1200
T(s)
Figure 12.12 Maximum tension variation of Cn (Ct = 0.015).
50
Z(m)
100 Ct=0.01
Ct=0.015
150 Ct=0.02
Ct=0.025
200
250
300
0 200 400 600 800 1000 1200

T(s)
Figure 12.13 Depth variation of towed body with Ct (Cn = 1.44).

50
Z(m)
100
Cn=1.2
Cn=1.44
150
Cn=1.53
Cn=1.84
200
250
300
0 200 400 600 800 1000 1200

T(s)
Figure 12.14 Depth variation of towed body with Cn (Ct = 0.015).
towed system is in the stable state, the tension of the towed cable is deter-
mined by Ct, the towed depth of the towed body is determined by Cn, as
well as the ocean current velocity and the density of the towed cable.
Here, I have to point out that as reference [11] does not use the same
cable and towed body specifications, it is impossible to make some com-
parisons between the model in this chapter and in reference [11]. Some
important parameters for the simulation in OrcaFlex cannot be found in
reference [11]. As a result, now it is hard for me to build the model of refer-
ence [11] in OrcaFlex. If enough parameters of the model can be obtained,
I think it is possible to run it in OrcaFlex. I will try it in the further research.
Although the simulation parameters and simulation methods are different,
they come to the same conclusions indeedly.
12.3.6 The Effect of the Added Mass Coefficient

of the Towed Cable
When the towed cable accelerates in water, it causes the surrounding fluid
to accelerate. Because the fluid has inertia, it will produce a reaction force to
the towed cable. At this time, the force that pushes the cable will be greater
than its own inertial force, which is seemed as the increase of the mass of
the towed cable itself. The additional force is the force that is greater than
the inertia of the cable itself. Generally, the velocity and acceleration of the
ocean currents change with time, resulting in additional inertial forces
on the towed cable, which results in changes in the tension and towed
configuration of the cable. So, it is important to analyze the influence of

added mass coefficient on the towed system in the design process.
Relevant research shows that the added mass coefficient Ca of the towed
cable is not only a function of the Reynolds number Red and the roughness
k, it is also related to the Keulegan-Carpenter number Kc. If the Reynolds
number Red > 1.5 × 106, then Ca is taken as 0.8. In some specifications, the
range and requirements of Ca are given directly. According to the relative
roughness and Reynolds number curve, the Ca value can be obtained in [12].
This section studies the influence of the added mass coefficient on the
tension of the towed cable. The towing ship is towed straight at a steady
speed, and the current speed is uniformly increased by 0.1 m/s2. The cur-
rent is lateral to the cable tangent, in the horizontal plane. To obtain the
statistical equilibrium of the system, 8 s is needed to build the static equi-
librium stage. Then, another 20 s is needed to make the current speed up
to 2 m/s. When the current speed reaches 2 m/s, the numerical simulation
stops. It means that the total time of the simulation model is 28 s. Then, the
cable will move sideways with the current that will make the results larger.
As is known to us, typically, the increasing current is introduced to verify
the effects of drag coefficient and added mass coefficient. But, here, I make
the current varying with time and not constant. If we keep the current
22 Difference between black and

Cd=1.44, Ca=0
blue lines shows the impact of
20 the drag coefficient, while Cd=1.44, Ca=0.8
difference between black and Cd=0, Ca=0.8
18
red lines shows the impact of
16 the added mass coefficient.
14
F(KN)
12
10
4
0 50 100 150 200 250 300
L(m)
Figure 12.15 Distribution of tension along the length direction of the 300-m cable (in
Matlab).
24
Cd=1.44, Ca=1
F(KN)
Cd=1.44, Ca=0.75
Cd=1.44, Ca=0.25
22
Cd=1.44, Ca=0
20
18
0 50 100 150 200 250 300

L(m)
Figure 12.16 Distribution of tension along the length direction of the 300-m cable (the
drag coefficient is kept constant, while added mass coefficient is varied).
constant and just change the additional mass coefficient in a certain range,
maybe the influence will not be clearly reflected. In my opinion, to some
extent, this will amplify the influence of the change of the additional mass
coefficient on the calculation results, which makes the calculation more
convenient to observe. The results are shown in Figures 12.15 and 12.16.
From Figure 12.15 to 12.16, it can be concluded that the added mass
coefficient affects the magnitude of the tension of the towed cable, but its
effect is not as significant as the drag coefficient. It can be seen from Figure
12.15 (in Matlab) that the proportion of the components of the drag force
in the tension distribution is larger than that of the additional inertial force.
Compared with the current amplitude, as the diameter of the towed cable
and the additional inertial force is very small, the drag force plays a major
role. Different added mass coefficients cause different tension values, but
the change is very small, just as shown in Figure 12.16.
12.4 Conclusions
1. During the whole process, there is a turning critical radius,
the towed body motion always lags behind the towing ship
motion; in the course of turning, the trajectory of the towed
body is located inside the trajectory of the towing ship, the
turning radius of the towed body and the time required for
steady rotation vary with the change of the turning radius of

the towing ship, and this confirms the findings of reference
[8]. The dynamic model of reference [8] can be reproduced
in OrcaFlex under certain assumptions. If enough parame-
ters of the model can be obtained, I think it is possible to run
the model of reference [11] in OrcaFlex, I will try it in the
further research.
2. When the total mass of the cable is constant, as the diameter
of the cable increases, the stable towed depth of the towed
body decreases; the larger the diameter of the towed cable,
the smaller the maximum tension with constant Ct at its
upper end.
3. When the towed system is in the stable state, the tension of
the towed cable is determined by Ct, the towed depth of the
towed body is determined by Cn, as well as the ocean cur-
rent velocity and the density of the towed cable.
4. The added mass coefficient affects the magnitude of the ten-
sion of the towed cable, but its effect is not as significant as
the drag coefficient. Therefore, in the actual project, a rea-
sonable choice of Cd and Ca is still worthy for further explo-
ration in the future.
Acknowledgments
Funding for this work was provided by the National Natural Science
Foundation of China (grant no. 11272160).
References
1. Zhu Keqiang, Li Daochang, Li Weiyang, 2002, “Lumped-parameter anal-
ysis method for time-domain of ocean cable-body systems”, The Ocean
Engineering, 20(2), pp.100–104.
2. Koh C G, Rong Y, 2004, “Dynamic analysis of large displacement cable
motion with experimental verification”, Journal of Sound and Vibration,
27(2), pp.1–2.
3. Ablow C M, Schechter S, 1983, “Numerical simulation of undersea cable
dynamics”, Ocean Engineering, 10(6), pp. 443–457.
4. Sun F J, Zhu Z H, La Rosa M, 2011, “Dynamic modeling of cable towed body

using nodal position finite element method”, Ocean Engineering, 38(4),
pp. 529–540.
5. Gerler M, Hagen G., 1967, Standard equations of motion for subma-
rine simulation, NSRDCR Report No. 2510, David Taylor Research Center,
Washington.
6. Abkowitz M A, 1969, Stability and motion control of ocean vehicles, MIT
Press, pp. 32–50.
7. Buckham B J, Driscoll F R, Nahon M, 2003, “Three dimensional dynamics
simulation of slack tether motion in an ROV system”, Offshore and Polar
Engineering Conference,pp. 127–134.
8. M. A. Grosenbaugh, 2007, “Transient behavior of towed cable systems during
ship turning maneuvers”, Ocean Engineering, 37(11-12), pp. 1532–1542.
9. White, F. M, 1972, “An analysis of axisymmetric turbulent flow past a long
cylinder”, J. Basic Eng., ASME, 94, pp. 200–206.
10. Egil Pedersen, AsgeirJ. Sørensen, 2001, “Modelling and control of towed
marine seismic streamer cables”, IFAC Control Applications in Marine
Systems, Glasgow, Scotland, UK, 34(7), pp. 89–94.
11. Wang Fei. Simulation and control research of marine towed seismic sys-
tem [D]. Shanghai Jiaotong University, 2006.(in Chinese)
12. Miao G P, 1995, Mechanical theory of Flexible Components. Shanghai:
Shanghai Jiaotong University Press, pp. 15–23.
13
Dynamic Analysis of Umbilical
Cable Under Interference
13.1 Introduction
The umbilical cable can assembly electric cable, optical cable, steel tube,
and other fillers in a certain way, which can realize the integrated trans-
mission of different media. It is mainly used for the connection of the float
underwater production equipment and subsea production system and it
also can provide power, signal transmission, and hydraulic and corrosion
inhibitors, and other reagents for underwater oil exploration equipment.
Since the early finite element technology and computer performance
were not yet mature, based on small deformation and linear elasticity
assumptions, some analytical analysis for umbilical cable were made by
some scholars (Fachri P et al., 2014) [1]. Combined with the experimental
results, the influence of the length of the finite element and the terminal
on the bending stress of umbilical cable was discussed by Witz and Tan Z
(1992) [2]. Considering material nonlinearity, gap formation, and the lat-
eral contact between helical component and helical component curvature
change and other factors, Custodio et al. derived an overall response non-
linear calculation method for the umbilical cable each layer stress and the
whole structure [3]. Sævik and Bruaseth (2005) [4] and Shunfeng Gong
et al. (2014) [5] made some theoretical and experimental research on the
complex cross-section of umbilical cable, then they put forward a finite
element analysis method that took the contact and internal pressure for
umbilical cable into account under different load conditions. A fatigue
assessment of the Foinaven dynamic umbilical including VIV was given
by F. Trarieux et al. [6]. Weidong Ruan et al. (2014) made a static analysis
of deepwater lazy-wave umbilical [7]. Yong Bai et al. (2015) [8] made an
analytical prediction of umbilical behavior under combined tension and
internal pressure, in which the contact problem between two adjacent lay-
ers has been taken into account. The umbilical deployment modeling for
267
tethered UAV detecting oil pollution from was made by Frédéric Muttin
(2011), in which two continuous models are presented using a Lagrangian
approach or an updated Lagrangian one, both including different kinds
of boundary condition at the umbilical bottom end [9]. When the umbil-
ical cable is on work status, the additional dynamic load caused by the
heave and rotation motion of the floating structures which is connected
with the umbilical cable and the combined action of the wind, wave and
current make the configuration of the umbilical cable be constantly chang-
ing under water. All these factors will inevitably result in the occurrence
of extrusion-compression, stretching, bending, and twisting in varying
degrees for the umbilical cable. The change of the configuration of the
umbilical cable can cause the collision with the adjacent risers. Collision
is a complex and strongly nonlinear dynamic response process, which is
a complex and strong nonlinear dynamic response of structures in very
short time. This kind of collision is called interference. Cycle and continu-
ous collisions will accelerate the wear and fatigue of the umbilical cable. At
present, there are few researches on the interference for the umbilical cable
in the domestic and foreign literatures.
A fast and accurate contact algorithm is very important to study the
dynamic analysis of the umbilical cable. The collision between the umbil-
ical cable and adjacent risers involves complex contact settings, and
the position of collision changes with time and the configuration of the
umbilical cable. The twist among the umbilical cable and other cables is
extremely easy to happen during the collision process. As the dynamic
operation process will result in large displacement, which makes the orig-
inal contact relations become more complex. Therefore, if using the finite
element method and finite element software ABAQUS to analyze the inter-
ference, due to the randomness and uncertainty of the collision, the con-
tact setting of model parts has become more difficult. To make it possible,
sometimes, a specific unit to form a grid position interpolation function
is needed in ABAQUS. What’s worse, the workload of the direct program-
ming is too huge, whose calculation process is extremely difficult to con-
vergence. There are many dynamic modeling methods for umbilical cable,
such as lumped mass method, finite element method, and finite difference
method, and so on. Among them, the physical meaning of lumped mass
method is clear, the algorithm is simple and easy to understand, with a
wide range of applicability and scalability. In consideration of the bend
moment and torque, the lumped mass method was derived. Based on the
hydrodynamic software OrcaFlex [10], combined with the lumped mass
method, the dynamic analysis model of umbilical cable under interference
has been made. The nonlinear dynamic characteristics of the interference
Umbilical Cable Dynamic Analysis Under Interference 269
have been obtained by time domain coupled dynamic analysis method.

Combined with the results of dynamic simulation, some guidance is given,
which is important to ensure the security during the operation.
13.2 Dynamic Model of Umbilical Cable

The method can be applied to the large deformation dynamic analysis of
marine flexible cable. As the effects of bending and torsion are considered,
it can also be used in the analysis of the submarine pipelines and risers. The
advantage of this method is its simplicity and generality. The advantage of
this method is that it is generally applicable to the coastal cable structure
and it only uses three translational degrees of freedom and a rotational
variable for each node. The node is connected by linear elastic element,
all the force (damping force, gravity, etc.) are considered to be acting on a
node. Then, the equations of the motion of the nodes are listed, with using
a computer to carry out numerical integration to calculate the instanta-
neous position and velocity components of each node and the tension
changing with time.
13.2.1 Establishment of Mathematical Model

In order to simplify the calculation, the structure of the umbilical cable is
considered as a flexible, smooth, and circular cross-section without any
attached components. The static equilibrium and moment equilibrium for-
mula of the umbilical cable are given by considering the external load of a
micro section see Figure 13.1:
  
∂Te ∂V   ∂2 R
+ + w + w eh = m 2 (13.1)
∂sε ∂sε ∂t
M+∂M/∂sε sε
–V
–M sε Te +∂T/∂sε sε
–Te
V+∂V/∂sε sε
W + Weh + Wg
Figure 13.1 Load analysis of pipeline.

 
∂( M + H )    
• ∇sε − V × ( Rk+1 − Rk ) + ∇sε • q = 0 (13.2)
∂sε

sε represents the length direction of the marine cable, w is the weight of

the unit length of the marine cable, and w eh is hydrodynamic load per unit
length.  
Torque vector H, moment vector M, and effective tension Te are respec-
tively related to torsional stiffness GIp, bending stiffness EI, and axial stiff-
ness EA:
Te = EAε (13.3)
  
( )
H = H ∂R ∂sε = GI Pθ ′ ∂R ∂sε( ) (13.4)
  
( ) (
M = EI ∂R ∂sε × ∂ 2 R ∂sε2 ) (13.5)
The torque H and the distributed load can be obtained from (13.2) mul-
tiplied by the unit tangent vector to get the scalar formula:

∂H  ∂R
+q• =0 (13.6)
∂sε ∂sε
The flow field around the pipeline is very complex due to the influence
of water flow and wave. Although the Navier-Stokes equations or discrete
vortex model (DVM)–based on the numerical solution of the CFD can be
used to determine the hydrodynamic load value, but the program did not
have enough economic benefits. The Morison equation of the semi empir-
ical method is presented as follows:
 1    π 
w eh = ρsw D(Cdn |vn|vn + π Cdt |vt|vt ) + D 2 ρswCanan (13.7)
2 4
where
  

((
 
vt = ∂R ∂sε vc + vw − ∂R ∂t • ∂R ∂t

) )
   
vn = vc + vw − ∂R ∂t − vt
13.2.2 The Discrete Numerical Method for Solving

the Lumped Mass Method
In general, we assume that the umbilical cable is slender, flexible, and
cylindrical. The numerical solution of the boundary problem for umbilical
cable is the discrete lumped mass method. The basic idea of the lumped
mass method is to divide the umbilical cable into N line segments, and
the mass of each segment is lumped to a node, so there are N + 1 nodes.
The tension Tat the end of each section and the shear force V can be seen
as a centralized function at a node, any external hydrodynamic loads and
other properties are all lumped to the nodes. That is also the foundation
of the mathematics model in OrcaFlex. In the continuity equation, the
lumped mass model is replaced by the finite difference method to obtain
the numerical solution see Figure 13.2.
Equations (13.2), (13.4), and (13.5) can also be written as
    
(
∂( M + H ) ∂sε + ∂R ∂sε × V = −q ) (13.8)
  
(
H = H ∂R ∂sε = Ht k ) (13.9)
     
( ) ( )
M = EI ∂R ∂sε × ∂ 2 R ∂sε2 = EIt k × (t k − t k−1 ) ∆sε k (13.10)
Equation (13.8) can also be written as the following expression:
Segment 2
Segment 1
Segment 3
Real Umbilical
Segment 2
Segment 1
Segment 3
Node 2 Node 3
Lumped Mass
Node 1
Figure 13.2 Schematic diagram of lumped mass method.

   
(M − Mk
k +1
+
) (
H k+1 − H k   ) 
+ t k × Vk = −qk (13.11)
∆sε k ∆sε k
Then, we put Eqs. (13.8) and (13.9) into Eq. (13.11), we can get the
expression form of Eq. (13.12):
       
EIt k+1 × (t k+1 − t k ) EIt k × (t k − t k−1 ) H k+1 • t k+1 − H k • t k   
− + + t k × Vk = −qk
∆sε k +1 ∆sε k ∆sε2k ∆sε k
(13.12)
Equation (13.12) can be simplified as Eq. (13.13):

     
EIt k+1 × t k EIt k × t k−1 H k+1 • t k+1 − H k • t k   
− − + + t k × Vk = −qk
∆sε k+1 ∆sε k 2
∆sε k ∆sε k
(13.13)

Then, we get Eq. (13.14) by t k × (13.13):
       
EIt k × (t k − t k+1 ) EIt k × (t k−1 − t k ) H k+1t k × t k+1     
− + + t k × (t k × Vk ) = −t k × qk
∆sε k +1 ∆sε k 2
∆sε k ∆sε k
(13.14)
   
As t k × (t k × Vk ) = −Vk , we can get the following equation:
       
 EIt k × (t k × t k+1 ) EIt k × (t k−1 × t k ) H k+1t k × t k+1  
Vk = − + + t k × qk (13.15)
∆sε k +1 ∆sε k ∆sε2k ∆sε k
After the numerical simulation analysis, the equation of motion of the

K-th node is expressed as
      
M Ak Rk = Tek − Tek−1 + FdI k + Vk − Vk−1 + w k ∆ sk (13.16)
where

Tek = EAε k ⋅ tk (13.17)
 π  π  
M Ak = ∆ sk  mk + Dk2 (Can − 1) I − ∆ sk Dk2 (Can − 1)(t k ⊗ t k−1 )
 4  4
(13.18)
( ))
   
1  
(   π 
)
FdI k = ρsw Dk 1 + ε k ∆ sk Cdnk vnk vnk + π Cdtk vtk vtk + Dk2 ρsw Cank ∆ sk aw k − aw k ⋅ t k t k
2 4
(
(13.19)
       
 EI k+1t k × (t k × t k+1 ) EI k t k × (t k−1 × t k ) H k+1t k × t k+1
Vk = − +
∆sε k ∆sε k+1 ∆sk2 ∆sε k
(13.20)
According to the mathematical relationship, Eq. (13.21) is shown in the

following:
           
t k × (t k × t k+1 ) = −(t k × t k+1 ) × t k = −t k+1 ⋅ (t k ⋅ t k ) + t k ⋅ (t k+1 ⋅ t k )
(13.21)

( )
As t k = t x k , t yk , t z k , Eq. (13.21) can be expressed as Eq. (13.22):
        
t k × (t k × t k + 1 ) = − t k + 1 ⋅ (t k ⋅ t k ) + t k ⋅ (t k + 1 ⋅ t k )
( )(
= − t x k +1 , t yk +1 , t z k +1 ⋅ t x2k + t y2k + t z2k )
( )(
+ t xk , t yk , t z k ⋅ t x k +1 ⋅ t x k + t yk +1 ⋅ t yk + t z k +1 ⋅ t z k )
( ( ) ( ) (
= −t x k +1 t y2k + t z2k − t yk +1 t x2k + t z2k − t z k +1 t x2k + t y2k ))
( ( ) ( ) (
+ t x k t yk t yk +1 + t z k t z k +1 + t yk t x k t xk +1 + t z k t z k +1 + t z k t x k +1 t x k + t yk +1 + t yk ))
(13.22)
The presented method can also be used for the calculation of the follow-
ing equation:
        
t k × (t k − 1 × t k ) = − t k (t k − 1 ⋅ t k ) + t k − 1 (t k ⋅ t k )
( )(
= − t x k , t yk , t z k ⋅ t x k −1 t x k + t yk −1 t yk + t z k −1 t z k )
(
+ t xk −1 , t yk −1 , t ) ⋅ (t
z k −1
2
xk +t +t2
yk
2
zk )
( ) ( ) (
= −t xk t yk −1 t yk + t z k −1 t z k − t yk t x k −1 t xk + t z k −1 t z k − t z k t xk −1 t xk + t yk −1 t yk )
(
+t x k −1 t + t 2
yk
2
zk ) + t (t
y k −1
2
xk +t 2
zk )
(13.23)
To make it easier to express, we put forward a variable symbol Qk:
 0 −t z k t yk   −t 2 − t 2 t x k t yk t xk tzk 
   yk zk

Qk =  t z k 0 −t x k 2 
 , Qk = t x t y −t z2k − t x2k t yk t z k 
   k k 
 t yk t xk 0   t x tz t yk t z k −t x2k − t y2k 
   k k 
(13.24)
Based on Eqs. (13.22), (13.23), and (13.24), each term of Eq. (13.19) can
be expressed as
  
EI k+1t k × (t k × t k+1 ) EI k+1  
= Qk2 ( Rk+2 − Rk+1 ) (13.25)
∆sε k ∆sε k+1 2
∆sε k ∆sε k+1
  
EI k t k × (t k−1 × t k ) EI k 2
 
= − Qk ( Rk − Rk−1 ) (13.26)
∆sε2k ∆sε2k ∆sε k−1
 
H k+1t k × t k+1 H k+1  
= Qk ( Rk+2 − Rk+1 ) (13.27)
∆sε k ∆sε k ∆sε k+1
The difference of the external force in Eq. (13.16) of motion can be

expressed as:
       
   EI t × (t k × t k +1 ) EI k t k (t k −1 × t k ) H k +1t k × t k +1 
Vk − Vk −1 =  k +1 k − + 
 ∆sε k ∆sε k +1 ∆sk2 ∆sε k 
       
 EI t × (t k × t k +1 ) EI k t k × (t k −1 × t k ) H k +1t k × t k +1 
−  k +1 k − + 
 ∆sε k ∆sε k +1 ∆sk2 ∆sε k 
 EI   EI  
= k +1
2
Qk2 ( Rk + 2 − Rk +1 ) + 2 k Qk2 ( Rk − Rk −1 )
 ∆sε k ∆sε k +1 ∆sε k ∆sε k −1
H k +1     EI k   EI
+ Qk ( Rk + 2 − Rk +1 ) −  Qk2−1 ( Rk +1 − Rk ) + 2 k −1
∆sε k ∆sε k +1 2
  ∆sε k −1 ∆sε k sε k −1 sε k − 2
  Hk     EI 
Qk2−1 ( Rk −1 − Rk − 2 ) + Qk −1 ( Rk +1 − Rk ) =  2 k −1 Qk2−1  Rk − 2
∆ssε k −1 ∆sε k   ∆sε k −1 ∆sε k − 2 
 EI EI
+  − 2 k Qk2 − 2 k −1 Qk2−1  Rk −1 +  k 2 k
  (
 EI Q 2 + Qk2−1
+
)Hk 
Qk −1  Rk
 ∆sε k ∆sε k −1 ∆sε k −1 ∆sε k − 2   ∆sε k ∆sε k −1 ∆sε k −1 ∆sε k 
 EI k +1 EI H H  
+ − Qk2 − k
Qk2−1 − k
Qk −1 − k +1
Qk Rk +1
2
 ∆sε k ∆sε k +1 ∆sε k −1 ∆sε2k ∆sε k −1 ∆sε k ∆sε k ∆sε k +1 
 EIk + 1 H k +1 
+ Qk2 + Qk  Rk + 2
2
 ∆sε k ∆sε k +1 ∆sε k ∆sε k +1 
(13.28)
   
Tek − Tek −1 = Tek t k − Tek −1 t k −1
   
( ) (
= Tek ( Rk +1 − Rk ) ∆sε k − Tek −1 ( Rk − Rk −1 ) ∆sε k −1 )
  
(
= Tek −1 ∆sε k −1 , −Tek −1 ∆sε k −1 ,Tek ∆sε k ,Tek ∆sε k  ⋅ Rk −1 , Rk , Rk +1 )
  
(
=  EAk −1ε k −1 ∆sε k −1 , − EAk −1ε k −1 ∆sε k −1 − EAk ε k ∆sε k , EAk ε k ∆sε k  ⋅ Rk −1 , Rk , Rk +1 )
(13.29)
By Eqs. (13.1)~(13.29), Eq. (13.16) can be represented in the form of a

matrix:
     
{
K k Rk−2 , Rk−1 , Rk , Rk+1 , Rk+2 } T
= Fek (13.30)
where Kk denotes the k-th segmental stiffness matrix, which consists of the
following five sub-matrices:
Kk = [Ak, Bk, Ck, Dk, Ek] (13.31)
EI k−1
Ak = 2
Qk2−1 (13.32)
∆s ∆sε k−2
ε k −1
EI k EI EA ε
Bk = − Qk2 − 2 k−1 Qk2−1 + k−1 k−1 I (13.33)
2
∆s ∆sε k−1
εk ∆sε k−1 ∆sε k−2 ∆sε k−1
Ck =
(
EI k Qk2 + Qk2−1
+
)Hk  EA ε EA ε 
Qk−1 −  k k + k−1 k−1  I
2
∆sε k ∆sε k−1 ∆sε k−1 ∆sε k  ∆sε k ∆sε k−1 
(13.34)
EI k +1 EI k Hk H k +1 EA ε
Dk = − Qk2 − Qk2−1 − Qk2−1 − Qk + k k I
∆sε k ∆sε2k +1 ∆sε k −1 ∆sε2k ∆sε k −1 ∆sε k ∆sε k ∆sε k +1 ∆sε k
(13.35)
EI k+1 H k+1
Ek = Qk2 + Qk (13.36)
2
∆sε k ∆sε k+1 ∆sε k ∆sε k+1
 1 0 0 
 
I= 0 1 0  (13.37)
 0 0 1 
Based on Eq. (13.16), the external forces of Eqs. (13.30) and (13.31)
are as the following expression:
   
Fek = − FdI k − w k ∆ sk + M Ak Rk (13.38)
The velocity and acceleration of the nodes in Eq. (13.40) can be obtained
by using the Newmark-β algorithm:
    1   
Rk(n ) = Rk(n−1) + R k(n−1)∆t +   − β  Rk(n−1) + β Rk(n )  ∆t 2 (13.39)
 2  
  1  
( )
R k(n ) = R k(n−1) + Rk(n−1) + Rk(n ) ∆t
2
(13.40)
Therefore, for the time step (n), the equation of motion in the inertial
frame is
[K](n){R}(n) = {Fe}(n) (13.41)
[K] is the inertial stiffness matrix of 3N × 3N, {R} and {Fe} are the
three-dimensional nodal position vector and the external force vector,
respectively. Through the numerical integration of the above-mentioned
single-node position vector and velocity vector, we can get the nodal dis-
placement, velocity vector, and pipeline tension with time response.
The analytical method can be applied to the large deformation dynamic
analysis of the ocean flexible cable, which is convenient for the analysis and
design of the submarine pipelines. It can solve a series of ocean cable and
pipeline problems by using three position degrees of freedom and an inde-
pendent torsion variable of each node, which has important significance
for the practical engineering design.
13.2.3 Calculation of the Clashing Force of Umbilical Cable

When the interference collision is considered, the umbilical cable and the
riser are respectively equivalent to a flexible body model of several cylin-
ders. When the two cylinders contact each other, the normal contact force
is calculated by the model based on the penalty function, the actual objects
in the collision process will be equivalent as a nonlinear spring damping
model based on penetration depth, which is the most commonly used in
collision model. In this collision model, the most direct impact force is
the normal contact force. In order to simplify the calculation process, the
influence of the normal contact force is mainly considered, and the tan-
gential contact force is not considered. R1 and R2 are the outer radius of
the umbilical cable and riser, respectively. To make it clear to understand,

we name the distance between the two middle axial axes of the umbilical
cable and riser δ. When δ < R1 + R2, the two collide and the interference
occurs; when δ > R1 + R2, the two don’t collide and the interference doesn’t
occur; when δ = R1 + R2, the contact relationship between the two is at the
edge where the collision occurs and does not collide; in other words, at this
time, the two cables will not collide, the two just begin to come into contact
with each other but also did not produce the role of force. Contact about
the relation between these three kinds of cable-to-cable, if placed in the
plane, the relationship between the two cables can be described as the con-
tact relationship of the intersection, separation, being externally tangent of
two circles see Figure 13.3.
The clashing force of interference between cable-to-cable in OrcaFlex
can be expressed as Fc:
Fc = k[δ – (r1 + r2)] + Fd (13.42)
Among them, δ is the distance between the two middle axial axes of the
umbilical cable and riser, Fd is the structural damping force, and K is the
coefficient of the contact stiffness and its expression is as the following:
1 1
k = 1/  +  (13.43)
 k1 k2 
K2 and K1, respectively, are the contact coefficient of the umbilical cable
and riser, the unit is kN/m.
Spline
z
y
Splined Line Segment x
Contact Surface
Line End Axes
y
z
x
Penetrating Line Nodes
Figure 13.3 Schematic diagram of contact relationship for cable-cable.

The expression of structural damping force is:
Fd = cv (13.44)
where c is the structural damping coefficient, v is the relative velocity of

the two during the collisions, when the relative velocity is less than zero,
it means that there is no damping force between the umbilical cable and
riser; when the relative velocity is greater than zero, it means that the two
cables are getting close with each other, there is the role of damping force.
It is important to note that in OrcaFlex, the coefficient c is directly assigned
a value of 0 in the calculation of the structural damping force if the inte-
gration method is chosen for the implicit integration. In this chapter, the
implicit integration is chosen to simplify the calculation results, so the
impact force of this chapter is not fully considered the influence of struc-
tural damping force. The influence of structural damping force will be fur-
ther considered in the next study.
13.3 The Establishment of Dynamic Simulation

Model in OrcaFlex
13.3.1 The Equivalent Calculation of the Stiffness
of the Umbilical Cable
As the umbilical cable is a multi-layer winding structure, the multi-layer
winding composite structure can’t be established in OrcaFlex. So, it needs
to be equivalent in the modeling process to obtain the equivalent bending
stiffness, equivalent tensile stiffness, and equivalent torsional stiffness. The
equivalent basic principle is that the original unit weight and mechanical
properties are the same before and after the equivalent transformation.
The equivalent bending stiffness is calculated by using Costello’s method
[11] based on LOVE screw theory which ignores the unit interaction con-
tact of single helical elements bending stiffness calculation formula. Then,
the bending stiffness of the equivalent components is calculated by the
integral bending stiffness of the structure as shown in formula (13.45):
K B = ( EI )0 + ∑ 2 2+Ev I+cossinαα
i =1
i i
i
2
i
i
(13.45)
where, KB is the overall equivalent bending stiffness; EiIi is the bending

stiffness of the i-th element; (EI)0 is the bending stiffness of the vertical
cylinder element; n is the number of helical units; αi and νi are winding
angle and Poisson ratio, respectively.
The equivalent axial stiffness and the torsional stiffness are calculated
by using Knapp’s two simplified cases of rigid core and incompressible soft
core. The tensile-torsional analysis model established when radial shrink-
age and relative sliding is neglected. The equivalent axial stiffness and the
torsional stiffness are shown in Eqs. (13.46) and (13.47), respectively:
n
 
KT = ( EA)0 + ∑ E A  1 − Θ2RR tan α  cos α
i =1
i i
c
i
3
i
3
(13.46)
K= ∑ A E R sin α ⋅ cosα ⋅+ J G
i =1
i i i
2 2
i i c c (13.47)
where KT is the overall equivalent axial stiffness of the pipeline; (AE)0 is the
axial stiffness of the vertical cylinder element; n is the number of helical
units; AiEi is the axial stiffness of the i-th element; Rc is the radius of the
core, Ri is the distance from the wire to the center of the cylinder. Θ is a
variable, when the core is assumed to be a rigid material, Θ is taken as 0,
and Θ is taken as 1 when the core is an incompressible soft material. In the
formula (13.47), K is the torsional stiffness, and JcGc is the torsional stiff-
ness of the center cylinder.
The outer layer of the steel tube umbilical cable is usually protected by a
polymer sheath, and the inner layer of the umbilical cable is composed of
nine steel tubes, four cables, and fillers, which are composed of a plurality of
functional members wound at a certain helical angle. The specific composi-
tion of the tube element comprises central tube elements and external tube
elements. In four 1-inch central tubes, one is for the methanol delivery, one
is for the hydraulic return, one is the ring hole tube, one is the spare tube; in
five external 1/2 inch tubes, four tubes are for the hydraulic transmission,
one is the scale inhibitor tube. Electrical unit: the internal is three copper
conductors, the external is insulation and filling layer. The rest is filler. The
specific section size of the umbilical cable is shown in Figure 13.4, the size
of the unit is mm. After equivalent calculation based on the above method,
the equivalent bending stiffness of the umbilical cable is 14. 4 kN•m2, the
equivalent axial stiffness is 509MN, and the torsional stiffness of 45 kN•m2.
67°
R16, 24
11
6 5
R66, 46
T2 T3
φ 32, 47 R65, 27
1 4
R16, 24
R13, 2
R3, 84
2 3
T1 9
10
7 8
Figure 13.4 Schematic diagram of umbilical cable.
13.3.2 RAO of the Platform

There are many different conventions for defining RAOs. Some attempts
were made for the standardization but such attempts all failed at last. So, it
remains different between the main computer programs and model basins.
Some establishments even use different conventions in reporting model
and computed data. The only safe approach is to obtain a complete descrip-
tion of the system used for the data in each case.
The convention is to use the amplitude of response (in length units for
surge, sway and heave, and in degrees for roll, pitch, and yaw) per unit wave
amplitude and to use the phase lag from the time the wave crest passes the
RAO origin to the time when the maximum positive excursion is reached (in
other words, the phase origin being at the RAO origin). It can be expressed as
x = R · a · cos(ωt − φ) (13.48)
where:
x is the platform displacement (in length units for surge, sway
and heave, and in degrees for roll, pitch, and yaw);
a and ω are wave amplitude (in length units) and frequency (in
radians/second), respectively;
t is time (in seconds);
R and φ are the RAO amplitude and phase, respectively.
13.3.3 The Choice of Wave Theory

Regular wave trains are defined in OrcaFlex by wave period, wave height,
and water depth. Which wave theory should be used for any given wave
train? For an infinitesimal wave in deep water then airy wave theory is
accurate. For finite waves, a non-linear theory should be used. In order to
decide which wave theory to use, one must calculate the U numerical value
which is given by:
U = HL2 / d3 (13.49)
where H is wave height; L is the wavelength; d is the water depth.

If U < 40, the wave is short, we should use the Stokes’ 5th theory, if U > 40
which is a constant wave, we use the cnoidal wave theory. Note that Dean
Stream theory applies to any wave, and if another theory is used, it should
be compared with Dean Stream theory to check its validity. But, the value
of 40 is not an accurate fixed standard. When the U value is close to 40,
using the Stokes’ 5th theory and the cnoidal wave theory are not accurate,
OrcaFlex recommends Dean Stream function theory. In summary, in most
cases, OrcaFlex recommends Dean Stream theory. In this chapter, in order
to ensure the accuracy of the calculation, we selected Dean Stream theory
(stream function order taken as 10).
13.3.4 Establishment of Model in OrcaFlex

The environmental parameters are as follows: the sea water density is
1,025 kg/m3, the water depth is 700 m, the seabed is flat, wave height is 8 m,
wave period is 9s, wave direction is from 0° to 180°, every 30° for a direction;
the umbilical cable length is 732 m, the outer diameter is 0.13292 m, the
inner diameter is 0m, the linear density is 35.3 kg/m, the Poisson’s ratio
is 0.5, the upper end is connected with the platform, the bottom end is
anchored to the seabed, the axial and normal friction coefficient between
umbilical cable and seabed are 0.5; The upper end of the adjacent riser is
connected with the platform, the bottom end is unrestricted, the length is
485 m, the outer diameter is 0. 65 m, the inner diameter is 0.6 m, the linear
density is 38.5 kg/m, the bending stiffness is 500,000 kN·m2, the axial stiff-
ness is 10,000 MN, the torsional stiffness is 0, and the Poisson ratio is 0.3.
The normal contact coefficient of the two pipelines is 5,000 kN/m. In order
to ensure the effective contact between node and node during the collision
and reflect the motion of the pipeline as much as possible, the stress sensi-
tive zone near the upper ends of the two pipelines is subdivided, the 0- to
10-m area is discretized into several segments of the length of 1-m segmen-
tation unit, the remaining length is divided into segments of the length of
Riser Umbilical
Figure 13.5 Schematic diagram of the model.
5-m segmentation unit. The model of the system completed in OrcaFlex is

as shown in Figure 13.5.
13.4 The Calculation Results

13.4.1 The Clashing Force of Interference
With the observation of the maximum clashing force along the umbilical
cable length in Figure 13.6, we have found that the umbilical cable and
riser don’t collide with each other in the 0- to 100-m length range; the
reason for this phenomenon is that the fixation and confinement of the
upper part of the platform for the umbilical cable and the riser is dominant
in this region. With the increase in length and depth, the flexibility of the
riser and umbilical cables is beginning to be manifested, the occurrence
of different degrees of bending and swing of the umbilical cable and riser
begins to appear, as the free end of the riser is unrestricted, the impact of
the wave causes the riser to start the lateral swing, which is somewhat sim-
ilar to the waving of the whip, which, in turn, makes its spatial structure
and configuration constantly changing; this motion is the most intense at
the free end, so the collision between the umbilical cable and riser is more
intense and frequent at a distance of 485 m (450–485 m) from the top zone,
the distribution of the maximum clashing force of the umbilical cable and
the riser along the length and the cable and the distribution of the standard
deviation of the clashing force of the riser and the umbilical cable along
the length direction illustrate this. Beyond that, the clashing force of the
umbilical cable and the riser rapidly changes from 0 to a large value near
100 0° 0°
30° 1.2 30°
The Maximum Clashing Force/KN
60° 60°
The Maximum Clashing Force/KN

140 90°
The Mean Clashing Force/KN

80 0° 90°
30° 120 120° 1.0 120°
60° 150° 150°
90° 180° 180°
60 120° 100 0.8
150° 80
180° 0.6
40 60
0.4
40
20
20 0.2
0 0 0.0
0 100 200 300 400 500 600 700 0 100 200 300 400 0 100 200 300 400 500 600 700
Arc Length of Riser/m
The Standard Deviation of the Clashing Force/KN

Arc Length of Umbilical/m Arc Length of Umbilical/m
0°
The Standard Deviation Clashing Force/KN
0°
0° 6 30° 8 30°
1.6 30° 60° 60°
60° 90° 90°
The Mean Clashing Force/KN
1.4 90° 5 120° 120°

120° 150° 150°
1.2 150° 6 180°
180° 4 180°
1.0
0.8 3 4
0.6 2
0.4 2
1
0.2
0.0 0 0
0 100 200 300 400 500 0 100 200 300 400 500 600 700 0 100 200 300 400 500
Arc Length of Riser/m Arc Length of Umbilical/m Arc Length of Riser/m
Figure 13.6 The distribution of the clashing force for interference.
the top of the riser and the umbilical cable at a distance around 150 m
(100–200 m), and the mutation point of the clashing force for different
wave will be slightly different. It is found that there is a certain similarity
between the clashing force curve geometrical shapes of the umbilical cable
and the riser along the length direction, which indicates that the range of
the interference occurs in the range of the riser length, no collision hap-
pened outside the length of the riser. The distribution of standard devi-
ation of the clashing force of the umbilical cable and the riser along the
length direction are also observed. It is found that the geometrical shapes
of the umbilical cable and the riser have some similarity, which indicates
that the trends of the change of the clashing force have similar distribution
and variation in the time and along the length direction. At the same time,
the numerical values and distribution positions of the clashing force under
different wave directions are slightly different because the change in the
wave direction does not change for the riser and the umbilical cable them-
selves, but that the change of wave direction will affect the motion of the
platform which is connected to the upper end of the system; that is to say,
the change of wave direction will affect the upper end restraint of the riser
and umbilical cable; in a certain range, the change of the motion trajectory
of the platform will also change the configuration of the riser and umbil-
ical cable and it can cause small amplitude fluctuations in the numerical
value size and position of the clashing force. At the same time, the form
of collision for the umbilical cable and the riser is not the same under dif-
ferent wave directions; when it is under 0° and 180° wave directions, the
special layout between the wave direction and the two pipelines makes the
umbilical cable and riser have a frontal collision, so the collision position
of the clashing force occurs more in advance than other wave directions.
What’s more, relevant measures should be taken to relieve the collision in
the relevant area of 150, 300, and 485 m from the upper end.
13.4.2 The Variation of the Effective Tension

Under Interference
Since the effective tension of the free end of the riser is 0 at any time, its
effective tension curve versus time has not been drawn. The effective ten-
sion time domain curve of the two ends of the umbilical cable is observed
in Figure 13.7. It is found that the upper end of the umbilical cable is in a
periodic tension state, and the anchored end is in a state of periodic axial
compression; this periodic variation changes with different wave directions
make its tension amplitude will be a slight change, but its period will not
change. At the same time, it is shown that the effective tension at the upper
end of the riser is far larger than that of the upper end of the umbilical
cable, but the change period of the tension of the riser is the same as that of
the umbilical cable. To observe the distribution of the effective tension of
the umbilical cable along the length direction, the tension of the umbilical
cable is in a linear decreasing trend from the upper end to the anchored
end, and the tension of the anchored end becomes negative, which indi-
cates that the anchored end of the umbilical cable keeps be bearing the
axial compressive force during the process. With the observation of the
distribution of the standard deviation of the effective tension of the umbil-
ical cable along the length direction under different wave directions, we
have come to the conclusion that the standard deviation of effective tension
of the umbilical cable along the length direction increases sharply from 0
to 50 m for all the wave directions and it increases to a maximum at 50 m,
then it decreases to a steady value with a small amplitude reduction, and
this stable value is no longer changed with the increase of the cable length,
indicating the farther away from the platform, the more rapid and frequent
changes the effective tension will make in the 0- to 50-m range and the
tension of each part is more and more dispersed in the time domain in the
0- to 50-m range with the increase of the length; the dispersion degree of
the effective tension in the 50- to 732-m range tends to be stable; the stan-
dard deviations of the effective tension of umbilical cable show that by the
The Effective Tension of the Anchored End for Umbilical/KN

The Effective Tension of the Top End for Umbilical/KN
0° 0° 0°
41 30° 30° 30°
60° 60° 60°
The Tension of the Top End for Riser/KN

90° 0 90° 3600 90°
40 120° 120° 120°
150° 150° 150°
180° –100
180° 3400 180°
39
–200
38 3200
–300
37
–400 3000
36
–500 2800
35 –600
0 20 40 60 80 0 20 40 60 80 –20 0 20 40 60 80
t/s t/s t/s
0° 0°
0° 30° 30°
30° 60°
The Maximum Effective Tension of Umbilical/KN
60°
60° 4000 90° 4000 90°
The Minimum Effective Tension/KN

90° 120°
120°
The Mean Effective Tension/KN
4000 120° 150°

150° 150°
180° 3000 180° 3000 180°
3000
2000 2000
2000
1000
1000
1000
0
0
0
0 100 200 300 400 500 600 700 0 100 200 300 400 500 600 700 0 100 200 300 400 500 600 700
Arc of Length of Umbilical/m Arc of Length of Umbilical/m Arc of Length of Umbilical/m
0°
30° 0° 0°
60° 30° 30°
90°
The Standard Deviation of the Effective Tension/KN
3500 60° 60°

120° 90° 3000 90°
180
The Maximum Effective Tension/KN
150° 120° 120°

160 180° 3000
The Mean Effective Tension/KN
150° 2500 150°

140 180° 180°
2500
120 2000
2000
100
1500
80 1500
60 1000
1000
40
500 500
20
0 0 0
0 100 200 300 400 500 600 700 0 100 200 300 400 500 0 100 200 300 400
Arc of Length of Umbilical/m Arc of Length of Riser/m Arc of Length of Riser/m
0° 0°
The Standard Deviation of the Effective Tension/KN
30° 30°
60° 60°
3000 90° 250 90°
The Minimum Effective Tension/KN
120° 120°
2500 150° 150°
180° 200 180°
2000
150
1500
100
1000
500 50
0 0
0 100 200 300 400 0 100 200 300 400
Arc of Length of Riser/m Arc of Length of Riser/m
Figure 13.7 The distribution of the effective tension under interference.
descending order of magnitude of value of the standard deviations under

different waves are: 0°, 180°, 30°, 150°, 60°, 120°, and 90°. With the contrast
observation of the standard deviations of the effective tension of the riser,
we have found that according to sort from largest to smallest order of the
standard deviations under different wave directions, the order is: 180°, 0°,
150°, 30°, 120°, 60°, and 90°. In sharp contrast to the standard deviation of
the effective tension of the umbilical cable, the numerical value of the stan-
dard deviation of the effective tension of the riser decreases linearly along
the length direction and reaches zero at the free end, indicating the more
gradual trend in the riser tension along the length, the discrete degree of
the effective tension is more and more weak, which shows the effective
tension of the riser along the length direction is approaching their respec-
tive averages under different wave directions in the time domain; in other
words, the closer to the free end, the smaller the amplitude of the tension
is in the time domain, the reason for this phenomenon is: with the distance
from the free end closer, the sharp and violent stretching action of the plat-
form’s motion on the riser is buffered by the damping action of the water,
which makes the sharp change in the time domain gradually approaching
to a constant load. At the same time, comparative observation of umbilical
cable and the riser effective tension distribution along the length direction,
the maximum effective tension along the riser length direction occurs at
the upper end, but the maximum effective tension of the umbilical cable
does not occur at the upper end, which is at the position which has a dis-
tance of 25 m from the upper end.
13.4.3 The Variation of Bending Under Interference

With the observation of the bending moment and curvature for the
umbilical cable along the length direction, we have found that the degree
of bending is greater at the upper end and near the anchored end of the
umbilical cable but the degree of bending at other positions of the umbili-
cal cable is gentle. The distribution of the standard deviation of the bending
moment of the umbilical cable along the length of the cable shows that the
standard deviation of the bending moments at the top and the anchored
ends are larger, indicating that the bending is more severe and frequent at
the top and anchored ends; the trend of the curve of the standard devia-
tion curve of the bending moment near the upper end is more than that
of the anchored end, which indicates that the increment and attenuation
of the unit length bending moment near the upper end is greater than
that of other parts of the umbilical cable. At the same time, with the
comparing observation of the distributions of the bending moment and
the curvature of the umbilical cable along the length direction, we have
found that the change of the curvature has some lag compared with the
change of the bending moment along the length, this phenomenon has
been shown in the figures of the mean curvatures and the mean bending
moment along the length direction; the results show that the curves of the
bending moment of the umbilical cable under different wave directions

are distinct but the curves of the curvature of the umbilical cable under
different wave directions appear to have a large degree of overlap. For a
specific wave direction, the curve is a segmented horizontal line parallel
to the horizontal direction of the cable length which has different discon-
tinuous curvature values and its distribution trend presents the charac-
teristic of ladder distribution except the two ends. At the same time, it
is necessary to take some measures to prevent excessive bending in the
zones where the maximum bending moment mutation is obvious in the
moment curve, such as the first end, the anchored end, and the position at
500 m along the cable length direction. For example, for the top end, the
effect of impact and alternating loads on the bending can be mitigated by
the addition of bellmouths, while for the bottom end, it can be strength-
ened by means of a combination of a limiter and a riprap. It can be found
that where the bending moment and the curvature are larger where the
standard deviations are also larger, indicating that for the umbilical cable,
the portion where the bending is relatively large is also the zone where the
degree of the fluctuating amplitude of variation of bending is more severe
and frequent see Figures 13.8-13.10.
As the minimum value of the curvature and the bending moment of the
riser are respectively 0, so it doesn’t need to show them in Figure 13.11.
But, the distribution of the curvature and bending moment along the
length direction of the riser is not all 0. The reason for this phenomenon
is that the load generated by the motions of the upper platform can be
effectively released to a certain extent by the free constraint of the riser, and
then every position of the riser can be restored after being bent. In contrast
to the riser, because both ends of the umbilical cable are always in a state of
0°
30° 0°
0.0025 60°
0°
0.00045 30°
0.0012 30°
The Minimum Curvature/(rad/m)
60°
The Maximum Curvature/(rad/m)
90° 0.0011 60° 0.00040 90°

120°
The Mean Curvature/(rad/m)
0.0020 90° 120°

150° 0.0010 120° 0.00035 150°
180° 0.0009 150° 180°
0.0015 0.0008 180° 0.00030
0.0007 0.00025
0.0010 0.0006 0.00020
0.0005
0.0004 0.00015
0.0005 0.0003 0.00010
0.0002 0.00005
0.0000 0.0001
0.00000
0.0000 0 100 200 300 400 500 600 700
−100 0 100 200 300 400 500 600 700 800 0 100 200 300 400 500 600 700
Arc Length of Umbilical/m
Figure 13.8 The distribution of the curvature of the umbilical cable under interference.
0°
0.0005 30°
The Standard Deviation of the Curvature/(rad/m)

60°
90°
0.0004 120°
150°
180°
0.0003
0.0002
0.0001
0.0000
0 100 200 300 400 500 500 700
Arc Length of Umbilical/m
Figure 13.9 The distribution of standard deviation of the curvature of the umbilical cable
under interference.
0° 0°
The Maximum Bend Moment/(KN.m)
30° 30°
60° 550 60°
The Mean Bend Moment/(KN.m)
1000 90° 500 90°

120° 120°
150° 450 150°
800 180° 400 180°
350
600 300
250
400 200
150
200 100
50
0 0
0 100 200 300 400 500 600 700 0 100 200 300 400 500 600 700
0° 0°
30° 30°
The Minimum Bend Moment/(kN.m)
220 60° 60°

The Standard Deviation of the Bend
90° 250 90°

200 120° 120°
180 150°
200 150°
160 180° 180°
Moment/(kN.m)
140 150
120
100 100
80
60 50
40
20 0
0
0 100 200 300 400 500 600 700 −100 0 100 200 300 400 500 600 700 800
Figure 13.10 The distribution of the bending moment of the umbilical cable under
interference.
0° 0°
30° 30°
0.0020 0.0005
The Maximum Curvature/(rad/m)
60° 60°
0.0018 90° 90°
The Mean Curvature/(rad/m)

120° 120°
0.0016 150° 0.0004 150°
180° 180°
0.0014
0.0012 0.0003
0.0010
0.0008 0.0002
0.0006
0.0004 0.0001
0.0002
0.0000 0.0000
0 100 200 300 400 0 100 200 300 400
Arc Length of Riser/m Arc Length of Riser/m
0°
30°
0.00035
The Maximum Bend Mument/(kN.m)

60° 0°
90°
The Standard Deviation of the
1000 30°
0.00030 120° 60°
150° 90°
180° 120°
0.00025
Curvature/(rad/m)
800 150°
180°
0.00020
600
0.00015
0.00010 400
0.00005 200
0.00000
0
0 100 200 300 400 0 100 200 300 400
0°
240 30° 0°
The Mean Bend Moment/(kN.m)
30°
220 60°
The Standard Deviation of the
90° 140 60°

200 120° 90°
Bend moment/(rKN/m)
120 120°
180 150° 150°
160 180° 180°
100
140
120 80
100 60
80
60 40
40
20 20
0 0
0 100 200 300 400 0 100 200 300 400
Figure 13.11 The distribution of the curvature and bending behavior of the riser under
interference.
constraint, the upper load generated by the motion of the platform cannot
be effectively released, so once the bending near the upper end and the
anchored end of the umbilical cable occurs, there is less room for recovery.
At the same time, it is found that the bending of the riser is more serious
and the bending moment is larger, which indicates that the bending of the
riser is more severe and frequent. With the comparative observation of
mean value distribution of the bending moment along the length direction
of the umbilical cable and the riser, it is different from the distributions of
the bending moment for the umbilical cable that other than 0° and 180°,
the distributions of the mean bending moment of the riser under different
wave directions along the length direction of the cable all have three peaks
(occurred at 50, 300, 400 m from the upper end), but there is no obvious
peaks of the mean bending moment of the umbilical cable under different
wave directions except the anchored end and the upper end; the causes of
this phenomenon is: the fixation at both ends of the umbilical cable makes
it more easily transfer the bending moment but also has a limit function
in the middle area of the bending of the umbilical cable; in sharp contrast
to the umbilical cable, as the lower end of the riser is a free end, with the
waves of ups and downs, it is more prone to drift with waves, so that there
is a greater degree of flexion at another position besides the upper end
and the bottom end. It is further observed that there are only two peaks
of the mean value of the bending moment of the riser along the length
direction under 0° and 180°wave directions; because of the layout relation
between the two pipelines and the wave angle of 0° and 180° so that the
collision between the riser and the umbilical cable is the frontal collision,
and the collision makes the bending of the riser cling to the umbilical cable
in some degree, as the umbilical cable is fixed at both ends, which acts as a
hindrance on the swinging and bending of the riser, so that a constraint is
imposed on the bending of the riser, as a result, the degree of curvature in
the intermediate zone of the riser is greatly reduced.
13.5 Conclusion
1. The umbilical cable and riser don’t collide with each other
in the 0- to 100-m length range, the reason for this phenom-
enon is that the fixation and confinement of the upper part
of the platform for the umbilical cable and the riser is dom-
inant in this region; with the increase in length and depth,
the flexibility of the riser and umbilical cables is beginning
to be manifested, the occurrence of different degrees of
bending and swing of the umbilical cable and riser begins to
appear, as the free end of the riser is unrestricted, the impact
of the wave causes the riser to start the lateral swing, which is
somewhat similar to the waving of the whip, which, in turn,
makes its spatial structure and configuration constantly
changing; this motion is the most intense at the free end.
2. There is a certain similarity between the clashing force curve

geometrical shapes of the umbilical cable and the riser along
the length direction, which indicates that the range of the
interference occurs in the range of the riser length, no col-
lision happened outside the length of the riser; the geomet-
rical shapes of the umbilical cable and the riser have some
similarity, which indicates that the trends of the change of
the clashing force have similar distribution and variation in
the time and along the length direction.
3. The form of collision for the umbilical cable and the riser
is not the same under different wave directions; when it
is under 0° and 180° wave directions, the special layout
between the wave direction and the two pipelines makes the
umbilical cable and riser have a frontal collision, so the col-
lision position of the clashing force occurs more in advance
than other wave directions.
4. The upper end of the umbilical cable is in a periodic ten-
sion state, and the anchored end is in a state of periodic axial
compression, this periodic variation changes with different
wave directions make its tension amplitude will be a slight
change, but its period will not change; the effective tension at
the upper end of the riser is far larger than that of the upper
end of the umbilical cable; the effective tension of the umbil-
ical cable is in a linear decreasing trend from the upper end
to the anchored end.
5. With the distance from the free end closer, the sharp and
violent stretching action of the platform’s motion on the riser
is buffered by the damping action of the water, which makes
the sharp change in the time domain gradually approaching
to a constant load.
6. The degree of bending is greater at the upper end and near
the anchored end of the umbilical cable but the degree of
bending at other positions of the umbilical cable is gentle;
the standard deviation of the bending moments at the top
and the anchored ends are larger, indicating that the bending
is more severe and frequent at the top and anchored ends;
which indicates that the increment and attenuation of the
unit length bending moment near the upper end is greater
than that of other parts of the umbilical cable.
7. The curvature has some lag compared with the change of the
bending moment along the length, the curves of the bend-
ing moment of the umbilical cable under different wave
directions are distinct but the curves of the curvature of the
umbilical cable under different wave directions appear to
have a large degree of overlap; for a specific wave direction,
the curve of the curvature for the umbilical cable is a seg-
mented horizontal line parallel to the horizontal direction of
the cable length which has different discontinuous curvature
values and its distribution trend presents the characteristic
of ladder distribution except the two ends.
8. The load generated by the motions of the upper platform can
be effectively released to a certain extent by the free con-
straint of the riser, and then, every position of the riser can
be restored after being bent; because both ends of the umbil-
ical cable are always in a state of constraint, the upper load
generated by the motion of the platform cannot be effec-
tively released, so once the bending near the upper end and
the anchored end of the umbilical cable occurs, there is less
room for recovery; the fixation at both ends of the umbilical
cable makes it more easily transfer the bending moment but
also has a limit function in the middle area of the bending of
the umbilical cable; in sharp contrast to the umbilical cable,
as the lower end of the riser is a free end, with the waves of
ups and downs, it is more prone to drift with waves, so that
there is a greater degree of flexion at another position besides
the upper end and the bottom end. It is further observed that
there are only two peaks of the mean value of the bending
moment of the riser along the length direction under 0° and
180° wave directions; because of the layout relation between
the two pipelines and the wave angle of 0° and 180°, so that
the collision between the riser and the umbilical cable is the
frontal collision, and the collision makes the bending of
the riser cling to the umbilical cable in some degree, as the
umbilical cable is fixed at both ends, which acts as a hin-
drance on the swinging and bending of the riser, so that a
constraint is imposed on the bending of the riser, as a result,
the degree of curvature in the intermediate zone of the riser
is greatly reduced.
References
1. Fachri P, Nasution, SveinSævik, et al. (2014). Finite element analysis of the
fatigue strength of copper power conductors exposed to tension and bending
.loads. International Journal of Fatigue. 59: 114–128.
2. Witz JA, Tan Z. (1992). On the axial torsional structural behavior of flexible
pipes, umbilicals and marine cables. Marine Structure. 5(2–3): 229–249.
3. Custodio AB, Vaz MA. (2002). A nonlinear formulation for the axisymmet-
ric response of umbilical cables and flexible pipes. Applied Ocean Research.
24: 21–9
4. Sævik S, Bruaseth S. (2005). Theoretical and experimental studies of the
axisymmetric behaviour of complex umbilical cross sections. Applied Ocean
Research. 27(2): 97–106.
5. Shunfeng Gong, PuXu, Sheng Bao, et al. (2014). Numerical modelling on
dynamic behaviour of deep water S-lay pipeline. Ocean Engineering. 88:
393-408.
6. F. Trarieux, G. J. Lyons, M. H. Patel. (2006). Investigations with a bandwidth
measure for fatigue assessment of the Foinaven dynamic umbilical including
VIV. Engineering Structures. 28: 1671–1690.
7. WeidongRuan, YongBai, Peng Cheng. (2014). Static analysis of deepwater
lazy-wave umbilical on elastic seabed. Ocean Engineering. 91: 73–83.
8. Yong Bai, Yutian Lu, Peng Cheng. (2015). Analytical prediction of umbilical
behavior under combined tension and internal pressure. Ocean Engineering.
109: 135–144.
9. Frédéric Muttin. (2011). Umbilical deployment modeling for tethered UAV
detecting oil pollution from Ship. Applied Ocean Research. 33: 332–343.
10. Orcina. (2014). OrcaFlex manual. (http: //www. orcina. com)
11. Costello G A, Philips J W. (1976). Effective modulus of twisted wire cables.
Journal of the Engineering Mechanics Division. 102(1): 171–181.
14
Fatigue Analysis of Flexible Riser
14.1 Introduction
In the deep-sea environment, flexible pipe has the advantages of strong
adaptability and weak coupling with the platform. Therefore, non-
bonded flexible pipes are mainly used at depths of 2,000 m, but its appli-
cation has even exceeded depths of more than 3,000 m. More and more
severe marine environmental conditions on the flexible pipe fatigue
design also put forward higher requirements. In addition, some of the
early installation has been completed in the installation of flexible pipe is
still part of the service. Some current marine environmental conditions
have exceeded the original environmental condition design standards.
In such a marine environment, the flexible pipe may be approaching, or
has reached, its ultimate fatigue life. It is necessary to evaluate its fatigue
life again and decide whether to continue using it. The fatigue life of the
flexible pipe is a key measure of its fatigue performance, which directly
determines the safety and reliability of the long-term service in deep-
water environments. Due to the complexity of the flexible riser struc-
ture and the marine environment, and taking into account the economics
and safety of ocean engineering, the stress fatigue analysis of flexible ris-
ers requires a safety factor of at least 10, according to the API Spec 17J
[1] specification. In order to predict the fatigue life of the flexible pipe
effectively, it is necessary to establish a reasonable fatigue analysis model
of the flexible pipe to provide the theoretical basis for the fatigue design
of the flexible pipe.
In this chapter, the nonlinear finite element software OrcaFlex [2] is
used to establish the global model of the flexible riser. The influence of
random wave, ocean current, and platform motion is taken into account.
Nonlinear finite element method is used to obtain the tension and curva-
ture. Then, based on the local model of flexible pipe under axial symmetry
and bending load, the load response time course of the flexible pipe is con-
verted into the stress response time of the tensile armor. Finally, the rain
295
flow counting method is used to calculate the stress cycle, and the fatigue
damage is obtained by combining the S-N curve to form a theoretical
method to calculate the fatigue life of the flexible pipe. Taking the 4-inch
diameter flexible pipe with 800-m depth as an example, the fatigue life
of the spiral strip of the flexible pipe is analyzed, and the influence of the
average stress, the friction coefficient, and the ellipticity is also analyzed.
14.2 Fatigue Failure Mode of Flexible Riser

The fatigue failure of flexible risers occurs mainly in the tensile armor
layer and the compressive armor layer. The tensile stress caused by plat-
form movement and wave action causes fatigue failure. The compressive
stress caused by the pressure change, which is caused by the pressure
armor layer, also causes fatigue failure. Among them, the compression
armor layer and the skeleton layer are large angle helical structures, mainly
to provide radial stiffness but does not provide axial stiffness and bending
stiffness. In the design process, they do not bear the tension and bending
moment, so the alternating stress caused by the alternating load is lim-
ited. The tensile armor layer is made of a high-strength steel strip with
a rectangular cross section wound at a helical angle between 20° and
55°. Under the bending load, the spiral strip in the tensile armor layer will
produce high local stress, so it is the most prone to stress fatigue failure of
the components [3].
In the marine environment, the following three fatigue failure modes
may occur for the spiral strips in the tensile armor layer:
1) The tensile strength of the spiral strip is caused by the alter-

nating load (axial tensile and bending moment), which leads
to fatigue damage of the spiral strip.
2) In the flexible riser and platform connection, due to the
greater axial tension and internal pressure, the tensile armor
layer and the adjacent layer will have a greater contact
pressure, and then under the cyclic load, the spiral will wear
to failure due to contact wear.
3) Sheath damage to the flexible riser will result in corrosion of
the tensile layer which is caused by seawater fatigue failure.
The latter two are more complex and difficult to simulate, so in marine
engineering, these are usually considered as the reduction of the material
fatigue performance.
Fatigue Analysis of Flexible Riser 297
In summary, the fatigue analysis of this chapter mainly focuses on the

stress fatigue failure of the tensile armor layer of the flexible pipe. JIP
(Joint Industry Project) has developed a reliable and effective method for
the fatigue analysis of the flexible pipes in combination with current indus-
try knowledge and engineering experience for the stress fatigue failure of
the tensile armor layer [4]. The guidelines have been applied in different
marine engineering projects, including the complex marine environment
of the North Sea project, the deep-sea project in Western Africa, and the
offshore project in Brazil. In addition, the guidelines are also widely used
in the professional analysis software of flexible pipes and local analysis
software. So, the guidelines have become a flexible pipe design and man-
ufacturing industry standards, mainly including the following four steps:
1) Collect and describe marine environmental load data.

2) The dynamic time domain analysis of the flexible pipe is car-
ried out, and the curves of the axial force and curvature of
the flexible pipe are obtained with time.
3) The curves of the axial force and the curvature in the flexible
pipe are transformed into the stress time-history curve of
the spiral strip in the flexible pipe by the local theoretical
model.
4) Calculate the fatigue life of flexible pipe tensile armor layer
based on rain flow method and Miner cumulative damage
criterion.
14.3 Global Model of Flexible Risers

14.3.1 Pipe Element
In recent years, numerical analysis and analysis tools have been devel-
oped rapidly and are widely used in practical engineering and scientific
research because of their rapid calculation speed and accurate simulation
results. In order to obtain the axial force and curvature of the flexible pipe
in the marine environment under the working conditions, the whole
finite element model of the pipeline is established based on the nonlinear
finite element software OrcaFlex. The pipeline is separated into a series of
pipeline units in the overall model, each consisting of a spring unit with
no mass and a node with mass at both ends. The spring unit simulates
the axial stretching, bending, and twisting of the pipe. The quality of the
pipeline, buoyancy and movement of all the characteristics of the focus
on the quality of the node, the pipeline by the axial force, bending, and
twisting and other loads will also act on the node.
Flexible pipe is a composite spiral wound structure made up of bearing
components such as skeleton and anti-armor layer and other high-strength
steel, and non-load-bearing components such as anti-wear and outer jacket
and other polymer materials. In OrcaFlex flexible pipe is simplified as a
tube unit, and the structural performance of the flexible pipe cannot be
defined by the constitutive relation of the material. Therefore, it is gener-
ally necessary to impart tension, twist, and bending stiffness to the spring
element between the nodes. According to the flexible axis axisymmetric
theoretical model proposed, the tensile stiffness and torsional stiffness of
the flexible pipe are linear. The spiral strip in the flexible pipe tensile layer
will undergo relative slip under the bending deformation, which makes the
bending stiffness of the flexible pipe have obvious nonlinearity. Although
OrcaFlex supports non-linear input of bending stiffness, this leads to the
convergence problem of the model and increases the computational cycle.
Therefore, in the actual engineering, to ensure the convergence of the cal-
culation, the calculation cycle is shortened making the results conserva-
tive in the global analysis model only consider the bending stiffness of the
flexible pipe after slippage.
14.3.2 Bending Stiffener

Due to the influence of the offshore platform movement and the wave load,
excessive bending deformation usually occurs at the connection of the
flexible pipe and the platform. As shown in Figure 14.1, in order to prevent
the bending curvature of the flexible pipe at the connection beyond the
(a) Bending stiffener (b) Bell mouth
Figure 14.1 Bending stiffener and bellmouth.

design range, a bending stiffener or bellmouth is generally provided here.

The bell-shaped mouth is a steel curved surface member that effectively
prevents the flexible pipe at the connection from exceeding the curva-
ture of the curved surface. The bending stiffener is a tapered cross-section
made of polyurethane material. The bending stiffness of the flexible pipe at
the connection is significantly increased by placing the bending fit on the
outer surface of the flexible pipe at the connection. Installation of bending
stiffener and bell-shaped mouth cannot only effectively limit the curvature
of bending, but also improve the fatigue life of flexible pipe. In this chapter,
the bending stiffener is used as the component to strengthen the bending
stiffness at the suspension point of the flexible riser in the global model.
14.3.3 Sea Condition

The service life of flexible risers is generally 25 years, not only to resist
the maximum wave during use but also to consider long-term wave load
fatigue damage, which requires the study of the long-term distribution of
waves and long-term wave statistical distribution of the values.
Using the long-term wave observations of the ocean observing station, a
joint profile of the wave height and period can be obtained, as shown in
the scatter diagram of Table 14.1. The plane Hs × Tp is divided into a grid,
each grid records the number of occurrences of the corresponding sea con-
ditions during the wave observation, each of which is represented by a set
of (Hs × Tp), where Hs represents the effective wave height of the wave and
Tp represents the wave peak period. The wave spectrum can be changed
to wave characteristic period Tz, that is, each group of sea conditions can
also be used to represent (Hs × Tz).
Table 14.1 Wave scatter diagram.

Tp(s) 3 4 5 6 7 8 9 10 11
Hs(m) 4 5 6 7 8 9 10 11 12 …
0.00 0.50 N1 N2 N3 N4 N5 N6 N7 N8 N9
0.50 1.00 N10 N11 N12 N13 N14
1.00 1.50 N15 N16 N17
1.50 2.00 N18 ..
2.00 2.50
… …
Table 14.2 Stochastic wave load.

Waveparameter
Effective wave Wave peak
Load case height Hs(m) period T p (s) Probability %
1 1.6 7.5 23.86
2 1.6 9.0 14.88
3 1.8 5.3 22.94
4 2.1 6.5 14.16
5 2.6 7.5 14.10
6 2.8 9.0 10.06
It is an important step in the fatigue analysis of the flexible pipe to

extract the wave input parameter of the global model from the wave pat-
tern of the sea condition. Whether the selected wave parameter is reason-
able and directly affects the result of the global analysis of the flexible pipe.
In the present study, there are two general methods: one is the regular
wave method. Based on the Longuet-Higgins distribution, the wave
scatter plot which characterizes the long-term sea condition is discretized
as the wave pattern of the regular wave. The analysis time of each regular
wave is five to six wave cycles. The other is a random wave method, based
on the JONSWAP spectrum will be long-term sea wave pattern is broken
down into several random wave spectrums, each random wave analysis
time is generally more than 1,200 s.
Regular wave analysis of the length of time is the common method of
ocean flexible pipe or marine pipe fatigue analysis. The final fatigue anal-
ysis results will be conservative. However, with the rapid development of
computer performance, in order to avoid excessive analysis of flexible pipe
fatigue analysis, random wave method has been more widely used.
In this chapter, the fatigue life of the flexible pipe is obtained, and the
random wave method is selected for the sea condition. Table 14.2 shows
the random wave load characteristics of a sea area in the South China Sea.
14.3.4 Platform Motion Response

In the ocean under the random waves, the floating platform will occur ran-
dom movement. Flexible pipe will be linked with the floating platform
y z
Platform motion
Wave x
Current
Water depth
Unbonded flexible pipes
Anchor point Touch down point

Seabed
Figure 14.2 Global configuration of flexible pipe.
during laying and operating conditions. Therefore, the random movement

of the floating platform will have a greater impact on the overall force and
deformation of the flexible pipe.
In order to consider the random motion of the floating platform, we
need to obtain the 6 degrees of freedom of the motion RAOs (Response
Amplitude Operator) along the three coordinate axes for linear motion of
surge, sway, heave, pitch, roll, and yaw that revolve around the axis. In the
global model, the input of RAOs data is taken as the boundary condition
of the top of the flexible pipe, so as to consider the influence of the random
motion of the floating platform on the whole force of the flexible pipe. As
shown in Figure 14.2, the Cartesian coordinate system xyz is established.
The wave direction is defined as trailing waves (0°) along the x-positive
direction, along the y-axis positive direction (90°), along the negative
direction of the x-axis (180°). RAOs calculate the wave angle from 0° to
180° and the interval is 45°. In this chapter, the motion response curve of
the top of the flexible riser is calculated by using the motion amplitude
response function of the nonlinear finite element software OrcaFlex and
the random wave load is shown in Table 14.2.
14.3.5 Time Domain Simulation Analysis

The dynamic analysis of the global model has the time domain method
and frequency domain method. Although the frequency domain method
is faster, it cannot deal with the problem of nonlinearity. Although the time
domain method is slower to calculate the geometrical, material, and load
nonlinearity of the structure problem, the rapid development of computers
has greatly increased the computing speed and the time domain method
has been greatly reduced. This reduction in computing time has resulted
in the time domain method being more widely used.
The global analysis of flexible pipes involves a large number of geomet-
ric nonlinearity, so this chapter uses time domain simulation analysis. The
whole analysis process is divided into two steps: the first step is the static
calculation to determine the equilibrium position of the pipeline to get
the initial configuration of the pipeline; the second part is the dynamic
calculation, the static calculation of the pipeline displacement and the
force condition as the initial conditions, the explicit integral method
is used to solve the internal force and deformation of the flexible pipe at
each time step according to the time increment step, and finally, the time
course response result of the axial force and curvature of the flexible pipe
is obtained.
14.4 Failure Mode and Design Criteria

The time-dependent response curves of the axial force and the curvature of
the flexible pipe can be obtained by the global model of the flexible pipe.
Under the combined load of axial force and curvature, the spiral strip in
the flexible pipe tensile armor layer will produce fatigue stress. In order to
calculate the fatigue stress, the load model of the flexible pipe is divided
into axisymmetric load and bending load, as shown in Figure 14.3.
14.4.1 Axisymmetric Load Model

The axisymmetric load includes axial force, axial torsion, internal pres-
sure, and external pressure load. The axial stress of the spiral strip under
the axisymmetric load can be obtained by the flexible axis axisymmetric
F T F T
M M
= +
Po Po
Pi Pi
Figure 14.3 Load model of flexible pipe.

e2
Oc e3
θ
x3
x2
Figure 14.4 Cross section of helical strip.
theoretical model. In addition, as shown in Figure 14.4, the spiral strip

in the flexible pipe tensile armor layer is generally rectangular in cross
section, and the variation of the curvature of the spiral strip itself at
the four corners of the section also causes the corresponding local bending
stress.
Therefore, under the axisymmetric load, the fatigue stress of the spiral
strip is
fa = E(εa + x2Δκ2 + x3Δκ3) (14.1)
where εa is the axial strain of spiral strip under axisymmetric load, x2

and x3 are the coordinates of the cross section of the spiral strip, Δκ2 and
Δκ3 are the changes of the curvature and the curvature of the spiral strip
under the axisymmetric load, respectively.
14.4.2 Bending Load Model

The stress of the spiral strip under the axisymmetric load is the initial con-
dition (consider the interlayer pressure caused by the axisymmetric load
between the tensile armor and the adjacent layer), and the bending load is
applied to calculate the bending fatigue stress. The slip mechanism of the
flexible pipe’s tensile layer makes the bending moment-curvature of the
flexible pipe appear obviously nonlinear, as shown in Figure14.5.
When the curvature of the flexible pipe is less than the critical curva-
ture, the spiral strip of the tensile armor layer does not slip, and the bend-
ing moment relationship of the flexible pipe remains linear. At this stage,
the flexural rigidity of the flexible pipe is also the largest for EIns. When the
bending deformation increases, the bending curvature of the flexible pipe
M
C
A
EIs
EIns
O κ
D
B
E
Figure 14.5 Bending hysteresis model of flexible pipe.
exceeds the critical curvature. The spiral strip begins to slip, the bending
moment-curvature relation of the flexible pipe starts to become nonlinear
and then becomes linear again. The bending stiffness of the flexible pipe is
also from the maximum gradually diminishes the last remain unchanged
for EIns. In order to simplify the fatigue stress calculation, the bend-
ing hysteresis curve with bilinear curve will be used in this chapter, as
Under random waves and during the process of loading and unloading,
the bending deformation of a flexible pipe will be constantly changing.
As shown in Figure 14.5, if the bending curvature of the next step is less
than the critical curvature, the bending stiffness of the flexible pipe is EIns
(path AB). If the critical curvature is exceeded, the bending stiffness
of the flexible pipe is EIns (path AC). If the curvature of the flexible
pipe begins to decrease, when the reduced amplitude is less than the crit-
ical curvature the spiral strip will be in a non-slip state and the bending
stiffness of the flexible pipe is EIns (path CD). When the reduced ampli-
tude is greater than the critical curvature, the spiral strip enters the slip
state and the flexural rigidity of the flexible pipe becomes EIns (path DB).
Under the influence of bending load, the change of the curvature of the
spiral strip at the four corners will cause the corresponding local bending
stress. Therefore, the fatigue stress of the spiral strip under the bending
load is
fb = E(εb + x2Δκ2 + x3Δκ3) (14.2)
where εb is the axial strain of spiral strip under axisymmetric load, Δκ2
and Δκ3 are the changes of the curvature and the curvature of the spiral
strip under the bending load, respectively.
14.5 Calculation Method of Fatigue Life

of Flexible Riser
When the structure is subjected to the cyclic load, local permanent dam-
age in structure will occur. The damage will gradually accumulate more
damage over time until destruction, that is, fatigue damage. In order to
calculate the fatigue damage, it is necessary to identify the stress cycles of
the fatigue damage. Then, the relationship between the magnitude of the
stress range and the number of actions, that is, the distribution of the stress
range must be determined. Therefore, the random fatigue load is required
to be counted.
14.5.1 Rainflow Counting Method

There are several ways to count the random fatigue load: peak counting
method, cross-mean peak counting method, cross-level counting method,
range counting method, and rain flow counting method. The rain flow
counting method has a certain mechanical concept as a well-known count-
ing method. In ocean engineering, the stress time history data are pro-
cessed by the rain flow counting method to identify the stress cycle, and
obtain the corresponding stress range and average stress.
14.5.2 S-N Curve

The relationship between S and N is commonly used in engineering to
indicate the fatigue strength of the structure. S is the range of alternating
stress, and N is the number of stresses required to reach the damage under
the constant amplitude of the stress range S, also known as fatigue life.
In the fatigue analysis, the median S-N curve is generally used. The
median S-N curve is usually obtained by means of a group test. Several
different stress ranges are selected. The specimen is tested at each stress
range. The test specimen is analyzed statistically and fitted with the curve.
The value of the fatigue life data points yields the median S-N curve. The
long-term fatigue test data study found that in the double logarithmic
coordinate system, the median S-N curve is usually close to the straight
line. For ease of analysis and use, the expression for the median S-N curve
is generally
log N = log a − m log ∆σ (14.3)

where a and m are two parameters that need to be determined by the test,
Δσ is the stress range.
In order to provide a large tensile stiffness, flexible pipe generally
use low-alloy high- strength steel, the ultimate strength is between 700
to 1,400 MPa. According to the DNV-RP-C203 specification, when the
yield stress of high-strength steel is greater than 500 MPa and the surface
roughness Ra = 3.2 or less, the fatigue life can be calculated using the fol-
lowing equation [5]:
log N = 17.446 – 4.70log Δσ (14.4)
The fatigue test is usually carried out at a moderate stress range. The
critical point of the applicable range of Eq. (14.4) in the dry environment
is Δσ = 235 MPa, N = 2 × 106, as shown in Figure 14.6.
When the stress range is less than 235 MPa, it is considered that
there is a definite fatigue limit. The S-N curve of the high life area is rep-
resented by a horizontal straight line and becomes a median S-N bilinear
curve. If the high-strength steel is in seawater but there is electrode pro-
tection, when the fatigue life exceeds 2 × 106, the S-N relationship of the
high life area is the same as that of the previous curve, and the S-N curve
is a single straight line.
In the long-term use of flexible pipes, seawater and other gases are gen-
erally present between the inner watertight layer and the outer watertight
1000
Air
Stress range (MPa)
Seawater with
100 cathodic protection
10
10000 100000 1000000 10000000 100000000
Number of cycles
Figure 14.6 S-N curve for high strength steel.

layer. Therefore, in this chapter, the gap between the watertight layer and
the outer watertight layer is considered to be a wet environment, that is, the
use of SN curve of a single linear form.
14.5.3 Miner’s Linear Cumulative Damage Theory

The fatigue failure of the flexible pipe tensile armor layer under alternating
stress is a process of fatigue damage accumulation, and every alternating
stress can cause some fatigue damage. The commonly used fatigue cumu-
lative damage model is based on the Miner linear cumulative damage the-
ory. Miner linear cumulative damage theory is easy to use and has a certain
degree of reliability, so it is widely used in countries of the ocean norms.
Miner’s linear cumulative damage theory holds that the total amount
of damage is the sum of the damage components under the stress range
when the structural fatigue failure occurs under the multi-stage constant
amplitude alternating stress.
d= ∑di
i (14.5)
where d is the total amount of damage in the event of fatigue damage, and
di is the amount of damage in the stress range of Si.
Miner’s linear cumulative damage theory holds that the ratio of the dam-
age component to the total amount of structural damage to damage at a
stress range is equal to the ratio of the number of cycles of the stress range to
the number of cycles of structural fatigue failure under a single stress range:
di ni
= (14.6)
d Ni
where ni is the number of cycles under the stress range Si, and Ni is the
number of cycles of structural fatigue failure under the stress range Si.
The following formula is usually defined as the fatigue cumulative dam-
age of the structure:
D= ∑ Nn
i
i
i
(14.7)
In the whole dynamic time domain analysis of the flexible pipe, the
dynamic analysis time is Td, and the fatigue cumulative damage degree
of the spiral strip of the tensile armor layer is D. When the fatigue cumu-
lative damage degree of the spiral strip reaches 1, the spiral strip reaches
its fatigue life, and the expression is
Td
Tf = (14.8)
D
14.5.4 Modification of Average Stress on Fatigue Damage

The S-N curve is usually obtained by means of a group test, where the
mean value of the alternating stress is usually zero. However, in the
fatigue life analysis of flexible pipe tensile armor layer, the flexible pipe
is subjected to the shaft pull load, so the average value of the alternat-
ing stress in the tensile armor layer is always positive. If the stress ratio is
within the applicable range of the S-N curve, no fatigue damage correc-
tion is required. If the stress ratio is not within the applicable range of the
S-N curve, the fatigue damage needs to be corrected. In order to consider
the effect of average stress on fatigue life, the stress ratio parameter Rσ is
defined [6]:
σ min
Rσ = (14.9)
σ max
From the engineering experience, it can be seen that the stress ratio of
the spiral strip in the flexible pipe tensile armor layer is between 0.1 and
0.5, and the Goodman correction theory and Gerber correction theory are
often used:
σa σm
+ =1 (14.10)
σe σu
2
σa σ 
+ m  =1 (14.11)
σe  σu 
σa
σe Ge
rbe
r
Go
od
ma
n
σu σm
Figure 14.7 Comparison between Goodman’s and Gerber’s theory.
where σa is the stress amplitude, σm is the average stress, σa is the corrected

stress amplitude, and σa is the ultimate tensile strength.
As shown in Eq. (14.10), Goodman’s theory assumes a linear relation-
ship between stress amplitude and mean stress, while Gerber argues that
the two are parabolic relationships, as shown in Figure 14.7.
In this chapter, the corrected stress amplitude σe is obtained by the aver-
age stress correction using the Gerber correction theory, and the cor-
rected fatigue life is obtained by substituting Figure 14.7.
14.6 Example of Fatigue Life Analysis

of Flexible Riser
Fatigue life analysis is a key factor in the fatigue design of flexible pipe.
The fatigue life of 4-inch diameter riser connected to FPSO is analyzed
by taking 800 m of water depth in an area of the South China Sea. The
geometrical parameters and material parameters of the flexible pipe are
shown in Table 14.3 [7]. The axial tensile stiffness of the flexible pipe and
the bending stiffness of the complete slip stage are calculated by the axially
symmetric load model and the bending theory model.
Flexible pipe in the global analysis is the use of free suspension of the
catenary structure, the initial suspension angle of 5.5°, as shown in Figure
14.2. The three-dimensional spatial coordinates (x, y, z) of the suspension
point and the anchor point of the seabed are (0, 0, 0) and (−600, −800, 0),
Table 14.3 Input parameters of flexible pipe.

Unit length Tension Bending stiffness
ID/OD (mm) weight (kg/m) stiffness (kN) (kN·m2)
101.6/154.0 83.20 1.79 × 105 8.11
respectively. The riser points are fixed, and the anchor points of the seabed
are hinged. In this numerical calculation model, 1,200-m long catenary
riser is divided into 795 differential units for mechanical analysis.
Figure 14.8 shows the plane of the bending stiffener, the figure shows
the detailed geometric parameters of the bending stiffener. This chapter
assumes that the material of the bending stiffener is linear, and the
modulus of elasticity usually takes 10% of the strain corresponding to
the secant modulus, which is 150 MPa.
This chapter assumes that the waves are incident in the 0° direction
along the floating platform. The JONSWAP random wave spectrum is used
as the input parameter for the global analysis model. A total of six sea con-
ditions are shown in Table 14.2. The velocity of surface current is 0.8 m/s
and the velocity of seabed current is 0 m/s, and its propagation direction
is along the positive x-axis direction. The normal stiffness of the seabed is
6 × 103 N/m2, and the friction coefficient between the seabed and the pipe-
line is simulated by the coulomb friction model. The friction coefficient is
0.55. The hydrodynamic coefficients in the model are shown in Table 14.4.
The dynamic time domain analysis is carried out using the nonlinear
finite element software OrcaFlex. The dynamic time domain analysis time
of the global model is 1,200 s. The static calculation of the axial force and
curvature distribution of the flexible riser is shown in Figure 14.9 before
dynamic time domain analysis. Due to the effect of the flexible pipe, the
maximum axial force of the riser is at the suspension point and the size is
563.55 kN. As the depth of water gradually increases, the axial force of the
flexible pipe is gradually reduced to 43.08 kN, as shown in Figure 14.9a.
0.2m 1.7m
0.325m 0.154m
Figure 14.8 Schematic of bending stiffener.
Table 14.4 Input parameters of hydrodynamic coefficient.

Normal Tangential Normal Tangential
drag force drag force added mass added mass
coefficient coefficient coefficient coefficient
1.2 0.008 1.0 0.008
600 0.06
Curvature (rad/m)
400 0.04
Tension (kN)
0.02
200
0.00
0
0 200 400 600 800 1000 1200 0 200 400 600 800 1000 1200
Length (m) Length (m)
(a) Tension (b) Curvature
Figure 14.9 Results of global analysis.
The curvature distribution of the riser is shown in Figure 14.9b. The

maximum curvature at the top of the riser is 0.06 m−1. With the extension
of the riser, the curvature decreases rapidly and approaches zero. When
the riser approaches the seabed, the curvature begins to increase rapidly
and reaches a maximum of 0.01 m−1 at the length of 899 m. From Figure
14.9b, it can be seen that the curvature of the suspension point is greater
than the bending curvature of the touch down point. The bending curva-
ture of the other positions of the riser is relatively small compared to the
suspension point and the touch down point.
Under the combination of the random wave, ocean current, and floating
platform, the axial force and bending curvature of the riser point and
touch down point fluctuate within a certain range. Taking the random
wave with the effective wave height of 2.8 m and the peak period of 9.
0 s as an example, the axial force and bending curvature of the flexible
riser point and the touch down point are obtained by dynamic analysis,
as shown in Figures 14.10 and 14.11. The axial force of the riser point
600 70
580 65
Tension (kN)
Tension (kN)
560 60
540 55
520 50
0 200 400 600 800 1000 1200 0 200 400 600 800 1000 1200
t (s) t (s)
(a) Hang-off point (b) Touch down point
Figure 14.10 Time history response of pipe tension.

0.15 0.013
0.10 0.011
κ (m–1)
κ (m–1)
0.05 0.009
0.00 0.007
0 200 400 600 800 1000 1200 0 200 400 600 800 1000 1200
t (s) t(s)
Figure 14.11 Time history response of pipe bending curvature.
fluctuates at 563.55 kN, and the corresponding amplitude reaches 59.46

kN. The axial force of the stand-off point fluctuates around 61.87 kN,
and the corresponding amplitude reaches 12.68 kN. By comparison, we
can see that the tension and fluctuation at the riser point are larger than
that of the touch down point. It can be seen from Figure 14.11 that the
bending curvature of the riser point reaches 1.11 × 10−1 m−1 and the
bending curvature of the touch down point reaches 3.10 × 10−3 m−1.
Similar to tension, by comparing the two we can see that the bending cur-
vature and fluctuation of magnitude at the riser point are also much larger
than the touch down point.
It can be seen from Figures 14.12a and 14.12b that the stress means val-
ues at the suspension point caused by the axisymmetric load in the inner
and outer spiral strips of the flexible riser are 372.40 and 314.02 MPa,
the volatility amplitude reached 18.21 and 15.35 MPa, respectively. It
can be seen from Figures 14.12c and 14.12d that the stress amplitude
caused by the bending load is near zero, and the stress amplitude caused
by the bending load in the inner spiral and outer spiral are 109.36 and
36.16 MPa. It can be seen from the comparison that the stress amplitude
caused by the bending load of the suspension point is larger than the stress
amplitude caused by the axisymmetric load, and the stress amplitude of the
inner layer is larger than the stress amplitude of the outer layer. In addi-
tion, the stress amplitude caused by the bending load at the suspension
point is relatively fixed compared to the magnitude of the stress caused by
the axisymmetric load. This is because the bilinear curve is used to sim-
ulate the hysteresis effect of the flexible pipe in the bending theoretical
model. Therefore, when the variation of the bending curvature is equal to
390 390
380 380
fa (MPa)
fa (MPa)
370 370
360 360
350 350
0 200 400 600 800 1000 1200 0 200 400 600 800 1000 1200
t (s) t (s)
(a) Stress caused by axisymmetric (b) Stress caused by axisymmetric
loads in the inner layer loads in the outer layer
80 80
40 40
fb (MPa)
fb (MPa)
0 0
−40 −40
−80 −80
0 200 400 600 800 1000 1200 0 200 400 600 800 1000 1200
t (s) t (s)
(c) Stress caused by the bending (d) Stress caused by the bending
load in the inner layer load in the outer layer
Figure 14.12 Time history response of stress of helical strip at top point.
or greater than the critical curvature, the stress amplitude caused by the
bending load will reach the maximum.
The stress response time of the flexible riser’s tensile layer is obtained.
The relationship between the average stress, the stress range, and the
number of actions is obtained by the rain flow counting method. Then,
the Gerber correction theory is used to modify the average stress.
According to the high-strength steel S-N curve in the DVN-C203
standard, combined with Miner linear cumulative damage theory, the
fatigue life of flexible pipe tensile layer can ultimately be obtained, as
shown in Figure 14.13. Fatigue life must consider a factor of safety
of 10.
Figure 14.13 shows the distribution of the fatigue life ofthe flexible
pipe’s tensile armor layer spiral strip along the ring. The minimum fatigue
life of the outer spiral strip at the suspension point of the flexible riser is
1E9 Inner Inner

Outer Outer
Fatigue lifes(years)
Fatigue lifes(years)
1E14
10000 1E9
0.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5 0.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5
θ θ
Figure 14.13 Results of global analysis.
6.34 × 103 years, and the minimum fatigue life of the inner spiral strip is
30.4 years. The fatigue life of the inner spiral strip is 99.52% lower than that
of the outer spiral strip. Therefore, in the suspension point of the flexible
pipe area, the inner spiral strip is more prone to fatigue damage than
the outer spiral strip. The fatigue life of the flexible standpoint is shown
in Figure 14.13b. The minimum fatigue life of the touch down point is
2.08 × 106 years, which is much greater than the minimum fatigue life
at the point of suspension. In addition, there was no significant differ-
ence in the fatigue life between the inner and the outer spiral strips at
the touch down point. From the above results, it can be seen that under
the catenary configuration, the most prone to fatigue damage is the
inner layer of the tensile layer at the hanging point, the fatigue life is
30.4 years.
References
1. API 17J. Specification for unbonded flexible pipe. American Petroleum
Institution.
2. Orcina. OrcaFlex User Manual, version9.7a[M]. Cumbria, UK: 2014.
3. Sævik, S. On stresses and fatigue in flexible pipes[D]. Norway: Norwegian
University of Science and Technology 1992.
4. Grealish, F.,Smith, R.,Zimmerman, J. New industry guidelines for fatigue
analysis of unbonded flexible risers[C]. In Offshore Technology Conference,
Houston, USA, 2006.
5. DNV-RP-C203. Fatigue Analyisis of Flexible Riser-Effect of Mean
Stress Correction Procedures[M], Norwegian University of Science and
Technology.
6. Zhao B, 2013, Fatigue Analysis of Flexible Riser -Effect of Mean Stress

Correction Procedures[M], Norwegian University of Science and
Technology.
7. Lu Y, 2017, Research on cross-sectional mechanical model and fatigue life
analysis of deepwater flexible risers[D]. Zhejiang University.
15
Steel Tube Umbilical and Control Systems
15.1 Introduction
15.1.1 General
The plan for umbilical delivery typically includes an overall schedule plan
for the following:
• Feasibility study;
• Umbilical specifications and request for quotation (RFQ);
• (Fatigue and other) qualification tests–specifications and
execution;
• Long-lead item procurement;
• Bid evaluation;
• Supplier selection;
• Project sanction and umbilical procurement;
• Detailed umbilical design and analysis by the supplier;
• 3rd party design verification by an analysis specialist;
• Prototype qualification tests;
• Umbilical manufacturing (typically requires a period of 1 year);
• System integration test;
• Umbilical delivery to host vessel;
• Commissioning;
• System start-up;
• Project management, QA/QC.
This chapter provides an overview of steel tube umbilical design. The

first task in an umbilical design is to size the umbilical cross-section. Some
design checks and calculations that are carried out for initial steel tube
sizing and cross-section design are described in Sections 15.3 and 15.4.
This part of the design process is followed by an extreme wave analysis,
which is described in Section 15.5. Following the feasibility study, a more
detailed design of the umbilical is carried out. This involves a detailed
317
fatigue analysis, aspects of which are discussed in Sections 15.6 to 15.8.

The fatigue analysis shall cover all sources of fatigue damage such as man-
ufacturing, in-place wave-induced stresses, installation, vortex-induced
vibrations (VIVs), and vortex-induced motions (VIMs).
One of the first papers dealing exclusively with steel tube umbilical design
was “Metal Tube Umbilicals–Deepwater and Dynamic Considerations” pre-
sented in OTC in Swanson et al. (1995). Another useful publication for fur-
ther information is ISO 13628-5 which is used as the standard for umbilical
design and operation.
The most recognized subsea umbilical manufacturers are Nexans, DUCO,
Oceaneering Multiflex, and Kvaerner Oil Products. Figure 15.1 shows an
umbilical cross-section, courtesy of Nexans (Bjornstad, 2004).
15.1.2 Feasibility Study

The project activities in a feasibility study phase include:
• Cross-section design and sizing;

• Analysis of extreme response, bend stiffener sizing, and riser
interference;
• Determination of umbilical azimuth angle, departure angle,
and lay route;
• Installation sequence and methods;
Figure 15.1 Umbilical cross-section (Bjornstad, 2004).

Steel Tube Umbilical and Control Systems 319
• Specification as part of RFQ;

• Umbilical delivery plan.
Umbilical design commences with a feasibility study, which involves
sizing the steel tubes that make up the umbilical. Tension-angle plots from
the extreme wave analysis are used to design an adequate bend stiffener or
bellmouth for the umbilical. A final aspect of feasibility study is an interfer-
ence analysis in order to ascertain that the umbilical does not collide with
adjacent risers, mooring lines or structures when under the influence of
extreme waves or currents. The main deliverables from the feasibility study
are the umbilical cross-section design, the required drag to diameter ratio
and azimuth angle derived from an interference analysis, as well as the
departure angle from extreme response analysis.
The umbilical procurement specification that goes with the RFQ docu-
ment includes:
• General requirements;
• Met-ocean data and vessel motion data;
• Umbilical requirements;
• Umbilical configuration;
• Umbilical components;
• Testing requirements;
• Failure analysis and rework;
• Quality control, testing, inspection, and QA/QC surveillance;
• Project management and documentation;
• Equipment packaging, shipping, and storage;
• Safety.
This procurement specification shall govern the design, materials, fab-
rication, inspection, and qualification of the umbilical. Umbilical suppliers
shall bid according to the RFQ specifications. The bids are then subject to
technical evaluation and price comparisons. The objective of the technical
evaluation is to identify any show stoppers (feasibility issues) and potentials
for contract deviations that may require extra cost and time period. Ideally
the technical quality for individual bids should also be ranked. When com-
paring prices, the technical quality and service quality as well as the prod-
uct reliability (simplicity) and durability should be given consideration.
15.1.3 Detailed Design and Installation

The most critical design challenge for an umbilical is usually fatigue. In
some designs it is necessary to iteratively design bend stiffeners such that
the fatigue design requirements are met. Another important fatigue

aspect to consider is fatigue damage due to VIVs and VIMs. The proce-
dure for analyzing VIV and VIM damage for an umbilical is similar to the
methods used for a SCR.
A detailed design shall be conducted by the umbilical supplier. Some
oil companies also require a third-party verification of the detailed design.
The design and design verification include a full set of dynamic analy-
sis such as: analyses of extreme response, interference, fatigue due to VIV,
VIM, wave loading and installations, etc.
We should be aware that the umbilical installation requires a large
installation vessel, and in certain scenarios, the availability for such a large
vessel may be an issue. This is particularly true for deep and ultra-deep
water umbilicals. The installation of flying leads and other components
may be conducted using smaller vessels.
15.1.4 Qualification Tests

The metal tubes, electrical cables, fiber optic conductors, and umbilical
section shall be subjected to tests as defined in the umbilical procurement
specifications. All component designs shall be qualified for their appropri-
ate requirements prior to manufacture.
In the past years, tube fatigue tests have been conducted in order to
qualify the S-N curves used for umbilical fatigue analysis. However, the
industry has now acquired enough confidence in the typical metals used to
construct the umbilical tubes. Hence, umbilical design now does not need
to include the documentation of the S-N curves used to qualify the metal,
unless special/new metals are used in the design.
15.2 Control Systems

15.2.1 General
The wells are controlled by hydraulic and/or electrical signals through an
umbilical, which also provides for the injection of chemicals.
The control umbilical is a bundled cable, connecting subsea equipment
with the topside facilities, containing all the required components to oper-
ate and serve the subsea equipment. Depending on the type of subsea con-
trol system being used, the control umbilical may contain required fluid
conduits (hoses), electrical power/signal conductors, chemical injection
lines, vent lines, and required spare lines. If the umbilical is subjected to
dynamic loads from wave action and currents, it may be equipped with
steel armoring. The composition of the control umbilical varies considerable
from one control system to the next, and it is determined primarily by the
requirements of the particular system being used.
15.2.2 Control Systems

The subsea production control system facilitates the operation of valves
and chokes on subsea completions and manifolds. Typically, five types of
control systems are used in the operation of subsea production systems,
those being:
• Direct hydraulic;
• Piloted hydraulic;
• Sequential hydraulic;
• Electro hydraulic;
• Multiplex electro hydraulic
The direct hydraulic control system is the least complex of the five alterna-
tives and has the fewest number of subsea components. Each subsea func-
tion requires a hydraulic flow path from the surface. Actuation of a valve on
the surface control panel results in pressurized fluid being routed through
a dedicated flow path to the selected subsea tree valve actuator. This system
with one line per subsea function is best suited for applications where the
control distance is relative short (actuation of a valve 4,000 m away from
the surface facility can take as long as three hours), and a limited num-
ber of subsea functions to be operated. As the number of subsea functions
increases, so does the outer diameter of the control umbilical and its cost.
Piloted hydraulic systems improve the response time by storing the
hydraulic pressure energy at the site. Actuators are then activated by send-
ing a hydraulic signal to a pilot valve, which opens and allows fluid from
the accumulated storage to activate the actuator.
The sequential hydraulic control system has, as with the piloted system,
an accumulator and control valves placed subsea. Control is then achieved
by sending a pressure signal to a sequence valve that is pre-set to operate
or shift at specific pressure levels. At this signal pressure level, the sequence
valve shifts and hydraulic fluid from the accumulators is routed to a group
of pre-selected gate valve operators. As signal pressure is increased in
a series of discrete steps, the sequence valve shifts and operates the next
group of pre-selected tree valves. This system is suitable for operating
at long distances and is cost effective at these distances due to the small
diameter umbilical. This system is normally configured to operate several

gate valves at a time placing the subsea tree in a particular operational
mode. The system can accommodate up to six modes. The disadvantage of
this type of system is the complex hydraulic circuitry and the inability to
operate each tree valve individually.
The electro hydraulic system is capable of controlling a limited number
of subsea valves very quickly at large distances. The system works, as with
the piloted and sequential system, by supplying hydraulic fluid at pressure
through a large supply line within the umbilical, which is accumulated in
a pressure vessel at the tree. A switch on the surface control panel directs
electrical current to a solenoid pilot valve. This valve shifts and directs
hydraulic fluid from the pressure vessel to the associated actuator. One
electrical signal conductor is required for each subsea valve or function to
be operated. As the number of valves to be operated increase so does the
size, cost, and complexity of the control umbilical.
The multiplex electro hydraulic system is capable of controlling large
numbers of subsea valves very quickly at large distances using a relative
small diameter control umbilical coupled to a multiplexer that normally
operates a large number of solenoid pilot valves. Hydraulic fluid at pressure
is made available through a supply line within the umbilical and stored in
pressure vessels or accumulators. When a signal is sent to the multiplexer it
energizes the selected solenoid valve, thereby directing the hydraulic fluid
from the supply umbilical and accumulators to the associated actuator. The
multiplexer makes the transmission of data such as pressure, temperature,
and valve position possible by means of electrical signals. The multiplex
system is very flexible but also very complex and incorporates a large num-
ber of subsea electrical components.
15.2.3 Elements of Control System

The main elements of a subsea production control system typically include
the following: A hydraulic power unit (HPU) provides a stable and clean
supply of hydraulic fluid to the remotely operated subsea valves. The fluid
is supplied via the umbilical to the subsea hydraulic distribution system
and to the SCMs to operate subsea valve actuators.
The master control station (MCS) is the central control “node” containing
application software required to control and monitor the subsea production
system and associated topside equipment such as the HPU and EPU. The
electrical power unit (EPU) supplies electrical power at the desired voltage
and frequency to the subsea users. Power transmission is performed via the
electrical umbilical and the subsea electrical distribution system.
The modem unit modulates communication signals for transmission to

and from the applicable subsea users.
The uninterruptible power supply (UPS) is typically provided to ensure
safe and reliable electrical power to the subsea production control system.
In a piloted-hydraulic or electro hydraulic control system, the subsea
control module (SCM) is the unit, which upon command from the MCS
directs hydraulic fluid to operate subsea valves. In an electro hydraulic sys-
tem the SCM also gathers information from subsea located sensors and
transmits the sensor values to the topside facility.
Subsea distribution systems covers distribution systems that distribute
electrical, hydraulic, and chemical supply from the umbilical termina-
tion(s) to the subsea trees, manifold valves, injection points, and the con-
trol modules of the subsea production control system.
Subsea located sensors are sensors located in the SCM, or on subsea trees
or manifolds, providing data to help monitor operation of the subsea pro-
duction system.
Control fluids are typically oil-based or water-based liquids that are used
to convey control and/or hydraulic power from the surface HPU to the
SCM and subsea valve actuators.
15.2.4 Umbilical Technological Challenges and Solutions

Some of the technological challenges are discussed below:
(1) Deepwater
The deepest umbilical installation to date is in 2,316 m of water, at Shell’s
Na Kika project. Some other deepwater umbilicals are the Thunder Horse
umbilical in 1,880 m water depth, and the Atlantis umbilical in 2,134 m
(Terdre, 2004). A challenge in design is that steel tubes are under high exter-
nal pressure as well as high tensile loads. At the same time, the increased
weight may also cause installation problems. This is particularly true for
copper cables as yield strength for copper is low. In ultra-deep water, a
heavy dynamic umbilical may present a problem to installation and opera-
tion as its hang-off load is high.
For design and analysis of ultra-deep water umbilical, it is important to
correctly model the effect of stress and strain on an umbilical and the fric-
tion effect. Sometimes, bottom compression may be observed for umbilical
under 100-year hurricanes. In this scenario, the design solution may be to
use lazy-wave buoyancy module or to use fiber carbon rods. The use of car-
bon fiber rods allows umbilicals in a simple catenary configuration, with-
out the need for expensive, inspection/maintenance demanding buoyancy
modules. The carbon fiber rods enhance axial stiffness as they have a Young’s
modulus close to the value of steel but with only a fraction of the weight.
One of the concerns for use of the carbon fiber rods is their capacity for com-
pressive loads. It is hence beneficial to conduct some tests that document the
minimum bending radius (MBR) and compressive strength of the umbilical.
If the currents in severe for ultra-deepwater umbilical, it might be nec-
essary to use strakes for VIV protection, although the use of strakes has so
far not been required yet. The strakes may for instance be a 16D triple start
helix with a strake height of 0.25D.
(2) Long-Distance
The length for Na Kika, Thunder Horse and Atlantis umbilicals is 130, 65,
and 45 km, respectively. The longest yet developed is 165 km in a single
length, for Statoil’s Snohvit development off northern Norway. One of the
constraints on umbilical length is the capacity of the installation equip-
ment. The Nexans-operated installation vessel, Bourbon Skagerrak, can
carry up to 6,500 tons of cable, that equals to a length of 260 km, assuming
umbilical unit weight is 25 kg/m.
(3) High Voltage Power Cables
The design constraints are the low yield strength of copper, which requires an
increasing amount of protection as depths increase, and the weight of steel
armoring employed to provide that protection as depths increase. Fatigue of
copper cables in dynamic umbilical is another technical challenge.
(4) Integrated Production Umbilical (IPU®)
Heggadal (2004) presented an integrated production umbilical (IPU®)
where the flowline and the umbilical are combined in one single line, see
Figure 15.2. The IPU cross- section consists of the following elements:
• A 10¾” flowline with a three-layer PP coating (its thick-

ness is 4 and 14 mm for static portion and dynamic portion,
respectively).
• Around the flowline, there is an annular shaped PVC matrix
that keeps in place the spirally wound umbilical tubes and
cables and provides thermal insulation to the flowline.
• Embedded in the PVC matrix, but sliding freely with it, the
various metallic tubes for heating, hydraulic and service
fluids, the electrical/fiber optic cables for power and signal,
and the high voltage cables for powering the subsea injec-
tion pump.
• An outer protective sheath of polyethylene 12-mm thick.
HV power cables
Flow line
Heating pipes
FO temperature
monitoring
PVC matrix
Electrical quads
Hydraulic/Service lines
Methanol injection
Weight Elements
Figure 15.2 IPU dynamic cross-section, super duplex flowline (Heggadal, 2004).
To qualify a new design concept like this, a series of analysis and quali-
fication tests were conducted as below (Heggadal, 2004):
(1) Analysis
• Global riser analysis and fatigue analysis;
• Corrosion and hydrogen induced cracking assessment;
• Thermal analysis;
• Structural analysis (prod. pipe, topside, and subsea termination);
• Reeling analysis;
• Electrical analysis;
• Reeling, trawl interaction, and on-bottom studies.
(2) Basic Tests

• Mechanical material tests, fatigue, corrosion, etc.
(3) Fabrication Tests

• Fabrication and closing test;
• STS injection test;
• QC tests and FAT;
• Pre/post installation tests.
(4) Prototype Tests

• External hydostatic test;
• Impact test;
• Model tensioner test;
• Reeling and straightening trials;
• Stinger roller trial;
• Repair trial;
• Vessel trial;
• System test;
• Dynamic riser full scale testing.
15.3 Cross-Sectional Design of the Umbilical

One of the initial stages in the design of an umbilical is the placing of the
components of the umbilical in the cross-section design. The cross-section
of an umbilical could include various items such as steel tubes for transport-
ing hydraulic and other fluids, electrical cables, fiber optic cables, steel rods
or wires for strength capacity, polymer layers for insulation and protection,
and polymer fillers to fill in the spaces between the components and keep
them in place.
Various issues need to be taken into consideration during this stage of
the design. A number of these issues are listed here:
• Tension and bending forces in the umbilical set up stress in

the steel tubes of the umbilical. This stress needs to be lim-
ited to an allowable value (defined in standards such as ISO
13628-5). If the stress exceeds the allowable value, the design
needs to be changed. One method to reduce the stress in the
steel tubes is to add steel rods or wires in the cross-section
so that the stress is distributed amongst them as well as the
steel tubes.
• The maximum yield is the limitation to determine the
umbilical breaking loading. The maximum breaking loading
calculation based on the steel tube and steel rods material
grade. In some cases, steel rods are excluded in the calcu-
lation of the maximum breaking loading. The maximum
working loading may be taken conservatively as 20% of the
maximum breaking loading. The maximum installation
loading may be taken as 55% of the maximum breaking
loading. However, all the loadings need to be confirmed by
the manufacture when the detail design is taken place.
• The umbilical drag to weight ratio is carefully controlled at
this stage of the design as this has a significant effect on the
dynamic behavior of the umbilical. If this ratio is unsatis-
factory and is expected to cause problems in the dynamic
behavior of the umbilical (such as clashing with adjacent
risers or structures), the drag to weight ratio can be modified

by changing the diameter of the umbilical or the weight. It
is much easier to increase the weight or the diameter, rather
than decrease them. The diameter can be increased by spac-
ing the components and adding more fillers, or adding more
components such as steel rods or steel tubes. The addition of
steel rods is the most straightforward method of increasing
the umbilical weight.
• Manufacturing constraints could have an impact on the
cross-section design. For example, some umbilical manu-
facturers are only able to place 12 functional components in
one layer of the umbilical.
• Components might need to be placed toward the outside
of the umbilical for easy access for repair and maintenance.
This is especially true for electrical and fiber optic cables.
• Electrical cables might need to be placed in a certain
order for them not to cause electrical interference with
each other. The use of insulation might also be required
for electrical cables. This is especially true for high volt-
age cables. The use of insulation could also have a detri-
mental effect by causing an increase in temperature in the
umbilical due to heat generated by the current flowing
through the electrical cables.
• Heavy components in the umbilical might need to be placed in
a certain order to prevent torsional unbalance of the umbilical.
Once the cross-section has been designed, an initial static analysis may
be performed for the umbilical in the mean vessel position. The length of
umbilical from the hang-off point to the touchdown point and its projected
horizontal length (which is the horizontal distance from the hang-off point
to the touchdown point) may be estimated using simple catenary theory.
The static catenary analysis may also predict the effective tension at the
hang off point, and at the TDP region, respectively. The minimum bend
radius (MBR) at the TDP region may be calculated using the catenary the-
ory, and be compared with an acceptance criterion.
15.4 Steel Tube Design Capacity Verification

The functional design capacities of the steel tubes of an umbilical need to
be evaluated during the design process. Specifically, calculations should
be performed to ascertain that the steel tubes are adequately designed in

terms of pressure containment, collapse, static loading, buckling, and oval-
ity. ISO 13628-5 is a useful reference for carrying out these calculations
and for determining the utilization ratios applicable to the umbilical. This
section describes some of the equations and methods that are used to ver-
ify the umbilical steel tube design.
15.4.1 Pressure Containment

The hoop stress in the steel tubes should be calculated in accordance with
the methods described in Sections 7.10.2.2.2 and 7.10.2.3.2 of ISO 13628-5.
Most likely, the wall-thickness of the steel tubes is determined by the pres-
sure containment requirements. However, buckling and collapse as well as
von Mises stress criteria need also to be checked for completeness.
15.4.2 Allowable Bending Radius

One of the processes during the design of a steel tube umbilical is the deter-
mination of the minimum allowable bending radius of the umbilical. This
bending radius is calculated by determining the stress that is experienced
by the umbilical during different phases of its life and also under various
operating conditions. The stress prevalent in the steel tubes depends on
three components: the pressure in the tubes, the axial stress due to the
tension in the tubes, and the bending stress.
(1) The pressure in the tubes generally remains constant at the

standard operating or design pressure. It is only likely to
change during extreme environmental or operational con-
ditions, or emergency situations.
(2) The tension and hence the axial stress in the tubes changes
depending on environmental conditions.
(3) Thus, this stage of the design process usually involves
calculating the minimum allowable bending radius of
the umbilical for a variety of environmental conditions
and top tensions, including the installation phase. These
allowable bending radii are not the same as the umbilical
storage MBR. The storage MBR is defined by the manu-
facturer and refers to the smallest radius that the umbili-
cal can be bent to when stored on reel. This storage MBR
is the minimum radius that does not cause any damage to
the steel tubes of the umbilical when stored on a reel, and

under no effect from internal pressure and tension in the
umbilical.
Knapp (1988) describes a method for calculating the allowable bending
stress in the umbilical, based on the pressure and axial stress experienced
by the steel tubes. The chapter gives a theoretical background to deter-
mining compound stress in helical wires of a cable bent through a planar
circular arc. The following assumptions have been made:
(1) Only round wires are considered;

(2) The geometry of deformation of a helical wire can be
described by its centurial axis;
(3) The helical wire lay angle and pitch radius remain constant;
(4) For the frictionless case, the transverse section of the cable
is allowed to warp as wires slip freely;
(5) For the full friction model, plane sections of the cable
remain planar after bending;
(6) Inter-wire contact stresses are neglected;
(7) The helical wires are homogeneous, isotropic and linearly
elastic.
The chapter shows both frictionless (lower bounding) and full friction
(upper bounding) models to evaluate the bend radius. It also indicates that
the frictionless model is more likely to be realistic, based on the experi-
ments conducted. The allowable MBR of the umbilical can be easily deter-
mined once the allowable bending stress is known.
15.5 Extreme Wave Analysis

An important aspect of the umbilical design process is an analysis of
extreme environmental conditions. A finite element model of the umbil-
ical is analyzed with vessel offsets, currents and wave data expected to be
prevalent at the site where the umbilical is to be installed. For example, in
the Gulf of Mexico, this would include an analysis for a 100-year hurri-
cane, 100-year loop current, and submerged current. The current and wave
directions are applied in a far, near, and cross condition. This analysis is
used to determine the top tension and angles that the hang-off location of
the umbilical is likely to experience. These values are then used to design
an adequate bend stiffener that will limit the umbilical movements and
provide adequate fatigue life for the umbilical. Design analysis resulted
from extreme analysis are:
(1) The touchdown zone of the umbilical is analyzed to ensure

an adequate bending radius that is larger than the minimum
allowable bending radius. It is also important to check that
the umbilical does not suffer compression and buckling at
the touchdown zone.
(2) A polyurethane bend stiffener has been designed to have a
base diameter of xxx inch, and cone length of yyy ft. This
design is based on the maximum angle and its associated
tension, and maximum tension and its associated angle
from dynamic analysis results using the pinned FE model.
(3) The maximum analyzed tension in the umbilical was found
to occur at the hang-off point for the 1,000-year hurricane
wind load case when the vessel is in far position.
(4) The minimum tension in the umbilical may be found to
occur in the TDP region for the 1,000-year hurricane wind
load case when vessel is in the near position.
(5) The MBR is estimated over the entire umbilical, over the
TDP region and the bend stiffener region, respectively.
They are to be larger than the allowable dynamic MBR.
(6) The minimum required umbilical on-seabed length is esti-
mated assuming that when it is subject to the maximum
value of the extreme bottom tensions.
15.6 Manufacturing Fatigue Analysis

A certain amount of fatigue damage is experienced by a steel tube umbil-
ical during manufacturing, and this needs to be evaluated during fatigue
analysis. The two main aspects of umbilical manufacturing fatigue analysis
that require attentions are accumulated plastic strain and low cycle fatigue.
These are explained below.
15.6.1 Accumulated Plastic Strain

Accumulated plastic strain is defined as “the sum of plastic strain incre-
ments, irrespective of sign and direction” (DNV, 2000). Accumulated plas-
tic strain can occur in the steel tubes of an umbilical during fabrication and
installation. The accumulated plastic strain needs to be maintained within
Coiling
Carousel
Carousel
Coating
Welding Transport Reel
Straightener Tensioner Stinger
Carousel
Figure 15.3 Diagram of deformations during fabrication and installation.
certain limits to avoid unstable fracture or plastic collapse for a given tube
material and weld procedure. Accumulated plastic strain is the general
criteria used by umbilical suppliers to determine whether the amount of
plastic loading on the steel tubes is acceptable. An allowable accumulated
plastic strain level of 2% is recommended for umbilical design.
Figure 15.3 shows a schematic of deformations that are likely to take
place during the fabrication and installation of a steel tube umbilical. All
the processes shown in this diagram are likely to induce plastic strain in
the umbilical.
15.6.2 Low Cycle Fatigue

The umbilical steel tubes are subject to large stress/strain reversals during
fabrication and installation. Fatigue damage in this low cycle regime is cal-
culated using a strain-based approach.
For each stage of fabrication and installation, the fatigue damage is cal-
culated by considering the contributions from both the elastic and plastic
strain cycles. The damage calculated from low frequency fatigue is added to
that from in-service wave and VIV conditions to evaluate the total fatigue
life of each tube of the umbilical.
15.7 In-Place Fatigue Analysis

The methodology used to assess wave induced in-place fatigue damage of
the umbilical tubes can be summarized as follows:
• Selection of sea state data from wave scatter diagram;

• Analysis of finite element static model;
• Umbilical fatigue analysis calculations;

• Simplified or enhanced approach;
• Generation of combined stress history;
• Rainflow cycle counting procedure or spectral fatigue damage;
• Incorporation of mean stress effects in histogram.
These aspects of fatigue analysis are described below. The main differ-
ence between fatigue analysis for an umbilical and a SCR is the effect of fric-
tion when the tubes in the umbilical slide against their conduits and each
other due to bending of the umbilical. The methodology discussed here
for umbilical in-place fatigue analysis is based on two OTC Papers: Paper
13203 by DUCO (Hoffman, 2001), and Paper 16631 by MCS (Kavanagh
et al., 2004). In-place fatigue analysis is required to prove that the fatigue
life of the umbilical is 10 times the design life.
15.7.1 Selection of Sea State Data From Wave Scatter Diagram

The wave scatter diagram describes the sea state environment for the
umbilical in service. It is not practical to run a fatigue analysis with all the
sea states described in a wave scatter diagram. Hence, the usual methodol-
ogy is to group a number of sea states together and represent these “point
sea states” with one significant wave height and wave period. The values of
the wave height and wave period are chosen to be conservative.
This methodology results in the reduction of the wave scatter diagram
to a “manageable” number of sea states (say about 20–50). This enables the
analysis to be carried out in a reasonable amount of time. It is also very
important to accurately consider the percentage of time that the umbilical
is expected to be affected by these different sea states.
15.7.2 Analysis of Finite Element Static Model

A finite element static analysis is carried out for a model representing the
steel tube umbilical. The static solution is used as a starting point for a time
domain or frequency domain dynamic finite element analysis.
15.8 Installation Analysis

The issues that need to be considered when dealing with fatigue damage
during installation of steel tube umbilicals are as follows:
• The contribution to accumulated plastic strain during reel-

ing and potential retrieval;
• Low cycle fatigue during reeling and potential retrieval;
• Dynamic wave frequency fatigue contributions during the
critical stages of installation, i.e., midlay and handover/pull-in.
The methodology for accounting for accumulated plastic strain and low
cycle fatigue has already been considered in Section 15.5. The calculations
for accumulated plastic strain and low cycle fatigue are carried out for both
fabrication and installation together.
The methodology for the calculation of wave induced fatigue damage
during the critical stages of installation is similar to the in-place fatigue
assessment described in Section 15.6. However, there are some aspects of
installation fatigue analysis that do not apply to in- place fatigue analysis.
Primarily,
• Since the umbilical changes configuration and is subject to

different loads during various stages of installation, different
umbilical models are needed to model the various stages of
installation that require analysis.
• For installation fatigue analysis, it is appropriate to use a
time-domain approach. A frequency domain analysis would
not adequately predict the fatigue damage suffered during
installation due to the highly irregular loading that the
umbilical experiences during this stage of its life.
15.9 Required On-Seabed Length for Stability

From the on-seabed stability point of view, before a curve may be taken in the
field layout, the length of the umbilical on the seabed should be long enough
so that the friction force provided by the on-seabed portion of the umbilical
is larger than the extreme bottom tension from the dynamic analysis.
Assuming an on-seabed length of L ft, the friction force is calculated as
follows:
Friction Force = Friction coefficient*Submerged weight*L
Therefore,
L = Friction Force/(Friction coefficient*Submerged weight)

Assuming that the maximum value of the bottom tension is 7.28 kips,
the submerged weight is 9.71(lbs/ft), the required on-seabed length for a
friction coefficient of 0.5 is
L = 7.28*1,000(lbs)/[0.5*9.71(lbs/ft)]
= 1,450 ft
Since the umbilical changes configuration and is subject to different

loads during various stages of installation, different umbilical models are
needed to model the various stages of installation that require analysis.
References
Almar-Naess, A. (1985), “Fatigue Handbook: Offshore Steel Structures”, Tapir.
API 17E: “Specification for Subsea Umbilical”, The American Petroleum Institute.
API Recommended Practice 2RD, (1998), “Design of Risers for Floating
Production Systems (FPSs) and Tension Leg Platforms”, 1st Edition.
Bjornstad, B. (2000), “Umbilical Stretches Subsea Performance”, E & P Magazine,
August 2004.
DNV Offshore Standard OS-F101(2000), “Submarine Pipeline Systems”.
DNV Recommended Practice, RP-C203(2000), “Fatigue Strength Analysis of
Offshore Steel Structures”.
GE Research & Development Center (2000), “Influence of Mean Stress on Low
Cycle Fatigue in High Temperature Water”, Doc. No. 2000CRD025, May 2000.
Heggdal, O. (2004): “Integrated Production Umbilical (IPU®) for the Fram Ost
(20 km Tie-Back) Qualification and Testing, Deep Oil Technology (DOT),
Dec. 2004, New Orleans.
Hoffman, J., Dupont W., Reynolds B. (2001), “A Fatigue-Life Prediction Model for
Metallic Tube Umbilicals”, OTC Paper 13203.
ISO 13628-5, 2000, “Petroleum and Natural Gas Industries–Design and Operation
of Subsea Production Systems”, Part 5: Subsea Umbilicals, 2002.
Kavanagh, W. K., Doynov, K., Gallagher, D., Bai, Y. (2004), “The Effect of Tube
Friction on the Fatigue Life of Steel Tube Umbilical Risers–New Approaches
to Evaluating Fatigue Life using Enhanced Nonlinear Time Domain Methods”,
OTC Paper 16631.
Knapp, RH, (1988): “Helical Wire Stresses in Bent Cables”, Journal of Offshore
Mechanics and Arctic Engineering, No. 110, pages 55–61, February 1988.
MCS (2004), Flexcom, Version 7, User’s Manual.
Stephens R, Fuchs HO(2001), “Metals in Fatigue Engineering”, 2nd Edition, Wiley

and Sons Inc., 2001.
Suresh, S (1991), “Fatigue of Materials”, Cambridge Solid State Science Series.
Swanson, RC, Rao, VS, Langner, CG, Venkataraman, G. (1995), “Metal Tube
Umbilicals–Deepwater and Dynamic Considerations”, OTC Paper 7713
(1995).
Terdre N (2004), “Nexans Looking beyond Na Kika to Next Generation of Ultra-
Deep Umbilical”, Offshore Magazine, March 2004.
16
Stress and Fatigue of Umbilicals
16.1 Introduction
A critical component in the wet tree development of deep water fields
is the umbilical riser. Failure of the umbilical can result in partial or total
loss of production, with resulting severe economic consequences. In many
recent deep water projects, steel tubes for control and injection lines have
become the preferred solution over traditional thermoplastic hose based
designs. The fatigue design of these steel tube umbilicals (STUs) with accu-
rate modeling of the steel tube interaction friction stresses is critical for
any development using STUs.
Simplified conservative approaches where the non-linearities of the
tube friction are ignored have frequently been used for STU fatigue design.
This chapter presents the theoretical basis for an enhanced non-linear time
domain fatigue model for deep water STUs. In developing this model, the
authors have proposed an analytical fatigue design approach that accounts
for the non-linearity of tube friction using slip theory combined with
time-domain global and local analytical methods.
A worked example is presented, where an ExxonMobil West of Africa
field development STU has been analyzed using both the enhanced non-
linear time domain fatigue model and an alternative simplified model.
The enhanced non-linear time domain fatigue model, has been used to
evaluate the influence on fatigue life of several parameters, including fric-
tion, water depth and STU cross-section geometry, some of which become
increasingly important in deeper water.
The merits of the enhanced nonlinear fatigue model over the simplified
model are critically evaluated, as are the lessons learned on the impor-
tance of critical design parameters and analysis assumptions affecting
STU fatigue life. Conclusions are also presented on the benefit of further
enhancement of the analytical approaches commonly used to design for
umbilical fatigue. The analysis in this chapter is quoted from reference [1].
337
16.2 STU Fatigue Models

In this section, equilibrium equations describing lateral buckling of tensile
armor in straight flexible pipe with no friction are derived on the basis of
the Principle of Virtual Work where a closed form expression for the
lateral buckling force is obtained by the Galerkin method.
In dynamic floating production STU risers, cyclic stresses occur in
the metal tubes within the umbilical due to the wave-induced dynamic
response of the umbilical and the motion of the floating vessel. These
dynamic stresses are a combination of the axial and bending cyclic stress
variations in tubes, induced by the global motions of the umbilical, com-
bined with the effect of friction between tubes and surrounding structure
in the umbilical cross-section.
Cross-section models and fatigue analysis techniques used to design
these tubes frequently make simplified assumptions about how to calculate
and combine cyclic stresses in these tubes to determine fatigue life. Such
assumptions typically include:
(i.) regular wave analysis of global riser response to fatigue scatter

diagram sea states compared to alternative irregular wave time or
frequency domain approaches;
(ii.) non-phase-consistent calculation of tube fatigue stresses under
global umbilical bending and tension;
(iii.)
fatigue calculations based on simplified tube friction/slip or
no-friction model;
(iv.) approximate cycle counting based on a regular wave response and
single dominant frequency instead of an alternative which cap-
tures actual sea state frequency content and multiple amplitude
cycles, for example, a rain flow counting irregular wave approach;
(v.) in-plane fatigue analysis based on analysis of umbilical response to
global motions in the same plane;
(vi.) prediction of friction stresses using static friction coefficient rather
than separate static and dynamic friction coefficients.
Cyclic stresses predicted by these cross-section models are typically

used in more conventional fashion to calculate fatigue life based on the
Palmgren-Miner approach of summing damage contributions from each
stress range as the number of stress cycles at each identified stress range
divided by the number of cycles to failure at that stress range. The number
of cycles is derived from the sea state scatter diagram while the number of
Stress and Fatigue of Umbilicals 339
cycles to failure is typically based on S-N curves derived from laboratory

fatigue tests.
The necessity for simplified assumptions depends on the availability
of enhanced cross-section stress models capable of more accurately rep-
resenting the behavior of the cross-section and ultimately depends on
achieving confidence in those models as representative of actual condi-
tions. The correct prediction of tube stress ranges, incorporating the effects
of tension, bending and friction, and their associated number of cycles are
thus the key to developing a reliable fatigue analysis model.
This chapter presents and compares two models of STU friction. The
first, a Simplified Model is derived from the approach described in refer-
ence [2] and summarized in the following section. In umbilical design, this
method is typically applied based on regular wave global analysis of umbil-
ical riser response, combined with a slip vs. no-slip condition defining tube
stress range under each of several fatigue sea states.
The authors have developed an alternative Enhanced Time-Domain Model
that combines tension and curvature induced stresses in a phase-consistent
manner, while accounting for the non-linearities of the tube friction and the
response of the umbilical to an irregular wave sea state in time domain. The
model represents an enhancement of the simplified cross-section fatigue
modeling approach by incorporating the phase-consistency of bending and
tension and a non-linear time domain tube-slip model implemented for an
irregular sea global response.
16.2.1 Simplified Model

This model is based on the theory developed by Hoffman et al. and pre-
sented in reference [2]. The theory is typically implemented based on reg-
ular wave approach and cyclic ranges of tension and curvature at critical
locations along the umbilical are derived from the regular wave global FE
analysis for use in determining stress cycles in the steel tubes. For this
comparison, the theory is modified to present stress ranges derived from
rms bending and tension for each cyclic load case.
16.2.1.1 Axial and Bending Stresses

In the absence of phasing information between tension and curvature, the
axial stress range ΔσA and bending stress range ΔσB may be calculated from
the standard deviations of tension SDT and curvature SDk, assuming a sinu-
soidal wave, as follows:
ΔσA = 2√2•SDT/A (16.1)
ΔσB = 2√2ER•SDS (16.2)
where E is Young’s modulus, A is steel cross-sectional area of the umbili-

cal, and R is outer radius of the critical steel tube.
Note that the factor of 2 2 in Eqs. (16.1) and (16.2) is taken from [3]
and this factor relates the stress range to the stress standard deviation. In
Eq. (16.2), the local tube curvature is assumed to vary as the global umbil-
ical curvature due to small lay angle of the local tube.
16.2.1.2 Friction Stress

Based on the slip theory presented in [2], the stress due to friction, σF,
is the minimum of the sliding friction stress, σFS, and the bending friction
stress, σFB.
In an umbilical where tubes are designed to move axially within con-
duits, the bending friction stress is the stress in the tube which occurs
before relative tube-conduit slip, where plane sections are still plane (i.e.,
the umbilical cross- section is behaving like a steel catenary riser, with
tube axial and bending stress range derived from the tube’s distance from
the umbilical neutral axis). In a frictionless conduit, this term is zero.
The sliding friction stress is the maximum axial stress that is taken by
the tube before it slips relative to its conduit. This term is proportional to
the high helical contact force and the friction coefficient between tube and
surrounding structure.
These terms may be expressed as
∆σ A = 2 2 • SDT /A (16.3)
σ FR = 2 ERL • SDk (16.4)
σF = min(σFS, σFR) (16.5)
where μ is friction coefficient, Fc is contact force between the helical steel

tube, At is cross-sectional area of the critical steel tube within the umbili-
cal, E is tube steel Young’s modulus, and RL is layer radius of the tube (this
being the distance from the center of the umbilical to the center of the
critical steel tube as shown in Figure 16.1).
T
2 π RL
RH
RL
Pitch Length, Lp = 2π RL/tanθ

Lp
Helix Length
= Lp/Cosθ
Figure 16.1 Schematic of helical geometry.
From the theory developed in [2], it is shown that the frictional contact
force, assuming a straight umbilical, is evaluated as
 T sin 2 ϕ EItube sin 4 ϕ 

Fc =  +  Lp (16.6)
 RL RL3 
where T is mean tension, Lp equals to(tube pitch length)/2, φ is tube lay

angle (this being the angle at which the tubes lie relative to the umbilical
neutral axis as shown in Figure 16.1), and EItube is individual tube bending
stiffness.
The first term of this equation relates to the helical radial force of the
steel tube on the material supporting it, caused by the tension in the helix.
The second term is derived from the bending of the steel tubes over the
supporting structure during manufacture and is a function of the bending
stiffness of the tube. Significant uncertainty relates to the existence and
the magnitude of the second term of Eq. (16.6) for two reasons. The first
is because of the likely reduction of this elastic bending stiffness effect
due to plastic deformation of steel tubes during manufacture. The second
is the potential reduction or disappearance of this stress contribution due
to stress relaxation after manufacture, when manufacturing tension is
removed and end fittings are fitted to the umbilical. For the examples
considered in this chapter, which related to an umbilical of small tube lay
angle, the term has been found to be very small relative to the first term
and the authors have chosen to retain the term according to the model
presented in [2]. Further model enhancement is considered necessary to
validate the existence of this term for more generalized application.
In the absence of capturing the correct phasing of bending and tension,
the Simplified Approach presented in [2] conservatively sums the stress
contributions from all sources before calculating a total stress range for
fatigue damage calculations.
16.2.1.3 Simplified Approach: Combining Stresses

The stress range, Δσ, acting on the largest steel tube within the umbilical
is obtained as the sum of the axial stress range ΔσA, bending stress
range, range, ΔσB, and friction stress ΔσF[2].
Δσ = ΔσF + ΔσA + ΔσB (16.7)
This conservatively combines all stresses assuming they are in phase

with one another.
16.2.1.4 Simplified (Combining Stresses) Fatigue Damage

The mean zero crossing period, Tz, is typically obtained based on its empir-
ical relationship with sea state spectral peak period, Tp. For the location
taken for the worked example described later in this chapter, the relation-
ship between Tz and Tp was calculated to be
Tz = Tp/1.18 (16.8)
and this is a state percentage occurrence, p, the number of cycles occurring

in one year, n, is easily calculated.
n = 365 × 24 × 3600 × p/Tz (16.9)
The accumulated fatigue damage is calculated based on the Palmgren-

Miner rule as follows:
Ns
 
D= ∑ Nn 
i =1
i
(16.10)
where Ns is the number of sea states, N is the number of cycles to failure

for the stress range from the ith sea state, and n is the actual number of
cycles for the ith sea state. For each sea state, N is calculated using the fol-
lowing S-N curve equation:
N = KΔσ−m (16.11)
where m and K are material constants depending on the environment and

welding conditions. Finally, the fatigue life of the STU is calculated as the
inverse of the damage.
1
life = (16.12)
D
16.2.1.5 Simplified Model Assumptions

Combining stresses presents a conservative evaluation of cyclic stress
by implicitly assuming that the axial, bending and friction stresses are in
phase.
The simplified method presents a simplified friction model by typically
representing the stress from each sea state with a single stress range and an
associated number of cycles which is calculated based on a single period
only, ignoring stress cycles at other frequencies.
Each sea state is assumed to represent either a slip or stick condition,
based on the outcome of Eq. (16.5), ignoring the existence of both slip and
stick conditions in a single sea state.
The Simplified Approach is, despite its simplifications, very simple to
implement since it typically combines regular wave global analysis with a
very simple friction/slip model.
16.2.2 Enhanced Non-Linear Time Domain Fatigue Model

For the enhanced nonlinear time domain model, an irregular sea time
domain global analysis is performed on the STU for each fatigue sea state
for a duration sufficient to get reliable damage statistics (say 0.5 to 3 hours).
The required output results from each sea state are time histories of ten-
sion and curvature in which the phasing information between tension and
curvature is retained.
16.2.2.1 Friction Stresses

The obtained time histories are used to generate a time history of sliding
friction stress σFS(t) and bending friction stress σFB(t) using the tube slip
theory from [2] as follows:
µ Fc (t ) (16.13)
σ FS (t ) =
At
σFB(t) = ERL•k(t) (16.14)
where Fc(t) is time history of contact force and k(t) is time history of global
curvature. The time history of frictional contact force in a straight umbili-
cal is calculated as follows:
 T (t )sin 2 ϕ EItube sin 4 ϕ 

Fc =  +  Lp (16.15)
 RL RL3 
where T(t) is time history of tension, Lp equals to (tube pitch length)/2.

Note that Eq. (16.15) is used to calculate the frictional contact force,
based on the time trace of tension at a single point along the umbilical.
Because the tension is maximum in the umbilical hang-off region and
reduces with depth, the global curvature variations at the point at which
Fc is calculated is what is relevant to fatigue at that location. For the time
trace of maximum tension, (16.15) represents the frictional contact force
time trace at the umbilical hang-off area.
16.2.2.2 Effect of Multiple Tube Layers

For the fatigue assessment of steel tubes in the inside layer(s) of multi-layer
umbilical, as illustrated in Figure 16.5, outer layer tubes, unless they are
fully supported by the cross-section conduit matrix itself, can contribute
to an increase in the contact force on inner tubes due to the additional
helical contact forces from the outer tubes. Figure 16.6 illustrates this sche-
matically for the case of a single small tube outside a large inner tube. In
such a case, a radial force term, Fc−OT, exists between the outer and inner
tube layers due to the tension carried by the outer tube helix. Furthermore,
the outer tube also contributes to an additional increase in contact force
between the large inner tube and the underlying support structure near
center of the umbilical.
For a structure of make-up similar to Figure 16.5, Fc may be calculated
as the inner tube contact force added to additional contact force contribu-
tions derived from the outer steel tube layer, as illustrated by Figure 16.6.
One way to estimate this enhanced Fc term is to first calculate the inner
tube Fc based on Eq. (16.15) and then augment that by an additional Fc
contribution from the outer layer tubes, as follows:
Fc−TOT(t) = Fc−OT(t) + [Fc−IT (t) + Fc−OT′(t)] (16.16)
where Fc−TOT is total radial contact term to use for friction stress calcula-
tions, Fc−TOT outer tube radial contact force from Eq. (16.15) applied to
outer layer, Fc−IT′(t) is inner tube radial contact force from Eq. (16.15)
applied to inner layer, Fc−OT′ is amount of the outer tube radial con-
tact force from Eq. (16.15) which is transmitted directly to the inner tube.
Assuming that the supporting structure surrounding the tube does not
share the radial load, a conservative approach to Eq. (16.16) is to assume
that Fc−OT(t) = Fc−OT′(t), thereby assuming that the outer tube radial load
is transmitted entirely to the inner tube, without any support from the
surrounding structure. This approach represents an analytical estimate of
this contribution only, for the case where plastic spacers exist between the
tubes. Additional investigation of the friction contribution of outer tubes
is required to achieve greater confidence in the magnitude of this term.
For the sake of the worked example presented later in this chapter, the
simple case of no outside tubes has been taken for the associated paramet-
ric study.
16.2.2.3 Combined Friction Stresses

The overall time history of friction stress is the minimum of the sliding or
bending friction stresses and is expressed as follows:
σF(t) = min(σFS(t), σFR(t)) (16.17)
16.2.2.4 Axial and Bending Stresses

From the time histories of tension T(t) and curvature k(t), the time his-
tories of axial stress σA(t) and bending stress of the tube bending about its
own axis σB(t) are calculated as follows:
σA(t) = T(t)/A (16.18)
σB(t) = ERT•k(t) (16.19)
where RT is the radius of the critical tube.

Note that a small tube lay angle is implicitly assumed in the friction
bending term of Eqs. (16.14) and (16.19) in the assumption that umbili-
cal bending is in the same plane as tube bending. For the lay angles con-
sidered here, this is generally a reasonable approximation. More rigorously,
stress could be derived based on tube force resolved in the direction of lay.
16.2.2.5 Combining Stresses

A combined stress time history ΔσC(t) for a particular tube which includes
friction stress and combines all stress with the correct phasing is calcu-
lated by adding the individual stress time histories together.
ΔσC(t) = ΔσF(t) + ΔσA(t) + ΔσB(t) (16.20)
16.2.2.6 Fatigue Life

Once a friction included combined stress time history has been developed
for each sea state, a rain flow counting algorithm, which also requires the
percentage occurrence of each sea state, is used to generate a single, friction
included, combined stress histogram. This rain flow counting procedure
is an ASTM standardized procedure and is discussed in detail in [4]. The
rain flow counting algorithm examines a time history and counts the stress
peaks organizing them into a combined stress histogram consisting of a
number of stress bins, each with a certain stress range, and a correspond-
ing number of cycles for each bin. From this combined stress histogram,
the fatigue damage, D, is calculated based on the Palmgren-Miner rule.
Nb
 
D= ∑ Nn 
i =1
i
(16.21)
Here, Nb is the number of stress bins, N is the number of cycle to failure

for the ith stress bin (Δσi), and n is the actual number of cycles in the
ith stress bin. For each bin, N is calculated using the following S-N curve
equation:
N = KΔσ−m (16.22)
where m and K are material constants. Finally, the fatigue life of the STU is
calculated as shown in Eq. (16.12).
16.2.2.7 Benefits of Enhanced Non-Linear Time Domain Fatigue

Model
Two significant benefits of the enhanced nonlinear time domain model
over the alternative simplified approach are as follows:
1. The time domain method of combining the friction stresses

ending with Eq. (16.17) more accurately accounts for the
time varying friction stress, which alternates between tube
slip and non-slip conditions. The alternative simplified regu-
lar wave approach assumes that the STU is either in a slip or
a stick condition for every cycle in a fatigue sea state.
2. Time domain explicitly accounts for the correct phasing of
the cycles of tension and bending of a STU.
Figure 16.2 presents an example time history of the friction stress expe-
rienced by a tube within an STU as predicted by the enhanced nonlin-
ear time domain fatigue model. The friction stress time history shown in
Figure 16.2 can be described as the bending friction stress bound by the
sliding (or tensile) friction stress. The friction stress equals the bending
friction stress until the tensile friction stress is exceeded and at this point
slip occurs. Physically, the tubes slip when the STU experiences large cur-
vatures. Here, the tubes are slipping longitudinally within the umbilical,
and friction stress experienced by the tube as it slips is dependent on the
umbilical tension. When the curvatures reduce and the bending friction
stress becomes smaller than the tensile friction stress, the tubes stick and
the bending friction stress dominates again.
STU stress behavior can be described by the curvature-stress plot pre-
sented in Figure 16.3. Initially, at smaller curvatures, the umbilical behaves
like a steel catenary riser (SCR), i.e., plane sections remain plane. Here,
the tubes are in a non-slip condition and bending occurs about the STU
neutral axis. At a certain curvature level, slip occurs and subsequently the
15
Bending Friction Stress
Tension Friction stress
Compression Friction Stress
10 Overall Friction Stress
5
Stress (MPa)
0
0 10 20 30 40 50 60 70 80 90 100
–5
–10
–15
–20
Time (s)
Figure 16.2 Time history of friction stress for increased tension model.
SCRStress
Slip
Umbilical
TubeStress
Figure 16.3 Stress behavior of STU versus SCR.
tubes are bending about their own neutral axis; hence, the stress no
longer increases linearly with STU curvature. This stress behavior is accu-
rately modeled in the enhanced non-linear time domain fatigue model.
16.3 Worked Example

This section presents a worked example of the application of the enhanced
time domain fatigue analysis compared to the simplified technique. The
example problem is based on an ExxonMobil West of Africa field devel-

opment study. The STU hangs in a lazy wave configuration from an FPSO
in 500-m water depth. Figure 16.4 shows a global FE model illustrating
the STU-FPSO system. The cross-section of the STU consists of 4 × 35
mm OD inner steel tubes with a layer of smaller 10-mm OD steel tubes
positioned outside the central tubes. A schematic of the STU cross-section
is shown in Figure 16.5. The STU cross-section data is shown in Table 16.1.
Figure 16.4 FPSO/STU system layout.
Small
Outer
Tubes
RL2
RL1
Polymer
Matrix
Between
Tubes
Large Inner Tubes
Figure 16.5 STU layup.

Table 16.1 STU layup data.

Units Value
Inner Tube Layer Radius, RL1 (m) 0.0277
Outer Tube Layer Radius, RL2 (m) 0.0550
Inner Tube Outer Diameter (m) 0.0348
Outer Tube Outer Diameter (m) 0.0100
Inner Tube Inner Diameter (m) 0.0318
Outer Tube Inner Diameter (m) 0.0080
Inner Tube Lay Angle (deg) 1
Outer Tube Lay Angle (deg) 1
Twenty-three fatigue sea-states, each with a defined percentage occur-

rence were considered for the STU fatigue analysis. Note that for the
purpose of this example, fatigue life results are normalized to the fatigue
life predicted by the base case time domain model.
A base case has been selected based on a STU with tubes that experi-
ences metal-to-plastic friction. An appropriate friction coefficient for this
friction has been taken as 0.2. The fatigue life of this base case analysis
using the enhanced non-linear fatigue model is assigned a value of 100%.
All other fatigue lives are then presented as a percentage of the base case
fatigue life, enabling direct comparisons between the different results.
This section first presents the effect of time domain vs. simplified mod-
els in their prediction of STU fatigue life. The remainder of the section
then evaluates several time domain sensitivities to the base case time
domain model which demonstrates the significance of several important
STU design variables on the fatigue life of a deep water umbilical.
16.3.1 Time Domain vs. Simplified Approaches

In order to establish the differences in predicted fatigue life between the
time domain and simplified approaches discussed in this chapter, the base
case model was analyzed using both approaches. Results from these analy-
ses are presented in Table 16.2.
Two simplified approaches, based on different assumptions of the ten-
sion and bending stress phasing, predict very different lives. The combined
Table 16.2 Comparing different STU fatigue approaches.

Analysis Description % variation from
(Note all analyses here include % of time domain time domain
friction with µ = 0.2) fatigue life fatigue life
Base Case Enhanced Non-linear 100% –
Time Domain Model
Simplified Model (Adding 27% −73%
Stresses)
Simplified Model (Adding 115% +15%
Damages)
stress simplified model, which makes the conservative assumption that

stress ranges due to bending and tension are additive, predicts a fatigue
life that is 73% less than that of the enhanced time domain model. The
primary reason for this reduction in fatigue life is the method in which
the stresses are combined as if they are exactly in phase. The combined
damages simplified model presents a fatigue life that is 115% that of the
time domain model. This model, in assuming that all stresses are com-
pletely independent of each other is potentially unconservative. In reality,
the combined stresses will consist of bending, tensile and friction stresses
with some phasing and neither of the limits explored by the simplified
models account for this phasing.
What is evident and intuitive from a comparison of the enhanced time
domain with the alternative simplified models is that the time domain
model prediction of fatigue life should be in between the predictions of
the simplified models and results show that it is giving confidence in the
predictions of the time domain approach.
Furthermore, for each sea state, the time domain accounts for the tube
alternating between slip and stick conditions. The simplified model will
assume one condition, either slip or stick, for an entire sea state. This
simplification inevitably also leads to inaccuracies in the results.
16.3.2 Effect of Friction on STU Fatigue

Two enhanced nonlinear time domain models, one with no friction and
one with infinite friction, were developed and results compared with the
base case model results so that the effect of accurately modeling friction
on STU fatigue life could be examined.
The first model, with no friction forces treats oscillating stresses as

if the individual metal tubes within the STU always bend about their own
axis. Hence, this physically represents an STU where the tubes are always
slipping longitudinally within the STU with no restraining frictional
forces being applied. The results from the No Friction and Infinite Friction
models show significant deviations from the base case model. Each of these
is described in this section. Finally, the friction effect of metal-to-metal vs.
metal-to-plastic friction is evaluated in terms of its effect on fatigue life.
Results illustrate the importance of modeling the correct amount of fric-
tion in an STU.
16.3.2.1 Influence of High Tube Friction on Umbilical Fatigue

An infinite friction model has been used to represent an STU with very
high friction forces, in this case representing perfect shear transfer in
the cross- section. This model assumes that plane cross-sections remain
plane, tubes never slip longitudinally and all tube bending occurs with
the tubes bending about the neutral axis of the STU itself (as, for exam-
ple, for an SCR). The fatigue lives obtained from these analyses are pre-
sented in Table 16.3.
In the Infinite Friction model where the tubes bend about the STU
neutral axis and no tube slipping is accounted for in the cross-section, an
87% reduction in fatigue life is predicted. Because slip is prevented in the
model, this fatigue life is considered highly conservative, especially for
shallow water umbilicals, where the tension required to generate tube fric-
tion is less.
16.3.2.2 Influence of Low Tube Friction on Umbilical Fatigue

An alternative No Friction model has been analyzed, where the tubes bend
about their own neutral axis (with no shear transfer between planes).
This predicts a fatigue life 72% longer than the base case model. Since
this case doesn’t include any friction, the result represents an unconserva-
tive view of life whose validity depends on the influence of friction as a
component of fatigue damage.
16.3.2.3 Influence of Metal-to-Metal Friction vs. Metal-to-Plastic

Contact on Umbilical Fatigue
An analysis was performed to check the effect of increasing the friction within
an STU due to the metal tubes being directly in contact with each other.
Outer Tube
Fc-OT
Friction due
to Fc-OT
Inner Tube
Fc-IT
Friction due to
Fc-IT + Fc-OT
Center of
Umbilical
Figure 16.6 Outer tube friction forces.
This type of STU is unlike the base case which assumes that the
metal tubes are in contact with polymer layers between the tubes. Here,
a friction coefficient of 0.3 has been taken to represent metal-to-metal
friction.
Results from the Metal-to-Metal Friction model, presented in Table
16.3, which also represent the case of a 50% increase in friction from any
source, show that increased friction forces due, for example, to metal-to-
metal contact or additional outer tube frictional forces, result in a 23%
reduction in fatigue life of the analyzed STU. This demonstrates the critical
importance of the selection of an appropriate friction coefficient, consis-
tent with the contact surfaces in an STU fatigue model, in determining
fatigue life.
Also illustrated by this result is the importance of considering the effect
of the contact force of an outer layer of tubes, as illustrated schematically
in Figure 16.6, in generating higher friction forces and fatigue damage in
inner tubes for a multi-layer tube makeup.
16.3.3 Effect of Increasing Water Depth

Increased water depth for a floating production system will typically lead
to increased tension at the top of the STU. This increased tension, in turn,
leads to increased frictional contact forces which decrease the STU fatigue
life. To investigate the effect of increasing the water depth by 500 m
on the fatigue analysis of the base case STU, tensions from the time
domain analyses were doubled to simulate a water depth of approximately

1000 m. The fatigue life obtained from this analysis is present in Table 16.3.
The increased water depth shows a 79% reduction in fatigue life due to
increased tension and associated friction forces generated by the helically
wound tube. This illustrates a very significant reduction in fatigue life and
demonstrates the large effect friction has on STU fatigue life in deeper
waters. The result suggests that an accurate STU friction model is nec-
essary for deep water floating production STUs.
16.3.4 Effect of Increasing the Tube Layer Radius

In order to examine the effect of different STU geometry (e.g., larger
distance from tube to centre of the umbilical) can have on STU fatigue
life, an analysis was performed at a layer radius twice that of the base
case model. The fatigue life obtained from this analysis is present in
Table 16.3.
Results from the increased layer radius analysis show that differences in
STU geometry have a large effect on the STU fatigue life. Doubling the
layer radius, in this case, had the positive effect of increasing the predicted
STU fatigue life by 32%. This result is further supported by Eq. (16.15),
and Figure 16.2 which demonstrates the tube contact force decreases with
increasing distance of the tube from the center of the umbilical.
Table 16.3 Comparing base case to sensitivities—all time domain.

% of base case % variation from
Sensitivity description fatigue life base case
Base Case 100% –
No Friction (µ = 0.2 0.0) 172% +72%
Infinite Friction (µ = 0.2 ∞) 13% −87%
Increased Friction (µ = 0.2 0.3) 77% −23%
Increased Water Depth (WD = 500 21% −79%
m 1,000 m)
Increased Layer Radius (RL = 0.027 132% +32%
m 0.055 m)
16.4 Conclusions
• An enhanced non-linear time domain STU fatigue model
has been developed. Comparisons have been made between
the enhanced non-linear time domain fatigue model and a
simplified approach. The enhanced model has been devel-
oped to account for non-linearities associated with the STU
friction as well as the correct phasing of the tension and
bending components of tube stress.
• The enhanced nonlinear time domain approach presents a
more advanced model of STU fatigue responses than sim-
plified models by modeling slip behavior and a phase con-
sistent time history of tube stress. A worked example of
an ExxonMobil West of Africa field STU has been used to
demonstrate the sensitivity of umbilical fatigue life to key
design inputs.
• The time domain modeling of friction where all non-lin-
earities are accounted for and correct phasing is used in the
analysis provides a more realistic model of fatigue behavior
than alternative simplified models.
• Differences in friction coefficient due to different materi-
als in contact produce significantly different fatigue lives,
emphasizing a) the importance of appropriate friction coef-
ficient selection and b) the potential benefit of achieving
contact surfaces with lower friction coefficients.
• Increased contact forces in a multi-layer STU result in
higher friction forces on inner layer tubes, and the potential
for higher fatigue damage in inner layer tubes. This effect
needs to be further quantified as an input to decision on the
radial placement of fatigue sensitive tubes.
• Fatigue design of STUs in deeper water present an increas-
ing challenge in terms of fatigue design, largely due to the
higher inter-layer contact force between tube and surround-
ing structure under higher hang-off tensions. Consequently,
for deepwater STUs, the accurate modeling of friction is an
important contributor to achieving confidence in umbilical
fatigue design.
• Frictionless modeling of the fatigue behavior of an STU
leads to unconservative predictions of fatigue life because it
ignores a potentially important contributor to tube fatigue
stress. This unconservatism is predicted to increase as the

water depth of the application increases.
• The selection of a suitable STU cross section design is
shown to be a significant input into achieving a feasible
fatigue design in deeper water. For a given friction coeffi-
cient, increasing tube layer radius is predicted to increase
fatigue life because of the lower inter layer contact force and
consequently lower friction force. This is typically the case
with larger diameter umbilicals and tubes further from the
umbilical center.
16.5 Recommendations
• The fatigue model presented in this chapter has been shown
to be a significant enhancement on existing simplified
methods. However, this work has highlighted several issues
worthy of further development, among them:
i. Further investigation of the uncertainties identified in the

second (bending) terms of Eq. (16.6) of this model.
ii. The enhancement of the existing enhanced friction model
to represent dynamic as well as static friction.
iii. The investigation of the potential for alternative enhanced
phase-consistent frequency domain methods as a predic-
tor of umbilical tube fatigue.
iv. The enhancement of the approach developed in this chap-
ter for the consideration of increased friction forces on the
inner layer tube under the effect of outer tube layer contact
forces in a multilayer umbilical.
v. The consideration of wave or motion directionality in
fatigue analysis of STUs.
vi. The consideration of the frictional contact force for a bent
umbilical by modifying Eqs. (16.6) and (16.15) with a con-
tact force determined for a bent helix.
• Ultimately, full scale testing of actual umbilicals would pro-

vide the best possible validation of enhanced STU fatigue
models such as is described in this chapter. Short of that,
the direct validation of tube friction coefficient through
more limited testing subject to varying tension, would also
contribute to a more thorough understanding of the design

issues related to how tubes interact in a STU.
References
1. Kavanagh W K, Doynov K, Gallagher D, et al. The effect of tube friction
on the fatigue life of steel tube umbilical risers-new approaches to evaluating
fatigue life using enhanced nonlinear time domain methods[C]//Offshore
Technology Conference. Offshore Technology Conference, 2004.
2. Hoffman J, Dupont W, Reynolds B. A Fatigue-Life Prediction Model for
Metallic Tube Umbilicals. Offshore Technology Conference, 2001.
3. Barltrop N D P, Adams A J. Dynamics of fixed marine structures. Butterworth-
Heinemann, 2013.
4. ASTM E-1049, “Standard Practices for Cycle Counting in Fatigue Analyses”,
American Society for Testing and Materials, Standard, 1997.
17
Cross-Sectional Stiffness for Umbilicals
17.1 Introduction
An umbilical cable is an important part of offshore oil and gas exploita-
tion project. It provides a communication and control link between a
submerged system and a surface vessel. It generally consists of various
functional elements for hydraulic, electrical power, and signal transmission
such as hydraulic tubes, electric cables, optical fibers, etc. The functional
elements surrounded by protective layers are normally wound in a heli-
cal manner around a central, large diameter, thin walled tube. The protec-
tive layers are designed to protect the functional elements from chemical
and mechanical damage which are generally either polymeric sheaths or
metallic armor wires. Generally, the helical structure is the main structure
of umbilical cables, and it has a great influence on mechanical behavior.
This helical structure is widely applied in the engineering as it is capable of
supporting high axial loads with comparatively low bending stiffness. As
a result, several researchers have worked on developing theoretical mod-
els for umbilical cables based on the knowledge of the helix material and
geometry.
The initial helical model was developed by Hruska [1–3]. The wires
in the model were assumed to be subjected to pure tensile forces (no
moments). The model was extended by Hruska to multi-strand wire
ropes for obtaining wire stress, and the interlayer pressure under tension
and torsion. In a 7 × 1 single strand model, Machida and Druelli [4]
accounted for the moments in helixes and gave explicit expressions of
axial force, bending and twisting moments in the helical wires. Knapp
[5, 6] adopted the well-known energy method to derive the equilibrium
equations and considered the compressibility and material nonlinearity
of the core element in helical armored cable. According to the above the-
ory, the resultant force and moment acting on the helix cross-sections
are obtained by accounting for the geometry and deformation of helical
359
wires. There is no resultant force and moment per unit length acting
along the helix.
Another fundamental approach for predicting the mechanical behavior
of helical structure is the application of Love’s [7] six nonlinear differential
equations describing the equilibrium of thin rods under both bending and
twisting. In a 7 × 1 single strand model, Costello and Philips [8] treated
the cables as groups of separate curved rods based on Love’s theory and
gave a rigorous derivation but assumed the extension of helical wires was
a small quantity of the order of the strains in the theory. Subsequently,
Costello and Philips [9] ended up removing this assumption. A math-
ematical model of a strand was developed by Utting and Jones [10] to
explore the importance of Poisson effects in wires, wire flattening under
interface pressure and the effect of friction between the core and helical
wires. In a comparable article [11], the theoretical results were compared
to previously published analytical work and the corresponding experi-
mental results. Witz and Tan [12] considered the umbilical cable as two
basic components: cylindrical elements and helical elements. The analy-
sis of cylindrical elements and helical elements are based on both thin
shell theory and Love’s theory. The continuity of interface pressure and
helical radius was considered to evaluate the load-displacement relation-
ship. Kumar and Botsis [13] made an attempt to experimentally test the
validity of the deformation derivation results earlier obtained for multi-
layered wire rope strands with metallic core. Earlier deformation relations
are utilized for predicting the maximum contact stress under tension and
torsion. Research undertaken in the mechanical behavior of umbilical
cables was presented by CP Pesce [14]. A mathematical model was also
developed combining equilibrium equations, geometric compatibility and
constitutive relations under external loads. This is generally the case when
working with complex cross-sectional designs. The behavior of a helical
wire on a frictionless cylindrical surface subjected to bending was studied
by Østergaard and Lyckegaard [15]. Gopinath [16] obtained the stiffness
matrix of a single-layered stranded cable by using the thin rod theory. The
linear elastic model under static loading conditions experiences the com-
bination of tension, torsion and bending.
In the theoretical model, lateral contact between wires as well as the fric-
tion conditions between the wires or between a wire and the core are always
negligible. To solve these problems finite element method is used to make
the model comparable with the practical condition of the umbilical cable.
Custodio and Vaz [17] presented a finite element formulation and
applied the principle of virtual work, as well as solving the Jacobi matrix by
Newton’s method for the umbilical model. The model takes into account
Cross-Sectional Stiffness for Umbilicals 361
a number of features, such as material nonlinearity, gap formation and

interface contact. A 3-D finite element formulation for predicting the
behavior of a complex umbilical cross-section exposed to combined load
was developed by Sævik [18]. Dixon and Zhao [19] proposed a 3D solid
model to capture local stress and contact between internal components
for fatigue analysis. Probyn [20] developed 3-D finite element analysis
models in ABAQUS. The components are modeled by C3D8R and S4R.
The analysis for umbilical under tension and bending is performed with
an explicit solver with general contact and coulomb friction. Lately, Corre,
and Probyn [21] used S4R and B31 instead of C3D8R and S4R to save the
calculation resources for complex sections of umbilical cables.
Most of researches done on helical structure were on helical wire
ropes. The materials and geometries of wires in each layer were the same.
The torsion in each helix was generally neglected when compared between
theoretical model and finite element model (FEM). Moreover, the finite
element model about umbilical cables applied dynamic explicit analysis.
In this chapter, based on the theoretical model of umbilical, the stiffness
and ultimate capacities of umbilicals under different loads are introduced.
17.2 Theoretical Model of Umbilicals

The typical structure of the umbilical is illustrated in Figure 17.1. For
the sake of solving the theoretical model, the umbilical structure can be
simplified to two parts: spiral unit (armored unit) and cylindrical unit
(cable core and polymer sheath). Since the cable core composed of inter-
nal components doesn’t withstand the load in point clamping, cable core
can be simplified to the cylinder by the equivalent of mechanical property.
Figure 17.1 Simplified umbilical model with winding angle.

As shown in Figure 17.1, the simplified umbilical model consists of the

cable core and n- layer spiral armored layers with winding angle αi(i = 1,
⋯, n).
To better describe the mechanical behavior of the umbilical, the theo-
retical model in the preliminary theoretical analysis makes the following
assumptions:
1. The elongation and torsion on the same cross-section of all

the layers are equal.
2. There is no slip between the layers when umbilical cable
is in the initial unstressed state.
3. All the materials are uniform and isotropic, and always are
within the linear elastic range.
4. The deformations of each part of the umbilical cable are
small deformations.
5. The inner wire is evenly distributed in the spiral armored
layer, and the bending and torsional stiffness are neglected.
6. Positive slip occurs between each layer.
17.3 Bending Stiffness of Umbilicals

The umbilical is a complex structure composed of multiple layers of dif-
ferent materials and spiral laying with different spiral angle. Due to the
unbonded assembly of the various structural layers in the umbilical, during
the bending process, with the increase of the curvature, the relative slip
between each layer will take place, and there is a lot of contact interface
and interface friction. For the different forms of the umbilical, the bending
deformation law is inconsistent and it will make it particularly difficult to
simulate the behavior of the umbilical. Due to the presence of interface
contact and friction, the bending stiffness of the umbilical cable is con-
stantly changing with the increase of bending curvature, which presents
nonlinear characteristics. From the theoretical prediction method, at pres-
ent, academia and industry generally assume that the sliding between each
unit is complete during the bending process of the umbilical. The complete
sliding model is adopted in this chapter to estimate the bending stiffness.
The bending theory model is as follows:
1. Witz bending theoretical model

This method assumes that the overall bending stiffness of the structure is
mainly from the spiral components and assumes that the bending behavior
of all the spiral components in the same layer is consistent, which is divided
into two kinds of bending states (before sliding and after sliding) to ana-
lyze respectively. There are two constraints to prevent the sliding between
components in this kind of the structure: structural constraint and fric-
tion constraint. Unless the structure is destroyed, the structure constraint
cannot be overcame to take place the sliding, and it can be overcame to
take place the sliding with a certain bending value. Therefore, when the
section of the spiral component is rectangular, due to the constraints of
the structure constraints, the spiral components will occur the torsional
deformation; when the cross-section is circular, the spiral structure will
overcome the friction constraint and produce the torsional deformation.
In the umbilical structure, the spiral component sections are all circular, so
its bending stiffness is obtained according to Witz bending theory.
( EI ) = ( EI )o + ∑ E2I (1 + cos α )
i =1
i i 2
i (17.1)
where (EI): overall bending stiffness of the structure;

(EI)o: bending stiffness of the cylinder structure;
n: number of the spiral structure;
EiIi: bending stiffness of the ith spiral structure;
αi: spiral angle of the ith spiral structure.
2. Costello bending theory model

From the overall structure, Costello extracts the single spiral structure to
conduct the pure bending study, and ignores the interaction and friction
slip between each component. Then the formula of bending stiffness of
the single spiral structure can be derived. Finally, the bending stiffness of
each single spiral structure can be superimposed and the overall bending
stiffness of the umbilical structure can be obtained.
( EI ) = ( EI )o + ∑ 22+EvI sin
i =1
cosα
α
i i
i
2
i
i
(17.2)

αi: spiral angle of the ith spiral structure;

vi: Poisson’s ratio of the ith spiral structure.
3. Roberto bending theoretical model

It is assumed that there is no circumferential contact between the spiral
components in the same layer and the strain is small enough. Once a series
of assumptions are satisfied, the bending stiffness of a spiral layer in the full
sliding state can be obtained. Then, the overall bending stiffness of all the
layers can be got by superposition calculation.
∑ G J + 32 (E I − G J )cos α  cosα

 2 
( EI ) = ( EI )o + i i i i i i i i (17.3)
i =1

GiJi: torsional stiffness of the ith spiral structure;
αi: spiral angle of the ith spiral structure.
In the full sliding model, it is assumed that the sliding occurs between
each layer. The bending stiffness consists of the spiral unit (armor, steel
pipe, cable, and optical cable) and cylindrical unit (sheath). The calculation
formula can be expressed:
kMB = η(kspiral + kcylinder) (17.4)
Since each structure layer exists the contact and friction in the full slid-
ing phase, according to the fold beam theory, this interlayer effect will influ-
ence the equivalent flexural center position of each component and has
impact on the bending stiffness. But the difficulty is that this influence is
difficult to derive and express through analytic formulas. Internationally, it
is generally used to construct a complex three-dimensional finite element
model to quantify the impact of this friction by building large computing
devices, such as large computers or fleets. However, because of the high
nonlinear calculation, a huge amount of calculation, and a very uncer-
tain process, the calculation strongly depends on the choice of computing
skills, which is still confined to the academic research and is different to
realize the popularization and application in engineering practice. To meet
the needs of a large number of modeling and calculation in engineering

design, the friction factor η is introduced to modify the above prediction
formula of bending stiffness, and some consider the effect of the interlayer
contact and friction. Specific value η needs combined with the production
process of umbilical and is determined by a lot of tests and engineering
application experience. In fact, this is also the technical route adopted by
the international large umbilical and marine cable manufacturers.
kspiral = ∑ 1 + cos cosα α sin α

2
i
i
2
i
(17.5)
i =1 +
2 EI 2GJ
m
E jπ
kcylinder = ∑ 64 (D
j =1
4
j − d 4j ) (17.6)
where EI: bending stiffness of the armored steel wire, for the circular
4
π Dsteel
cross-section of the steel wire, I = ; for the rectangle cross-section
64
bh3
of the steel wire, I = ; GJ: torsional stiffness of the armored steel wire,
12 4
π Dsteel
for the circular cross-section of the steel wire, J = ; for the rectangle
3 32
bh
cross-section of the steel wire, J = ;
3
j: number of the cylindrical layer;
D: outer radius of the cylindrical layer;
d: inner radius of the cylindrical layer.
From the formula above, it can be seen that:
1. The smaller bending stiffness and torsional stiffness of the

armor steel wire section, the larger the armor winding angle,
the smaller the bending stiffness of the umbilical, and the
better the bending stiffness.
2. The smaller the thickness of the polymer cover, the
smaller the bending stiffness, and the better the bending
performance.
17.4 Tensile Stiffness of Umbilicals

Since the umbilical has the helical structure, the axial tension and torsion
are mutually coupled. When the axial displacement occurs, it will result in
the axial elongation and torsion. The relationship between the axial tension
and displacement is expressed as below:
 ∆L 
 F   k11k12   
L 
 =    (17.7)
 M1   k21k22   ∆ϕ 
 L 
where F: axial tension;

M1: torque;
ΔL: axial tensile displacement;
Δφ: torsion angle;
L: umbilical length.
The expression of the stiffness matrix can be expressed as
n
 
k11 = ∑ A E  1 − Θ2 RR tan α  cos α + A E
i =1
i i
c
i
2
i
3
i c c (17.8)
k12 = ∑ A E R sinα cos α

i =1
i i i i
2
i (17.9)
n
 
k21 = ∑ A E  1 − Θ2 RR tan α  cos α sinα
i =1
i i
c
i
2
i
3
i i (17.10)
k22 = ∑ A E R cosα sin α + J G

i =1
i i i
2
i
2
i c c (17.11)
When Θ is equal to 0, the central cylinder is rigid; When Θ is equal to 1,

the central cylinder is incompressible;
where n: number of spiral units;

Ai: cross-section area of the ith spiral unit;
Ei: elasticity modulus of the ith spiral unit;
Rc: radius of the central cylinder;
Ri: circle radius of the ith spiral unit;
αi: spiral angle of the ith spiral unit;
AcEc: tensile stiffness of the central cylinder;
JcGc: torsional stiffnessof the central cylinder.
When calculating the tensile stiffness of the umbilical, two constraint
conditions are mainly considered: (1) No torsion can occur at both ends of
the structure; (2) Torsion can occur at both ends of the structure.
For the umbilical that no torsion can occur at both ends of the structure,
its tensile stiffness can be expressed as
 nL
 Θ Rc  
k11 = 

∑
i =1
ni Ai Ei  1 −
 2 Ri
tan 2 α i  cos3 α i  + Ac Ec
 
(17.12)
For the umbilical that torsion can occur at both ends of the structure,
its tensile stiffness can be expressed as
∆ϕ
k = k11 + k12 (17.13)
∆L
Generally speaking, when the umbilical cord occurs the axial displace-
ment, the angle of the negative direction will happen. Therefore, the cal-
culation result of the case that no torsion can occur at both ends of the
structure would be larger than the case that torsion can occur at both ends
of the structure.
For the cable wires with two fixed-ends, the tensile stiffness can be
expressed as
  nL
 Θ Rc  3  
kT = ksteel + kelse = χ  
 
∑
i =1
ni Ai Ei  1 −
 2 Ri
tan α i  cos α i  + Ac Ec 
2
  
(17.14)
From Eq. (17.8), the tensile rigidities of the steel pipe, cable wire, filling,
sheath units are related to the elastic modulus of the unit material and area.
According to the tensile stiffness expression of the armored wire, the
influencing factors affecting the tensile stiffness can include the following:
1. Layout of the cable core.

The greater the radial stiffness of the cable core, the better the radial
contraction can be controlled, and the better tensile properties of cables.
2. Property of armored steel wire.

The thickness, section shape, and material properties of the armored steel
wire can influence the number, area, and elastic modulus of the steel
wire, thus affecting the tensile stiffness of the cord cable. Through further
study, it is found that the influence of the tensile stiffness of the armored
wires is the spiral angle.
17.5 Torsional Stiffness of Umbilicals

According to the classical twist theory of spiral winding structure, the tor-
sional stiffness formula is as follows:
k= ∑ A E R sinα cosα + J G
j =1
i i i
2 2
i i c c (17.15)
where n: number of the spiral structure;

Ai: cross-section of the ith spiral structure;
Ei: elasticity modulus of the ith spiral structure;
Ri: circle radius of the ith spiral structure;
GiJi: torsional stiffness of the ith spiral structure.
17.6 Ultimate Capacity of Umbilicals

17.6.1 Minimum Bending Curvature
In the process of bending, the plane cross-section assumption takes the
yield strain of each unit as the failure criteria and chooses the maximize
value as the global minimum bending radius. The minimum bending

radius can be expressed as
D
Rmin = cosα i − R (17.16)
2ε max
where D: outer radius of the unit;

R: distance between the unit center and cable center;
εmax: material yield strain.
17.6.2 Minimum Tensile Load

Choosing the strain failure for different units (steel pipe unit, armored
unit, and polymer unit) as the failure criteria of umbilical, the maximum
tensions of each component are given. Selecting the minimum value of the
three as the minimum tensile load, the minimum tensile load formula can
be expressed as
ε max
F= ( EA)tensile (17.17)
cosα i
where F: minimum tensile load;

(EA)tensile: global stiffness matrix of the umbilical;
εmax: material yield strain.
17.6.3 Tensile Capacity Curve

In the application process of the umbilical, the most common load case is the
combination of the tensile and bending load. These loads work on the bear-
ing component of the umbilical-steel wire/steel tube to produce the greater
stress, which will cause the component to fail. For umbilical products, the
internal/external pressure is usually fixed, and the tensile-bending load is a
random combination. The capacity of the umbilical product to withstand the
tensile-bending combination load is an important indicator of its mechanical
properties. Therefore, the accurate assessment of the tension-bending capacity
curve of the umbilical is of great importance to the umbilical application.
The umbilical capacity to withstand the tensile-bending load can be

analyzed and evaluated through the key load bearing components. For
the umbilical, the steel pipe unit is both the functional component of the
hydraulic transmission or liquid conveying and also is the bearing compo-
nent of the internal pressure, external pressure, and tensile- bending load.
The tensile-bending load of the umbilical can be obtained through the
stress analysis of the steel tube.
1. Hoop stress
The hoop stress of the steel pipe is produced by the internal pressure of the
pipe, and the formula is
(D 2 + d 2 )
σ hoop = p (17.18)
(D 2 − d 2 )
where p: design internal pressure (vector of internal pressure and external

pressure);
D: outer diameter of the steel pipe;
d: inner diameter of the steel pipe.
2. Radial stress
Radial stress is produced by the internal pressure and external pressure,
and the calculation formula is
σradial = −p (17.19)
where p: design internal pressure (vector of internal pressure and external

pressure).
3. Axial stress
Axial stress can be produced by the tensile and bending load. Because of
the end sealing of the steel pipe, internal pressure can produce the axial
stress inside the steel tube. The method of determining the axial stress
components can be concluded to the following:
1. Calculation formula for the axial stress generated by the

direct tensile load:
F
σ ATension = (17.20)
A
where F: tensile load acting on the steel pipe;

A: total area of the steel pipe.

bending load:
(d + 2t )E
σ ABending = (17.21)
2ρ
where E: elasticity modulus of the steel pipe;

ρ: bending radius of the umbilical;
d: inner diameter of the steel pipe;
t: thickness of the steel pipe.

internal pressure acting on the end cap:
ρπ ID 2
σ AEndCap = (17.22)
4 ATube
where p: design internal pressure;

D: outet diameter of the steel pipe;
ATube: area of the steel pipe.
The axial stress of the steel pipe is the sum of each of these components,
that is to say:
σAxial = σATension + σABending + σAEndCap (17.23)
The load capacity of the steel pipe is determined by its ultimate stress
state. The hoop, radial, and axial stress were combined with the von Mises
yield criterion to check the stress was satisfied with the yield strength of
the material. The von Mises stress of the steel tube is met with the yield
strength of the internal pressure, stretching, and bending load, which is
the curve of the tensile–bending bearing capacity under certain internal
pressure of the umbilical.
The relation between the von Mises stress and each stress component
is shown as below:
2 2 2
σ e = σ radius + σ hoop + σ axial + σ hoopσ radius − σ hoopσ axial − σ radiusσ axial
(17.24)
When σe = γσyield (γ is the safety factor), the steel pipe reaches the critical
state.
When the internal pressure of the umbilical is determined, the hoop
stress component and radial stress component can be determined directly
by the formula above. Choosing the equivalent stress value in the critical
state of the steel pipe, Eq. (17.25) is a binary equation of the axial stress
σaxial:
2 2 2
σ axial − (σ hoop + σ radius )σ axial + σ radius + σ hoop − σ hoopσ radius − σ e2 = 0
(17.25)
To solve this quadratic equation, the critical value of the axial stress
of the steel pipe σaxial[σhoop(p), σradius(p), σe] is obtained under the speci-
fied internal pressure. Under the conditions of steel pipe parameters, the
critical value of the axial stress σaxial(p) is only the function of internal
pressure.
Eq. (17.23) shows that the axial stress of the steel pipe is a linear com-
bination of the axial stress component. Therefore, the axial stress is also a
function of tensile load and bending radius, that is to say, σaxial(p, σe) = f(f, ρ).
Therefore, the tensile-bending capacity curve of the umbilical is obtained by
the combination of different tensile-bending load.
References
1. Hruska F H. Calculation of stresses in wire ropes[J]. Wire and wire prod-
ucts, 1951, 26: 766–767, 799–801.
2. Hruska F H. Tangential forces in wire ropes[J]. Wire and wire products,
1953, 28(5): 455–460.
3. Hruska F H. Radial forces in wire ropes[J]. Wire and wire products,
1952, 27(5): 459–463.
4. Machida S, Durelli A J. Response of a strand to axial and torsional displace-
ments[J]. Journal of Mechanical Engineering Science, 1973, 15(4): 241–251.
5. Knapp R H. Derivation of a new stiffness matrix for helically armoured cables
considering tension and torsion[J]. International Journal for Numerical
Methods in Engineering, 1979, 14(4): 515–59.
6. Knapp R H. Nonlinear analysis of a helically armored cable with nonuniform
mechanical properties in tension and torsion[C]//OCEAN 75 Conference.
IEEE, 1975: 155–164.
7. Love A E H. A treatise on the mathematical theory of elasticity[M].
Cambridge university press, 2013.
8. Phillips J W, Costello G A. Contact stresses in twisted wire cables[J]. Journal

of the Engineering Mechanics Division, 1973, 99(2): 331–341.
9. Costello G A, Phillips J W. Effective modulus of twisted wire cables[J].
Journal of the Engineering Mechanics Division, 1976, 102(1): 171–181.
10. Utting W, Jones N. 1987a. The response of wire rope strands to axial tensile
loads—Part I. Experimental results and theoretical predictions. International
journal of mechanical sciences. 29: 605–619.
11. Utting W, Jones N. 1987b. The response of wire rope strands to axial tensile
loads—Part II. Comparison of experimental results and theoretical predic-
tions. International journal of mechanical sciences. 29: 621–636.
12. Witz J A, Tan Z. On the axial-torsional structural behaviour of flexible pipes,
umbilicals and marine cables[J]. Marine Structures, 1992, 5(2–3): 205–227.
13. Botsis J. Contact stresses in multilayered strands under tension and tor-
sion[J]. 2001.
14. Pesce C P, Ramos Jr R, Yamada da Silveira L M, et al. Structural Behavior
of Umbilicals—Part I: Mathematical Modeling[J]. ASME Paper No.
OMAE2010-20892, 2010.
15. Østergaard N H, Lyckegaard A, Andreasen J H. A method for prediction
of the equilibrium state of a long and slender wire on a frictionless toroid
applied for analysis of flexible pipe structures[J]. Engineering Structures,
2012, 34: 391–399.
16. Gopinath D, Shibu G, Thanigaiyarasu G. Theoretical estimation of stiffness
of stranded cable subjected to constrained bending[J]. International Journal
of Offshore and Polar Engineering, 2012, 22(04).
17. Custódio A B, Vaz M A. A nonlinear formulation for the axisymmet-
ric response of umbilical cables and flexible pipes[J]. Applied Ocean
Research, 2002, 24(1): 21–9
18. Sævik S, Gjøsteen J K. Strength analysis modelling of flexible umbilical
members for marine structures[J]. Journal of Applied Mathematics, 2012,
2012.
19. Dixon M A J, Zhao T. 3D modeling improves deepwater umbilical design
dependability[C]//Offshore Technology Conference. Offshore Technology
Conference, 2008.
20. Probyn I, Dobson A, Martinez M. Advances in 3-D FEA techniques for
metallic tube umbilicals[C]//The Seventeenth International Offshore and
Polar Engineering Conference. International Society of Offshore and Polar
Engineers, 2007.
21. Le Corre V, Probyn I. Validation of a 3-dimensional finite element analysis
model of a deep water steel tube umbilical in combined tension and cyclic
bending[J]. ASME Paper No. OMAE2009-79168, 2009.
18
Umbilical Cross-Section Design
18.1 Introduction
18.1.1 General
Early umbilical cable ducts used thermoplastic hoses, but as the water
depth increased, problems gradually emerged, such as fluid penetration,
water tightness, and the limitations of the hose body’s resistance to crush-
ing and stretching. Therefore, in the 1990s, the steel pipe umbilical cable
was introduced. The steel pipe material was carbon steel, duplex stainless
steel, super duplex stainless steel, etc., which greatly enhanced the mechan-
ical properties of the umbilical cable, and also increased the water depth
range of the umbilical cable.
18.1.2 Sectional Composition of the Umbilical Cable

The structure of a typical umbilical cable is shown in Figure 18.1. It mainly
includes functional components (such as steel pipes, cables, and optical
fibers), reinforcing members (such as armored steel wires, etc.) fillers, jack-
ets, etc. It should be noted that steel pipes can also be used. As a reinforcing
member, the mechanical properties of the umbilical cable are enhanced.
Steel pipe unit: Usually, the inner part is steel pipe, and the
outer layer is wrapped with polymer layer to avoid direct
contact between steel pipes. The main function is to pro-
vide hydraulic pressure for underwater production system
and chemical agents required for oil field development,
Cable unit: The inside is a stranded copper conductor, the outer
part is covered with an insulating layer and a shielding layer,
and the outermost layer is a polymer sheath layer mainly
for supplying electric energy or transmitting signals to the
underwater production system.
375
Figure 18.1 Example of subsea umbilical cable structure [1].
Fiber unit: The fiber material is easily damaged. It is usu-

ally placed in a seamless stainless steel tube and filled with
a water-blocking paste. The outer spiral-wound galvanized
steel wire is used for protection. In order to balance the tor-
sion, a double-layer steel reverse spiral is used. Winding, the
outermost layer, is a polymer sheath layer; the main func-
tion is to transmit the control signals of the water module
and the sensor data of the underwater production system.
Armoured steel wire: Galvanized steel wire, spirally wound
around the outer layer of umbilical cable, generally double-
layer, four-layer armor, is the main bearing member in the
umbilical cable and protects the internal functional unit.
Filler: There are many forms, such as plastic rope, polypro-
pylene rope, carbon fiber composite, etc., which makes the
structure more compact and compact.
Outer Sheath: Generally, used as a polymer material, mainly to
make the structure intact, reduce tube wear, and protect the
internal structure of the umbilical cable.
18.1.3 Umbilical Cable Structure Features

Spiral Structure
Spiral structures are widely used in engineering, such as steel strands and
marine flexible risers. A significant advantage of this type of construction
is the ability to withstand relatively high tensile loads while having a low
bending stiffness. Similarly, the umbilical cable design uses a spiral struc-
ture to combine components with different mechanical properties, which
not only ensures that the urethane cable has strong tensile properties but
also ensures that the umbilical cable has good bending flexibility so that
Umbilical Cross-Section Design 377
the umbilical cable can be entangled. It easy to transport and install on

the reel.
Multi-Layer, Multi-Contact, Slidable

Another feature of the umbilical cable structure is its multi-layered nature,
which is clearly defined from the inside to the outside, and the outer mem-
ber is spirally wound on the inner layer at an angle. The outer layer and the
inner layer member and the layer members are in contact with each other in
a non-bonded form. From the perspective of the umbilical cable as a whole,
the contact can be divided into two types: one between the layers and the
layers in the radial direction of the umbilical cable. Contact 2 is the contact
between adjacent members in a layer in the direction of the umbilical cable
loop. Because they are combined in a non-bonded manner, the umbilical cable
may be relatively slid between adjacent members when the force is applied
and even separation occurs between the members in contact with each other,
thereby analyzing the structure of the cable.
Designability
The design of the umbilical cable is embodied in two aspects, first of all, the
designability of the internal functional components. It can contain only a
single functional component, such as only the cable inside, the transmission
of electric or electrical signals, or the integration of multi-functional compo-
nents, such as cables, fibers, steel pipes, and other functional components so
that electricity can be used. Integrated transmission of signals and hydrau-
lics. The design of the internal functional components is mainly determined
by the needs of the umbilical cable to be served.
Secondly, the design can be embodied in the designability of the umbili-
cal cable structure, mainly to enhance the designability of the components.
The reinforcing member is mainly for providing mechanical protection
to the internal functional components of the umbilical cable so that the
umbilical cable meets the design requirements of strength and fatigue.
18.2 Umbilicals Cross-Section Design Overview

18.2.1 Umbilical Cross-Section Design Flowchart
According to the functional requirements and environmental load require-
ments put forward by the actual needs, the preliminary design of the
umbilical cable cross-section mainly includes functional unit design, sec-
tional layout design, and reinforcement mode design. The reinforcement
design is based on the requirement of the maximum tensile load of the
Unit Size
Functional Requirements Cross-Section Preliminary

Design Layout Design
Load Requirements
Reinforcement
Design
Bend Stiffness
Stiffness Analysis
Tensile Stifness
Response
Satisfaction NO
YES
Global Dynamic Cross-Section Detail

Analysis Design
Extreme Response Strength Check
Response
Satisfaction NO
YES
Fatigue Response Fatigue Check
Response
Satisfaction NO
YES
Finish
Figure 18.2 Design flowchart of umbilical cable cross-section [2].
umbilical cable in place, and the overall tensile mechanical properties

of the umbilical cable are designed to obtain a preliminary plan for the
section design, and then, the tensile and bending mechanical properties
are analyzed. The load requirements of the umbilical cable in place and
installation are meet. The general flow of the umbilical cable section design
is shown in the Figure 18.2.
18.2.2 Load Analysis

During use, the umbilical cable is first subjected to the laying and instal-
lation phase and then undergoes the in situ operation phase. The entire
process is subjected to tensile and bending loads caused by the functional

environment alone or in combination.
The installation and laying of the umbilical cable are carried out by
using a pipe-laying ship. As shown in the figure, the installation diagram
is shown in Figure 18.3. The umbilical cable is gradually laid on the seabed
through the escrow frame at the tail of the pipe-laying ship. If the pipe
unit is included in the umbilical cable, to enhance the pressure resistance
of the pipe during installation, the pressure installation is usually adopted
to ensure the safety of the pipe unit. While the pipelaying vessel moves, the
umbilical cable is lowered to the bottom of the sea. The umbilical cable is
in the form of a catenary in the water. A small radius of curvature is pres-
ent at the top and bottom, respectively, and the umbilical cable is subjected
to its gravity and gravity in the upper curved section. The tensile load
generated by the installation is subjected to the tensile load generated by
the installation in the lower curved section, so the most prone to failure is
usually located at the upper curved section connected to the hull and the
lower curved section connected to the seabed. A detailed load analysis
to determine the load size is performed.
After the umbilical cable is laid and put into use, it is in place and will
be affected by the function and environmental load. The area where the
umbilical cable is connected to the floating body bears the entire suspen-
sion gravity of the umbilical cable, and the connection is susceptible to
the effects of marine environmental loads such as waves and currents. The
umbilical cable must be distributed in the water in a certain line. Due to
the influence of the line type, the radius of curvature of a certain area of
the umbilical cable will be too small. The bending load is more serious
Tensioners
Overbend region S-lay barge
Waterline
Unsupported span Stinger
Touchdown point
Sagbend region
Seabed
Figure 18.3 Design flowchart of umbilical cable cross-section [3].

in this area, so the bending performance of the umbilical cable is also

designed. An important aspect of the umbilical cable section design must
be carried out.
In summary, the umbilical cable section design load is mainly the ten-
sile and bending loads encountered during the installation and in-position
process, so the tensile mechanical properties and bending mechanics of
the umbilical cable under the installation and in-position conditions are
required. Performance is designed and analyzed.
18.3 Umbilical Cable Cross-Section Design

The design phase of the umbilical cable section is a crucial part of the
overall design process of the umbilical cable. It determines the geomet-
ric properties, material properties, and basic mechanical properties
of the umbilical cable. It is related to the effectiveness of the umbil-
ical cable during use and safety. The umbilical cable section design
includes many aspects, including functional unit design, section lay-
out design, section tensile performance design, and section bending
performance design.
18.3.1 Umbilical Cable Cross-Section Layout Design

The design of the umbilical cable functional unit is mainly designed
according to the functions to be performed by the umbilical cable.
Generally, there is a single function and a multi-functional difference.
For example, the marine power umbilical cable has a single function,
and the internal-only contains the cable functional unit, and the func-
tion is only the submarine equipment transportation and electric power.
When the internal functional components of the umbilical cable are
designed, the components will be laid out for integration to make their
structure more rational and optimized for further design and analysis.
At present, there is no fixed standard for the layout of the umbilical cable
section, mainly based on the engineering experience. According to the
recommendations of the specification, the layout section should be as
compact and symmetrical as possible and can be filled with additional
components or fillers, generally filled with anti-crush material. For
high-performance umbilical cable, it can be filled with high-strength
and light-weight materials, which can increase the tensile strength of the
umbilical cable and make the cross-section full weight meet the require-
ments. The cable should generally be in the center of the umbilical cable.
If it contains steel pipes or hoses, the crushing force of the steel pipe
should be considered.
18.3.2 Tensile Performance Design

At present, offshore oil exploration and development is gradually devel-
oping in the deepwater area with richer oil reserves. The depth of the
deep-water marine environment also challenges the umbilical cable sec-
tion design technology. The increase in a water depth of the umbilical
cable application will inevitably lead to an increase in the weight of the
suspension at the top of the umbilical cable, thus placing higher demands
on the tensile performance of the umbilical cable. This requires increas-
ing the tensile strength properties of the umbilical cable while reducing
the gravity of the umbilical cable. At the same time, the subsea produc-
tion system is not only the demand for electricity but also fluid trans-
portation and signal control and transmission. As the functionality of
the umbilical cable increases, the number of internal functional units
will also increase, while the functions of cables and optical fibers will
increase. The increase in the unit will only increase the outer diameter
and weight of the umbilical cable without increasing the tensile strength
properties of the umbilical cable, thereby placing higher demands on the
tensile properties of the umbilical cable. In general, the umbilical cable
section tensile performance design is an important part of the umbilical
cable section design mainly depends on the reinforcement of the section,
including armored steel wire reinforcement, steel bar reinforcement, and
high-strength material reinforcement.
The traditional umbilical cable cross-section structure is generally a
functional unit inside, and the outer part is protected by the armor of low
carbon steel material. On the one hand, the tensile mechanical properties
of the umbilical cable are increased, and on the other hand, the lightweight
of the umbilical cable can be increased. It meets the needs of subsea sta-
bility. However, the armored umbilical cable is generally used in shallow
water areas. With the increase of water depth, traditional steel wire armor
cannot increase the tensile strength of the umbilical cable while increas-
ing the weight and outer diameter of the umbilical cable. Due to the large
water depth in the deepwater area and the ultra-deepwater area, to enhance
the tensile strength of the umbilical cable and make the weight of the
umbilical cable meet the requirements, the external reinforcement is not

carried out by steel wire armor but is strengthened by high-strength and
low-density materials.
If the internal functional unit of the umbilical cable contains a steel pipe
functional unit, the super-duplex stainless steel material with high tensile
strength is usually used to replace the traditional material to make the steel
pipe, so that the steel pipe unit serves as both a functional unit and a rein-
forcing unit, thereby avoiding low strength. The low carbon steel wire is
armored. The table gives a comparison of the properties of various steel
materials. It can be seen from the table that the three materials have the same
elastic model and density, but the yield strength is very different. Therefore,
for the umbilical cable of the same structural form, the three kinds of steel
materials are respectively used, and the overall section tensile stiffness is the
same, that is, the same tensile load of the umbilical cable is the same for the
same tensile load within the allowable range, but the strength of the umbil-
ical cable using the super duplex stainless steel is obviously larger than the
two outer two, which is suitable for deeper sea area.
In addition, the umbilical cable used in the ultra-deepwater region in
the world is made of high elastic modulus and low-density materials to
make the component unit for filling and strengthening, so that the overall
section tensile stiffness mechanical properties of the umbilical cable are
greatly improved, and the weight is not increase too much. The material
property values generally used to fill the reinforcing members are shown
in the table. It can be seen from the table that polymer filled rope and
carbon fiber composite have better mechanical properties than steel wire
rope and are more suitable for deepwater marine umbilical cable design.
18.3.3 Bending Performance Design

Another critical mechanical property of the umbilical is the bending per-
formance, which is subject to bending loads during storage, installation,
and in situ operation of the umbilical. Therefore, the umbilical cable struc-
ture should be designed to ensure sufficient flexibility of the umbilical.
The conditions of bending loads under installation and in-place condi-
tions are meet. In the umbilical cable cross-section design, spiral structure
winding and non-bonding are usually used to improve the flexibility of
the umbilical cable structure without reducing the tensile properties of the
umbilical cable.
Spiral Structure Winding

The spiral structure has been used in a large number of practical projects
and life, the most common such as steel strands. For a straight cylin-
drical steel bar of the same cross-sectional area, the bending stiffness
of the steel strand is much smaller, so the strand can be wound on the
disc at a certain radius without failure. Drawing on the structural form
and characteristics of steel strands, the spiral winding is also widely used
in umbilicals.
Straight members are bent and deformed along the neutral axis when
subjected to a bending load. When the spiral member is subjected to bend-
ing, not only bending deformation but also torsional stress generated
by the bending load is generated in the cross-section, thereby generat-
ing torsional deformation, and the final bending deformation of the spiral
member under the bending load will be two. The superposition of the per-
son, therefore, for the spiral member and the straight member of the same
section, the bending rigidity of the former is smaller than the latter, and
the larger the helix angle, the smaller the bending rigidity of the member,
and the better the flexibility.
Unbonded Composite
Another way to increase flexibility in the umbilical is that the compo-
nents are combined in a non-bonded manner, which means that when
the umbilical is bent to a certain extent, relative sliding between the
components occurs. If the members in the umbilical cable adopt a spiral
structure, but the components are not combined in a non-bonded man-
ner but are completely consolidated together, then when bending, the
components in the umbilical cable will be integrated as a whole, and If
the central axis of the cable is used as the curved neutral axis, if the non-
bonded form is adopted, the bending will also gradually deform around
the respective axes of the members after the sliding occurs between the
members, so that the bending moment of inertia is significantly reduced
and the bending rigidity is also drastically reduced. Studies have shown
that in the initial stage of bending of the umbilical cable, there is no slid-
ing between the components due to the contact friction. At this time,
the members are bonded together, and the bending stiffness is at a large
value. When the bending continues to a certain extent after that, the
members gradually overcome the maximum static friction and slide, and
the bending stiffness will decrease rapidly.
References
1. API 17E. Speccification for Subsea Production Control Umblica. 2010.
2. ISO 13628-5:2009 Petroleum and natural gas industries - Design and opera-
tion of subsea production systems - Part 5: Subsea umbilicals.
3. Beattie, M., et al., Caesar-Tonga Project Steel Lazy Wave Riser Design, OTC
24232, 2013.
Part 3
FIBER GLASS REINFORCED DEEP
WATER RISERS
19
Collapse Strength of Fiber Glass
Reinforced Riser
19.1 Introduction
Fiber glass reinforced bonded flexible pipe (FGRFP) has a lot of advan-
tages compared with other composite pipes. According to the produc-
tion process of composite pipes, composite pipes can be divided into two
types: unbonded flexible pipe and bonded flexible pipe. The remarkable
difference between unbonded flexible pipe and bonded flexible pipe is
that there is a hollow annuls in the pipe wall of unbonded flexible pipe,
while bonded flexible pipe have not hollow annuls in its pipe wall. This
difference makes the unbonded flexible pipe is easier to form high pres-
sure in hollow annuls than bonded flexible pipe. The high pressure in
annuls will lead to inner layer and outer cover burst. In addition, FGRFP
also has an advantage on cost. Therefore, FGRFP has a great value in
ocean engineering.
Many scholars have developed some mechanic analysis methods on cyl-
inder shells since nineteenth century. Bryan, G.H. [1] used the energy test
to analyze the collapse of long-thin tubes under external pressure. The form
of collapse pressure conducted by Bryan is P = c(t/d)3. Carman [2] studied
the collapse pressure of short-thin tubes by using experiment methods. He
also deduced a critical length formula. Windenburg et al. [3] discussed the
collapse pressure of three classes of pipes, which are infinite length pipe,
finite length pipe subjected to external pressure only, and finite length pipe
with external pressure and axial pressure.
In addition, there has been a lot of research done about flexible pipes
under external pressure. Kaveh Arjomandi et al. [4–7] have studied
the stability, post-buckling, and elastic buckling capacity of sandwich
pipes under eternal pressure. Yong Bai et al. [8–13] has analyzed buck-
ling stability of steel strip reinforced thermoplastic pipe and unbonded
multi-layer pipe subjected to external pressure and bending moment.
387
Leone Corradi et al. [14, 15] studied the collapse capability of thick
cylinders under radial pressure and axial load. Gong Shunfeng et al.
[16–18] have investigated asymmetric buckling of offshore pipelines
under combined tension, bending and external pressure. Rouzbeh
Hashemian et al. [19, 20] studied the buckling capability of sandwich
pipes by means of finite difference model and finite element formu-
lation. At present, the study was focus on the buckling capability and
limit strength of flexible pipes.
However, little research has been done on the mechanics of FGRFP.
The collapse capability and limit strength of FGRFP is still not so clear.
As we all known, the external pressure will increase along with the
depth of the water. Therefore, it is urgent to analyze the mechanism
of FGRFP for better usage. This chapter will establish the equilibrium
equations of FGRFP under external pressure based on the theories of
nonlinear buckling and virtual work principle for numerical analysis.
In addition, the external pressure test and finite element analysis will
be also conducted.
19.2 External Pressure Test

We have conducted external pressure tests to obtain the buckling pressure
of FGRFP and verify the accuracy of the numerical analysis.
19.2.1 Testing Specimen

To conduct the external pressure test, we have taken three testing speci-
mens of FGRFP. The dimensions of the specimens are listed in Table 19.1.
Table 19.1 Dimension of specimens.

Specimen 1 Specimen 2 Specimen 3
Length (mm) Inner 885 1,000 1,000
Diameter (mm) Outer 50 50 50
Diameter (mm) 76 76 76
Collapse Strength of Fiber Glass Reinforced Riser 389
19.2.2 Testing System

The external pressure testing system mainly consist of two parts: the first
part is the External Pressure Cylinder, as shown in Figure 19.1a; the second
part is the recording system for recording the external pressure and inner
volume as the FGRFP changes, as shown in Figure 19.1b.
The testing procedure is as follows. First, we put the specimens with
flanged joint into the external pressure cylinder and seal the joint. Then,
we apply the external pressure to FGRFP by using a pressure water pump.
The rate of external pressure is controlled at 1 MPa/min. At the same time,
the reading of pressure gage will be recorded. The FGRFP will reach the
buckling pressure as the external pressure continues increasing. The read-
ing of the pressure gage will decrease suddenly when the FGRFP reaches
the buckling pressure. What’s more, the inner volume of FGRFP will also
be changed suddenly when the FGRFP reaches the buckling pressure. The
change of FGRFP’s inner volume can be recorded by the inner volume
recording equipment.
19.2.3 Testing Results

The failure signal of FGRFP under external pressure is the abrupt change
of inner volume and external pressure. The buckling shape of FGRFP is
shown in Figure 19.2. In addition, we can also obtain the buckling pressure
and inner volume change of FGRFP from the external pressure test. The
(a) External Pressure Cylinder (b) Inner volume of FGRFP recording

equipments
Figure 19.1 External pressure testing system.

Figure 19.2 Testing specimens after buckling.
Table 19.2 The results of external pressure test.

Specimen 1 Specimen 2 Specimen 3 Average
Buckling Pressure 18 20 17 18.33
(MPa)
Volume change (L) 0.51 0.56 0.46 0.51
average buckling pressure of three specimens is 18.33 MPa, as shown in

Table 19.2.
From Figure 19.2, we can see that the bucking shape of FGRFP is flat-
tened by the external pressure. The mainly reason is that there are material
or geometric imperfections in the body of FGRFP.
19.3 Theoretical Analysis

19.3.1 Fundamental Assumptions
Being a kind of composite pipe, the microstructure of FGRFP is complex.
Therefore, the theoretical analysis is based on the following assumptions:
• The macrostructure of FGRFP is constructed by three layers,

which are inner PE layer, outer PE layer, and reinforced layer.
• The material of the inner and outer layers is homogenous

and continuous.
• The reinforced layer is regarded as homogeneous and con-
tinuous at macro level.
• There is no slippage among the interface between the three
layers.
• The cross-section of FGRFP keeps perpendicular to the cen-
tral axis of the pipe.
• The effect on materials caused by temperature and load time
can be neglected.
19.3.2 Constitutive Model of Materials

1. Constitutive Model of Inner and Outer Layers
According to basic assumptions and the isotropic hardening J2 flow theo-
ries, without considering the influence of σ r , τθ z and τ rz. σ r represents the
radial stress of the inner and outer layers, τθ z and τ rz represent shear stress,
the constitutive formula can be expressed as follows:
 σ   D D12 D13   ε z 
 z   11  
 σ θ  =  D21 D22 D23   εθ  (19.1)
 τ   D D32 D333   γ θr 
 θr   31 
where, Dij = φ(σij, Q, υ), (i, j=1,2,3);
 0, σ e < σ e ,max

Q= 1  E 
 4σ 2  E − 1 σ e ≥ σ e ,max
 e t
υ is Poisson’s ration; E is elasticity modulus; Et is tangent modulus; σe is

Mises equivalent stress; σe,max is maximum equivalent stress.
2. Constitutive Model of Reinforcement Layer

According to the classical lamination theory, we can regard the reinforce-
ment layer as homogeneous and continuous at macro level. According to
Halpin-Tsai’s model, we took a representative volume unit of reinforced
layer to obtain the material parameter of the reinforced layer. The repre-
sentative volume unit of the reinforced layer can be shown in Figure 19.3.
Figure 19.3 Representative volume unit of reinforced layer.
Therefore, the material parameter of reinforcement layer can be obtained

from the following formulas:
EL = EPEVPE + Efg(1 – VPE) (19.2)
μL2 = μPEVPE + μfg(1 − VPE) (19.3)
M 1 + ξηV fg
= (19.4)
M m 1 − ηV fg
 M fg   M fg 
η= −1  M + ξ  (19.5)
 Mm  m
ξE = 2, ξG = 1 (19.6)
where EL represents the elasticity modulus of reinforced layer along the

fiberglass direction. μL2 represents the Poisson’s ratio of the reinforced
layer along the fiberglass direction. VPE is the volume fraction of PE in the
reinforced layer. Vfg is the volume fraction of fiberglass in the reinforced
layer. M represents one of the parameters E2, μ2L, and GL2. The parameters
E2, μ2L, and GL2 are the elastic constant of reinforced layer perpendicular
to the fiberglass’s direction. Mfg represents corresponding elastic constant
of fiberglass, such as Efg, μfg, and Gfg. Mm represents corresponding elastic
constant of HDPE, such as EPE, μPE, and GPE. ξ represents the reinforce effect
of fiberglass for reinforced layer. ξ is an experience factor. It depends on
the shape of cross-section of fiberglass, the arrangement mode of fiber-
glass and loading methods. In this study, we take the experience factor of
elasticity modulus ξE and Poisson’s ratio ξμ as 2, the experience factor of

shear modulus ξG as 1.
19.3.3 Establish the Equations of Motion

We simplified the cross-section of FGRFP to deduce the equations of
motion. The simplified schematic of FGRFP can be shown in Figure 19.4.
ε x = ε x0 + ζκ (19.7)
ζ = (R + w)cos θ – v sin θ + z cos θ. (19.8)
ε x0 is the axial strain at neutral plane. ζ is the distance between the point
of εx and neutral plane.
The formula of hoop strain is
2
 v′ + w  1  v′ + w  1  v + w′   γ′
εθ =   +   +   + z ⋅  κ θ + θ  (19.9)
 R  2 R  2 R   R
2
 v ′ − w ′′   v − w′ 
κθ =  1−  (19.10)
 R 2   R 
y w.z
t
R
θ
Figure 19.4 Cross-section of FGRFP.

19.3.4 Establish Virtual Work Equations

According to the nonlinear buckling theories and virtual work principles,
we established the equilibrium equations as shown in the following:
2π  N l t2 (i ) 
∫ ∑∫
0


 i =1 t1 (i )
(σ xδε x + σ θδεθ + τ θrδγ θr )(R + z )dz  dθ = δWe

(19.11)
2π
 1 
 ∫
δWe = PδV = PR δ w + (2wδ w + 2vδ v + wδ v ′ + vδ w − vδ w ′ − wδ v ) dθ
2R
0

(19.12)
2π
 1 
V = π R2 + R
∫0
w + 2 R (v − vw + vw + w ) dθ (19.13)
19.4 Numerical Analysis

According to the constitutive equations, motion equations, and virtual
work equations, it’s feasible to obtain the stress and strain of FGRFP under
certain external pressure. However, the equations as foresaid are compli-
cated for hand calculation. Therefore, we take the software Matlab, which
has powerful calculation and program capability, as a tool to solve those
equations. The flow chart of Matlab program is shown in Figure 19.5.
The dimension and material parameters of Fiber Glass Reinforced
Flexible Pipe are listed in Tables 19.3 and 19.4.
According to the theoretical analysis, the ovality and external pressure
curve of FGRFP can be obtained by using Matlab calculation procedure.
The ovality and external pressure curve of FGRFP is shown in Figure 19.6.
From the ovality and external pressure curve, we can see that this curve can
be divided into three parts. The first stage is the external pressure increases
linearly along with ovality. We can call this stage the linear elastic stage.
The second stage is the external pressure increases slowly while the ovality
increases fast. We can call the second stage the plastic stage. The exter-
nal pressure will reach the maximum. The third stage is when the external
Start
Entering dimension and

material parameters
No
End
Load external pressure

Yes
Terminate
Joint displacement: Ui load step?
Joint strain and stress: εi , σi
Equivalent stress: σe
σi>σe Update and output:

No Current pressure: Pi
Displacement: U
Unload Stress: σ
Update
Strain: ε
Newton-Raphson Iteration Equivalent stress: σe
No Yes
Convergence
criterion
Figure 19.5 The flow chart of Matlab program.
pressure decreases as the ovality increases. We can call this stage the buck-
ling stage. The ultimate external pressure bearing capacity is 17.58 MPa.
19.5 Finite Element Analysis

To verify the validity of the numerical analysis method, we established a
3D finite element model of FGRFP by using ABAQUS which is a software
for finite element analysis. We used the dimension and material parame-
ters in Tables 19.3 and 19.4 to establish the finite element model.
Table 19.3 Dimension of FRGFP.

Inner Outer
Structural diameter Thickness diameter
layer Material and no. (mm) (mm) (mm)
Inner layer HDPE 50 4 58
Reinforced Fiber Glass 58 6 70
layer reinforced
HDPE
Outer layer HDPE 70 3 76
Table 19.4 Material parameters of FGRFP.

Elasticity Yield strength
Material Position modulus (MPa) (MPa)
HDPE Inner layer and 850 23
Outer layer
Fiber Glass Reinforced layer 33,000
20
External Pressure (MPa)
15
10
0
0 0.05 0.1 0.15 0.2 0.25
Ovality (%)
Figure 19.6 Ovality-external pressure curve of FGRFP by using numerical analysis

method.
19.5.1 Establish the Finite Element Model

The cross-section structure of FGRFP mainly consists of three layers: inner
layer, outer layer, and reinforced layer. In the modeling process, we used
the element type C3D8R to simulate the mechanical behavior of the each
layer. Then, we assigned the material parameters to each layer individually.
It should be noted that we use the Composite Layup Module in ABAQUS

to establish the finite element model of the reinforced layer. The reinforced
layer consists of fiber glass and matrix resin. Therefore, we regarded the
reinforced layer as a composite layer. There are two analysis methods in
ABAQUS for analyzing composite materials. The first method is to estab-
lish the fiber model and the matrix resin model and set the contact method
between them. The second method is use the Composite Layup Module
in ABAQUS to establish the model of composite material. The second
method does not need to establish the fiber and matrix resin model. The
first method is suitable for analyzing the micromechanical behavior of
composite materials. However, the second method is suitable for analyzing
the macro mechanical behavior. Therefore, we use the Composite Layup
Module for this analysis.
In addition, we divided the finite analysis step into two steps to obtain
the mechanical behavior of FGRFP under external pressure. The first step
is modal analysis to obtain the buckle mode of FGRFP and provide the
geometric imperfections for the next analysis. The second step is Static-
Riks analysis to get the load-deformation curve and the buckling load
under geometric imperfections conditions.
19.5.2 The Results of the Finite Element Analysis

1. Buckling Mode of FGRFP
We can obtain the buckling mode of FGRFP by means of the first step anal-
ysis of ABAQUS. The first mode of FGRFP subjected external pressure is
shown in Figure 19.7. Comparing Figures 19.7 and 19.2, we can see that the
buckling shape of finite element model is similar as the external pressure
testing. Then, we can apply the results of buckling model to the Static-Riks
analysis as an imperfection to obtain the ultimate strength of FGRFP.
2. The Result of Static-Riks Analysis

Through the Static-Riks analysis, we can get the relationship between exter-
nal pressure and the displacement when the geometric imperfections and
external pressure added on the finite element mode of FGRFP. The curves
of the external pressure and node displacement of FGRFP are shown in
Figures 19.8 and 19.9.
From Figure 19.8, we can see that the node displacement at the top of
pipe’s section decreasing along with the external pressure at first. Then, the
node displacement at the top of the pipe’s section is increasing along with
the external pressure when the external pressure is around at 18.86 Mpa.
That is because there is an imperfection in the radius direction loaded at
Step: buckle0523-1
Mode 1: EigenValue = 1.2871
Figure 19.7 First buckling mode of FGRFP subjected external pressure.
25
20
15
10
0
–0.5 0 0.5 1 1.5
The node displacement at the top of pipe’s cross section (mm)
Figure 19.8 The curve of external pressure and node displacement at the top of pipe
section.
the waist of the pipe’s cross-section, as shown in Figure 19.10. When FGRFP
is subjected to external pressure and there are already imperfections at first,
the external pressure is in a dominant position. The node at the top of the
pipe’s cross-section will move to the center of the cross-section under the
external pressure. Therefore, the node displacement at the top of the pipe’s
cross-section will be negative at first. Because of the imperfection at the
waist of the pipe’s cross-section, the node displacement at the waist of the
pipe’s cross-section increased faster than the node displacement at the top
of the pipe’s cross-section, as shown in Figure 19.9. The difference in node
displacement at the top and waist of the pipe’s cross-section gives a raising
force to the top node of the pipe’s cross-section. The node displacement at
25
20
15
10
0
–3.5 –3 –2.5 –2 –1.5 –1 –0.5 0
The node displacement at the waist of pipe’s cross section (mm)
Figure 19.9 The curve of external pressure and node displacement at the waist of pipe.
Imperfection Imperfection
Figure 19.10 Diagrammatic drawing for the imperfection’s location.
the top of the pipe’s cross-section will gradually become positive when the
raising force is greater than external pressure. From Figures 19.8 and 19.9,
we can get that the buckling external pressure of FGRFP with imperfection
is around 20.99MPa.
3) Stress of FGRFP
The FGRFP is consisted of three layers: inner layer, reinforced layer, and
outer layer. The stress nephogram of each layer is shown in Figure 19.11.
From the stress nephogram of FGRFP, we can see that the inner layer
and outer layer will reach the ultimate strength, 23Mpa. The reinforced
layer has eight layers. We can name the eight layers from inside to outside
S, Mises
(Avg: 75%)
+2.300e+01
+2.269e+01
+2.237e+01
+2.205e+01
+2.174e+01
+2.142e+01
+2.110e+01
+2.079e+01
+2.047e+01
+2.015e+01
+1.984e+01
+1.952e+01
+1.920e+01
(a) The stress nephogram of outer layer
S, Mises
(Avg: 75%)
+2.300e+01
+2.217e+01
+2.134e+01
+2.051e+01
+1.968e+01
+1.886e+01
+1.803e+01
+1.720e+01
+1.637e+01
+1.554e+01
+1.471e+01
+1.388e+01
+1.305e+01
(b) The stress nephogram of inner layer
S, Mises
Envelope (max)
(Avg: 75%)
+3.339e+02
+3.143e+02
+2.948e+02
+2.752e+02
+2.557e+02
+2.361e+02
+2.165e+02
+1.970e+02
+1.774e+02
+1.579e+02
+1.383e+02
+1.188e+02
+9.921e+01
(c) The stress nephogram of reinforced layer
Figure 19.11 The stress nephogram of FGRFP.
as the first reinforced layer to the eighth reinforced layer. The stress at bot-
tom and top side of the eighth reinforced layer has the minimum value
in the reinforced layer, 99.21Mpa. The stress at the bottom and top side
350
top
300 waist
250
Stress (MPa)
200
150
100
50
1 2 3 4 5 6 7 8
The layer number
Figure 19.12 The stress curve at top and waist of pipe’s cross-section change with
thickness.
of the first reinforced layer has the maximum value in reinforced layer,
333.9 Mpa. We can extract the stress at top side and waist side of the pipe’s
cross-section for each reinforced layer. The stress change along with the
thickness of FGRFP is shown in Figure 19.12. The stress at the top of pipe’s
cross-section decreases from the inner reinforced layer to the outer rein-
forced layer. The stress at waist of pipe’s cross-section is increass from the
inner reinforced layer to the outer reinforced layer.
19.6 Conclusion
The buckling pressure of FGRFP is 18.33 MPa in external pressure
test. The buckling pressure of FGRFP is 17.58 MPa by using numerical
method. The relative error is 4.09% between external pressure test and
numerical method. The buckling pressure of FGRFP is 20.99 MPa by using
finite element method. The relative error is 14.51% between external pres-
sure test and finite element method.
The reasons causing the difference between the three methods are listed
as follows:
• The boundary conditions of external pressure test, numer-

ical method, and finite element method are different from
one another.
• There are some assumptions when we conduct the numeri-

cal analysis.
• The specimens are not perfect pipes. There are some geo-
metrical imperfections or material imperfection in the
specimens.
Even though there are some differences in those three methods, the
results of buckling pressure of those three methods can verified each other
to some extent. Therefore, those three methods in this chapter can be used
to analyze the buckling pressure of FGRFP.
References
1. G.H., B., 1888. Application of the Energy Test to the Collapse of a Long Thin
Pipe Under External Pressure. Cambridge Phill. Soc. Proc. 6, 287–292.
2. Carman, A.P., 1916. LXI. The collapse of short thin tubes. The London,
Edinburgh, and Dublin Philosophical Magazine and Journal of Science 32
(192), 559–566.
3. Windenburg, D.F., Trilling, C., 1934. Collapse by instability of thin cylindri-
cal shells under external pressure. Trans Asme 11, 819–825.
4. Arjomandi, K., Taheri, F., 2010. Elastic buckling capacity of bonded and
unbonded sandwich pipes under external hydrostatic pressure. Journal Of
Mechanics Of Materials And Structures 5 (3), 391–408.
5. Arjomandi, K., Taheri, F., 2011a. The influence of intra-layer adhesion con-
figuration on the pressure capacity and optimized configuration of sandwich
pipes. Ocean Engineering 38 (17-18), 1869–1882.
6. Arjomandi, K., Taheri, F., 2011b. A new look at the external pressure capacity
of sandwich pipes. Marine Structures 24 (1), 23–42.
7. Arjomandi, K., Taheri, F., 2011c. Stability and post-buckling response of
sandwich pipes under hydrostatic external pressure. International Journal
Of Pressure Vessels And Piping 88 (4), 138-148.
8. Bai, Y., Igland, R.T., Moan, T., 1997. Tube collapse under combined external
pressure, tension and bending. Marine Structures 10 (5), 389–410.
9. Bai, Y., Liu, T., Cheng, P., Yuan, S.A., Yao, D.Z., Tang, G., 2016a. Buckling
stability of steel strip reinforced thermoplastic pipe subjected to external
pressure. Composite Structures 152, 528–537.
10. Bai, Y., Tang, J.D., Xu, W.P., Cao, Y., Wang, R.S., 2015a. Collapse of reinforced
thermoplastic pipe (RTP) under combined external pressure and bending
moment. Ocean Engineering 94, 10–18.
11. Bai, Y., Wang, N.S., Cheng, P., Qiao, H.D., Yu, B.B., 2015b. Collapse and
Buckling Behaviors of Reinforced Thermoplastic Pipe Under External
Pressure. Journal Of Offshore Mechanics And Arctic Engineering-

Transactions Of the Asme 137 (4), 9.
12. Bai, Y., Wang, P., Xiong, H.C., Tang, G., 2016b. Mechanical behavior of pipe
reinforced by steel wires under external pressure. Journal Of Reinforced
Plastics And Composites 35 (5), 398–407.
13. Bai, Y., Yuan, S., Cheng, P., Han, P.H., Ruan, W.D., Tang, G., 2016c. Confined
collapse of unbonded multi-layer pipe subjected to external pressure.
Composite Structures 158, 1–10.
14. Corradi, L., Ghielmetti, C., Luzzi, L., 2008. Collapse of Thick Tubes
Pressurized From Outside: An Accurate Predictive Formula. Journal of
Pressure Vessel Technology 130 (2), 021204-021204-021209.
15. Corradi, L., Luzzi, L., Trudi, F., 2004. Collapse of Thick Cylinders Under
Radial Pressure and Axial Load. Journal of Applied Mechanics 72 (4),
564–569.
16. Gong, S.F., Hu, Q., Bao, S., Bai, Y., 2015. Asymmetric buckling of offshore
pipelines under combined tension, bending and external pressure. Ships and
Offshore Structures 10 (2), 162–175.
17. Gong, S.F., Ni, X.Y., Bao, S., Bai, Y., 2013. Asymmetric collapse of offshore
pipelines under external pressure. Ships and Offshore Structures 8 (2),
176–188.
18. Gong, S.F., Sun, B., Bao, S., Bai, Y., 2012. Buckle propagation of offshore pipe-
lines under external pressure. Marine Structures 29 (1), 115–130.
19. Hashemian, R., Mohareb, M., 2016a. Buckling finite element formulation for
sandwich pipes under external pressure. International Journal Of Pressure
Vessels And Piping 147, 41–54.
20. Hashemian, R., Mohareb, M., 2016b. Finite difference model for the buckling
analysis of sandwich pipes under external pressure. Ocean Engineering 122,
172–185.
20
Burst Strength of Fiber Glass
Reinforced Riser
20.1 Introduction
Composite pipes have been widely used in marine oil and gas transpor-
tation. They are normally divided into two main categories: unbonded
flexible pipes and bonded flexible pipes [1–4]. Fiber Glass Reinforced
flexible pipes (FGRFPs), as a new-styled bonded flexible pipe, overcome
some flaws like high-priced production and overweight property belong to
conventional unbonded flexible pipes, having better industry application
potential and development value [5–7].
Typical FGRFP are composited of inner PE layer, outer PE layer,
and fiber glass reinforced layers sandwiched in the middle, as shown in
Figure 20.1.
Internal pressure is the most common load during the service of
flexible pipes and risers, Bai et al. [8, 9] did some theoretical analy-
sis about the mechanical behavior of metallic strip flexible pipes and
umbilical under internal pressure based on elastic theory and principle
of virtual work. Fan et al. [10] investigated the change of internal pressure
while filament wound reinforcement tube has been damaged. However,
the analysis of mechanical behavior of FGRFP under internal pressure is
still not enough. From the above, to obtain the mechanical responses of
FGRFP under internal pressure, a short-term burst pressure test is carried
out and a 3D finite element model is established. Then, an analytical
model considering plane stress state of fiber glass reinforced layers is
established. The detailed stress of each layer and the ultimate internal
stress of FGRFP are calculated by analytical method. Finally, parametric
analyses like winding angle of fiber glass and radius-thickness ratio are
studied.
405
External PE
Internal PE
Fibre glass reinforcements
Figure 20.1 Typical Construction of FGRFP.
20.2 Experiment
20.2.1 Dimensions and Material Properties of FGRFP
Specimens used in this research were produced by Shang Hai Fei Zhou
Bo Yuan Petroleum Equipment company. Components of the FRFP are:
an inner UHMWPE, an outer HDPE, and eight reinforced layers consist
of HDPE and fiber glass. The fiber glass in odd reinforced layers wind in
one direction, while the fiber glass in other reinforced layers wind in the
converse direction. Detailed material properties and geometric parameters
of the specimens are listed in Tables 20.1 and 20.2.
Table 20.1 Geometric parameters of testing specimens.

Parameter Value
Inner radius (mm) 25
Outer radius (mm) 38
Thickness of inner PE layer (mm) 4
Number of reinforced layer 8
Winding angle of the fiberglass (º) ±55
Thickness of reinforced layers (mm) 6
Burst Strength of Fiber Glass Reinforced Riser 407
Table 20.2 Material properties of each layer.

Material Secant modulus Tensile strength (MPa)
Fiberglass 33 GPa 765
Outer layer HDPE 850 MPa 23
Inner layer UHMWPE 570 MPa 26~30
20.2.2 Experiment Device

A short-term burst pressure test machine is used in this experiment. The
largest burst pressure and the largest loading speed can reach 120 MPa
and 5 MPa/min.
20.2.3 Experiment Results

The loading speed used in the three specimens are maintained as
0.61, 0.77, and 0.92 MPa/s, as shown in Table 20.3. Obvious necking phe-
nomenon happens in the burst areas when the pipes get into failure state.
The distances between the burst areas and the nearer connectors are 160,
378, and 202 mm, respectively, which is accorded with the experiment
requirements of ASTM D1599 [11].
The time-burst pressure relationship is shown in Figure 20.2, which
shows that the internal pressure drops sharply after the pipes reach its
failure state, and the ultimate burst pressures almost not influenced by the
loading speed.
20.3 Numerical Simulations

20.3.1 Mesh and Interaction
Finite element software ABAQUS is used to build the models. C3D8R solid
element is used to simulate the inner layer, outer layer, and the matrix of
Table 20.3 Burst pressure of the testing specimen.

#1 #2 #3 Average value
Burst Pressure/MPa 79.46 81.46 80.54 80.49
100
0.61 MPa/s
0.92 MPa/s
80
0.77 MPa/s
60
/MPa
40
20
0
0 50 100 150
/s
Figure 20.2 The curves of burst pressure and time.
reinforced layers while Truss element is used to simulate the fiber glass.
Since FGRFP belongs to bonded flexible pipes, fiber glass is then embed-
ded into the matrix. As shown in Figure 20.3, fiber glass is winded as the
same angle but in opposite directions.
20.3.2 Load and Boundary Conditions

When FGRFP is subjected to internal pressure, only the deformation in
axial direction is considered. As shown in Figure 20.4, the right section
of the model is totally fixed while the left section can only move in axial
direction. The displacement is added from zero at constant speed inside
the pipes.
Y Y
Z X Z X
Figure 20.3 FEM of FGRFP.

Y Y
X Z
Z X
(a) (b)
Figure 20.4 Boundary conditions of FGRFP under internal pressure.
20.3.3 Numerical Results

When the pipes are subjected to internal pressure, the deformation and
stress of inner layer, outer layer, and reinforced layers are relatively aver-
age, the cross-section still stay circular. Figure 20.5 gives the axial stress
distributions of four layers. It can be seen from this figure that the axial
stresses of reinforced layers far larger than the ones in PE layers, and the
axial stresses in the innermost reinforced layer is larger than the ones in
outermost reinforced layer. This happens because the secant modulus of PE
layers is smaller than the ones of fiber glass. Anti-tension ability of PE layers
is stronger than fiber glass. Numerical results prove that reinforced layers
bear the main force during the internal pressure process. When the axial
force of the fiber glass in the innermost reinforced layer reaches 765 MPa,
the tensile strength of fiber glass, FGRFP get its burst pressure 85.5 MPa.
20.4 Analytical Solution

20.4.1 Basic Assumptions
FGRFP are a type of composite structure. They not only have obvious
mesostructured properties but also have macroscopic properties. It is nec-
essary to make the following assumptions before analytical solution.
The materials used in FGRFP are homogeneous and continuous without
any void and crack.
(1) F
iber glass and matrix in reinforced layers are bonded with-
out any slippage during deformation process; Each layer is
bonded closely without any slippage.
S,Sll
Bottom Left Corner
(Avg: 75%)
+1.035e+03
+8.615e+02
+6.875e+02
+5.135e+02
+3.395e+02
+1.655e+02
–8.529e+00
–1.825e+02
–3.565e+02
–5.305e+02
–7.045e+02
–8.785e+02
–1.053e+02
Innermost reinforced layer
S,Sll
Bottom Left Corner
(Avg: 75%)
+1.012e+03
+8.758e+02
+7.396e+02
+6.035e+02
+4.674e+02
+3.313e+02
+1.952e+02
+5.906e+01
–7.705e+01
–2.132e+02
–3.493e+02
–4.854e+02
–6.215e+02
Outermost reinforced layer
S,Sll
(Avg: 75%)
+1.035e+03
+8.615e+02
+6.875e+02
+5.135e+02
+3.395e+02
+1.655e+02
–8.529e+00
–1.825e+02
–3.565e+02
–5.305e+02
–7.045e+02
–8.785e+02
Inner layer
S,Sll
(Avg: 79%)
+1.012e+03
+8.758e+02
+7.396e+02
+6.035e+02
+4.674e+02
+3.313e+02
+1.952e+02
+5.906e+01
–7.705e+01
–2.132e+02
–3.493e+02
–4.854e+02
Outer layer
Figure 20.5 Axial stress distributions of each layer under internal pressure.
(2) Th
e cross-section of FGRFP always perpendicular to the
center axis during deformation process.
(3) The mechanical properties stay unchanged during com-
posite processing, and they do not change with time or
temperature.
(4) All the materials are considered as linear elasticity. Inner
layer and outer layer are homogeneous while reinforced lay-
ers are anisotropy.
20.4.2 Stress Analysis

Figure 20.6 shows the global and local coordinate systems of analytical
model. In which Z, θ, and r mean the axial direction, the hoop direction,
an d the radial direction in the global coordinate system.
In global coordinate system, the displacements of FGRFP are denoted as
ur = ur(r), uθ = uθ(r, z), uz = uz(z) (20.1)
In which, ur, uθ, and uz are the axial displacement, the hoop displace-
ment, and the radial displacement, respectively. The constitutive relation
of the kth layer in global coordinate system is
(k ) (k ) (k )
 σz   C11 C12 C133 0 0 C16   εz 
     
 σθ   C21 C22 C23 0 0 C26   εθ 
     
σr C31 C32 C33 0 0 C36  εr
  =  
 τ θr   0 0 0 C44 C45 0   γ θr 
     
 τ zr   0 0 0 C45 C55 0   γ zr 
 τ zθ   C16 C26 C36 0 0 C66   γ zθ 
    
(20.2)
In which, Cij is the of- axial stiffness matrix. From elastic mechanics,
the stress-displacement relationship in kth layer is
r, r
T
θ θ
L
a
–a
Z
R
HDPE
Fibre glass rope
Figure 20.6 Global and local coordinate systems.
∂ur( k ) ( k ) 1 ∂uθ( k ) ur( k )

ε r( k ) = , εθ = + ,
∂r r ∂θ r
∂u( k ) 1 ∂uz( k ) ∂uθ( k )
ε z( k ) = z , γ z(θk ) = + ,
∂z r ∂θ ∂z (20.3)
∂u( k ) ∂u( k )
γ zr( k ) = z + r ,
∂r ∂z
1 ∂ur (k )
∂  u( k ) 
γ θ( kr ) = +r  θ 
r ∂θ ∂r  r 
Without the body force considered, the equilibrium equations in

global coordinate system are denoted as
∂σ r( k ) 1 ∂τ θ( kr ) ∂τ zr( k ) σ r( k ) − σ θ( k )
+ + + = 0,
∂r r ∂θ ∂z r
∂τ θ( kr ) 1 ∂σ θ( k ) ∂τ z(θk ) 2τ θ( kr )
+ + + = 0, (20.4)
∂r r ∂θ ∂z r
∂τ zr( k ) 1 ∂τ z(θk ) ∂σ z( k ) τ zr( k )
+ + + =0
∂r r ∂θ ∂z r
Substituting Eq. (20.1) into Eq. (20.3), the equilibrium equations lead to
∂ur( k ) ( k ) ur( k ) ( k ) ∂uz( k )

ε r( k ) = , εθ = ,εz = ,
∂r r ∂z
(20.5)
∂u( k ) ∂  u( k ) 
γ z(θk ) = θ , γ zr( k ) = 0, γ θ( kr ) = r  θ 
∂z ∂r  r 
Substituting Eqs. (20.1), (20.2), and (20.5) into Eq. (20.4), the equilib-
rium equations lead to
∂σ r( k ) σ r( k ) − σ θ( k ) ∂τ ( k ) 2τ ( k )
+ = 0, θ r + θ r = 0 (20.6)
∂r r ∂r r
Displacements become as
B( k )
ur( k ) = A( k )r + ,
r
uθ( k ) = C ( k )zr , (20.7)
uz( k ) = D ( k )z + E ( k )
Substituting Eq. (20.7) into Eq. (20.5), displacement can be simplified as
(k ) B( k ) ( k )
(k ) (k ) B( k )
ε = A − 2 , εθ = A + 2 ,
r r
(k ) (k )
(20.8)
εz = D ,
γ z(θk ) = C ( k )r , γ zr( k ) = 0, γ θ( kr ) = 0
Strains of reinforced layers become

(k ) (k ) B( k ) ( k ) (k ) B( k )
ε r =A − 2 , εθ = A + 2 ,
rk rk
(20.9)
ε z( k ) = D ( k )
γ zkθ = C ( k )rk , γ zr( k ) = 0, γ θ( kr ) = 0
The constitutive relation of the kth reinforced layer in global coordinate

system is
(k ) (k ) (k )
 σ   Q11 Q12 Q13   ε 
z
     z 
 σθ  =  Q21 Q22 Q23   εθ  (20.10)
 τ   Q Q32 Q333   γ 
 zθ   31  zθ 
While the constitutive relation in local coordinate system is
(k ) (k ) (k )
 σ   Q11 Q12 Q13   ε 
L L
     
 σT  =  Q21 Q22 Q23   εT  (20.11)
 τ   Q Q32 Q333   γ 
 LT   31  LT 
Eqs. (20.10) and (20.11) denote the stresses in global coordinate

system and in local coordinate system, respectively.
20.4.3 Boundary Condition

The boundary condition is continuous, as shown in the following:
σ (1) (r0 ) = − P0 , σ r(n ) (ra ) = 0 (20.12)
τ θ(1r) (r0 ) = τ zr(1) (r0 ),τ θ(nr ) (rn ) = τ zr(n ) (rn ) (20.13)
Interactions between each surface are
ur( k ) (rk ) = ur( k+1) (rk ) (20.14)
uθ( k ) (rk ) = uθ( k+1) (rk ) (20.15)
τ zr( k ) (rk ) = τ zr( k+1) (rk ) (20.16)
σ r( k ) (rk ) = σ r( k+1) (rk ) (20.17)
τ θ( kr ) (rk ) = τ θ( kr +1) (rk ) (20.18)
Equilibrium equation of axial force is
∑∫
rk
σ z( k ) • 2πr dr = πr02 P0 (20.19)
rk −1
k =1
Equilibrium equation of torsion
∑∫
rk
τ z(θk ) • r 2 dr = 0 (20.20)
rk −1
k =1
Eqs. (20.13), (20.15), (20.16), and (20.18) are constant under axial
symmetry loads. 2n + 2 equations can be obtained from the continuous
Table 20.4 Comparison of three methods.

Experiment/MPa FEM/MPa Analytical solution/MPa
80.49 85.5 82.82
boundary condition. Because the secant modulus of fiber glass is way too
larger than the ones of PE, fiber glass is the first to fail when its axial force
reaches the biggest tensile strength.
20.5 Results and Discussion

The burst pressures from three methods are shown in Table 20.4. It can be
seen that the difference of burst pressures between analytical method and
experimental method is 2.81%, analytical value is larger than the experi-
mental value, which is because in the analytical analysis, the failure crite-
rion is that the fiber glass in innermost reinforced layer reaches its tensile
strength. However, in practical pipes, due to the confines of production
technology, some fiber glass will first fail when it reaches its tensile strength.
For example, the amount of fiber glass in practical pipes is less than the
amount in design, the nonuniform of the winding angle and the thickness
of each reinforced layer. The difference of burst pressures between numer-
ical method and analytical method is 3.13%, which is because the both
sides of the model in analytical analysis can move in the axial direction
while just one side of the model in numerical analysis can move in the axial
direction, another side is totally fixed. Axial stress- pressure relationships
of inner reinforced layer from analytical solution and numerical solution
are shown in Figure 20.7.
Figure 20.8 gives the axial stress-pressure relationships of each rein-
forced layer. The innermost reinforced layer is considered as Layer 2 while
the outermost reinforced layer is considered as Layer 9. It can be seen from
this figure that the axial forces in reinforced layers linearly increase with
800
600
/MPa
400
200
0
0 50 100
/MPa
Figure 20.7 Axial stress-pressure curve of inner fiber glass reinforced layer (analytical
solution and FEM).
800
600
/MPa
400
2
4
200 5
8
9
0
0 50 100
/MPa
Figure 20.8 Axial stress-pressure curve of each layer.
the increasement of internal pressure. The axial force decreases gradually

from inside layers to outside layers.

Winding angle of fiber glass and diameter-thickness ratio can lead to
the change of stress and strain of FGRFP under internal pressure, and
then influence the mechanical behavior of FGRFP; therefore, they are quite
important in the cross-section design of FGRFP.
20.6.1 Winding Angle of Fiber Glass

Figure 20.9 shows the burst pressure of FGRFP under internal pressure
resulted by different winding angle of fiber glass. As shown in this figure,
when the winding angle is from 45° to 50°, the burst pressure of FGRFP
increases gradually; when the winding angle is from 50° to 75°, the burst
pressure of FGRFP soars up; when the winding angle overreaches 80°, the
burst pressure of FGRFP decreases gradually.
Figure 20.10 shows the winding angle-hoop strain curve and winding
angle-axial strain curve. With the increasement of the winding angle, hoop
strain decreases quickly and then increase slowly. Axial strain increases
quickly before the winding angle is less than 80°. It increases slowly after
the winding angle is over 80°. The two curves overlap when the winding
angle is 59°. Hoop strain and axial strain equal each other at this time,
which means 59° is the optimum winding angle of fiber glass.
170
/MPa
150
130
45 50 55 60 65 70 75 80 85 90
/°
Figure 20.9 Effect of winding angle on internal pressure.
0.2
0.1
0
40 60 80
−0.1
Figure 20.10 Effect of winding angle on hoop strain and axial strain.
/MPa
90
40
4 6 8 10
D/t
Figure 20.11 Effect of D/t ratio on internal pressure.
20.6.2 Diameter-Thickness Ratio

Figure 20.11 shows the effect of diameter-thickness (D/t) ratio on burst
pressure. As can be seen, with the increase of D/t ratio, the burst pres-
sure of FGRFP decreases significantly. When D/t ratio increases from 4 to
10, the burst pressure of FGRFP decreases from 128.82 to 45.89 MPa by
64.38%. Within the requirements range, it is reasonable to decrease the D/t

ratio to increase the burst pressure of FGRFP.
20.7 Conclusions
To investigate the mechanical behavior of FGRFP under internal pressure,
a short-term burst pressure test, numerical simulation, and analytical solu-
tion are performed in this chapter, and the conclusions are as follows:
(1) Th
e difference of burst pressure from experimental results
and analytical results is 2.81%, which is acceptable in prac-
tical engineering. This analytical solution can be a reference
for engineering application.
(2) The difference of burst pressure from numerical solution
and analytical solution is 3.13%, and the innermost rein-
forced layer damages first under certain internal pres-
sure. The reinforced layers bear more forces than the PE
layers do, indicating the reinforced layers bear the main
forces under internal pressure.
(3) When the winding angle of fiber glass changes from 45° to
80°, with the increase of winding angle, the burst pressure
becomes larger. The optimum winding angle is 59°.
(4) With the increase of D/t ratio, the burst pressure decreases
significantly.
References
1. Bai Y, Liu T, Cheng P, et al. Buckling stability of steel strip reinforced ther-
moplastic pipe subjected to external pressure[J]. Composite Structures,
2016, 152: 528–537.
2. Arikan H. Failure analysis of (± 55°) filament wound composite pipes
with an inclined surface crack under static internal pressure[J]. Composite
Structures, 2010, 92(1):182–187.
3. Bai Y, Chen W, Xiong H, et al. Analysis of steel strip reinforced thermo-
plastic pipe under internal pressure[J]. Ships & Offshore Structures, 2016,
11(7):766–773.
4. Bai Y, Lu Y, Cheng P. Analytical prediction of umbilical behavior under
combined tension and internal pressure[J]. Ocean Engineering, 2015,
109:135–144.
5. Zheng J Y, Gao Y J, Xiang L I, et al. Investigation on short-term burst

pressure of plastic pipes reinforced by cross helically wound steel wires[J].
Journal of Zhejiang University-Science A(Applied Physics & Engineering),
2008, 9(5):640–647.
6. Zheng J, Li X, Xu P, et al. Analyses on the short-term mechanical properties
of plastic pipe reinforced by cross helically wound steel wires[J]. Journal of
Pressure Vessel Technology, 2009, 131(3): 031401.
7. Wakayama S, Kobayashi S, Imai T, et al. Evaluation of burst strength
of FW-FRP composite pipes after impact using pitch-based low-modulus
carbon fiber[J]. Composites Part A: Applied Science and Manufacturing,
2006, 37(11): 2002–2010.
8. Bai, Yong, et al. Analysis of steel strip reinforced thermoplastic pipe under
internal pressure. Ships and Offshore Structures 11.7 (2016): 766-773.
9. Bai, Yong, Yutian Lu, and Peng Cheng. Analytical prediction of umbilical
behavior under combined tension and internal pressure. Ocean Engineering
109 (2015): 135-144.
10. Wakayama, Shuichi, et al. Evaluation of burst strength of FW-FRP composite
pipes after impact using pitch-based low-modulus carbon fiber. Composites
Part A: Applied Science and Manufacturing 37.11 (2006): 2002-201.
11. Standard Test Method for ÿesistance to Short-Time Hydraulic Pressure of
Plastic PipeÿTubingÿand fittings: ASTM D1599ÿSÿÿ 2005.
21
Structural Analysis of Fiberglass
Reinforced Bonded Flexible
Pipe Subjected to Tension
21.1 Introduction
Fiber reinforced bonded pipes are extensively applied in various engi-
neering practices, such as oil transportation (Gibson and Arun, 1989). It
has been shown that fiber offers an attractive quality of enhanced strength
and deformability of bonded pipes. Normally, these pipes are composed
of two kinds of materials: fiber and resin. Fibers such as kevlar, glass,
and carbon have excellent tensile strength and specific modulus, which
have often been selected for reinforced layers. Resins have the ability of
transferring stress among fibers, which have always selected for matrix.
However, kevlar and carbon fiber are not usually used in deep-sea pipe-
lines because of high cost and electrochemical corrosion. It is widely rec-
ognized that fiberglass reinforced flexible pipe (FGRFP) is a favorable
choice for oil and gas transportation due to its advantages in being rela-
tively light weight, corrosion, and abrasion resistance. Moreover, FGRFP
can be easily coiled in thousands of meters, which is convenient for quick
installation and greatly reduces the usage of joints. However, they will
inevitably confront tensile loads that are important aspects affecting
the integrity and security of the FGRFP’s utilization in installation and
service.
In the past decades, many achievements have been made in the field of
composite pipes subjected to tensile loads (Yue et al., 2013; Bai et al., 2017).
However, the classical laminated-plate theory has been adopted in most
studies on the mechanical properties of fiber helically reinforced bonded
composite pipes (Rosenow 1984; Xia et al., 2001; Bai et al., 2012; Hu et al.,
2015; Xing et al., 2015). This theory holds that the fiber and matrix are
merged into an integral part. Equations of thin-laminate plane deforma-
tion are established based on the Kirchhoff-Love assumption. The pipe
421
is considered as a 3D orthotropic cylinder, and then, stress and deforma-

tion of sandwich pipe subjected to different loads are deduced. However,
classical laminated-plate theory assumes that the material is elastic rather
than considering the elastic-plastic properties. In addition, this method
always regards reinforced material as a part of the reinforced layer that
is equivalent to an anisotropic plate. The stiffness of the reinforced layer
is determined by the rule of mixture. Thus, contribution of each material
cannot be known clearly.
Over the last few decades, there has been a shift in attention from a focus
on the classical laminated-plate theory to emphasis on composite pipes or
optical cables with helical structures (Custodio et al., 1999; Custodio and
Vaz, 2002; de Sousa et al., 2013; Yue et al., 2013; de Sousa et al., 2014;
Weppenaar et al., 2014). Knapp RH (1975) played an important role in the
nonlinear analysis of optic cable under tension and torsion. This theory
established the equations of equilibrium based on a helically armored cable
model. Geometric nonlinearities are exact formulations, and material non-
linearity of the core element is treated by a secant modulus approach which
utilizes the stress-strain data obtained from a simple uniaxial tension test.
In 1979, a simplified analysis procedure was proposed by using a simplified
linear stiffness matrix (Knapp, 1979). A new stiffness matrix was derived
for cables consisting of layers of helical armoring wires wrapped around a
central elastic core.
As can be seen from the current research situation, most theories don’t
consider elastic-plastic properties of materials. Besides, mechanical behav-
ior of every winding structure cannot be obtained by applying the classical
laminated-plate theory. In other words, the investigations on mechanical
properties of fiber reinforced bonded flexible pipes are still very much
in their infancy. In this chapter, an analytical model of fiber reinforced
bonded pipes that consider material nonlinearity is established. This ana-
lytical model is validated by a full-scale test of FGRFPs subjected to pure
tension. Finite element analysis is accomplished by ABAQUS to better
study the mechanical properties of the target pipe. Detailed analysis
of the experimental and simulative results is done, and some interesting
phenomena are noted.
Additional parametric studies are also conducted. A profound under-
standing about the function of FGRFPs can be achieved, which can pro-
vide valuable advice for this kind pipe’s design and application.
Structural Analysis of Fiberglass 423
21.2 Experiment
Figure 21.1 shows that FGRFPs are typically made up of an external
coating, reinforced tapes, and internal liner. Detailed manufacturing
dimensions of the specimens are pictured in Figure 21.2 and Table 21.1.
Internal Liner
Reinforced
Tapes
External
Coating
Figure 21.1 Construction of FGRFP.
4 mm 8*0.75 mm 3 mm
Inter diameter 50 mm
Outer diameter 76 mm
Figure 21.2 Cross-section of FGRFP.

Table 21.2 shows specifications of FGRFPs. The test specimens used in this
chapter were produced by the continuously helical tape wrapping method.
There is a total of eight fiberglass reinforced layers laying helically on the
inner ultra-high molecular weight polyethylene (UHMWPE) tube. Each
reinforced layer has 18 to 21 prepreg tapes that are made of 60% (volume
Table 21.1 Specifications of the tensile test specimen.

Inter Outer
diameter Thickness diameter
Parts Material (mm) (mm) (mm)
Internal UHMWPE 50 4 58
Liner
Reinforced Layer No. Winding Fiberglass
Layers Angle Prepreg Tape
Amount
1 +55° 18 59.5
2 -55° 18 61.0
3 +55° 19 62.5
4 -55° 19 64.0
5 +55° 20 65.5
6 -55° 20 67.0
7 +55° 20 68.5
8 -55° 21 70.0
External HDPE 70 3 76
Coating
Table 21.2 Geometric parameters of the tensile test specimen.

Label Outer diameter (mm) Effective length (mm)
A-1 76.63 1,013
A-2 76.55 996
A-3 76.06 999
Average 76.41 1,003
Figure 21.3 Tensile test of the specimen in process.
percent) fiberglass and 40% high-density polyethylene (HDPE). Those are

subsequently wrapped with HDPE as external coating.
A full-scale test was prepared and tested in the Marine Engineering
Experimental Center at the Zhejiang University. According to the
authors’ published paper (Xu et al., 2017), HDPE is the main compo-
nent resisting tensile load when the axial elongation is less than 7.7%. The
test was conducted according to the international standard ISO6259-1,
“Thermoplastic pipes-Determination of tensile properties”. As shown in
Figure 21.3, the tension tests were performed using a 10,000 kN capacity
testing machine under displacement control at a constant rate of 1 mm/s.
The loads and displacements were recorded using a data logging system
and were stored in a computer.
21.2.2 Material Characteristics

The material properties of FGRFP are shown in Table 21.3. This kind of
FGRFP consists of three kinds of material: UHMWPE used for the internal
Table 21.3 Material properties of FGRFP.

Material Young’s Modulus (MPa) Poisson’s ratio
HDPE 988 0.4
UHMWPE 1,004 0.4
Fiberglass 72,607 0.3
30
25
20
Stress (MPa)
15
10
5
Nominal
True
0
0.00 0.02 0.04 0.06 0.08 0.10 0.12 0.14 0.16
Strain
Figure 21.4 Stress-strain data of HDPE from tensile test.
liner to resist hydrodynamic pressure and corrosion, fiberglass heat welded

in the reinforced layers, and HDPE used for the external coating with
aging resistance and durability. The elastic modulus of HDPE is calculated
as the secant modulus when the true strain is 0.05% and 0.25% based on
the standard ISO527-2012. As shown in Figure 21.4, the true stress and
strain that are calculated through the corresponding nominal stress and
strain obtained from the specimen tensile tests shall be used to define
the material properties in numerical analysis.
21.2.3 Experimental Results

The failure process of the FGRFP during the experiment can be concluded
as follows: In the initial stage, the fiberglass snapping was heard. With
the sound becoming more and more acute, creep in the external coating
showed up. After that, an external crack developing gradually and then
suddenly snapped. At the same time, the reinforced layers were exposed
along with some fiberglass fracture. Finally, the fiberglass began to snap
from outside to inside layer-by-layer. Three specimens showed similar
failure patterns, though the time between the fracture of fiberglass and
the cracking of HDPE was quite different. However, there was no appar-
ent damage in the internal liner in the entire loading process. The failure
modes of the three specimens are shown in Figure 21.5.
The corresponding tension-extension curves for the test specimens are
illustrated in Figure 21.6. For this testing machine, the tensile forces as well
as the elongation displacements are negative. It can be observed from
Figure 21.5 Failure mode of the specimen.
the curves that the tendency of tension-extension relation is similar to the

HDPE stress-strain relation when the elongation is below 7.7%. The yield
stress of HDPE is calculated as 23 MPa, which is very close to that of the
test specimens. The ensuing descending trajectory is due to the HDPE out
of efficacy. As the axial displacement increases, the spiral fiberglass gradu-
ally unwinds, which directly contributes to the segment stage. In practical
applications of FGRFP, once the material cracks, the pipe is regarded as
completely unserviceable. API 17J specifies that the maximum allowable
strain for PE shall be 7.7% in utilization. Thus, it is important to study
the structural response of FGRFP before the strain in the HDPE reaches
7.7%. Table 21.4 shows the ultimate tensile strength of three specimens.
The maximum error rate is 11.3%, which further proves the validity of
the experiment.
21.3 Theoretical Solution

This chapter proposes an analytical method that not only takes mate-
rial nonlinearity into consideration but can also obtain the contribution of
each material.
The conformity between the proposed solution and Knapp’s theory
(Knapp 1975, 1979) can be classified into three major categories. First,
both cables and pipes are composed of spiral winding and cylindrical
structure. Second, they both take material nonlinearity into consideration.
Third, they are both subjected to axisymmetric loads. Therefore, the ana-
lytical solution is developed on the basis of Knapp’s theory.
The reason why modification of Knapp’s theory is required can be
concluded as follows: The geometry of composite pipes is more complex
Extension (mm)
0 –100 –200 –300 –400 –500 –600
0
–20
–40
0 –20 –40 –60 –80
0
Load (kN)
Fiberglass snapped
–10
–60
–20
–30
HDPE cracked
–80 –40
–50
–60
–100 –70
–80 (a) A1
Extension (mm)
0 –100 –200 –300 –400 –500 –600
00 0 –20 –40 –60 –80
–10
–20
–20
–30
–40
–50
Load (kN)
–40
–60
Fiberglass snapped
–60
HDPE cracked
–80
(b) A2
Extension (mm)
0 –100 –200 –300 –400 –500 –600
0
–20
–40 0 –20 –40 –60 –80

0
Load (kN)
–20
–60
–40
Fiberglass snapped
–80 HDPE cracked
–60
–80
–100
(c) A3
Figure 21.6 Tension-extension relation of three specimens.

Table 21.4 Ultimate tensile strength (kN).

A1 A2 A3
83.94 80.88 90.00
than the cable, which contributes to the difference in kinematics relations,

deformation deduction, and equilibrium equations. Besides, different loads
bring about different assumed boundary conditions and loading regimes.

The assumptions used in the analytical method are listed as follows:
1. In general, bonded pipes are composed of an inner sheath,

reinforced layers, and an external sheath, while the rein-
forced layers are always made by spiral winding in two
opposite angles. So, only the analytical solution of bonded
pipes with two reinforced layers is necessarily deduced.
2. Layers are assumed to be fully bonded.
3. Only tensile deformation is considered. Bending defor-
mation, torsion deformation, and compression deformation
are not considered.
4. The sheaths and matrix, used in the four layers, are assumed
to be same.
5. Radial displacement and axial displacement are allowed;
torsion angle is restrained.
21.3.2 Cross-Section Simplification

The reinforced layers are modeled as shown in Figure 21.7, and the
cross-section simplify cation is mainly for reinforced layers. The matrix
part (black solid squares in Figure 21.7b) between fibers is deemed to be a
spiral winding rope, as is the fiber. That means two kinds of rope arranged
at intervals around the same way, only differing in cross-sectional area and
materials. Therefore, the whole section can be simplified as two spiral rope
layers between two homogeneous cylinders.
r4
r4
r3
R2
R1
r2
r1
r1
(a) (b)
Figure 21.7 Treatment of reinforced layers
21.3.3 Fiber Deformation

The axial strain can be expressed as follows:
∆u
εz = (21.1)
L
Δu = u1 – u2 (21.2)
where u1 and u2 represent axial displacement at both ends, and L is pipe

length. Without considering bending strain, compressive strain and tor-
sion strain, only tensile strain is considered. Axial tensile strain ε ia can
be expressed as follows:
1
ε = (1 + ε z )2 cos 2 α i + βi2 sin 2 α i  2 − 1
a
i (21.3)
Ri′
βi = (21.4)
Ri
sin α i′ β
= ia
sin α i 1 + ε i
cosα i′ 1 + ε z (21.5)
=
cosα i 1 + ε ia
tan α i′ β
= i
tan α i 1 + ε z
where αi and α i′ irepresent the winding angle before and after deformation,
Ri and Ri′ represent the pitch radius before and after deformation, and βi
represents the pitch radius variation.
21.3.4 Cross-Section Deformation

There are two reasons that can cause the cross-section deformation
when the pipe is subjected to tension. On one hand, the radial shrinkage
is caused by the Poisson’s Ratio. On the other hand, fibers in the reinforced
layer creates radial compression on the internal polymer.
Therefore, the radial pressure of each layer can be calculated as follows:
1  nN1 tan α1 sin α1′ nN 2 tan α1 sin α1′ 

P1 = +
2π R1′  R1 (1 + ε Z ) R1 (1 + ε Z ) 
(21.6)
1  nN1 tan α 2 sin α 2′ nN
N tan α 2 sin α 2′ 
P2 =  + 2
2π R2′  R2 (1 + ε Z ) R2 (1 + ε Z ) 
N1 = E1ε1A1
(21.7)
N2 = E2ε2A2
where N1 and N2 represent the axial tension of reinforced rope and matrix
rope respectively, and n represents the amount of fiber rope in each rein-
forced layer. ε1 and ε2 represent the corresponding axial strain. A1 and
A2 represent the corresponding cross-section area. Eq. (21.6) is, however,
only an implicit expression for the radial pressure, Pi, since the expression
for Ri′ given by Eq. (21.4) also involves Pi (Knapp, 1979). Thus, it is appar-
ent that an iterative procedure will be required to achieve a solution.
Ignoring the influence of the reinforced rope on the sectional stiffness,
the interaction force between layers can be deduced byelastic mechanics
method.
Figure 21.8 shows the interaction force between different layers. The
deduction principle is similar, only different in section radius and bound-
ary condition. According to the plane strain assumption (that is, εz is
constant), Hooke’s Law and equilibrium equations are employed. The
r4
r3
r3
r2
r2
σ2
r1
P1 σ1 σ2 P2
σ1
Figure 21.8 Interaction force between layers.
differential equation of displacement can be mathematically described as

follows:
d 2u 1 du u
+ − =0 (21.8)
dr 2 r dr r 2
Substituting different boundary conditions of Figure 21.8 into Eq.

(21.8), radial displacements can be obtained.
In the internal layer r = R1,
R1  − R12 ( P1 − σ 1 )  1 r12 R1 (− P1 + σ 1 )
uR1 =  − λε Z  + (21.9)
2(λ + G )  R12 − r12  2G R12 − r12
In the middle layer r = R1,
R1  R12σ 1 + R22σ 2  1 R22 R1 (σ 1 + σ 2 )

uR1 =  − λε Z  + (21.10)
2(λ + G )  R12 − R22  2G R12 − R22
In the middle layer r = R2,
R2  R12σ 1 + R22σ 2  1 R12 R2 (σ 1 + σ 2 )

uR2 =  − λε Z +
 (21.11)
2(λ + G )  R12 − R22  2G R12 − R22
In the outer layer r = R2,

R2  R22 ( P2 − σ 2 )  1 r42 P2 ( P2 + σ 2 )
uR2 =  − λε Z +
(21.12)
2(λ + G )  R22 − r42  2G R2 − r4
2 2
where λ and G represents Lame’s constant in Hooke’s Law, and G is also

known as shear modulus.
According to the reciprocal theorem of reaction, Eq. (21.9) is equivalent
to Equation (21.10), so as Eqs. (21.11) and (21.12). Thus, the interaction σ1
and σ2 can be obtained. Then, the pitch radius variation βi can be given as
follows:
β1 =
1
R (λ + 2G )( r − r )( R − r )(λ + G )G
2 2 2 2 2
( 2 ( R − r )( ( r
2
1 1
2
1
2
)
− r42 R12G 3 +
1 1 4 1 1
1
4
 3 1  1  2 1 2 1 2
 P1 R1 +    − ε Z  λ + P1  r1 + r4 (ε Z − 3)λ + ( P1 + P2 )r4 +
4  2 2 4  2 4
1
4
 1
4
( )) 1
P2 R22  R12 + r12 ( P1 + P2 )r42 + P2 R22 G 2 + λ((((2 − ε Z )λ + P1 )r12 +
 4
  
( ) 1
(ε Z − 2)λ + P1 + P2 )r42 R12 + 2  ( P1 + P2 )r42 + P2 R22  r12  G +
 2  
1 2 2 2 
r1 r4 λ ( P1 + P2 )  (21.13)
4 
 1  1 3
β2 =
1
( )
 r12 − r42 R22G 3 +  P2 R24 +    ε Z −  λ
2
( 2 2
)
R2 (λ + 2G ) r1 − r4 (λ + G )G  4  2 2
1  1 1 1  1
+ P2  r42 − r12 (ε Z − 3)λ + ( P1 + P2 )r12 + P1R12  R22 + r42 (( P1 + P2 )r12
4  2 4 4  4
)) 1
+ P1R12 G 2 + ((((ε Z − 2)λ + P2 )r42 + r12 ((2 − ε Z )λ + P1 + P2 ))R22
4
( )) 1
+r4 2( P1 + P2 )r12 + P1R12 λG + r12r42λ 2 ( P1 + P2 ))
2
4
(21.14)

21.3.5 Equilibrium Equations

Using the method of deducing the equilibrium equations of optical cables
under tension and torsion proposed by Knapp, the equilibrium equations
of the fiberglass reinforced bonded flexible pipe under tension can be
deduced as follows.
According to principle of minimum potential energy, the following
equation should be satisfied:
δU + δV = 0 (21.15)
where U represents internal strain energy of the whole system, and V rep-
resents external load potential energy. U and V can be listed as follows:
4N
δU = δU ic + δU mc + δU oc + ∑ δU
i =1
i (21.16)
δV = −Tδ(Δu) (21.17)
where Ui represents the strain energy of each rope. The subscripts ic, mc,
and oc represent the inner layer, middle layer, and outer layer of the matrix
ring, respectively, as shown in Figure 21.7. T represents the axial load. The
equilibrium equation can be obtained by next equation.
4n
T = N ic + N mc + N oc + ∑  N cosα ′
i =1
i i (21.18)
where N represents the axial tension.
21.4 Finite Element Model

The FEM in ABAQUS is imported by a script file written in UItraEdit soft-
ware, which is very convenient to change parameters. As seen from Figure
21.9, all HDPE cylinders are assigned solid elements (C3D8R), while fiber-
glass is assigned beam element (B31). Two layers of fiberglass embedded in
the matrix with the same winding angle (55°) but opposite winding direc-
tion form like a network (Figure 21.10).
RP-1
Z X
Figure 21.9 Pipe model in FEM.
X
Z
Figure 21.10 Network of fiberglass in FEM.
In the manufacturing process, the fiberglass is firstly heated and

welded into a flake shape, then wrapped outside the UHMWPE cylin-
der. Therefore, it is more accurate to use beam element rather than truss
element to simulate the fiberglass in FEM. Figure 21.11 shows the cross-
section of a fiberglass prepreg tape in FEM needs be converted to eight
fiberglass ropes in the proposed analytical model.
0.75mm
6mm
Figure 21.11 Equivalent cross-section.
The assumed stress-strain relation of HDPE under tension has an

obvious elastic-plastic relation. The von Mises yield criterion is used to
determine the boundary of the elastic and plastic HDPE deformation.
The fiberglass is regarded as elastic material with high tensile stiffness.
Radial displacement and axial displacement of FGRFP are allowed
meanwhile the torsion angle is completely limited. The center of the end
sections is chosen as reference points that are coupled with both ends of
the section. The coupling includes five directions except the radial direc-
tion, which means that the radial displacement of the pipe wall is only
related to the central axis.
Boundary conditions are applied on reference points. One reference
point is fixed and the other one is set to allow displacement in axial direc-
tion. The whole loading process uses static calculation, which involves ten-
sile displacement increases from 0 to 77 mm at a constant speed.

In this section, both a finite element model using ABAQUS and the analyt-
ical solution using MATLAB program are established to verify the exper-
imental results.
21.5.1 Tension-Extension Relation

According to the above modeling method, the finite element and analytical
solution are verified with experimental results. The tension-extension
relation is shown in Figure 21.12. It should be noticed that the measured
axial load fluctuates around zero in the initial stage of the loading pro-
cedure. This is due to the instability and resolution measurement of the
testing machine when the applied load is relatively small.
It can be observed from this graph that the curves share the same trend as
the HDPE material relation in general. However, it should also be noticed
that, the results from Specimen 1 and Specimen 3 are obviously higher
80
70
60
50
Load (kN)
40
Specimen 1
30 Specimen 2
Specimen 3
20
Analytical Method
FEM
10
0
0 2 4 6 8
Extension (%)
Figure 21.12 Tension-extension relation from three methods.
than that of the analytical method and FEM. This is due to the instability
of the initial loading process and material discrete influence. Although
the curves may have a certain range of allowable error, they are still com-
parable. Curves from those three sets of methods coincide with each other,
further implying the accuracy and reliability of virtual work.
The reason the FEM result differs from the analytical result can be clas-
sified into three major categories. Each flake of fiberglass is simplified as
eight fiberglass ropes, as shown in Figure 21.11, which makes the HDPE
area bigger than that in the experiment. The HDPE makes a large contri-
bution to tension at the early stage, so the discrepancy between the ana-
lytical results and the FE results may arise from the HDPE area. Another
reason is that the fiberglass is embedded into reinforced layers in FEM,
that means fiberglass and HDPE are fully bonded. Instead of considering
bonding between fiberglass and HDPE, the reinforced layer is simplified
as two kinds of rope arranged at intervals around the same way in the ana-
lytical model. Additionally, UHMWPE is assumed as HDPE to be better
application of analytical model, so slight differences may be caused due
to this.
21.5.2 Cross-Section Deformation

Deformation of FGRFP after extension is shown in Figure 21.13. The
cross-section shows a large contraction as evidenced by some apparent
necking. Ignoring the end effect and selecting the middle section, the
radial displacement is obtained by FEM. Comparing the FEM result and
ODB: 20170804-1.odb Abaqus/Standard 6.14-1 Fri Aug 04 09:07:15 GMT+08:00 2017

Y
Step: Step-1
Z X Deformed Var: U Deformation Scale Factor: +1.000e+00
Figure 21.13 Deformation of FGRFP after extension.
40
36
32
Radius (mm)
28
24
20
Analytical Method
FEM
16
0 2 4 6 8
Extension (%)
Figure 21.14 Radius-extension relation (analytical model and FE model).
the analytical result, the radius (consistent with Figure 21.7) variation
tendency is given in Figure 21.14. The radius of each layer decreases with
the increase of axial displacement not only in the analytical analysis but
also in the FE analysis.
Curves from analytical method and FEM closely coincide with each
other when the elongation is less than 7.7%. Actually, the results from
FEM are slightly lower than that of the analytical method. The difference
between the two methods can be explained as follows. At the first stage
of tension, the contraction of radius is mainly caused by HDPE volume

compression. Therefore, the analytical result and the FEM result show
high coincidence in this period. Then, fibers begin to squeeze each other,
which is not considered in the analytical analysis. Moreover, the elastic
modulus of the fiberglass is much larger than that of HDPE. That is why
the FE curve decreases faster than the analytical curve.
21.5.3 Fiberglass Stress

The proposed analytical method enhances previous methods by studying
every single fiberglass mechanical property. Numbering the reinforced
layers 1 to 8 from inside to outside in the FEM, and the corresponding
force curve is described in Figure 21.15. From the graph, it can be noticed
that the adjacent layers have opposite winding angle but similar loading.
In general, fiberglass stress increases from the inside to the outside layer,
meanwhile the load on the odd layers is slightly larger than that on
the even layers. Furthermore, it appears that compressive stress occurs in
the innermost layers, which varies the analytical method. The analytical
results show that the simplified fiberglass in the inner layer displays
compressive resistance, while the outer layer provides tension resistance
in the loading cycle.
21.5.4 Contribution of Each Material

In order to deeply understand the contribution of each material, HDPE
and fiberglass tensile loads calculated by the analytical method are
separately depicted in Figure 21.16.
240 #Odd layers #7

200 #Even layers #8
160
#5
120 #6
Load (N)
80
40 #3
#4
0
0 2 4 6 8
–40
–80
Extension (%) #2
–120 #1
Figure 21.15 Fiberglass stress variation in different layers (FE model).

70
Total
Fiber Glass
60
HDPE
50
40
Load (kN)
30
20
10
0
0 2 4 6 8
Extension (%)
Figure 21.16 Tension-extension relation of each material (analytical model).
According to Figure 21.16, all three curves show a rising trend.

HDPE makes more than 80% contribution to the tension resistance, while
fiberglass makes about 15%. Although the fiberglass contribution increases
gradually, it is still far less than HDPE (the maximum contribution is only
17.6%). In practical engineering, the axial elongation of flexible pipe is
always less than 7.7% because too much strain can be judged as losing
effectiveness. In a word, HDPE is the main component to resist tensile
load when the axial elongation is less than 7.7%, while, at the same time,
contribution of the fiberglass is very small. For better contrast, the pure
HDPE pipe and FGRFP with same geometric parameters is conducted
in the program. From Figure 21.17, it can be seen that the curve for
pure PE pipe is slightly above the one for FGRFP at the beginning stage.
At the later stage, the two curves remain close but the FGRFP appears to
have higher tensile resistance than pure PE. The appearance shows that
the role of fiberglass in carrying the tension load is not that obvious in the
concerned stage.
21.5.5 Summary
According to the above analysis, the operating principle of each com-
ponent in FGRFP can be inferred. A flexible rope under tension has the
following three conditions.
70
FGRFP
Pure PE
60
50
Load (kN)
40
30
20
10
0
0 2 4 6 8
Extension (%)
Figure 21.17 Comparison between pure PE pipe and FGRFP with the same thickness
(analytical model).
1. If the rope is straight, it can carry the axial tension directly.

2. If the rope is bent and allowed to have free displacement, it
carries tension only after being total straight.
3. If the rope is bent but with a fixed shape, it works as soon as
tension is applied.
At the initial stage, the HDPE volume is compressed, while the

spiral wound fiberglass rope is allowed to displace with shape change. The
stress state is between the last two situations, that is, fiberglass can with-
draw part of tension but not too much. Then, with the axial displace-
ment increasing continuously, HDPE cannot be compressed any more.
Fiberglass starts to squeeze each other, which shows an ascent stage in the
experimental curves. Figure 21.18 chooses one of fiberglass layers to show
the corresponding deformation during the loading process.
ODB: 20170803-2.0db Abaqus/Standard 6.14-1 Wed Oct 25 20:18:22 GMT+08:00 2017

ODB: 20170803-2.0db Abaqus/Standard 6.14-1 Wed Oct 25 20:18:22 GMT+08:00 2017 Y Y
ODB: 20170803-2.0db Abaqus/Standard 6.14-1 Wed Oct 25 20:18:22 GMT+08:00 2017
Y
Step: Step-1
Increment 5: Step Time = 0.5000 Step: Step-1
Step: Step-1 Increment 10: Step Time = 1.000
Z X Deformed Var: U Deformation Scale Factor: +1.000e+00 Z X Deformed Var: U Deformation Scale Factor: +1.000e+00
Figure 21.18 The unwinding process of fiberglass (FE model).


There are many parameters that can affect FGRFP mechanical behavior,
so parametric study is of great significance to engineering practice. Not
only can the variation of mechanic properties under different parameters
be observed but optimization design can also be obtained by taking cost,
service life, etc., into consideration.
The influence of different parameters (winding angle, fiberglass amount,
and diameter-thickness ratio) on the mechanical behavior of FGRFP under
tension is analyzed by using the analytical method.
21.6.1 Winding Angle

It can be observed from Figure 21.19 that, in general, the tensile stiffness
in the four curves decreases with the increase of axial tensile displacement.
As the fiberglass winding angle increases from 45° to 60°, the tensile
resistance of FGRFP decreases gradually.
As seen from Figure 21.20, the contribution of the fiberglass to tensile
resistance increases as the winding angle declines. Stress of the fiberglass
is quite large when the winding angle is 45°. Contribution of the fiber-
glass is even more than the HDPE, which is up to 71.3% when winding
120
45°
100 50°
55°
80 60°
Load (kN)
60
40
20
0
0 2 4 6 8
Extension (%)
Figure 21.19 Tension-extension relation of each material in different winding angles

(analytical model).
Fiberglass
HDPE
1.0 60°
0.8 55°
Contribution 45°
0.6 50°
50°
0.4
45°
0.2 55°
0.0 60°
0 2 4 6 8
Extension (%)
Figure 21.20 Contribution of each material in different winding angles (analytical

model).
angle is 45°. The calculations suggest that the fiberglass makes a negative
contribution to the resistance when winding angle reaches 60%, which is
unreasonable in practical terms. Therefore, it is reasonable to choose 55°
as the proper winding angle.
21.6.2 Fiberglass Amount

As can be seen from Figure 21.21, increasing or decreasing the fiber-
glass amount has no significant effect on the tension-extension relation.
70
Total
Fibrglass
60 HDPE
50
40
Load (kN)
30
20
10
0
0 2 4 6 8
Extension (%)
Figure 21.21 Tension-extension relation of each material in different fiberglass amount

(analytical model).
Contribution of HDPE has a downward trend, but the magnitude is very

small. Moreover, the stress of each fiberglass increases with the reduction
of the fiberglass amount not only in internal layers but also in external
layers. Contribution of all the fiberglass under tension increases, but the
value is still small. In a word, the tensile capacity of FGRFP cannot be
raised by increasing the fiberglass amount but the stress of each fiber-
glass can be decreased by that way.
21.6.3 Diameter-Thickness Ratio

The larger the diameter of the pipe, the greater the tensile resistance.
Though larger diameter means greater flow capacity, weight and cost are
Table 21.5 Different diameter-thickness ratios of FGRFP.

Radius-
thickness
ratio r1 (mm) r2 (mm) r3 (mm) r4 (mm) R1 (mm) R2 (mm)
4 13 17 23 26 18.5 21.5
6 26 30 36 39 31.5 34.5
8 39 43 49 52 44.5 47.5
90
Total
D/T=8
HDPE
75 Fiberglass
D/T=8
60 D/T=6
Load (kN)
D/T=6
45
D/T=4
30
D/T=4
15 D/T=8
D/T=6
D/T=4
0
0 2 4 6 8
Extension (%)
Figure 21.22 Tension-extension relation of each material in different diameter-thickness

ratio (analytical model).
still a key factor needed to be considered in the manufacturing process.

It is of great significance to find a balance between performance and cost.
Therefore, it is important to choose a reasonable radius-thickness ratio for
marine flexible piping. The experiment uses diameter-thickness of 5.8
(76/13). The wall thickness is kept constant and the radius is enlarged, then
different diameter-thickness ratios can be obtained, as shown in Table 21.5.
As seen in Figure 21.22, the tendencies of curves are very similar
to each other. The larger the diameter-thickness ratio is, the higher the
tension-extension relation reaches. However, the degree of variability in
the fiberglass is still minimal.
21.7 Conclusions
In this chapter, the mechanical behavior of FGRFP subjected to tension is
investigated by experimental, theoretical, and finite element methods. A
prototype test with three specimens has been verified by both analytical
and finite element methods. Considering material nonlinearity, the pro-
posed numerical method represents a valuable alternative way to under-
stand the mechanics properties of each component. The above three
methods validate each other with good accuracy and reliability.
This procedure can be used in predicting the tension behavior of a
fiberglass reinforced pipe and should provide a useful design tool for such
pipes. Generally, the tension-extension curve shares the same trend as the
HDPE material relation. The simplified fiberglass in the inner layer shows
compressive resistance, while the outer layer provides tension resistance in
the analytical analysis.
The influence of various parameters is determined in a parametric study.
As the fiberglass winding angle increases, the tensile stiffness of FGRFP
decreases gradually. Increasing or decreasing fiberglass amount has no
significant effect on the tension-extension relation. But, the stress of each
fiberglass can be decreased by increasing the fiberglass amount. Diameter-
thickness ratio reflects flow capacity of FGRFP. The larger diameter-
thickness ratio is, the higher tension-extension relation reaches.
The results presented in this chapter are able to provide a macro concept
for understanding this newly emerging pipe subjected to tension and serve
as the primitive work of other possible loads. The described theory, finite
method, and experimental method can provide some references for the
related engineering practice to a degree and for future research on the
more comprehensive analytical derivation.
Acknowledgement
The authors would like to acknowledge Kevin Conroy who helped us to
examine the grammar of this chapter. In addition, the authors express their
gratitude to Shanghai Fei Zhou Bo Yuan Petroleum Equipment Technology
Co., Ltd. for their help on conducting the experimental studies with all
necessary equipment.
References
Bai, Y., Wang, Y., Cheng, P., Liu, Q., Jin, X. H., & Xu, L. B. 2012. Axial strength of
reinforced thermoplastic pipe (RTP). In Proceedings of the ASME 2012 31st
International Conference on Ocean, Offshore and Arctic Engineering.
Bai Y, Liu T, Ruan W, Chen W. 2017. Mechanical behavior of metallic strip flexible
pipe subjected to tension. Composite Structures. 170.
Custodio AB, Vaz MA., Estefen SF. 1999. A survey on the response of subsea
umbilical cables under axisymmetric loads. Proceedings of the 15th Brazilian
Congress of Mechanical Engineering. p. 22–26.
Custódio, AB, Vaz MA. 2002. A nonlinear formulation for the axisymmetric
response of umbilical cables and flexible pipes. Applied Ocean Research.
24(1):21–29.
de Sousa JRM, Magluta C, Roitman N, Londoño TV, Campello GC. 2013. A
study on the response of a flexible pipe to combined axisymmetric loads.
Proceedings of ASME 2013 32nd International Conference on Ocean,
Offshore and Arctic Engineering. American Society of Mechanical Engineers.
p. V04BT04A040-V04BT04A040
de Sousa JRM, Campello GC, Kwietniewski CEF, Ellwanger GB, Strohaecker TR.
2014. Structural response of a flexible pipe with damaged tensile armor wires
under pure tension. Marine Structures. 39:1–38.
Gibson AG, Arun S. 1989. Composite materials in the offshore industry.
Comprehensive Composite Materials. 459–478.
Hu HT, Lin WP, Tu FT. 2015. Failure analysis of fiber-reinforced composite lami-
nates subjected to biaxial loads. Composites Part B: Engineering. 83:153–165.
Knapp RH. 1975. Nonlinear analysis of a helically armored cable with nonuniform
mechanical properties in tension and torsion. Proceedings of OCEAN 75
Conference. IEEE. p. 155–164.
Knapp RH. 1979. Derivation of a new stiffness matrix for helically armoured
cables considering tension and torsion. International Journal for Numerical
Methods in Engineering. 14(4):515–529.
Rosenow MWK. 1984. Wind angle effects in glass fibre-reinforced polyester fila-
ment wound pipes. Composites. 15(2):144–152.
Weppenaar N, Andersen B. 2014. Investigation of Tensile Armor Wire Breaks in

Flexible Risers and a Method for Detection. ASME 2014 33rd International
Conference on Ocean, Offshore and Arctic Engineering. American Society
of Mechanical Engineers.
Xia M, Kemmochi K, Takayanagi H 2001. Analysis of filament-wound compos-
ite pipes under combined internal pressure and thermomechanical loading.
Composite structures. 51(3):273–283.
Xing, J., Geng, P., & Yang, T. 2015. Stress and deformation of multiple winding
angle hybrid filament-wound thick cylinder under axial loading and internal
and external pressure. Composite Structures. 131, 868–877.
Xu Y et al. 2017. Analysis of Fibre Glass Pipe Subjected to Tensile Load. The
27th International Ocean and Polar Engineering Conference. International
Society of Offshore and Polar Engineers.
Yue Q, Lu Q, Yan J, et al. 2013. Tension behavior prediction of flexible pipelines in
shallow water. Ocean Engineering. 58:201–207.
22
Fiberglass Reinforced Flexible
Pipes Under Bending
22.1 Introduction
Recently, fiberglass reinforced flexible pipe (FRFP) has been widely used
in offshore transportation due to its high corrosion resistance, light weight
characteristic, and relatively low fabrication and facility cost. Compared
to other types of bonded pipes, it has a higher density and shows better
subsea mechanical behavior.
FRFP studied in this chapter, shown in (Figure 22.1), consists of a poly-
ethylene liner, eight layers of reinforced tape made of polyethylene and
fiberglass wrapping around the liner and an outer polyethylene coating. The
inner liner pipe is ultra-high molecular weight polyethylene (UHMWPE)
and the outer coating pipe is high-density polyethylene (HDPE), while the
reinforced tape is made of HDPE and fiberglass.
There are numerous papers on analyzing the responses of metallic cyl-
inder tube under pure bending. Brazier’s [1] studies clarified that the oval-
ity of cross-section leads to deterioration of flexure rigidity under bending
load. Weinitschke and Reissner [2] proved the Brazier’s work through
more formulations. The strain-displacement of infinitely-long cylinder
shell under pure bending was investigated by Gellin [3].
Shaw [4] and Kyriakides [5] analytically predicted the growth of ovality
for tubes under cyclic bending by using a number of nonlinear-hardening
plasticity models. Kyriakides and Ju [6, 7] studied the interaction of bifur-
cation and localization instability of thin-wall tube by experiments and
mathematical methods. Lee [8] used Hill’s quadratic anisotropic yield
function to predict bifurcation curvatures and wrinkle wavelengths which
are found to be in excellent agreement with experiment results. Karamanos
[9] proposed the nonlinear finite element method to investigate infinitely
long thin-wall tubes. Hauch [10] presented the predictions of ultimate
limit under various loading such as external pressure and pure bending
449
Inner UHMWPE Layer
Eight Reinforcement Layers
Outer HDPE Cover
Figure 22.1 Structure of FRFP.
and used the experiments to prove them. Takano [11] presented the
displacement-strain relation by investigating the buckling of moderately
thick anisotropic cylinders. Gong [12] proposed a two-dimensional the-
oretical model to study the buckling response of offshore pipeline under
several conditions like tension and bending.
However, the studies on bending behavior of bonded flexible pipe are
still not enough. Corona and Rodrigues [13] proposed the formulation
considering nonlinear pre-buckling response, the possibility of shell-type
bifurcation, and the material failure criterion by studying the responds
and first failure of long cross-ply composite tubes under pure bending.
Xia [14] studied the bending behavior of fiber-reinforced sandwich pipes
with filament-wound. Arjomandi and Taheri [15] investigated the buck-
ling and post-buckling response of sandwich pipe line under pure bend-
ing load by using finite element mothed. Li [16] used theoretical method
to predict the buckling moments of the plastic pipe reinforced by cross-
winding steel wire (PSPs) under bending. Bai [17] studied the collapse
of reinforced thermo-plastic pipe under external pressure and bending
moment by theoretical method and finite element method, and those two
methods showed great agreement with each other [18].
Fiberglass Reinforced Flexible Pipes Under Bending 451
In this chapter, based on the nonlinear ring theory and the principle
of virtual work, a model which simplifies the infinite length glass fiber
pipe into a two-dimensional plane is established in the theoretical part.
This model takes the nonlinear effect of materials into consideration by
using J2 flow theory and considers the contribution of glass fiber to rein-
forced layers by introducing the reduction factor. The Newton-Raphson
method is used to solve the equilibrium equations in MATLAB. In the
FEM part, a detailed FEM is established with the glass fiber embedded.
The result obtained from those two methods showed agreement with
experiment and it proves the using of theoretical method in engineering
is possible.
22.2 Experiment
The bending experiments mothed is usually known as three-point bending
test and four-point bending test. Kagoura [17] used the three-point bend-
ing test to investigate the bending stiffness of metallic flexible pipe, and
Zheng [19] used the same experiments mothed to get the bending stiffness
of other type of flexible pipes. Since the test section is not under pure bend-
ing load in three-point bending test, there is a stress concentration on the
test section. In this chapter, the bending moment-curvature relationship of
FRFP is obtained by four-point bending test.
22.2.1 Experimental Facility

A typical four-point bending test is carried out on an experimental facility,
whose diagrammatic sketch is shown in Figure 22.2. As shown in Figure
22.3 and 22.4, the four-point bending test is carried out in the horizon-
tal plane to eliminate the influence of the weight of the pipeline. Loading
beam which is pushed by a jack moves on the slider and the displacement
of the beam is recorded by a displacement gauge. The load generated by the
jack is recorded by a force sensor. According to the load and the displace-
ment which are recorded by the force sensor and the displacement gauge, a
moment-curvature relationship can be figured out. The dimensions of the
experimental facility are shown in Table 22.1.
The mechanical characteristics data of fiberglass shown in Table 22.1
are provided by the manufactory of the pipe. It should be noted that the
elongation at break of this kind of fiberglass is 1%~1.5%.
30 25
Engineering Stress-strain Engineering Stress-strain
True Stress-strain True Stress-strain
25 20
20
Stress (MPa)
Stress (MPa)
15
15
10
10
5
5
0 0
0.00 0.04 0.08 0.12 0.16 0.00 0.02 0.04 0.06 0.08 0.10 0.12 0.14 0.16
Strain Strain
(a) HDPE (b) UHMWPE
Figure 22.2 Stress-strain curve from tensile test.
Figure 22.3 The four-point facility.
2F
displacement gauge
Loading beam
Loading roller
Rigid region
r
Δ
Test section
θ
r
Support roller
Test section
L
l L1 l
L2
Figure 22.4 Diagrammatic sketch of facility.

Table 22.1 Dimensions of test facility.

Symbol Value (mm) Note
L 600 Length of test section
L1 800 Distance between loading points
L2 1,400 Distance between support points
l 300 Horizontal distance between loading point
and support point
Table 22.2 Material properties of testing specimens.

Materials Secant modulus E (MPa) Poisson’s ratio μ
Fiberglass 72607 0.3
HDPE 850 0.4
UHMWPE 570 0.4
22.2.2 Specimen
The FRFP specimen is formed by an inner UHMWPE, an outer HDPE, and
eight reinforced layers. HDPE and fiberglass constitute the eight reinforced
layers in which the winding angle of fiberglass in odd layers is +54.7° and
the winding angle of fiberglass in even layers is −54.7° (the converses direc-
tion compared to +54.7°). Material properties and manufacturing dimen-
sions of the specimens are shown in Tables 22.2 and 22.3.
22.2.3 Experiment Process

The thickness, outer diameter, and length of the test specimens were mea-
sured before the test. The measured dimensions of specimens are shown
in Table 22.4. Table 22.4 shows that the deviation between the maximum
and minimum diameter is 0.52%, and the average of the outer diameter is
0.5% deviated from the scheming manufacturing size. It also can be found
that the deviation of the measured thickness is 2.53%, which is also within
the error limit. Thus, it can be concluded that the experiment results of the
three specimens are comparable.

Parameter Value
Thickness of outer PE layer (mm) 3
Number of reinforced layers 8
Winding angle of the fiberglass (°) ±55
Table 22.4 Valid length and diameter of specimens.

Specimen Diameter (mm) Thickness (mm)
#1 76.53 13.47
#2 76.48 13.10
#3 76.13 13.41
During the test, the loading beam is pushed by a jack generating an

increasing thrust (2F) and slide forward at a constant speed. The displace-
ment of the loading beam is recorded as Δ by the displacement gauge.
When the loading beam is sliding forward, the FRFP tends to be bent
under the bending moment generated by the four rollers. The rigid region
is formed by inserting a cylindrical rigid rod into the section of specimens
between the support roller and the loading roller. According to the loading
condition of specimens, the test section can be regarded as a pure bending
region. In order to make sure that the pipe is subjected to a static load, the
loading process must be slow, stable and constant by keeping the speed of
the loading beam at about 0.2~0.4 mm/s. After the test, the displacement
of the loading beam (Δ) and the load (2F) is converted into the curvature
κ and moment M from the following expression:
 2r + D0 
∆ − l ⋅ tanθ + (2r + D0 ) − =0 (22.1)
 cosθ 
2θ
κ= (22.2)
L
F  1 
M= ⋅ − (2r + D0 )tanθ  (22.3)
cosθ  cosθ 
where θ is the inclination angle of the rigid region with respect to the hor-
izontal line, r is the diameter of the roller, D0 is the outer diameter of the
rigid region, l is the horizontal distance between the loading roller and the
support roller, L is the length of the test section, and F is half the thrust
generated by the jack.
22.2.4 Experimental Results

In the end of the test, the picture of samples was recorded and the
curvature-moment relationships were obtained. From Figure 22.5, it can
be observed that the samples were bent excessively under the ultimate
bending moment. Additionally, there is no sign of sagging on the pipe wall
and that fits the theory of localization instabilities of cylindrical shells.
Figure 22.6 shows that the moment gradually rises as curvature increases,
and three test curves are closed to each other. Although the test curves have
a slight fluctuation, they maintain a relatively steady rise. This means that
the test result is reliable. The summary of the bending test data is shown
in Table 22.5.
Figure 22.5 The four-point facility.

(a) Specimen 1
(b) Specimen 2
(c) Specimen 3
Figure 22.6 Bending deformation of three specimens.

Table 22.5 Summary of bending test data.

Specimen Displacement (mm) Curvature (m−1) Moment (N·m)
#1 263 2.94 1,879
#2 253 2.85 1,988
#3 262 2.93 1,993

22.3.1 Fundamental Assumption
Based on the actual stress characteristics of the samples, the following
assumptions are proposed to focus on the bending problem with the engi-
neering application point of view.
a. The FRFP consists of outer layers, inner layers and rein-

forcement layers, and the reinforcement layers consists of
HDPE matrix and filament glass fibers.
b. The materials including HDPE and filament glass fibers are
continuous, homogeneous, and flawless.
c. The layers are bonded tightly, which means no slippages
and gaps exist between the layers and interface of filament
glass fibers and HDPE matrix.
22.3.2 Kinematic Equation

A long, nearly circular pipe with radius R and wall thickness t loaded by
pure bending load is shown in Figure 22.6. The curvature and deformation
of the cross-section are uniform along the length of the pipe under a pure
bending load. As shown in Figure 22.7, w and z are respectively circumfer-
ential and radial displacements of the mid-surface. The Kirchhoff assump-
tion of “plane section remains plane and perpendicular to the mid-surface”
is adopted in the axial direction. Since the diameter-to-thickness (D/t)
ratio of the FRFP is relatively moderate, the effect of the transverse shear
on the cross-section must be taken into consideration and γθ represents
2500
1#
2#
2000 3#
Moment (N·m)
1500
1000
500
0
0.0 0.5 1.0 1.5 2.0 2.5 3.0
ĸ (m–1)
Figure 22.7 Curvature-moment curves of three test specimens.
the transverse shear deformation through the thickness. Under the small
deformation assumption, the axial strain is given:
ε x = ε x0 + ζκ (22.4)
ζ = (R + w)cos θ – v sin θ + z cos θ (22.5)
where ε x0 is the strain of the axis of the pipe and ζ is the distance from any
point on the deformed section to the neutral surface under bending, as
The circumferential strain is given:
2
 v′ + w  1  v′ + w  1  v − w′   γ′
εθ =   +   +   + z ⋅ κ θ + θ 
 R  2 R  2 R   R  (22.6)
2
 v ′ − w ′′   v − w′ 
κθ =  1− 
 R 2   R 
(22.7)
where (*)′ = (*),θ.

The transverse shear deformation is given:
γ θr = γ θ
(22.8)
22.3.3 Material Simplification

In this part, the glass fibers, in each of the reinforcement layer, are simpli-
fied as equidistant from each other according to the number of fibers in the
unit volume of the reinforcement layer, as shown in Figure 22.8.
As shown in Figure 22.8, a cylindrical coordinate system is established.
The coordinate axes r, θ, and z denote the radial, hoop, and axial direction,
respectively. The local material coordinate system of the reinforced layers
is designated as (1, 2, 3), where 1 is the direction of the fiber, 2 is the direc-
tion perpendicular to the aramid strand in the plane, and 3 is the normal
direction in the cylindrical coordinate system.
Based on a representative volume unit as shown in Figure 22.8, nine
elastic constants of the reinforced layers E1, E2, E3, G12, G13, G23, μ12, μ13, and
μ23 can be determined [20].
r
θ
φ
Glass fiber
2
2
1
Matrix (HDPE)
Glass fiber
1
glass fiber
II matrix
I
Representative volume unit
Figure 22.8 Simplification of reinforced layer.

E1 = EFBVFB + EPEVPE (22.9)
EFB EPEVI
E2 = + EPE (1 − VI ) (22.10)
VFB  VFB 
EPE + EFB  1 −
VI  VI 
VFB
EFB EPE
VI  V 
E3 = + EPE  1 − FB  (22.11)
EPEVFB + EFB (1 − VI )  VI 
VFB
GFBGPE
VI  V  (22.12)
G12 = G13 = + GPE  1 − FB 
GPEVI + GFB (1 − VI )  VI 
GFBGPEVI
G23 = + GPE (1 − VI ) (22.13)
VFB  VFB 
GPE + GFB  1 −
VI  VI 
ν12 = ν12 = νFBVFB + νPE(1 − VFB) (22.14)
EFB
ν FBVFB + ν PE (VI − VFB )
EPE
ν 23 = + ν PE (1 − VI ) (22.15)
VFB EFB  VFB 
+ 1−
VI EPE  VI 
where EPE, νPE, and GPE is the Young’s modulus, Poisson’s ratio, and shear
modulus of HDPE matrix; EFB, νFB, and GFB is the Young’s modulus,
Poisson’s ratio, and shear modulus of the fiberglass; VPE, VFB, and VI is
the volume fraction of the matrix, fiber and section I, as shown in Figure
22.8. In specific, the volume of the representative unit equals one and the
shaded part is the glass fiber.
(VPE + VFB = 1, VI + VII = 1)

The flexibility matrix defines the relation between stress and strain
vector for the i-th reinforced layer and can be expressed as the following
formulations:
i
i  1 ν12 ν13  i
 ε11   − − 0 0 0   σ 11 
   E1 E2 E3   
   ν12 1 ν 23   
 ε 22   − − 0 0 0   σ 22 
   E1 E2 E3   
   ν13 ν 1   
 ε 33   − − 23 0 0 0   σ 33 
  i i  E1 E2 E3   
  =S σ = 1   
 γ 23   0 0 0 0 0   τ 23 
   G23   
     
1
 γ 13   0 0 0 0 0   τ 13 
   G13   
     
 1  τ 12 
γ 12  

0 0 0 0 0 

G12
(22.16)
In the global cylindrical coordinate system, the flexibility matrix can

express as
s i = Ts i TT (22.17)
where
 cos 2 ϕ sin 2 ϕ 0 0 0 cosϕ sinϕ 

 
 sin 2 ϕ cos 2 ϕ 0 0 0 − cosϕ sin ϕ 
 
0 0 1 0 0 0 
T =
 0 0 0 cosϕ sinϕ 0 
 
 0 0 0 − sin ϕ cosϕ 0 
 −2 cosϕ sin ϕ
 2 cosϕ sinϕ 0 0 0 cos 2 ϕ − sin 2 ϕ 
(22.18)
Without taking σr, τzθ, and τrz into consideration, the stiffness matrixes
can express as
−1
 Si S12i S14i   Ci C12i C13i 
11 11
i
   
C =  S21i S22i S24i  =  C21
i i
C22 i
C23 
 i   
 S41 S42i S44i  i
 C31
i
C32 i
C33 
(22.19)
The circumferential stiffness effecting the cross-section’s ovality is

selected to estimate the degree of reduction factor [21].
K i = C22
i i
/E Matrix (22.20)
i
where E Matrix is the Young’s modulus of the matrix for the i-th layer.
22.3.4 Constitutive Model

In this chapter, the HDPE is modeled as non-elastic materials and the rein-
forced layers are regarded as homogenous materials of matrix (HDPE)
with a specified reduction factor Ki. Apparently, the reduction factor Ki of
the inner and outer layers equal one because there is no need of reduction
for isotropy material.
The stress-strain curve of the HDPE is provided by the FRFP manufac-
tures. By fitting the stress-strain curve, the fitting formula of the stress-
strain relations is given [17].
 σ  
2
σ σ
= EPE 1 − 0.7 + 0.001   (22.21)
ε  σ0  σ 0  
where EPE is the elastic modulus of HDPE and σ0 is the yield stress of HDPE,
the two of them can be calculated by using the method on ISO527-2012
[22].
Based on the J2 flow theory with isotropic hardening and only three
stress components are considered, the strain rate can be expressed as
 σ   ε 
 z   z 
 σ θ  = [D ]  εθ
i −1
 (22.22)
 τ   γ 
 θr   θr 
 1 + Q(2σ z − σ θ )2
1
Di =  −ν + Q(2σ z − σ θ )Q(2σ θ − σ z ) (22.23)
E
 3Q(2σ z − σ θ )τ θr
−ν + Q(2σ z − σ θ )Q(2σ θ − σ z ) 6Q(2σ z − σ θ )τ θr 


1 + Q(2σ θ − σ z )2 6Q(2σ θ − σ z )τ θr 

3Q(2σ θ − σ z )τ θr 1 + ν + 18Qτ θ2r 
(22.24)
 0, σ e < σ e ,max

Q= 1  E  (22.25)
 4σ 2  E − 1 , σ e ≥ σ e ,max
 e t
where ν is Poisson’s ratio, E is the elastic modulus, Et is the tangent modu-

lus, σe is the von Mises equivalent stress, and σe,max is the maximum equiv-
alent stress.
E = EPEKi, Et = EPE,tKi (22.26)
where E is the elastic modulus and Et is the tangent modulus of the

i-th layer of FRFP. Although the result of the numerical simulation
agrees with the experimental result very well, it should be noted that
the process of the numerical simulation is rather time-consuming and
demands great computing power. Therefore, using the simplified theo-
retical method to predict the ultimate bending moment and curvature
in bearing capacity limit states is a more efficient way for engineering
applications.
22.3.5 Principle of Virtual Work

As in the experiments, the pipe is bent by incrementally prescribing the
curvature, κ.
Under this loading, the Principle of Virtual Work (PVW) can be
expressed as
t2k
∑∫ ∫
2π
(σˆ xδε x + σˆθδεθ + τˆθrδγθr )(R + z )dz dθ = 0 (22.27)
0 t1k
f
where (• ) denotes a variable increment and (ˆ) • = (• + • ) . The problem

domain is discretized, using the following series expansions for the two
displacements:
w ≅ Ra0 + R ∑[a cos(nθ ) + b sin(nθ )]

n=1
n n (22.28)
v≅R ∑[c cos(nθ ) + d sin(nθ )]

n= 2
n n (22.29)
γθ ≅ R ∑[e cos(nθ ) + f sin(nθ )]

n=1
n n (22.30)
22.3.6 Algorithm of Analytical Solutions

The simultaneous equations consisting of Eqs. (22.27) to (22.30) has
6N nonlinear algebraic equations, and a computer program based on
the Newton-Raphson method is used to solve those 6N unknowns
{ε 0
}
, an , bn , cn , dn , en , fn .
x
The specific algorithm is as follows.

Step 1. Necessary parameters should be inputted in to this program,
such as, initial geometric dimensions, initial imperfections, and mate-
rial properties, etc. Gauss integration method is used to achieve the finite
integration problem and the pipe cross-section is discretized into l and m

elements along radial and circumferential directions.
Step 2. For the first iterative operation, the initial of 6N unknowns
are determined appropriately. For the specific iteration, the value of 6N
unknowns are calculated by calculation results of the previous step.
Step 3. The strain increment {ε } of the integral point is calculated by
using the kinematic equations, while the stress increment {σ } of the inte-
gral point is calculated by using the constitutive model.
Step 4. When the stress of the integral point is obtained, the 6N nonlin-
ear algebraic equations can be solved by using Newton-Raphson method,
and the solution of the equation is the value of 6N unknowns.
Step 5. Once the Newton-Raphson scheme converged, the actual dis-
placements, strains and stresses were updated automatically. Finally, the
bending moment at the current curvature can be calculated as
∑∫ ∫
2π t2
M= f
σˆ xζ ( R + z )dz dθ (22.31)
0 t1
f
22.4 Finite Element Method

In this part, a finite element model (FEM) is built to study the mechanical
behavior of FRFP by using ABAQUS/Standard nonlinear finite element
analysis tool. The geometrical dimensions of the FEM are in accordance
with those from manufacturing.
As shown in Figure 22.9, 3D solid element (C3D8I) is selected for inner
layers, outer layers, and the matrix of reinforced layers. This type of element
is more effective to provide relatively high accuracy than second-order ele-
ments [23]. Truss element (T3D2) is selected for the winding glass fibers.
Elastic-plastic material model with the von Mises criteria is used for HDPE.
As the fibers can only subject to tension, the glass fiber is regarded as linear
elastic materials and the compression capacity is removed. The truss elements
are embedded in the solid elements to model the perfect bonded condition.
As shown in Figure 22.10, symmetric boundary condition about the
axial direction (Z axis) are applied to the right side of the pipe and a kine-
matic coupling constraint (U1 = UR2 = UR3 = 0) is established at the
other end. This kind of constraint can apply a uniform rotation on the left
side and freely allow the ovality on the right side. The bending process is
achieved by applying the rotation at the reference point of the coupling
RP-1
Y Inner Layer
Matrix Y
Z X Outer Layer
Z X
(a) Inner, outer layer and matrix (b) one layer of glas-fibers
RP-1
Y Y
Z X Z X
(c) embedded fiber in matrix (d) the whole FEM model
RP-1
Y Y
Z X Z X
(e) 3D solid element (38CDI) (f) Truss element (32TD)
Figure 22.9 The FEM model of FRFP.
and a nonlinear static analysis (Nlgeom option) is selected for large dis-
placement effects.
22.5 Result and Conclusion

The comparison of curvature-moment curves between the experiment,
theoretical method, and finite element method is shown in Figure 22.11
This cure shows nonlinearity with the increasing bending, which is due to
U1 = U2 = 0
Coupling
On z-axis
symmetry
Reference
point
Y Y
X U1= 0 X
(a) Right side (U3 = 0) (b) Left side (U3 = 300)
Figure 22.10 Boundary condition of the FEM model.
2500
1#
2#
2000 3#
FEM
Theoretical
Moment (N·m)
1500
1000
500
0
0.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5 4.0 4.5
κ (m–1)
Figure 22.11 Comparison of results from bending test, theoretical method, and
numerical simulation.
the nonlinearities of material and geometrical deformation (mainly ovality

of cross-section). Due to the manual loading in the experiment, some test
data shows small fluctuations. Especially when the curvature is larger than
2m−1, the deviation between the experiment and FEM is a little bit large.
More concretely, the biggest deviation between experiment and FEM is
about 15.59% when the curvature is 3m−1. This phenomenon may mainly be
caused by the increasing friction between specimen and the rollers, and the
S, Mises
(Avg: 75%)
+4.272e+00
+3.920e+00
+3.569e+00
+3.218e+00
+2.867e+00
+2.516e+00
+2.164e+00
+1.813e+00
+1.462e+00
+1.111e+00
+7.565e-01
+4.083e-01
+5.707e-02
(a) Stress distribution of outer layer (κ = 1.05m–1)
S, Mises
(Avg: 75%)
+2.147e+01
+1.974e+01
+1.800e+01
+1.627e+01
+1.454e+01
+1.281e+01
+1.108e+01
+9.344e+00
+7.612e+00
+5.880e+00
+4.148e+00
+2.417e+00
+6.845e-01
(b) Stress distribution of outer layer (κ = 4.01m–1)

S, Mises
(Avg:75%)
+1.054e+01
+9.661e+00
+8.783e+00
+7.905e+00
+7.027e+00
+6.149e+00
+5.271e+00
+4.393e+00
+3.515e+00
+2.636e+00
+1.758e+00
+8.801e-01
+2.017e-03
(c) Stress distribution of the eighth layer fiber (κ = 1.05m–1)
S, Mises
(Avg: 75%)
+5.272e+01
+4.833e+01
+4.395e+01
+3.956e+01
+3.517e+01
+3.079e+01
+2.640e+01
+2.201e+01
+1.762e+01
+1.324e+01
+8.850e+00
+4.463e+00
+7.598e-02
(d) Stress distribution of the eighth layer fiber (κ = 4.01m–1)
Figure 22.12 Stress distribution in FEM simulation.

uneven speed of loading is another reason for it. In general, the result shows
great agreement with each other at the period whose curvature is equal to
or less than 2m−1, and the difference between the theoretical method and
finite element method is close to each other at this period. It proves that
the simplification made in the theoretical method is reasonable when the
curvature is relatively small. When the curvature is larger than 2m−1, the
deviation of the theoretical method from finite element method increases a
little and the biggest deviation is about 4.68% less than 5% which is accept-
able in practical engineering. The main reason of this phenomenon is that
the transverse shear deformation is assumed as one-order linear in the the-
oretical method. As shown Figure 22.12, when curvature is small, the stress
of fiber is no more than 2.64 MPa basically; however, the stress of the outer
layer is generally larger than 3.21 MPa. It means that the fiber contributes
a little to the bending stiffness of pipe when the curvature is small. Instead,
when the curvature is large, the stress of the both sides of the layer of fibers
reach 30.79 MPa, which is larger than the maximum stress (21.47 MPa) of
the outer layer. It means that the fibers can play a function of constraining
to the radial deformation of the cross-section to prevent ovalization when
the curvature is relatively large. Therefore, using the reduction factor Ki
to consider the fiber’s contribution to constraint is another reason for the
difference between the theoretical method and finite element method. In
conclusion, the contribution of fibers is too small to be considered before
the curve reaches the ultimate bending moment and the result of curva-
ture-bending moment relations obtained from three methods agree well
with each other, which can prove that the simplified analytical model and
FEM are accurate and reliable.

The previous part proves the theoretical method is reliable. Therefore, this
method is used to analyze some parameters in this part.
22.6.1 D/t Ratio

D/t Ratio is a key parameter to estimate the bending stiffness of FRFP. In the
manufactory of FRFP, the larger the diameter of FRFP is, the more the glass
fibers would be used on the matrix, so it is hard to estimate the influence
of the diameter of FRFP on the bending stiffness by using the theoretical
method. Therefore, changing the thickness of the pipe wall is an easy way
to change D/t ratio. As shown in Figure 22.13, the limit moment and limit
3.4
κc (m–1) 3.2
3.0
2.8
2.6
6 7 8 9 10
D/t
(a) limit curvature--D/t curve
1800
1700
1600
1500
Mc (N•m)
1400
1300
1200
1100
6 7 8 9 10
D/t
(b) limit moment--D/t curve
Figure 22.13 Effect of wall-thickness.
curvature decrease along with the increasing thickness of FRFP, whlie the
decreasing rate slows down. More concretely, the D/t ratio raises from 6
to 10, it means that the thickness of the pipe decreases from 13 to 7.8 mm
when the diameter is 78 mm. Along with the increase of the D/t ratio, the
limit curvature and moment decreases 27% and 36% separately. This illus-
trates that the D/t ratio has a great effect on the limit moment and curvature.
22.6.2 Initial Ovality

A perfect circular cross-section of pipe doesn’t exist in engineerings, while
a main reason for initial geometric imperfection of pipe is initial ovality.
The influence of the initial ovality on bending response is related to the
orientation between the bending axis and the major axis of the ellipse, and
the most dangerous situation happens when the two axes coincide. As shown
in Figure 22.14, the limit cuvature-initial ovality and the limit moment-
initial ovality relationship are nealy linenear. More concretely, along with the
increase of the initial ovality from 0% to 5%, the limit curvature and moment
3.55
3.54
3.53
3.52
κc (m–1)
3.51
3.50
3.49
3.48
0 1 2 3 4 5
∆0 (%)
(a) limit curvature--∆0 curve
1840
1820
1800
1780
Mc (N•m)
1760
1740
1720
1700
1680
1660
0 1 2 3 4 5
∆0 (%)
(b) limit moment--∆0 curve
Figure 22.14 Effect of Δ0.

decreases only 1.4% and 8.2% separately. It can be concluded that the influ-
ence of the initial ovality on the limit curvature and moment is small.
22.7 Conclusions
This chapter presents experimental studies of the three 10-layer FRFP on
the four-point bending experiment machine in Research Lab for Civil
Engineering in China. The curvature-moment relations were recorded
during the test. Based on the nonlinear ring theory, and by introducing
the effect of the transverse shear deformation and the simplified method of
reinforced layers, the theoretical method is proposed to estimate the oval-
ization instability of FRFP under the pure bending load. Then, a 10-layer
FEM with different materials was established for a further study of the
mechanical behavior of FRFP under the bending load. Moment-curvature
relations obtained from the three methods agree with each other very well.
It proves that the theoretical method can accurately predict the response of
ovalization instability under pure bending. From comparing and discuss-
ing, the key conclusions can be drawn.
a) Comparisons are carried out between the theoretical method, numer-
ical simulation using ABAQUS and the bending test, which proves that the
54.7° winding glass fiber contributes a little to stiffness of FRFP before
collapse. Thus, the flexibility of FRFP is almost the same with the unrein-
forced HDPE pipe.
b) The theoretical method can predict mechanical behavior of the
FRFP before collapse of the pipe wall, and the reasons of slight differences
between the theoretical method and finite element method after collapse
are as follow: 1) the use of the reduction factor considers the fiber’s contri-
bution to constraint and 2) the assumption of one-order linear transverse
shear deformation.
c) Thickness of the pipe wall is a key parameter which contributes a lot
to the stiffness under pure bending, and the increasing effects on stiffness
drops down along with the increase of thickness. It means that increasing
the thickness of pipe wall is an efficient way to improve the limit curvature
of FRFP under pure bending.
d) Initial ovality has a relatively little effect to the limit curvature and
moment when the initial ovality is less than 5%. It means that the loss of
stiffness caused by initial ovality when its value is less than 5%, which is
acceptable in practical engineering.
References
1. Brazier L G. On the flexure of thin cylindrical shells and other “thin” sec-
tions[J]. Proc. R. Soc. Lond. A, 1927, 116(773): 104–114.
2. Reissner E, Weinitschke H J. Finite pure bending of circular cylindrical
tubes[J]. Quarterly of applied mathematics, 1963, 20(4): 305–319.
3. Gellin S. The plastic buckling of long cylindrical shells under pure bend-
ing[J]. International Journal of Solids and Structures, 1980, 16(5): 397–407.
4. Shaw P K, Kyriakides S. Inelastic analysis of thin-walled tubes under cyclic
bending[J]. International Journal of Solids and Structures, 1985, 21(11):
1073–1100.
5. Kyriakides S, Shaw P K. Inelastic buckling of tubes under cyclic bending[J].
Journal of Pressure Vessel Technology, 1987, 109(2): 169–178.
6. Kyriakides S, Shaw P K. Inelastic buckling of tubes under cyclic bending[J].
Journal of Pressure Vessel Technology, 1987, 109(2): 169–178.
7. Ju G T, Kyriakides S. Bifurcation and localization instabilities in cylindrical
shells under bending—II. Predictions[J]. International journal of solids and
structures, 1992, 29(9): 1143–1171.
8. Corona E, Lee L H, Kyriakides S. Yield anisotropy effects on buckling of cir-
cular tubes under bending[J]. International Journal of Solids and Structures,
2006, 43(22–23): 7099–7118.
9. Karamanos S A. Bending instabilities of elastic tubes[J]. International Journal
of Solids and Structures, 2002, 39(8): 2059–2085.
10. Hauch S R, Bai Y. Bending moment capacity of pipes[J]. Journal of Offshore
Mechanics and Arctic Engineering, 2000, 122(4): 243–252.
11. Takano A. Buckling of thin and moderately thick anisotropic cylinders under
combined torsion and axial compression[J]. Thin-Walled Structures, 2011,
49(2): 304–316.
12. Gong S, Yuan L, Jin W. Buckling response of offshore pipelines under
combined tension, bending, and external pressure[J]. Journal of Zhejiang
University-SCIENCE A, 2011, 12(8): 627–636.
13. Corona E, Rodrigues A. Bending of long cross-ply composite circular
cylinders[J]. Composites Engineering, 1995, 5(2): 163–182.
14. Xia M, Takayanagi H, Kemmochi K. Bending behavior of filament-wound
fiber-reinforced sandwich pipes[J]. Composite structures, 2002, 56(2):
201–210.
15. Arjomandi K, Taheri F. Bending capacity of sandwich pipes[J]. Ocean
Engineering, 2012, 48: 17–31.
16. Li X, Zheng J, Shi F, et al. Buckling analysis of plastic pipe reinforced by
cross-winding steel wire under bending[C]//ASME 2009 Pressure Vessels
and Piping Conference. American Society of Mechanical Engineers, 2009:
259–268.
17. Bai Y, Tang J, Xu W, et al. Collapse of reinforced thermoplastic pipe

(RTP) under combined external pressure and bending moment[J]. Ocean
Engineering, 2015, 94: 10–18.
18. Kagoura T, Ishii K, Abe S, et al. Development of a flexible pipe for pipe-in-
pipe technology[J]. Furukawa Review, 2003, 24: 69–75.
19. Zheng Jiexin. Verification and Buying Research on Design and Analysis of
Marine Non-cohesive Flexible Pipe [D]. Master’s Degree Thesis, Dalian
University of Technology, 2010.
20. Zhu Yancong. Study on the External Pressure Instability of Wire-wound
Reinforced Plastic Composite Pipes [D]. Master’s Degree Thesis of Zhejiang
University. 2007
21. Bai Y., Wang N. S., Cheng P. Collapse of RTP (reinforced thermoplas-
tic pipe) subjected to external pressure[C]. International Conference on
Pipelines and Trenchless Technology, 2012.
22. ISO, ISO527-2012, Plastics—Determination of Tensile Properties, Geneva,
Switzerland.
23. ABAQUS 6.10, 2010, ABAQUS Analysis of User’s Manual, Dasssult
Systemes, Velizy-Villacoublay, France.
23
Fiberglass Reinforced Flexible
Pipes Under Torsion
23.1 Introduction
Recently, fiberglass reinforced flexible pipe (FRFP) has been widely used
in offshore transportation due to its high corrosion resistance, light
weight characteristic, and relatively low fabrication and installation cost.
Compared to other types of bonded pipes, it has a higher density and
shows better subsea mechanical behavior.
FRFP studied in this chapter, shown in (Figure 23.1), consists of a poly-
ethylene liner, eight layers of reinforced tape made of polyethylene and
fiberglass wrapping around the liner and an outer polyethylene coating. The
inner liner pipe is ultra-high molecular weight polyethylene (UHMWPE)
and the outer coating pipe is high density polyethylene (HDPE), while the
reinforced tape is made of HDPE and fiberglass.
Flexible pipes under various loads have been studied by many scholars;
however, most of the research is confined to unbonded flexible pipes. On
the numerical side, A. Bahtui et al. [1, 2] modeled an unbonded flexible
riser under various loads in which all layers are separated with contact
interfaces between each layer. Hector E. M. Merino et al. [3] created a
FEM under torsion considering the friction and adhesion between layers,
the results show that finite element estimations agree quite well with the
experimental measurements. Then, M. S. Liu et al. [4] imported a model
into ABAQUS to simulate the riser’s mechanical behavior under torsion.
This model takes into consideration material nonlinearity and nonlin-
ear boundary conditions. On the analytical side, Bahtui et al. [2] created
an analytical model based on the hypothesis of small displacements and
strains. REN Shao-fei [5] exhibited an analytical model taking local bend-
ing and torsion of helical strips into consideration. Results show these
two factors may have great influence on torsional stiffness. Yong Bai et al.
[6] developed a modified theoretical model based on the model initially
475
External Coating
Reinforced Tapes
Internal Liner
Figure 23.1 Structure of FRFP.
proposed by Knapp et al. [7]. An FEM under tension-torsion load was

established and compared with the theoretical model. The results of these
two methods agree well with each other.
As for the bonded flexible pipes, M. Xia et al. [8] developed an analysis
method for multi-layered filament-wound composite pipes under internal
pressure in the elastic range by assuming each layer of the pipes as aniso-
tropic. Yong Bai et al. [9] investigated the mechanical properties of rein-
forced thermoplastic pipe under internal pressure. The analytical part was
extended from the theory initially proposed by M.P. Kruijer et al. [10]. It
introduced a new model based on a plane strain characterization and non-
uniform strain distribution through the thickness. Y. Zhao et al. [11] devel-
oped an analytical model based on the mechanics of composite materials
and performed a finite element analysis using the software COSMOS/M,
then composite pipe samples were tested to obtain the relationship between
applied torque and torsional angles. Results shows that the relationship
of these three methods agree with each other well.
However, research on bonded flexible pipes is still not quite enough.
Previous studies have not given detailed solutions about bonded flexible
pipes. In the analysis of this chapter, a 3D analytical model of FRFP was
established to explore the mechanical properties in the elastic range. FRFP
in the analytical solution was considered as a thick cylinder and the stress
situation in the pipes was characterized as generalized plane strain. The
inner layer and outer layer were both considered to be homogeneous, con-
tinuous, and isotropic while the reinforced layers were considered to be
homogeneous, continuous, and transverse isotropic. Because all the layers
are bonded together firmly, the strains of all layers were assumed to be
continuous at the interface, and the interface remains constant during the
Fiberglass Reinforced Flexible Pipes Under Torsion 477
analysis. In addition, the cross-section of the FRFP always remained per-

pendicular to its central axis and the deformation in the axial direction was
assumed to be uniform. Then, an FEM was established to simulate the real
situation when the FRFP was subjected to torque. Finally, three specimens
were subjected to torque to obtain the torque-torsion angle relationship.
Results from these three methods were compared, ultimately proving the
reliability of the analytical model and FEM.
23.2 Experiments
A full-scale laboratory test was conducted at the Ningbo-OPR factory in
China to study the mechanical behavior of FRFP subjected to pure torsion.
Specimens used in this research were produced by Shang Hai Fei Zhou Bo
Yuan Petroleum Equipment Company. Components of the FRFP are: an
inner UHMWPE, an outer HDPE, and eight reinforced layers consist of
HDPE and fiberglass. The fiberglass in odd reinforced layers winds in one
direction, while the fiberglass in other reinforced layers winds in the con-
verse direction. Detailed material properties and geometric parameters of
the specimens are listed in Tables 23.1 and 23.2.
Experimental Process
a) Measured the valid length and diameter of the specimens
and recorded the data (see Table 23.3).
b) Conditioned the specimen at room temperature for no less
than 2 hours.
c) Assembled the samples onto the torsion testing machine and
took care to align the axis of the specimen with the end con-
nectors of torsion machine.
d) Tightened the end connectors evenly and firmly to the
necessary degree in order to prevent slippage between
Table 23.1 Material properties of testing specimens.

Materials Secant modulus E (MPa) Poison’s ratio μ
Fiberglass 72,607 0.3
HDPE 850 0.4
UHMWPE 570 0.4

Parameter Value
Number of reinforced layers 8
Winding angle of the fiberglass (°) ±55
Table 23.3 Valid length and diameter and diameter of specimens.

Specimen Valid length (mm) Diameter (mm)
#1 1,005 76.53
#2 1,008 76.48
#3 998 76.13
the specimen and end connectors of the testing machine.

Adjusted the test machine dial pointer to zero.
e) Set the speed of the machine at 0.18 deg/min and started the
machine. Tested the specimen until the pipe failed.
f) Recorded the data of the relation between torque and twist
angle, found the failure angle of the sample pipes.
23.3 Experimental Results

During the experiment process, the samples were completely fixed at
one end, and only a torsion angle was applied on the other end. It can be
observed that the samples were twisting along with the fixed side and slight
bulge was gradually appearing on the surface of the pipe, followed by con-
tinuous snapping sound of fiberglass in the pipe. There were small areas of
the pipe that showed twisting deformation. However, these deformations
were located at different points on the pipe for each of the three samples.
The bulge area of the first sample was located at 1/4 of distance from the
pipe’s fixed end, and the bulge area of the second sample was located at 2/5
of distance from the fixed end, while the bulge area of the third sample was
located at 1/5 of distance from the fixed end. The deformations of three
samples are shown in Figure 23.2.
In order to determine reasons for the failure of FRFP, a section along the
circumferential direction at the twist shape of the pipe was cut so that the
deformation situation inside the pipe could be examined. It can be seen
from Figures 23.3 and 23.4 that some gaps appear in the reinforced layer
and there is some arcuate deformation on the inner layer; however, these
gaps on the reinforced layers appear at different areas for each specimen.
The cross-section form of specimen 1 is similar to Figure 23.3. By remov-
ing the outer PE layer of specimen 1, it can be seen in Figure 23.5 that the
outermost reinforced layer separates slightly from the inner layers.
As shown in Figure 23.6, the torque-torsion angle relationship of three
test specimens closely aligns with one another. In the beginning phase,
Specimen 1
Specimen 2
Specimen 3
Figure 23.2 Torsion deformation of three specimens.

Figure 23.3 The cross-section in the bulge area of Specimen 2.
Figure 23.4 The cross-section in the bulge area of Specimen 3.
Figure 23.5 The damage of reinforced layers of Specimen 1 after parting from outer layer.
3000
2500
2000
Torque (N·m)
1500
1000
500 specimen 1
specimen 2
specimen 3
0
0.0 0.1 0.2 0.3 0.4 0.5 0.6
Torsion Angle (rad)
Figure 23.6 Torque-torsion angle curves of three test specimens.
there are some fluctuations of the first sample which may be due to the fact
that connectors of the torsion machine were not tight enough. Therefore,
the torsion angle and small amount of torque for the first specimen in the
beginning period is not counted in the results. After passing the unstable
state, the torque goes up steadily until it drops suddenly at a peak point,
and the maximum torque recorded is regarded as the failure torque. The
failure torque obtained from the tests are 2,343 N•m for the first sample,
2,650 N•m for the second sample, and 2,777 N•m for the third one, which
demonstrate low variability. The average value of the failure torque is 2,590
N•m. The maximum torque difference of these three specimens is 15.6%,
and the maximum rotation angle difference is 8.24%, demonstrating the
accuracy and repeatability of the experiment. The above differences came
into being because of the different directions the torque applied on the
pipe’s end, the different winding angle of fiberglass in the outermost rein-
forced layer, the different amount of fiberglass, and the different distance
between each fiberglass during fabrication.

23.4.1 Coordinate Systems
A cylindrical coordinate system was used, as shown in Figure 23.7. The coor-
dinate axes r, θ, and z denote radial, hoop, and axial direction, respectively.
The local material coordinate system of the reinforced layers is desig-

nated as (L, T, r), where L is the direction in which the fibre winded, T is
the direction perpendicular to the aramid strand in the plane, and r is the
normal direction just as in the cylindrical coordinate system.
The relation of the local material coordinate system (on-axis coordi-
nate) and the cylindrical coordinate system (off-axis coordinate) is shown
in Figure 23.8.
θ r
φ –φ
σzz
σrr
σθθ
Figure 23.7 Cylindrical coordinate system.
r, r
T Z
θ L
Ф
Figure 23.8 Relationship between on-axis coordinate (L, T, r) and off-axis coordinate
(z, θ, r).
23.4.2 Elastic Constants of Reinforced Layers

(k = 2, 3 … (n − 1))
In order to investigate the mechanical properties of the reinforced layers,
elastic parameters of reinforced layers should be determined first.
Based on a representative volume unit (RUV), nine elastic constants of
the reinforced layer EL, ET, Er, GLT, GLr, GTr, μLT, μLr, μTr can be determined
[12]. It is assumed E2 = E3 due to the similar property in these two direc-
Π
tions. In these formulas, VI = , VII = 1 – VI, VFB = 60% and VPE = 40%.
2
The results of elastic constants are listed in Table 23.4.
EL = EFBVFB + EPEVPE (23.1)
EFB EPEVI
ET = Er = + EPE (1 − VI ) (23.2)
VFB  VFB 
EPE + EFB  1 −
VI  VI 
VFB
GFBGPE
VI  V  (23.3)
GLT = GLr = + GPE  1 − FB 
GPEVI + GFB (1 − VI )  VI 
Table 23.4 Elastic constants of reinforced layers.

Elastic constants Value
EL 27400MPa
ET 1379MPa
Er 1379MPa
GLT 1204MPa
GLr 1204MPa
GTr 493MPa
μLT 0.36
μLr 0.36
μTr 0.40
GFBGPEVI
GTr = + GPE (1 − VI ) (23.4)
VFB  VFB 
GPE + GFB  1 −
VI  VI 
μLT = μLr = μFBVFB + μPE(1 − VFB) (23.5)
EFB
µ FBVFB + µ PE (VI − VFB )
EPE
µTr = + µ PE (1 − VI ) (23.6)
VFB EFB  VFB 
+ 1−
VI EPE  VI 
23.4.3 Reinforced Layers Stiffness Matrix k = 2, 3...(n – 1)

The reinforced layers are considered transversely isotropic with the same
elastic properties on the plane (T − r), the on-axis stiffness matrix C(k) of
k-th layer is given by
C(k) = (S(k))−1, k = 2,3…n – 1 (23.7)
 1 µ LT µ LT 
 − − 0 0 0 
 EL EL EL 
 µ 1 µ 
 − LT − Tr 0 0 0 
 EL ET EL 
 µ µ 1 
 − LT − Tr − 0 0 0 
EL EL ET 
S( k ) = 
 1 
 0 0 0 0 0 
 GTr 
 1 
 0 0 0 0 0 
 GLT 
 1 
 0 0 0 0 0 
 GLT 
(23.8)
where the symbol S is on-axis flexibility matrix. The superscript “−1” in Eq.
(23.7) denotes the inverse matrix.
By introducing the flexibility and stiffness transformation matrices Tε
and Tσ which are a function of the angle α, the relationships between the
off-axis and on-axis elastic constants are expressed as
(
S ( k ) = Tε( k )S ( k ) Tε( k ) )
T
(23.9)
(
C ( k ) = Tσ( k )C ( k ) Tσ( k ) )
T
, k = 2, 3…n − 1 (23.10)
where
(k )
 m2 n2 0 0 0 −mn 
 2

 n m2 0 0 0 mn 
 
Tε( k ) =  0 0 1 0 0 0
 (23.11)
 0 0 0 m n 0 
 0 0 0 −n m 0 
 2mn −2mn 0 0 0 m − n2
2 
 
(k )
 m2 n2 0 0 0 −2mn 
 2

 n m2 0 0 0 2mn 
 
Tσ( k ) =  0 0 1 0 0 0
 (23.12)
 0 0 0 m n 0 
 0 0 0 −n m 0 
 mn −mn 0 0 0 m − n2
2 
 
m = cos α, n = sin α, where α is the winding angle of the fiberglass.

According to Eqs. (23.7) and (23.8), on-axis stiffness matrix can be
written as
(k )
 C11 C12 C12 0 0 0 
 
 C12 C22 C23 0 0 0 
 C C23 C22 0 0 0 
C =  12
(k )  (23.13)
 0 0 0 C44 0 0 
 
 0 0 0 0 C55 0 
 0 0 0 0 0 C55 
 
According to Eq. (23.10), off-axis stiffness matrix can be written as
(k )
 C11 C12 C13 0 0 C16 
 
 C12 C22 C23 0 0 C26 
 C13 C23 C33 0 0 C36 
C (k ) =   (23.14)
 0 0 0 C44 C45 0 
 
 0 0 0 C45 C55 0 
 C16 C26 C36 0 0 C66 

23.4.4 Inner Layer and Outer Layer Stiffness Matrix (k = 1, n)

As a homogenous isotropic material, the off-axis stiffness matrix C ( k ) of
k-th layer is given as
(k )
 C11 C12 C13 0 0 0 
 
 C21 C22 C23 0 0 0 
 C C32 C33 0 0 0 
−(k )
C =  31  (23.15)
 0 0 0 C44 0 0 
 
 0 0 0 0 C55 0 
 0 0 0 0 0 C66 
 
It can be calculated by
C ( k ) = (S ( k ) )−1 , k = 1, n. (23.16)
(k )
 1 µ µ 
 − k − k 0 0 0 
 Ek Ek Ek 
 µ 1 µk 
 − k − 0 0 0 
 Ek Ek Ek 
 µ µk 1 
 − k − 0 0 0 
Ek Ek Ek
S ( k ) =  
 (23.17)
1
 0 0 0 0 0 
 Gk 
 1 
 0 0 0 0 0 
 Gk 
 1 
 0 0 0 0 0 
 Gk 
where the symbol S is the off-axis flexibility matrix. Ek is the elastic mod-
ulus of PE and μk is the Poisson’s ratio of PE. The superscript “−1” in Eq.
(23.16) denotes the inverse matrix.
23.4.5 Stress and Deformation Analysis

When the pipe is subjected to axial symmetric loads ∂
∂θ ( )
= 0 , the stresses
and strains are independent of θ. In addition, the radial and axial displace-
ments do not depend on the axial z and radial r directions, respectively.
With the above assumption, the displacement fields can be expressed by
ur = ur(r), uθ = uθ(r, z), uz = uz(z) (23.18)
where, ur, uθ, and uz are radial, hoop, and axial displacements, respectively.
The strain-displacement relations can be described as
(k ) ∂ur( k ) (k ) 1 ∂uθ( k ) ur( k ) (k ) ∂uz( k )

ε = r , εθ = + , εz = ,
∂r r ∂θ r ∂z
1 ∂uz( k ) ∂uθ( k ) ∂u( k ) ∂u( k )
γ z(θk ) = + , γ zr( k ) = z + r , (23.19)
r ∂θ ∂z ∂r ∂z
1 ∂ur (k )
∂ u ( k )
γ θ( kr ) = +r  θ 
r ∂θ ∂r  r 
Substituting Eq. (23.18) into Eq. (23.19), Eq. (23.19) can be simplified as
∂ur( k ) ( k ) ur( k ) ( k ) ∂uz( k )

ε r( k ) = , εθ = , εz = = ε 0( k ) ,
∂r r ∂z
∂  u( k )  ∂u( k ) u( k ) ∂u( k )
γ zr( k ) = 0, γ θ( kr ) = r  θ  = θ − θ , γ z(θk ) = θ = γ 0r
∂r  r  ∂r r ∂z
(23.20)
where γ0 is twist of pipe per unit length which is independent of z. The

axial deformation of all layers is equal to a constant ε0.
In the absence of body forces, equilibrium equations in cylindrical coor-
dinates can be described as
∂σ r( k ) 1 ∂τ θ( kr ) ∂τ zr( k ) σ r( k ) − σ θ( k )
+ + + =0 (23.21a)
∂r r ∂θ ∂z r
∂τ θ( kr ) 1 ∂σ θ( k ) ∂τ z(θk ) 2τ θ( kr )
+ + + =0 (23.21b)
∂r r ∂θ ∂z r
∂τ zr( k ) 1 ∂τ z(θk ) ∂σ z( k ) τ zr( k )

+ + + =0 (23.21c)
∂r r ∂θ ∂z r
Reinforced Layers Analysis (k = 2,3…(n − 1))

The stress-strain relations of the k-th layer are given as
{σ }( k ) = C ( k ) {ε−}( k )
(k) (k) (k)

 σz   C11 C12 C13 0 0 C16   εz 
     
 σθ   C12 C22 C23 0 0 C26   εθ 
 σr   C13 C23 C33 0 0 C36   εr 
   
  =  
 τ θr   0 0 0 C44 C45 0   γ θr 
     
τ zr  0 0 0 C45 C55 0  γ zr
    
 τ zθ   C16 C26 C36 0 0 C66   γ zθ 
Substituting Eqs. (23.20) and (23.23) into Eq. (23.21), the equilibrium
equations lead to
∂σ r( k ) σ r( k ) − σ θ( k )
+ =0 (23.24a)
∂r r
∂τ θ( kr ) 2τ θ( kr )
+ =0 (23.24b)
∂r r
∂τ zr( k ) τ zr( k )
+ =0 (23.24c)
∂r r
Substituting the stress-strain relation of Eq. (23.23) into Eq. (23.24),

and using Eq. (23.20), we get the expressions of ur, uθ, and uz of reinforced
layers through solving equation
uθ( k ) = γ 0rz + A( k )
uz( k ) = ε 0 z + B( k )
(k) (k)
ur( k ) = D ( k )r β + E ( k )r − β + α1( k )ε 0r + α 2( k )γ 0r 2
(23.25)
where, A(k), B(k), D(k), E(k) are unknown integration constants to be

determined.
β ( k ) = C22
(k ) (k )
/ C33 (23.26a)
C12( k ) − C13( k )
α1( k ) = (k ) (k ) (23.26b)
C33 − C22
(k ) (k )
C26 − 2C36
α 2( k ) = (k ) (k ) (23.26c)
4C33 − C22
For the anisotropic materials (reinforced layers) used in this study, there
exists
(k ) (k ) (k ) (k )
C22 /C33 > 0 and C22 /C33 ≠ 1.
Using transformation matrix Tσ( k ) and Tε( k ), strains and stresses in local
material coordinates can be obtained:
{ε }( k ) = Tε( k ) {ε }( k )
(23.27)
{σ }( k ) = Tσ( k ) {σ }( k )
Inner Layer and Outer Layer Analysis (k = 1, n)

For isotropic materials, the stress-strain relations of the k-th layer are
given as
{σ }( k ) = C ( k ) {ε }( k )
(k) (k) (k)

 σz   C11 C12 C12 0 0 0   εz 
     
 σθ   C12 C11 C12 0 0 0   εθ 
 σr   C C12 C11 0 0 0   εr 
   
  =  12   
 τ θr   0 0 0 C44 0 0   γ θr 
     
τ zr  0 0 0 0 C44 0  γ zr
   
 τ zθ   0 0 0 0 0 C44   γ zθ 
 
Substituting Eqs. (23.20) and (23.29) into equilibrium equation

(23.21), the equilibrium equations lead to
∂σ r( k ) σ r( k ) − σ θ( k )
+ =0 (23.30a)
∂r r
∂τ θ( kr ) 2τ θ( kr )
+ =0 (23.30b)
∂r r
(k) (k)
As an isotropic material, there exits C22 / C33 = 1, C12( k ) = C13( k ) and
(k) (k)
(k)
C26 (k)
= C36 = 0, which leads to β(k) = 1, α1 = α 2 = 0 (k = 1, n) in Eq.
(23.26).
Hence, the solutions of ur, uθ, and uz in the inner and outer layers can
be expressed by
uθ( k ) = γ 0rz + A( k )
uz( k ) = ε 0 z + B( k )
(23.31)
E(k )
ur( k ) = D ( k )r +
r
where, A(k), B(k), D(k), E(k) are unknown integration constants to be determined.

The unknown integration constants in Eqs. (23.25) and (23.31) can be
determined by substituting these equations into boundary conditions and
solving the algebraic equations.
The traction-free condition at the inner surface and outer surface are
written as
σ r(1) (r0 ) = 0, σ r(n ) (rn ) = 0 (23.32a)
τ θ(1r) (r0 ) = τ zr(1) (r0 ) = 0, τ θ(nr ) (rn ) = τ zr(n ) (rn ) = 0 (23.32b)
where r0 and rn are the inner and outer radius, respectively.

23.4.7 Interface Conditions

Assuming that the interfaces between the core and skin layers are perfectly
bound, continuity conditions for the displacements and stresses at the
interfaces lead to
ur( k ) (rk ) = ur( k+1) (rk ) (23.33a)
uθ( k ) (rk ) = uθ( k+1) (rk ) (23.33b)
σ r( k ) (rk ) = σ r( k+1) (rk ) (23.33c)
τ zr( k ) (rk ) = τ zr( k+1) (rk ) (23.33d)
τ θ( kr ) (rk ) = τ θ( kr +1) (rk ) (23.33e)
where, k = 1, 2⋯n − 1.
Axial equilibrium for a cylinder with closed ends
∑∫
rk
2π σ z( k ) (r )r dr = 0 (23.34)
rk −1
k =1
Torsion condition
∑∫
rk
2π τ z(θk ) (r )r 2 dr = T (23.35)
rk −1
k =1
Using Eq. (23.32b), Eqs. (23.33d) and (23.33e) can derive that the
integration constant, A(k) = 0. Therefore, the hoop displacement can be
expressed as follows:
uθ = γ0rz (23.36)
For N-layered FGRFP, there are 2n + 2 unknown constants of integra-

tion, D(k), E(k) (k = 1, 2⋯n), and ε0, γ0. Eqs. (23.32a), (23.33a), (23.33c),
(23.34), and (23.35) can give 2n + 2 equations to determine these unknown
constants.
The integration constants for the multi-layered FGRFP can be obtained
from the solution of the simultaneous equation as Eq. (23.37). Elements
in the matrix M can be obtained through transformations of 2n + 2
equations.
 D (1)   
  0
D(2)  
   0 
     
   
 D (n )  0
 
 E (1)   0 
  = [ M ]2n+2 ,2n+2   (23.37)
 E(2)   0 
     
 E (n )   0 
   
 ε0   0 
 γ0   T / (2 * pi ) 
  2n+ 2 ,1
Once their values are determined, the strains and displacements are
thus obtained from Eqs. (23.20), (23.25), and (23.31), respectively. Then,
the stresses can be calculated as products of strains and stiffness matrices.
23.4.8 Geometric Nonlinearity

As a result of the deformation, the change of cross-section dimension and
the fiber angles are no longer negligible. Nonlinearity should be taken
into account in the analysis. The change of cross-section dimension can

be obtained from the variable radius displacements. The change of fiber
angles in the reinforced layers can be calculated from the following equa-
tion proposed by Kruijer (2005) [1]:
α ( k ) = α 0( k ) + (−ε z + εθ )sin α 0( k ) cosα 0( k ) + γ zθ cos 2 α 0( k ) (23.38)
where α 0(k ) is the initial wound angle of the k-th reinforced layer. Change
of the wound angle leads to a nonlinear stiffness matrix of the reinforced
layer in the cylindrical coordinate system.
23.5 Numerical Simulations

In this part, a finite element model (FEM) is presented studying the
mechanical behavior of FRFP by using ABAQUS/Standard nonlinear
finite element analysis tool. The geometrical dimensions of the FEM are in
accordance with those from manufacturing.
As shown in Figures 23.9 and 23.10, the FRFP consists of 10 layers, an
inner layer, an outer layer, and eight reinforced layers. It is assumed that
the interfaces between layers are connected closely. Extrusion and parti-
tion commands were used to separate the reinforced layer into eight layers.
The eight-node linear brick reduced integration element C3D8R was used
to mesh the pipe. HDPE was considered isotropic and fiberglass was con-
sidered transverse isotropic; therefore, the orientation of reinforced layers
was assigned according to different layers. A global rectangular system was
defined and X, Y, and Z represents three directions. An orientation dis-
crete field defining a spatially varying orientation was selected, on which
a cylindrical coordinate system was defined, and 1, 2, and 3 denote the
axial, hoop, and radial direction, respectively. For example, one layer of
the model before aligning orientation is shown in Figure 23.11. An addi-
tional rotation of 55° was assigned to this layer along the radial direction
3, which is shown in Figure 23.12.
Since the ends of FRFP were fixed with end-fittings in the test machine,
in order to make the loading conditions similar to the ones from test spec-
imens, one of the ends was totally fixed, and the other end was considered
only rotated along the Z axis.
Z X
X
Z
Figure 23.9 Front view of FRFP.
Z X
Z X
Figure 23.10 Side view of FRFP.
On one end, the cross-section was coupled to a reference point RP2 at

which point the displacement of u1, u2, u3, ur1, ur2, and ur3 were con-
strained to zero. The other end was coupled to another reference point
RP1, which was located at the center of the cross-section. At last, only a
displacement loading was applied in the ur3 direction at the reference
point RP1 to simulate the torsion of FRFP.
3 s s 3 3
2 s
3 s s
2
s s
3 3 2 1
s 2 s
3
s s 2s
3 2
3 2 1
s 2 1
2 3
3 s 2 3 s
1 s 3 2
2 s 3
3 2 1
s 2s 1 3
1 2
3 s 2 3
1 s 1 2
s 3
2 1 1 3
s 2 s
2
13 2
s
3 1
2 1 s
s 2 2 12 s
Y 1 s
13 1 s 2 3
s s 32
Z X 2 1 2
s 2 1
Figure 23.11 Discrete field of one layer before aligning orientation.
3 s s 3 3
2 s
3 s 2
s
2 s
3 s
3 2 s
s 2
2 3
s s s 3
3 2
2 s
2 21 3
31 3 s
s
s 3 1 2
2 s 3
3 2
2 2 s
2 1 s 3
1 s 1 3
3 2
s 1
s 32
2 s s 3
2 1 1
1 2
s 1
2 3 s
1 s 2 2
Y 1 2
1 s3 1
s 1
s s 3
Z X 2 2
2 1
1 s
Figure 23.12 Discrete field of one layer after aligning orientation.
23.6 Results and Discussions

In order to investigate the accuracy and reliability of the results obtained
from the analytical solution and FEM, the torque-torsion angle curve from
these two methods were compared with the curve from experiment. The
theory process was solved by MATLAB. In MATLAB, the total torque
applied on the pipe was 3,000 N·m which was added evenly over 500 steps,
so 6 N·m torque was applied on the pipe every step. While in the FEM,
on one of the reference point, UR3 was rotated 0.4 rad. The deformation
and von Mises stress distribution of the model is shown in Figure 23.13.
The torque-torsion angle relationship of these three methods agrees with
each other well enough in the elastic phase; however, there are still some
differences among these three methods mainly because of the material
properties. The average torsional rigidity of these three methods remains
about 8,333 N·m/rad, shown in Figure 23.14.
S, Mises
(Avg: 75%)
+1.787e+02
+1.642e+02
+1.496e+02
+1.350e+02
+1.204e+02
+1.058e+02
+9.127e+01
+7.669e+01
+6.211e+01
+4.753e+01
+3.295e+01
+1.837e+01
+3.795e+00
Y Standard 6.14-1 Sat Mar 31 14:50:42 GMT+08:00 2018
8: Step Time = 1.000

Primary Var: S, Mises
Figure 23.13 Deformation and von Mises stress distribution.
3500
3000
2500
torque/N·m
2000
1500
1000
specimen 1
specimen 2
500 specimen 3
numerical solution
analytical solution
0
0.0 0.1 0.2 0.3 0.4 0.5 0.6
torsion angle/rad
Figure 23.14 Torque-torsion angle relationship of three methods.


23.7.1 Effect of Winding Angle
The failure torsion angles of experiment specimens during the elastic
period do not exceed 0.4 rad. In order to save calculation time, the tor-
sion angle was only taken to 0.4 rad in this parametric analysis. Firstly, the
winding angle was modified in ABAQUS while the other parameters were
maintained same as the original model. The winding angle changed from
30° to 70° and the corresponding torque-torsion angle curve is shown in
Figure 23.15. The differences between 50° and 40°, between 30° and 60°,
and between 35° and 55° is quite small, so, in the picture, these curves are
overlapped together. It can be seen from this figure that the highest torsion
rigidity appears when the winding angle of fiberglass is 45° and the value is
8,990 N·m/rad, 9.35% higher than 8,149 N·m/rad, at the winding angle of
55°. The lowest torsion rigidity is 4,801 N·m/rad, when the winding angle
of fiberglass is 70°. Therefore, torsion rigidity can be influenced greatly by
the winding angle of fiberglass. By decreasing the winding angle the tor-
sion resistance can be increased.
23.7.2 Effect of Thickness of Reinforced Layers

The single-layer thickness was modified in ABAQUS as 0.5, 0.75, and
1 mm; therefore, the entire thickness of reinforced layer was changed
as 4, 6, and 8 mm, respectively. As shown in Figure 23.16, the thicker
4000
3500
3000
2500
Torque (N·m)
2000 30°
35°
1500 40°
45°
1000 50°
55°
500 60°
65°
70°
0
0.0 0.1 0.2 0.3 0.4 0.5
Torsion Angle (rad)
Figure 23.15 Torque-torsion angel relations under different winding angles.

5500
5000
4500
4000
3500
Torque (N·m)
3000
2500
2000
1500
1000 4mm
500 6mm
8mm
0
0.0 0.1 0.2 0.3 0.4 0.5
Torsion Angle (rad)
Figure 23.16 Torque-torsion angel relations under different reinforced layer thicknesses.
the reinforced layer is, the higher the torsion rigidity becomes. The tor-
sion rigidity is 11,844 N·m/rad when the thickness of reinforced layer
is 8 mm, 57.55% higher than only 5,028 N·m/rad when the thickness is
4 mm. Therefore, torsion rigidity can be influenced greatly by the thick-
ness of reinforced layer, which, in return, can remarkably increase the
torsional resistance.
The radius-thickness ratio was modified in ABAQUS while the other
parameters were maintained the same as the original model. The inner
diameter was set up as 40, 50, and 60 mm, so the radius-thickness ratios of
these three models became 66/13, 76/13, and 86/13, respectively. As seen
in Figure 23.17, the higher the radius-thickness is, the higher the torsion
rigidity becomes. The highest torsion rigidity is 12,598 N·m/rad when
the radius-thickness ratio is 86/13, 61% higher than only 4,890 N·m/rad
when the radius-thickness ratio is 66/13. Therefore, increasing the radius-
thickness would be an effective way to enhance torsion resistance in prac-
tical engineering.
23.8 Conclusions
This chapter presents experimental studies of three 10-layer FRFP on the
torsional experiment machine in Ningbo OPR (Offshore Pipelines and
6000
5000
4000
Torque (N·m)
3000
2000
1000 66/13
76/13
86/13
0
0.0 0.1 0.2 0.3 0.4 0.5
Torsion Angle (rad)
Figure 23.17 Torque-torsion angel relations under different radius-thickness ratios.
Risers Inc.) factory. The torque-torsion angle relations were recorded

during the test.
A 10-layer mathematical model based on 3D classical lamination theory
was established to study the mechanical behavior of FRFP under torsion.
This model can give stresses, strains, and torsion rigidity of FRFP.
Then, a 10-layer FEM with different material orientation assigned to
each layer was established to further study the mechanical behavior of
FRFP during the elastic period. Torque-torsion angle relations obtained
from these three methods turned out to be in good agreement with each
other, which proved that the mathematical model and FEM can be used to
predict the torsion rigidity of FRFP.
In the parametric analysis, changes of the winding angle, thickness of
reinforced layers, and radius-thickness ratio were taken into consider-
ation. Results find that increasing the thickness of reinforced layers and
radius-thickness can obviously improve the torsional rigidity of FRFP, and
the optical winding angle remains at around 45°.
Acknowledgments
The authors would like to thank Shang Hai Fei Zhou Bo Yuan Petroleum
Equipment Company and OPR (Offshore Pipelines and Risers Inc.) for
providing the specimens and laboratory equipment. Also, we appreci-

ate Kevin Conroy who seriously help us to examine the grammar of this
chapter.
References
1. Bahtui A, Bahai H, Alfano G. A finite element analysis for unbonded flex-
ible risers under torsion[J]. Journal of Offshore Mechanics and Arctic
Engineering, 2008, 130(4): 041301.
2. Bahtui A, Bahai H, Alfano G. Numerical and analytical modeling of
unbonded flexible risers[J]. Journal of Offshore Mechanics and Arctic
Engineering, 2009, 131(2): 021401.
3. Merino H E M, Sousa J R M, Magluta C, et al. Numerical and experi-
mental study of a flexible pipe under torsion[C]//Proceedings of the 29th
International Conference on Offshore Mechanics and Arctic Engineering.
OMAE2010-20902. 2010.
4. Liu M S, Liu X W, Li J Y, et al. Numerical Simulation of Flexible Multilayered
Pipe/Riser Under Torsion[J]. Strength of Materials, 2017, 49(1): 180–187.
5. Ren S, Xue H, Tang W. Analytical and numerical models to predict the behav-
ior of unbonded flexible risers under torsion[J]. China Ocean Engineering,
2016, 30(2): 243–256.
6. Bai Y, Lu Y, Cheng P. Theoretical and finite-element study of mechanical
behaviour of central, large-diameter umbilical cables under tension and tor-
sion[J]. Ships and Offshore Structures, 2015, 10(4): 393–403.
7. Knapp R H. Nonlinear analysis of a helically armored cable with nonuniform
mechanical properties in tension and torsion[C]//OCEAN 75 Conference.
IEEE, 1975: 155–164.
8. Xia M, Takayanagi H, Kemmochi K. Analysis of multi-layered filament-
wound composite pipes under internal pressure[J]. Composite Structures,
2001, 53(4): 483–491.
9. Bai Y, Xu F, Cheng P. Investigation on the mechanical properties of the
Reinforced Thermoplastic Pipe (RTP) under internal pressure[C]//The
Twenty-second International Offshore and Polar Engineering Conference.
International Society of Offshore and Polar Engineers, 2012.
10. Kruijer M P, Warnet L L, Akkerman R. Analysis of the mechanical proper-
ties of a reinforced thermoplastic pipe (RTP)[J]. Composites Part A: Applied
Science and Manufacturing, 2005, 36(2): 291–300.
11. Zhao Y, Pang S S. Stress-strain and failure analyses of composite pipe under
torsion[J]. Journal of Pressure Vessel Technology, 1995, 117(3): 273–278.
12. Zhu Y C. Buckling Analysis of Plastic Pipe Reinforced by Winding Steel
Wires under External Pressure[J]. Zhejiang University, 2007.
24
Cross-Section Design of
Fiberglass Reinforced Riser
24.1 Introduction
Compared with traditional risers, fiberglass reinforced risers have the
advantages of light weight, good corrosion resistance, and long service life,
which are widely used in risers. Therefore, it is necessary to propose a sys-
tematic method for designing the cross-section of composite pipes.
In order to meet the requirement of external pressure, thick-walled
cylindrical shell is usually used as composite pipe in engineering. This
chapter is mainly based on fiberglass reinforced riser with thickness and
diameter ratio greater than 0.1.
24.2 Nomenclature
d Diameter of fiberglass
N Total number of fiberglass
i Internal radius of riser
ro External radius of riser
α Angle between the winding direction and the axial
direction
k Coefficient, K = ro/ri
σbg Ultimate strength of fiberglass
σbp Stress of HDPE
pB Burst pressure
n Reduction factor
pdesign Design pressure
pLTHS Long-term hydrostatic pressure
k1 Design coefficient
503
E(k) Elastic Modulus (k=1, 4)

v(k) Poisson’s ratio (k = 1, 4)
EL( k ) Longitudinal elastic modulus in local coordinate system
(k = 2, 3)
(k)
ET Lateral elastic modulus in local coordinate system

(k = 2, 3)
Do External diameter
Eθ Circumferential elastic modulus
Pcr External pressure instability pressure
r Raduis of fiberglass
d Pitch
Rj Fiber glass bonding resin layer radius
L1 Length of fiber on the winding direction with an axial
length of L1
L2 Length of wire on the winding direction with an axial
length of L2
β Undetermined constant
r0 Fiber radius
Lpl Critical length of connector insert sleeve
(k)
GLT Shear modulus (k = 2, 3)
(k)
vTL Poisson’s ratio (k = 2, 3)
Qij( k ) Stiffness coefficient in local coordinate system (k = 2, 3)
Qij( k ) Stiffness coefficient in the whole coordinate system (k = 2, 3)
k Layer number
ri Radius of each layer (i = 0, 1, 2, 3, 4)
T1 The tensile capacity provided by the innermost and the
outermost layers
T2 Tensile strength provided by strengthened layer
Tdesign Design pulling force
T Yield tension
k
5
Design coefficient
t Wall thickness of whole pipe
D m
Mean diameter of tube
μzr Poisson’s ratio in Z-R plane
Ro Outer diameter of pipe
t
1
Inner wall thickness
Rdesign Design minimum bending radius
R crit,b
Minimum bending radius
Cross-Section Design of Fiberglass Reinforced Riser 505
μθz Poisson’s ratio in plane θ-z

μzθ Poisson’s ratio in plane z-θ
Pdesign Design of external pressure instability pressure
τ The maximum anti debonding shear stress that the adhe-
sive resin can achieve is measured by the test.
ε Tensile strain of fiber in the middle part of pipe
Af Cross-sectional area of fiber
Ef Elastic modulus of fibers
e The strain far away from the fiber in the matrix is mea-
sured by the test.
Gm Shear modulus of bonded resin
R0 Equivalent circle radius of binder layer
24.3 Basic Structure of Pipe

24.3.1 Overall Structure
The pipe is composed of a dielectric layer, an inner lining layer, a rein-
forced layer, and an outer protective layer [1]. The enhancement layer
is bonded to the inner and outer layers through an adhesive, and the
structure is shown in Figure 24.1.
Figure 24.1 Schematic diagram of pipeline structure.

24.3.2 Material
1. Dielectric layer
The dielectric layer can be made of PVDF, which has good chemical and tem-
perature tolerance in the temperature range from 0°C (32 F) to 130°C (266 F).
2. Inner lining layer

Ultra-high molecular weight polyethylene (UHMWPE) with a melt mass flow
rate of not more than 0.1g/10 min (190°C, 21.6 kg) is used as the lining layer.
The necessary additives such as antioxidants, anti-ultraviolet, and colorants
can be added to UHMWPE, and all additives should be dispersed evenly.
3. Reinforced layer
The reinforced layer is made of fiberglass, and the tensile strength of the
same batch of fiberglass varies within 5% and should be coated with suit-
able adhesive [2].
4. Outer protective layer

High-density polyethylene (HDPE) of no less than PE100 grade is used
as the outer protective layer. The necessary additives such as antioxidants,
anti-ultraviolet, and colorants can be added to HDPE. All additives should
be dispersed evenly [3].
24.4 Strength Failure Design Criteria

24.4.1 Burst Pressure
The axial burst pressure of fiberglass reinforced pipe is
z Nd 2 (σ bg cos 2 α − σ bp )
p =
B 2
+ σ bp ( K 2 − 1) (24.1)
4ri cosα
The circumferential burst pressure of fiberglass reinforced pipe is
Nd 2 (σ bg sin 2 α − σ bp )
pθB = + σ bp ( K − 1) (24.2)
4ri (ri + ro )cosα
Long-term hydrostatic pressure reduction factor is showm in Table 24.1.

The short-term burst pressure is the minimum value of the circumfer-
ential burst pressure and the axial burst pressure [4]. That is
(
pB = min pBz , pθB ) (24.3)
The ultimate strength σbg of fiberglass and the calculated strength σbp of
polyethylene can be obtained by tensile test of glass fiber and HDPE sam-
ples. Temperature reduction factor
 y = 1 0 ≤ T ≤ 20°C
 (24.4)
 y = −0.005T + 1.1 20 ≤ T ≤ 60°C
Long-term hydrostatic pressure reduction factor:
pLTHS = pB × n (24.5)
Long-term hydrostatic pressure design criteria:
PLTHS
Pdesign ≤ (24.6)
k1
Table 24.1 Long-term hydrostatic pressure

reduction factor for composite pipes at
different temperatures.
Temperature (°C) Reduction factor
20 0.59
40 0.62
60 0.64
24.4.2 Burst Pressure Under Internal Pressure Bending

Moment
The axial burst pressure of fiberglass reinforced pipe is
z Nd 2[(σ bg − 0.1σ st )cos 2 α − σ bp + 0.1σ z 1 ]

p =
B 2
+ (σ bp − 0.1σ z 1 )( K 2 − 1)
4ri cosα
(24.7)
The circumferential burst pressure of fiberglass reinforced pipe is
Nd 2[(σ bg − 0.1σ st )sin 2 α − σ bp ]

pθB = + σ bp ( K −1) (24.8)
4ri (ri + ro )cosα
The short-term burst pressure is the minimum value of the circumfer-

ential burst pressure and the axial burst pressure [5]. That is
(
pB = min pBz , pθB ) (24.9)
The ultimate strength σbg of fiberglass and the calculated strength σbp of
polyethylene can be obtained by tensile test of glass fiber and HDPE sam-
ples [6].
Design criteria for burst pressure under combined load of internal pres-
sure and bending moment
PB
Pdesign ≤ (24.10)
k3
24.4.3 Yield Tension

In the process of laying fiberglass pipes, there is a certain requirement for
the maximum tensile force that fiberglass pipes can bear; so, it is necessary
to check the yield tension of fiberglass pipes [7].
When the strengthen layer is two layers, the tensile load is
T = T1 + T2 (24.11)
( )
T1 = π X 4 r42 − r32 + r12 − r02 ε z (24.12)
 
 ( 2
)
8 X1 2r23 − r13 − r33 ( X 2 X 3 − X1 )(1 + X 2 ) 
 (
2 2 2 2
) 4
T2 = 2π  9 X 4 r4 − r3 + r1 − r0 + 2 X5 (1 + X 2 ) r3 − r1 (
4
) 
εz
 
1 2
(
 + 2 ( X 6 − X 2 X 7 ) r3 − r1
2
) 
  (24.13)
X1 = m3nQ11 + (mn3 − m3n)Q12 − mn3Q22 − 2mn(m2 − n 2 )Q33 (24.14)
X2 = ν(1) (24.15)
X4 = E(1) (24.16)
X 3 = mn3Q11 + (m3n − mn3 )Q12 − mn3Q22 + 2mn(m 2 − n 2 )Q33 (24.17)
X5 = m 2n 2Q11 − 2m 2n 2Q12 + m2n 2Q22 + (m 2 − n 2 )Q33 (24.18)
− − − −
X 6 = m 4 Q11 + 2m 2n 2 Q12 + n 4 Q 22 + 4m 2n 2 Q 33 (24.19)
X 7 = m 2n 2Q11 + (m 4 + n 4 )Q12 + m 2n 2Q22 − 4m 2n 2Q33 (24.20)
m = cos φ, n = sin φ (24.21)
EL ET
Q11 = , Q22 = (24.22)
1 − ν LT νTL 1 − ν LT νTL
ELνTL
Q12 = , Q33 = GLT (24.23)
1 − ν LT νTL
When the strengthened layer is four layers, the tensile load is
T = T1 + 2T2 (24.24)
Yield tensile design criteria
T
Tdesign ≤ (24.25)
k5
24.5 Failure Criteria for Instability Design

24.5.1 Minimum Bending Radius
The minimum bending radius of the fiberglass tube should be checked
because of the varying degrees of bending during the coil transportation
and laying process.
The minimum bending radius is
Ro × Dm × µzr 1
Rcrit ,b = ×
0.28 × t ε (24.26)
ε = −0.000612 × N + 0.00109 × α + 0.01206 × d

+0.0386 × t1 + 1.335 (24.27)
Minimum bending radius design criteria
Rcrit ,b
Rdesign ≤
k2 (24.28)
24.5.2 External Pressure Instability Pressure
3
2 Eθ  t 
Pcr = (24.29)
1 − µθ z µzθ  Do 
Design criterion of external pressure instability pressure
Pcr
Pdesign ≤
k4 (24.30)
24.6 Design Criteria for Leakage Failure
ld rd csch β l2 (σ f max − σ 0 ) + σ 0 coth β l2

L pl = = ×
2π R j 4π R jτ coth β l1 + coth β l2
(24.31)
L1 L
Among which: l1 = 2π R j , l2 = 2π R j 2 , σ0 = Ef Af e, σfmax = Ef εT,
d d
H Gm 2π
β= =
E f Af Ef R0
A f ln
r0
References
1. Syed Naqvi, Karam Mahmoud, Ehab El-Salakawy. Effect of Axial Load and
Steel Fibers on the Seismic Behavior of Lap-Spliced Glass Fiber Reinforced
Polymer-Reinforced Concrete Rectangular Columns[J]. Engineering
Structures, 2017,134.
2. Theeranan Thummakul, Dimitri Gidaspow, Pornpote Piumsomboon,
Benjapon Chalermsinsuwan. CFD simulation of CO2 sorption on K2CO
3 solid sorbent in novel high flux circulating-turbulent fluidized bed riser:
Parametric statistical experimental design study[J]. Applied Energy, 2017,190.
3. K.D. Do. Boundary control design for extensible marine risers in three
dimensional space[J]. Journal of Sound and Vibration,2017,388.
4. Xiaoyuan Wang, Yalu Zhu, Minzhao Zhu, Yuezhao Zhu, Hongtu Fan,Yinfeng
Wang. Thermal analysis and optimization of an ice and snow melting system
using geothermy by super-long flexible heat pipes[J]. Applied Thermal
Engineering, 2017,112.
5. Liang Zhen, Pizhong Qiao, Junbin Zhong, Qingyuan Chen, Jin-Jian Chen,
Jian-Hua Wang. Design of steel pipe-jacking based on buckling analysis
by finite strip method[J]. Engineering Structures, 2017,132.
6. Jian Qu, Xiaojun Li, Yingying Cui, Qian Wang. Design and experimental
study on a hybrid flexible oscillating heat pipe[J]. International Journal of
Heat and Mass Transfer, 2017,107.
7. Stephen H. Taylor, Suresh V. Garimella. Design of electrode arrays for 3D
capacitance tomography in a planar domain[J]. International Journal of Heat
and Mass Transfer, 2017,106.
25
Fatigue Life Assessment of Fiberglass
Reinforced Flexible Pipes
25.1 Introduction
Fiberglass reinforced flexible pipe (FGRFP) belongs to a new type of com-
posite pipe. It is mainly composed of three layers, which are inner PE layer,
fiberglass reinforced layer, and outer PE layer. Among them, the fiberglass
reinforced layer is formed by winding the fiberglass reinforced belt around
the lining layer at a certain angle. Usually, the fiberglass reinforced layer is
composed of multi-layer structure. The three layers of structure are closely
bonded together to form the pipe body of the FGRFP. Because FGRFP is
made of non-metallic materials and PE, it has many advantages. Such as, it
has strong deformation ability, small bending radius, corrosion resistance,
and fatigue resistance.
For a long time, researchers have done a lot of research on the fatigue
analysis of flexible pipes. Sævik S et al. [1] have studied the fatigue test,
stress analysis and fatigue modeling of two kinds of non-bonded 4-inch
flexible pipe specimens. The correlation between the experimental failure
mode and the condition of fatigue model prediction has been theoretically
studied, and a good correlation between the experimental and theoretical
behavior has been observed.
Tan ZM et al. [2] studied the fatigue evaluation of flexible risers in bimodal
and proposed a more accurate irregular wave method for evaluating the
fatigue of flexible pipes. Through the OrcaFlex and its tensile line stress model,
the model produces the bending hysteresis behavior and stress results in the
dynamic excitation process, and the fatigue damage is evaluated by the rain
flow counting method.
Kraincanic I et al. [3] considered the influence of nonlinear sliding on the
bending behavior of marine flexible pipelines based on the principle of virtual
work and Coulomb friction. The mechanism of relative sliding of the unit in
the whole bending process of marine flexible pipelines is studied in detail.
513
Yue QJ et al. [4] proposed an analytical expression of the tensile stiffness

of offshore flexible pipelines. At the same time, the effect of radial stiffness on
the tensile stiffness of offshore flexible pipelines under tension was explored.
Bahtui A et al. [5, 6] used ABAQUS software to carry out finite ele-
ment analysis of offshore flexible pipelines under internal pressure, exter-
nal pressure, torsion, tension, and bending loads.
Tan ZM et al. [7] used ANSYS software to study the high-order geomet-
ric nonlinearity of helical layers in flexible offshore pipelines. Considering
the cross-section characteristics of steel strips, a more accurate energy
analysis model for axial stretching ratio was proposed.
Jiang KE et al. [8] mainly studied a theoretical method for calculat-
ing the fatigue life of flexible pipeline tension armor layer. In this method,
the local model of flexible pipeline is adopted, and the force and moment
obtained from the global analysis in time domain are transformed into the
stress of the spiral reinforcement in the stretched armor layer. It is found
that the fatigue damage of the stretched armor in the suspension point is
most likely to occur.
Ruan WD et al. [9] take into account the effects of stiffener constraints
and bending hysteresis characteristics in the riser overall model on the
basis of the bar system theory. They found that the bending stiffener can
optimize the curvature distribution and amplitude of the top region; the
apex angle can change the curvature distribution; and the bending hystere-
sis characteristic has an important influence on the curvature.
Ebrahimi A et al. [10] studied the mechanical response of unbonded
flexible pipes under axisymmetric loads. By establishing a continuous
finite element model, it was found that the ideal numerical model based
on analysis and structure may be incomplete. The importance of predict-
ing bifurcation response and contact mechanics between layers by using
implicit solver was emphasized.
Djihad Rial et al. [11] have carried out a series of fatigue tests on hydro-
forming metal bellows. Among them, the amplitude of mechanical fatigue
load is constant, besides a constant internal pressure similar to working
conditions, a three-dimensional stress- strain state has been generated, and
the life of the structure has been extracted from the test results.
Hyunmin Bae et al. [12] studied the slow crack growth behavior of poly-
ethylene tubular resins by static fatigue (stress fracture) and cyclic fatigue
tests. They found that in some cases, in cyclic fatigue tests, resins with higher
slow crack resistance rank were lower than resins with another static fatigue.
Yoo DH et al. [13] evaluate the overall length of flexible pipes used in
offshore power riser applications. Under the action of axial tension and
compression load, they adopt the practical and stable methods to evaluate
Life Assessment of Fiberglass Flexible Pipes 515
the ultimate strength of eight-layer and five-layer three-dimensional finite

element models, respectively. In addition, they also studied the failure
mechanism and interlayer interaction under incremental loading.
Skeie G et al. [14] established the mathematical analysis model of pipe-
line, and the slip between helical layers of pipeline was considered in the
fatigue analysis of pipeline.
Zhu LF et al. [15] introduced two buckling mechanisms, radial bird-
cage buckling, and lateral buckling. For each failure mode, an analytical
buckling prediction theory is proposed and the driving parameters are dis-
cussed. At the same time, they have developed an alternative air drill to
simulate inclination tests in controlled laboratory environments.
It can be seen from the previous discussion that there are few studies
on fatigue damage of FGRFP. Therefore, the fatigue damage of FGRFP
should be analyzed for a better understanding of the capability of FGRFP.
Firstly, the tension and bending moment time histories curve of FGRFP is
obtained by using global analysis method. Then, the stress time histories
are obtained by using the finite element method. According to the stress
time histories and SN curve, the fatigue life of FGRFP can be evaluated.
The method presented in this chapter can provide a reference for fatigue
analysis of FGRFPs. In this chapter, we do not consider the impact of mate-
rial creep on the analysis.
25.2 Global Analysis

In order to analyze the fatigue life of FGRFP, it is first necessary to make
a global analysis of the FGRFP used for riser under marine conditions. In
the global analysis of the model, there are two methods: frequency domain
method and time domain method. In this chapter, the time domain method
is used for analysis because it performs well in dealing with non-linear prob-
lems. In the global analysis of FGRFP, it is necessary to determine the mate-
rial parameters, size parameters, riser type, end fitting type, and marine
environmental parameters. In this example, the riser type is selected as lazy
wave type, the top is connected with the floating production platform, and
the bottom is anchored on the seabed. The geometric parameters and mate-
rial properties of FGRFP are shown in Tables 25.1 and 25.2, and the marine
environmental parameters and working conditions are shown in Table 25.3.
Through the global analysis, we can get the tension time histories and bend-
ing moment time histories curve of the FGRFP.
The global analysis of the model is divided into two steps. The first
step is the static calculation, by which the equilibrium position and initial
Table 25.1 Size parameters of EGRFP.

Inner Outer
Structural diameter Thickness diameter
layer Materials (mm) (mm) (mm)
Inner layer HDPE 50 4 58
Reinforced Fiberglass 58 6 70
layer Reinforced
Outer layer HDPE 70 3 76
Table 25.2 Material parameters of EGRFP.

Elasticity modulus Yield strength
Material Position (MPa) (MPa)
HDPE Inner layer and 850 23
Outer layer
Fiberglass Reinforced layer 33,000
Table 25.3 Environmental parameters.

Depth (m) Wind (m/s) Wave Hs (m) Wave Tz (s) Current (m/s)
2,100 17.27 1.74 4.07 0.3
condition of the pipeline in the sea can be obtained. In addition, the max-
imum position of the internal force on the riser section can be obtained
by the static calculation. The second step is dynamic calculation, which
imposes a 10-year-once ocean current condition on the pipeline. The
implicit integral method is used to solve the deformation and stress of the
flexible pipe. Finally, the maximum tension and curvature of the force in
the FGRFP changing withtime is obtained. From Figures 25.1 and 25.2, it
can be seen that the tension and bending moment of flexible pipes fluctu-
ate within a certain range under the combined action of random waves,
currents, wind loads, and FPSO.
85
80
Tension (kN)
75
70
0 100 200 300
Time (s)
Figure 25.1 FGRFP tension time histories.
17.5
Bend Moment (kN·m)
17.0
16.5
0 100 200 300

Time (s)
Figure 25.2 FGRFP bending moment time histories.
25.3 Rain Flow Method

The rain flow counting method was proposed in the 1950s by two British
engineers, M. Matsuishi and T. Endo. The main function of this counting
method is to simplify the measured load history into several load cycles
for fatigue life estimation and fatigue test load spectrum. It is based on the
two-parameter method and considers two variables of dynamic strength
(magnitude) and static strength (mean), which are intrinsic to the fatigue
load itself. The rain flow counting method is mainly used in the engineer-
ing field, and is widely used in the calculation of fatigue life.
The basic counting rules for rain flow counting are as follows:
(1) The rain flow sequentially flows from the inside of the peak
position of the load time history down the slope.
(2) The rain flow starts from a certain peak point and stops
when it encounters a peak larger than its initial peak.
(3) When the rainwater encounters the rain flowing down, it
must stop flowing.
(4) Take out all the full cycles and note the amplitude of each cycle.
(5) The time history of the divergent convergence load remain-
ing after the first stage count is equivalent to a convergent
divergence load time history, and the second stage rain flow
count is performed. The total number of count cycles is
equal to the sum of the count cycles of the two count phases.
By using the rain flow counting method, the tension time history curve
and the bending moment time history curve obtained by the global analy-
sis are arranged, and the histogram of the tension and bending moment
frequency distribution such as Figures 25.3 and 25.4 can be obtained.
12
10
Number of Cycles
10
8
5 6
4
0
79 78 2
77 76 10
5 0
75 0
Mean Tension(kN) Tension Amplitude(kN)
Figure 25.3 Rain flow histogram of tension.
8
Number of Cycles
6
5
4
0 2
17.1
17 1
16.90.516.8 0
0
Bending Moment Amplitude (kN·m)
Mean Bending Moment (kN·m)
Figure 25.4 Rain flow histogram of bending moment.

25.4 Local Analysis

After the global analysis, the internal force time histories of the FGRFP is
calculated and analyzed locally, and the strain time histories of the FGRFP
is obtained. Then, we chose the most disadvantageous position from the
global analysis. In the local analysis, the explicit dynamic analysis of
commercial software ABAQUS is used. The local finite element model of
the FGRFP is established, and then, the tension time histories and bending
moment time histories are loaded onto the finite element model to solve
the problem.
25.5 Modeling
In this chapter, the fiberglass reinforced flexible composite pipe is stud-
ied, so the composite material modeling method is adopted. At present,
there are three kinds of commonly used ABAQUS composite modeling
methods:
1. Modeling according to the type of laminate. The shell ele-

ment model is adopted in this method, which is suitable
for the dynamic analysis of the macroscopic force analysis
of the structure. There are two advantages of this modeling
method: one is that the modeling is easy to operate, and the
other is that the calculation speed is fast.
2. Composite layup method modeling. This method of solid
element modeling is suitable for static analysis in macro-
mechanics analysis, and the solid element model is complex
in calculation and takes a long time.
3. Modeling of fiber and matrix separately. This modeling
method is suitable for meso-mechanical analysis and is mainly
used to study the meso-mechanical properties between fibers
and matrix.
Because the dynamic analysis is used in the local analysis of FGRFP, the
shell element is used to establish the finite element model, and the consti-
tutive relationship between fiberglass reinforced layer and PE is established
by using the material properties of ABAQUS embedded monolayer. The
property of ABAQUS monolayer is that fiberglass reinforced layer is equiv-
alent to orthotropic macroscopic homogeneous material. This equivalence
Table 25.4 Specific Parameters of the layer.

No. Layer Thickness (mm) Angle (⁰)
1 Inner PE Layer4 4 0
2 Fiberglass Reinforced Layer 1 0.75 55
3 Fiberglass Reinforced Layer 2 0.75 −55
10 Outer PE Layer 3 0
requires comprehensive consideration of the material properties of matrix

materials and fiberglass, and calculation of the material properties of
single-layer plates by using the theory of single-layer plates. The material
parameters and dimension parameters of FGRFP are the same as those of
global analysis. Tables 25.1 and 25.2 data are used.
The length of the model is 500 mm, which is about 6.5 times the outer
diameter. According to Saint Venant’s principle, the influence of bound-
ary factors can be neglected. As the fiberglass reinforced layer is usually
formed by multi-layer structure winding at both positive and negative
angles, the model uses eight layers of fiberglass winding to form the rein-
forced layer. The specific parameters of the layer are shown in Table 25.4.
25.6 Result Discussion

The purpose of this analysis is to investigate the stress changes of FGRFP
layers under the action of internal force time histories and to evaluate the
fatigue life of FGRFP layers. Therefore, 0–300s tension time histories and
bending moment time histories are selected for local analysis. The results of
simulation calculation by ABAQUS are shown from Figures 25.5 and 25.10.
The stress nephogram is shown in Figure 25.5. From the stress nepho-
gram, we can see that the stress in the middle part of the FGRFP is more
Z X
Figure 25.5 Stress nephogram.
uniform, and the stress concentration in the end part is limited by the
constraint conditions. Due to the long length of the FGRFP, the influence
of the end is not taken into account according to the principle of Saint-
Venant. Therefore, we take the element in the middle of the FGRFP as the
reference element for each layer of stress and strain.
From Figure 25.6 we can see that the stress time histories of the inner
and outer layers of PE show that the trend of the stress changes with time
is very similar. The stress of the inner and outer layers of PE fluctuates
around 3.8 MPa, and the maximum fluctuation amplitude is 0.27 MPa.
Inner PE Layer
4.2 Outer PE Layer
4.0
Stress (MPa)
3.8
3.6
3.4
3.2
50 100 150 200 250 300
Time (s)
Figure 25.6 Stress time histories of inner PE layer and outer PE layer.
This Figure 25.7 shows the stress time histories of all structural layers.
It can be seen that the stress change trend of the eight fiberglass reinforced
layers are similar. The stress of the fiberglass reinforced layers fluctuates
around 95.88 MPa, and the maximum fluctuation amplitude is 8.08 MPa.
The mean stress of the inner and outer PE layers is 3.8 MPa, and the stress
of the fiberglass reinforced layers is far greater than that of the PE layers.
In addition, the stress change amplitude of the inner and outer PE layers
is smaller than that of the fiberglass layers. It can be concluded that the
fiberglass reinforced layers are the main stress layers and the key part of
the fatigue failure of the FGRFP under the marine environment. Therefore,
the strength of the flexible pipe can be greatly enhanced by increasing the
strength of the fiberglass reinforced layers.
The Figures 25.8 and 25.9 show that the stress produced by the rein-
forced layers laid at the same angle are very similar, and the thickness has
little influence on it, compared with the reinforced layer laid at 55° of four
layers and the reinforced layer laid at −55° of four layers.
Because the stress time histories are almost the same when the laying
angle is the same, in order to compare and analyze the difference between
the stress time histories of the fiberglass reinforced layer with 55° and that
of the fiberglass reinforced layer with −55°, only two layers of the reinforced
layer with different angles are taken as the object of analysis in Figure 25.10.
By comparing the two curves, it can be seen that the stresses produced by
the two layers are different when they are subjected to the same tension
and bending moment. The stresses produced by the reinforced layer laid at
55° are slightly larger than those produced by the reinforced layer at −55°.
120
100
80
Stress (MPa)
Inner PE Layer
Fiber Glass Reinforced 1
60 Fiber Glass Reinforced 3
40 Fiber Glass Reinforced 6
20 Outer PE Layer
0
0 50 100 150 200 250 300
Time (s)
Figure 25.7 Stress time histories of all structural layers.

55° Layer 1
55° Layer 3
110 55° Layer 5
55° Layer 7
100
Stress (MPa)
90
0 50 100 150 200 250 300

Time (s)
Figure 25.8 Stress of 55° layers.
–55° Layer 2
–55° Layer 4
105 –55° Layer 6
–55° Layer 8
100
Stress (MPa)
95
90
85
0 50 100 150 200 250 300
Time (s)
Figure 25.9 Stress of −55° layers.
The mean stress of the fiberglass reinforced layer with 55° of paving angle
is 96.6 MPa and the change amplitude is 6.88 MPa. The mean stress of the
fiberglass reinforced layer with −55° of paving angle is 95.37 MPa and the
change amplitude is 6.81 MPa. The mean difference of fiberglass reinforced
55° Layer 1
104 –55° Layer 2
100
Stress (MPa)
96
92
88
0 50 100 150 200 250 300

Time (s)
Figure 25.10 Comparison of 55° layer and −55° layer (0.75 mm).
layer between 55° and −55° laying angle is 1.22 MPa, and the variation
amplitude difference is 0.07 MPa.
25.7 Sensitivity Analysis

In the previous analysis, we set the thickness of each fiberglass reinforced
layer to 0.75 mm. In order to find out the law of the thickness and stress
of the fiberglass reinforced layer, the thickness of the fiberglass reinforced
layer is set to 0.5 and 0.25 mm, respectively, and compared with the stress
produced by the fiberglass reinforced layer of 0.75 mm, the following
results are obtained.
The three lines in Figure 25.11 are the stress produced by PE layer when
the thickness of fiberglass reinforced layer is 0.25, 0.5, and 0.75 mm, respec-
tively. When the thickness of fiberglass reinforced layer is 0.25 mm, the
stress of PE layer fluctuates around 10.4 MPa, and the maximum amplitude
is 0.96 MPa. When the thickness of fiberglass reinforced layer is 0.5 mm, the
stress of PE layer fluctuates around 5.55 MPa, and the maximum amplitude
is 0.43 MPa. When the thickness of fiberglass reinforced layer is 0.75 mm,
the stress of PE layer fluctuates around 3.8 MPa, and the maximum ampli-
tude is 0.27 MPa. It can be seen that the thickness of fiberglass reinforced
layer has a greater impact on the stress of PE layer. The thinner the fiberglass
reinforced layer, the greater the stress of PE layer.
Inner PE Layer (0.75mm)

12 Inner PE Layer (0.5mm)
Inner PE Layer (0.25mm)
10
8
Stress (MPa)
0
0 100 200 300
Time (s)
Figure 25.11 Stress comparison of PE layer under different thickness of fiberglass

reinforced layers.
Figure 25.12 shows the stress time histories produced when the thick-
ness of reinforced layer is 0.5 mm and the laying angle is 55° and −55°. It
can be seen that the fluctuation trend of the two is similar. Among them,
the stress produced by 55° fiberglass reinforced layer fluctuates around
141.1 MPa, the maximum amplitude is 10.37 MPa, and the stress produced
55° Layer 1(0.5mm)

155 -55° Layer 2(0.5mm)
150
145
Stress (MPa)
140
135
130
125
0 50 100 150 200 250 300
Time (s)
by −55° fiberglass reinforced layer fluctuates around 139.5 MPa, the maxi-
mum amplitude is 11.41 MPa.
Figure 25.13 shows the stress time histories of fiberglass reinforced layer
with thickness of 0.25 mm and laying angle of 55° and −55°, respectively.
Similarly, the fluctuation trend of both is similar. Among them, the stress
produced by 55° fiberglass reinforced layer fluctuates around 263.2 MPa,
the maximum amplitude is 23.11 MPa, and the stress produced by −55°
55° Layer 1(0.25mm)

290 −55° Layer 2(0.25mm)
280
Stress (MPa)
270
260
250
240
0 50 100 150 200 250 300
Time (s)
55° Layer 1 (0.75mm)

55° Layer 1 (0.5mm)
300 55° Layer 1 (0.25mm)
250
Stress (MPa)
200
150
100
50
0 100 200 300
Time (s)
Figure 25.14 Stress contrast of reinforcement layer under different thicknesses.

fiberglass reinforced layer fluctuates around 261.3 MPa, the maximum

amplitude is 24.19 MPa.
The stress in the fiberglass reinforced layer with the thickness of 0.25,
0.5, and 0.75 mm laid at 55° is shown in Figure 25.14. It can be seen that
the thickness of the fiberglass reinforced layer has a greater impact on the
stress. The thinner the fiberglass reinforced layer is, the greater the stress is,
and the easier the fatigue failure will happen.
25.8 Fatigue Life Assessment

Different from metal-based flexible pipes, the fatigue life of composite flex-
ible pipes has its own complex characteristics: (1) the fatigue life charac-
teristics of the same material vary greatly in different directions and laying
modes; (2) the dispersion of fatigue strength or fatigue life is so great that it
is difficult to draw S-N curves in some cases; (3) some composite materials
have no obvious fatigue limit. In order to obtain S-N curves of composites,
many researchers have proposed many models from different angles, such
as theory, experiment and their combination. The model adopted in this
chapter is an S-N curve model with strong fitting ability [16].
   lg( N + 1)  a  
S = 1 + m  exp  −    − 1
   b  
In the formula, S = σmax/σult, σmax is the tensile strength. The physical mean-
ing of the other three parameters is also very clear: parameter a is the slope
of the curve, which is related to the static strength of the material and its
structure, reflecting the fatigue resistance of the material itself; parameter B
is similar to the characteristic life, which more reflects the internal damage
extension of the material under the external load; when n →∞, S → 1 − m, so
parameter m reflects the fatigue limit of the material.
For the FGRFP which is selected in this chapter, m = 1, a = 2.1, b = 4.7.
The S-N curve is shown in Figure 25.15.
The tensile strength of glass fiber is 798 MPa. In this chapter, when the
thickness of glass fiber reinforced layer is 0.75 mm, the maximum stress is
101 MPa. Therefore,
S = σmax/σult =101/798 = 0.13

0.9
0.8
0.7
0.6
0.5
S
0.4
0.3
0.2
0.1
0
0 2 4 6 8 10
Ig (N)
Figure 25.15 S-N Curve of FGRFP.
Based on the rain flow method and S-N curve, we can obtain that N =
3,000,000. Because the single damage is equal to 1 N, so the single damage
rate of FGRFP in this marine environment is 3.33 × 10−7.
25.9 Conclusion
Under the influence of marine environment, the stress produced by fiber-
glass reinforced layer in FGPFP is obviously greater than that produced by
PE layer, both of which fluctuate within a certain range. Compared with
the PE layer, the stress fluctuation of fiberglass reinforced layer is larger.
When the reinforced layer is laid at the same angle, the stress changes of
each layer are very similar. The stress of the reinforced layer laid at the neg-
ative angle has little change compared with the reinforced layer laid at the
positive angle. Fiberglass reinforced layer is the main force component of
fiberglass pipe, and it is also the key part of fatigue failure of fiberglass pipe.
The thickness of fiberglass reinforced layer has great influence on the
stress of PE layer and fiberglass reinforced layer. Therefore, the strength of
flexible pipe can be greatly improved by increasing the strength of fiber-
glass reinforced layer and choosing the appropriate thickness of fiberglass
reinforced layer.
After global analysis and local analysis, the S-N curve of FGRFP is drawn,
and the fatigue life is found out according to the calculation results and rain
flow method. It is found that under this condition, FGRFP can withstand
3,000,000 cycles of loading with a single damage rate of 3.33 × 10−7.
References
1. Sævik, Svein. “Theoretical and experimental studies of stresses in flexible
pipes.” Computers and Structures 89.23-24(2011):2273-2291.
2. Zhimin Tan, Yucheng Hou, John Zhang, Terry Sheldrake. “Irregular Wave
Simulation and Fatigue Damage Evaluation of a Flexible Riser Subjected to
Bi- Modal Sea States.” 31st International Conference on Ocean, Offshore and
Arctic Engineering 2012.
3. Kraincanic, I, and E. Kebadze. “Slip initiation and progression in helical
armouring layers of unbonded flexible pipes and its effect on pipe bend-
ing behaviour.” The Journal of Strain Analysis for Engineering Design
36.3(2001):265-275.
4. Qianjin Yue, et al. “Tension behavior prediction of flexible pipelines in shal-
low water.” Ocean Engineering 58(2013):201-207.
5. Bahtui, A., H. Bahai, and G. Alfano. “A Finite Element Analysis for Unbonded
Flexible Risers Under Torsion.” Journal of Offshore Mechanics & Arctic
Engineering 130.4(2007):169-173.
6. Bahtui, A., H. Bahai, and G. Alfano. “Numerical and Analytical Modeling
of Unbonded Flexible Risers.” Journal of Offshore Mechanics and Arctic
Engineering 131.2(2009): 021401.
7. Zhimin Tan, Michael Case, Terry Sheldrake. “Higher Order Effects on
Bending of Helical Armor Wire Inside an Unbonded Flexible Pipe “ 24th
International Conference on Offshore Mechanics and Arctic Engineering
2005: Volume 3:447-455.
8. Kaien Jiang, Yutian Lu, Yong Bai. “A Theoretical Method to Estimate
the Fatigue Life of Tensile Armors of Flexible Pipes” 37th International
Conference on Ocean, Offshore and Arctic Engineering 2018: Volume 5.
9. Weidong Ruan, Yong Bai, and Shai Yuan. “Dynamic analysis of unbonded
flexible pipe with bend stiffener constraint and bending hysteretic behavior.”
Ocean Engineering 130(2017):583-596.
10. Ebrahimi, Alireza, S. Kenny, and A. Hussein. “Finite Element Investigation
on the Tensile Armour Wire Response of Flexible Pipe for Axisymmetric
Loading Conditions Using an Implicit Solver.” 37th Journal of Offshore
Mechanics and Arctic Engineering (2018).
11. Rial, Djihad, et al. “Multiaxial fatigue analysis of a metal flexible pipe.”
Materials & Design 54(2014):796-804.
12. Bae, Hyunmin, et al. “Abnormality in using cyclic fatigue for ranking static
fatigue induced slow crack growth behavior of polyethylene pipe grade res-
ins.” Polymer Testing 55(2016):101-108.
13. Yoo, Dong Hyun, B. S. Jang, and K. H. Yim. “Nonlinear finite element analysis
of failure modes and ultimate strength of flexible pipes.” Marine Structures
54(2017):50-72.
14. Nils Sodahl, et al. “Efficient Fatigue Analysis of Helix Elements in Umbilicals
and Flexible Risers.” 29th International Conference on Ocean American
Society of Mechanical Engineers, 2010.
15. Linfa Zhu, Zhimin Tan, Victor Pinheiro Pupo Nogueira, Jian Liu and Judimar
Clevelario. “Prediction and Qualification of Radial Birdcage and Lateral
Buckling of Flexible Pipes in Deepwater Applications.” 34th International
Conference on Ocean, Offshore and Arctic Engineering 2015.
16. Fuqiang Wu, WeiXing Yao. “A fatigue damage model of composite materials.”
International Journal of Fatigue 32(2010):134-138.
Part 4
ANCILLARY EQUIPMENTS FOR
FLEXIBLES AND UMBILICALS
26
Typical Connector Design for Risers
26.1 Introduction
In a connector, to enable contacts to other structures, all of the flexi-
ble pipe layers must be terminated. The end-fitting must have the same
structural integrity as the flexible pipe itself. As mentioned above, all
of the flexible pipe layers must be terminated so that they are capable
of transmitting the pipe induced loads such as bending moments. The
fluid barrier must be terminated in a way, which ensures no leakage
during the lifetime when considering relevant factors such as pressure,
temperature, and transported medium. Particularly, connectors used on
floating systems are subjected to large varying loads, which may cause
fatigue damage with leakage of explosive and poisonous gases and flu-
ids as a probable result. Due to the sealing requirements involved, the
connector sealing mechanism should ensure that the combined strain
induced by the in-service end loads does not result in the failure of the
sheath during the service life [1]. In this article, we have demonstrated
how the seal system used in the termination of the rough bore pipes
and smooth bore pipes have been developed. It needs to be considered
that in rough bore pipes, the carcass is the support of gaskets and also
contact pressures are carried by it. It was found that only the pipeline
size and the yield stress of carcass material effects should be considered;
thus, we can decrease the carcass radius. It means that changing the car-
cass profile has no effect in the sealing system function. Meanwhile, if
the carcass is fixed against axial movements, we can ignore the inner
liner holder. Smooth bore pipes have no carcass, so we have to use an
inner liner expander. The inner liner expander diameter is the same
as the pipe size and is a tube stretch which is pressed into the inner
liner. Therefore, the carcass was exposed to hoop stress and inner liner
expander was exposed to hoop and bending stresses.
533
This chapter has used the principle of the TechnipFMC end-fitting

design according to the standard of American Petroleum Institute specifi-
cation 17J [2], to assess the carcass roll in the seal system of the connector.
26.2 Carcass
The carcass is designed to support the inner liner of the riser against
external pressures. The external pressure can subject the inner liner in
two ways. If there is gas diffusion through the inner liner, a pressure is
built-up in the annulus between the two polymeric layers. The build-up
of this pressure usually takes more than a few days. As long as the pres-
sure inside the riser is maintained the inner liner is in mechanical equi-
librium. However, if the internal pressure is dropped quickly from high
to low, the gas is unable to diffuse back through the inner liner immedi-
ately. Accordingly, the pressure on the outside of the inner liner is much
higher than on the inside. The inner liner itself is not capable of carrying
any specific external pressure and will consequently collapse if it is not
supported on the inside. The carcass is installed to provide the inner liner
with this support. Another way the inner liner is subjected to external
pressure is if the outer sheath is damaged, and seawater enters the annu-
lus. If the pipe is unprescribed, the inner liner is subjected to an external
pressure equal to the water pressure, so the carcass is necessary to pre-
vent the collapse of the inner liner. The carcass and other components of
a riser are shown in Figure 26.1.
Nevertheless, the carcass has a significant role in the connection part
of the riser as well. The carcass role in the sealing system of a connector is
undeniable. In the following, this issue will be discussed.
Carcass
Liner
bonded to
composite
Pressure
armor
Metallic
tensile armor
Insulation
Sheath
Figure 26.1 Typical flexible riser structure.

Typical Connector Design for Risers 535
26.3 Typical Connector

The connector which is assessed in this chapter is the end-fitting type. The
French Coflexip1 is one of the three worldwide unbounded flexible pipe
manufactures (besides the FMC, Wellstream of UK2, and Danish NKT
Flexibles3 are in the market). Technically, end-fitting is the most compli-
cated part of the riser design. In the end-fitting, under consideration of
the operational profile of the pipe within its specified lifetime, the pipe
layers are terminated. The end-fitting has been designed to transfer loads
from the pipe to the connecting system unit and also to ensure sealing
of the pipe fluid. Particularly, end-fitting used on floating systems is sub-
jected to large varying loads, which cause fatigue damage with leakage of
toxic and explosive fluids and gases as a possible outcome. It is extremely
important to the security of the platform that the end-fitting is designed
and installed appropriately. Several end-fitting were failed over the recent
years; therefore, the Health and Safety Authorities pay much more atten-
tion to the terminating flexible pipes subject.
The connector has different components based on the manufacturer.
The components of the FMC end-fitting are described in Figure 26.2
(details are as below).
Electrical insulation between the carcass, the carcass holder and the steel
in the connector is vital; therefore, the seal ring is mounted. Since the car-
cass and the carcass holder are made of stainless steel, while the other steel
components are made of carbon steel, this insulation is necessary. The two
steel types have different electrical potentials, so if they are in direct con-
tact, galvanic corrosion happen [3]. Thus, the seal ring is installed to avoid
this corrosion. The seal ring retainer screw is a component used in the seal
system. It is screwed to fix the seal ring. Pin mandrel profile is the structure
connecting of the end-fitting. Mandrel-type construction provides stable
engagement before energization. The piston is used in order to align the
mandrel. The lock port is the interface to the connecting structure. The
purpose of lock and primary lock ports is to transfer global loads from
the pipe to connecting structure. Connecting rod houses the ring joint gas-
kets which seal the pipe contents in the connector. Cam ring is installed as
a barrier to the pipe contents [4]. Locking dog distributes the load evenly
throughout the body and mandrel of the connector. Because of the possi-
ble gas penetration through the inner liner of the pipe, a pressure in the
1
https://www.technipfmc.com/
2
https://www.bhge.com/
3
https://www.nkt.com/
SEAL RING
LOCK PORT
PISTON
PRIMARY
SEAL RING LOCK PORT
RETAINER
SCREW VENT PORT
CAM RING SECONDARY

RELEASE
CONNECTING PORT
ROD
PRIMARY
RELEASE
LOCKING DOG PORT
PIN
MANDREL
PROFILE
Figure 26.2 Drawing of FMC end-fitting.
annulus may be built up. The outer sheath itself is not capable of carry-
ing high annulus pressure. Hence, the gas is relieved when the differential
pressure has reached a certain limit, in the order of 0.4 MPa. The gas is
relieved through the ports (vent, primary release, and secondary release),
which open at a predefined differential pressure [5].
26.4 Seal System

The seal system is subjected to different types of loading during the life-
time of the riser. There are two main types of loading, the first is during
activation that is mechanical, and the second one is during service that is a
combination of thermomechanical and physiochemical [6].
Manufacturers have two different ways of terminating the inner liner
in risers. Crimping seal mechanism is a famous one. In this mechanism,
the inner liner anchoring and the sealing are combined in one device,
where a seal ring is swaged against the body of the connector by a collar. In
this way, the inner liner is compressed between the carcass and the collar.
The friction force between the crimp seal and the inner liner is the basis
of anchoring [7].
This chapter reviews the method which considers termination into two
devices. That means the seal system is not combined with the crimp ring seal.
The seal system contains many ring joints gaskets separated by mechanic
rings. The gaskets and the rings are placed on the outside of the inner liner
in a sealing insert. This seal system has also seal seats, which create sufficient
contact pressure between the gaskets and the surrounding surfaces.
26.5 Termination of the Carcass

In order to mount the end-fitting on the pipe, the carcass and the inner liner
should be cut in two. The inner liner surface is prepared for the gaskets,
including the check of surface roughness and measurement of the outer
diameter. Then, with a purpose-built tool, it is prepared for installation of
the inner liner holder. The inner liner holder is temporarily installed [8]
and the rest of the inner liner is removed from the carcass. The inner liner
holder is removed for installing the inner gaskets, the inner casing, the seal-
ing insert, and the inner ring. After installation of these components, the
carcass holder, the inner liner holder and the insulation rings are installed.
When a flexible pipe is used as a riser, it is necessary to anchor the car-
cass. The reason for that is the layers of the pipe are, in principle, free to
move relative to each other. In addition, the carcass has an interlocked
structure (as shown in Figure 26.3), so a relative displacement between
Figure 26.3 Interlock structure of carcass.

the profile is possible. Depending on the profile type, the length of the car-
cass can be elongated or shortened up to 10%, due to the relative displace-
ment [9]. Theoretically, if a carcass is pulled out of a topside riser connector
and shortened due to its own weight it will not support the inner liner of
the top section of the riser as it should support. It is necessary to transfer
the force from the carcass weight to the connector; this is done by the car-
cass holder. In fact, this is done by screwing a steel bushing on the outside
of the carcass and utilizing the inherent spiral grooves. The carcass holder
is a device for mechanical fixation of the carcass; it has a thread on the
inside with the same pitch as the carcass [10].
The carcass holder is screwed on to the carcass and fixed by a number
of screws to avoid any rotations and movements. In order to prevent the
galvanic corrosion, the carcass holder is made of the same material as the
carcass. The outer diameter of the carcass holder, including the insulation
bushing, is given by the inner liner holder since both components have to
fit into the end body. The inner diameter is based on measurements of the
carcass after manufacturing.
The thread dimensions of the carcass holder depend on the carcass
dimensions. The carcass weight induces shear stress; for carrying this
stress, the teeth must have a certain thickness. This thickness can be calcu-
lated by the following equations [11]:
Wcar 3
b> (26.1)
σ y ,car Lt
2
 s
2
Lt = N π d +   (26.2)
π
In the above equations, Wcar indicates to carcass weight, which is carried

as shear stress in the teeth. The thread length is shown by Lt, and also N and
s are the number of pitches in carcass holder and carcass pitch, respectively.
For the evaluation of the carcass role in the end-fitting design, once the
carcass is cut two, some elastic spring-back might be there, which cause
problems of screwing on the carcass holder. In its place, the carcass holder
is divided into split rings as the sheath holder and inner liner holder, still
utilizing the threated structure of the carcass. The split rings are bolted and
mounted together, instead of screwing the carcass holder on the carcass.
This kind of design is not too much sensitive to variations in the carcass
outer diameter.
26.6 Smooth Bore Pipe

The pipes which have no carcass are described as smooth bore pipes. As we
mentioned above, the carcass has a significant role in the seal system of a
connector. The seal system basically is the same for all risers; therefore, a
replacement as the carcass function is essential. The carcass is substituted
by an inner liner expander, which is a tube pressed into the inner liner.
Figure 26.4 is a schematic of an inner liner expander.
Since the expander is only a substitution for the carcass on the first
section of the pipe, the sheath holder and the outer gaskets can con-
tract the riser. Specifically, when the riser does not have pressure armor.
Consequently, there isn’t any radial support to carry the pressure from the
outer gaskets and the sheath holder.
Hence, the sealing of the smooth bore flexible pipe has many differ-
ences with the rough bore flexible pipe. This chapter has shown that carcass
carries the external contact pressure on the inner liner from the gaskets.
Meanwhile, the inner liner itself is not capable of carrying the contact pres-
sure and also the smooth bore pipe does not have the carcass, so a substi-
tution for the carcass is necessary.
Therefore, for solving this problem, using an inner liner expander is the
best choice.
The inner diameter of the expander is equal to the pipe size. In this way,
when the expander is installed, the inner liner will be radially extended.
Sealing insert Inner casing
Inner
liner
Inner liner expander
Figure 26.4 Inner liner expander.

Installing the expander in the inner liner is a challenging task. Therefore, for
solving this problem, the inner liner will be heated up. Consequently, deter-
mining the final outer diameter of the inner liner (used to design the com-
ponents of the seal system) after installation of the expander is too difficult.
As long as the seal system is activated, choosing a criterion for the time is
another challenge. Most of the times the expander is stiff, so during activa-
tion, it does not deform. Instead, the axial displacement of the inner casing
as a function of the applied torque is analyzed during the activation [12].
Carcass and expander have many differences. While the carcass can be
regarded as a number of two-dimensional rings, due to the high pitch, as
well as, the expander has a longitudinal dimension. Hence, the carcass is
only exposed to hoop stress, while the expander is exposed to both hoop
and bending stresses. The carcass is more porous than the expander. There
is a contact pressure between the inner liner and the expander; therefore,
the pressurized fluid in the pipe doesn’t improve the seal system certainty,
as it does in rough bore pipes. Drilling some small holes in the expander
under the gaskets to make it more porous can solve this problem. However,
regarding stress concentration around the holes, drilling the expander
reduces the strength of that. Consequently, the wall thickness of the
expander should be larger.
26.7 Rough Bore Pipe

There is a relation between gasket contact pressure and carcass deforma-
tion. The carcass profile carries all of the pressure. It is assumed that the
inner liner does not carry any pressure, and also contact pressure to the
inner liner is constant along the gasket. The carcass is modeled as a thin-
walled ring in which in the profile, the external pressure is carried as axial
stresses. An equivalent carcass thickness (teq) is defined as [13]
Acar
teq = (26.3)
Scar
In the above equation, the cross-sectional area of the carcass profile

and the carcass pitch are demonstrated by (Acar) and (Scr), respectively. The
compressive carcass hoop strain (εcar) is
δ car
ε car = (26.4)
Rcar
where (δcar) is the radial deformation of carcass relative to the initial radius
(Rcar).
This strain induces stress (σcar) as below:
δ car
σ car = Eε car = E (26.5)
Rcar
Figure 26.5, shows the vertical direction force equilibrium and it’s given
by
teq
Pc = σ θ (26.6)
Rcar Inner liner expander
For obtaining the contact pressure, Eqs. (26.4) and (26.5) is substituted
in Eq. (26.6):
teq
Pc = E 2
δ car (26.7)
Rcar
In the above equation, it seems that in the carcass deformation (δcar),

the contact pressure (Pc) is a linear function. Once the carcass stress (σθ)
and the carcass material yield stress (σy), be equal, the maximum Pc is
deduced as
teq
Pc ,max = σ y (26.8)
Rcar
Pc
σθ σθ
Rcar
dθ
Figure 26.5 Carcass under external pressure.

Table 26.1 Carcass properties.

Carcass profile Acar (mm2) Scar (mm) teq (mm)
40 × 1.0 40 14.4 2.8
40 × 1.2 48 14.0 3.4
55 × 1.4 77 19.8 3.9
Also, the maximum deformation of the carcass can be calculated as
σy
δ car ,max = Rcar (26.9)
E
Eqs. (26.8) and (26.9) do not consider some effects, like strain hard-
ening of the carcass material or contact between neighboring profiles of
the carcass. The equations show when the same carcass profile is used,
with increasing pipe size the maximum contact pressure will be decreased.
Table 26.1 presents three various typical carcass profiles, and by replacing
this amounts in the above equations valuable results are derived.
26.8 Discussion
According to Eq. (26.7), the seal system under varying pressure has been
analyzed. A pressure cycle between P = 0 MPa and P = 20 MPa is applied
in fixed steps of ΔP = 0.4 MPa over a total of 100 increments. The radial
deformation of the carcass during activation and pressure loading is shown
in Figure 26.6. It should be noticed that when the carcass will be removed
by the pressure, the inner liner and the gasket move back to their last posi-
tion, which they had before the activation.
0
–0.1
Deformation (mm)
–0.2
–0.3
–0.4
–0.5
–0.6
–0.7
–0.8
–0.9
100 80 70 60 50 40 30 20 10 0
Increments
Figure 26.6 Deformation of carcass during pressure cycling.

The variation of Pc as a function of the carcass deformation (δcar) is shown

by Figure 26.7, based on the combinations of pipe size and carcass profile
stated in Table 26.2. There is a linear relationship between the contact pres-
sure and the carcass deformation till the carcass starts to yield. Pc remains
constant for the large diameter pipes, while it differs for the more rigid
pipes. Eqs. (26.8) and (26.9) give the maximum contact pressure and corre-
sponding carcass deformation for the equal carcass profiles and pipe sizes
as presented in Figure 26.7. According to Figure 26.7, for some of the pipes
δcar (mm)
0 –0.2 –0.4 –0.6 –0.8 –1 –1.2 –1.4 –1.6 –1.8 –2
0
–2
–4
–6
–8
Pc (MPa)
–10
–12
–14
–16
–18
–20
1 2 3 4 5 6 7
Figure 26.7 Pc and δcar relationship for different carcass profiles and pipe sizes (values are
shown in Table 26.2).
Table 26.2 Maximum possible contact pressure.
No Pipe size (in) Rcar (mm) Profile Pc,max (MPa) δcar,max (mm)
1 6 76.2 40 × 1.0 10.7 0.11
2 6 76.2 40 × 1.2 12.9 0.11
3 6 76.2 55 × 1.4 14.8 0.11
4 8 101.6 40 × 1.2 9.7 0.14
5 8 101.6 55 × 1.4 11.1 0.14
6 10 127.0 55 × 1.4 8.9 0.18
7 12 152.4 55 × 1.4 7.4 0.22

Pc is affected by neighboring profiles and strain-hardening after the carcass

yield, which Eqs. (26.3) to (26.6) are not able to predict. Nevertheless, Eqs.
(26.8) and (26.9) give fine values for an approximation of the contact pres-
sure for various carcass profiles and pipe sizes.
26.9 Conclusions
This chapter discussed the role of the carcass in the sealing system of the
riser connector. The termination of the carcass is a significant matter in
the end-fitting of the riser. From Eqs. (26.8) and (26.9), the pressure at
which carcass yield is the largest possible contact pressure. The initial con-
tact pressure from crimping the gasket on the inner liner has no effect on
this contact pressure.
According to Table 26.2, if the carcass profile of the pipes is same, the
small diameter pipes are able to carry a higher contact pressure than the
larger pipes, since they are more rigid. Likewise, for equal pipe sizes, when
the contact pressure is maximum, the maximum deformation of the car-
cass is independent on the carcass profile. This is seen both for the 6-in
(40 × 1, 40 × 1.2, and 55 × 1.4 profiles) and 8-in (40 × 1.2 and 55 × 1.4 pro-
files) pipes. In addition, with increasing pipe sizes, Pc is decreased.
As a matter of fact, for the different combinations of pipe size and car-
cass profile, the pressure on the inner liner is generally smaller than the
pressure on the sealing insert, although the difference is in a range of 50%,
except for the 8-in 40 × 1.2 pipe where the difference is up to 100%. The
rigidity of the sealing insert caused this difference, whereas the supporting
carcass and the inner liner are deformable.
The gap between the inner liner and the sealing insert and also between
the inner liner and the inner casing remain independent and persistent of
carcass profiles and pipe sizes. Regarding to determine the correct activa-
tion, deformation of the carcass for all carcass profiles and pipe sizes is the
same.
It is from Figure 26.7 that in the maximum contact pressure, the carcass
deformation is not a function of the carcass profile; actually, the only func-
tion of the yield stress of the carcass material and the pipe size affects it.
The carcass must be fixed against the axial movements, in the end-fitting;
otherwise, the seal system function is failed. Because the carcass is threated,
so external pressure causes the carcass to rotate and expand axially, which
decrease the diameter and then reduce the contact pressure. While, the
axial position of the gasket relative to the carcass is not important.
This chapter showed that the carcass has a notable effect in the sealing
system, so for smooth bore pipe which has no carcass, inner liner expander
is the best substitution. It has many differences with carcass but still it is
required for improving the seal system of a riser connector.
The riser connector structure is so complicated; therefore, for future
work, finite element study for considering the carcass role in the seal-
ing system of the riser connector and comparing it with theoretical
methods would be useful for better understanding the applicability of
theoretical models.
References
1. American Petroleum Institute, API Publishing Services, 1220 L Street, N.W.
Washington D.C. 2005, “Recommended Practice for Flexible Pipe”, API RP
17B, 5th edition, 2014.
2. American Petroleum Institute, API Publishing Services, 1220 L Street, N.W.
Washington D.C. 2005, “Specification for Unbonded Flexible Pipe”, API Spec
17J, 4th edition, 2014.
3. American Society for Testing and Materials, ASTM, Bar Harbor Drive,
West Conshohocken, Pennsylvania 19428, USA. Test Methods for Tensile,
“Compressive and Flexural Creep and Creep Properties”, 1995.
4. Berge, S. and Eide, O. “Thermal Cycling of High Temperature Flexible Pipe”.
In Conference Papers – An International Conference on Oilfield Engineering
with Polymers. MERL, 1996.
5. Berge, S. and Eide, O. “Facility for Thermal Cycling of End Terminations of
Flexible Pipe”. In Proceedings of the Third European Conference on Flexible
Pipes, Umbilicals and Marine Cables – Material Utilization for Cyclic and
Thermal Loading. Marinflex, 1999.
6. Dansk Standard, Charlottenlund, Denmark. Geometric Product Specifications
(GSP) Surface Texture: Profile Method – “Rules and Procedures for the assess-
ment of Surface Texture”, 1st Edition, 1997.
7. Fox, R.W. and McDonald, A.T. “Introduction to Fluid Dynamics”. John Wiley &
Sons, Inc, 3rd Edition, 1985.
8. Kalman, M., Belcher, J., Chen, B., Fraser, D., Ethridge, A. and loper, C.
“Development and Testing of Non-Bonded Flexible Pipe for High Temperature/
High Pressure/Deep Water/Dynamic Sour Service Applications”. In Proceedings
of Offshore Technology Conference. Offshore Technology Conference, 1996.
9. MacFarlane, C.J. “Flexible Riser Pipes: Problems and Unknowns”. Engineering
Structures, 11, 1998.
10. Patel, M.H., Witz, J.A. and Tan, Z. “A Flexible Risers Design Manual”. Dilke
House, 1 Malet Street, London WC1E 7JN, UK, 1993.
11. Pettani-Auzire, C. “Flexible Dynamic Risers”, State of the art. Technical

Report, Norwegian Petroleum Society, Oslo, Norway, 1986.
12. Stevenson, A. and Campion, R. “Thermoplastic Performance Properties
for Flexible Pipes and Umbilicals”. In Proceedings of the Second European
Conference on Flexible Pipes, Umbilicals and Marine Cables – Structural
Mechanic and Testing. Marinflex, 1995.
13. Young, W.C. “Roark’s Formulas for Stress and Strain”. McGraw-Hill, 6th
Edition, 1989.
27
Bend Stiffener and Restrictor Design
27.1 Introduction
One of the critical areas of a flexible riser is the top part of the riser, just
before the hang-off arrangement. This area is prone to overbending, and
hence, an ancillary device is incorporated into the design to increase the
stiffness of the riser and prevent bending of the riser beyond its allow-
able bend radius. For this application, flexible pipe manufacturers tend to
use bend stiffeners to provide a better performance in applications with
high-motion vessels. Bend stiffeners also provide a moment transition
between the riser and its rigid end connection. Bend stiffeners are designed
separately from the pipe cross-section analysis, and specialized software
is used for this purpose. Global loads from the flexible riser analysis are
used as input to the bend stiffener design, and the design in detail can refer
to reference [1].
Bend stiffeners are normally made of polyurethane material, and their
shape is designed to provide a gradual stiffening to the riser as it enters
the hang-off location. The bend stiffener polyurethane material is itself
anchored in a steel collar for load transfer. Bend stiffeners are sometimes
utilized subsea, such as in steep-S or steep wave applications to provide
support to the riser at its subsea end connection and to prevent overbend-
ing at this location. Design issues for bend stiffeners include polyurethane
fatigue and creep characteristics. The cone dimensions are determined
with the most severe tension and angles combination to satisfy a mini-
mum bending radius acceptance criteria and an acceptable fatigue life for
the pressure vault and the armor wires of the flexible structure. Figure 27.1
shows an example of a bend stiffener. Note that bend stiffeners longer than
20 ft have been manufactured and are in operation in offshore applications.
547
Figure 27.1 An example of a bend stiffener.
27.2 Response Model

The key to efficient design/analysis of bend stiffeners is that it is possible
to establish a local quasi-static response model of the bend stiffener. This
is because the bend stiffener only will govern the local behavior close to
the support; the overall global behavior of the riser/umbilical will not be
influenced by the bend stiffener.
The local response model as shown in Figure 27.2 consists of a bend-
ing stiffener of length SB and a riser/umbilical part of length SR. The
SB
EI (s) SR
s
EIR
αR
Figure 27.2 Local response model.

Bend Stiffener and Restrictor Design 549
bending stiffness of the bend stiffener is denoted EI(s), where s is the

length coordinate from the tip of the bend stiffener and EIR is the bending
stiffness of the riser/umbilical.
The purpose of including a small riser/umbilical part in the local model
is to enhance the accuracy of local analyses using results from global anal-
yses as boundary conditions. The riser/umbilical part in the local model
should be sufficiently long to ensure that any boundary effects from the
bend stiffener are described by the local model to obtain a complete sepa-
ration of the local and global response models.
For the purpose of bend stiffener sizing, it is assumed that the bend
stiffener and riser/umbilical is fully integrated in bending. This means that
a beam model including material and geometric stiffness contributions
can be applied in the local analyses. A so-called “shooting method” for
solution of two point boundary value problems is well suited for efficient
local analyses [2, 3]. The local analyses can alternatively be based on a
Finite Element (FE) approach using beam elements and a non-linear static
solution scheme.
27.3 Extreme Load Description

The loads on the local bend stiffener response model are governed by
effective tension T and relative angle αR, see Figure 27.2. Combinations
of effective tension and relative angle aggregated for all relevant extreme
load conditions define the design loads on the bend stiffener. The following
definitions are introduced to describe the loading:
• The design contour is defined as the right side outer bor-

der of the design space, see Figure 27.3. The load contour is
most conveniently formulated in terms of relative angle as
function of the associated tension: α Rc (T ) where Tmin ≤ T
≤ Tmax.
• The design line is defined as the total tension range, Tmin ≤ T
≤ Tmax, at the maximum relative angle αmax.
The design line represents the simplest, conservative approximation

to the design contour. The degree of conservatism depends on actual
correlation between αR and T. A major advantage is that the parameters
describing the design line can be established by standard extreme value
statistics. Furthermore, this parameterization of the loading allows for
non-dimensional bend stiffener design as described later.
T
Load contour
Tmax
Load space Design line
Tmin
αR
αmin αmax
Figure 27.3 Extreme load description.
The design contour gives a more consistent description of the combined

loading on the bend stiffener. However, advanced statistical estimation
techniques will, in general, be required to establish accurate estimates of
the design contour from irregular global dynamic analyses. Rational con-
servative approximations are hence needed to establish design contours
for practical design applications. This may involve linear/bilinear conser-
vative approximations based on the actual correlation between tension and
relative angle. Since capacity curves are generally convex, the convex
approximation to the design contour can be useful. Reference [4] describes
how this can be obtained from irregular time series data.
27.4 General Optimization Scheme

The bend stiffener geometry is given in terms of the outer diameter Ro(s)
as function of the length coordinates. This gives Eq. (27.1) for the
bending stiffness distribution along the bending stiffener
Ebπ 4
EI (s ) = EI r +
4
(
Ro (s ) − Ri4 ) 0 ≤ s ≤ SB (27.1)
where Eb is the modulus of elasticity of the bend stiffener and Ri is the

inner radius. A shape function describing the variation of outer diame-
ter in terms of a few parameters (i.e., degrees of freedom) is needed to
find a practical solution. These parameters are in an optimization scheme
considered as the unknowns that need to be determined. As an example,

the simplest choice of shape function is to assume a conical geometry
 ( R − R )s 4

Eπ 4
EI (s ) = EI r + b  B i
+ Ri  − Ri4  0 ≤ s ≤ SB (27.2)
4  SB  
where RB is the radius at the root end of the bend stiffener. Thus, the bend-
ing stiffness distribution is a function of the length and root diameter of
the bend stiffener. These parameters are in an optimization scheme consid-
ered as the unknowns that need to be determined.
In a general formulation, xs denote the vector of shape parameters used
to define the bend stiffener geometry over the length 0  s  SB . The shape
function can in this case, e.g., describe a piecewise linear variation of outer
radius of the bend stiffener or any other parameterized outer geometry.
Two criteria are applied to establish optimal bend stiffener geometry:
(1) O
ptimized bend stiffener shall obtain maximum allowable
curvature given by the capacity curve for at least one loca-
tion along the load contour.
(2) The bend stiffener length shall be minimized.
The margin to the capacity curve for given values of the geometry
parameters can be expressed as
( )
m( x s , SB ) = min κ m x s , SB ,T ,α Rc (T ) − κ R (T )) for Tmin ≤ T ≤ Tmax
(27.3)
where;
α Rc (T ) : Load contour;
κR(T): Capacity curve;
κm(xs, SB, T, αR): Maximum curvature along bend stiffener for loads
(T, αR) and given geometry parameters (xs, SB).
Thus, the bend stiffener optimization is recognized as a classic opti-
mization problem: Minimization of the objective function, SB with
constraints m(xs, SB) = 0. A tailor-made software, Benito, has been
established to solve this problem. The following main functionalities
have been included:
(1) A general purpose optimization procedure using a quasi-

Newton solution [5].
(2) Non-linear static FE analysis scheme for the local response
model.
It has been experienced that the described optimization challenge can be

complex depending in the complexity of the load contour and the capacity
curve. The solution space can be quite limited with potential non-unique
solutions. A robust optimization algorithm is hence essential. A robust FE
solver for evaluation of the constraints is also crucial for the numerical
performance of the software. The FE approach may also easily be
extended to handle non-linear material properties. This applies to the
bend stiffener material as well as the riser/umbilical itself.
27.5 Application Example

A deepwater umbilical with an outer diameter of 0.14 m is selected as
example case. The input is given in terms of the load contour as specified
in Figure 27.4 and the capacity curve as specified in Figure 27.5. The opti-
mized conical geometry is shown in Figure 27.6 giving a length of 4.45 m
and a root diameter of 0.9m.
The bend stiffener curvature response evaluated along the load contour
is compared to the capacity curve in Figure 27.5. It is seen that the maxi-
mum curvature is touching the capacity curve confirming full utilization.
load contour
800
750
700
650
tension
600
550
500
450
400
0 2 4 6 8 10 12 14 16 18 20
relative angle
Figure 27.4 Load contour.

Bend stiffener curvature margin

0.14
0.13
Capacity curve
0.12
maximum curvature
0.11
0.1
0.09
0.08
Maximum bend
0.07 stiffener curvature
along load contour
0.06
0.05
0.04
400 450 500 550 600 650 700 750 800
Applied tension
Figure 27.5 Capacity curve and bend stiffener performance.
Bend stiffener geometry

0.45
0.4
0.35
Outer radius
0.3
0.25
0.2
0.15
0.1
0.05
0 1 2 3 4 5 6
Length coordinate
Figure 27.6 Bend stiffener geometry.
27.6 Non-Dimensional Bend Stiffener Design

Non-dimensional design curves for easy bend stiffener sizing can be con-
structed for a simplified subset of the general optimization scheme as out-
lined in the previous section. The following restrictions are introduced:
(1) Th
e entire tension range is conservatively assumed for
the maximum relative angle (i.e., design line approach, see
Figure 27.3);
(2) Capacity is described by constant curvature independent of
the effective tension;
(3) A stiffness modeling of the bend stiffener is applied (i.e.,
geometry not considered).
The capacity curves are restricted to a constant critical curvature κc

independent of the effective tension. The associated critical curvature is
given as κc = 1/Rc.
To reduce the number of parameters describing the bend stiffener prop-
erties to a minimum, it is required to consider the bending stiffness vari-
ation along the bending stiffener as the primary outcome of the sizing.
This allows for excluding the modulus of elasticity from the sizing, which
is required if geometry is applied as the basis for sizing as shown in the
previous section.
A convenient parametric representation of the bending stiffness distri-
bution can be established as the bending stiffness distribution that will give
a constant radius of curvature, RD, all along the bend stiffener for a given
loading, (αR, TD). By assuming that the deflection of the centerline of the
bend stiffener is a circle segment with a specified radius, RD, the bending
stiffness distribution can be found by simple equilibrium considerations
[2, 3]
 s 
EI (s ) = EI r + 2TD RD2 sin 2  0 ≤ s ≤ SB = RDα R (27.4)
 2 RD 
It should be noted that this is the shortest possible bend stiffener for the
load case described by (αR, TD) if RD = Rc. The derived stiffness distribu-
tion given by Eq. (27.4) is for this reason also denoted the design equation
as it provides an optimal bend stiffener for one specific load case.
For more complex load cases, optimization is required considering TD
and RD as unknown parameters to determine the optimal bend stiffener
geometry. This means that an optimal configuration is sought within the
class of functions described by the design equation. The general proce-
dure as outlined in the previous section can be applied. Alternatively, a
tailor-made procedure for this simplified optimization problem can be
applied [2, 3].
Figure 27.7 shows typical results for a bend stiffener geometry obtained
by the described optimization procedure, the specified critical radius of
curvature is Rc = 3m. Figure 27.7 shows the curvature along the local
model for 50 discrete points on the design line. Distinct curvature peaks
are observed close to the bend stiffener tip and at the root end. Both
peaks agree with the specified curvature capacity confirming optimal
performance.
The basis for constructing non-dimensional design curves is to apply
the described procedure to establish design parameters TD and RD for
relevant ranges of all input design parameters.
The design parameters TD and RD can hence be expressed as functions
of Tmin, Tmax, αmax, EIr, Rc, and Sr. Thus, the bend stiffener design prob-
lem is described by a functional relationship between eight parameters
expressed in terms of the basic dimensions of force [N] and length [m].
According to the principles of standard dimensional analysis, an equiva-
lent description can be obtained by a functional relationship between six
independent non-dimensional parameters. Furthermore, it can be shown
that the bend stiffener design is independent of Sr provided that Sr is suf-
ficiently long, typically Sr* = Sr/Rc [2]. The bend stiffener design is hence
completely described by the five non-dimensional parameters given in
Table 27.1.
For practical applications, it is convenient to express the nondimen-
sional design parameters TD* and RD* as function of αmax and θ* for a
0.4
Curvature
Rc = 3m
Tmax
0.3
0.2
0.1
0.0 Tmin
0 1 2 3 4
Length-coordinate
SR SB
Figure 27.7 Bend stiffener performance.

Table 27.1 Non-dimensional parameters.
TD Non-dimensional design force

TD* =
Tmax
RD Non-dimensional design radius

RD* = of curvature
Rc
Non-dimensional stiffness
Tmax Rc2
θ* = parameter
EI r
Tmin Force ratio

TR* =
Tmax
θmax Maximum force angle
given force ratio, TR*. Examples of such non-dimensional design curves

are presented graphically in Appendix A for TR* values in the range of 0 to
0.8. A rather limited parameter range is considered for presentation pur-
poses. Extended design curves covering a wider parametric range is used
in practical design. Intermediate values can be obtained by interpolation
with sufficient accuracy. The non-dimensional design curves provide the
main bend stiffener geometry with a minimum of computational efforts.
This information provides the basis for detailed practical design of the
bend stiffener.
In practical implementations, it might be desirable to use a conical
bend-stiffener design to approximate the optimal bend stiffener described
by the design equation. Furthermore, it might be required to perform
detailed FE analyses of the bend stiffener taking into account possible
non-linear material properties of the bend-stiffener.
27.7 Alternative Non-Dimensional Parameters

The five non-dimensional parameters given in the previous section and
Table 27.1 are not unique. It follows from the principles of dimensional
analysis that any set of five independent dimensionless parameters will
represent the same solution.
It is for this reason useful to investigate the bend stiffener curvature
for small angles at a particular tension level. Intuitively, for small angles,
we expect linearity, i.e., the curvature is proportional to the angle. This is

described in the next section. The unit of this gradient parameter is curva-
ture per angle. Since κc/αmax has the same units as κc = 1/Rc, one alterna-
tive nondimensional parameter should contain Rcαmax.
Figure 27.8 presents all the results of Appendix A in a more compact
manner. The following set of alternative nondimensional parameters is
used: 1/TD*, RD*, TD*, αmax, and θ*α2max.
A main finding from Figure 27.8 is that relative force angle, αmax, is
not very important. The figure includes data for αmax from 10° to 90°. The
small effect of the angle variation is indicated by the rather small variation
seen for each of the curve branches. Thus, this choice of non-dimensional
parameters has for all practical purposes eliminated αmax. It is therefore
sufficient to consider only one characteristic relative angle when producing
design charts. Another striking feature is the shape similarity of the curves
for Tmax/TD and RD/RC.
For a non-dimensional bend stiffener load less than 20, where the riser
bend stiffness dominates, the optimal bend stiffener length is shorter than
RCαmax and is practically independent of the force ratio (the tension
range). This is mentioned in /6/.
2.5
RD Tr =
Tmin =0
Tmax
2.0 RC 0.2
1.5 0.4
0.6
1.0 0.8
0.5
0.0
1 10 100 1000 10000
Tmin
3.5 Tmax Tr = Tmax
=0
3.0 TD 0.2
2.5
2.0 0.4
1.5 0.6
1.0 0.8
0.5
0.0
1 10 100 1000 10000
R2
Tmaxα2max C
EIr
Figure 27.8 Design chart for bend stiffener.

For non-dimensional bend stiffener load of about 30, the value of both
Tmax/TD and RD/RC is about one. The bend stiffener length is therefore
about RCαmax.
For non-dimensional bend stiffener loads greater than 30, the riser load
dominates and there is a strong dependency on the force ratio (the tension
range). The bend stiffener is longer than RCαmax for this region.
27.8 Conclusions
The bend stiffener is a vital component to ensure the structural integrity
of the interior functional and strength elements of risers and umbilicals.
A general optimization scheme has been outlined which gives the short-
est bend stiffener for a given extreme loading environment. Of particular
importance for umbilicals is the capability to include the capacity curve
and a complex extreme load description in the bend stiffener optimiza-
tion scheme. A bend stiffener design procedure suitable for use in delivery
projects is outlined. The methodology presented and the knowledge of the
shortest possible bend stiffener is also of great benefit in such projects. The
methodology is also very useful for feasibility studies.
References
1. SøDahl N, Ottesen T. Bend Stiffener Design for Umbilicals[C]// ASME
2011, International Conference on Ocean, Offshore and Arctic Engineering.
2011:449–460.
2. Sødahl N, ‘Methods for Design and Analysis of Flexible Risers’, Phd Thesis,
Div of Marine Structures – NTNU 1991.
3. Sødahl N, Larsen C M, ‘Design procedure for bending stiffeners in flexible
riser systems’, 1989, PRADS 89, Varna, Bulgaria.
4. Ottesen T, Aarstein J A, ‘The statistical boundary polygon of a two parameter
stochastic process’, OMAE 2006, Hamburg.
5. Polak E, ‘Computational methods in optimization’, Academic Press, 1971.
Appendix A: Non-dimensional bend stiffener design curves
Tr*= 0 Tr*= 0.2

1.4 2.0
TD* RD*
1.5 1.9 90
50
1.2 1.8
1.1
1.7
1.0 25
1.6
0.9 20
10 1.5
0.8
1.4
0.7 15
1.3
0.8
15 1.2
0.5
1.1 10
0.4
20 1.0
0.3 90 50 30 25
0.2 0.9
0 200 400 600 800 1000 1200θ*1400 0 200 400 600 800 1000 1200 1400
θ*
Tr*= 0 Tr*= 0.2
2.2 1.4
RD* TD*
2.1 90 1.3
50 30 25
2.0 1.2
1.9 20
1.1
1.8
1.0
1.7
1.6 0.9
10
1.5 0.8
15
1.4
0.7
1.3
0.6 15
1.2
0.5 20
1.1 10 25
30
1.0 0.4
90 50
0.9 0.3
0 200 400 600 800 1000 1200θ*1400 0 200 400 600 800 1000 1200θ* 1400
Tr*= 0.4 Tr*= 0.6
1.4 1.4
TD* TD*
1.3
1.5
1.2
1.2
1.1
1.1
1.0
10
0.9 1.0 10
0.8 0.9
15 15
0.7 20 20
25 0.8 25
30 30
0.6 50 50
90 90
0.7
0.5 0 200 400 600 800 1000 1200 1400
0 200 400 600 800 1000 1200θ*1400 θ*
Tr* = 0.4 Tr* = 0.6

1.5 1.25
90 90
R*D R*D
50 50
1.20
1.4
30 30
25
1.15 25
1.3
20
20
1.10
1.2 15
15
1.05
1.1
1.00
10 10
1.0
0.95
0.9 0.90
0 200 400 600 800 1000 1200 θ*1400 0 200 400 600 800 1000 1200 1400
Tr* = 0.8 *
θ*
Tr = 0.8
1.40 1.10
*
TD
R*D 90
1.35
50
1.30
30
1.05
1.25
20
1.20
1.15 15
1.00
1.10
10
1.05 10
1.00 0.90
15
0.95 20
30
0.90 50
90 0.80
0.85 0 200 400 600 800 1000 1200 1400
0 200 400 600 800 1000 1200 θ*1400 θ*
28
End Termination Design for Umbilicals
28.1 Introduction
The end termination for umbilicals is mechanical fitting attached to the
end of an umbilical which provides a means of transferring installation
and operating loads, fluid, and electrical services to a mating assembly
mounted on the subsea installation or surface facility. Armored umbilicals
shall be terminated with end terminations with a minimum loading capa-
bility equal to or exceeding the maximum working load of the umbilical.
The terminations shall be designed for use in a marine environment.
Subsea Umbilical Termination (SUT) is the mechanism for mechani-
cally, electrically, optically, and/or hydraulically connecting an umbilical or
jumper bundle to a subsea system. The SUT contains the UTA (Umbilical
Termination Assembly) and STI (Subsea Termination Interface) but does
not include bend restrictors or stiffeners. Figures 28.1 and 28.2 illustrate
SUT for steel tube umbilical and thermoplastic umbilical.
In the design of subsea termination assembly, all parts of the SUT
need to be studied under the various loading scenarios, and for each load-
ing scenario the tensile, bending, shear, bearing, and equivalent stresses are
calculated. The calculated stresses are then compared against the allowable
stresses in the materials and if the ratio is found to be below 1, and then,
the part is considered fit for purpose under the loading scenario studied.
28.2 Umbilical Termination Assembly

28.2.1 General
UTA, or named SUTA, is used to distribute hydraulic and electrical power,
chemicals, and communication signals from an electro/hydraulic umbili-
cal to multiple subsea assemblies. UTA contains the typical components
as follows [2]:
561
BEND RESTRICTOR SPOOL UTA

CABLE
ARMOUR BODY TUBE SERVICE TUBING
WELDING LOOP
ELECTRICAL/FIBRE
OPTIC CONNECTOR HYDRAULIC CONNECTOR
(MQC PLATE)
Bend Restrictor STI UTA
SUT
Figure 28.1 Steel tube umbilical or thermoplastic umbilical with spool [1].
BEND RESTRICTOR UTA HOSE FITTINGS &

BULKHEAD PLATE
CABLE
ARMOUR BODY SERVICE TUBING
LOOP
ELECTRICAL/FIBRE
OPTIC CONNECTOR HYDRAULIC CONNECTOR
(MQC PLATE)
Bend Restrictor STI UTA
SUT
Figure 28.2 Thermoplastic umbilical [1].
• Electrical/fiber optic wet-mateable connectors

• Hydraulic/chemical junction plates
• Bend restrictor
• Mud mat
28.2.2 UTA Design

The UTA design to choose is largely determined by field size and installa-
tion method. The UTA design to choose is largely determined by field size
and installation method.
End Termination Design for Umbilicals 563
There are various types of UTA design and two of them are more
common:
• Stab and Hinge-Over

• Bolted UAT and Mudmat Assembly
Stab and Hinge-Over consists of Umbilical Termination Head (UTH)

and mudmat. The S&HO design allows mudmat to be installed separately
from the UTH (Figures 28.3 and 28.4). After mudmat is laid on seafloor,
the UTH is lowered down and maneuvered with application of external
horizontal forces, so that the stab is aimed and inserted into the guide fun-
nel in the mudmat. ROV locks UTH in place on mudmat. Padeye is used
as the lifting point for UTH/umbilical installation and recovery.
The hinging stab shall be attached to the frame and is the interface
between the UTH and mud mat structure. The installation/recovery padeye
shall be designed to take the full recovery load of the umbilical and UTH.
The UTH frame size shall be minimized in order to simplify handling
and over boarding. The UTH shall be kept small to fit inside most
umbilical over boarding chutes and can be maneuvered through most
vessel umbilical handling systems. The UTH, combined with the umbilical
split barrel (supplied by umbilical manufacturer), shall be designed to
sit on the umbilical reel of an installation vessel at the end and/or at the
beginning of the umbilical.
UTH
Stab Pin
UTH MQC Plate
Parking Stab Funnel

MQC Plate
Mudmat

UMBILICAL LIFTING YOKE

FLANGE/SPLIT BARREL
(SUPPLIED BY OTHERS) LOGIC CAP RECEIVERS
ELECTRICAL FLYING LEAD
ROV OPERATED
LOCKING PINS STAB AND HINGEOVER ASSY
HYD FLYING LEAD

(IN PARKED POSITION) PARKING PLATE FOR
ELECTRICAL FLYING LEAD
HYD FLYING LEAD

(IN OPERATIONAL POSITION)
MUD MAT
(SIZE TO BE DETERINED)
For Bolted UAT and Mudmat Assembly, the hydraulic manifold

structure bolted to the mudmat on board before they are lowered
down on sea floor together with umbilical from installation vessel.
The installation vessel should be able to handle the weight and size
of the SUTA and mudmat assembly. There will be no need for ROV
assistance during the operation. Arm shape of yoke which is attached
either SUTA box or mudmat is used as the lifting device for subsea
installation and recovery.
All spare tubes in umbilical shall be routed to MQC positions on the
SUTA (Figure 28.5). The SUTA shall be designed to overall dimensions
that allow for ground transportation from fabrication facility to final des-
tination. The yoke shall be designed to take the full recovery load of the
umbilical, SUTA box and mudmat, with appropriate safety factor. Center
of Gravity (CG) of SUTA and mudmat assembly should be such that
minimum entry angle be maintained before the assembly is landed on
sea floor. The connection between SUTA and mudmat should be strong
enough so that the mudmat will first break away from mud suction force
when the assembly is lifted up to surface.
The following controls equipment used to terminate umbilical services
have the following influence on UTA primary dimensions.
Used to terminate hydraulic and chemical services, they tend to affect
the width dimension of the UTA in the main as they are generally fitted
Mudmat
Umbilical
Electrical Yoke Bend Limitor
Connectors SUTA Box
Bolted/Pinned Sea
Connection Floor
to the vertical sides of a UTA but have to accommodate small bore tubing
running horizontally into the back of their hydraulic couplers. The small
bore tubing has to reach down the length of the UTA (from the Back Face)
and turn at right angles to interface with MQC plate couplers and provide
compliance for couplers to float. This has a greater size constraint than the
height of plate hence affects UTA width more than height. Where MQC
plates require to be mounted on opposing sides of a UTA, this greatly
affects the final width dimension.
Typically, it affects the height, width, and length of the UTA as small
bore tubing has to route into and then out of the isolation valve before
running onto MQC plates.
Typically, cable termination type units should be packaged into the UTA
along with hose extension to output connectors. This tends to affect the
length of UTAs as umbilical cables must be run inside the UTA before ter-
minating. A level of cable over length is required to enable the cables to be
terminated outside of the enclosure and to ensure a second opportunity at
terminating if the first attempt fails.
28.2.3 UTA Structural Design Basis

When considering the design of a UTA, the responsible party should assess
the available industry standards and elect a suitable standard on which to
base the design. For structural design, refer to guidance given in API 17P.
When considering the design of a UTA, the responsible party should
assess the available industry standards and elect a suitable standard on
which to base the design. For structural design, refer to guidance given in
API 17P.
The UTA design should identify all the relevant mechanical loads and load
combinations to be experienced by the UTA and present this in a Design
Basis document. The Design Basis document is the statement of applied
loads and load resistance requirements for the UTA load bearing structure.
The following is a list of potential candidate industry standards that may
be elected as a design specification for a UTA. Special care should be taken
when selecting a structural standard for use in a design basis as not all
sections of the standard may be applicable [6].
• DNV 2.7-3
• DNV-OS-C201
• DNV VMO Standard (DNV-OS-H101 to H206)*
• API 17A Appendix K
• API 17D Appendix K
• API RP 2A WSD
• AISC 13th Edition (ASD)
It should be noted that all the above design standards work on the basis
of defining a static mass or weight (Static Rating) for the equipment to be
lifted/handled. All then apply design factors to the static number in order
to cover dynamic accelerations that will be experienced in an offshore lift-
ing and handling environment. Given that the identity of the installation
vessel/vessel crane/installation spread is rarely available at the time of
design of the UTA, this is the most practical way forward for UTA struc-
tural design.
28.3 Subsea Termination Interface

A STI is a structural unit which forms an interface between the umbilical
and the UTA, and it features a bolted flange for physical connection to
the UTA (Figure 28.6). The STI is typically supplied by the umbilical sup-
plier. Within the STI, all tensile-strength members of the umbilical such as
armour wires, rods, or metallic tubes themselves are physically terminated
by use of an approved method.
A STI is a structural unit which forms an interface between the umbili-
cal and the UTA, and it features a bolted flange for physical connection to
the UTA. The STI is typically supplied by the umbilical supplier. Within
the STI, all tensile-strength members of the umbilical such as armour
wires, rods, or metallic tubes themselves are physically terminated by use
of an approved method.
UTA
Transition
Spool
Bend Restrictor STI
Figure 28.6 Thermoplastic Umbilical [7].
The STI will normally include a separate cathodic protection system.

The STI may be equipped with lifting points, yet these are normally
intended for onshore handling only, as a UTA will be equipped with lifting
points rated for offshore handling and use during installation.
The size of the STI is mainly driven by one or more of the following
parameters:
• Type and number of elements within the umbilical;

• Method (size) of the physical termination of the tensile
strength members and any other umbilical element to be
terminated within the STI
• Sufficient space to allow for tube-jointing beyond the STI
(i.e., access for welding apparatus/welding operators to per-
form the work, and also for NDE to be performed).
Total rigid length is described as in Figure 28.6.

To reduce the STI length, the interface communication between UTA
and umbilical manufacturer is established early. UTA tubes are normally
distributed on circular patterns through the interface opening on the UTA
to match the umbilical orientation. A static umbilical will present itself to
the two UTA’s in opposite handed configurations. This must be planned for
and managed with early interface agreements. Where it can be technically
justified, using the same dimension and material on both UTA tubing and
umbilical tubing at the interface can reduce or remove the need for weld
adaptors inside the STI and might reduce the STI length. The UTA man-
ufacturer should build the interfacing UTA tubing within the Transition
Spool. The position of the thermoplastic hose interface connection to a
UTA bulkhead plate should be located to allow routing and interfacing of
the hoses within the UTA.
An STI may comprise a Transition Spool (split barrel) housing the tran-
sition between umbilical tubing and UTA tubing in case there is lack of
space to perform tube joints inside the UTA. The use of a Transition Spool
will increase the total length of the STI, yet in theory, the length of the UTA
may be reduced accordingly.
References
1. Thomas Worzyk. Submarine Power Cables - Design, Installation, Repair,
Environmental Aspects, Springer, 2009.
2. ISO 13628-5/API 17E - Petroleum and Natural Gas Industries - Design and
Operation of Subsea Production Systems - Part 5: Subsea Umbilicals 2009.
3. Beedle A, Stansfield J. Current trends and design limitations of subsea
control hardware. Offshore Technology Conference, 20663, Houston, 2010.
4. IEC 60502-2 - Power cables with extruded insulation and their accessories
for rated voltages for voltages from 6kV up to 30kV, Second Edition.
5. BS EN 10244-2 2001: British Standard Steel wire and wire products. Non-
ferrous metallic coatings on steel wire. Zinc or zinc alloy coatings.
6. BS EN 10257-2 1998: British Standard Zinc or Zinc alloy coated non-alloy
steel wire for armouring either power cables or telecommunication cables.
Submarine cables.
7. International standard organization. ISO 13628 - 15 Petroleum and nat-
ural gas industries-design and operation of subsea production systems - Part
15: Subsea structures and manifolds, 2009.
29
Mechanical Properties of Glass
Fibre Reinforced Pipeline
During the Laying Process
29.1 Introduction
With the development of the offshore oil industry, offshore exploration
has progressed from shallow seas to deep waters. Composite materials are
being used to improve the performance of pipelines [1]. The properties
of polymeric composite materials, including their low weight, corrosion
resistance, and extended lifetime against fatigue phenomena, have ren-
dered them a competitive candidate for offshore pipeline engineering
[2, 3]. The application of glass fibre reinforced flexible pipes can improve
both production operations and deep water exploration [4]. Further, the
mechanical performance of flexible laying is better than that of other risers.
Pipeline laying methods mainly include the S-type pipe laying method,
J-type pipe laying method, rolling pipe laying method, and dragging pipe
method. Each method has its own characteristics and is suitable for differ-
ent pipeline laying conditions [5]. Pipeline laying process analysis methods
include static and dynamic analysis [6].
A variety of methods have been proposed to study the mechanical prop-
erties of metal pipelines during laying, including the catenary method,
singular perturbation method, finite difference method, finite element
method, and more [7]. In 1968, the rigid catenary method was used by
Dixon and Rutled to solve the stress and shape of deepwater submarine
pipeline S-shaped laying [8]. Gong established the static equilibrium gov-
erning the differential equation of a pipe element, using the stiffened cate-
nary theory, and the solution equations of the total pipeline configuration
from a lay-barge over a stinger to the seabed were derived [9]. Guarracino
and Mallardo applied a singular perturbation technique to the elastic
deflection of submerged pipelines [10]. Wang Lizhong et al. considered
the seabed as an elastic foundation. Based on the catenary theory and
569
linear beam theory, segmented models were used to solve the S-shaped
and J-shaped laying problems [11, 12]. Szczotka presented the Rigid Finite
Element Method (RFEM) to study the laying of J-pipes and compared it
with finite element ANSYS simulation results [13]. Based on the lumped
mass method, the riser was divided into a series of straight massless line
segments with a node at each end [14]. At present, scholars use ORCAFLEX
commercial software to conduct a large number of numerical simulations
on the dynamic response of offshore engineering anchor chains and risers
of different configurations and different pipeline laying methods [15–21].
Previous studies have focused on the mechanical performance of rigid
pipes during laying, while few have studied the dynamic behavior of glass
fibre reinforced pipelines. In this paper, through the finite element numer-
ical simulation method, the dynamic response characteristics of glass fibre
reinforced pipeline laying under different wave conditions were system-
atically studied. The influence of the lay angle and submerged weights on
the dynamic response of pipeline laying was discussed. Finally, we sum-
marized the factors which affect the pipeline installation stability, and put
forward suggestions for practical engineering.
29.2 Theoretical Analysis

29.2.1 Wave Load
When calculating the wave load of a slender structure with a smaller size
than the wavelength, the Morison equation is widely used in the engineer-
ing field. This theory assumes that the existence of the cylinder has no
obvious influence on the motion of the wave, and the effect of the wave on
the cylinder is mainly an inertial term and a drag term.
The inertial force of the cylinder needs to be multiplied by an iner-
tia coefficient, CM. The unit inertial force fI acting at the depth z can be
expressed as:
π D 2 ∂u (29.1)
f I = CM ρ
4 ∂t
The drag force is related to the velocity of the fluid particle. The drag
force on the cylinder per unit height at the depth z can be expressed by the
drag coefficient CD as:
Glass Fibre Pipelines During the Laying Process 571
ρD
f D = CD µ |µ| (29.2)
2
The speed of the current changes slowly with time. Usually, the current
is simplified as a steady current, and it is considered that the force of the
current on the cylinder is only a drag force. Applying Morrison’s equation,
the current force f per unit height of the riser at depth z is:
ρD 2 (29.3)
f = CD µz
2
where ρ is the density of seawater, and CD is the drag coefficient.

The coupling effect of waves and currents is very complicated, and drag
forces cannot be simply superimposed linearly. The current engineering
design is calculated according to the following formula. Assuming that the
velocity vectors of wave and current are v and vc, respectively, the angle
between the direction of the flow and wave propagation is ψ:
1
FD = CD ρD(v + vc )|v + vc | (29.4)
2
The components are:
1
FDx = CD ρD(v x + vc cosψ )|v + vc | (29.5)
2
1
FDy = CD ρDvc sinψ |v + vc| (29.6)
2
The wave force of the cylinder considering the influence of the current is:
1 π D2
dFx = ρCD D(v + vc cosψ )|v + vc | ds + ρC M v x ds (29.7)
2 4
1
dFy = ρCD Dvc sinψ |v + vc | ds (29.8)
2
29.2.2 Motion Response of the Vessel

As the top boundary condition of pipeline laying in the overall analysis,
the motion response of the vessel is a crucial external load in the design of
submarine pipelines. Under the action of wind, waves, and current load,
the vessel produces dynamic motion responses of six degrees of freedom,
namely Surge, Sway, Heave, Roll, Pitch, and Yaw.
The roll, pitch, and heave motions have a restoring force or moment due
to the action of the hydrostatic force. They have a stable static equilibrium
position. After the structure deviates from the equilibrium position under
the action of external interference, it can return to the balance point if
the external force disappears. However, the three motions of surge, sway
and yaw have no restoring force or moment. The vessel will not automati-
cally return to the initial equilibrium point. Floating structures must have a
mooring system or dynamic positioning system to maintain their position.
Figure 29.1 shows the degrees of freedom of floating structures.
The motion response of the floating structure mainly includes the
first-order high frequency (wave frequency) response and the second-order
low frequency (wave difference frequency) response. The wave frequency
response is mainly determined by the nature of the floating structure and
has nothing to do with the movement of the riser. It can be calculated by
the response amplitude operator (RAO) of the floating structure.
Vessel Axis Definition
BOW BOW
PORT PORT
STARBOARD
STERN STARBOARD
STERN
HEAVE
YAW SURGE
SWAY
ROLL
PITCH
Figure 29.1 Degrees of freedom of floating structures.

x = R · a · cos(ωt − φ) (29.9)
where x is the displacement response of the floating body; a is the ampli-

tude of the wave; ω is the frequency of the wave; R is the RAO response;
and φ is the phase.
29.2.3 Dynamic Numerical Solution

The Newmark integral can be regarded as a generalization of a linear acceler-
ation method. The equation of motion adopted by Newmark is expressed as:
Mr + Cr + Kr = F (29.10)
where M is the mass matrix of each element, C is the damping matrix, K is

the stiffness matrix, and F is the load matrix of each element. Suppose that
the displacement, velocity, and acceleration at time t = 0 are known. In the
Newmark method, there are:
1 1  1 
rt +∆t = (r − rt ) −
2 t +∆t
rt −  − 1 rt (29.11)
β ∆t β ∆t  2β 
1 
rt +∆t = rt + ∆ trt +  − β  ∆ t 2rt + β ∆ t 2rt +∆t (29.12)
2 
where the coefficient γ provides the weight of the linear variation between
the effect of the initial and final accelerations on the change in veloc-
ity. Similarly, the coefficient β provides the weight of the contribution of
these initial and final accelerations to the change in displacement. In the
Newmark method, γ = 1/4 and β = 1/2, which is an average acceleration
method. It is solved from the above formula:
1 1  1 
rt +∆t = (r − rt ) −
2 t +∆t
rt −  − 1 rt (29.13)
β ∆t β ∆t  2β 
rt +∆t = rt + (1 − γ ) ∆ trt + γ ∆ trt +∆t (29.14)

Replace the above two equations into the equation of motion at time
t + Δt:
Mrt +∆t + Crt +∆t + Krt +∆t = Ft +∆t (29.15)
It can be further obtained:
 1 γ   1   γ  
 M + C + K  rt + ∆t = Ft + ∆t +   − 1 M +  − 1 C ∆ t  rt
 β∆t
2
β∆t    2β   2β  
 1 γ    1 γ 
+ M +  − 1 C  rt +  M+ C  rt
 β ∆ t  β    β∆t
2
β ∆t 
(29.16)
From the above formula, rt + Δt is obtained. rt + t and rt + t are calculated
by Equations (29.18) and (29.19).
The Newmark method is an implicit integration method. When γ≥1/2
and β≥1/4(1/2 + γ)2, the algorithm converges unconditionally. The calcu-
lation steps of the Newmark method can be summarized as follows:
Step 1: Initial calculation

List stiffness matrix K, mass matrix M and damping matrix C.
The r0 is calculated from the initial conditions r0 and r0
r0 = M −1 ( F0 − Cr0 − Kr0 ) (29.17)
Select the time step and parameters to calculate the integral

constant:
2
1 1 1
γ ≥ , β ≥ γ + 
2 4 2
1 γ 1
A1 = , A2 = , A3 =
β ∆t 2
β ∆t β ∆t
1  γ  γ 
A4 = − 1, A5 =  − 1 ∆ t , A6 =  − 1
2γ  2β  β 
(29.18)
Calculate the effective stiffness matrix Keff.

Keff = K + A1M + A2C (29.19)
Step 2: Calculate the effective load vector Feff at time t + Δt
Fefft +∆t = Ft +∆t + [ A1rt + A3rt + A4rt ]M + [ A2rt + A6rt + A5rt ]C

(29.20)
Solve the displacement at time t + Δt:
rt+Δt = Keff −1Fefft+Δt (29.21)
Calculate the acceleration and speed at t + Δt:
1 1  1 
rt +∆t = (r − r∆t ) −
2 t +∆t
rt −  − 1 rt (29.22)
β ∆t β ∆t  2β 
rt +∆t = rt + (1 − γ ) ∆ trt + γ ∆ trt +∆t (29.23)
The displacement, velocity, and acceleration values at time t + Δt are

obtained as the initial state of the next time step, and the cycle continues
for the duration.
29.3 Static Analysis

An analysis of the laying of 4 inch glass fibre reinforced pipeline was per-
formed. The pipe parameters of the glass fibre reinforced pipeline are given
in Table 29.1.
According to the catenary theory, it is first necessary to assume the ini-
tial horizontal tension To1 at the touch down point. The initial catenary
curve coordinates are calculated by Equation (29.5). According to the top
tension T1, we calculated the new horizontal tension To2, then compared
the difference between To2 and To1 to determine whether to continue iter-
ating. Finally, a stable horizontal tension Ton was obtained to control the
initial shape of the segment from the top of the pipe to the touch point.
According to the data in Table 29.1, the results obtained by the above
method, to lay the flexible pipe at a water depth of 150 m and a departure
angle of 80°, are shown in Table 29.2.
Figure 29.2 shows the shape of the pipeline from the inclined slide of
the pipe-laying ship to the contact point, under the conditions of a water
Table 29.1 Pipe parameters.

Parameter Unit Value
Outer diameter m 0.125
Wall thickness m 0.0138
Outer layer thickness m 0.003
Inner layer thickness m 0.0063
Reinforced layer thickness m 0.0045
Reinforcement layer tensile modulus MPa 30,000
Unit weight of empty pipe kg/m 4.8
Unit weigh of pipe filled with water kg/m 12.44
Table 29.2 Static analysis of flexible laying.

Parameter Symbol Unit Value
Horizontal projection length S m 76.8
Suspended span length L m 178.8
Horizontal tension Tr kN 3.54
Top tension Ttop kN 20.39
depth of 150 m and a departure angle of 80°. This shape is used as the initial
layout of the suspension section in the flexible laying analysis.
Figure 29.3 shows the shape of the pipeline entering the water from the
inclined slide of the pipe-laying vessel to the contact point under the con-
ditions of a water depth of 150 m and lay angles of 50, 60, 70, and 80°. This
overhang length was used as the initial layout for the dynamic analysis of
flexible laying.
Table 29.3 shows the total length, static horizontal tension, and top ten-
sion of the pipeline at different lay angles.
It can be seen from the static analysis that as the lay angle increases, the
horizontal tension and the top tension decrease. The second-order poly-
nomial fitting results are shown in Figure 29.4, through the polynomial
fitting. When the lay angle is small, the top tension is more sensitive to
changes in the lay angle. According to the maximum allowable laying ten-
sion of the flexible pipe, considering the safety factor SF = 2, a lay angle of
50° leads to a top tension greater than the allowable tension of Ta = 85kN.
150
100
Vertical position(m)
50
0
0 20 40 60 80
Horizontal position(m)
Figure 29.2 The static state of the pipeline under a water depth of 150 m.
150
V er t i c a l p o s i t i o n ( m )
100
lay angle 80°

50 lay angle 70°
lay angle 70°
lay angle 50°
0
0 50 100 150 200 250 300
Horizontal position(m)
Figure 29.3 The static state of pipelines at different lay angles.
In the actual laying, it is necessary to select a larger lay angle. However, the
larger lay angle will lead to a smaller bending radius of the pipe contact
section, which may lead to bending failure under the action of the hull and
waves. Further dynamic analysis of flexible laying is needed to investigate
the reliability of flexible laying.
Table 29.3 Top tension at different lay angles.

Lay angle Total length (m) Horizontal tension (kN) Top tension (kN)
50° 321.7 30.32 47.16
60° 259.8 16.85 33.69
70° 214.2 8.76 25.6
80° 178.7 3.51 20.39
50
Data point
45 Fitting curve
40
T o p ten s i o n ( k N )
35
30
25
20
1 2 3 4
Different lay angles
Figure 29.4 Top tension at different lay angles.
Table 29.4 Top tension at different water depths.

Horizontal Top tension
Water depth (m) Total length (m) tension (kN) (kN)
80 95.3 1.89 10.88
100 119.2 2.36 13.60
120 143.0 2.83 16.31
140 166.8 3.30 19.03
Data point
20 Fitting curve
18
Top tension (kN)
16
14
12
10
1 2 3 4
Different water depth
Figure 29.5 Top tension at different water depths.
Table 29.4 shows the static horizontal tension and top tension values
at an 80° lay angle and 80, 100, 120, and 140 m water depths, respectively.
According to Table 29.4, the top tension increases linearly with the
increase in the water depth, and the linear fitting results are shown in
Figure 29.5. Therefore, based on the maximum allowable tension of the
pipeline, the applicable water depth can be calculated.
29.4 Dynamic Characteristic Analysis

The water depth of flexible pipe installation ranges from 100 to 150 m. The
wave environmental parameters are shown in Table 29.5, and the ocean
current parameters are given in Table 29.6.
Considering the waves and currents acting on the vessel in the same
direction, the load cases were defined according to the wave direction
angle. Wave direction distribution is shown in Figure 29.6.
29.4.1 Influence of the Wave Direction

Based on the above theory, a dynamic analysis of a lay angle of 70° at a
water depth of 150 m was performed. Results of the maximum tension
at the top, minimum tension in the touch zone, and minimum bending
radius for different wave directions were obtained, as shown in Table 29.7.
We concluded that the top tension is at the maximum when the wave
direction is a stern sea, and the top maximum tension is relatively small
Table 29.5 Wave environmental parameters.

Return periods Significant wave Highest wave Spectrum Peak
(year) height (m) height (m) Periods (s)
1 4.2 8.1 12.1
10 5.05 9.4 11.7
100 5.9 10.7 11.3
Table 29.6 Ocean current parameters.

Return periods
(year) Ocean surface Middle depth Seabed
1 0.85 0.75 0.5
10 1.2 0.97 0.60
100 1.47 1.18 0.72
Quarter sea Beam sea Bow sea
Stem sea Head sea
Figure 29.6 Wave direction distribution.
Table 29.7 Dynamic response results in different wave directions.

Quarter
Stern sea sea Beam sea Bow sea Head sea
Maximum 40.25 37.89 34.61 34.35 34.02
tension at the
top (kN)
Minimum 3.59 4.78 4.89 1.21 −1.18
tension at
TDP (kN)
Minimum 19.61 32.79 40.77 12.33 4.54
bending radius
ar TDP (m)
when the wave direction is a bow sea. The statistical results of the last three
wave cycles were taken, and the change value of the maximum tension
along the flexible length coordinates is given in Figure 29.7. It can be seen
that the maximum tension value at the top of a stern sea is 40.25 kN, which
meets the requirements of the maximum allowable tension. The maximum
tension along the pipe gradually decreases. Because the first 25 m of the
hose are on the incline slide, there is a curve turning point when the pipe
detaches from the incline slide. When the pipe touches the ground, it is
subjected to axial and normal soil friction, and the tension along the pipe
decreases gradually.
Figure 29.8 gives the curve of minimum tension change along the pipe
for different wave directions during lay-up. The variation of tension at the
top is similar in the three conditions, stern sea, quarter sea, and beam sea,
45
Maximum effective tension(kN)
40
35
30
35
20
15
10
5
0
0 50 100 150 200 250 300
Arc length(m)
Stem sea Quarter sea Beam sea Bow sea Head sea
Figure 29.7 Maximum tension at a lay angle of 70°.
30
25
20
15
10
5
0
0 50 100 150 200 250 300
–5
Arc length(m)
Figure 29.8 Minimum tension at a lay angle of 70°.

Minimum bending radius(m) 1.00E+08
1.00E+06
1.00E+04
1.00E+02
1.00E+00
0 50 100 150 200 250
Arc length(m)
Figure 29.9 Minimum bending radius at a lay angle of 70°.
and the location of the touchdown point changes due to hull lift and sink-
ing. The minimum tension in the touchdown area is negative in the head
sea, which will result in a possible local buckling of the pipe under pressure.
Figure 29.9 shows the curve of the minimum bend radius along the pipe
for different wave directions. It can be seen that a smaller bend radius occurs
at the top of the pipe at the point where it breaks away from the inclined slip.
In the vicinity of the pipeline touchdown, the bending radius of the head sea
and bow sea is smaller than under other conditions. The minimum bending
radius is 4.54 m in the head sea, and the minimum value of the bending
radius occurs at the position where the negative tension occurs.
29.4.2 Influencing of Different Lay Angle

During the laying process, the lay angle can be controlled to improve the
dynamic response. Two conditions with lay angles of 70° and 80° were
compared at a water depth of 150 m. The results of the maximum tension
at the top, minimum tension in the touch zone, and minimum bending
radius for different lay angles are shown in Table 29.8.
Figures 29.10, 29.11, and 29.12 give the distribution curves along the
pipe for the maximum tension, minimum tension, and minimum bending
radius. A lay angle of 80° will reduce the tension of the pipeline. The maxi-
mum top tension is 32.87 kN, which is significantly lower than the tension
of 40.25 kN at a lay of 70°. Compared with a lay angle of 70°, the minimum
tension under the head sea condition is –1.28 kN, a relative increase of
8.5% at the touch down point. The minimum bending radius correspond-
ing to this point is 3.11 m. 32% less than the minimum bending radius of
4.54 m at a lay angle of 70°.
Table 29.8 Dynamic response results with a lay angle of 80°.

Stern Quarter Beam Bow Head
sea sea sea sea sea
Maximum tension at 32.87 31.02 28.33 27.86 27.48
the top (kN)
Minimum tension at 1.30 1.78 1.01 –0.2 –1.25
TDP (kN)
Minimum bending 11.68 15.45 16.74 5.77 3.11
radius ar TDP (m)
35
30
25
20
15
10
5
0
0 50 100 150 200 250 300
Arc length(m)
Figure 29.10 Maximum tension at the lay angle of 80°.
25
Minimum effective tension(kN)
20
15
10
0
0 50 100 150 200 250 300
–5
Arc length(m)
Figure 29.11 Minimum tension at the lay angle of 80°.

1.00E+12
Maximum bending radius(m)
1.00E+10
1.00E+08
1.00E+06
1.00E+04
1.00E+02
1.00E+00
0 50 100 150 200 250
Arc length(m)
Figure 29.12 Minimum bending radius at a lay angle of 80°.
29.4.3 Influencing Submerged Weight

The submerged weight of the pipeline can be adjusted and controlled by water
injection and arrangement of counterweights. The change in submerged
weight has an impact on the stability of pipeline laying. The sensitivity analy-
sis of the submerged weight of the pipeline was carried out for a water depth
of 120 m and the lay angle of 70°. The influence of submerged weight on
the dynamic response of pipeline laying was investigated by changing the
wet weight by –20, –10, %, +10, and +20%. The results of the maximum top
tension, the minimum tension in the touch down point, and the minimum
bending radius under different load conditions are shown in Table 29.9.
Figure 29.13 shows the relationship between the change rate of the
minimum bending radius and submerged weight under different wave
directions at the touch down point. For stern sea conditions, the linear
relationship between the rate of change in the minimum bending radius
and submerged weight is relatively strong; while the change in the sub-
merged weight in a beam sea has a relatively small effect on the minimum
bending radius. In the direction of the head sea, with a two performance
quadratic polynomial relationship, the change in the submerged weight
has the greatest influence on the minimum bending radius.
29.5 Conclusions
This paper discusses the dynamic response of glass fibre hose laying under
different wave directions. The effects and sensitivity analysis of the lay angle
Table 29.9 Sensitivity analysis results of different submerged weights.

Change the wet weight
Stern sea −20% −10% 0% +10% +20%
Maximum tension (kN) 35.03 37.66 40.33 42.99 45.65
Change rate relative to 0% −13% −7% 0% 7% 13%
Minimum tension at TDP (kN) 2.14 2.88 3.61 4.36 5.10
Minimum bending radius at 13.82 16.38 19.74 22.74 25.35
TDP (m)
Beam sea
Minimum tension at TDP (kN) 3.58 4.23 4.91 5.57 6.21
TDP (m)
Head sea
Minimum tension at TDP (kN) −2.16 −1.72 –1.16 –0.56 0.07
TDP (m)
0.6
Stern sea
Beam sea
Head sea
0.4
Fitting curve of stern sea
Different submerged weights
Fitting curve of beam sea

Fitting curve of head sea
0.2
0.0
-0.2
-0.4
0 2 4 6
Change rate relative to 0%
Figure 29.13 The relationship between the minimum bending radius and the submerged
weight.
and submerged weight are discussed. Reducing the laying angle properly
will avoid negative tension at the contact point and increase the minimum
bending radius. Adding the minimum bending radius will reduce the pos-
sibility of over bending failure in the process of laying. However, reducing
the laying angle will significantly increase the maximum tension at the top
of the pipe, which may cause it to exceed the maximum allowable tension
of the hose and cause tensile failure. Therefore, the laying angle should be
adjusted dynamically according to the wave monitoring data, to ensure the
safety of hose laying.
For the stern sea direction, the linear relationship between the rate of
change in the minimum bending radius and submerged weight is strong; in
the beam sea direction, the influence of the submerged weight on the mini-
mum bending radius is relatively small. Meanwhile, in the head sea direction,
a quadratic polynomial relationship is manifested, and the influence of the
submerged weight on the minimum bending radius is the most significant.
References
1. Amaechi, C.V., Gillett, N., Odijie, A.C., Hou, X., Ye, J., Composite risers for
deep waters using a numerical modelling approach, Composite Structures,
210, 486–499, 2019, doi:10.1016/j.compstruct.2018.11.057.
2. Rafiee, R., On the mechanical performance of glass-fibre-reinforced ther-

mosetting-resin pipes: A review, Composite Structures, 143, 151–164, 2016,
doi:10.1016/j.compstruct.2016.02.037.
3. Hollaway, L.C., A review of the present and future utilisation of FRP com-
posites in the civil infrastructure with reference to their important in-ser-
vice properties, Construction and Building Materials, 24, 2419–2445, 2010,
doi:10.1016/j.conbuildmat.2010.04.062.
4. Williams, J.G.,Sas-Jaworsky, A., Composite Spoolable Pipe Development,
Advancements, and Limitations, In Proceedings of the Offshore Technology
Conference; Offshore Technology Conference: Houston, Texas, 2000.
5. Gong, S., Zhang, T., Wang, X., Liu, C., Numerical simulation on dynamic
behaviour of deepwater J-lay systems, Ocean Engineering, 196, 106771, 2020,
doi:10.1016/j.oceaneng.2019.106771.
6. Senthil, B., Selvam, R.P., Dynamic Analysis of a J-lay Pipeline, Procedia
Engineering, 116, 730–737, 2015, doi:10.1016/j.proeng.2015.08.358.
7. Gu, Y.N., Analysis of pipeline behaviours during laying operation, China
Ocean Eng, 3, 477–486, 1989.
8. Dixon, D.A., Rutledge, D.R., Stiffened catenary calculations in pipeline lay-
ing problem, 1968.
9. Gong, S.,Chen, K., Chen, Y., Jin, W., Li, Z., Zhao, D., Configuration analysis
of deepwater S-lay pipeline, China Ocean Engineering, 25, 519, 2011.
10. Guarracino, F., Mallardo, V., A refined analytical analysis of submerged pipe-
lines in seabed laying, Applied Ocean Research, 21, 281–293, 1999.
11. Wang, L., Yuan, F., Guo, Z., Li, L., Numerical analysis of pipeline in J-lay
problem, Journal of Zhejiang University-SCIENCE A, 11, 908–920, 2010.
12. Wang, L.-Z., Yuan, F., Guo, Z., Numerical analysis for pipeline installation
by S-lay method, In Proceedings of the International Conference on Offshore
Mechanics and Arctic Engineering; Vol. 49132, pp. 591–599, 2010.
13. Szczotka, M., A modification of the rigid finite element method and its appli-
cation to the J-lay problem, Acta Mechanica, 220, 183–198, 2011.
14. Sun, L., Qi, B., Global analysis of a flexible riser, J. Marine. Sci. Appl., 10,
478–484, 2011, doi:10.1007/s11804-011-1094-x.
15. Sarda, E.I., Dhanak, M.R., A USV-Based Automated Launch and Recovery
System for AUVs, Ieee Journal of Oceanic Engineering, 42, 37–55, 2017,
doi:10.1109/JOE.2016.2554679.
16. Elosta, H., Huang, S., Incecik, A., Dynamic response of steel catenary riser
using a seabed interaction under random loads, Ocean Engineering, 69,
34–43, 2013, doi:10.1016/j.oceaneng.2013.05.022.
17. Sethuraman, L., Venugopal, V., Hydrodynamic response of a stepped-spar
floating wind turbine: Numerical modelling and tank testing, Renewable
Energy, 52, 160–174, 2013, doi:10.1016/j.renene.2012.09.063.
18. Pestana, R.G., Roveri, F.E., Franciss, R., Ellwanger, G.B., Marine riser emer-
gency disconnection analysis using scalar elements for tensioner modelling,
Applied Ocean Research, 59, 83–92, 2016, doi:10.1016/j.apor.2016.05.004.
19. Gong, S., Xu, P., Bao, S., Zhong, W., He, N., Yan, H., Numerical modelling
on dynamic behaviour of deepwater S-lay pipeline, Ocean Engineering, 88,
393–408, 2014, doi:10.1016/j.oceaneng.2014.07.016.
20. Wang, J., Duan, M., He, T., Jing, C., Numerical solutions for nonlinear large
deformation behaviour of deepwater steel lazy-wave riser, Ships and Offshore
Structures, 9, 655–668, 201,4doi:10.1080/17445302.2013.868622.
21. Tan, Z., Quiggin, P., Sheldrake, T., Time Domain Simulation of the 3D
Bending Hysteresis Behavior of an Unbonded Flexible Riser, Journal of
Offshore Mechanics and Arctic Engineering-Transactions of the Asme, 131,
031301, 2009, doi:10.1115/1.3058698.
Index
Aging, 426 Collapse pressure, 19, 25, 27–29, 33,

Analytical modeling, 143, 501, 529 85, 97, 98, 150, 154, 387
Analytical solution, 57, 409, 415, 416, Combined bending and external
419, 427, 429, 436, 457, 465, 476, pressure, 144
481, 496, 497 Combined external pressure, 402, 474
Ancillaries, 157 Combined load, 119, 124, 302, 361,
Ancillary, 222, 531, 547 508, 550
Annulus venting system, 153 Combined tension, 116, 267, 294, 373,
API 17B, 60, 64, 84, 85 388, 403, 419, 474
API 17K, 86 Component, 8, 10, 39, 84, 87, 97, 103,
Axial stiffness, 39, 103, 147, 155, 215, 108, 133, 152, 222, 259, 267, 299,
217, 225, 250, 270, 280, 282, 296, 307, 320, 337, 352, 363–364,
324, 411 369–372, 377, 382, 425, 440, 445,
Axial tension, 89, 164, 167–169, 192, 528, 535, 558
296, 366, 431, 434, 441, 514 Confined collapse, 402
Axisymmetric loads, 10, 80, 84, 94, Critical curvature, 104, 109, 111–114,
101, 103, 427, 446, 514 215, 221, 303, 304, 313, 554
Cross-section design, 10, 77, 85, 86,
Bending stiffness, 3, 10, 11, 18, 27, 103, 317–319, 326, 327, 375, 377,
104, 108, 109, 114, 115, 147, 155, 379–383, 417, 503, 505, 507, 509,
213–215, 220–224, 230, 234–235, 511
238, 240, 242, 244, 245, 252, 270,
279, 280, 282, 296, 298, 299, 304, Design criteria, 9, 10, 154, 156, 302,
309, 341, 359, 362–365, 376, 382, 506–508, 510, 511
383, 451, 469, 470, 549–551, 554 Diameter-to-thickness, 457
Bonded flexible, 119, 143, 148, 158,
247, 387, 405, 408, 421, 422, 434, Elastomer, 147
450, 476, 545 Ellipticity, 296
Burst pressures, 407, 416 End fittings, 341
Environmental loads, 161, 187, 379,
Carcass layer, 8, 19, 22, 86, 98, 117 384
Case study, 55, 83–85, 87, 89, 91, 93, Equivalent stiffness, 24, 99
95, 97, 99–101, 215, 225, 230, 231, External pressure, 8, 11, 12, 15, 19, 22,
234, 242, 245 24, 25, 27, 28, 32, 33, 38, 57, 58,
Circumferential stiffness, 462 64, 67, 68, 75, 77, 79, 80, 81, 85,
589
590 Index
86, 89, 91, 92, 94, 96, 110, 119, 453, 457, 460–463, 466, 472, 475,
143, 144, 217, 302, 323, 369, 370, 477, 494, 503, 506
387–390, 394–399, 401–403, 419,
447, 449, 450, 474, 501, 503–505, Initial imperfection, 16, 19, 21, 29, 64,
510, 511, 514, 534, 540, 541, 544 465
Initial ovality, 16, 470, 473
Failure criteria, 368–369, 510 Internal pressure, 36, 38, 45–50, 53–55,
Failure mode, 9, 12, 84, 85, 98–100, 85, 87–91, 94, 96–98, 213, 214,
118, 149, 154, 296, 302, 426, 427, 217, 267, 294, 296, 302, 329,
513, 515, 530 370–372, 405–410, 417–419, 447,
Fatigue damage, 6, 153, 222–223, 231, 476, 501, 508, 514, 534
233, 245, 296, 299, 305, 307, 308, Isotropic hardening, 22, 391, 463
313, 314, 318, 320, 330–333, 342,
346, 352–353, 355, 513–515, 529, J2 flow theories, 391
530, 533, 535
Finite element analysis, 83, 116, 119, Laminated-plate theory, 421, 422
120, 267, 294, 332, 361, 373, 388, Lay angle, 8, 18, 19, 31, 32, 60, 64, 87,
395, 397, 422, 465–476, 494, 501, 108, 151, 329, 340, 341, 346, 350
514, 529–530 Limit state, 5, 31
Finite element model, 4, 25, 43, 46, 55, Load case, 12, 94, 96, 104, 153–155,
120, 153, 154, 174, 187, 214, 297, 216, 300, 330, 339, 369, 554
329, 361, 364, 395–397, 405, 434,
436, 465, 494, 514, 515, 519 Material properties, 20, 22, 30, 66, 368,
Flexibility matrix, 461, 485, 487 380, 406, 407, 425, 426, 453, 465,
Flexible composite pipe, 519 477, 497, 515, 519, 520, 552, 556
Four-point bending, 123, 451, 470 Material selection, 85, 189
Four-point bending test, 451 Mechanical behavior, 5, 24, 57, 61, 63,
80, 89, 91, 92, 99, 100, 117, 119,
Gauss integration method, 465 120, 134, 139, 142, 166, 167, 174,
Geometry properties, 123 176, 187, 188, 195, 212, 240, 359,
Global analysis, 6, 84, 126, 142, 189, 360, 362, 396, 397, 402, 406, 417,
196, 212, 214, 216, 224, 236, 298, 419, 422, 442, 445, 446, 449, 465,
300, 302, 309–311, 314, 339, 343, 471, 472, 475, 477, 494, 500
514–515, 518–520, 528 Mechanical properties, 5, 39, 59, 80,
84, 85, 125, 142, 279, 369, 372,
Helical layer, 37, 91, 92, 104–106, 108, 375, 378, 380–382, 411, 420–422,
109, 114, 115, 245, 514, 515 446, 476, 483, 501, 519
Helical wire, 103–105, 107–114, 116, Metallic strip, 12, 57, 77, 80, 85, 117,
329, 334, 359–361 143, 144, 405, 446
High density polyethylene (HDPE), Minimum bending radius, 324, 369,
31, 32, 57, 58, 63, 64, 69, 70, 73, 504, 510, 547
77, 84, 95, 117, 119–123, 150, 392,
396, 406, 407, 412, 424–428, 434, Nonlinear analysis, 80, 98, 162, 243,
436, 437, 439–445, 449, 450, 452, 244, 372, 422, 446, 501
Index 591
Nonlinear behavior, 120, 174 Tensile test, 120, 122, 217, 424–426,
No-slip, 104, 112, 113, 115, 339 452, 507, 508
Tension and internal pressure, 267,
Ovalization instability, 471 294, 296, 419
Tension load, 104, 231, 440
Parametric study, 77, 134, 187, 345, Thick tubes, 403
422, 442, 445 Thin tubes, 387, 403
Pipe manufactures, 535 Time history, 123, 231–236, 238, 240,
Plastic pipe reinforced, 450, 474, 501 252, 297, 305, 311–313, 343–348,
Plastic strain, 330, 331, 333, 384 355, 515–522, 525, 526
Polymer sheath, 148–154, 280, 361, Torsion, 5, 9–11, 33, 38, 44, 55, 80,
375, 376 84, 92, 100, 101, 107, 118, 119,
Pure bending, 104, 109, 123, 363, 143, 151, 269, 270, 277, 279, 280,
449–451, 454, 457, 471, 473 282, 294, 298, 302, 327, 359–368,
372, 373, 376, 383, 415, 422, 429,
Reduction factor, 451, 462, 469, 472, 430, 434, 436, 446, 473, 475–479,
503, 507 481, 483, 485, 487, 489, 491–493,
Reinforced thermoplastic pipe, 15, 24, 495–501, 514, 529
33, 34, 51, 53–55, 80, 94, 143, 212,
388, 402, 419, 446, 474, 476, 501 Unbonded flexible pipe, 8, 10, 55, 79,
Reliabiliy, 4, 54, 64, 85, 86, 117, 143, 80, 83, 84, 91, 101, 103–105, 107,
175, 198, 227, 296, 307, 319, 437, 109, 111, 113–116, 147–149, 158,
445, 477, 496 213, 214, 216, 217, 220, 222, 223,
Representative volume unit, 391, 392, 225, 238, 241, 245, 247, 314, 387,
459, 460, 483 405, 475, 514, 529, 545
Uncertainties, 10, 356
Simple method, 84
S-N, 296, 305, 306, 308, 313, 314, 320, Virtual work, 38, 90, 338, 360, 388,
334, 339, 343, 347, 357, 527, 528 394, 401, 437, 451, 464, 513
Steel reinforcement, 87
Steel strip, 9, 15, 19, 24, 30–34, 51–55, Winding angle, 11, 32, 41, 43, 51–53,
77, 79, 80, 94, 117, 120–122, 143, 58, 59, 61, 69, 74, 83–85, 135,
296, 387, 402, 419, 514 138, 280, 361, 362, 365, 405, 406,
Steel wire, 79, 87, 150, 151, 365, 368, 416–419, 424, 431, 434, 439, 442,
369, 375, 376, 381, 382, 402, 420, 443, 445, 447, 453, 478, 481, 485,
450, 474, 501, 568 498, 500
Also of Interest
Check out these other titles from Scrivener Publishing

Structural Dynamics, by Yong Bai and Zhao-Dong Xu, ISBN
9781119605607. Written by two experts across multiple disciplines,
this is the perfect reference on structural dynamics for veteran engi-
neers and introduction to the field for engineering students. NOW
AVAILABLE!
Flexible Pipes, by Qiang Bai, Yong Bai, and Weidong Ruan, ISBN
9781119041269. Written by one of the most well-respected teams of
scientists in the area of pipelines, this revolutionary approach offers
the engineer working in the energy industry the theory, analysis, and
practical applications for applying new materials and modeling to
the design and effective use of flexible pipes. NOW AVAILABLE!
Corrosion Engineering, by Volkan Cicek, ISBN 9781118720899. This
timely volume covers the state-of-the art in corrosion engineering
today, for use in industrial applications and as a textbook. NOW
AVAILABLE!
Cathodic Protection: Industrial Solutions for Protecting Against
Corrosion, by Volkan Cicek, ISBN 9781118290408. The most up-to-
date and timely treatment on cathodic protection available. NOW
AVAILABLE!
Corrosion Chemistry, by Volkan Cicek and Bayan Al-Numan, ISBN
9780470943076. The causes and results of corrosion in industrial
settings are some of the most important and difficult problems that
engineers and scientists face on a daily basis. Coming up with solu-
tions, or not, is often the difference between success and failure, and
can have severe consequences. This timely volume covers the state-
of-the art in corrosion chemistry today, for use in industrial applica-
tions and as a textbook. NOW AVAILABLE!
Fracking 2nd Edition: Further Investigations into the Environmen
tal Considerations and Operations of Hydraulic Fracturing, by
Michael D. Holloway, ISBN 9781119363422. This revised, expanded
edition of an instant classic explores the history, techniques, and
materials used in the practice of induced hydraulic fracturing,
while examining the environmental and economic impact. NOW
AVAILABLE!
Fundamentals of Drilling Engineering, by M. Enamul Hossain, ISBN
9781119083566. The most comprehensive learning tool for drilling
engineering ever conceived, this must-have volume is essential for
any student, veteran engineer, or any engineer transitioning into the
energy industry. NOW AVAILABLE!

Flexíveis

Uploaded by

Document Informationclick to expand document information

Document Informationclick to expand document information

Copyright:

Available Formats

Flexíveis

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Flexíveis

Uploaded by

Copyright:

Available Formats

Deepwater Flexible

Risers and Pipelines

Wiley Global Headquarters

Limit of Liability/Disclaimer of Warranty

Library of Congress Cataloging-in-Publication Data

Cover design by Kris Hackerott

Printed in the USA

7 Coiling of Flexible Pipes 117

Part 2: Riser Engineering 145

9 Lazy-Wave Static Analysis 159

11 3D Rod Theory for Static and Dynamic Analysis 213

13.2.1 Establishment of Mathematical Model 269

15 Steel Tube Umbilical and Control Systems 317

16.2.2.1 Friction Stresses 344

18.2 Umbilicals Cross-Section Design Overview 377

Part 3: Fiber Glass Reinforced Deep

20 Burst Strength of Fiber Glass Reinforced Riser 405

20.4 Analytical Solution 409

21 Structural Analysis of Fiberglass Reinforced Bonded Flexible

22 Fiberglass Reinforced Flexible Pipes Under Bending 449

22.2.1 Experimental Facility 451

24 Cross-Section Design of Fiberglass Reinforced Riser 503

Part 4: Ancillary Equipments for Flexibles

26.9 Conclusions 544

Part II addresses global response of the riser systems in deepwater. We

Part IV summarizes experience we have for ancillary equipment design.

Prof. Yong Bai

1.1 Flexible Pipelines Overview

in comparison to axial tensile stiffness. Besides, the economic benefits,

1.2 Environmental Conditions

forces if considering the substantial pipe length in deep water, which is

(a) Free-hanging Catenary

(b) Steep-S (c) Lazy-S

(d) Steep Wave (e) Lazy Wave

Figure 1.1 Employment of flexible pipelines as risers.

1.3 Flexible Pipeline Geometry

• Internal polymeric sealing layer provides insulation from

They can be characterized in two main groups: the metal-based flexible

Figure 1.2 Unbonded composite-based flexible pipe.

Figure 1.3 Unbonded metal-based flexible pipe.

Other elements, such as anti-wear and bird-caging tapes, can be con-

1.4 Base Case-Failure Modes and Design Criteria

If unbonded flexible pipes are subjected to axisymmetric loads, the stiff-

Figure 1.4 Interlocked carcass.

Figure 1.5 Pressure armor cross-section profiles.

1.6 Project and Objectives

3. Garrett D L. Dynamic analysis of slender rods[J]. J. Energy Resour.

2.2 Theoretical Models

where radial displacements uR are considered to be small, is the circumfer-

M = pR(uR + uR1 cos 2ϑ) (2.2)

uR1 = δ1R (2.3)

The initial ovality is caused by manufacturing tolerances or bend-

d 2uR p[(uR + uR1 cos(2ϑ )]R 3

for which the solution results in

where pcr is the critical buckling pressure for a perfect ring.

Finally, the need of a reliable theoretical model suitable for practical

Figure 2.2 Carcass profile-principal outline.