Erochko Jeffrey A 201306 PHD Thesis PDF

Improvements to the Design and Use of Post-Tensioned
Self-Centering Energy-Dissipative (SCED) Braces
by
Jeffrey A. Erochko
A thesis submitted in conformity with the requirements

for the degree of Doctor of Philosophy,
Graduate Department of Civil Engineering,
University of Toronto
© Copyright by Jeffrey A. Erochko, 2013

Improvements to the Design and Use of Post-Tensioned
Self-Centering Energy-Dissipative (SCED) Braces
Jeffrey A. Erochko
Doctor of Philosophy,
Department of Civil Engineering,
University of Toronto
2013
ABSTRACT
The self-centering energy dissipative (SCED) brace is an innovative cross-bracing system that elimi-
nates residual building deformations after seismic events and prevents the progressive drifting that other
inelastic systems are prone to experience under long-duration ground motions. This research improves
upon the design and use of SCED braces through three large-scale experimental studies and an associated
numerical building model study. The first experimental study increased the strength capacity of SCED
braces and refined the design procedure through the design and testing of a new high-capacity full-scale
SCED brace. This brace exhibited full self-centering behaviour and did not show significant degradation
of response after multiple earthquake loadings. The second experimental study extended the elongation
capacity of SCED braces through the design and testing of a new telescoping SCED (T-SCED) brace that
provided self-centering behaviour over a deformation range that was two times the range that was achieved
by the original SCED bracing system. It exhibited full self-centering in a single storey full-scale frame that
was laterally deformed to 4% of its storey height. The third experimental study confirmed the dynamic
behaviour of a multi-storey SCED-frame in different seismic environments and confirmed the ability of
ii
computer models of differing complexity to accurately predict the seismic response. To achieve these goals,
a three-storey SCED-braced frame was designed, constructed, and tested on a shake table. Lastly, a numer-
ical six-storey SCED-braced building model was constructed. This model used realistic brace properties
that were determined using a new software tool that simulates the full detailed mechanics of SCED and T-
SCED braces. The building model showed that initial SCED brace stiffness does not have a significant
effect on SCED frame behaviour, that T-SCEDs generally perform better than traditional SCEDs, and
that the addition of viscous dampers in parallel with SCED braces can significantly reduce drifts and accel-
erations while only causing a small increase in the base shear.
iii
Acknowledgements
ACKNOWLEDGEMENTS
The author would first and foremost like to thank Professor Constantin Christopoulos, for his years of
personal guidance and support. Without him this work would not have been possible and without his
steady feedback this would be a far lesser work.
The author would also like to specially thank Professor Robert Tremblay who provided support and
valuable feedback during all of the phases of this work.
The author sincerely appreciates the time and feedback provided by Professors Evan Bentz, Oya Mer-
can, and Oh-Sung Kwon in the review of this work.
For the experimental work that was performed in the laboratory at Ecole Polytechnique, the author
would like to extend special thanks to Antoine Le Bec, who was vital to the timely completion of the tests
in Montreal and who also took the author to the hospital in the middle of the night, Martin Leclerc, whose
technical expertise and eagerness to help was largely responsible for the success of the shake table tests, and
Michael Montgomery who helped keep the author sane and also took him to the hospital one day. The
author would also like to acknowledge the help granted to him by the other members of the laboratory
staff at Ecole Polytechnique: Patrice Bélanger, Marc Charbonneau, David Ek, and Guillaume Cossette.
For the experimental work that was performed in the laboratory at the University of Toronto, the
author would like to extend special thanks to Viswanath Kammula and Nikolas Kyriakopoulos for their
dedicated assistance and would also like to sincerely thank the laboratory staff: Renzo Basset, Giovanni
Buzzeo, John MacDonald, Alan McClenaghan, Xiaoming Sun and Bryant Cook.
The author would like to acknowledge the support of his colleagues and friends, especially Michael
Montgomery, Michael Gray, Graeme Kennedy and Lydell Wiebe for their technical and moral support.
Finally, the author would like to extend a heartfelt thanks to his family for their support during this
doctoral work. Specifically, he would like to thank his wife Sarah for her constant support, his parents Rob-
ert and Caterina, his brother Andrew, his sister Krysten, his mother-in-law Jane, and his newborn daughter
Madeleine for providing him with an incentive to finish in a timely manner.
iv
Table of Contents
TABLE OF CONTENTS
Abstract .................................................................................................................ii
Acknowledgements.............................................................................................. iv
Table of Contents.................................................................................................. v
List of Tables........................................................................................................ xi
List of Figures..................................................................................................... xiv
List of Symbols.................................................................................................... xx
Chapter 1: Introduction ........................................................................................1
Chapter 2: Background and Motivation ................................................................5
2.1 Performance-Based Design – 5

2.2 Residual Drifts – 9
2.2.1 The Importance of Residual Drifts – 9
2.2.2 Residual Drifts in Performance-Based Design – 11
2.2.3 Reducing Residual Drifts – 13
2.3 Self-Centering Systems vs. Traditional Inelastic Systems – 13
2.4 Buckling-Restrained Braces (BRBs) – 15
2.5 Self-Centering Rocking Walls and Moment Frames – 16
2.5.1 Rocking Walls – 18
2.5.2 Self-Centering Moment Frames – 20
2.6 Previous Self-Centering Cross-Braces – 21
2.6.1 The Energy Dissipating Restraint (EDR) (Nims et al., 1993) – 21
2.6.2 Fluid Restoring Force / Damping Device (Tsopelas & Constantinou, 1994) – 23
2.6.3 Friction Spring Seismic Damper (Filiatrault et al., 2000) – 23
2.6.4 Self-Centering Friction Damping Brace (Zhu and Zhang, 2008) – 26
2.7 The Self-Centering Energy-Dissipative (SCED) Brace (Christopoulos et al., 2008) – 27
2.7.1 The SCED Brace Concept – 27
2.7.2 Prior Prototypes (Christopoulos et al., 2008) – 29
2.7.3 Dynamic Modelling – 31
v
Table of Contents
2.8 Impetus for the Research / Necessary Improvements – 31

2.9 Other Recently Developed Self-Centering Braces based on the SCED – 34
Chapter 3: Mechanics, Modelling and Design of the Original and New Telescoping
SCED (T-SCED) Braces ................................................................................35
3.1 The Components of a SCED Brace – 36

3.1.1 Axial Members – 38
3.1.2 Internal Friction Damper – 38
3.1.3 Optional External Friction Fuse – 44
3.1.4 Long-Term Behaviour of the Friction Devices – 46
3.1.5 Tendons and Tendon Anchors: Behaviour and Design Considerations – 47
3.1.6 End Plates – 53
3.1.7 Connections and Accommodation of Frame Rotation – 54
3.2 Original (O-SCED) Brace Mechanics – 55
3.2.1 Simplified Step-wise Hysteretic Response of the SCED Brace – 55
3.2.2 Full Step-wise Hysteretic Response of the SCED Brace – 61
3.3 Concept for a Telescoping (T-SCED) Brace – 71
3.4 The SCED Mechanics Simulator - a Detailed Simulation of SCED and T-SCED Brace
Hystereses – 78
3.4.1 SCED Mechanics Simulator Algorithmic Implementation – 79
3.4.2 User Interface – 83
3.4.3 Validation of the SCED Mechanics Simulator – 89
3.5 The Effect of Fabrication Length Tolerances – 90
3.5.1 The Effect of Fabrication Length Tolerances on an Original SCED Brace – 90
3.5.2 Estimating the Effective Initial Stiffness of an Original SCED Brace – 93
3.5.3 The Effective Initial Stiffness of a T-SCED Brace – 98
3.5.4 Rapid Estimation Method for Determining the Effective Initial Stiffness of
SCED and T-SCED Braces for Preliminary Design – 101
3.6 SCED Brace Modelling using Effective Hysteretic Parameters – 103
3.6.1 Ruaumoko Implementation (Carr, 2005a) – 103
3.6.2 SAP2000 Implementation (Le Bec et al., 2010) – 103
3.6.3 New OpenSees Implementation – 105
3.7 SCED-Braced Frame Design Considerations – 106
3.7.1 Reasonable Construction Tolerances and the Determination of the Effective
Initial Stiffness of SCED Braces – 106
3.7.2 Force-Based Design Considerations – 108
3.7.3 The Effect of Non-Structural Components – 111
3.7.4 The Ultimate Limit State of SCED Braces – 112
3.7.5 Sample SCED-Frame Design Method – 113
3.8 Conclusions – 115
vi
Table of Contents
Chapter 4: Design and Testing of a Prototype High-Capacity (HC-SCED) Brace ..

117
4.1 High-Capacity SCED (HC-SCED) Brace Design and Innovations – 117

4.2 High-Strength External Friction Fuse Subcomponent Design and Testing – 123
4.2.1 External Friction Fuse Subcomponent Specimen and Calibration – 123
4.2.2 Velocity-Dependance Tests – 127
4.2.3 Temperature Dependance Tests – 129
4.2.4 Higher Velocity Dynamic Tests – 130
4.2.5 Test Issues – 131
4.3 HC-SCED Tendon Pretensioning – 132
4.3.1 Assumed Tendon Properties – 132
4.3.2 Pretensioning Method – 133
4.4 HC-SCED Testing – 136
4.4.1 Load Frame Setup and Instrumentation – 137
4.4.2 SCED Brace Axial Force Calculation – 138
4.4.3 Internal Friction Damper Calibration – 146
4.4.4 Wind Response Simulation – 147
4.4.5 Simulation of SCED Brace Response and Alternate Calculation of HC-SCED
Axial Force – 149
4.4.6 Dynamic Earthquake Response - Design-Level Earthquakes – 152
4.4.7 Dynamic Earthquake Response - Maximum-Considered Earthquake – 154
4.4.8 Dynamic Protocol Tests – 156
4.5 External Friction Fuse Issues – 160
4.5.1 External Friction Fuse Initial Calibration and Conditioning – 160
4.5.2 Velocity Dependance of the External Friction Fuse – 161
4.6 Tendon Anchor Performance Issues – 164
4.6.1 Anchor Spikes Dislodged During Shipping – 164
4.6.2 Problems Encountered During the Tendon Tensioning Process – 165
4.6.3 Tendon Anchor Failure during the First Attempt at Tendon-Only
Characterization – 166
4.6.4 Step-Response Observed due to Tendon Debris – 167
4.6.5 Testing the Remaining Tendons from the First Set – 167
4.6.6 Failure of Tendon Anchor during Quasi-Static Test – 169
4.6.7 Simulation of Step Behaviour – 171
4.6.8 Ultimate Testing of Remaining Tendon Anchors – 174
4.6.9 Tendon Performance Summary – 174
Chapter 5: Design and Testing of a Prototype Telescoping (T-SCED) Brace ....177
5.1 Telescoping SCED (T-SCED) Design and Construction – 178
vii
Table of Contents
5.1.1 T-SCED Design – 178

5.1.2 Tendon Assembly – 181
5.1.3 Tendon Tensioning – 184
5.2 Test Frame Design – 185
5.3 Test Instrumentation – 188
5.4 Bare Test Frame Tests – 188
5.5 T-SCED Braced-Frame Tests – 190
5.5.1 T-SCED Brace Axial Force Calculation – 190
5.5.2 Brace Calibration – 198
5.5.3 T-SCED Brace Test Protocols – 199
5.5.4 Wind Response Simulation – 200
5.5.5 Deformation-Limited Dynamic Protocol Test – 201
5.5.6 Full-Deformation Dynamic Protocol Tests – 202
5.5.7 Dynamic Earthquake Response – 208
5.6 Lower Connection Rotational Stiffness and Retrofit – 210
5.7 Moving Part Interference Issues and Retrofits – 212
5.7.1 The Effect of Beam Connection Slip – 212
5.7.2 End Plate Slot Interference and Installation of End Plate Guides – 213
5.7.3 Internal Friction Damper Bolt Bearing at High Drifts – 215
5.7.4 The Effect of Friction Between the Inner and Outer End Plates – 216
5.8 Hybrid Testing Program (Kammula, 2012) – 220
5.9 Tendon Anchor Failure due to Repeated Loading Cycles – 220
Chapter 6: Shake Table Testing of a SCED Braced Frame ................................224
6.1 Design Structure – 224

6.1.1 Location and Prototype Design using ASCE 7-05 (Kim, 2009a) – 224
6.1.2 Scaling of Prototype for Design of Test Structure (Kim, 2009a) – 226
6.2 SCED Brace Design (Kim, 2009a) – 229
6.3 Test Frame Design and Setup – 229
6.4 Preliminary Brace Tests – 232
6.4.1 Test Setup – 233
6.4.2 SCED Brace Response Issues As-Shipped – 234
6.4.3 Load Frame Testing Program and Brace Recalibration – 238
6.4.4 Simulation of the SCED Brace Response using the SCED Mechanics Simulator
– 246
6.5 Shake Table Test Modeling – 249
6.5.1 SCED Brace Simplified Modeling – 249
6.5.2 OpenSees Test Frame Model – 250
6.5.3 OpenSEES Assumptions / Simplifications – 251
6.5.4 SAP2000 Test Frame Model – 252
6.6 Selection and Adaptation of Ground Motions – 253
viii
Table of Contents
6.7 Shake Table Test Results and Discussion – 254

6.7.1 SCED Brace Axial Force Calculation – 258
6.7.2 SCED Brace Activation Force in Shake Table Test vs. in Load Frame – 259
6.7.3 SCED Parameters used for Computer Modelling – 262
6.7.4 Overall Comparison of Experimental and Model Results – 263
6.7.5 Hysteretic Response, Drift Results, and Comparison to Model Predictions – 263
6.7.6 Acceleration Results and Comparison to Model Predictions – 266
6.7.7 Storey Shear Results and the Effect of Inherent Damping – 270
Chapter 7: SCED Frame Building Model Comparisons ....................................274
7.1 Six-Storey Building Design (Choi et al., 2008) – 275

7.2 Previous Studies into Combining Viscous Damping with Self-Centering Systems – 275
7.3 Building Modelling – 277
7.4 Design of SCED Braces – 279
7.4.1 Design of SCEDs with Viscous Dampers – 286
7.5 Building Periods and Pushover Analyses – 294
7.6 Building Damping – 296
7.7 Time History Ground Motions – 297
7.8 Time History Analysis Results – 298
7.8.1 Individual Record Response and Hystereses – 299
7.8.2 Viscously Damped SCED Modelling Comparison – 306
7.8.3 Response Summary – 307
7.8.4 Comparison of Time History Results to V-SCED Models with No Friction –
310
7.8.5 C-SCED versus Realistic Designed O-SCED – 311
7.8.6 Theoretical Stiffness versus Realistic Stiffness – 313
7.8.7 Realistic Designed O-SCED versus T-SCED – 316
7.8.8 T-SCED versus Viscously Damped V-SCED – 316
Chapter 8: Conclusions .....................................................................................324
8.1 Summary of the Research – 324

8.2 Primary Contributions of the Thesis – 327
8.3 Suggestions for Future Research – 329
8.3.1 Tendon and Tendon Anchor Behaviour – 329
8.3.2 Long-Term Behaviour of the Friction Devices – 330
8.3.3 The Effect of Member Length Construction Tolerances – 330
8.3.4 High-Capacity External Fuse Development – 330
8.3.5 Rayleigh Damping – 331
ix
Table of Contents
8.3.6 More Extensive Nonlinear Modelling – 331

8.4 Current State of SCED Brace Technology – 331
References .........................................................................................................333
Appendix A: SCED Mechanics Simulator Source Code ....................................343
Appendix B: HC-SCED Drawings ....................................................................357
Appendix C: Friction Subcomponent Test Drawings ........................................363
Appendix D: T-SCED Drawings.......................................................................369
Appendix E: Vertical Test Frame Drawings.......................................................386
Appendix F: Shake Table Test Prototype Design Report Excerpt (Kim, 2009a) 414
Appendix G: Shake Table SCED Drawings (Kim, 2009a).................................422
Appendix H: Shake Table Frame Drawings.......................................................428
x
List of Tables
LIST OF TABLES
Table 2.1: Vision 2000 Performance Based Design (adapted from SEAOC, 1995) – 7
Table 2.2: NEHRP/FEMA 356 Performance Based Design (adapted from ASCE, 2000) – 8
Table 3.1: Full Solution Step 0 - Initial Tendon Pretension – 64
Table 3.2: Full Solution Step 1T - First Activation (Tension) – 64
Table 3.3: Full Solution Step 1C - First Activation (Compression) – 65
Table 3.4: Full Solution Step 2T - Full Activation (Tension) – 66
Table 3.5: Full Solution Step 2C - Full Activation (Compression) – 67
Table 3.6: Full Solution Step 3T - Ultimate Force (Tension) – 68
Table 3.7: Full Solution Step 3C - Ultimate Force (Compression) – 68
Table 3.8: Full Solution Step 4T - First Reverse Activation (Tension) – 69
Table 3.9: Full Solution Step 4C - First Reverse Activation (Compression) – 69
Table 3.10: Full Solution Step 5T - Full Reverse Activation (Tension) – 70
Table 3.11: Full Solution Step 5C - Full Reverse Activation (Compression) – 71
Table 3.12: Full Solution Step 6T - First End Plate Re-Contact (Tension) – 72
Table 3.13: Full Solution Step 6C - First End Plate Re-Contact (Compression) – 73
Table 3.14: Full Solution Step 7T - Full End Plate Re-Contact (Tension) – 74
Table 3.15: Full Solution Step 7C - Full End Plate Re-Contact (Compression) – 75
Table 3.16: O-SCED Brace Designs Used to Check the Effective Initial Stiffness Calculation – 98
Table 3.17: Effective Initial Stiffness Sample Calculations – 99
Table 3.18: Initial Stiffness of Equivalent SCED Brace and Buckling-Restrained Brace Designs (kN/
mm) – 109
Table 4.1: HC-SCED Target Design Parameters – 119
Table 4.2: External Friction Fuse Subcomponent Testing Summary – 127
Table 4.3: External Friction Fuse Subcomponent Test Result Summary – 128
Table 4.4: Assumed Tendon Properties - 22mm Tendons – 133
Table 4.5: HC-SCED Revised Target Design Parameters – 137
Table 4.6: HC-SCED Brace Load Frame Testing Summary - First Set of Tendons – 138
Table 4.7: HC-SCED Brace Load Frame Test Result Summary - First Set of Tendons – 139
Table 4.8: HC-SCED Brace Load Frame Testing Summary - Second Set of Tendons – 140
Table 4.9: HC-SCED Brace Load Frame Test Result Summary - Second Set of Tendons – 142
Table 4.10: Mechanics Simulator Model Inputs – 150
xi
List of Tables
Table 4.11: Combined ASCE/AISC Deformation Protocol – 158

Table 4.12: Tendon Anchor Performance Comparison – 175
Table 5.1: T-SCED Target Design Parameters – 179
Table 5.2: Assumed Tendon Properties - 17mm Tendons – 185
Table 5.3: T-SCED Brace Load Frame Testing Summary – 191
Table 5.4: T-SCED Brace Load Frame Test Result Summary – 194
Table 5.5: Combined ASCE/AISC Deformation Protocol – 199
Table 5.6: Test DPE-2 Simulation Inputs – 205
Table 6.1: Prototype SCED Frame Properties (after Kim, 2009a) – 226
Table 6.2: Prototype Building Section Properties (after Kim, 2009a) – 227
Table 6.3: Scaling Factors for the Test Structure (after Kim, 2009a) – 228
Table 6.4: Test Frame SCED Properties (after Kim, 2009a) – 228
Table 6.5: Extra SCED Brace Load Frame Testing Summary – 239
Table 6.6: Extra SCED Brace Load Frame Test Result Summary – 240
Table 6.7: First to Third Storey SCED Brace Load Frame Testing Summary – 241
Table 6.8: First to Third Storey SCED Brace Load Frame Test Result Summary – 242
Table 6.9: Mechanics Simulator Model Inputs – 247
Table 6.10: Earthquake Record Properties and Scaling – 253
Table 6.11: Experimental Shake Table SCED Brace Response – 255
Table 6.12: Experimental Results vs. Model Results
Experimental Result (OpenSees Result with 1% Damping) [SAP2000 Result with 1% Damping] – 256
Table 6.13: Comparison of Shake Table Test and Load Frame Test SCED Response – 260
Table 7.1: Mechanics Simulator Model Inputs for Designed SCEDs – 281
Table 7.2: SCED Brace Property Summary – 283
Table 7.3: Complete SCED Brace Hysteretic Parameters – 284
Table 7.4: Participating Mass Ratios for the
Starting-Point T-SCED Structure – 291
Table 7.5: Velocity Comparison for the Starting-Point T-SCED Structure – 292
Table 7.6: Analysis Model Periods – 294
Table 7.7: Percentage of Critical Damping in Each Building Model
(values do not include the effect of friction damping from the SCED braces) – 297
Table 7.8: Frequently Considered (FOE) Hazard Level Earthquake Records
(50% Probability of Exceedence in 50 years) – 298
Table 7.9: Design-Basis (DBE) Hazard Level Earthquake Records
xii
List of Tables
Table 7.10: Maximum Considered (MCE) Hazard Level Earthquake Records

Table 7.11: Maximum Response Quantities - Median [Mean ] (Mean + One Standard Deviation ) –
308
Table 7.12: Number of Earthquake Records that Caused
Maximum Drifts Greater than 5% – 310
Table 7.13: Comparison of Model Results with Response Spectrum Predictions
for the V-SCED Models with No Friction – 311
Table 7.14: Response Quantities Relative to T-SCED (Realistic Stiffness) - Median Values – 318
xiii
List of Figures
LIST OF FIGURES
Figure 1.1: Nonlinear hysteretic response to earthquakes: Yielding/Friction Systems vs. Self-Centering
Systems – 2
Figure 2.1: The Effect of Residual Drift on the Drift Response of Structures Subjected to the Same
Earthquake Twice (adapted from Erochko et al., 2011) – 11
Figure 2.2: Maximum Residual Drifts in Buckling-Restrained Braced Frames and Special Moment-
Resisting Frames (adapted from Erochko et al., 2011) – 14
Figure 2.3: Nonlinear hysteretic response of Yielding/Friction Systems and Self-Centering Systems – 15
Figure 2.4: Example Buckling-Restrained Braces (BRBs) (adapted from Tremblay et al., 2006) – 17
Figure 2.5: Example Buckling-Restrained Brace Hysteresis (adapted from Tremblay et al., 2006), – 18
Figure 2.6: Behaviour of Rocking Walls and Self-Centering Moment Frames – 19
Figure 2.7: The Energy Dissipating Restraint (from Nims et al., 1993) – 22
Figure 2.8: The Fluid Restoring Force / Damping Device (from Tsopelas and Constantinou, 1994) –
24
Figure 2.9: Friction Spring Seismic Damper (adapted from Filiatrault et al., 2000) – 25
Figure 2.10: The Self-Centering Friction Damping Brace (from Zhu and Zhang, 2008) – 26
Figure 2.11: SCED Brace Mechanics (adapted from Christopoulos et al., 2008) – 28
Figure 2.12: The First SCED Brace Prototype (from Christopoulos et al., 2008) – 30
Figure 2.13: The Second SCED Brace Prototype (from Christopoulos et al., 2008) – 32
Figure 3.1: Sample SCED Brace Design – 37
Figure 3.2: Friction Damper General Design – 40
Figure 3.3: Friction Interface Material Tests (from Kim et al., 2004) – 41
Figure 3.4: Friction Damper General Design (T-SCED Prototype) – 42
Figure 3.5: Bolt Length Measurement Device – 45
Figure 3.6: Shake Table Test External Fuse Bolt Torquing Setup – 46
Figure 3.7: Tendon Anchor – 49
Figure 3.8: Tendon and Tendon Anchor Photos – 51
Figure 3.10: Generalized Tendon Behaviour – 52
Figure 3.9: Sample 22mm Tendon Test (adapted from Kim, Personal Communication January 5,
2010) – 52
Figure 3.11: Sample Linear Finite Element Analysis Results for T-SCED End Plate Design – 54
Figure 3.12: SCED Brace Mechanics (adapted from Christopoulos et al., 2008) – 56
Figure 3.13: SCED Brace Simplified Hysteresis – 57
xiv
List of Figures
Figure 3.14: SCED Brace Detailed Hysteretic Response – 62

Figure 3.15: SCED Brace Detailed Mechanics - Behavioural Stages – 63
Figure 3.16: T-SCED Brace Mechanics – 76
Figure 3.17: SCED Mechanics Simulator Model.
Note: Both models are one-dimensional, but are shown here separated into two dimensions for clarity
– 80
Figure 3.18: SCED Mechanics Simulator Inputs – 84
Figure 3.19: SCED Mechanics Simulator Results – 87
Figure 3.20: SCED Mechanics Simulator Output Data – 88
Figure 3.21: Detailed Mechanics Results vs. Mechanics Simulator Model for a Sample SCED Brace – 89
Figure 3.22: Sample SCED Mechanics Simulator Results Compared to Test Data – 90
Figure 3.23: Sample SCED Brace Hystereses - Effect of Length Tolerance – 91
Figure 3.24: Sample SCED Brace Hysteresis Breakdown - Effect of Length Tolerance – 94
Figure 3.25: Sample T-SCED Brace Hystereses - Effect of Length Tolerance – 100
Figure 3.26: Flag Shaped Bilinear Hysteresis for Ruaumoko (from Carr, 2005b) – 104
Figure 3.27: Constructing a Flag-Shaped Hysteresis in SAP2000 (adapted from Le Bec et al., 2010) –
106
Figure 3.28: OpenSees Self-Centering Hysteresis – 107
Figure 4.1: High-Capacity SCED Design Frame Geometry – 118
Figure 4.2: High-Capacity SCED Schematic – 120
Figure 4.3: Flexing of Member Walls in Earlier Designs due to Eccentric Bolt Loading – 122
Figure 4.4: High-Capacity SCED Cross-Section with Clearances – 123
Figure 4.5: High Capacity SCED Assembly Photos – 124
Figure 4.6: External Friction Fuse Subcomponent Test Specimen – 125
Figure 4.8: External Friction Fuse Subcomponent Calibration Tests – 126
Figure 4.7: External Friction Fuse Subcomponent Test Specimen Photos – 126
Figure 4.9: Effect of Velocity in Friction Fuse Subcomponent Tests – 129
Figure 4.10: Effect of Temperature in Friction Fuse Subcomponent Tests
and Recalibration of Fuse Slip Force – 130
Figure 4.11: Friction Fuse Subcomponent Sine Displacement History Tests – 131
Figure 4.12: Gouge in Friction Surface of Outer Yoke Plate – 132
Figure 4.14: Tendon Tensioning Order – 134
Figure 4.13: Tendon Tensioning Plot - First Set of Tendons – 134
Figure 4.15: Tendon Pretensioning Setup – 135
Figure 4.16: Tendon Tensioning Plot - Second Set of Tendons – 136
xv
List of Figures
Figure 4.17: HC-SCED Test Setup and Instrumentation – 144

Figure 4.18: HC-SCED Test Setup Photo – 145
Figure 4.19: HC-SCED Internal Friction Damper Calibration – 146
Figure 4.20: Calibrated HC-SCED Characterization and Tendon Anchor Failure – 147
Figure 4.21: Wind Loading Simulation Results – 148
Figure 4.22: Effect of Wind Loading Comparison – 149
Figure 4.23: Simulation of Brace Response and Calculation of Brace Element Forces for Test SC8-A –
151
Figure 4.24: Input Deformation Records for DBE Earthquake Tests – 153
Figure 4.25: Dynamic Earthquake Response - Design-Level Earthquake Records – 155
Figure 4.27: Dynamic Earthquake Response - Maximum Considered Earthquake Record – 156
Figure 4.26: Input Deformation Records for MCE Earthquake Tests – 156
Figure 4.28: Input Deformation Records for Dynamic Protocol Test – 159
Figure 4.29: Dynamic Displacement Protocol Response – 160
Figure 4.30: External Friction Fuse Surface Conditioning Tests – 162
Figure 4.31: External Friction Fuse Calibration – 163
Figure 4.32: Effect of Displacement Velocity on External Friction Fuse Response – 163
Figure 4.33: Dislodged Tendon Anchor Spike - First Set of Tendons – 165
Figure 4.34: Failure of Tendon Anchor during Tendon-Only Test – 167
Figure 4.36: Third Tendon Anchor Failure – 168
Figure 4.35: Internal Friction Damper Calibration with Step Behaviour – 168
Figure 4.37: Third Tendon Anchor Failure Photos – 169
Figure 4.38: Tendon Anchor Failure Photos – 170
Figure 4.39: Changes in HC-SCED Hysteresis due to Tendon Anchor Failure – 171
Figure 4.40: Effect of Tendon Debris on Hysteretic Response – 172
Figure 4.41: Simulation of Brace Response and Calculation of Brace Element Forces for Test SC16-A –
173
Figure 4.42: Attempt to Fail Tendon Anchors – 174
Figure 5.1: T-SCED Brace Schematic – 180
Figure 5.2: T-SCED Cross Section with Clearances – 181
Figure 5.3: T-SCED Assembly Photos – 182
Figure 5.4: Tendon Assembly Procedure (from Linear Composites Ltd., 2011) – 183
Figure 5.5: Tendon Assembly Photos – 184
Figure 5.6: T-SCED Tendon Tensioning Plot – 186
xvi
List of Figures
Figure 5.7: T-SCED Tendon Tensioning Order – 186

Figure 5.8: T-SCED Test Frame – 187
Figure 5.9: T-SCED Test Frame Photo – 187
Figure 5.10: T-SCED Test Instrumentation Summary – 189
Figure 5.11: Bare Frame Hysteretic Response with no T-SCED Brace – 190
Figure 5.12: T-SCED Brace Progressive Calibration – 198
Figure 5.13: Wind Simulation Response – 200
Figure 5.14: Effect of Wind Simulation on Hysteretic Response – 201
Figure 5.15: Deformation Limited Test Protocol Results up to – 202
Figure 5.16: Comparison of Test DPE-2 (recalibrated brace) to Mechanics Model Results – 204
Figure 5.17: Full Test Protocol Hysteretic Results up to – 207
Figure 5.18: Full Test Protocol Tendon Deformation Results – 207
Figure 5.19: Simulated Earthquake Drift History Hysteretic Results and Comparison to HC-SCED
Results – 209
Figure 5.20: Effect of Lower Gusset Connection Retrofit – 211
Figure 5.21: Effect of Beam Connection Slip and Internal Friction Fuse Plate Jam on the Outer End
Plate – 213
Figure 5.22: Elongated Slotted Holes in Beam South Connection – 214
Figure 5.23: Location of Outer End Plate Gouged Edge – 215
Figure 5.24: End Plate Guides – 216
Figure 5.25: Bolt Bearing in Internal Friction Damper – 217
Figure 5.26: Comparison of Test DPD-11 (with interfering bolts removed) to Mechanics Model Results
– 219
Figure 5.27: End Plate Edge Damage due to Friction Between Inner and Outer End Plates – 220
Figure 5.28: Full Protocol Results with Tendon Anchor Rupture – 221
Figure 5.29: Photos of Tendon Anchor Rupture – 222
Figure 6.1: Prototype Building Geometry (after Kim, 2009a) – 225
Figure 6.2: Small-scale SCED brace for Shake Table Test – 230
Figure 6.3: Test Structure Overview – 230
Figure 6.4: Test Structure Photographs – 231
Figure 6.5: Test Structure Instrumentation – 233
Figure 6.6: SCED Brace Calibration Load Frame Test Setup – 234
Figure 6.7: SCED Brace Load Frame Test Instrumentation – 235
Figure 6.8: Photo of SCED Brace Press Test Setup – 236
Figure 6.9: Step Response and Unsymmetric hysteresis in the Third Storey SCED brace – 237
xvii
List of Figures
Figure 6.10: End Plate Retrofit (Third Storey SCED Brace) – 238
Figure 6.12: Full Initial Hysteretic Behaviour of the Extra Brace – 243
Figure 6.11: Extra Brace External Fuse Tests (all slip levels) – 243
Figure 6.13: Preliminary SCED Brace Tests Hysteresis Summary – 245
Figure 6.14: Photo of Tendon Tensioning Setup – 246
Figure 6.15: Simulation of Brace Response and Calculation of Brace Element Forces for Test 2N-4 – 248
Figure 6.16: Simulation of Brace Response and Calculation of Brace Element Forces for Test E-12 – 250
Figure 6.17: SCED shake table test structure OpenSEES models
(a) with the shake table - horizontal accelerations applied at the fixed restraint on the left, and
(b) without the shake table - horizontal accelerations applied at the bases of the columns – 251
Figure 6.18: First Storey Brace Mechanism – 257
Figure 6.19: Sample Shake Table Test Results - Drift Time Histories and SCED Brace Hystereses – 258
Figure 6.20: Comparison between SCED Brace Axial Force Calculation Methods – 260
Figure 6.21: Korean Test Result - First Storey SCED Brace Response (from Kim, 2009b) – 261
Figure 6.22: Comparison of Axial Force Calculation Method for First Storey SCED Brace Test 2-4 – 262
Figure 6.23: Comparison of SAP2000 and OpenSees model results – 264
Figure 6.24: Comparison of Test Results with both Computer Models – 265
Figure 6.25: Sample Drift and Hysterestic Responses (Experiment vs. Model) – 266
Figure 6.26: First Floor Dynamic Equilibrium and the Effect of Velocity on the
Force Imbalance between Adjacent Storeys – 267
Figure 6.27: Sample Acceleration Spike Response – 269
Figure 6.28: Reduction of Free Vibration Base Shear Amplitude with Added Inherent Damping – 271
Figure 7.1: Six-Storey SCED Building Design (after Choi et al., 2008) – 276
Figure 7.2: Six-Storey SCED Building OpenSees Model – 278
Figure 7.3: Six-Storey SCED Building OpenSees Model – 280
Figure 7.4: Energy Dissipated by the SCED versus a Linear Viscous Damper – 289
Figure 7.5: Building Model Pushover Analysis Comparison – 295
Figure 7.6: Spectral Acceleration Plots for the Considered Earthquake Records – 299
Figure 7.7: Hysteretic Response Comparison for First Storey Brace (MCE Earthquake LA21) – 301
Figure 7.8: Hysteresis Breakdown for the First Storey Braces in the V-SCED designs (MCE Earthquake
LA21) – 302
Figure 7.9: Comparison of SCED Brace Response under Different Seismic Hazard Levels – 303
Figure 7.10: Time History Comparison of First and Top Floor Displacements (MCE Earthquake LA21)
– 304
Figure 7.11: Time History Comparison of Second Floor Acceleration and Base Shear (MCE Earthquake
LA21) – 305
xviii
List of Figures
Figure 7.12: Time History Comparison of Second Floor Acceleration and Base Shear (MCE Earthquake
LA21) – 307
Figure 7.13: Result Comparison - Choi et al. SCED versus O-SCED – 312
Figure 7.14: Result Comparison - Theoretical versus Realistic Initial Stiffness of O-SCEDs – 314
Figure 7.15: Result Comparison - Theoretical versus Realistic Initial Stiffness of T-SCEDs – 315
Figure 7.16: Result Comparison - O-SCED versus T-SCED (Realistic Initial Stiffness) – 317
Figure 7.17: Result Comparison - T-SCEDs versus Second Mode Period V-SCEDs – 319
Figure 7.18: Result Comparison - V-SCEDs First Mode Period Design versus Second Mode Period
Design – 321
xix
List of Symbols
LIST OF SYMBOLS
A - cross-sectional area
Ab - nominal cross-sectional area of a bolt
Ai - cross-sectional area of the inner member of a SCED brace
Am - cross-sectional area of the intermediate member of a SCED brace
Ao - cross-sectional area of the outer member of a SCED brace
Ap - nominal cross-sectional area of pretensioning tendon
A pt - total nominal cross-sectional area of all of the pretensioning tendons in a SCED brace or of
each set of pretensioning tendons in a T-SCED brace
b - width of cross section
Cd - deflection amplification factor as given in Tables 12.2-1, 15.4-1 or 15.4-2 of
ASCE 7-05(ASCE, 2005)
CL - linear viscous damping constant
Cs - seismic response coefficient as given by clause 12.8.1.1 of ASCE 7-05(ASCE, 2005)
Cu - coefficient for upper limit on calculated period from ASCE 7-05(ASCE, 2005)
d - depth of cross section
db - nominal diameter of steel bolt
d ts - minimum cross-sectional area of a bolt in the trough of the threads
D - diameter
Es - Young’s modulus of steel
Ep - effective cyclic Young’s modulus of a pretensioning tendon
E p0 - initial Young’s modulus of a pretensioning tendon
E flag - energy dissipated by a flag-shaped hysteresis in each hysteretic cycle
E flag d - energy from the flag shaped hysteresis that is to be replaced by viscous damping
E vd - energy dissipated by a linear viscous damper in each hysteretic cycle
F Et - external fuse activation force of a SCED brace
F , F It - internal friction damper total slip force of a SCED brace
F I1 - internal friction damper slip force at end 1 of a SCED brace
F I2 - internal friction damper slip force at end 2 of a SCED brace
FN - total normal force on a friction interface
hs - storey height
HB - height of a bolt head
xx
List of Symbols
HN - height of a bolt nut

Ix - second moment of area (‘moment of inertia’) about the strong bending axis
Iy - second moment of area (‘moment of inertia’) about the weak bending axis
I x req - required second moment of area (‘moment of inertia’) (x or y direction)
I x sel - selected second moment of area (‘moment of inertia’) (x or y direction)
k1 - theoretical initial stiffness of a SCED brace
k 1 eff - effective initial stiffness of a SCED brace (secant stiffness between positive and negative acti-
vation points)
k 1 efftol - effective initial stiffness of a SCED brace (secant stiffness between positive and negative acti-
vation points) including the effect of the construction tolerance on the length of the axial
members.
k 1 LFE - initial stiffness of short linear frame element (for SAP2000 SCED modelling)
k 1 MLE - initial stiffness of multi-linear elastic element (for SAP2000 SCED modelling)
k 1 MLP - initial stiffness of multi-linear plastic element (for SAP2000 SCED modelling)
k i sec - secant stiffness of a SCED brace at deformation level i
k 1 tot - total initial stiffness of a SCED brace including the effective initial stiffness k 1 eff and the
stiffness of the connections in series with the brace k conn
ka - post-activation stiffness of a SCED brace
+
ka - post-activation stiffness of a SCED brace in the tension direction
-
ka - post-activation stiffness of a SCED brace in the compression direction
k a MLE - post-activation stiffness of multi-linear elastic element (for SAP2000 SCED modelling)
k a MLP - post-activation stiffness of multi-linear plastic element (for SAP2000 SCED modelling)
k conn - connection stiffness
ki - axial stiffness of the inner member
km - axial stiffness of the intermediate member
ko - axial stiffness of the outer member
kp - total axial stiffness of all the pretensioning tendons in a SCED brace or of each set of tendons
in a T-SCED brace
k sec d - secant stiffness of a SCED brace at design deformation
LB - effective bolt length
Li - length of the inner member of a SCED brace
LJ - total joint length in a bolted connection (from the inner side of the bolt head to the inner side
of the nut)
Lm - length of the intermediate member of a SCED brace
Lo - length of the outer member of a SCED brace
Lp - effective pretensioning tendon length
xxi
List of Symbols
L p0 - initial unloaded pretensioning tendon length

LS - length of the unthreaded shank of a bolt within the joint
n 5% - number of earthquake records with maximum drift exceeding 5% of storey height
P - axial force
+
P - axial force in the tension direction
-
P - axial force in the compression direction
Pa - activation force of a SCED brace
P a MLE - activation force of multi-linear elastic element (for SAP2000 SCED modelling)
P a MLP - activation force of multi-linear plastic element (for SAP2000 SCED modelling)
P ai - activation force of a SCED brace at storey i
Pb - axial force in bolt
Pd - axial force in SCED brace at design drift
Pi - absolute axial force in the inner member of a SCED brace
P i0 - total precompression force in the inner member of a SCED brace
P ir - axial force in the inner member of a SCED brace relative to the initial axial force caused by
the pretensioned tendons
P is - force on the inner member of a SCED brace that is required to make the inner and outer
members the same length (up to the maximum pretension load provided by the tendons)
P mr - axial force in the inner member of a SCED brace relative to the initial axial force caused by
Po - absolute axial force in the outer member of a SCED brace
P o0 - total precompression force in the outer member of a SCED brace
P or - axial force in the outer member of a SCED brace relative to the initial axial force caused by
P os - force on the outer member of a SCED brace that is required to make the inner and outer
members the same length (up to the maximum pretension load provided by the tendons)
Pp - absolute total axial force in all the tendons of a SCED brace
P p0 - total pretension force in all the tendons of a SCED brace or of each set of tendons in a T-
SCED brace
P pr - total axial force in all the tendons of a SCED brace relative to the initial axial force caused by
the pretension
P pir - total axial force in the inner set of tendons of a T-SCED brace relative to the initial axial force
caused by the pretension
P por - total axial force in the outer set of tendons of a T-SCED brace relative to the initial axial force
caused by the pretension
P piu - absolute axial strength of a single tendon of a SCED brace
xxii
List of Symbols
Ps - external friction fuse slip force of a SCED brace

PT - target activation force of a SCED brace
Pu - ultimate force capacity of a SCED brace
Pv - viscous damper force
P v max - peak viscous damper force
r - post-activation stiffness ratio of a SCED brace
R - response modification coefficient as given in Tables 12.2-1, 12.14-1, 15.4-1, or 15.4-2 of
ASCE 7-05(ASCE, 2005)
Sa - scaling factor for acceleration
SA - scaling factor for area
S DS - design, 5 percent damped, spectral response acceleration parameter at short periods as defined
in Section 11.4.4 of ASCE 7-05(ASCE, 2005)
S D1 - design, 5 percent damped, spectral response acceleration parameter at a period of 1 s as
defined in Section 11.4.4 of ASCE 7-05(ASCE, 2005)
SE - scaling factor for elastic modulus
SF - scaling factor for force
SI - scaling factor for second moment of area (‘moment of inertia’)
SL - scaling factor for length
SM - scaling factor for mass
ST - scaling factor for time
SZ - scaling factor for section modulus
tf - thickness of flange
tw - thickness of web
T , T 1 - the fundamental period of a building
Ta - approximate fundamental period of the building as determined in Section 12.8.2 of
ASCE 7-05(ASCE, 2005)
Tb - bolt torque
T eff - effective building period
Ti - the period of mode i of a building, or the period at drift level i
V cLi - left column shear at storey i
V cRi - right column shear at storey i
V xi - storey shear at storey i
W - effective seismic weight of the building as defined in Section 12.7.2 of ASCE 7-05
(ASCE, 2005)
xxiii
List of Symbols
 - SCED brace energy dissipation capacity parameter

 - elongation/deformation
·
 - velocity
·
 max - maximum velocity
+
 - elongation/deformation in the tension direction
-
 - elongation/deformation in the compression direction
a - elongation of a SCED brace at activation
+
a - elongation of a SCED brace at positive (tension) activation
-
a - elongation of a SCED brace at negative (compression) activation
b - bolt elongation
d - elongation of a SCED brace at design drift
 EPi1 - movement of the inner end plate relative to the outer member at end 1 in a T-SCED brace
 EPi2 - movement of the inner end plate relative to the outer member at end 2 in a T-SCED brace
 EPo1 - movement of the outer end plate relative to the outer member at end 1 in a T-SCED brace
 EPo2 - movement of the outer end plate relative to the outer member at end 2 in a T-SCED brace
i - inner member elongation, or horizontal deformation of floor i
+
 ia - SCED brace elongation at inner member positive (tension) activation
-
 ia - SCED brace elongation at inner member negative (compression) activation
 i0 - initial inner member elongation after tendon pretensioning
o - outer member elongation
+
 oa - SCED brace elongation at outer member positive activation
-
 oa - SCED brace elongation at outer member negative activation
 o0 - initial outer member elongation after tendon pretensioning
p - pretensioning tendon elongation
 p0 - initial pretensioning tendon elongation after pretensioning
 pi - inner pretensioning tendon elongation
 po - outer pretensioning tendon elongation
 pu - effective ultimate axial deformation in the tendons of a SCED brace
u - total elongation capacity of a SCED brace
 - storey drift
b - control deformation for the AISC buckling-restrained brace protocol (AISC 2005b)
 bm - deformation of specimen at first yielding for the AISC buckling-restrained brace protocol
(AISC 2005b)
xxiv
List of Symbols
 by - deformation corresponding to the design storey drift for the AISC buckling-restrained brace
protocol (AISC 2005b)
i - drift of storey i
 io - initial, unloaded difference in length between the inner and outer members of a SCED brace
 io ant - anticipated difference between the lengths of the members of a SCED brace. Equal to twice
the construction length tolerance.
 io r - difference in length remaining between the inner and outer members of a SCED brace after
the load from the pretensioned tendons has been applied
 rel - relative movement of the two SCED members
r - residual drift
 ri - residual drift of storey i
T - target storey drift
L p - change in effective pretensioning tendon length
L pa - change in absolute pretensioning tendon length
i - average axial strain in the inner member of a SCED brace
m - average axial strain in the intermediate member of a SCED brace
o - average axial strain in the outer member of a SCED brace
p - service axial strain in the tendons of a SCED brace
 p0 - reference zero axial strain in the tendons of a SCED brace for calculation of service axial strain
 pa - absolute axial strain in the tendons of a SCED brace
 pu - ultimate service axial strain in the tendons of a SCED brace
b - brace angle
 - mean
d - coefficient of dynamic friction
s - coefficient of static friction
 - percentage of critical damping
 - standard deviation
 pu - ultimate axial stress in the tendons of a SCED brace
 - strength reduction factor
 - angular frequency
 eff - effective angular frequency used for the design of viscous dampers
0 - overstrength factor as defined in Tables 12.2-1, 5.4-1, and 15.3-1 of ASCE 7-05(ASCE,
2005)
xxv
CHAPTER 1: INTRODUCTION
The current state-of-the-art for the seismic design of building structures typically incorporates the use
of pre-determined and well-detailed inelastic mechanisms (or structural 'fuses') to protect the integrity of a
structure and to reduce and limit the force demands on structural elements. Some examples of common
structural fuses include buckling/yielding steel braces, buckling-restrained braces (BRBs), yielding beams
in steel or concrete moment-resisting frames, and the bases of concrete walls. The load reduction and limi-
tation benefits that these fuses provide typically come at the cost of localized damage in the inelastic fuse
itself. This is desirable because the fuse is specifically designed and detailed to accommodate that damage.
To avoid damage elsewhere in the structural system, all other structural elements are designed according to
the principals of capacity design, whereby elements are designed to resist the highest possible load from the
inelastic fuses. This protects the gravity elements and helps to prevent collapse of the structure.
This structural fuse design paradigm has proved to be a significant step forward in the seismic design of
structures, but it suffers from two important drawbacks. First, buildings that use typical inelastic fuses are
likely to sustain permanent deformations after a significant seismic event. These permanent deformations
are generally referred to as residual drifts. Previous studies have recognized the importance of residual drifts
as a key performance index for building design and have investigated the magnitude of residual drifts for
various nonlinear single- and multiple-degree-of-freedom-systems (Christopoulos et al., 2003; Pampanin
et al., 2003; Ruiz-Garcia & Miranda, 2006a-b). A recent analytical study by Erochko et al. (2011) has sug-
gested that the magnitude of the residual drift may be estimated by subtracting the recoverable elastic drift
(that is, the drift at yield) from the maximum expected drift. These residual drifts are important because
they pose significant challenges to continued building occupancy and repair after an earthquake. A recent
study by McCormick et al. (2008) suggested that, in a Japanese context, a residual drift greater than 0.5%
means that it will likely be less expensive to completely replace a building than it would be to retrofit it.
The second drawback is that for earthquakes of significant duration, typical inelastic systems may experi-
1
Chapter 1: Introduction 2
ence a progressive collapse. This problem arises because drifts in these systems tend to accumulate in one
direction due to a bias that is created in the direction of any inelastic lean. Under the influence of P-Delta
effects, successive cycles tend to pull a building predominantly in this same direction, eventually leading to
collapse (MacRae & Kawashima, 1997). This type of progressive collapse mechanism is particularly dan-
gerous in areas of the world that are proximate to subduction-type faults, which tend to produce long dura-
tions of significant ground accelerations on the order of two to three minutes. Some examples of such sites
include the western coast of North America near Vancouver and Seattle, the western coast of South Amer-
ica near Chile and the eastern coast of Japan.
Self-centering systems arose primarily to mitigate these drawbacks in the current design paradigm by
attempting to reduce or eliminate residual drifts. The use of self-centering systems allows a designer to
develop building systems that can achieve higher levels of performance under earthquake loading. The
common feature of all self-centering systems is that, unlike typical yielding or friction elements in build-
ings that exhibit a parallelogram-shaped response, self-centering systems exhibit a flag-shaped hysteretic
response as shown in Figure 1.1. This flag-shaped response dissipates less energy per cycle than a compara-
ble yielding device (at most, half the energy dissipation capacity), but it returns to zero displacement at the
end of every cycle. This lack of energy dissipation capacity is not in itself a significant drawback, since large
earthquakes tend to be characterized by one or two maximum peaks in acceleration. As a consequence, it
has been shown that buildings equipped with either yielding or self-centering systems tend to have similar
maximum storey drift demands (Christopoulos et al., 2002a; Choi et al., 2008; Tremblay et al., 2008).
Yielding/Friction System Self-Centering System
Force Force
Maximum Maximum
Elastic Responses Elastic Responses
Hysteretic
Energy
Hysteretic
Energy
Displacement Displacement
Residual No Residual
Figure 1.1: Nonlinear hysteretic response to earthquakes: Yielding/Friction Systems vs. Self-Centering Systems
Jeffrey Erochko, Department of Civil Engineering, University of Toronto

A number of different practical structural configurations have been devised that are able to produce the
type of global self-centering behaviour shown in Figure 1.1. These include self-centering moment-resisting
frames and rocking wall systems, which will be discussed in Chapter 2. Both of these types of systems
require that the building’s structural system must be modified to accommodate significant localized defor-
mations. This is particularly important at the interface between the lateral load resisting system and the
gravity supporting elements such as adjacent gravity columns and floor slabs. This problem may be
avoided through the use of an alternative self-centering system that has been developed at the University of
Toronto and Ecole Polytechnique in Montreal. The Self-Centering Energy-Dissipative (SCED) brace pro-
duces a full flag-shaped response within the form-factor of an easily-accommodated bracing member
(Christopoulos et al., 2008).
SCED brace prototypes have been previously built and tested both axially and within a single-storey
frame subjected to dynamic loading by Christopoulos et al. (2008); however, following those original pro-
totype tests, a number of issues remained which have hampered the adoption of the technology in real
structures.
This doctoral thesis improves upon the design and use of self-centering energy dissipative (SCED)
braces through three large-scale experimental studies and an associated numerical building model study.
The first experimental study was conducted to increase the strength capacity of SCED braces and to refine
the design procedure. To achieve these goals, a new high capacity (3000kN axial capacity) full-scale SCED
brace was designed and tested pseudo-dynamically in a load frame. The second experimental study was
conducted to extend the elongation capacity of SCED braces. This was achieved by designing a new
enhanced-elongation telescoping SCED (or T-SCED) brace that provides a self-centering response over
two times the deformation range that was achieved by the original SCED bracing system. This design was
fabricated and tested dynamically in a full-scale vertical steel frame. The third experimental study was con-
ducted to evaluate the dynamic response of a multi-storey SCED-frame system subjected to multiple types
of ground motions from different seismic environments. An additional goal of this third experimental
study was to assess the ability of computer models of differing complexity, including commercial software
that is widely used in practice, to accurately predict the seismic response of a multi-storey SCED-braced
frame. To these ends, a three-storey SCED-braced frame was designed, constructed, and tested in real time
on a shake table. Lastly, a numerical six storey SCED-braced frame building model was created and tested
to evaluate the dynamic behaviour of the new SCED designs, to assess the effect of brace construction tol-

erances on the dynamic frame behaviour, and to evaluate the benefit of adding viscous dampers to every
storey in parallel with the T-SCED braces to reduce the structure’s floor accelerations.
Each of the major studies introduced above will be discussed in a separate chapter; however, some
important practical and theoretical background information will first be provided in Chapters 2 and 3.
Chapter 2 will present the previous research that has been conducted in the area of self-centering systems,
will present the previous SCED brace prototypes, and will elaborate on the motivation and objectives of
the research. Chapter 3 will then provide the theoretical context for the research by presenting the mechan-
ics and design methods for SCED braces as well as the modified mechanics of the new proposed telescop-
ing (T-SCED) brace system. In addition, Chapter 3 will discuss the modelling of SCED and T-SCED
braces and will describe a new software tool that has been developed to model the detailed mechanical
behaviour of SCED and T-SCED braces, which includes modelling the effect of construction tolerances.
Chapter 4 will then describe the design and experimental testing of the new high-capacity (HC-SCED)
brace, Chapter 5 will discuss the design and experimental testing of the enhanced-elongation T-SCED
brace, and Chapter 6 will describe the design, testing and modelling of the shake table tests. To tie the
experimental results together, Chapter 7 will present the results from the six storey numerical building
model that was used to assess and compare the behaviour of realistically modelled SCED and T-SCED
braces. These models were calibrated and designed based on the results of the analyses presented in Chap-
ters 4 to 6 and modelled using the software tool and methods described in Chapter 3. This numerical
model also evaluated the effect of adding viscous damping to SCED-braced frames. Chapter 8 will then
summarize the contributions and limitations of this research and present recommendations for future
research, including the practical challenges that remain to be solved before SCED braces can be widely
used in real building structures.

CHAPTER 2: BACKGROUND AND MOTIVATION
Since the Self-Centering Energy-Dissipative (SCED) brace eliminates residual drifts in structures, it
provides an higher level of performance than conventional seismic force resisting systems. This elimination
of the residual drifts increases building performance by increasing the likelihood that a building will still be
usable after an earthquake; however, the ability to take advantage of this in design relies on a designer’s abil-
ity to quantify the benefit that is achieved above and beyond the minimum code requirements. This may
be done by using a performance-based design methodology. A few such methodologies have been devel-
oped and improved over the past twenty years, which have led to increased demand for high-performance
systems such as the SCED brace.
A number of other pioneering self-centering systems for buildings have been developed over that time.
These include rocking piers and walls, self-centering moment frames, and other self-centering braces. A
brief summary of these previously developed systems will be provided in this chapter.
Even though other self-centering braces have been proposed, the SCED brace is the first large-deforma-
tion, high-capacity, self-centering cross-brace that is practical for use in a real building. Two prior full-scale
SCED brace prototypes have been designed, built and tested (Christopoulos et al., 2008). These proto-
types provided proof of the concept and confirmed the dynamic response of the brace in the context of a
full-scale steel frame. In parallel with those physical tests, two major numerical studies were conducted
which examined the seismic response of SCED braced frames and compared their behaviour to that of
other inelastic seismic force resisting systems for buildings (Tremblay et al., 2008; Choi et al. 2008). The
results from these prototype tests and models provide the background for this thesis work.
2.1 PERFORMANCE-BASED DESIGN

The use of modern building codes results in structures that protect life-safety very well during an earth-
quake. During the recent earthquake in Christchurch, New Zealand (22 February 2011), no structures
5
Chapter 2: Background and Motivation 6
that were built with modern earthquake-resistant building codes collapsed (Wilkinson et al., 2013). Simi-
larly, during the recent earthquake in Van, Turkey (23 October 2011), the well-designed modern buildings
performed well, whereas buildings that were not code-compliant experienced poor performance and
tended to collapse (Erdik et al., 2012). Even though these earthquakes have shown that modern design
codes protect life safety and prevent collapse, it has been clear that they do not provide any guarantee that
a building will be usable after an seismic event. After the Christchurch earthquake it was estimated that in
the central business district 900 buildings would need to be completely demolished and that outside of
that district 10 000 residential homes would similarly have to be demolished (Kaiser et al., 2012).
Prior to 1995, there was no broadly accepted method for designing buildings to have a specified level of
performance greater than that prescribed by building codes. The only way to design a building for a higher
level of performance was to use an ‘importance factor’ such as the one that is included in building codes
such as ASCE 7: Minimum Design Loads for Buildings and Other Structures (ASCE, 2005) and the
National Building Code of Canada (NRC, 2005). This factor simply assigns higher design loads to soci-
etally important structures such as hospitals, police stations, fire stations, schools, power generation and
distribution facilities, and emergency shelters. It does not explicitly consider the behaviour of those impor-
tant buildings when subjected to seismic hazards in order to ensure that the desired performance is
achieved. Nor does the importance factor consider any other reasons why an owner or designer would
desire a higher level of performance, such as the case of a private business that cannot fiscally tolerate long
periods of downtime. These codes only consider a single level of seismic hazard based on the type of occu-
pancy and seismic risk at the structure’s location.
In 1995, the Structural Engineers Association of California (SEAOC) published the Vision 2000 docu-
ment which explicitly defined multiple possible levels of seismic performance for different levels of seismic
hazard (SEAOC, 1995). This resulted in the performance/hazard matrix shown in Table 2.1. In the table,
each column represents a different performance objective from fully operational, which means that there is
no down-time in building use after an earthquake, to near collapse, which means that the building is a total
loss and is uninhabitable. The rows in the table represent different levels of seismic hazard which are char-
acterized by different earthquake return periods, from the very frequent earthquake that has a probability
of occurrence of 50% in 30 years to the very rare earthquake that has a probability of occurrence of 10% in
100 years. The Vision 2000 document specifies a base level of performance whereby the building is fully
operational after a frequent earthquake, operational following some minor down time after an occasional
earthquake, preserves life safety during a rare earthquake, and prevents collapse under a very rare earth-

quake. Higher levels of performance are also shown in the table which are roughly equivalent to the high
importance categories defined in the codes, with essential and hazardous facilities being designed for a
higher performance level and safety critical facilities such as nuclear power stations designed so that they
are operational even when subjected to a very rare seismic event. The Vision 2000 document provides cri-
teria for each level of performance including building drift limits and the amount of damage permitted to
specific structural parts such as the plastic hinges in steel moment frames; however, it does not specify any
analytical approaches that may be used to determine performance.
Table 2.1: Vision 2000 Performance Based Design (adapted from SEAOC, 1995)
Performance Objective
Earthquake Fully
Operational Life Safe Near Collapse
Probability Operational
Frequent
Basic Unacceptable Unacceptable Unacceptable
50% in 30 years
Facilities Performance Performance Performance
(43 year return)
Occasional Essential/
Basic Unacceptable Unacceptable
50% in 50 years Hazardous
Facilities Performance Performance
(72 year return) Facilities
Rare Safety Essential/
Basic Unacceptable
10% in 50 years Critical Hazardous
Facilities Performance
(475 year return) Facilities Facilities
Very Rare Safety Essential/
Basic
10% in 100 years Critical Hazardous
Facilities
(970 year return) Facilities Facilities
Performance-based design methodologies were further developed under mandates from the American
Federal Emergency Management Agency (FEMA) as laid out in the documents FEMA 273: NEHRP
Guidelines for the Seismic Rehabilitation of Buildings (BSSC, 1997) and FEMA 356: Prestandard and
Commentary for the Seismic Rehabilitation of Buildings (ASCE, 2000). These expanded upon the foun-
dation laid by Vision 2000 by adding an explicit recognition of the uncertainty inherent in the perfor-
mance-design process and by focusing on a displacement-based design approach. The revised performance-
based design matrix proposed by FEMA 356 is shown in Table 2.2. Instead of defining specific combina-
tions of performance for each level of seismic hazard, the FEMA-356 approach permits a more flexible
combination of different performance criteria as shown in the note below the table. The use of criteria k
and p in the table represent the basic level of safety that should be present in all designs. This basic perfor-
mance level may then be enhanced as shown. To determine the response of the structure, the FEMA-356

guideline specifically prescribes that nonlinear dynamic and static pushover analyses may be used. Like the
Vision 2000 guidelines, performance criteria for both structural and non-structural components are pro-
vided. This includes specific consideration of the permanent (residual) drift remaining in a structure after
an earthquake.
Table 2.2: NEHRP/FEMA 356 Performance Based Design (adapted from ASCE, 2000)
Performance Objectivea
Seismic Hazard Immediate Collapse
Operational Life Safety
Level Occupancy Prevention
(FOE)
50% in 50 years a b c d
(72 year return)
20% in 50 years
e f g h
(225 year return)
BSE-1 (DBE)
10% in 50 years i j k l
(475 year return)
BSE-2 (MCE)
2% in 50 years m n o p
(2475 year return)
a. Rehabilitation / Design Objectives:
k + p = Basic Safety Objective (BSO)
k + p + any of a, e, i, b, f, j, or n = Enhanced Objectives
o alone or n alone or m alone = Enhanced Objective
k alone or p alone = Limited Objectives
c, g, d, h, l = Limited Objectives
The seismic hazard levels shown in Table 2.2 are provided in terms of probability of exceedance in 50
years. These same seismic hazard levels will be used in later chapters. The 50% in 50 year level is called the
frequently occurring earthquake (FOE), the 10% in 50 year level is called the design-basis earthquake
(DBE) and the 2% in 50 year level is called the maximum considered earthquake (MCE). The perfor-
mance levels in the FEMA-356 approach are defined as follows (ASCE, 2000): operational means that
there is very little damage and the building can still operate using backup systems, immediate occupancy
means that only minor repairs are necessary and the building is safe to use, life safety means that the struc-
ture requires major repairs but it is stable and not in danger of collapse, and collapse prevention means that
the structure is close to collapse and will likely need to be demolished.

The future of performance based design relies on a more comprehensively probabilistic approach. This
approach was first proposed by Cornell and Krawinkler (2000) at the Pacific Earthquake Engineering
Research Center (PEER) and is generally referred to as the PEER performance-based design methodology.
A draft guideline based on this PEER approach, ATC-58, has been developed by the Applied Technology
Council (ATC), under a mandate provided by FEMA (ATC, 2011a; ATC, 2011b). The ATC-58 guideline
provides a much broader definition of what performance means (ATC, 2011a). Instead of being specified
in terms that relate primarily to structural performance, in ATC-58 performance is primarily defined in
terms of casualties, repair cost, repair time, and the probability that a building will be determined to be
unsafe after an earthquake. As in the earlier guidelines, the performance assessment is partially based on the
seismic hazard level and dynamic building response; however, it also considers the vulnerability of the
building contents to damage, the number of people who are likely to be present in the building at the time
of the earthquake, and the anticipated reactions of building inspectors after an earthquake. By using a
probabilistic approach, the assessment of the building performance may be computed not only based on
the spectral intensity of the earthquake, it may also be computed for different hazard scenarios or for differ-
ent periods of time. Since this probabilistic performance-based design requires complex and tedious analy-
ses, a companion software tool called the Performance Assessment Calculation Tool (PACT) was created as
part of the ATC project (ATC, 2011b).
2.2 RESIDUAL DRIFTS

One of the key parameters that is used to represent the performance-level of a building is the lean that
remains in a building after a seismic event. This lean is generally expressed as a ratio of the horizontal
deformation of a storey to the storey height at the end of the earthquake and is called the permanent or
residual drift  r . The residual drift of storey i (  ri ) is equal to:
i – i – 1
 ri = --------------------
- = i (Eq. 2.1)
hs time = end of EQ
time = end of EQ
where the  i and  i – 1 are the horizontal floor deformations for the floor levels above and below storey i
h s is the storey height, and  i is the storey drift of storey i .
2.2.1 THE IMPORTANCE OF RESIDUAL DRIFTS
The importance of residual drift as a performance measure stems directly from the progress that has
been made in the development of performance-based design methodologies as described in the previous

section. A strictly code-based design, with no explicit consideration of the post-earthquake performance of
a building, will have no interest in the residual drifts; it will only be concerned with the life safety of the
occupants under the design-based seismic hazard level earthquake and with preventing the structure from
collapsing. Using performance-based design, buildings may be designed for a higher performance level at
both the design-basis and maximum-considered seismic hazard levels. A designer may also determine the
anticipated performance of the structure when subjected to lower hazard earthquakes. For all of these
design scenarios, the residual drift will be a good indicator of the post-earthquake performance of the
structure.
Several previous studies have established this importance of residual drifts as a key performance index.
Residual drifts effect a building’s post-earthquake performance in two ways. The first is that they effect a
structure’s ability to sustain aftershocks. A case study of an eight storey moment frame commissioned by
the Pacific Gas & Electric Company and reported by Maffei et al. (2008) found that residual drifts in the
structure may significantly reduce the effective ductility capacity that is available to resist future earth-
quakes. A study of multi-storey special moment-resisting frames (SMRFs) and buckling-restrained braced
frames (BRB frames) by Erochko et al. (2011) showed that if the maximum residual drift in a structure is
less that 0.5% of the storey height then the building will not likely have a significantly increased risk of
damage when subjected to a second identical earthquake record; however, if the residual is between 0.5%
and 1.0%, then the building drifts resulting from a second identical earthquake would generally be
increased by 10-15%. If the residual is greater than 1.0%, then the drifts in the structure could be
increased by more than 50%. The Erochko et al. (2011) results are shown for the two studied building
heights in Figure 2.1. In the figure, the incremental drift ratio (IDR) represents the drift caused by the sec-
ond earthquake minus the initial residual drift caused by the first earthquake, divided by the drift that was
originally caused by the first earthquake. The authors suggest that, based on these results, buildings with
residual drifts between 0.5% and 1.0% would be safe to enter but not likely safe to occupy after an earth-
quake and that buildings with residual drifts greater that 1.0% would not be safe. They also determined
that, as the figure shows, the buckling-restrained braced frames were more sensitive to the effect of the
residual drift than the special moment resisting frames and that taller buildings were likewise more greatly
affected than shorter buildings.
The second effect that residual drifts may have on the post-earthquake performance of a building is that
they may significantly increase the expense that is required to repair a structure so that it may be used again
after an earthquake. For example, Rosenblueth and Meli (1986) reported after the 1985 Mexico earth-

6 Storey Models 12 Storey Models

(All Stories and EQs) (All Stories and EQs)
3.0 12 Story
Data Points
for IDR>3.0
not shown
2.5 BRB for clarity of
average data
SMRF
BRB BRB
2.0
Incremental
Drift Ratio
(IDR) SMRF
1.5
( )
∆max2 - ∆r1
∆max1
BRB
SMRF
Moving
1.0 Averages
SMRF Moving (eval. every
Averages 0.5%)
0.5
(eval. every
0.5%)
0 1% 2% 3% 0 1% 2% 3%
Story Residual Drift after 1st EQ (∆r1)
Figure 2.1: The Effect of Residual Drift on the Drift Response of Structures Subjected to the Same Earthquake Twice
(adapted from Erochko et al., 2011)
quake a number of reinforced concrete buildings were demolished due to the cost associated with repairing
residual drifts. McCormick et al. (2008) conducted a case-study of one occupied building at Kyoto Univer-
sity in Japan and also reviewed previous research into the effects of residual drifts in Japanese buildings.
Their study included consideration of both the physiological and psychological effects of residual drifts on
building occupants. They concluded that residual drifts of 0.5% are generally perceivable by occupants,
and that occupants may experience dizziness and nausea as residual drifts approach 1.0%, making build-
ings with such large residual drifts practically unusable. This level of residual drift is also not easily
repaired; the study concluded that in Japan it was generally more expensive to repair a building with resid-
ual drifts greater than 0.5% than it would be to tear it down and replace it.
2.2.2 RESIDUAL DRIFTS IN PERFORMANCE-BASED DESIGN
The effect of residual drifts on building performance has gradually come to be fully incorporated into
performance-based design methodologies. This process started with Kawashima et al. (1998) who devel-
oped residual drift spectra based on the post-yield stiffness of single-degree-of-freedom (SDOF) bilinear
oscillators. These spectra allow the estimation of the residual drift response of SDOF structures. They were

developed with the understanding that residual drifts are an important factor in the repair and reconstruc-
tion of a building after an earthquake and that, therefore, the quantification of the underlying causes and
effects of residuals is important for the quantification of building seismic performance.
Christopoulos et al. (2003) also recognized that, along with the peak response quantities such as drifts
and accelerations, residual drifts provide an important measurement of building performance. In the con-
text of performance-based design, they developed a framework to quantify the damage to structural and
non-structural components caused by both peak and residual deformations in simplified SDOF systems.
They also highlighted the benefits of self-centering systems in reducing or eliminating residual drifts and
argued that valid comparisons between traditional inelastic systems and self-centering systems should
include consideration of the effect of residuals on building performance. In a companion study by Pam-
panin et al. (2003), these residual drift damage indices were extended to multiple-degree-of-freedom
(MDOF) systems including the consideration of P-Delta effects and higher mode effects. Later, Christo-
poulos and Pampanin (2004) proposed a displacement-based design procedure for MDOF systems that
considered residual drifts along with suggestions of ways to mitigate them.
Ruiz-Garcia and Miranda (2006a) developed a method to estimate the residual deformations of elasto-
plastic SDOF systems on firm soil conditions while taking into account the uncertainty in the seismic
demand and system response. A companion MDOF analysis of the residual drift response of two-dimen-
sional inelastic moment-frames with flexural beam hinges showed that the uncertainty in the determina-
tion of the residual drift for these systems is significantly larger than the uncertainty in the peak drifts
(Ruiz-Garcia & Miranda, 2006b). They emphasize that this larger uncertainty in the residual drift values
should be considered in performance-based design.
To accommodate the modern probabilistic PEER performance-based design methodologies described

in the previous section, Uma et al. (2010) developed a probabilistic approach to estimate residual drifts in
SDOF systems and Ruiz-Garcia and Miranda (2010) developed a probabilistic method for multi-storey
frames. Ramirez and Miranda (2012) further improved upon the PEER performance-based design meth-
odology by explicitly including a new term that accounts for building demolitions due to excessive residual
drifts. Their study assumed, based on engineering judgement, that, on average, a residual drift of 1.5%
would result in building demolition after an earthquake and that the probability of demolition due to
residuals was lognormally distributed with a logarithmic standard deviation of 0.3. Case studies of four-
and twelve-storey concrete moment resisting frames were conducted that suggested that previous method-
ologies significantly overestimate building performance by not considering the effect of residual drift.

Erochko et al. (2011) compared the residual drift response of special moment-resisting frames (SMRFs)
and buckling-restrained braced frame (BRB frames) to develop a method for estimating the residual drifts
in multi-storey structures. They found that BRB frame buildings tended to experience significantly greater
residual drifts than comparable SMRF buildings and that a reasonable estimate of the residual drift may be
attained by finding the maximum drift minus the elastic recoverable drift that results from the elastic
deformations of the inelastic elements. This trend is shown in Figure 2.2. The solid line represents the
upper bound on the drift due to that elastic recoverable drift. They proposed that a better estimation of the
residual drift could be achieved by considering the drift concentration factor, which measures the degree to
which drift is concentrated in specific storeys instead of being spread equally over the height of the build-
ing. The results of this study emphasize the importance of residual drift because they showed that braced-
frame and moment frame buildings with peak drifts that are greater than 1.0% to 1.5% (significantly less
than typical design drift levels) are likely to experience residual drifts greater than 0.5% and that this may,
in turn, result in a building that cannot be occupied after an earthquake.
2.2.3 REDUCING RESIDUAL DRIFTS
Previous residual drift studies have identified that one of the most important factors affecting residual
drift is the post-yield stiffness of the inelastic mechanism (MacRae & Kawashima, 1997; Christopoulos et
al., 2003; Ruiz-Garcia & Miranda, 2006a). To take advantage of this fact, Pettinga et al. (2007) have pro-
posed that residual drifts may be reduced in a structure by purposefully increasing the post-yield stiffness of
the structure. They suggest that this may be achieved by changing the steel reinforcement design in rein-
forced concrete structures or by adding a secondary linear lateral force-resisting system in parallel with the
primary inelastic elements. Numerical modelling results showed that both of these approaches would be
feasible and that they do indeed significantly reduce residual drifts in a building.
The most obvious way to reduce residual drifts and the performance costs that they impose on a system
is to use a lateral force-resisting system that is designed specifically to avoid them. Such systems are called
self-centering systems. A number of different strategies have been developed to reduce or eliminate residual
drifts and these will be described in future sections.
2.3 SELF-CENTERING SYSTEMS VS. TRADITIONAL INELASTIC SYSTEMS

Self-centering systems come in many different forms, but they all generally combine two fundamental
behaviours. The first behaviour is that they contain a mechanism that provides a restoring force to always

5.0%
Buckling-Restrained Braced Frames (BRB Frames)
Data for Stories with Largest Δr
ine)
4.0%
(Circle Size represents Building Height) : 1L
(1
12 Story ma
x
10 Story
8 Story
=Δ
Δr
6 Story
3.0% 4 Story
Residual 2 Story
Interstory
Drift Δr = Δmax - Δel
Δr 2.0%
1.0%
5.0%
Special Moment-Resisting Frames (SMRFs)
Data for Stories with Largest Δr
in e)
4.0%
(Circle Size represents Building Height) :1L
(1
12 Story ma
x
10 Story
8 Story
=Δ
Δr
6 Story
3.0% 4 Story
Residual 2 Story
Interstory
Drift
Δr 2.0%
Δr = Δmax - Δel
1.0%
0 1.0% 2.0% 3.0% 4.0% 5.0%

Peak Interstory Drift Δmax
Figure 2.2: Maximum Residual Drifts in Buckling-Restrained Braced Frames and Special Moment-Resisting Frames
(adapted from Erochko et al., 2011)
recenter the system. The earliest types of self-centering systems utilized the rocking of massive elements
such as bridge piers or concrete walls to provide this restoring force. Other systems, including the SCED
brace, utilize pretensioned cables or tendons. The second fundamental behaviour is that self-centering sys-
tems dissipate energy. This energy dissipation is required to control the accelerations and displacements of
the structure. It may be provided by viscous dampers, yielding steel sections, friction dampers or shape
memory alloys.

The combination of these two behaviours typically produces the flag-shaped hysteretic behaviour
shown on the right side of Figure 2.3. The left side of the figure shows the hysteretic behaviour of a typical
high-performance yielding or friction element such as the buckling-restrained brace that will be discussed
in the next section. A yielding or friction system is likely to experience residual drift at the end of the earth-
quake as previously discussed and shown in Figure 2.2. In a self-centering system such as a SCED brace,
the restoring force brings the displacement of the self-centering system back to zero after every cycle, while
the energy dissipation increases the width of the flag-shaped hysteresis. The energy contained within this
hysteretic shape is equal to the energy dissipated by the system. Too much energy dissipation would cause
the width of the flag to extend past the horizontal axis, compromising the self-centering effect. This occurs
when the restoring force element does not provide enough restoring force to overcome the slip or yield
force of the energy dissipator.
Yielding/Friction System Self-Centering System
Force Force
Maximum Maximum
Elastic Responses Elastic Responses
Hysteretic
Energy
Hysteretic
Energy
Residual No Residual
Figure 2.3: Nonlinear hysteretic response of Yielding/Friction Systems and Self-Centering Systems
2.4 BUCKLING-RESTRAINED BRACES (BRBS)

Before discussing the SCED brace system and other previously designed self-centering systems, this sec-
tion will briefly introduce the system that is generally considered to be the current state-of-the-art seismic
force-resisting system for braced frames: the buckling-restrained brace or BRB.
Typical steel cross-braces in buildings yield in tension but then buckle in compression. Over multiple
cycles, this alternation between yielding and buckling causes strength and stiffness degradation in the
brace. To avoid this degradation, buckling-restrained braces surround a yielding steel core with a buckling
restraint mechanism that prevents the steel core from buckling in compression, allowing it to yield in com-
pression instead (Christopoulos & Filiatrault, 2006). This mechanism allows the BRB to exhibit the full

hysteretic behaviour similar to that shown on the left side of Figure 2.3.
Two types of buckling restrained brace designs are shown in Figure 2.4 which shows two of the BRB
designs that were tested by Tremblay et al. (2006). The design shown at the top of the figure is a typical
mortar-filled tube design for a BRB. In this design, the steel core is restrained from buckling by the outer
steel tube and the concrete or mortar that fills the space between the tube and the core. When the BRB is
in compression, the steel core tries to buckle but is restrained by the mortar (Della Corte et al., 2011). This
creates friction between the mortar and the core which generally means that BRBs have higher strength in
compression than they do in tension (Della Corte et al., 2011). In addition, in compression the cross-sec-
tion of the steel core expands due to Poisson’s effect, which can further increase this friction at the core/
mortar interface (Della Corte et al., 2011). To prevent the mortar from chemically bonding to the steel and
to leave a small gap to allow for the cross-sectional expansion, a de-bonding material is typically located in
between the core and the mortar (Della Corte et al., 2011). A sample hysteretic response attained by Trem-
blay et al. (2008) for one of the mortar-filled BRBs is shown in Figure 2.5. This hysteresis is similar to the
ideal hysteresis shown in Figure 2.3. The lower half of Figure 2.4 shows a sample all-steel BRB design with
no de-bonding material. An all-steel design reduces construction time because there is no need to wait for
the concrete to cure.
With good connection detailing, a recent experimental study by Fahnestock et al. (2007) was able to
achieve a maximum storey drift of 4.8% of the storey height in a four-storey mortar-filled BRB frame sub-
jected to an MCE seismic hazard level earthquake; however, the maximum residual drift in that same test
was 2.7% of the storey height, which would certainly represent a complete loss of the building from an
economic perspective. At the DBE level the maximum drift was 3.0% and the maximum residual drift was
1.3%, which would also likely represent a total building loss from a cost standpoint as previously discussed
in Section 2.2.2.
A more comprehensive review of the current state of buckling-restrained brace technology may be
found in Della Corte et al. (2011).
2.5 SELF-CENTERING ROCKING WALLS AND MOMENT FRAMES

Self-centering rocking walls and self-centering moment frames both operate on a similar principle. As
shown in Figure 2.6, both systems form gaps between their individual elements when they are displaced. In
the rocking wall system, the gap is formed between the wall and the foundation when the system rocks. In
the moment frame, the gaps form between the beams and the columns. These gaps elongate the tendons,

Mortar-Filled Buckling-Restrained Brace (BRB)

Core Member (Plan) Steel Core Connection Stiffener
Buckling-Restraint Member (Plan) Steel Tube
A B
Complete Assembly (Plan)
Section A Section B
Steel Tube Connection

Stiffener
Concrete/
Mortar Fill
Debonding
Steel Core Material
All-Steel Buckling-Restrained Brace (BRB)

Core Member (Plan)
Steel Core Connection Stiffener
Buckling-Restraint Assembly (Plan) Steel Tube Guide Plate
A B
Complete Assembly (Elevation)
Section A Section B
Steel Tube Connection
Stiffener
Guide Plate Tubes Welded to
Shim Plate Guide Plates (typ.)
Filler Plate
Bolted
Steel Core Connection (typ.)
Figure 2.4: Example Buckling-Restrained Braces (BRBs) (adapted from Tremblay et al., 2006)

Figure 2.5: Example Buckling-Restrained Brace Hysteresis (adapted from Tremblay et al., 2006),
which provide a restoring force to bring the system back to the at rest position. Energy dissipation mecha-
nisms take advantage of the concentrated deformation that occurs at the location of the gaps as shown in
the figure. For the rocking walls, the self-weight of the wall provides a portion of the restoring force. If that
self-weight is large enough that it can overcome the yield or slip force of the energy dissipation mecha-
nisms, then the tendons may not be necessary.
For both rocking walls and moment frames, the systems may be made of either steel or concrete. In Fig-
ure 2.6, the rocking wall is shown as concrete, but the concrete wall could be replaced by a stiff steel braced
frame. The moment frame in the figure is shown with steel beams and columns, but they could alterna-
tively be made of precast concrete. The energy dissipation mechanisms may be friction mechanisms as
shown in the figure, yielding bars, viscous dampers, or any other practical system that can dissipate energy
under small deformations.
2.5.1 ROCKING WALLS
After the Chilean earthquake in 1960, it was noticed that many unstable-looking tall water tower struc-
tures surprisingly survived the shaking (Housner, 1963). Housner derived the dynamic behaviour of such a
structure and found that, due to the rocking motion at the base, these tall slender structures were more sta-
ble than they seemed. He also speculated that if a rocking structure had bolts at the base that stretched and
dissipated energy, then the structure would be even more likely to survive; however, he provided no sugges-
tions about how an engineer could purposefully take advantage of this rocking behaviour in design.

Rocking Wall Self-Centering Moment-Resisting Frame

Tendon
Anchorage Tendon
Anchorage
Concrete Wall
or Stiff
Steel Frame Concrete or
Steel Column
Unbonded Energy
Pretensioned Dissipation
At Tendon Mechanism
Tendon
Rest Tendon
Duct
Discontinuity
Concrete or Hole in Column
Energy Steel Beam Flange for Tendon
Dissipation
Mechanism
Foundation Foundation
Gap
Opening
Top
Displaced
Tendon Tendon
Elongates Gap
Opening Elongates
Bottom
Gap
Opening
Gap
Gap Opening
Opening Top
Gap
Opening
Bottom
Figure 2.6: Behaviour of Rocking Walls and Self-Centering Moment Frames
A decade later, Beck and Skinner (1974) presented the design of an A-frame bridge pier that was pur-
posefully allowed to rock or ‘step’ when resisting an earthquake. They proposed the inclusion of a yielding
steel torsion beam as part of the stepping connection to dissipate energy and mitigate the high lateral
deformations that the rocking behaviour caused. Their analysis of the dynamic behaviour of their design
showed that forces in the bridge were significantly reduced and the lateral deformations could be con-
trolled using the steel torsion dampers. A similar design was later incorporated into the design of the Ran-

gitikei railway bridge in New Zealand which opened in 1981 (Skinner et al., 1991). Kelly and Tsztoo
(1977) extended the stepping concept to the base of building frame columns.
The ten-year Precast Seismic Structural Systems (PRESSS) research program was intended to extend
the use of precast concrete in seismic areas (Priestley et al., 1999). One of the systems that the program
developed was a dual pre-cast rocking wall system (Kurama et al., 1999). This system was post-tensioned
to provide a restoring force and energy dissipation was provided by U-shaped flexural yielding plates that
deformed based on the relative movement of the two adjacent precast walls. A prototype five-storey build-
ing with the dual rocking walls behaved well under earthquake loading, experiencing minimal damage
even under the largest hazard scenarios.
The PRESSS program paved the way for other new rocking wall systems for buildings. Palermo et al.
(2005) developed single and dual rocking wall systems for wood buildings using laminated veneer lumber
(LVL) and using similar U-shaped flexural yielding plates to provide energy dissipation. Toranzo et al.
(2009) developed a rocking wall using confined masonry with bending steel energy dissipators. Sause et al.
(2010) developed a rocking steel braced frame with friction energy dissipators. Single and dual rocking
steel braced frames were also further developed by Deierlein et al. (2011). These frames used a new type of
butterfly-shaped steel fuses. Wiebe et al. (2012a; 2012b) extended the rocking frame concept by adding a
second rocking mechanism halfway up the height of the building. This mechanism reduced higher mode
effects in the structure. They also replaced the first storey linear steel brace in the rocking frame with a
SCED brace, which considerably reduced the storey shears and overturning moments.
2.5.2 SELF-CENTERING MOMENT FRAMES
Along with the precast rocking wall described in the previous section, the PRESSS program also devel-
oped a self-centering moment frame using precast concrete beams (Magdy et al., 1999). These beams were
post-tensioned to the columns in a similar way to the frame shown previously in Figure 2.6. The energy
dissipation was provided by debonded reinforcing bars which were grouted into place at the beam/column
interface. These frames experienced higher maximum drifts than comparable concrete moment frames, but
had much smaller residual drifts.
Ricles et al. (2001) applied the self-centering moment frame concept to steel frames. Steel angles con-
nected the top and bottom of the beams to the columns in the same locations as the friction dampers that
are shown in Figure 2.6. When the frame was displaced, these steel angles bent and deformed inelastically,
dissipating energy. This post-tensioned frame provided full self-centering capability. The analytical models

of buildings incorporating this system showed that its performance exceeded that of traditional steel
moment-resisting frames. Christopoulos et al. (2002b) developed a similar concept using energy dissipat-
ing bars instead of yielding angles. These bars were restrained from buckling in compression by steel cylin-
ders, allowing them to have good energy-dissipation characteristics in both tension and compression. Rojas
et al. (2005) modified the self-centering moment frame concept by using friction devices for energy dissi-
pation instead of steel devices. Kim and Christopoulos (2008) further refined the use of friction energy dis-
sipation in self-centering steel moment frames by using a non-asbestos organic (NAO) friction material in
the friction interface. This is the same type of material that is used in the friction dampers for SCED
braces. They also investigated the ultimate behaviour of the frames and demonstrated that the beams could
be detailed to yield at the ultimate limit state, avoiding tension rupture of the tendons.
2.6 PREVIOUS SELF-CENTERING CROSS-BRACES

Four notable self-centering brace designs were developed prior to the SCED brace; however, with the
exception of the fluid restoring force / damping device none of these was able to achieve the level of axial
force and deformation capacity that would make them practical for use in a full-scale structure. These four
braces, the energy dissipating restraint, the fluid restoring force / damping device, the friction spring seis-
mic damper, and the self-centering friction-damping brace will be described in the following sections.
2.6.1 THE ENERGY DISSIPATING RESTRAINT (EDR) (NIMS ET AL., 1993)
The energy dissipating restraint (or EDR) (Nims et al., 1993) is a self-centering bracing member that
was developed at the University of California, Berkeley and was intended to be used for the seismic retrofit
of building structures. A schematic showing the EDR and examples of the brace’s hysteretic behaviour are
shown in Figure 2.7. The body of the brace consists of a steel cylinder that is connected to the end cap on
the left end of the brace in the figure. The right end of the brace is connected to a rod which is housed
within the cylinder. Two sets of friction wedges surround the rod at each of its ends. These friction wedges
provide energy dissipation between the wedges and the outer cylinder. The two sets of wedges are separated
by a large spring. This spring may be precompressed during assembly. At the left end of the rod, there is a
nut that forms a left stop for the friction wedges and at the right end of the rod it becomes thicker to pro-
vide a right stop for the wedges. In addition, there is another set of internal stops affixed to the inner sur-
face of the outer cylinder as shown in the figure.

Figure 2.7: The Energy Dissipating Restraint (from Nims et al., 1993)
When the brace is in compression and the force is large enough to overcome the slip force of the fric-
tion wedges, the rod and the friction wedges shift to the left until the wedges contact the internal stops, at
which point the left friction wedges are held in place, the spring compresses, and the right friction wedge
shifts to the left. When in tension, the rod and wedges shift to the right until the right friction wedges con-
tact the end of the steel outer cylinder at which point the right friction wedges are held in place, the spring
compresses, and the left friction wedge shifts to the right. Two possible hystereses for an EDR device are
shown at the bottom of Figure 2.7. The flag-shaped hysteresis on the left is attained by eliminating the

gaps on the left and right sides of the friction wedges and precompressing the spring. Single-degree-of-free-
dom dynamic numerical analyses performed by Nims et al. (1993) showed that the flag-shaped EDRs were
able to significantly reduce displacement and accelerations when used to retrofit a pre-existing frame.
Unfortunately, since the EDR relies on the stiffness of an axial spring, the concept does not scale well
and it is not practical to design one with an axial capacity that would be large enough to work in a full scale
structure.
2.6.2 FLUID RESTORING FORCE / DAMPING DEVICE (TSOPELAS & CONSTANTINOU, 1994)
The fluid restoring force/ damping device (Tsopelas & Constantinou, 1994) is a self-centering bracing
member that was originally developed for the United States Military in the 1970s but was adapted for use
in combination with base isolation systems for civil engineering structures. In unpublished military appli-
cations, these devices have been built to have axial capacities of up to 1500 kN; however, the devices pre-
sented by Tsopelas and Constantinou (1994) have a maximum axial capacity of only approximately 15 kN.
A schematic showing the fluid restoring force / damping device and an example of the device’s hyster-
etic behaviour are shown in Figure 2.8. Like the EDR, the body of the brace consists of a steel cylinder.
Within that cylinder there is a compartment which contains a compressible silicone fluid. A piston head
slides with the fluid compartment. That piston head contains orifices to allow the fluid to flow between the
two sides of the compartment which are separated by the piston. The movement of this fluid as the piston
head moves within the compartment creates a velocity-dependant viscous force in the brace. The restoring
force in the system is provided by the compressible fluid. The piston rod is thick, so when the piston is
pushed to the right side of the fluid compartment (as shown in the figure), the volume taken up by the rod
increases, which compresses the fluid, providing a restoring force. Likewise, when the piston is pulled to
the left, the volume taken up by the rod decreases. During assembly, the fluid is precompressed so that the
precompression in the fluid must be exceeded before the piston can move. These behaviours combine to
create the overall hysteretic response shown in the lower part of Figure 2.8.
2.6.3 FRICTION SPRING SEISMIC DAMPER (FILIATRAULT ET AL., 2000)
The friction spring seismic damper (Filiatrault et al., 2000) is similar in concept to the EDR but instead
of using friction wedges and an axial spring, the functions of both of these elements are provided by a ring
spring (also called a friction spring). The ring spring concept is shown at the top of Figure 2.9. The ring
spring consists of two sets of rings with mating wedge-shaped cross sections as shown in the figure. Due to

Figure 2.8: The Fluid Restoring Force / Damping Device (from Tsopelas and Constantinou, 1994)
this wedge shape, when the spring is loaded in compression the outer rings are put in elastic circumferen-
tial tension and the inner rings are subjected to elastic circumferential compression. Friction between the
inner and outer rings at their contact surfaces dissipates energy.
The axial spring and friction wedges from the EDR are replaced with these ring springs to get the fric-
tion spring seismic damper as shown in the middle of Figure 2.9. In the figure, the left rod end, tie bar and
sliding sleeve are connected together to form a single part. The left cup and right cup are not connected to
the other parts are allowed to slide within the housing. These two cups are separated and held apart by the

Ring Spring / Friction Spring

Before Loading After Loading
d F
Outer Ring
Inner Ring
Inner Half Ring
Friction Spring Seismic Damper

Sliding Sleeve Friction Springs (typ.)
Rod End
Left Cup Tie Bar Right Cup
Sample Hysteresis
150
100
50
Force
0
(kN)
-50
-100
-150
-30 -20 -10 0 10 20 30
Mean Displacement (mm)
Figure 2.9: Friction Spring Seismic Damper (adapted from Filiatrault et al., 2000)
friction springs, which may be precompressed. This arrangement results in a flag-shaped hysteresis such as
the one shown at the bottom of Figure 2.9.
Although the friction spring seismic damper can resist higher axial loads than the EDR, the axial capac-
ity of the prototype damper is still an order of magnitude lower than the capacity that would be required
for use in full-scale building applications.

2.6.4 SELF-CENTERING FRICTION DAMPING BRACE (ZHU AND ZHANG, 2008)
A third notable self-centering brace that has been developed is the self-centering friction damping brace
(Zhu and Zhang, 2008). This brace relies on the inherent self-centering behaviour of a new class of mate-
rials: shape memory alloys (SMAs). The self-centering friction damping brace uses a simple friction
damper in parallel with wire strands made of a shape memory alloy called nitinol to provide the flag-
shaped hysteretic behaviour as shown in Figure 2.10. In the figure, block A and block B are connected
along a long friction interface with the normal force on the interface being provided by the bolts. The hys-
teretic behaviour of the friction interface alone for different amounts of normal force is shown in the lower
three plots (d), (e), and (f ). When the friction interface slides, the nitinol wire strands elongate. These
strands have an inherent flag-shaped behaviour as shown in the top left plot (a). The combination of the
wires and the friction produce the full flag-shaped hysteretic responses shown in plots (b) and (c).
Figure 2.10: The Self-Centering Friction Damping Brace (from Zhu and Zhang, 2008)

Although the concept for the self-centering friction damping brace works in principle, similar to the
other two previous self-centering braces, the axial capacity is too low for use in a real structure and no full-
scale prototype has been designed or tested.
2.7 THE SELF-CENTERING ENERGY-DISSIPATIVE (SCED) BRACE (CHRISTOPOULOS ET AL.,

2008)
As the previous section demonstrated, prior self-centering braces have all shared the same problem: they
are difficult to scale up to the axial capacities that are necessary for them to be used in a full scale building.
The self-centering energy dissipative (SCED) brace (Christopoulos et al., 2008) solves this problem by
reversing the self-centering mechanism of the EDR. Instead of relying on a spring in compression to pro-
vide the self-centering restoring force, which has a low stiffness and a low capacity for precompression, the
SCED brace uses a cable or tendon in tension to provide this restoring force. This allows the use of rela-
tively high stiffness, high strength tendons which can accommodate the high axial capacities that are neces-
sary for a building cross-brace application. The mechanics of the SCED brace will be briefly described
below; however, these mechanics will be described in much greater detail in the next Chapter.
2.7.1 THE SCED BRACE CONCEPT
The mechanics of the original SCED brace were first explained by Christopoulos et al. (2008) and are
depicted in Figure 2.11. The brace mechanism consists of four main elements: (1) an inner steel member
(usually an I-beam section or a steel tube), (2) an outer steel member (usually a steel tube), (3) an energy
dissipating device that activates based on the relative movement of the inner and outer members, and (4) a
set of tendons that are pretensioned and which axially clamp both the inner and outer members at both
ends via a pair of free end plates. Although the figure shows the inner and outer members adjacent to one
another, in reality the inner member is usually smaller in cross-section and is located within the outer
member. At one end of the brace, the structure is connected to the inner member, and at the other end of
the brace, the structure is connected to the outer member. Examples of energy dissipating devices that may
be used with a SCED brace are: metallic yielding devices, friction slip devices, viscous dampers, viscoelastic
dampers, or shape memory alloys.
The flag-shaped hysteretic behaviour is achieved through the interaction of the four main elements as
shown in Figure 2.11. The tendons are pretensioned during assembly and, therefore, provide a restoring
force that constantly pulls all of the elements back toward their initial positions. Thus, when the brace is at
rest (a) there is a tension force in the tendons and a compression force in the inner and outer members.

Hysteretic Response
With Hysteretic Damper
Yield/Slip Force = F
a) No Load Hysteretic, Viscous or Shape P
Outer Member Memory Alloy Energy Dissipation Pp0+F
Tendon 2F
Anchor
Tendon (Preload = Pp0 ) δ

End Plate
Inner Member -(Pp0+F)
b) Initial Stiffness P
P
c) Tension Energy Dissipator P

Brace Strain = +δ Strain = +δ
Outer Member Force = F P
Tendon Force = Pp0+P - F δ
Inner Member Force = F

Tendon Strain = +δ
d) Initial Stiffness P
e) Compression P
Brace Strain = -δ Energy Dissipator
Strain = -δ
Outer Member Force = P P
Inner Member Force = P

Tendon Strain = +δ
Figure 2.11: SCED Brace Mechanics (adapted from Christopoulos et al., 2008)
Due to the arrangement of the brace elements, the tendons elongate relative to their initial pretensioned
length regardless of whether the brace is in tension or compression. When the brace is in tension (c), the
right end plate abuts the outer member and separates from the inner member; however, the left end plate
remains stationary because it is held in place by the inner member. When the brace is in compression (e),
the opposite case occurs: this time the left end plate abuts the outer member and separates from the inner
member while the right end plate remains stationary because it is held in place by the inner member. The
initial stiffness in the hysteretic response (b) and (d) is caused by of all of the elements deforming together
until the tension or compression force overcomes the pretension in the tendons, allowing the right or left

end plate to separate from the inner member. Before this can happen though, the axial force in the brace
must also overcome the slip or yield force in the energy dissipater. After the pretension force plus the slip or
yield force is exceeded, the SCED brace has 'activated' and the stiffness of the brace is approximately equal
to the stiffness of the tendons alone. The energy dissipating device causes the width of the flag hysteresis.
This width is equal to 2F in Fig. 2 where F is the slip or yield force of the energy dissipating device. In
order for the brace to be fully self-centering, the only criterion is that the pretension in the tendons ( P p0 )
must be greater than the activation/yield force of the energy dissipation mechanism ( F ).
2.7.2 PRIOR PROTOTYPES (CHRISTOPOULOS ET AL., 2008)
The first SCED brace prototype was a proof of concept. The brace design, test setup and sample test
results are shown in Figure 2.12. This brace consisted of: an inner steel tube, an outer steel tube, four ara-
mid tendons which are pretensioned against free steel end plates that abut the ends of the inner and outer
tubes, and two friction interfaces that dissipate energy based on the relative movement of the inner and
outer tubes. The aramid tendon behaviour and the design of the friction interfaces will be described in the
next chapter. The stable friction damper hysteretic behaviour is shown in the bottom right plot. The
amount of friction damping may be varied to create the different self-centering brace hystereses as shown
in the top row of plots. This prototype proved that the SCED brace concept worked when subjected to
quasi-static loading and also showed that high axial strength capacity up to at least 800kN was feasible.
Following the initial prototype test, a second SCED brace prototype was built that was tested within a
full-scale steel braced frame as shown in Figure 2.13. The design of this SCED brace was similar to the first
prototype except that it was longer and that it was equipped with an external friction fuse in series with the
SCED brace. This end fuse allowed the lateral deformation of the system to exceed the lateral deformation
that the brace itself could accommodate, which was approximately equivalent to 2.0% of the storey height.
The test frame with the SCED brace was tested both statically and dynamically and it was subjected to
both qualification protocols and earthquakes deformation histories. Sample test results are shown at the
bottom of Figure 2.13. The bottom left plot shows the storey hysteresis for a limited deformation protocol
test, which produced a symmetric flag-shaped response. The two plots in the middle of the bottom row
show the storey hystereses for two earthquake deformation histories, one with a balanced deformation
demand in the positive and negative directions and the other with an unbalanced response, with all of the
brace activation occurring in the negative direction. The second earthquake shown caused a maximum
interstorey drift of 2.0%, which is the brace’s maximum deformation capacity. The plot on the bottom

Figure 2.12: The First SCED Brace Prototype (from Christopoulos et al., 2008)

right shows the hysteretic response of the storey when subjected to a large amplitude quasi-static deforma-
tion protocol. This plot shows the response of the end fuse, which slips when the storey reaches 2.0% drift,
changing the stiffness of the system to zero and limiting the force in the brace. As this plot shows, the acti-
vation of the end fuse caused some residual drift in the storey. This end fuse is only likely to slip under the
largest maximum-considered earthquakes. The second prototype test showed that the SCED brace worked
properly when subjected to realistic dynamic loading and within a full-scale realistic steel frame.
2.7.3 DYNAMIC MODELLING
Two significant building modelling projects have been conducted which have assessed the dynamic per-
formance of buildings equipped with SCED braces. The first was performed by Tremblay et al. (2008)
who compared the response of SCED buildings to that of buildings equipped with buckling-restrained
braces (which were previously discussed in Section 2.4). They studied two dimensional building frames
that had five different heights between two and sixteen storeys. These frames were subjected to suites of
earthquakes at three different seismic hazard levels. The results showed that the SCED braced frame had
lower peak storey drifts than the BRB frame. They also found that the SCED frames had no residual drift
at the low and medium seismic hazard levels and experienced much lower residual drifts than the BRB
frames at the highest hazard level. Floor accelerations were found to be significantly larger in the SCED
braced frames due to the sharp stiffness transitions assumed in the modelling of the SCED braces.
A second major analytical study of SCED-braced frames was performed by Choi et al. (2008). This
study also compared SCED and BRB buildings, but expanded upon the previous study by comparing
these to special moment-resisting frames (SMRFs) and also by studying both two and three dimensional
models, including the effect of building torsion. Similar to the other study, these analyses showed that the
SCED brace had much better residual drift response than the other two systems even though the maxi-
mum drift response for the three systems was similar; however, also like the previous study, this study
showed that the SCED braced frames had higher accelerations than the other two systems. The source of
the high accelerations that are encountered in these models will be addressed in Chapter 6.
2.8 IMPETUS FOR THE RESEARCH / NECESSARY IMPROVEMENTS

As was clear from the discussion of the development of previous self-centering braces in Section 2.6, in
order for a brace to be useful in a real building, it has to be practical to build one with a sufficiently high
axial capacity. Previous SCED brace prototypes have achieved axial capacities that are an order of magni-

Figure 2.13: The Second SCED Brace Prototype (from Christopoulos et al., 2008)

tude larger than those of previous self-centering braces, up to approximately 800kN. While this capacity is
large enough for use in some structures, for SCED braces to be used in taller or larger buildings they must
be available in the same range of axial capacities as buckling restrained braces. To address this limitation,
the first goal of this thesis was to design, build and dynamically test a SCED brace that has an axial capac-
ity of approximately 3000kN. This goal was completed successfully and the high-capacity (HC-SCED)
brace will be discussed in Chapter 4.
The second major goal of the thesis was to eliminate the need for a friction fuse in series with the
SCED brace to accommodate high seismic hazard earthquakes. To achieve this goal, a SCED brace was
developed that could accommodate storey drifts of approximately 4% of the storey height, which is equiv-
alent to approximately double the brace deformation demand achieved by previous prototypes. This was
done by designing a new configuration for the SCED brace that had a telescoping mechanism (the T-
SCED brace). This T-SCED brace prototype was constructed, and tested dynamically in a full scale verti-
cal steel test frame. It successfully achieved a storey drift of 4% as will be discussed in Chapter 5.
The third major goal of the thesis was to address the absence of any system-level testing of SCED
braced frames. Previous prototypes were tested either axially, or as part of a single-storey frame, driven with
a demand displacement. Neither of these types of tests accounts for the effect of dynamic inertial forces or
for the interaction between SCED braces at different levels of the structure. To address this, a one-third
scale shake table test of a three-storey SCED braced frame was designed and constructed. This test con-
firmed that multi-storey SCED frames respond as expected to earthquake shaking at a system level and
that the numerical models of those frames did a good job predicting their dynamic behaviour. In addition,
these tests provided an opportunity to investigate the high accelerations encountered in the models that
were studied by Tremblay et al. (2008) and Choi et al. (2008). The shake table tests will be discussed in
Chapter 6.
Tests of the prototype SCED and T-SCED braces described in Chapters 4 to 6 revealed that the initial
stiffness estimates determined using the simplified equations described by Christopoulos et al. (2008) did
not match the measured initial brace stiffnesses that were observed during testing. It seemed that complex
brace mechanics caused by inexact SCED brace member lengths were causing this disconnect between the
predicted and measured stiffnesses. Therefore, the final goal of the thesis was to develop analytical and
numerical tools to better predict and model the behaviour of the SCED brace and of the newly-developed
T-SCED brace. To this end, the idealized mechanics of the SCED brace were derived and a new software
tool was created to take into account the effect of construction length tolerances on the behaviour of the

brace. These will be presented in Chapter 3. Then, to assess the effect of these more realistic SCED brace
initial stiffnesses on the behaviour of a SCED braced frame, a six storey building numerical model was
constructed and tested using various different SCED brace initial stiffness values. The model results
showed that the initial stiffnesses did not have a large effect on the building behaviour. The numerical
model also permitted comparison between the behaviour of the original SCED braces and the enhanced-
elongation T-SCED braces described above. The T-SCED braces were found to perform better than com-
parable traditional SCED braces. In addition, previous SCED braces have used friction energy dissipators
exclusively, so these models were also used to investigate the feasibility of using viscous dampers instead of,
or in combination with, friction dampers. It was found that the use of viscous dampers could result in sig-
nificant performance improvements in terms of drift, accelerations and base shear. These numerical model
results will be discussed in detail in Chapter 7.
2.9 OTHER RECENTLY DEVELOPED SELF-CENTERING BRACES BASED ON THE SCED

In parallel with the research that will be discussed in this thesis, two other notable projects have been
conducted elsewhere that extend upon the concept of the SCED brace. The first is a modification of the
SCED brace concept that uses a SCED brace with shape memory alloy rods instead of tendons to provide
the self-centering restoring force to the system and uses a BRB instead of friction dampers for energy dissi-
pation (Miller et al., 2012). Chou and Chen (2012) developed a dual-core SCED system that is similar to
the T-SCED that will be discussed in Chapter 5; however, their brace was only tested statically and could
only consistently achieve equivalent storey drifts of approximately 2.0%, which is similar to the perfor-
mance achieved by the original SCED brace prototypes presented by Christopoulos et al. (2008).

CHAPTER 3: MECHANICS, MODELLING AND DESIGN OF THE
ORIGINAL AND NEW TELESCOPING SCED (T-SCED)
BRACES
The previous chapter introduced the SCED brace and described the function of each of its components
and, conceptually, how those components interact to produce a flag-shaped hysteretic behaviour. In this
chapter, these SCED brace components will now be described in detail to provide context for an exposition
of the precise mechanics and behaviour of the SCED brace.
Christopoulos et al. (2008) previously introduced the SCED brace concept and presented a simplified
step-wise derivation of the SCED brace hysteretic response. This simplified hysteretic behaviour will be
summarized in this chapter; however, it does not consider the effect of the differing stiffness of the inner
and outer members. This stiffness difference produces multilinear stiffness transitions in the hysteretic
response, which in turn results in a lower effective initial stiffness for the SCED brace. To investigate this
effect and to attain a more complete closed-form solution for the ideal SCED brace response, this chapter
will present a new, full step-wise derivation of the SCED brace hysteresis.
To improve upon the SCED brace concept, a new telescoping configuration (referred to as the T-SCED
brace) will be proposed that doubles the deformation capacity of the brace. A prototype of this brace sys-
tem was designed and tested which will be discussed in Chapter 5. The mechanics of this new T-SCED
brace configuration are significantly more complex than those of the original SCED brace. Therefore,
instead of attempting to derive a prohibitively complex closed-form solution to describe the T-SCED hys-
teretic behaviour, a software tool called the SCED Mechanics Simulator was created in HTML and Javas-
cript to directly model the detailed mechanics of the brace using a nonlinear incremental stiffness method
analysis.
35
Chapter 3: Mechanics, Modelling and Design of the Original and New Telescoping SCED (T-SCED) Braces 36
This software tool was also designed to be able to model the effect of construction tolerances by permit-
ting the main axial members to have different lengths in the model. To take advantage of this, the effect of
member length construction tolerances on the hysteretic behaviour of SCED and T-SCED braces was
investigated and an method was developed to estimate the effective initial stiffness of original SCED braces
for a given member length difference.
In addition to modelling the detailed mechanical behaviour of the SCED and T-SCED braces them-
selves, it is also important to be able to model these braces efficiently as part of a larger building structure.
The nonlinear structural analysis package Ruaumoko (Carr, 2005a) has the ability to model a self-center-
ing flag-shaped hysteresis; however, modelling an external friction fuse, which may be desired in series with
a SCED brace, is difficult and often results in unstable models. In North America, many research engineers
prefer to use the nonlinear structural analysis package OpenSees (McKenna et al., 2000) to model dynamic
systems subjected to earthquake excitations. OpenSees is an open-source package which allows users to
construct custom material types. Since no self-centering material previously existed for use in OpenSees, a
new material was developed that included both the flag-shaped hysteretic behaviour of the SCED brace
and the ability to model an external friction fuse in series with the brace.
After discussing the individual components of SCED and T-SCED braces, the braces’ detailed hyster-
etic behaviour, and a software tool that may be used to reliably simulate that behaviour, this chapter will
present some of the key design considerations for the design of SCED-braced frames. These will include
the determination of the effective initial stiffness, challenges involved with the use of force-based design
and the effect of non-structural components. A sample design procedure will also be presented to provide
some guidance in order to minimize the amount of design iterations that are required to develop a full
SCED brace design.
The mechanics, modelling and design of SCED and T-SCED braces presented in this chapter will pro-
vide the context necessary to discuss the design and behaviour of the experimental SCEDs and T-SCEDs
that will be introduced in Chapters 4, 5 and 6 and the OpenSees building model that will be described in
Chapter 7.
3.1 THE COMPONENTS OF A SCED BRACE

Having previously introduced and described the concept and mechanics of the SCED brace previously
in Section 2.7.1, this section will describe each of the SCED’s key elements: (1) the inner, intermediate and

outer members, (2) the internal friction damper, (3) the optional external friction fuse, (4) the tendons and
tendon anchors, (5) the end plates, and (6) the connection to a braced frame.
Full final designs of the different SCED brace prototypes that were developed for this thesis will be
described in Chapters 4, 5 and 6. To illustrate the arrangement of the different brace elements in a real
SCED brace, a sample schematic of the newly-designed high-capacity SCED brace (the HC-SCED) that
will be described in Chapter 4 is shown in Figure 3.1. The novel aspects of this brace design will be dis-
cussed in detail in that chapter.
7.7m
Internal ly
Friction Damper e mb
Ass
te
ple
Friction Interface
Plate (welded to m
Inner Section) Co
External
Friction Fuse
ly
Interface B
emb
Tendon Ass
Inner Member er
Outer Member e mb
terM
Friction Ou
Interface
Angle
Internal Friction bly
Section A
Interface Angle ss em
Internal Friction Damper be rA
M em
In ner
Plastic Guide
Friction Interface es
Plate (with Stainless P lat
Steel Surface) End
&
m bly
A sse
don
Ten
Section B Tendon
External Friction Fuse
Tendon Anchor
End Plate
Figure 3.1: Sample SCED Brace Design

3.1.1 AXIAL MEMBERS
When a SCED brace is activating in compression, most of the external applied axial force on the brace
travels through the axial members and tendons in series. In the portion of those members that are between
the internal friction dampers, the force in the members is lessened because some of the external force trav-
els through the friction dampers instead of the axial members and tendons; however, the ends of the mem-
bers experience the full externally-applied axial force. Therefore, for the purposes of design, the inner and
outer members are generally designed for the full axial force capacity of the brace.
The critical design case for the axial members is typically their resistance to compression buckling. The
effective length of the member is assumed to be equal to its total length. This is a conservative assumption
because each axial member will have some bending restraint at one end at the connection.
Due to the need to design each axial members for compression buckling for the full axial capacity of the
brace, the axial members tend to require a large quantity of steel. This large amount of steel could theoret-
ically be used to construct a high-capacity conventional steel brace instead of a SCED brace; however, even
if this conventional brace could resist the full earthquake force elastically, which would technically make it
self-centering, it would dramatically increase the capacity design forces for the rest of the structure by up to
seven times (if the force-reduction factor for a SCED brace is assumed to be seven). This would drastically
increase the structural cost. If the alternative conventional brace could not be designed to resist the full
earthquake force elastically, then it would buckle in compression, resulting in residual drifts and eliminat-
ing the primary benefit provided by the SCED brace.
3.1.2 INTERNAL FRICTION DAMPER
As previously mentioned, a number of different devices may be used in a SCED brace to provide
energy-dissipation such as metallic yielding devices, friction slip devices, viscous dampers, viscoelastic
dampers, or shape memory alloys; however, all of the previously designed and constructed SCED proto-
types, and all of the SCED and T-SCED braces that were designed as a part of this work, use friction slip
devices to provide the energy dissipation. Friction mechanisms are desirable because a well-designed fric-
tion damper has a slip load that is easy to control and does not become damaged during cyclic loading. A
SCED brace that is equipped with friction slip devices for energy dissipation is essentially a damageless sys-
tem up to its design deformation level. This avoidance of damage is particularly attractive for laboratory
testing since it allows the SCED braces to be tested many times without recalibration or retrofit of the
energy-dissipation mechanism.

The internal friction damper design for use in SCED braces was developed by Kim et al. (2004) to pro-
vide energy dissipation for his self-centering steel moment-resisting frames. The resulting friction interface
design was previously adapted to be used in the first SCED prototypes (Christopoulos et al., 2008). The
general layout of a friction damper as developed by Kim et al. (2004) is shown in Figure 3.2. These damp-
ers consist of six primary elements; (1) and (2): the inner and outer interface plates form the backbone of
the interface, with the inner plate connected to one side of the interface and the outer plates connected to
the other side. The inner plate has a slot to accommodate the movement of the bolts that provide the nor-
mal force. Two outer interface plates are used to avoid eccentricity in the connection and to provide two
sliding surfaces for the friction interface. (3): A friction material is cut into pads which are then set into
identically-sized recesses which are machined into the outer interface plates. A small amount of epoxy is
used to hold each pad inside its recess. (4): Stainless steel surfacing plates are welded to either side of the
inner interface plate. These plates provide a well-controlled sliding surface for the friction pads. (5): High
strength steel bolts hold the connection together, enforce the interface alignment and provide the normal
force to the friction interface. These bolts pass through holes in the outer plates and the slot in the inner
plate. (6) Washer plates spread the normal force from the bolts so that it extends across the surface of the
friction pads.
The friction damper shown in Figure 3.2 uses two normal force bolts. Each bolt is partnered with two
friction pads on each side of the inner interface plate. Since both the friction material and the bolts have
limited axial stress capacities, for a given friction material and friction pad size the slip force in the connec-
tion is proportional to the number of bolts. A minimum of two bolts are used so that the interface slips in
a straight line. The length of the slot in the inner interface plate must be designed to accommodate the
amount of friction damper deformation that is required by the SCED brace design.
Kim et al. (2004) tested multiple different friction interface materials by fabricating test interfaces that
had a similar layout to the interface shown in Figure 3.2. All of the materials that he tested were non-asbes-
tos organic (NAO) friction materials produced by Carlisle Brake and Friction (2012). A sample of the
resulting friction damper hystereses for the different materials are shown in Figure 3.3. For the self-center-
ing moment frame application, Kim chose the NF-916 material because it had a sufficient axial stress
capacity and the most stable hysteretic response with respect to deformation velocity and temperature. All
of the previous SCED prototypes and all the SCED and T-SCED prototypes that will be presented in this
thesis also use the NAO NF-916 friction material for the internal friction dampers.

Hole through Washer Plate

Stainless Steel and Outer Interface Plate
Surfacing Plate
Outer Friction
Inner Friction Interface Plates
Interface Plate
Slot in Inner
Non-Asbestos Organic (NAO) Interface Plate
Friction Material
Machined Recess in
High-Strength Outer Interface Plate
Bolt
Steel Bolt Washer Washer Plate
Inner Outer
Friction Friction
Interface Interface
Plate Plate
Stainless Steel Slot in Inner Machined Recess in

Surfacing Plate Interface Plate Outer Interface Plate
Tension Sliding Compression Sliding
Figure 3.2: Friction Damper General Design
In SCED brace designs, the inner interface plate of the friction damper is connected to the inner mem-
ber as shown in Figure 3.1. To facilitate assembly of the friction interface, the outer interface plates for
SCED braces are generally replaced by steel angles that are connected to the outer member as shown in
Figure 3.1. An example of a fully-designed outer interface angle and inner interface plate are shown in Fig-
ure 3.4. This figure shows the friction damper design for the T-SCED brace that will be discussed in Chap-
ter 5. The geometry of the friction pad and the recess that holds the pad are shown in the figure. Each
recess is designed to have a depth equal to half the thickness of the friction pad. The holes and slots were
sized to accommodate 3/4” high strength steel bolts, and the holes in the angles are actually short vertical

Figure 3.3: Friction Interface Material Tests (from Kim et al., 2004)
slots in order to accommodate outer and inner member size and shape tolerances. The end of the steel
angle is cut on an angle to accommodate the movement of the brace within the test frame.
The outer interface angles in the sample connection were sized to carry the friction slip force via shear
in the angle leg. The inner interface plate was also designed for the total plate shear caused by the friction
slip force. The distance between the end of the slots and the edge of the plate was designed to be long
enough to resist half of the total friction slip force in shear due to the friction from the pads located above
the slot.
The slip force in the internal friction dampers was calculated and calibrated by assuming simple Cou-
lomb friction behaviour in the interfaces whereby:
F It =  s F N (Eq. 3.1)
where F It is the slip force of the damper,  s is the coefficient of static friction and F N is the total normal
force on the connection provided by the bolts. As Figure 3.3 shows, the slip behaviour of the NF-916

Outer Friction Interface Angle (other angle is a mirror image)

Plan
89 L152x89x12.7
A 79 313
Elevation
9.5 Slotted 22 diameter hole

152 64 Hole for 19mm bolt
30
41
70
A
721
Section A
13
3D View
33
3.2 Recess
Recess Depth
30
152
Slot
32
0.8 between
Recess Slot and Recess
30
26
13
89
Inner Friction Interface Plate Stainless Steel Surfacing Plate
Elevation
22 diameter slot 22 diameter slot
121
184 60
Slot 283
Slot 283
2 Plan
476
Friction Pad (NF-916 NAO Friction Material)

Plan
30
R6.4
6.4 Elevation
70 All dimensions in mm
Figure 3.4: Friction Damper General Design (T-SCED Prototype)

material is very stable. The static slip force is roughly equal to the dynamic slip force, therefore the static
coefficient of friction  s and the dynamic coefficient of friction  d are assumed to be equal. For the NF-
916 pads, Kim et al. (2004) found the coefficient of friction to be between 0.11 and 0.15. Tests of the
interfaces that were used in the prototype SCED and T-SCED braces that will be described in Chapters 4,
5 and 6 found that the coefficient of friction of the NF-916 pads was approximately 0.15 under practical
conditions.
Knowing the coefficient of friction, the only remaining unknown element in the interface is the normal
force provided by the bolts F N . The NF-916 pads have a nominal axial stress capacity of 136 MPa, mean-
ing that the pair of pads that straddle each bolt as shown in Figure 3.4 have a total axial capacity of approx-
imately 571kN. The A325 3/4” steel bolts that are used in the interface have a specified minimum tensile
strength of 177kN and a factored tensile resistance of 141kN, both of which are much lower than the
capacity of the friction pad. This margin of safety is important to ensure that the friction pads do not dete-
riorate when there are subjected to considerable shear stress that the friction pads in combination with the
axial stress from the bolt. If the bolt is torqued to 70% of its specified minimum tensile strength, the nor-
mal force will be 124kN. Using Equation 3.1 with an assumed coefficient of friction of 0.15, this gives a
slip force per bolt of 19kN. Use of higher strength 3/4” A490 bolts increases this slip force per bolt to
approximately 25kN.
For the SCED and T-SCED prototypes that will be discussed in Chapters 4, 5 and 6, the normal force
in the interior friction damper bolts was controlled using a combination of bolt torque measurement using
a torque wrench and direct bolt elongation measurement using a C-shaped frame equipped with a dial
gauge. The bolt torque was only used to determine the bolt load approximately. The bolts were preten-
sioned by first loading them up to a target torque value and then adjusting the load in the bolt using the
direct elongation measurement. The Machinery’s Handbook (Oberg et al., 2008) provides approximate
bolt torque values for ASTM A325 and A490 bolts to achieve 60-70% of their specified minimum tensile
strength:
2.922 + 2.893 log d b

For A325: T b = 10 (Eq. 3.2)
3.092 + 3.057 log d b

For A490: T b = 10 (Eq. 3.3)
where T b is the torque in foot-pounds, and d b is the nominal diameter of the bolt in inches. For the 3/4”
steel bolts used in the sample friction damper, Equations 3.2 and 3.3 give bolt torques of 364 ft-lb (494

Nm) for the A325 bolts and 513 ft-lb (696 Nm) for the A490 bolts. The direct elongation measurement of
the bolt was used to calculate the axial load by using the following equations from the Machinery’s Hand-
book (Oberg et al., 2008):
 b EA b
P b = --------------
- (Eq. 3.4)
LB
where P b is the axial force in the bolt,  b is the elongation of the bolt, E is the Young’s modulus of steel,
A b is the nominal cross-sectional area of the bolt, and L B is the effective bolt length given by the equation:
d ts H HN
L B =  -----  L S + ------B- + L J – L S + -------
- (Eq. 3.5)
 db   2  2
where d ts is the minimum cross-sectional area of the bolt caused by the threads, d b is the bolt shank diam-
eter, L S is the length of the unthreaded shank of the bolt within the joint, L J is the total joint length (from
the inner side of the bolt head to the inner side of the nut), H B is the height of the bolt head, and H N is
the height of the nut. This effective bolt length takes into account the effect of the threaded portion of the
bolt, the deformation of the head of the bolt, and the deformation of the portion of the bolt that is within
the nut.
The bolt deformation was measured using the custom designed bolt-measurement device shown in Fig-
ure 3.5. This measurement device consisted of a C-shaped steel frame equipped with a 0.0001”
(0.0025mm) precision dial gauge. The dial gauge position was adjustable using two threaded rods. The
dial gauge probe and the dead-end probe both had a domed ends (not shown in the figure) which fit into
chamfered holes in both ends of the bolt (as shown the in lower half of the figure). All of the internal fric-
tion damper bolts were machined and measured ahead of time to determine their unloaded length.
3.1.3 OPTIONAL EXTERNAL FRICTION FUSE
In past SCED prototypes, the external friction fuses were designed using the same methods that were
used to design the internal friction dampers. The main difference for those external friction fuses is that
they were designed for much higher slip loads because they must resist the full axial load of the brace at the
maximum design displacement. Due to this higher load, the NF-336 friction material (see previous Figure
3.3) was often used instead of the NF-916 material because the NF-336 material has a higher axial load
capacity. Although the NF-336 material has a less stable hysteretic response, its use was justified because

Figure 3.5: Bolt Length Measurement Device
the external fuse is not generally expected to undergo more than a few cycles of slip, even when subjected
to extreme maximum-considered seismic hazard level earthquakes.
The high-capacity SCED that will be described in Chapter 4 uses a new external friction fuse design
that does not use any non-asbestos organic friction material (this new external friction fuse is shown in pre-
vious Figure 3.1). The development and design of this fuse will be fully described in that chapter. The low-
capacity SCEDs that were used in the shake table tests described in Chapter 6 used a more conventional
external fuse design that uses the NF-916 friction pad. Figure 3.6 shows the torquing setup for the external
friction fuse bolts for the shake table test SCED braces.

Figure 3.6: Shake Table Test External Fuse Bolt Torquing Setup
3.1.4 LONG-TERM BEHAVIOUR OF THE FRICTION DEVICES
The use of both internal friction dampers and external friction fuses in SCED braces requires an aware-
ness and understanding of the long-term behaviour of friction devices. Long-term studies of the behaviour
of the friction interfaces in SCED braces have not been performed and such tests are not within the scope
of this thesis; however, five main characteristics of the SCED brace and the friction interfaces mitigate the
effect of the uncertainty about the long-term friction behaviour. First, the friction interfaces use a special-
ized and engineered friction material (Carlisle NF-916 or NF-336 as described in Sections 3.1.2 and 3.1.3)
and stainless steel sliding surfaces. This reduces the uncertainty associated with the behaviour of the fric-
tion materials themselves and reduces the possible material degradation over the life of the device. Second,
prior to installation, SCED braces may be tested axially to confirm their hysteretic behaviour. This removes
any uncertainty with respect to the initial brace response. Third, wind simulation tests that will be
described in Chapters 4 and 5 will show that the friction devices do not experience any significant wear
due to wind loading that may occur prior to any brace activation due to an earthquake. This removes the
uncertainty associated with premature wear over the life of the brace. Fourth, for the internal friction
dampers, the friction itself always contributes less that half of the overall activation force. This means that
even if the friction were to vary by 25%, the overall brace activation force would change by less than 13%.
To reduce the uncertainty associated with the behaviour of an external fuse, the use of an external fuse may
be avoided by using the T-SCED brace concept that will be introduced in Section 3.3 and tested in Chap-
ter 5. Fifth, to accommodate any remaining uncertainty in the use of friction devices with SCED braces,

an upper bound / lower bound approach for the friction coefficient may be used, similar to the approach
that is currently used for base isolator design as described by Christopoulos and Filiatrault (2006).
The above characteristics reduce the uncertainty associated with the long-term behaviour of the friction
devices but do not completely eliminate it. Some work remains to be done in characterizing the long term
friction behaviour. In particular, the long term effect of bolt relaxation on this type of friction device con-
figuration is not known. In addition, it is not known how much the specialized friction material itself is
susceptible to long term creep, which would increase the bolt relaxation. These are both important ques-
tions which should be addressed in future research.
3.1.5 TENDONS AND TENDON ANCHORS: BEHAVIOUR AND DESIGN CONSIDERATIONS
The tendons for all SCED braces, (previous prototypes as well as the new SCED and T-SCED proto-
types that will be described in Chapters 4, 5 and 6), are made out of a parallel lay, straight fibre Technora
aramid material. The Technora material is a high strength, relatively high-modulus, fatigue- and heat-resis-
tant, polymer that is manufactured by Teijin Aramid in Japan and assembled into straight fibre cables of
various diameters by Linear Composites Ltd. in the U.K (Teijin Aramid, n.d. B; Linear Composites Ltd.,
n.d.). In the form of a woven yarn cable, the fibres have a stated tensile strength of 3400 MPa, and a
Young’s modulus of 78 GPa (Teijin Aramid, n.d. B), which is approximately half as stiff as steel; however,
as will be seen below, in the straight fibre tendons that are used for SCED braces the response of the ten-
dons is not consistently linear. The fibre has a listed tensile elongation capacity of 4.6% (Teijin Aramid,
n.d. B), making it attractive for use in SCED braces, which require a large elongation capacity to function
effectively. This is the reason why the aramid fibres were originally chosen for the SCED braces instead of
steel tendons or glass/carbon composites; typical steel braced-frame geometries require the SCED to use
tendons that have an elongation capacity on the order of 1.5-2.0% in order for the frame to be able to
accommodate reasonable frame lateral displacements. More traditional materials are not capable of this
level of elastic strain.
–6
The coefficient of thermal expansion for Technora fibres is – 6 10 m/m per C , meaning that the
material contracts when heated (Teijin Aramid, n.d. C). This is significantly different than steel, which has
–6
a coefficient of thermal expansion of approximately 12 10 m/m per C (CISC, 2003). In a SCED
brace, the Technora tendons are pretensioned against steel axial members, which are put into compression
by the tendons. Therefore, if the temperature increases, the tendons will try to contract and the steel mem-
bers will try to expand. These thermal strains will result in mechanical strains in the tendons and axial

members. Both the tendon contraction and the steel expansion have the effect of increasing the pretension
force in the tendons and the resulting precompression force in the axial members. Likewise, a drop in tem-
perature will decrease the force in the tendons. For a 30C increase or decrease in temperature, this ther-
mal expansion effect will increase the effective tendon strain by approximately 0.05% strain. If a tendon
with a strain capacity of 1.7% is pretensioned to 25% of its nominal capacity, this 30C increase or
decrease would therefore result in an approximate 11% change to the tendon pretension. If a SCED brace
was installed outdoors in a cold climate, these temperature effects have the potential to be significant and
should be considered in the brace design; however, SCED braces that are installed within a building enve-
lope should not experience such drastic changes in temperature during their operational life, and they may
therefore be neglected.
Technora fibres are resilient at high temperatures. At 250°C, the tendons retain more than half of their
strength and thermal decomposition does not start until 500°C (Teijin Aramid, n.d. C). The ignition tem-
perature of the fibres is 600°C (Teijin Aramid, n.d. C). When a SCED brace is at rest, the tendons are pre-
tensioned to only a fraction (~30%) of their maximum strength, therefore the tendons should perform well
in a fire.
The Technora properties brochure (Teijin Aramid, n.d. C) describes the fibre as having low creep and
low stress relaxation and high dimensional stability. Creep and relaxation plots are provided with a time
range of up to one thousand hours and these do indeed show that the short-term creep and relaxation of
the Technora fibres is low. Recently, the first major studies on the long-term creep of Technora fibres were
conducted by Giannopoulos and Burgoyne (2012a&b). They assessed the long-term creep behaviour of
the fibres by using a conventional one year creep test as well as two different accelerated creep methods.
The conventional one year creep test showed that fibres with lower levels of sustained axial stress experi-
enced higher creep after one year than those with higher levels of axial stress did. One test that applied a
sustained stress equal to 20% of the average breaking load for the fibres experienced a creep of 0.28% after
one year (0.23% after two months). At 30% of the breaking load the creep was 0.23% after one year
(0.20% after one month), for 50% it was 0.19% after one year (0.17% one month) and for 65% it was
0.15% after one year (0.14% after two months) (Giannopoulos & Burgoyne, 2012a&b). Technora ten-
dons in SCED braces are typically designed to have sustained loads in the range of 25% to 40% of the
assumed nominal breaking load; the nominal breaking load is approximately 10% to 15% less than the
average breaking load. The tests by Giannopoulos and Burgoyne (2012a&b) showed that the Technora
fibres exhibit nonlinear viscoelastic behaviour at sustained loads below 40% of the average breaking load,

meaning that the material creeps faster at those lower levels of axial load. Above that 40% point, they
exhibit a linear viscoelastic behaviour. Their accelerated creep tests were only conducted for loads above
50% of the average breaking load to avoid the nonlinear viscoelastic behaviour range. These accelerated
tests seem to suggest that for a sustained load of 55% of the average breaking load of the Technora fibres, a
total creep strain of 0.25% over one hundred years is possible. This creep is small compared to the total
elastic elongation of approximately 2.2% that is present in the fibres at 55% of the average breaking load;
however, this creep has the potential to significantly relax the pretension in the tendon. Due to the chang-
ing stiffness behaviour of the SCED brace tendons that will be discussed, it is not clear what the quantita-
tive effect of this creep is on the tendons used in SCED braces. This is an critical avenue for future study.
The tendons are anchored at each end by aluminum spike and barrel anchors. The anchor design is
shown in Figure 3.7. When the tendons and anchors are assembled, the tendon fibres are arranged evenly
around the circumference of the spike. The spike and tendon fibres are then slipped into the barrel. The
fibres are held in place by friction between the spike, the fibres and the barrel. The tension on the tendon
fibres pulls the spike into the barrel (towards the left in the figure), which, in turn, increases the clamping
force on the fibres, keeping them in place. The tendon anchors have threads machined on the inside of the
barrel to allow a threaded rod to pull on the anchor to facilitate pretensioning of the tendon. They also
have threads machined on the outside of the barrel to accommodate a nut that locks the tendon in place
after pretensioning. The nut is typically either round or hexagonal. In Figure 3.7 some selected dimensions
are provided for the tendon anchors for the two sizes of tendons that have been used in SCED braces,
which have either a 17mm or 22mm nominal diameter.
Tendon Anchor 17mm [22mm] Anchor Nut Anchor Nut

may be circular
Threads for or hexagonal
Tendon Anchor Spike Nut
Barrel
18 [23] 50 [68]
Tendon
Fibres Enter Threads for
Here Pretensioning
206 [272] 80 [95]
Figure 3.7: Tendon Anchor

Photos of a completed tendon and of the tendon anchors are shown in Figure 3.8. The assembled ten-
don at the top of the figure shows that the aramid fibres are contained within a black polyethylene sheath
to protect them from abrasion and ultraviolet light. This short tendon has a 17mm nominal diameter. The
tendon anchors shown at the bottom of the figure are for the larger 22mm nominal diameter tendons. The
anchor on the left is shown without the tendon and the point of the anchor spike may be seen protruding
from the bottom. The anchor on the right has a tendon installed which is cut so that the tendon cross sec-
tion is visible. This cross section shows the tendon fibres on the inside, surrounded by the black polyethyl-
ene sheath.
Figure 3.9 shows a sample test result for a single 22mm tendon (Kim, Personal Communication Janu-
ary 5, 2010). This test result demonstrates the nonlinear behaviour of the Technora tendons. During the
first loading of the tendon, the stiffness is relatively low; however, when the tendon is unloaded, the stiff-
ness becomes significantly higher. If the tendon is reloaded, this higher stiffness is maintained until the pre-
vious maximum load is exceeded, at which point the stiffness returns to the low initial value until the
tendon is unloaded again. This stiffening behaviour is caused by molecular-level realignments within the
polymer (Christopoulos et al., 2008). Since the tendons must undergo cyclic loading during the operation
of a SCED brace, the higher effective cyclic stiffness is the governing stiffness for design. In order to
achieve linear tendon behaviour in SCED braces, when the tendons are installed, they must be first loaded
up to the maximum load that they are expected to experience during the operation of the brace, and then
released to the design pretension load. This has the added benefit that all of the tendons in a SCED brace
are effectively tested up to their maximum load during the pretensioning process, which reduces the uncer-
tainty in the tendon behaviour. If the tendon exceeds this maximum load during the operation of the
brace, the linear cyclic response portion of the hysteresis will shift to the right as shown in the figure, result-
ing in a loss in the tendon pretension. The figure also shows that below a certain load, approximately 25%
of the nominal strength, the unloading stiffness drops and is no longer linear. For this reason, the preten-
sion load of the tendon is generally not designed to be less than 25% of the nominal strength. In addition,
it is clear from the test results that, while the total tendon strain at the nominal tendon strength is approx-
imately 4.4%, due to the stiffening behaviour the usable strain within the linear cyclic response portion is
on the order of 1.5-2.0%.
The Technora tendon behaviour may be generalized as shown in Figure 3.10. Based on test results, the
initial Young’s modulus of the tendon E p0 is approximately 44 GPa and the effective cyclic Young’s modu-
lus E p0 is approximately 102 GPa. The maximum design load for the tendons is generally designed to be

Figure 3.8: Tendon and Tendon Anchor Photos
85-95% of the total nominal tendon strength in order to provide a factor of safety for tendon failure. The
lock-off point, which is equivalent to the tendon pretension load P p0 , is generally limited to be within the
range of 25-40% of the nominal tendon strength P piu . The lower limit of this range avoids the nonlinear
unloading behaviour below 25% and the upper limit conserves tendon deformation capacity between the
lock-off point and the maximum load. For the design of a SCED brace, the tendon may be assumed to
behave linearly after pretensioning, with a stiffness equal to the effective cyclic modulus E p . Using this
stiffness, the effective ultimate axial deformation of the tendons (  pu ) may be determined by assuming

22mm Diameter Technora Tendon Test

1.2
Nominal Tendon Strength 300kN

1.0
As load exceeds previously

0.8 attained level, linear cyclic
Normalized range shifts to the right
Strength Low Initial Stiffness
0.6 Higher Effective
Cyclic Stiffness
0.4
Lock-off Load
0.25(300kN)=75kN
0.2 Reloading
Unloading
0 1 2 3 4 5 6
Tendon Strain (%)
Figure 3.9: Sample 22mm Tendon Test (adapted from Kim, Personal Communication January 5, 2010)
that the tendon unloads linearly to zero load as shown by the dashed line in the figure. Since the tendon
load never drops below the pretension load during the operation of a SCED brace, the deformation capac-
ity and the stiffness measured from this point define the tendon’s full effective behaviour.
Axial
Force 100% Ppiu (Nominal Strength)
Pp
85-95% Ppiu
Ep Ap /Lp
Assumed Linear
Pp0 = 25-40% Ppiu Cyclic Behaviour
Unloading Lock-off Point

Ep0 Ap /Lp
Reloading
Deformation δp
δpu
Figure 3.10: Generalized Tendon Behaviour

Since the tendons in a SCED brace are tensioned up to their maximum possible deformation during
the pretensioning process, they are unlikely to fail during operation of the brace. If a tendon does fail dur-
ing operation, the axial force capacity of the brace will immediately drop since the capacity provided by
that tendon will have been removed. Since lateral building deformations are inertia-driven during earth-
quakes, the axial force demand may remain constant after the failure of a tendon. This has the potential to
lead to the failure of others tendons in sequence; however, since the rupture of a tendon would be accom-
panied by a loss of stiffness, the dynamic behaviour of the system would also change when a tendon rup-
tures. This may partially mitigate a progressive failure by modifying the inertial force demand. To reduce
the chance of SCED brace failure due to a progressive rupture of the tendons, SCED braces are always
designed to be equipped with at least four tendons. This provides some redundancy to the system. In com-
parison, a traditional yielding system such as a BRB consists of a single force resisting element; therefore, if
it exceeds its maximum deformation capacity, it will fail completely.
Once SCED brace tendons are installed, they are inaccessible and do not require any maintenance.
3.1.6 END PLATES
Since the end plates have a complex geometry with multiple holes and discontinuities, the cannot gen-
erally be considered to be simple beams or plates. Therefore, the end plates for the SCEDs and T-SCEDs
in Chapters 4, 5 and 6, were designed using a simple linear finite element model. This allowed the design
to consider the full geometry of the plate as well as realistic boundary conditions. Each end plate was ana-
lyzed using to two different boundary conditions, each of which corresponded to the outline of one of the
two axial members that the end plate was designed to abut. These boundary conditions on the plate were
applied through contact with a stub model of the relevant axial member which was fixed in place. The lay-
out of the holes and the size of the plate were determined based on geometric constraints and the finite ele-
ment model was used to determine the plate thickness. Some sample analysis results for the end plates of
the T-SCED brace that will be described in Chapter 5 are shown in Figure 3.11. The stub members that
provide the boundary conditions for the end plates are not shown in the figure. The darkest shade in both
of the plots represents a Von Mises stress of 250 MPa. Since the analyses were linear, a small amount of
minor yielding at the edges of the holes was permitted, but the end plates were otherwise designed to
remain elastic.

Figure 3.11: Sample Linear Finite Element Analysis Results for T-SCED End Plate Design
3.1.7 CONNECTIONS AND ACCOMMODATION OF FRAME ROTATION
For erection, the connections at both ends of a SCED brace are generally bolted to gusset plates in the
corners of a steel braced frame. On the inner member connection side (left in Figure 3.1) the SCED brace
connection generally consists of a single thick plate that is welded to the inner member and passes through
a slot in the end plate at that side. This connection plate must resist the full tension and compression load
of the brace and is designed for net section fracture, block shear, axial buckling and shear lag where appro-
priate (if necessary at the connection between the plate and the inner member). On the outer member con-
nection side, the connection is generally formed by extending the outer interface angles from the friction
connection. These angles are already welded to the outer member and are extended past the end plate and
drilled with holes to connect to a gusset plate. Like the inner-side connection, the outer-side connection
must resist the full tension and compression load of the brace and is designed for net section fracture, block

shear, axial buckling and shear lag. The connections at both sides of the brace must also be designed to
accommodate the rotations that are imposed on them by the frame geometry.
For the design of the connections to accommodate frame rotation, the orientation of the connections is
an important consideration. This is especially true for the outer-side connection, where the angles on the
top and bottom surfaces of the outer member may be separated from each other by a significant distance
(up to half a metre apart). If this connection is oriented vertically, the large separation may cause the con-
nection to have a significant moment resistance and it may be necessary to introduce a true pin to release
this moment. This issue was encountered in the design of the T-SCED brace and will be described in detail
in Chapter 5.
3.2 ORIGINAL (O-SCED) BRACE MECHANICS

The behaviour of the individual components described in the previous section combine to produce a
self-centering hysteresis brace response as described previously in Section 2.7.1. This section will outline
two closed-form step-wise solutions for the hysteretic behaviour of a SCED brace, each of which is for a
different level of complexity. First, the traditional simplified solution will be presented, which was previ-
ously summarized by Christopoulos et al. (2008). Then, a new full solution to the hysteretic behaviour will
be presented which includes the effect of differing member stiffnesses and non-simultaneous slippage of
the friction dampers. The layout and behaviour of the main SCED brace elements are shown again in Fig-
ure 3.12 for reference in the following sections.
3.2.1 SIMPLIFIED STEP-WISE HYSTERETIC RESPONSE OF THE SCED BRACE
The hysteretic behaviour that results from a common simplification of the mechanics of the SCED
brace is shown in Figure 3.13 and described below. This derivation is adapted from Christopoulos et al.
(2008). These simplified mechanics are derived with the assumption that both end plates separate from the
inner or outer members simultaneously. As will be seen in the derivation of the detailed SCED mechanics,
this is not actually realistic; however, the use of this assumption provides an approximation of the hysteretic
shape that is sufficiently close to the true shape to suffice for design. It will be assumed that the energy dis-
sipation is provided by a friction slip mechanism with a total slip force F and zero post-slip stiffness.
Starting from the origin of the hysteresis in tension, the initial stiffness of the brace is k 1 which is the
sum of the stiffness of the three axial elements in the brace: the inner and outer members and the tendons.
k1 = ki + ko + kp (Eq. 3.6)

Hysteretic Response
No Load Hysteretic, Viscous or Shape P
Outer Member Memory Alloy Energy Dissipation Pp0+F
Tendon 2F
Anchor
Tendon (Preload = Pp0 ) δ

End Plate
Inner Member -(Pp0+F)
Initial Stiffness P
P
Tension Energy Dissipator P

Brace Strain = +δ Strain = +δ
Outer Member Force = F P

Tendon Strain = +δ
Initial Stiffness P
Compression P
Brace Strain = -δ Energy Dissipator
Strain = -δ
Outer Member Force = P P

Tendon Strain = +δ
Figure 3.12: SCED Brace Mechanics (adapted from Christopoulos et al., 2008)
where k i and k o are the stiffnesses of the inner and outer members, respectively, and k p is the total stiff-
ness of the pretensioning tendons. These values are equal to the axial stiffness of each element:
Es Ai Es Ao E p A pt
k i = ---------- k o = ---------- k p = ------------
- (Eq. 3.7)
Li Lo Lp
where E s is the Young’s modulus of steel, E p is effective cyclic Young’s modulus of the pretensioning ten-
dons, A i and A o are the cross-sectional area of the inner and outer members, respectively, A pt is the total
nominal cross-sectional area of all of the pretensioning tendons, and L i , L o and L p are the lengths of the

P Optional k1
External r=
Fuse Slip ka
Activation Pu
ka+
1 k1 βPa
Pa 1
k1
1 Reverse
1
ka+ Activation
δ
Gaps Closed
1
k1
Pa
- 1
ka
Pu
Figure 3.13: SCED Brace Simplified Hysteresis
inner and outer members and tendons, respectively. For this simplified derivation, the lengths of the three
elements are assumed to be exactly equal.
As previously mentioned, the brace activates and changes stiffness when the total external force on the
brace P overcomes the initial tendon pretension P p0 plus the friction slip force F ; however, there is a
small additional term in the activation force expression that takes into account the elongation of the ten-
dons when the system has reached the activation load. Together, these terms give an brace activation force
P a equal to:
P a = P p0 + F + k p  a (Eq. 3.8)
where  a is the deformation of the entire brace at activation which is equal to:
P
 a = -----a (Eq. 3.9)
k1
Substituting Equation 3.9 into 3.8 gives:
Pa
P a = P p0 + F + k p  ----- (Eq. 3.10)
 k 1
P p0 + F
P a = -----------------------------------------  P p0 + F (Eq. 3.11)
kp
1 –  --------------------------
 k i + k o + k p

Since k p is generally much smaller than k i and k o , the term in the denominator may generally be omitted,
providing the simplified equation shown on the right side of Equation 3.11. The deformation of the brace
at activation  a is therefore equal to:
Pa P p0 + F
 a = -----  -------------------------- (Eq. 3.12)
k1 ki + ko + kp
After brace activation in the tension direction, a gap opens between the right end plate and the inner
member and a corresponding gap opens between the left end plate and the outer member as shown previ-
ously in Figure 3.12. After that point, the only remaining stiffness in the brace is provided by the tendons.
+
Therefore, the post-activation stiffness in the tension direction k a is equal to the stiffness of the tendons
kp :
+
ka = kp (Eq. 3.13)
+
If the brace is extended post-activation up to a given deformation  , the axial force in the brace at that
deformation will be:
+ +
P = P p0 + F + k p  (Eq. 3.14)
This expression is exact because it includes the entire deformation of the tendons including the deforma-
tion that they experienced pre-activation.
When the SCED brace reaches its ultimate load capacity P u , which has an upper limit that is generally
dependant on the load capacity of the tendons, some SCED brace designs include an external friction fuse
in series with the brace to extend its deformation capacity. The hysteretic behaviour that results from the
inclusion of such a fuse is shown in Figure 3.13. When the external friction fuse activates, the stiffness of
the combined system becomes zero and all of the additional applied deformation is taken in the fuse. The
full magnitude of the deformation of the fuse results in a permanent deformation in the system. Thus, the
extra deformation capacity that the fuse provides comes at the cost of residual deformation in the system.
The hysteretic response of the SCED brace itself is unaffected by the external fuse slip, the only change
being that the entire hysteresis shifts in the x-direction by a distance that is equal to the deformation in the
external fuse.
When the load reverses at the peak and returns towards the origin of the hysteresis, the friction mecha-
nisms stop slipping and must be activated in the reverse direction before the brace can self-center. During
this process, the friction mechanisms are locked and the brace reverts to its initial stiffness k 1 (Equation

3.6). Since the friction mechanism must both unload and then reload in the opposite direction before the
brace can return to zero, the vertical width of the flag hysteresis is equal to 2F . Generally, it is convenient
to express this width in terms of the activation force P a such that:
2F = P a (Eq. 3.15)
2F 2F
 = ------  ----------------- (Eq. 3.16)
P a F + P p0
where  represents a percentage of the activation force. The parameter  is called the energy dissipation
capacity parameter. The convenience of the  parameter is that as long as   1.0 , the SCED brace will be
fully self-centering. If   1.0 , then there will be some residual drift when the brace returns to zero load
because the tendon pretension force cannot completely overcome the slip force of the friction mechanism.
+
If the brace attained a deformation of  before reversing direction, then when both friction dampers slip
again and the brace stiffness decreases (shown in Figure 3.13 as the “Reverse Activation” point), the defor-
mation of the brace is equal to:
+ 2F P a
 =  –  a =  – ------  ----- =  – --------------------------
+ + 2F
(Eq. 3.17)
P a  k 1 k +k +k i o p
and the force at that point is equal to:
2F P
P = P – 2F –  a k p = P p0 + F + k p  – 2F – ------  -----a  k p 
+ +
(Eq. 3.18)
P a  k 1
kp
P = P p0 – F + k p  – 2F  --------------------------
+
(Eq. 3.19)
 k i + k o + k p
Assuming again that k p is generally much smaller than k i and k o :
+ +
P  P p0 – F + k p   P – 2F (Eq. 3.20)
at that “Reverse Activation” point shown in Figure 3.13.
After the friction mechanism slips and the brace activates in the reverse direction, the stiffness of the
brace is once again equal to the stiffness of the tendon alone ( k p ).When the deformation returns to the ini-
tial stiffness branch of the hysteresis near the origin, the gaps between the end plates and the inner and
outer members close. At this point, the force in the brace is:
P = P a – 2F  P p0 – F (Eq. 3.21)

and the deformation of the brace is:
P a – 2F P p0 – F
 =  a  1 –   = ------------------  -------------------------- (Eq. 3.22)
k1 ki + ko + kp
The initial stiffness of the brace is then recovered and the brace can accumulate load in order to activate
in the compression direction. Since all of the axial elements of the brace (the axial members and the ten-
dons) deform together when the brace is not activated, the activation load and deformation in the com-
pression are the same as they were for the tension direction. After the brace activates in the compression
direction, though, the post-activation stiffness is slightly different. As shown in Figure 3.12, when the
brace is deforming in compression, the inner and outer members and the tendons are acting in series.
-
Therefore, the true post-activation stiffness of the brace in compression k a is given by the series stiffness
expression:
- 1
k a = --------------------------- (Eq. 3.23)
1 1
---- + ---- + ---- 1
ki ko kp
Continuing to determine the other points on the hysteresis using this expression for the activated stiffness
involves an increased level of complexity because the deformations now depend on the location of the one
or more friction mechanisms in the brace. For example, if the friction mechanism is located at the left end
of the inner and outer members (at the end where the inner member is restrained), then the post-activation
-
deformation in the compression direction  is given by:
- - -
- P – P o0 P – P i0 – F P – P p0 – F
 = ------------------- + ---------------------------- + ----------------------------- (Eq. 3.24)
ko ki kp
where P o0 and P i0 are the initial precompression load in the outer and inner members, respectively, due to
-
the pretensioned tendons, and P is the total compression force in the brace. Since k p is generally much
smaller than k i and k o , the hysteresis in the compression direction may be simplified by assuming that the
series stiffness in Equation 3.23 may be simplified to:
-
ka  kp (Eq. 3.25)
This assumption causes the hysteretic response of the brace to become symmetric, where the positive and
negative quadrants of the hysteretic response are mirror images of each other. Alternatively, the series stiff-
ness from Equation 3.23 may be used to directly estimate the force and deformation in the brace on the

compression side, by assuming that when the brace is activated, the full force is transferred through all the
series elements such that:
- - -
P = –Pa – ka   + a  (Eq. 3.26)
where P a and  a are the positive values of the activation force and deformation derived for the tension
side of the hysteresis.
3.2.2 FULL STEP-WISE HYSTERETIC RESPONSE OF THE SCED BRACE
In reality, during the activation of the SCED brace, the end plates at either end of the brace do not sep-
arate from the inner or outer members simultaneously. This is caused by the difference between the inner
and outer member stiffness and the locations and slip forces of the friction mechanisms. This makes the
behaviour of the brace more complex than the simplified response introduced in the previous section. The
activation of the brace does not actually occur at a single point; there is a multilinear transition between the
initial stiffness and the post-activation stiffness as shown for an example SCED brace design in Figure
3.14. While these transitional stiffnesses may generally be significantly different from each other, their
effect is not easily apparent in a SCED hysteresis when it is shown at a larger deformations magnitude as
shown on the right side plot in the figure. The left side plot shows the same hysteretic response at small
deformations so that the transitional stiffnesses may be easily seen. As this plot shows, the stiffness is not
only multilinear during the forward activation of the brace, but it is also multilinear during the other
branches of the response, such as the “reverse activation” point and the point where the end plate gaps close
and the inner and outer members re-contact the end plates. To fully characterise these effects, a full deriva-
tion of the SCED brace mechanics has been developed. A similar procedure was previously implemented
in a design spreadsheet; that spreadsheet calculated the hysteretic response numerically and did not include
a full exposition of the force and deformation equations that describe the response. The stages of the new
derivation are shown all together in Figure 3.15 and the resulting force and deformation equations are
given in Tables 3.1 to 3.15 (Equations 3.27 to 3.153). This derivation assumes that the inner and outer
members have different stiffnesses, but are the exact same initial length. It also assumes that the outer
member is stiffer than the inner member ( k 0  k i ). While this may seem counter-intuitive since the design
of these axial members is typically governed by global buckling and they both have the same design load,
geometric constraints and available section sizes typically result in the outer member having a larger cross-
sectional area than the inner member, making the outer member stiffer. This has been true for all of the

SCED brace designs in this thesis. Additionally, the derivation assumes that the energy dissipation is pro-
vided by two friction mechanisms, one at each end of the brace, which each have a slip force of F  2 , giv-
ing a total slip force of F .
Sample SCED Brace Hysteresis from Detailed Mechanics

ki = 645kN/mm, ko = 1129kN/mm, kp = 29.2kN/mm, F = 760kN, Pp0 = 890kN
2000 4000
Small-Amplitude 3 Large-Amplitude 3
2
1500 Deformations 1
3000 Deformations
Pu = 1700kN Pu = 3500kN 4
on
1000 2000
2 sion 5
n si
1 Ten
Brace 500 Te 4 1000
Axial 7 6 5 7 6
Force 0
5
0 6 7
P 6 7
-500 4 -1000
ion
(kN) 1
1-First Activation
ssion
pre
s
2-Full Activation 2
res
5
-1000 3-Ultimate Force -2000 m
Co
mp
1 4
4-First Reverse Activation
Co
-1500 5-Full Reverse Activation -3000

3 2 6-First End Plate Re-Contact
7-Full End Plate Re-Contact 3
-2000 -4000
-3 -2 -1 0 1 2 3 -80 -60 -40 -20 0 20 40 60 80
Brace Deformation δ (mm) Brace Deformation δ (mm)
Figure 3.14: SCED Brace Detailed Hysteretic Response
For each direction of loading (tension and compression), the hysteretic behaviour of the brace has seven
key points as shown in Figure 3.14. Each of these points is associated with a transition between the differ-
ent behavioural stages shown in Figure 3.15. Each stage shown in the Tables 3.1 to 3.15 has a criterion
associated with it that dictates when the stage transition occurs. The value of the total axial force P in both
the figure and the table is positive when the brace is in tension and negative when it is in compression.
Up to the first activation point (Step 1 - Tables 3.2 and 3.3), the stiffness of the brace is equal to the
sum of the stiffness of all three axial elements (the inner and outer members and the tendons). The first
activation in the brace occurs when one of the two internal friction dampers reaches the slip force of F  2 .
This point occurs on the tension side of the hysteresis when the force in the outer member is also equal to
F  2 . As Figure 3.15 shows, if the outer member is stiffer than the inner member, as previously described,
then the friction damper that is closest to the connection for the inner member (on the left in the figure)
will slip first. This is because the external force is distributed in each axial element according to their rela-
tive stiffnesses up to the first activation. Activation of that left internal friction damper allows the inner and
outer members to deform separately and therefore allows a gap to open between the left end plate and the

Outer Member Energy Dissipation

Tendon
Anchor
Tendon
End Plate
Inner Member
Tension Compression
Outer Member Force Compression Negative
Friction Damper Total External
Force F/2 F/2 (T) <F/2 P Force >P+F/2 (C) Force
P(-)
1-First
Activation
<F/2 (T) Inner Member F/2 P+F/2 (C) <F/2
Tension
Force
F/2 (T) P F/2 P+F/2 (C) F/2 P(-)
2-Full
Activation
F/2 F/2 (T) F/2 P+F/2 (C)
F/2 F/2 (T) F/2 P F/2 P+F/2 (C) F/2 P(-)
3-Ultimate
Force P-F (T) P-F (T)
F/2 F/2 (T) F/2 Tendon Force F/2 P+F/2 (C) F/2
F/2 -F/2 (C) <F/2 P F/2 -(Pp -F/2) (C) <F/2 P(-)
4-First
Reverse Pp (T)
Activation
F/2 >-F/2 (C) <F/2 F/2 <-(Pp-F/2) (C) <F/2
F/2 -F/2 (C) F/2 P F/2 -(Pp -F/2) (C) F/2 P(-)
5-Full
Reverse Pp (T)
Activation
F/2 -F/2 (C) F/2 F/2 -(Pp-F/2) (C) F/2
δp- δp0 = δo- δo0 δp- δp0 = δo- δo0
P P(-)
6-First
Re-Contact
P P(-)
7-Full
Re-Contact
δp- δp0 = δi- δi0 δp- δp0 = δi- δi0
Figure 3.15: SCED Brace Detailed Mechanics - Behavioural Stages
outer member. This chain of events is similar for the first activation on the compression side. The left fric-
tion damper still activates first; however, this time the gap on the left opens between the end plate and the
inner member and it is therefore the inner member which reaches the stage criterion of P + F  2 when the
friction damper slips. The activation criterion for compression has the full external applied force P (which
is negative) summed with F  2 (which is positive) because in compression the elements are in series, mean-
ing that the full force must always travel through all three elements, even when the gaps are open.
The full activation point (Step 2 - Tables 3.4 and 3.5) is reached in the brace when both friction inter-
faces reach the force of F  2 . In tension, this point occurs when the inner member attains a force of F  2 .

Table 3.1: Full Solution Step 0 - Initial Tendon Pretension

0 - INITIAL TENDON PRETENSION
Stage Criterion: P p = P p0
Force Deformation
Inner P p0 k i P p0
Member P i = – --------------
- (Eq. 3.27)  i0 = – --------------
- (Eq. 3.28)
ki + ko ki + ko
Outer P p0 k o P p0
Member P o = – --------------
- (Eq. 3.29)  o0 = – --------------
- (Eq. 3.30)
ki + ko ki + ko
Tendon P p = P p0 (Eq. 3.31) P p0
 p0 = ------- (Eq. 3.32)
kp
TOTAL P = 0 (Eq. 3.33)  = 0 (Eq. 3.34)
Table 3.2: Full Solution Step 1T - First Activation (Tension)

1 - FIRST ACTIVATION (TENSION)
F/2 F/2 (T) <F/2 P
Stage Criterion: P o = F  2
Transitional Stiffness: k eff = k i + k o + k p
F/2 <F/2 (T) <F/2
Force Deformation
Inner F k F
P i =  ---   ----i  (Eq. 3.35)  i = ------- (Eq. 3.36)
Member  2   k o 2k o
Outer Po = F  2 (Eq. 3.37) o = i (Eq. 3.38)

Member
Tendon k p   F   k p
P p = P p0  1 + --------------  p = P p0  ---
1-
+ --------------- + -------
1 F
- + --- ---- (Eq. 3.39)  k p k i + k o 2k o
(Eq. 3.40)
 k i + k o  2   k o
TOTAL P p0 k o F  k i + k o + k p P p0
P =  --------------
- + --- -------------------------- (Eq. 3.41)  = -------------- F
- + ------- (Eq. 3.42)
 k i + k o 2  ko  k i + k o 2k o
In compression, it occurs when the outer member attains a force of P + F  2 . Between the first and full
activation points, the stiffness is a transitional value which is less than the initial stiffness but greater than
the post-activation stiffness, as shown in the table and in Figure 3.14. On the tension side, this transitional
stiffness is equal to the sum of the inner member stiffness and tendon stiffness, with the outer member not
accumulating any force between first and full activation. For the compression side, the transitional stiffness
is equal to the inner member stiffness alone, since only that inner member accumulates force during that
stage as shown in figure 3.15. Since the tendon stiffness is much smaller than the inner member stiffness,

Table 3.3: Full Solution Step 1C - First Activation (Compression)

1 - FIRST ACTIVATION (COMPRESSION)
F/2 >P+F/2 (C) <F/2 P(-)
Stage Criterion: P i = P + F  2 ,
F/2 P+F/2 (C) <F/2
Force Deformation
Inner ki
P i =  ---------------    i =  ---------------  
1
(Eq. 3.43) (Eq. 3.45)
Member  ko + kp   ko + kp 
k i + k o + k p F 
 – P  ------------------------- k i + k o + k p F 
 – P  -------------------------
- – --- - – ---
 p0  k i + k o  2   p0  k i + k o  2 
Pi = P + F  2 (Eq. 3.44)
Outer ko  o = i (Eq. 3.47)
P o =  ---------------  (Eq. 3.46)
Member  ko + kp 
k i + k o + k p F 
 – P  -------------------------
- – ---
 p0  k i + k o  2 
Tendon k i + k o + k p  k o  P p0  k i + k o + k p  k o 
P p = P p0  -------------------------
- --------------- (Eq. 3.48)  p = ------- -------------------------- --------------- (Eq. 3.49)
 ki + ko   ko + kp  kp  ki + ko   ko + kp 
kp 
– ---  --------------
F
– ---  --------------- 
- F 1
2 ko + kp 
 2  ko + kp 
TOTAL k i + k o + k p ki  F 
P =  -------------------------  =  ---------------  – P p0  --------------
1
-  (Eq. 3.50) - – --- (Eq. 3.51)
 ko + kp   k o + k p   k i + k o 2 
ki  F 
 – P  --------------- – ---
 p0  k i + k o 2 
the transitional stiffness between the first activation point and full activation is approximately the same in
tension and compression. As the table shows, the detailed expression for the force at which full activation
occurs is approximately equal to the sum of the tendon pretension P p0 and the total internal friction
damper slip load F for both the tension and compression sides of the hysteresis (Equations 3.59 and 3.68).
This is the same as the simplified formulation discussed in Section 3.2.1 (Equation 3.11).
Once the brace has activated, gaps have opened up at both ends of the brace and the brace deforms
under a much lower stiffness. On the tension side, this stiffness is equal to the tendon stiffness alone, and
on the compression side the stiffness is equal to the stiffness of all three axial elements in series. Since the
tendon stiffness is always significantly smaller than the inner and outer member stiffnesses, this series stiff-
ness is approximately equal to the tendon stiffness alone as discussed in Section 3.2.1. The brace and ele-

Table 3.4: Full Solution Step 2T - Full Activation (Tension)

2 - FULL ACTIVATION (TENSION)
F/2 F/2 (T) F/2 P
Stage Criterion: P i = F  2 ,
Transitional Stiffness: k eff = k i + k p
F/2 F/2 (T) F/2
Force Deformation
Inner Pi = F  2 (Eq. 3.52) F
 i = ------- (Eq. 3.53)
Member 2k i
Outer Po = F  2 (Eq. 3.54) F
 o = ------- (Eq. 3.55)
Member 2k o
Tendon kp   F   kp 
P p = P p0  1 + --------------  p = P p0  ---- + --------------- + -------
1 1 F
- + --- ---- (Eq. 3.56)  k p k i + k o 2k i
(Eq. 3.57)
 k i + k o  2   k i 
TOTAL P p0  k i + k o + k p  P p0 F
P = --------------------------------------
- (Eq. 3.58)  = --------------
- + ------- (Eq. 3.60)
ki + ko k i + k o 2k i
F kp + ki
+ ---  --------------- + ---
F
2  ki  2
If k p « k i k o , then P  P p0 + F (Eq. 3.59)
ment forces and deformations when the brace reaches its ultimate force (Step 3) are shown in Tables 3.6
and 3.7.
After the peak force has been reached and the deformation changes direction, the internal friction
dampers lock-up, the brace returns to its initial stiffness and the SCED brace must be activated in reverse
before it can return to zero deformation. This ‘reverse activation’ (Steps 4 and 5 - Tables 3.8 to 3.11) was
previously shown in Figure 3.14. The external load in the brace does not have to reverse direction to
accomplish this reverse activation because of the restoring force that is present in the tendons. Similar to
the first and full activation in the forward direction (Steps 1 and 2), the reverse activation occurs when the
friction dampers each attain their slip force of F  2 . Also similar to the forward activation point, if the
outer member is stiffer than the inner member, the interface closest to the inner member connection (the
left hand side in the figures) will slip first. On the tension side of the hysteresis, the criterion for first reverse
activation (Step 4) is simple, it occurs when the outer member reaches a compression force of – F  2 , and
then full reverse activation (Step 5) occurs when the other member also reaches this compression load. On
the compression side, the criterion for reverse activation is more complex. Equilibrium around the left fric-
tion damper when the force in it reaches – F  2 dictates that when it slips, the sum of the force in the outer

Table 3.5: Full Solution Step 2C - Full Activation (Compression)

2 - FULL ACTIVATION (COMPRESSION)
F/2 P+F/2 (C) F/2 P(-)
Stage Criterion: P o = P i = P + F  2 ,
Transitional Stiffness: k eff = k i
F/2 P+F/2 (C) F/2
Force Deformation
Inner ko ko
P i =  ---------------    i =  ---------------   ---- 
1
(Eq. 3.61) (Eq. 3.62)
Member  ko + kp   ko + kp   ki
k i + k o + k p F 
 – P  ------------------------- k i + k o + k p F 
 – P  -------------------------
- – --- - – ---
 p0  k i + k o  2   p0  k i + k o  2 
Outer Po = Pi (Eq. 3.63)

 o =  ---------------  
1
(Eq. 3.64)
Member  ko + kp 
k i + k o + k p F 
 – P  -------------------------
- – ---
 p0  k i + k o  2 
Tendon k i + k o + k p  k o  P p0  k i + k o + k p  k o 
P p = P p0  -------------------------
- --------------- (Eq. 3.65)  p = ------- -------------------------- --------------- (Eq. 3.66)
 ki + ko   ko + kp  kp  ki + ko   ko + kp 
kp
– ---  --------------- 
F
– ---  --------------- 
F 1
2 ko + kp 
 2  ko + kp 
TOTAL k i + k o + k p  k o 
P = – P p0  -------------------------
- ---------------
 =  --------------- 
(Eq. 3.67) 1
 ki + ko   ko + kp  (Eq. 3.69)
 k o + k p
F 2k o + k p
– ---  ------------------- 
2  ko + kp 
2
  k i k p – k o – k o k p F  k o 
 P p0  ------------------------------------ – ---  ---- 
  ki  ki + ko   2 ki 
If k p « k i k o , then P  P p0 + F (Eq. 3.68)
member (compression) and the tendon (tension) force must be equal to F  2 . Similarly, when the right
friction damper slips, the force in the inner member and the force in the tendon add up to the same value.
These equilibrium equations allow for the determination of the forces in the other members and the total
force and deformation in the brace. As Tables 3.10 and 3.11 show, the detailed expression for the force at
which full reverse activation occurs is approximately equal to the difference between the peak brace load
P u and the total internal friction damper slip load F for both the tension and compression sides of the
hysteresis (Equations 3.109 and 3.118). Again, this is equivalent to the simplified formulation discussed in
Section 3.2.1 (Equation 3.20). The transitional stiffness between the first and full reverse activation for
both the tension and compression sides are the same as they were for the positive activation direction.

Table 3.6: Full Solution Step 3T - Ultimate Force (Tension)

3 - ULTIMATE FORCE (TENSION)
F/2 F/2 (T) F/2 P
Stage Criterion: P = P u ,
Transitional Stiffness: k eff = k p P-F (T)
F/2 F/2 (T) F/2
Force Deformation
Inner Pi = F  2 (Eq. 3.70) F
 i = ------- (Eq. 3.71)
Member 2k i
Outer Po = F  2 (Eq. 3.72) F
 o = ------- (Eq. 3.73)
Member 2k o
Tendon Pp = Pu – F (Eq. 3.74) Pu – F
 p = -------------- (Eq. 3.75)
kp
TOTAL P = Pu (Eq. 3.76) P u – P p0 – F
 = ---------------------------
- (Eq. 3.77)
kp
Table 3.7: Full Solution Step 3C - Ultimate Force (Compression)

3 - ULTIMATE FORCE (COMPRESSION)
Stage Criterion: P = – P u , F/2 P+F/2 (C) F/2 P(-)
–1
Transitional Stiffness: k eff =  ---- + ---- + ----
1 1 1 P-F (T)
 k i k o k p
F/2 P+F/2 (C) F/2
Force Deformation
Inner F F P
P i = – P u + --- (Eq. 3.78)  i = ------- – ----u- (Eq. 3.79)
Member 2 2k i k i
Outer Po = Pi (Eq. 3.80) F P
Member  o = ------- – ----u- (Eq. 3.81)
2k o k o
Tendon Pp = Pu – F (Eq. 3.82) –Pu – F
 p = -----------------
- (Eq. 3.83)
kp
TOTAL P = – Pu (Eq. 3.84)
2
  k i k p – k o – k o k p F  k o 
 =  ---------------   P p0  -----------------------------------
1
- – --- ----
 k o + k p   k i  k i + k o   2  k i  
(Eq. 3.85)
ki + ko + kp ko F 2k o + k p
+  ---- + ---- + ----   – P u + P p0  --------------------------  ---------------  + ---  -------------------  
1 1 1
 k i k o k p   ki + ko   ko + kp  2  ko + kp  

Table 3.8: Full Solution Step 4T - First Reverse Activation (Tension)

4 - FIRST REVERSE ACTIVATION (TENSION)
F/2 -F/2 (C) <F/2 P
Stage Criterion: P o = – F  2 ,
F/2 >-F/2 (C) <F/2
Force Deformation
Inner F F  ki  F F
 i = ------- – ---- (Eq. 3.87)
Member P i = --- – -----------
- (Eq. 3.86)
2 ko 2k i k o
Outer Po = –F  2 (Eq. 3.88) F

 o = – ------- (Eq. 3.89)
Member 2k o
Tendon F  kp  Pu – F F
P p = P u – F – ------------
- (Eq. 3.90)  p = -------------- – ---- (Eq. 3.91)
ko kp ko
TOTAL F  ki + ko + kp  P u – P p0 – F F
P = P u – ---------------------------------
- (Eq. 3.92)  = ---------------------------
- – ---- (Eq. 3.93)
ko kp ko
Table 3.9: Full Solution Step 4C - First Reverse Activation (Compression)

4 - FIRST REVERSE ACTIVATION (COMPRESSION)
F/2 -(Pp -F/2) (C) <F/2 P(-)
Stage Criterion: P o + P p = F  2 ,
Pp (T)
F/2 <-(Pp-F/2) (C) <F/2
Force Deformation
Inner ki  F –P
P i = – P u + F  --------------
F F
- + --- (Eq. 3.94)  i = --------u + --------------- + ------- (Eq. 3.95)
Member  k o + k p 2 ki k o + k p 2k i
Outer ko  F –P
P o = – P u + F  --------------
F + -------
F
- + --- (Eq. 3.96)  o = --------u + --------------- (Eq. 3.97)
Member  k o + k p 2 k o k o + k p 2k o
Tendon kp  P
P p = P u + F  ---------------
F F
–F (Eq. 3.98)  p = ----u- + --------------- – ---- (Eq. 3.99)
 k o + k p kp ko + kp kp
TOTAL k i + k o + k p
P = – P u + F  -------------------------
- (Eq. 3.100)
 ko + kp 
2
  k i k p – k o – k o k p F  k o 
 =  ---------------   P p0  -----------------------------------
1
- – --- ---- + F
 k o + k p   k i  k i + k o   2  k i
(Eq. 3.101)

k + k o + k p  k o  F  2k o + k p  
+  ---
1-  – P P  -------------------------
1- + ---
1- + ---  + p0 i - --------------- + --- -------------------
 k i k o k p  u  ki + ko   ko + kp  2  ko + kp  

Table 3.10: Full Solution Step 5T - Full Reverse Activation (Tension)

5 - FULL REVERSE ACTIVATION (TENSION)
F/2 -F/2 (C) F/2 P
Stage Criterion: P i = – F  2 ,
Transitional Stiffness: k eff = k i + k p
F/2 -F/2 (C) F/2
Force Deformation
Inner Pi = –F  2 (Eq. 3.102) F
 i = – ------- (Eq. 3.103)
Member 2k i
Outer Po = –F  2 (Eq. 3.104) F
 o = – ------- (Eq. 3.105)
Member 2k o
Tendon F  kp  P
P p = P u – F – ------------
- (Eq. 3.106)  p = ----u- – F  --- 1- 
1- + --- (Eq. 3.107)
ki kp  kp ki 
TOTAL F  ki + kp  P u – P p0 – F F
P = P u – F – ----------------------
- (Eq. 3.108)  = ---------------------------
- – ---- (Eq. 3.110)
ki kp ki
If k p « k i , then P  P u – 2F (Eq. 3.109)
The post-activated stiffness after reverse activation is the same as it was after positive activation: tendon
only stiffness in tension, series stiffness in compression. The simplest criteria that may be used to deter-
mine when the end plates recontact the ends of the inner and outer members are based on the deforma-
tions of the members. Member and brace deformations at each stage of the hysteretic response are shown
in Tables 3.1 to 3.15. Similar to activation, re-contact of the end plates (Steps 6 and 7 - Tables 3.12 to
3.15) does not happen simultaneously at both ends of the brace. If the outer member is stiffer than the
inner member, then the left gap will close first for the tension side of the hysteresis and the right gap will
close first for the compression side as shown in Figure 3.15. The gap closes when the total relative deforma-
tions of the outer member and the tendon are the same as shown in Figure 3.15. These deformations are
calculated relative to the initial deformations that existed after the initial pretensioning of the tendons. As
Tables 3.12 and 3.13 show, the detailed expression for the force at which first recontact occurs is approxi-
mately equal to the difference between the tendon pretension force P p0 and the total internal friction
damper slip load F for both the tension and compression sides of the hysteresis (Equations 3.127 and
3.136). This is the same as the simplified formulation discussed in Section 3.2.1 (Equation 3.21). After the
first recontact, the effective stiffness of the brace increases to the sum of the outer member and tendon
stiffnesses on the tension side of the hysteresis and to the inner member stiffness alone for the compression

Table 3.11: Full Solution Step 5C - Full Reverse Activation (Compression)

5 - FULL REVERSE ACTIVATION (COMPRESSION)
F/2 -(Pp -F/2) (C) F/2 P(-)
Stage Criterion: P i + P p = F  2 ,
Pp (T)
F/2 -(Pp-F/2) (C) F/2
Force Deformation
Inner ko –P k0
P i = – P u + F  --------------- + ---  i = --------u + F  ------------------------- + ------- (Eq. 3.112)
F F
(Eq. 3.111)
Member  k o + k p 2 ki  k i  k o + k p  2k i
Outer Po = Pi (Eq. 3.113) –P F F
Member  o = --------u + --------------- + ------- (Eq. 3.114)
k o k o + k p 2k o
Tendon ko  P ko
P p = P u – F  --------------- (Eq. 3.115)  p = ----u- – F ------------------------- (Eq. 3.116)
 k o + k p kp kp  ko + kp 
TOTAL 2k o + k p 
P = – P u + F  ------------------
- (Eq. 3.117)
 ko + kp 
If k p « k i k o , then P = – P u + 2F (Eq. 3.118)
2
  k i k p – k o – k o k p F  k o 
 =  ---------------   P p0  -----------------------------------
1
- + --- ----
 k o + k p   k i  k i + k o   2  k i  
(Eq. 3.119)
k i + k o + k p  k o  F  2k o + k p  
+  --- 1-   – P + P  -------------------------
1- + ---
1- + --- - --------------- + --- -------------------
 k i k o k p  u p0 
ki + ko   ko + kp  2  ko + kp  
side. These transitional stiffnesses are then used to find the forces and deformations at full end plate recon-
tact, which occurs when the total relative deformations of all three elements are equal. At this point the
stiffness of the brace reverts to the initial stiffness and the next hysteretic cycle begins.
3.3 CONCEPT FOR A TELESCOPING (T-SCED) BRACE

The deformation capacity of a SCED brace is directly limited by the deformation capacity of its ten-
dons. As previously discussed in Section 3.1.5, the aramid tendons that are used within SCED braces have
a maximum usable strain of approximately 1.5-2.0%. Since the tendon length in a SCED brace is approx-
imately equal to the length of the main axial members, the elongation capacity of a SCED brace depends
solely on the tendon deformation capacity and on the length of the brace. The elongation capacity of the
aramid tendons limits the storey drift under which a SCED braced frame can fully self-center to approxi-
mately 1.5% to 2.0% of the storey height. All of the previous SCED brace prototypes have been designed

Table 3.12: Full Solution Step 6T - First End Plate Re-Contact (Tension)
6 - FIRST END PLATE RE-CONTACT (TENSION)
δp- δp0 = δo- δo0
Stage Criterion:  p =  p0 +   o –  o0  , P
Transitional Stiffness: k eff = k p
Force Deformation
Inner Pi = –F  2 (Eq. 3.120) F
 i = – ------- (Eq. 3.121)
Member 2k i
Outer Po = –F  2 (Eq. 3.122) F
 o = – ------- (Eq. 3.123)
Member 2k o
Tendon kp 
 p = P p0  ---- + ---------------  –  ------- 
kp  F
P p = P p0  1 + --------------
- – F  -------
1 1
(Eq. 3.124) (Eq. 3.125)
 ki + ko   2k o   k p k i + k o   2k o 
TOTAL kp  kp  P p0
P = P p0  1 + --------------
- – F  1 + -------
F
(Eq. 3.126)  = --------------
- – ------- (Eq. 3.128)
 ki + ko   2k o  k i + k o 2k o
If k p « k i , then P  P p0 – F (Eq. 3.127)
to have drift capacities in this range. This includes the HC-SCED prototype that will be discussed in
Chapter 4 and the shake table test SCEDs that will be discussed in Chapter 6. As previously mentioned, to
design a SCED brace that can exceed that 2.0% drift limit for use in high-seismic areas, external friction
fuses must be designed which act in series with the brace. These fuses are designed to slip when the SCED
brace reaches its deformation capacity. Since the external friction fuse has zero stiffness when it slips, any
additional external deformation that is applied to the combined SCED brace and fuse system goes into the
fuse instead of the main SCED brace. These fuses are typically designed to permit the combined system to
achieve storey drifts up to 4.0%, a drift level that is sufficient for most high-seismicity applications; how-
ever, the addition of such a friction fuse results in some residual drift in the frame when it is subjected to
large maximum credible seismic hazard earthquakes (MCE level earthquakes with a probability of exceed-
ance of 2% in 50 years). Although it has been shown that these residual drifts are lower than those that
would be experienced by comparable yielding systems (Choi et al., 2008), under large earthquakes a SCED
brace equipped with an external friction fuse is no longer completely self-centering.
In order to increase the elongation capacity of the SCED system and allow it to fully self-center even
when subjected to MCE seismic hazard level earthquakes, a novel telescoping configuration has been
developed. This new telescoping type of SCED is called a T-SCED. It relies on the same general mechani-

Table 3.13: Full Solution Step 6C - First End Plate Re-Contact (Compression)
6 - FIRST END PLATE RE-CONTACT (COMPRESSION)
δp- δp0 = δo- δo0
Stage Criterion:  p =  p0 +   o –  o0  , P(-)
–1
Transitional Stiffness: k eff =  ---- + ---- + ----
1 1 1
 k i k o k p
Force Deformation
Inner ko ko
Member P i = 2-------------------------------------------
- (Eq. 3.129)  i = -----------------------------------------------
- (Eq. 3.130)
 ki + ko   ko + kp  2k i  k i + k o   k o + k p 
  F  k i + k o  – 2P p0  k i + k o + k p     F  k i + k o  – 2P p0  k i + k o + k p  
Outer Po = Pi (Eq. 3.131) 1

 o = -------------------------------------------- (Eq. 3.132)
Member 2  ki + ko   ko + kp 
  F  k i + k o  – 2P p0  k i + k o + k p  
Tendon 1 1
P p = -------------------------------------------- (Eq. 3.133)  p = ------------------------------------------------- (Eq. 3.134)
2  ki + ko   ko + kp  2k p  k i + k o   k o + k p 
  Fk p  k i + k o  + 2P p0 k o  k i + k o + k p     Fk p  k i + k o  + 2P p0 k o  k i + k o + k p  
TOTAL F  2k o + k p   k i + k o  – 2P p0 k o  k i + k o + k p 
P = ----------------------------------------------------------------------------------------------------
- (Eq. 3.135)
2  ki + ko   ko + kp 
If k p « k i , then P  – P p0 + F (Eq. 3.136)
2
Fk o  k i + k o  – 2P p0  k o + k o k p – k i k p 
 = --------------------------------------------------------------------------------------
- (Eq. 3.137)
2k i  k i + k o   k o + k p 
cal principles as the original SCED, except that it adds an extra element. As shown in Figure 3.16, a T-
SCED contains one or more new 'floating' intermediate steel members that are parallel to the existing
inner and outer members. Each new intermediate member necessitates the inclusion of an additional set of
tendons. In the revised configuration with one intermediate member, the two sets of tendons and the inter-
mediate member are effectively in series, allowing the two sets of tendons to share the brace deformation
equally between them. Therefore, if the strain capacity of each individual set of tendons is the same as it
was for the original SCED, then the T-SCED will have double the total elongation capacity. Although only
one intermediate member is shown in Figure 3.16, the concept may easily be extended to include two or
more intermediate members.

Table 3.14: Full Solution Step 7T - Full End Plate Re-Contact (Tension)
7 - FULL END PLATE RE-CONTACT (TENSION)
P
Stage Criterion:  p =  p0 +   i –  i0  ,
Transitional Stiffness: k eff = k o + k p
δp- δp0 = δi- δi0
Force Deformation
Inner Pi = –F  2 (Eq. 3.138) F
 i = – ------- (Eq. 3.139)
Member 2k i
Outer F k o = i (Eq. 3.141)
P o = – ---  ----o (Eq. 3.140)
Member 2  ki
Tendon kp
 p = P p0  ---- + ---------------  –  ------- 
kp  F
- – F  ------- 
P p = P p0  1 + --------------
1 1
(Eq. 3.142) (Eq. 3.143)
 ki + ko   2k i   k p k i + k o   2k i 
TOTAL kp  P p0
P = P p0  1 + -------------- - (Eq. 3.144)  = -------------- F-
- – ------ (Eq. 3.145)
 ki + ko  k i + k o 2k i
F  ki + ko + kp 
– ---------------------------------
-
2k i
The simplified step-wise hysteretic response of the T-SCED brace is similar to the simplified step-wise
derivation for the original SCED brace described in Section 3.2.1. The overall hysteretic behaviour may be
calculated in terms of the same parameters that were previously shown in Figure 3.13. The full step-wise
hysteretic response of the T-SCED brace is significantly more complex than for it was for the original
SCED brace described in Section 3.2.2. Since there are dozens of different combinations of relative mem-
ber stiffnesses and gap opening progressions, it is not practical or desirable to derive a full closed form step-
wise solution for the T-SCED brace. Instead, the full simulated hysteretic response of a T-SCED brace may
be calculated by using the SCED Mechanics Simulator software tool that will be described in a future sec-
tion (Section 3.4). For the purposes of design, the simplified step-wise response that is derived below is suf-
ficient. As for the original SCED, the simplified T-SCED hysteresis expressions presented here assume that
the energy dissipation is provided by a friction slip mechanism with a total slip force F and zero post-slip
stiffness. All of the gaps between the end plates and the axial members are assumed to open simultaneously.
Starting from the origin of the hysteresis in tension, the initial stiffness of the brace is k 1 which is the
sum of the stiffness of the four axial elements comprising the brace: the inner, intermediate, and outer
members and the tendons.

Table 3.15: Full Solution Step 7C - Full End Plate Re-Contact (Compression)
7 - FULL END PLATE RE-CONTACT (COMPRESSION)
P(-)
Stage Criterion:  p =  p0 +   i –  i0  ,
δp- δp0 = δi- δi0
Force Deformation
Inner ki 1
Member P i = 2-------------------------------------------
- (Eq. 3.146)  i = ------------------------------------------------ (Eq. 3.147)
 ki + ko   ko + kp  2k i  k i + k o   k o + k p 
  F  k i + k o  – 2P p0  k i + k o + k p     F  k i + k o  – 2P p0  k i + k o + k p  
Outer ko o = i (Eq. 3.149)

Member P o = 2-------------------------------------------
 ki + ko   ko + kp 
- (Eq. 3.148)
  F  k i + k o  – 2P p0  k i + k o + k p  
Tendon 1 1
P p = -------------------------------------------- (Eq. 3.150)  p = ------------------------------------------------- (Eq. 3.151)
2  ki + ko   ko + kp  2k p  k i + k o   k o + k p 
  Fk p  k i + k o  + 2P p0 k o  k i + k o + k p     Fk p  k i + k o  + 2P p0 k o  k i + k o + k p  
TOTAL  k i + k o + k p   F  k i + k o  – 2P p0 k i 
P = --------------------------------------------------------------------------------
- (Eq. 3.152)
2  ki + ko   ko + kp 
F  k i + k o  – 2P p0 k i
 = --------------------------------------------
- (Eq. 3.153)
2  ki + ko   ko + kp 
k 1 = k i + k m + k o + 2k p (Eq. 3.154)
where k i , k m and k o are the axial stiffnesses of the inner, intermediate, and outer members, respectively,
and k p is the total stiffness of each set of pretensioning tendons. The T-SCED brace activates and changes
stiffness when the total external force on the brace P overcomes the initial tendon pretension of a single set
of tendons P p0 and the friction slip force F , while also taking into account the elongation of the tendons.
These terms give a T-SCED brace activation force P a equal to:
P a = P p0 + F + 2k p  a (Eq. 3.155)
where  a is the deformation of the brace at the activation point which is equal to:
P
 a = -----a (Eq. 3.156)
k1

Hysteretic Response
Tension Energy Dissipator
Brace Strain = +2δ Strain = +2δ
Outer Member Force = F P P
South Outer Pp0+F
2F
Dual Tendon Force = Pp0 + P - F
End Plates North Outer
South Inner Intermediate Member Force = P - F δ
Tendon Force = Pp0 + P - F North Inner -(Pp0+F)

Tendon Strain = +δ
Compression
Brace Strain = -2δ Energy Dissipator
Strain = -2δ
Outer Member Force = P P P
South Outer
Tendon Force = Pp0 + P - F North Outer
South Inner Intermediate Member Force = P - F δ
Tendon Force = Pp0 + P - F North Inner

Tendon Strain = +δ
Figure 3.16: T-SCED Brace Mechanics
Substituting Equation 3.156 into 3.155 gives:
Pa
P a = P p0 + F + 2k p  ----- (Eq. 3.157)
 k 1
P p0 + F
P a = ---------------------------------------------------------  P p0 + F (Eq. 3.158)
2k p
1 –  ------------------------------------------
 k i + k m + k + 2k p
o
The deformation of the brace at activation  a is therefore equal to:
Pa P p0 + F
 a = -----  ------------------------------------------ (Eq. 3.159)
k 1 k i + k m + k o + 2k p
After brace activation in the tension direction, a gap opens between the north (right) outer end plate
and the intermediate member, the north inner end plate and the inner member, the south (left) outer end
plate and the outer member, and the south inner end plate and the intermediate member as shown in Fig-
ure 3.16. The stiffness in the brace is then provided by the two sets of tendons in series with the intermedi-
+
ate member. Therefore, the post-activation stiffness in the tension direction k a is equal to:

+ 1 1 kp
k a = ----------------------------- = -----------------  ---- (Eq. 3.160)
1 + ----- 1- + ---- 1 2 + -----
---- 1- 2
----
kp km kp k p k m
+
If the brace is extended post-activation up to a given deformation  , the axial force in the brace at that
deformation will be:
+
+ kp 
P = P p0 + F + ---------
- (Eq. 3.161)
2
Similarly to the original SCED response, when the load reverses from the peak and returns towards the
origin, the brace reverts to its initial stiffness k 1 (Equation 3.154). The calculation of the energy dissipa-
tion capacity parameter may also be simplified to the expression given in Equation 3.16. If the brace
+
attained a deformation of  before reversing direction, at the reverse activation point the deformation in
the brace is equal to:
+ 2F
 =  – -----------------------------------------
- (Eq. 3.162)
k i + k m + k o + 2k p
and the force at that point is equal to:
+
kp  2F P
- – 2F – ------  -----a  2k p 
+
P = P – 2F – 2 a k p = P p0 + F + --------- (Eq. 3.163)
2 P a  k 1
+ +
kp   2k p  kp 
P = P p0 – F + ---------- – 2F ------------------------------------------  P p0 – F + ---------- (Eq. 3.164)
2  k i + k m + k + 2k p 2
o
After the friction mechanism slips and the brace activates in the reverse direction on its return to zero,
+
the stiffness of the brace is once again equal to the stiffness k a from Equation 3.160. When the deforma-
tion returns to the initial stiffness branch of the hysteresis near the origin, the gaps between the end plates
and axial members close and the force at that point is equal to:
P = P a – 2F  P p0 – F (Eq. 3.165)
and the deformation of the brace is:
P a – 2F P p0 – F
 =  a  1 –   = ------------------  ------------------------------------------ (Eq. 3.166)
k1 k i + k m + k o + 2k p

As for the original SCED brace, the activation of the T-SCED brace is the same in compression as in
tension. As shown in Figure 3.16, when the T-SCED brace is deforming in compression post-activation
the inner, intermediate and outer members and both sets of tendons are all in series. Therefore, the true
-
post-activation stiffness of the brace in compression k a is given by the series stiffness expression:
- 1
k a = --------------------------------------- (Eq. 3.167)
1 1 1 2
---- + ------ + ---- + ----
ki km ko kp
Since k p is generally much smaller than k i , k m and k o , the series stiffness may be simplified to:
- kp
k a  ---- (Eq. 3.168)
2
As for the original SCED, the design of T-SCED braces may be simplified by using this assumption to
conclude that the hysteresis may be considered to be symmetric, such that the positive and negative quad-
rants of the hysteretic response are mirror images of each other.
3.4 THE SCED MECHANICS SIMULATOR - A DETAILED SIMULATION OF SCED AND T-

SCED BRACE HYSTERESES
The equations that described the full step-wise hysteretic behaviour of the SCED brace that were dis-
cussed in Section 3.2.2 were complex and dependant on the relative stiffness of the inner and outer mem-
bers and the relative slip loads of the left and right friction dampers. That derivation was only valid if the
outer member is stiffer than the inner member and if the friction damping force is split evenly between the
left and right dampers. It also assumed that the inner and outer members are exactly the same length. In
reality, there will be some finite dimensional tolerance between the length of the two members which will
introduce even more complex hysteretic behaviour that cannot reasonably be solved for using a closed-
form step-wise solution. With the addition of a third intermediate member, the real mechanics of the T-
SCED brace are even more complex.
To be able to determine this complex hysteretic response of SCED and T-SCED braces, a new software
tool was created using HTML5 and JavaScript. This program is called the SCED Mechanics Simulator
and it can simulate the hysteretic effect of any combination of member stiffnesses, arbitrary internal fric-
tion damper slip forces at either end of the brace, and member length construction tolerances. It also
includes the ability to add viscous damping in parallel with the brace and simulate the effect of that damp-
ing at a specific sinusoidal frequency.

The SCED Mechanics Simulator requires a web browser in order to run. It was programmed in JavaS-
cript so that it is cross-platform compatible and may be served to users over the web. As such, any com-
puter system with a modern web browser can use the software whether the system is Windows, Mac, or
Linux-based. All of the programs inputs, outputs and interactive elements are contained within a single
dynamic webpage. The open-source JavaScipt package Sylvester was used to provide linear algebra capabil-
ities for the mechanics calculations (Coglan, 2012). A number of other open-source JavaScript packages
were used to facilitate other various user interface capabilities. The package JQuery was used to add
dynamic functionality to the HTML page (JQuery Foundation, 2012a). JQueryUI was used to provide
the visual user interface elements for the input forms, buttons and sliders (JQuery Foundation, 2012b).
The JQuery extension Flot was used to provide dynamic data plotting and charts (Laursen, 2012). The
dynamic schematic of the SCED brace was drawn using the HTML5 canvas element, but backwards-com-
patibility support for the canvas element in older browsers was provided using the package ExplorerCanvas
(Google Inc., 2006). Input box validation to restrict users to appropriate input values and ranges was pro-
vided using the LiveValidation javascipt package (Hill, 2008).
3.4.1 SCED MECHANICS SIMULATOR ALGORITHMIC IMPLEMENTATION
The SCED Mechanics Simulator models SCED brace hysteretic behaviour using a nonlinear incremen-
tal stiffness method analysis. The stiffness matrix for the analysis contains elements that are connected and
arranged as shown in Figure 3.17. The model itself is effectively one-dimensional; however, for clarity, the
figure shows the model elements spaced out in two dimensions. As the figure shows, two different stiffness
matrices are possible depending on whether the brace being modelled is an original SCED or a T-SCED.
The elements that represent the inner, intermediate and outer members ( k i , k m , k o ), and the tendons
( k p , k p1 and k p2 ) have a permanent linear stiffness which is dictated by the input parameters. The con-
nection stiffness element ( k conn ) also has a permanent stiffness dictated by an effective series connection
stiffness provided in the inputs. The end plate contact elements ( k g   ) represent the contact between the
end surfaces of each of the axial members and the end plates. The stiffness of these contact elements is set
at each analysis step to be equal to either the bearing stiffness provided in the inputs (during contact), or a
small non-zero stiffness (when not in contact). The internal friction damper elements ( k f1 and k f2 ) repre-
sent the conditional stiffness of the friction dampers. Similar to the contact elements, at each analysis step
the stiffness of each friction damper element is set to the friction interface stiffness provided in the inputs
when the damper has not slipped and a small non-zero stiffness when the damper is slipping. For the T-

SCED model, the end plate friction elements ( k ep1 and k ep2 ) work similarly, changing stiffness based on
the friction slip between the inner and outer end plates. These elements simulate the effect of friction
between the T-SCED’s inner and outer end plates. These elements will be used in some of the T-SCED
brace models discussed in Chapter 5. For each analysis step, the simulator calculates the locations, lengths,
axial force, and elongation for every element in the model.
End 1 Node ko Outer Member kconn Applied

Internal
kf1 Element Deformation
Friction
Damper 4 kg,po2 5 6
Node Number F
Element Permanent F
(typ.)
kg,po1 kp Stiffness
(typ.) δ
End Plate δ
2 kg,ip2 3 Linear
Contact Tendon Element Nonlinear Connection
Element Gap (typ.) Stiffness
(typ.) kg,ip1 ki Element
F
0 1
kf2
F Inner Member
Element End 2 δ
δ Nonlinear
Linear Members Friction (typ.)
and Tendons
T-SCED Model
End 1 ko Outer Member kconn Applied
Internal Element Deformation
Friction kf1
Damper 4 kg,po2 5 6
Element
(typ.) kp2
kg,po1 End Plate
kep1 Contact
9 kg,mp2 10 Element
Outer Tendon (typ.)
Element
kg,mp1 km
7 kg,pm2 8
Intermediate
Member End
kg,pm1 kp1 Element Plate
Friction
kep2
2
Inner Tendon
kg,ip2 3 Element
(typ.)
Element
kg, ip1 ki
kf2
0 1
Inner Member
Element End 2
Figure 3.17: SCED Mechanics Simulator Model.

Note: Both models are one-dimensional, but are shown here separated into two dimensions for clarity
An excerpt from the software source code that contains the full simulation calculations may be found in
Appendix A; however, the simulation algorithm will be summarized below. The stiffness method analysis
was performed using the Payne and Irons method (Zienkiewicz and Taylor, 2000).
The first stage of the analysis is to apply the pretensioning to the tendon elements to determine the ini-
tial state of the brace prior to external loading. During this stage, it is assumed that the internal friction

dampers are not active and, therefore, do not provide any stiffness. The pretensioning force is applied in
small force steps. The step increment for this pretensioning force is defined in the software input. The left
end of the brace (node 0 in Figure 3.17) is considered to be restrained. For each step in the pretensioning
analysis the following steps are performed:
1. Construct the stiffness matrix for the current step based on the positions of the
left and right ends of each axial member with respect to the tendon end plates.
2. Construct a fixed end force vector that contains the force increment applied to
each end of the tendon element. (There is no external force during the
pretensioning.)
3. Multiply the diagonal value in the stiffness matrix that is associated with the
restrained node (nodes 0) by a large number (Payne and Irons).
4. Invert the stiffness matrix.
5. Multiply the resulting inverted matrix by the fixed end force vector to solve for
the displacement vector, which gives all the nodal displacements for the current
step.
6. Add these nodal displacements to the previous displacement values for each
node and then calculate the new locations of the axial members and end plates
(which will be used to define the stiffness matrix during the next step).
7. Find the axial forces in all of the brace elements by multiplying the
displacement vector of each element by that element’s local stiffness matrix and
then adding the element’s fixed end force vector.
8. If the load in the tendon exceeds the pretension load defined in the input, stop
the pretensioning.
Once the pretension is fully applied, the full brace is deformed in the directions and magnitudes speci-
fied in the program input. During this single cycle analysis, node 0 is restrained and the axial deformation
of the brace is controlled by applying incremental displacements to the free end of the brace (node 6 in Fig-
ure 3.17). The deformation step increment is defined in the software input. For each step in the hysteretic
analysis the following steps are performed:

1. Check the force in each of the internal friction dampers to see if they exceed the
slip load specified in the input.
a) If a damper’s force exceeds the slip load, set the stiffness of the element to
zero.
b) If not, set the element stiffness to the friction interface stiffness specified
in the input.
2. If it is a T-SCED analysis, check the end plate friction elements in the same way
and set the element stiffnesses accordingly.
3. If viscous damping is specified in the input, calculate the total viscous force for
the current step using the sinusoidal deformation frequency and SCED brace
deformation amplitude defined in the input.
4. Construct the stiffness matrix for the current step based on the positions of the
left and right ends of each axial member with respect to the tendon end plates,
and the current stiffness of each of the friction damper elements.
5. Construct a fixed end force vector using all of the internal element forces from
the previous analysis step. Include the current force from the viscous damper as
a fixed end force between the restrained node and the deformed node (nodes 0
and 6).
6. Multiply the diagonal values in the stiffness matrix that are associated with the
restrained node and the deformed node (nodes 0 and 6) by a large number
(Payne and Irons).
7. Construct an external force vector with a single value for the deformed node
(node 6) that is equal to the new diagonal value from the stiffness matrix for
that node multiplied by the incremental deformation value for the current step
(not the total deformation).
8. Invert the stiffness matrix.

9. Multiply the resulting inverted matrix by the difference between the external
force vector and the fixed end force vector to solve for the displacement vector,
which gives all the nodal displacements for the current step.
10. Add these nodal displacements to the previous displacement values for each
node and then calculate the new locations of the axial members and end plates
(which will be used to define the stiffness matrix during the next step).
11. Find the axial forces in all of the brace elements by multiplying the
displacement vector of each element by that element’s local stiffness matrix and
then adding the element’s fixed end force vector.
12. Determine the total force in the brace by calculating the reaction force at the
restrained node (node 0). This is done by multiplying the nodal displacement
vector by the unmodified stiffness matrix (without the large numbers) and then
adding the fixed end forces.
13. If the brace deformation has reached the peak deformation or returned to zero,
set a new target deformation or stop the analysis.
The analysis does not cap the forces in any of the elements, including the friction damper and contact
elements. Therefore, it is prone to overshoot force and deformation values if the step increment is too
large. A small deformation step increment value is necessary to attain high-quality hysteretic behaviour
simulation results.
3.4.2 USER INTERFACE
The input form for the SCED Mechanics Simulator is shown in Figure 3.18. The form begins with
inputs for the Young’s modulus for the steel sections and the pretensioned tendons. This is followed by
inputs for the geometry of each steel member. Input parameters describing the geometry of the inner and
outer members are always required, and the ‘Middle’ member which represents the T-SCED intermediate
member is optional. The ‘Middle’ checkbox effectively selects between analyzing an original SCED or a T-
SCED. The member section geometry inputs allow the user to specify member length, area and initial
position. By inputting different initial section lengths for each member, the Mechanics Simulator can sim-
ulate the effect of member length tolerances on the behaviour of the brace. The initial position inputs
allow the user to specify where along the length of the brace the shorter sections begin. For each section,

the initial length plus the initial position must be less than the initial length of the longest section; this
requirement is enforced by the software.
Figure 3.18: SCED Mechanics Simulator Inputs
The friction inputs specify the slip force for the internal friction dampers at each end of the brace.
These may be different from each other and may be set to zero to eliminate internal friction damping. End

plate friction is only available for T-SCED analyses and represents the effect of the inner and outer end
plates rubbing against each other during the operation of the brace. This option was introduced to the pro-
gram to simulate an effect that was seen in the T-SCED prototype testing that will be described in Chapter
5. Tendon inputs are provided for the cross-sectional area of the tendons and the tension pretension during
brace assembly. For T-SCED analyses, these parameters are assumed to be the same for each set of tendons.
For example, if the T-SCED brace that is being simulated has two sets of six tendons (twelve total), each set
of tendons has the tendon area specified and each set is pretensioned to the specified input pretension
force.
Viscous damping inputs are optional. The viscous damping that is specified by the input damping con-
stant adds a force to the model at each step that depends on the specified cycle frequency and the analysis
deformation limits.
The last set of model inputs, the connections in series, provides the ability to simulate the effect of a
spring in series with the brace. Up to three different part stiffnesses may be provided by inputting a length
and area for each part. These connection parts are assumed to be made out of steel with the Young’s modu-
lusthat is provided in the materials input box. These parts are used to calculate an effective spring stiffness
that will be modelled in series with the SCED brace, the value of which is displayed dynamically below the
connection inputs. Zero values in the input boxes cause the simulator to assume infinite stiffness.
To the right of the main brace inputs, a dynamic plot shows the simplified behaviour of the brace as
predicted using the simplified equations given in Sections 3.2.1 and 3.3. This plot updates dynamically as
the inputs are modified and the plot range representing the maximum deformation of the brace may also
be modified dynamically using the slider below the plot.
Below the brace inputs there are a set of inputs which determine how the analysis is conducted. The
first line of analysis inputs allows the user to specify the bearing stiffness and friction interface stiffness.
The default stiffness values (100000 kN/mm for both) tend to provide the most stable analytical results for
most practical design cases. The next line of analysis inputs specifies the force step that is used for applying
the tendon pretension as described in the previous section. After the tendon pretensioning, the entire
SCED brace is deformed using a displacement-controlled analysis. The deformation step input determines
the size of the deformation steps for the analysis as described in the previous section. The small deforma-
tion step is used when the analysis is within 4mm of the origin and the large deformation step is used the
rest of the time. These two differing deformation steps allow a higher-resolution analysis when the brace is
undergoing complex element interactions, and lower resolution otherwise to reduce the size of the result

output. The two deformation limit inputs in the last line of the analysis inputs determine the maximum
deformations that are applied to the brace model in each direction. These limits may be either positive or
negative. Usually, to get a full hysteretic response, one of the limits is positive and the other is negative.
During the analysis, deformation limit A is applied first and then deformation limit B is applied.
After the ‘Calculate SCED’ button is pressed, the pretensioning and hysteretic behaviour calculations
are performed and the results of the analyses are presented as shown in Figure 3.19. The top two plots
depict the pretension load in each of the axial members (inner, outer and intermediate if a T-SCED analy-
sis is being performed). The left plot shows the axial force in each member as a function of the load step,
which shows when each member begins to take tendon pretension force. The right plot shows the axial
force in each member as a function of each member’s axial deformation. This plot represents the hysteretic
behaviour of each member during tendon pretensioning. These plots may both be panned by clicking and
dragging with the mouse and may be zoomed in and out using the mouse wheel. Hovering the mouse
pointer over a line in the plot shows the x- and y-coordinates of the plot line at the mouse pointer location.
Below the pretensioning analysis plots, there is a dynamic schematic showing the forces and deforma-
tions of each element of the SCED brace. This schematic is similar in form to the brace schematics shown
previously in Figures 3.12 and 3.16. The horizontal slider that is below the schematic allows the user to
select an analysis step. The state of the brace for the selected step is then shown in the schematic. Text
input boxes above and below the slider allow the user to select a step by the step number and to change the
horizontal scale of the dynamic schematic so that deformations may be exaggerated. The coloured rectan-
gles immediately below the schematic represent the force present in the internal friction dampers. Both
pretension analysis steps and hysteretic behaviour analysis steps may be shown by the dynamic brace sche-
matic.
The large bottom plot in the results output shows the total hysteretic behaviour of the SCED brace as
well as the hysteretic behaviour of the individual brace elements (the axial members and the tendons). For
the individual elements, the hysteretic behaviour is plotted as the force in that element versus the total
deformation of the brace. For each hysteresis shown in that plot, a circular marker identifies the point on
the hysteresis that is represented by the step selected by the slider above and the state shown by the
dynamic schematic above that. Individual hystereses may be shown or hidden using the radio buttons
above the plot. As was the case for the pretensioning plots, this plot may be panned and zoomed and the x-
and y-coordinates of any point may be determined by hovering over a plot line with the mouse pointer.

Figure 3.19: SCED Mechanics Simulator Results

After the calculation is complete, the full result data set may be output in text format by clicking the
‘Show Data...’ button below the large hysteresis plot. The output format is shown in Figure 3.20. The data
set may be output in either a space-delimited or comma-delimited format. The header of the data set
includes the analysis date and time and the full set of input parameters. The output data is given in col-
umns with the appropriate column headers above. The ‘Select All Data’ button selects the full contents of
the text box containing the data set which may then be copied by right-clicking with the mouse. Once cop-
ied, the data may be pasted into a text editor to save it for external post-processing.
Figure 3.20: SCED Mechanics Simulator Output Data

3.4.3 VALIDATION OF THE SCED MECHANICS SIMULATOR
To validate the SCED Mechanics Simulator, the example that was previously used to compute the hys-
teretic behaviour of a sample original SCED brace that was derived based on the full closed-form step-wise
equations (as shown previously in Figure 3.14) was simulated using the SCED Mechanics Simulator. A
comparison of these two analyses is shown in Figure 3.21. As the figure shows, both the full step-wise cal-
culations and the simulated response produce identical hystereses. Since there are no practical closed-form
solutions for the T-SCED detailed mechanics or for the effect of member length tolerance, there is no way
to perform a theoretical comparison for those scenarios; however, real experimental test results in later
chapters are used to validate the predictions from the SCED Mechanics Simulator to assess its performance
when used to model those types of scenarios.
2000
1500
Full Closed-Form
Step-wise
Hysteresis
1000
(grey line behind)
Brace 500
Axial
Force 0
P
-500
(kN) SCED Mechanics
-1000
Simulator Model
Hysteresis
(black line front)
-1500
-2000
-3 -2 -1 0 1 2 3
Brace Deformation δ (mm)
Figure 3.21: Detailed Mechanics Results vs. Mechanics Simulator Model for a Sample SCED Brace
Although detailed comparisons between SCED Mechanics simulator results and real test data will be
presented in Chapters 4, 5, and 6, two different sample comparisons from those chapters are shown in Fig-
ure 3.22. These show that the SCED Mechanics Simulator is able to model the complex behaviour of real
SCED braces. The force vibration in the T-SCED comparison plot on the right is a result of an phenome-
non that is unrelated to the modelling and will be discussed in Chapter 5. The detailed discussion of the
comparisons in future chapters will include comparisons of the individual element hysteresis predictions as
well, with comparisons showing the responses of the inner, intermediate and outer members and the ten-
dons. It will be seen that the SCED Mechanics Simulator predicts the response of these elements well too.

100 1000
Shake Table Test SCED T-SCED
75
Test E-12 800 Test DPD-11
Mechanics 600
50 Simulator
400
Brace 25
200
Axial
Force 0 0 Mechanics
(kN) Simulator
Test -200
-25
-400
-50 Force Vibration
Step due to -600
due to Beam
-75 Member Length Connection Slip
Difference -800
Test
-100 -1000
-5.0 -2.5 0 2.5 5.0 -120 -60 0 60 120
Brace Axial Deformation (mm)
Figure 3.22: Sample SCED Mechanics Simulator Results Compared to Test Data
3.5 THE EFFECT OF FABRICATION LENGTH TOLERANCES

As mentioned previously, in a real SCED brace, the length of the inner, intermediate and outer mem-
bers of the brace will not be identical. The axial members will have a construction tolerance that applies to
their length. In practice, the manufacturer of the brace may be asked to use any length tolerance to within
a fraction of a millimeter; however, tighter tolerances are associated with higher costs, especially for ele-
ments that are the size and length of typical SCED brace elements. Since the SCED Mechanics Simulator
can effectively simulate the effect of different construction tolerances, it is possible to attain a better under-
standing of the effect of construction length tolerances on the behaviour of SCED and T-SCED braces.
This will help a designer of SCED braces to determine what tolerance is appropriate. The end result of this
analysis will be a simplified procedure to estimate the magnitude of the effect in SCED braces.
3.5.1 THE EFFECT OF FABRICATION LENGTH TOLERANCES ON AN ORIGINAL SCED BRACE
The same sample SCED brace that was used in Figure 3.14 and later in Section 3.4.3 is used to study
the effect of differing axial element lengths as shown in Figure 3.23. The top plot in the figure shows a
comparison of the low amplitude hysteretic response of the brace under different member length scenarios.
These hystereses are only shown for a deformation amplitude of +/- 2.25mm so that the different transi-
tional stiffnesses are clearly visible. In addition, each hysteresis in this plot is labelled with the stiffness of
each linear branch in terms of the stiffnesses of the inner (i) and outer (o) members and the tendons (p).
Combinations of the symbols represent the sum of the stiffnesses of those elements. For example, ‘ip’ is the

sum of the inner member stiffness and the tendon stiffness. On the compression side, the post-activation
stiffness for all of the hysteresis is equal to the series stiffness of the three axial elements (Equation 3.23),
which is approximately equal to the tendon stiffness alone, and is therefore shown as ‘~p’.
Sample SCED Brace Hysteresis from SCED Mechanics Simulator

ki = 645kN/mm, ko = 1129kN/mm, kp = 29.2kN/mm, F = 760kN, Pp0 = 890kN, L = 6200mm
Mech. Sim. Mech. Sim. Mech. Sim.
Simple Equal Inner Mem. Outer Mem.
Calc. Lengths Short (-0.7) Short (-1.5)
2000
k1,eff 1803 kN/mm 1300 kN/mm 1106 kN/mm 1046 kN/mm
Activation p p p Step p
Point ip
1500 op Behaviour
ip
iop
1000 iop iop
iop iop
ip
Brace iop 0
500 iop
ip ip ip
Axial Origin p p p iop
i i o p
Force 0 op
P ~p i ~p i ~p op ~p ip
i Branch Stiffness, e.g.:
(kN) -500 iop = inner + outer + tendon
0 p = tendon
iop iop
iop iop iop ~p = tendon approximately
-1000 iop i iop
o Effective
i Step Initial Stiffness
-1500 Behaviour
~p ~p ~p ~p k1,eff
-2000
-3 -2 -1 0 +/-1 0 +/-1 0 +/-1 0 1 2 3
Mech. Sim. Mech. Sim. Mech. Sim.
Equal Inner Mem. Outer Mem.
Lengths Short (-0.7) Short (-1.5)
2000
Same Hystereses
as Above at
1500
Larger
Amplitudes
1000
Brace 500
Axial
Force 0
P
(kN) -500
-1000
-1500
-2000
-15 -10 -5 0 +5/-10 +10/-5 0 +5/-10 +10/-5 0 5 10 15
Figure 3.23: Sample SCED Brace Hystereses - Effect of Length Tolerance

The left hysteresis in the plot represents the idealized simplified behaviour of the brace with a simple
initial stiffness equal to the sum of all three axial elements and a post-activation stiffness equal to the stiff-
ness of the tendon. This is the brace behaviour that would be predicted by the simplified equations given
in Section 3.2.1. To the right of that simplified hysteresis is the realistic hysteresis of a SCED brace in
which the inner and outer members have different stiffnesses but identical lengths. The initial stiffness
before activation is multilinear. This is the same as the hysteretic behaviour that is predicted by the detailed
mechanics of the SCED brace given in Section 3.2.2.
The two hystereses on the right of the plot show the effect of different inner and outer member lengths
due to construction tolerances. The third hysteresis from the left shows the behaviour of the brace if the
inner member is 0.7mm shorter than the outer member and the fourth hysteresis shows the behaviour if
the outer member is 1.5mm shorter than the inner member. These values were chosen such that both
members still take some load during pretensioning, but most of the load is taken by the longer member.
These two hystereses show that the behaviour of the brace becomes significantly more complicated when
the lengths are not equal. The initial stiffness branches of these hystereses do not show a consistent pattern
with respect to the different element stiffnesses. In addition, there are some portions of the hysteresis that
have zero stiffness, causing ‘steps’ in the response. As will be seen, this step behaviour becomes much more
pronounced as the difference between the member lengths becomes greater.
The bottom plot in Figure 3.23 shows the same hystereses at a greater deformation amplitude (but
without the simple calculation hysteresis). The +/-10mm amplitude of these hystereses is on the order of
one sixth of the total deformation capacity of the sample SCED brace design. These larger amplitude hys-
tereses are shown so that the stiffness variations of the upper plot may be seen in context. All of these larger
amplitude hystereses show that the member length differences do not have any effect on the activation
force of the SCED brace. The most significant effect that the member length differences cause is that they
change the initial stiffness of the brace. This effect may be taken into account in higher-level building
models by using an effective initial stiffness k 1 eff when modelling the SCED braces as single elements
(using, for example, the single element models described in Section 3.5). This effective stiffness may be
determined by finding the slope between the positive and negative activation points. The effective stiff-
nesses of the different brace scenarios are shown in the top plot of Figure 3.23.

3.5.2 ESTIMATING THE EFFECTIVE INITIAL STIFFNESS OF AN ORIGINAL SCED BRACE
A method to calculate the effective initial stiffness of a SCED brace with differing member lengths may
be arrived at by studying the behaviour of a brace’s individual elements under different member length sce-
narios. Five such scenarios are shown in Figure 3.24. For each scenario, the plots in the figure show the
overall brace hysteresis, the inner and outer member hystereses with respect to overall brace deformation,
and the tendon hysteresis with respect to overall brace deformation. These hystereses are for the same sam-
ple brace that was presented in the previous section, for which the detailed hysteretic response was gener-
ated using the SCED Mechanics Simulator with a deformation amplitude of +/-5mm. For all of the plots,
the initial position of both members was assumed to be zero. This means that the members start adjacent
at the end that is closest to the inner member connection.
The differing member lengths have a significant effect on the initial forces in the members that are
caused by the pretensioning of the tendons. If the members are exactly the same length, they share the pre-
tension force in proportion to their stiffness:
ki
P i0 = – P p0  --------------- (Eq. 3.169)
 k i + k o
ko
P o0 = – P p0  --------------- (Eq. 3.170)
 k i + k o
where P i0 and P o0 are the initial forces in the inner and outer members, respectively, after pretensioning
of the tendons and P p0 is the total pretension force in all the tendons. If the inner and outer members are
not the same length, then the forces in the two members depend on the difference in length as follows:
 io = L o – L i (Eq. 3.171)

 0 if  io  0
P is =  (Eq. 3.172)
 k i  io  – P p0 if  io  0


 – k o  io  – P p0 if  io  0
P os =  (Eq. 3.173)
 0 if  io  0


Sample SCED Brace Hysteresis Breakdown from SCED Mechanics Simulator

ki = 645kN/mm, ko = 1129kN/mm, kp = 29.2kN/mm, Pp0 = 890kN, L = 6200mm, FI1 = FI2 = F/2 = 380kN
Equal Lengths Brace
2000
Tendon
Full Activation Coincides with the
1000 Largest Activation Deformation of the
Axial Two Axial Members
Force 0
(for both Tension and Compression)
(kN) Outer Member
-1000 Inner Member
Initial Force due to Activation Forces for the
Pretensioning (typ.) Axial Members are the
-2000 same in all cases
Inner Member Shorter (-0.7mm) Outer Member Shorter (-1.0mm)
2000
1000
Axial
Force 0
F/2 (typ.)
(kN)
-1000 Pp0 + F/2 (typ.)
-2000
Inner Member Shorter (-3.0mm) Outer Member Shorter (-3.0mm)

2000
1000
Axial
Force 0
(kN)
-1000
-2000
-6 -4 -2 0 2 4 6 -6 -4 -2 0 2 4 6
Brace Deformation δ (mm) Brace Deformation δ (mm)
Figure 3.24: Sample SCED Brace Hysteresis Breakdown - Effect of Length Tolerance
where L i and L o are the initial lengths of the inner and outer members,  io is the difference in length
between those members, and P is and P os are the forces on the inner and outer members required to make
them the same length. The longer member is compressed by the tendons until the members are the same
length. If the tendon pretension is not large enough to make up the difference (i.e P os  – P p0 or
P is  – P p0 ), then some difference will remain between the two member lengths. The value of this differ-
ence (  io r ) is equal to:


  –P p0
-------
- if  io  0 and – k o  io  – P p0
 io k o

 io r =  P p0 (Eq. 3.174)
  io + -------- if  io  0 and k i  io  – P p0
 ki

 0 otherwise
Then, to find the total force in the inner and outer members due to the tendon pretension, the remainder
of the pretension force that was not required to make the members of equal length is distributed according
to each member’s relative stiffness:
ki
P i0 = P is –  P po + P is + P os   --------------- (Eq. 3.175)
 k i + k o
ko
P o0 = P os –  P po + P is + P os   --------------- (Eq. 3.176)
 k i + k o
Note that in these equations, P p0 will have a positive value because the tendons are in tension and P is and
P os will be negative because both axial members are in compression.
After determining the precompression force in each member due to the tendon pretension, an approxi-
mate method may be determined for estimating each brace’s effective stiffness, using the element hystereses
in Figure 3.24 as a guide. As the figure shows, for all cases, the activation forces in each direction coincide
with the activation point of either the inner or outer member hysteresis. The positive (tension) activation
of each member hysteresis occurs at a load that is equal to the slip load of the friction damper at one end of
the brace. If the internal friction damper slip force at each end of the brace is the same, then both members
experience positive activation at a force of F  2 ; however, if the slip loads are different, the outer member
activates at the slip load of the damper nearest the inner member connection F I1 (end 1, the left side in
Figure 3.15) and the inner member activates at the slip load of the damper nearest the outer member con-
nection F I2 (end 2, the right side in Figure 3.15). Using these positive activation forces for each member,
the deformation of the brace at that point may be determined by using the stiffness of the members and
the initial load in the members that was previously calculated in Equations 3.171 to 3.173, 3.175 and
3.176. Additionally, these deformations must take into account the effect of any remaining difference in
length that was not taken up by the member precompression (  io r from Equation 3.174). For the situa-
tion shown here, where the inner and outer member have an initial position of zero (i.e. the members are
lined-up at the inner member connection side), this remaining length difference affects only the inner

member response. If the members were lined up at the other end of the brace at the start, the remaining
length difference would affect only the outer member; however, the derived effective brace stiffness would
be the same because there would be an equal shift in the activation deformations. Using these parameters,
the positive activation deformations for each member may be calculated as follows:
 – P i0 + F I2
 ------------------------ if  io  0
+  ki
 ia =  (Eq. 3.177)
 – P i0 + F I2
 - –  io r
----------------------- if  io  0
 ki
+ – P o0 + F I1
 oa = ------------------------ (Eq. 3.178)
ko
+ +
where  ia is the activation deformation of the inner member in tension and  oa is the activation deforma-
tion of the outer member in tension. The negative (compression) activation of each member hysteresis
occurs at a load that is equal to the total tendon pretension plus the slip load of the friction damper at one
end of the brace. This is equal to –  P p0 + F I1  for the outer member and –  P p0 + F I2  for the inner
member as shown in Figure 3.24. Again, using the initial load in the members and the remaining differ-
ence in length between the two members after pretensioning, the negative activation deformations for each
member are:
 –  P i0 + F I2 + P p0 
 ------------------------------------------- –  io r if  io  0
-  ki
 ia =  (Eq. 3.179)
 –  P i0 + F I2 + P p0 
 ------------------------------------------- if  io  0
 ki
- –  P o0 + F I1 + P p0 
 oa = -------------------------------------------
- (Eq. 3.180)
ko
- -
where  ia is the activation deformation of the inner member in compression and  oa is the activation
deformation of the outer member in compression.
Using these activation brace deformations for each member, the full activation deformation of the brace
in each direction may be determined by selecting the largest of the two member deformations in each load-
ing direction:

 +  -
+   -  
a = max  ia and a = min  ia (Eq. 3.181)
 +  -
 oa  oa
If it is assumed that the activation load of the brace is equal to P p0 + F , then the activation deformations
may be used to calculate an approximate effective initial stiffness for the SCED brace:
2  P p0 + F 
k 1 eff = ------------------------- (Eq. 3.182)
+ -
a – a
For the purposes of modelling, the stiffness of an effective SCED brace element should include the stiff-
ness of the connection in series with the brace. This effect is considered by the SCED Mechanics Simulator
as a series stiffness. To include this connection stiffness in the determination of an effective initial stiffness,
the total effective stiffness of the brace ( k 1 tot ) may be calculated as follows:
1
k 1 tot = -----------------------------
- (Eq. 3.183)
1 1
----------- + -----------
k 1 eff k conn
where k conn is the total axial stiffness of the connections in series with the brace. Note that the connection
stiffness changes the effective stiffnesses of the entire SCED brace hysteresis, not only the initial stiffness.
To test this method for estimating the initial effective stiffness, the results from the equations above
were compared to the SCED Mechanics Simulator results for five fully-designed SCED braces. These
SCEDs were designed for the building frame models that are described in Chapter 7. For now, only the
details of the braces that are pertinent to the determination of the effective initial stiffness are shown. A full
detailed discussion of the design of these braces is provided in Chapter 7. These original-style SCEDs (O-
SCEDs) were designed for a six-storey building frame. Storeys one and two used the same brace, so there
were five different brace designs. The brace designs are summarized in Table 3.16. The braces are referred
to by their storey location in that model.
Each of the designs was modelled using the SCED Mechanics Simulator to determine the detailed pre-
dicted hysteretic behaviour. This modelled hysteresis was then used to calculate the effective initial stiffness
by finding the secant stiffness between the full activation points in tension and compression. For each
SCED design, the effective initial stiffness was found for three different scenarios: (1) the members having
equal lengths, (2) the inner member being 1.5mm shorter than the outer member, and (3) the outer mem-
ber being 1.5mm shorter than the inner member. The resulting modelled values are shown in the right-

Table 3.16: O-SCED Brace Designs Used to Check the Effective Initial Stiffness Calculation
Inner/Outer Inner Outer Connection Tendon End 1 Friction End 2 Friction

Member Member Member Stiffness Pretension Damper Slip Damper Slip
Storey
Length Stiffness k i Stiffness k i k conn P p0 Force F I1 Force F I2
(mm) (kN/mm) (kN/mm) (kN/mm) (kN) (kN) (kN)
S1/S2 6500 618 637 3000 1053 450 450
S3 6500 618 637 3000 894 400 400
S4 6500 618 824 2500 787 350 350
S5 6500 618 824 2500 617 250 250
S6 6500 418 824 2000 398 200 150
most column of Table 3.17. The remaining columns in that table show the intermediate and final results
from the approximate calculation of the effective initial stiffness using the approach that was developed in
this section. This table shows that the approximate method for determining the effective initial stiffness
provides a good prediction of the value that is calculated using the SCED Mechanics Simulator results.
The difference between the Simulator results and the estimated values was a maximum of 4.2%.
For the same-length scenarios in Table 3.17 the effective stiffness values are in the range of approxi-
mately 80-95% of the theoretical initial stiffness calculated by simply finding the sum of the inner and
outer member stiffnesses. When a 1.5mm length construction tolerance is taken into account, the effective
initial stiffness values drop to a range of approximately 50-70% of the theoretical values (or 65-80% of the
modelled same-length effective stiffness). These results show that there is a potential for significant
dynamic behavioural differences between models that assume the theoretical initial stiffness versus those
that use an effective stiffnesses and that take construction tolerances into account. The effect of these dif-
ferent initial stiffnesses on the dynamic response of a structure will be studied in the context of a full six
storey model structure in Chapter 7.
3.5.3 THE EFFECTIVE INITIAL STIFFNESS OF A T-SCED BRACE
As was the case for the detailed mechanics of the T-SCED braces with equal member lengths, the deter-
mination of a closed-form estimate for a T-SCED brace’s effective initial stiffness is a more complex task
than it was for the O-SCED brace above. The introduction of a third axial element (the intermediate
member) adds a multitude of additional possible member length scenarios. The hysteretic response of a
sample T-SCED brace for four such scenarios are shown in Figure 3.25. The left hysteresis in the top plot
of this figure shows the small amplitude hysteretic behaviour of a sample T-SCED brace using the simpli-

Table 3.17: Effective Initial Stiffness Sample Calculations

Estimated Estimated Calc. SCED
Element Inner Outer Remaining Activation Activation
Effective Mech. Calc.
Length Member Member Length Def. Def. Initial Initial vs.
Storey Diff. Precomp. Precomp. Diff. (Tens.) (Comp.) Stiffness Stiffness SCED
 io P i0 P o0  io r + -
a a k 1 tot k 1 tot Mech.
(mm) (kN) (kN) (mm)
(mm) (mm) (kN/mm) (kN/mm)
S1/S2 0 -519 -534 0 1.57 -1.59 876 849 +3.2%
-1.5 -989 -64 0 2.33 -2.26 663 650 +2.0%
1.5 -48 -1005 0 2.28 -2.35 658 645 +2.0%
S3 0 -440 -453 0 1.36 -1.38 876 850 +3.1%
-1.5 -894 0 -0.06 2.15 -2.03 638 627 +1.8%
1.5 0 -894 0.10 2.03 -2.19 633 623 +1.6%
S4 0 -338 -450 0 1.11 -1.29 828 803 +3.1%
-1.5 -787 0 -0.23 2.07 -1.38 642 629 +2.1%
1.5 0 -787 0.54 1.38 -2.38 600 586 +2.4%
S5 0 -265 -352 0 0.83 -0.97 828 795 +4.2%
-1.5 -617 0 -0.50 1.90 -1.05 580 571 +1.6%
1.5 0 -617 0.75 1.05 -2.15 545 535 +1.9%
S6 0 -114 -264 0 0.68 -0.99 619 607 +2.0%
-1.5 -398 0 -0.55 1.86 -0.73 449 445 +0.9%
1.5 0 -398 1.02 0.73 -2.33 394 389 +1.3%
Mean +2.2%
Standard Deviation 0.9%
fied brace behaviour expressions from Section 3.3. The other four hystereses in this plot show the small
amplitude hysteretic response from the SCED Mechanics Simulator for the four different length scenarios:
equal length members, short inner and intermediate members, short outer and intermediate members, and
short intermediate members. This is not an exhaustive set of member length scenarios, but it does provide
some representative cases that result in low effective initial stiffness. As this figure shows, the stiffness tran-
sitions are significantly different for each scenario. Also, the resulting effective initial stiffness for the differ-
ent length scenarios vary significantly from each other, but are all much lower than the theoretical initial
stiffness calculated as the sum of the stiffness of all the axial members.
Since the initial stiffness transitions of T-SCED braces are caused by the complex interaction of the
brace elements, especially when member length tolerances are taken into account, the best option to deter-
mine the effective initial stiffness of a T-SCED brace is to use the SCED Mechanics Simulator. Multiple

Sample T-SCED Brace Hysteresis from SCED Mechanics Simulator

ki = 301kN/mm, km = 396kN/mm, ko = 779kN/mm, kp = 9.3kN/mm, F = 170kN, Pp0 = 200kN, L = 5030mm
Mech. Sim. Mech. Sim. Mech. Sim. Mech. Sim.
Simple Equal Inner (-1.0) Outer (-1.0) Inter. (-1.0)
Calc. Lengths Inter. (-0.5) Inter. (-0.5) (Others = 0)
400
k1,eff 1485 kN/mm 401 kN/mm 314 kN/mm 431 kN/mm 226 kN/mm
Activation
300 Point
Step
Behaviour
200
Brace 100
Axial Origin
Force 0
P
(kN) -100
-200
Effective
-300 Initial Stiffness
k1,eff
-400
-6 -4 -2 0 +/-2 0 +/-2 0 +/-2 0 2 4 6
Mech. Sim. Mech. Sim. Mech. Sim. Mech. Sim.
Equal Inner (-1.0) Outer (-1.0) Inter. (-1.0)
Lengths Inter. (-0.5) Inter. (-0.5)
400
300
200
Brace 100
Axial
Force 0
P
(kN) -100
-200
-300
-400
-15 -10 -5 0 +5/-10 +10/-5 0 +5/-10 +10/-5 0 +5/-10 +10/-5 0 5 10
Figure 3.25: Sample T-SCED Brace Hystereses - Effect of Length Tolerance
different length scenarios should be simulated to determine a realistic range of effective initial stiffnesses for
use in modelling. This design exercise using the SCED Mechanics Simulator was conducted as part of the
design of the building models that will be described in Chapter 7. The resulting brace designs had effective
initial stiffnesses that were in the range of approximately 35-40% of the theoretical initial stiffness for equal
member length scenarios and approximately 25-35% for scenarios that considered a 1.5mm construction
length tolerance (70-100% of the equal member length scenario). These results show that for a T-SCED

the theoretical initial stiffness, which is equal to the sum of the stiffnesses of the inner, intermediate and
outer members, cannot be used to determine the initial stiffness of the brace, even if the members are
exactly the same length; however, the results of the six storey building models constructed and analyzed in
Chapter 7 will show that using the realistic initial stiffness instead of the theoretical stiffness results in a
moderate decrease in the model accelerations, but has no significant effect on the storey drift or base shear.
This suggests that extremely tight controls on the length tolerances that were used for the high capacity
SCED brace prototype that will be described in Chapter 4 and the T-SCED prototype described in Chap-
ter 5, may not be necessary for practical brace designs; however, it is important to note that the initial stiff-
ness is still important because it affects the elastic period of the building, which determines the earthquake
design loads. In addition, a lower initial stiffness will increase the lateral deformations due to wind loading.
3.5.4 RAPID ESTIMATION METHOD FOR DETERMINING THE EFFECTIVE INITIAL STIFFNESS OF SCED
AND T-SCED BRACES FOR PRELIMINARY DESIGN
During the initial iterations for SCED-braced frames, it may be desirable to be able to determine quick
estimates for the effective initial stiffness of SCED and T-SCED braces without having to use the SCED
Mechanics Simulator or the simplified stiffness calculation method for SCED braces that was described in
Section 3.5.2. Such a rapid estimation method would facilitate the estimation of approximate building
frame modal periods in order to calculate refined brace loads at each level. It would also facilitate the esti-
mation of the lateral stiffness of a building frame for use in the determination of approximate lateral defor-
mations due to wind loading. Once these approximate loadings and deformations are determined, the
SCED braces may be more fully designed and the more accurate methods may be used to find the final
brace properties including the initial stiffness.
Using the sample SCED and T-SCED braces that were modelled using the SCED Mechanics Simula-
tor, upper- and lower-bound effective initial stiffness values may be estimated relative to the theoretical ini-
tial stiffness of the brace. The theoretical initial stiffness k 1 of SCED and T-SCED braces was previously
provided in Equations 3.6 and 3.154, respectively. This stiffness depends only on the stiffness of the axial
members and the tendons. The axial member stiffnesses may in turn be estimated based on the maximum
anticipated axial load in the brace and prior design experience. Since, the tendon stiffness in these equa-
tions has a very small effect on the total theoretical stiffness, it may be neglected. Using the SCED
Mechanics Simulator results from some typical SCED and T-SCED brace designs that will be fully
described in Chapter 7 and the prototype brace test results that will be described in Chapters 4, 5 and 6,
upper- and lower-bound estimates of the effective initial stiffness k 1 eff may be taken as follows:

1. For SCED braces, the lower-bound is approximately 75% of the theoretical

initial stiffness and the upper-bound is approximately 100% of the theoretical
initial stiffness
2. For T-SCED braces, the lower-bound is approximately 30% of the theoretical

initial stiffness and the upper-bound is approximately 50% of the theoretical
initial stiffness
The lower-bound estimates would be useful for the conservative calculation of building drifts and the
upper-bound would be useful for the conservative calculation of building forces due to modal periods.
The above estimates assume that the lengths of the axial members are identical. If the designer wishes to
estimate the effect construction tolerances on the lengths of the axial members at the preliminary design
stage, the following method may be used:
1. Starting with the effective stiffness estimate using the approximate percentages
provided above, and using the target activation load P a , calculate the
approximate elongation at the activation point  a :
Pa
 a = ----------
- (Eq. 3.184)
k 1 eff
2. Determine the anticipated maximum difference between the lengths of the

SCED or T-SCED brace’s axial members  io ant . This should typically be
equal to two times the construction length tolerance value since one member
could be longer by the magnitude of the construction tolerance and the other
member could be shorter by the same amount.
3. Find the a new effective stiffness by assuming that the anticipated length
difference increases the total activation deformation from the negative
activation point to the positive activation point:
2P a
k 1 efftol = -----------------------------
- (Eq. 3.185)
2 a +  io ant
This new effective stiffness estimate modifies the previous estimates to take into account the effect of con-
struction length tolerances.

3.6 SCED BRACE MODELLING USING EFFECTIVE HYSTERETIC PARAMETERS

To model SCED braces within the context of full frame or building designs, an implementation of a
multi-spring SCED brace analysis such as the one used by the SCED Mechanics Simulator would not be
practical. Therefore, for dynamic structural modelling, a SCED brace is typically modelled as a single non-
linear spring that is programmed to replicate its effective flag-shaped response. Two existing implementa-
tions of such models previously exist for the nonlinear structural analysis software packages Ruaumoko
(Carr, 2005a) and SAP2000 (Computers and Structures Inc., 2009; Le Bec et al., 2010). To extend upon
this previous work, a robust single element flag-shaped hysteresis formulation that includes the option to
model external friction fuse behaviour in series with a SCED brace has been developed in this thesis for the
nonlinear structural analysis package OpenSees (McKenna et al., 2000).
Traditionally, the properties that were input into one of these models were based on the simplified
mechanics determined in Section 3.2.1; however, for a more representative response, effective values for the
initial stiffness, post-activation stiffness and activation forces may be determined using the results from the
mechanics simulator model. This approach will be taken for the construction of the new building models
that will be discussed in Chapter 7.
3.6.1 RUAUMOKO IMPLEMENTATION (CARR, 2005A)
Ruaumoko is a nonlinear structural analysis package developed by Professor Carr at the University of
Canterbury in Christchurch, New Zealand (Carr, 2005a). It has been widely used by academics in the
earthquake engineering field to model the dynamic response of structures and soils. Ruaumoko already
contains the ability to model a self-centering hysteretic behaviour within a single element.The figure from
the Ruaumoko manual (Carr, 2005b) depicting the hysteretic response of the flag model is shown in Figure
3.26. This simple model allows the definition of a flag-shaped hysteretic response based on the initial stiff-
ness ( k 1 which is named K 0 in the Ruaumoko model), post-activation stiffness ratio ( r = k a  k 0 ), activa-
tion force ( P a which is named F y in the Ruaumoko model), and the energy dissipation capacity parameter
(  shown as ‘Beta’ in the figure). The Ruaumoko program using this hysteretic model has been previously
used to successfully model SCED-braced frames by Tremblay et al. (2008) and Choi et al. (2008).
3.6.2 SAP2000 IMPLEMENTATION (LE BEC ET AL., 2010)
SAP2000 (Computers and Structures Inc., 2009) is a commercial structural engineering analysis pack-
age that is widely used by practicing engineers. SAP2000 does not have any built-in hysteretic behaviour

Figure 3.26: Flag Shaped Bilinear Hysteresis for Ruaumoko (from Carr, 2005b)
option that is able to model the flag-shaped hysteretic behaviour of SCED braces. Le Bec et al. (2010) out-
lined a method to model SCED braces in SAP2000 using two parallel nonlinear elements in series with a
single linear frame element. While this is not a single element brace model, it is still significantly simpler
than the model used by the SCED Mechanics Simulator. This method will be used to construct a model of
the SCED braced frame shake table specimen that will be described in Chapter 6. The modelling steps
developed by Le Bec et al. (2010) are summarized briefly below.
The layout of the elements that are necessary to model a SCED brace in SAP2000 and the individual
hysteretic shapes and parameters for those elements are shown in Figure 3.27. The simulated SCED brace
consists of one very short linear frame element (LFE) with an axial stiffness of k 1 LFE , that is in series with
two parallel link elements; one multi-linear elastic (MLE) element with an initial axial stiffness of k 1 MLE ,
a post-activation axial stiffness of k a MLE , and an activation force of P a MLE ; and one multilinear plastic
(MLP) element with an initial axial stiffness of k 1 MLP , a post-activation axial stiffness of k a MLP , and an
activation force of P a MLP . Since the model requires the use of some high-stiffness values to create the flag
shape, a stability parameter (  ) is used to put a limit on that high-stiffness and ensure the numerical sta-

bility of the model. This parameter does not affect the hysteretic behaviour of the brace model and it was
found that a value of 20 resulted in a stable model behaviour. Given the target SCED properties (the initial
stiffness k 1 , activation load P a , post-activation stiffness k a , and the energy dissipation capacity parameter
 , the properties for each element in the SAP model should be:
k 1
k 1 LFE = ------------ (Eq. 3.186)
–1
k
k 1 MLE = k 1 MLP = --------1- (Eq. 3.187)
2
k 1 k a
k a MLE = -------------------------------------
- (Eq. 3.188)
k 1 –   – 1 k a
k a MLP = 0 (Eq. 3.189)
P
P a MLP = ---------a (Eq. 3.190)
2
P a MLE = P a – P a MLP (Eq. 3.191)
To ensure model stability, the short linear frame element should have a length that is on the order of 0.1%
of the total length of the brace.
3.6.3 NEW OPENSEES IMPLEMENTATION
OpenSees (McKenna et al., 2000) is a nonlinear structural analysis package developed at the University
of California at Berkeley. It has become the standard analysis package for academics in the earthquake engi-
neering field. A self-centering hysteresis was not previously available for OpenSees, and so one was created
in the form of a new uniaxial material, that is now included with the default OpenSEES distribution. This
material includes an optional parameter that allows the hysteresis to model the effect of an external friction
fuse in series with a SCED brace, as well as additional optional parameters to model the effect of bearing
due to limited deformation capacity of the brace or the external fuse.
The hysteretic response of the new material model is shown in Figure 3.28. In the figure, the parame-
ters that begin with a ‘$’ are the model inputs. $k1 is the initial stiffness of the brace, $k2 is the post-acti-
vation stiffness of the brace, $sigAct is the activation stress/force, $epsSlip is the external fuse slip strain/
deformation, $beta is the energy dissipation capacity parameter, $epsBear is the total strain/deformation

Multi-Linear Elastic
Element (MLE)
P
Pa,MLE ka,MLE
k1,MLE
δ Flag-Shaped
Hysteretic Response
-Pa,MLE P
Pa ka
kMLE βPa
k1
kLFE
Linear Frame
kMLP = δ
Element (LFE)
Multi-Linear Plastic -Pa
P
Element (MLP)
k1,LFE P ka,MLP = 0
Pa,MLP
δ k1,MLP
δ
-Pa,MLP
Note: Element lengths are not to scale
Figure 3.27: Constructing a Flag-Shaped Hysteresis in SAP2000 (adapted from Le Bec et al., 2010)
that will cause the external friction fuse to bear (since it may have a limited amount of travel), and $rBear
is the bearing stiffness as a ratio of the initial stiffness $k1.
3.7 SCED-BRACED FRAME DESIGN CONSIDERATIONS

Based on the complete understanding of the elements and behaviour of SCED and T-SCED braces pre-
sented in this chapter, this section will briefly introduce the design issues and considerations that relate to
the incorporation of SCED braces into building frames. These design considerations include the determi-
nation of an appropriate construction tolerance and the resulting effective initial brace stiffness, the identi-
fication and design of the ultimate limit state response, the selection of a response modification coefficient
(R-factor) for design and the comparison between the behaviour and stiffness of SCED frames and BRB
frames. The section will conclude with a discussion of practical SCED-braced frame design procedures.
3.7.1 REASONABLE CONSTRUCTION TOLERANCES AND THE DETERMINATION OF THE EFFECTIVE

INITIAL STIFFNESS OF SCED BRACES
Section 3.5 outlined the effect of axial member construction length tolerances on the behaviour of
SCED and T-SCED braces. It included examples of realistic SCED brace hysteretic responses, identifica-

Stress or
Force
Post-Strain
Linear Bearing
Slip Strain $epsSlip
(If = 0, No Slip) Stiffness
($rBear)($k1)
Activation Stress $k2

$sigAct ($beta)($sigAct)
$k1
Strain or
Deformation
Bearning Strain
$epsBear
(If = 0, No Bearing)
Figure 3.28: OpenSees Self-Centering Hysteresis
tion of the hysteretic ‘step behaviour’ caused by the difference between member lengths, and three different
methods that may be used to determine the effective initial stiffness of SCED braces. Those three main
methods (in order of increasing accuracy) are: the rapid estimation method (Section 3.5.4), the estimation
method (Section 3.5.2 - SCED brace only), and use of the SCED Mechanics Simulator (Section 3.4).
There is no single correct way to determine an appropriate construction tolerance for the design of
SCED braces; however, there are a two important factors that a designer should consider when deciding
how much tolerance is acceptable for a given project. The first factor is cost. Attaining very precise length
tolerances for the long SCED brace axial members may add significant expense. In addition, a designer
should consider that to achieve an effective initial stiffness that is close to the theoretical initial stiffness of
a SCED brace, extremely precise manufacturing methods are necessary. Even for the laboratory prototype
specimens, which were manufactured with the most stringent tolerances possible (Chapters 4 and 5), the
measured initial stiffness was significantly less than the theoretical initial stiffness.
The second factor that a designer should consider when determining an appropriate construction toler-
ance is the potential for the brace to exhibit the ‘step behaviour’ that was described previously in Section

3.5.1. This step behaviour reduces the effective initial stiffness of the SCED brace; although perhaps more
importantly, it also introduces additional nonlinearity and stiffness transitions that would not be captured
by a building frame model. It is important to note that the step behaviour never occurs at a axial load that
is lower than the total friction slip force provided by the internal friction dampers, so it is not likely to pro-
duce any zero-stiffness brace elongations when subjected to normal service loads. If the wind load on the
structure is greater than the friction damper slip load, then the step behaviour may cause small zero-stiff-
ness movements during wind storms. If there are no internal friction dampers in the SCED brace and
damping is instead provided entirely by viscous devices, this step behaviour may theoretically occur under
service loading. Even then, they may not be significant since the magnitude of such movements cannot be
greater than the total length tolerance for the axial members. Thus, the dynamic effect would be similar to
the slippage of a bolted joint in a braced frame.
For the full SCED brace designs that will be described in Chapter 7, the effective initial stiffness of each
SCED brace design was determined by assuming that the length of each steel member could vary from the
length of the other members by up to 1.5mm. This is equivalent to a variation of 1/32 of an inch in either
direction, which is the tolerance for columns with both ends finished for contact bearing in the AISC Code
of Standard Practice for Steel Buildings and Bridges (AISC, 2005c). This level of tolerance may provide
designers with a good balance between cost and performance, since it is commonly used in practice for
long members (columns) and should therefore be relatively easy to achieve. The tolerance will provide at
most a 1.5mm amplitude step in the hysteretic response. The true step length will be shorter by an amount
equal to the elastic deformation of the longer member under the tendon pretension load.
3.7.2 FORCE-BASED DESIGN CONSIDERATIONS
Even when the reduction effect of the effective initial stiffness is taken into account, SCED-braced
frames are still very stiff. Compared to buckling-restrained braced frames (previously discussed in Section
2.4), SCED-braced frames are significantly stiffer. The analytical study by Choi et al. (2008) compares the
dynamic behaviour of SCED-braced frames to equivalent BRB frames. The six storey building from this
study was adapted as a case study model that will be presented in Chapter 7. New fully-designed original
SCED (O-SCED) braces and telescoping T-SCED braces were designed for this six-storey building. The
theoretical and effective stiffness of each of these braces is compared to the stiffness of the SCEDs and
BRBs from the Choi et al. (2008) study in Table 3.18. The realistic effective initial stiffness values for those
new designs were determined using the SCED Mechanics Simulator, assuming a maximum 1.5mm differ-

ence in length between the axial members. The SCED braces from the Choi et al. (2008) study were not
fully designed, the brace properties were determined using a design spreadsheet; however, as the table
shows, the stiffnesses of the braces are similar to those of the more realistic newly designed T-SCED braces.
The table shows that the effective realistic initial stiffness for both types of SCED braces are much higher
than the stiffness of the BRBs.
Table 3.18: Initial Stiffness of Equivalent SCED Brace and Buckling-Restrained Brace Designs (kN/mm)
First Mode
Brace Storey 1/2 Storey 3 Storey 4 Storey 5 Storey 6
Period T 1
O-SCED (Theoretical) 885.0 885.0 914.6 914.6 766.3 0.76
O-SCED (Realistic) 645.0 623.0 586.0 534.5 388.7 0.83
T-SCED (Theoretical) 2232.8 2232.8 1161.4 1161.4 1166.4 0.68
T-SCED (Realistic) 674.0 560.0 387.0 348.0 280.0 0.86
SCED (Choi et al., 2008) 494.7 468.3 411.4 362.0 221.2 0.90
BRB Equivalent 277.0 243.0 210.5 172.1 113.4 1.055
(Choi et al., 2008)
In typical inelastic building design, a response modification coefficient R , also called the force reduc-
tion factor (or more generally called the ‘R-factor’), is used to reduce the design forces to account for the
ductility capacity of the inelastic elements. The assumption implicit in the use of the R-factor is that the
maximum displacement of an inelastic system is the same as the maximum displacement of an elastic sys-
tem with the same initial stiffness, also called the equal displacement approximation (Christopoulos & Fil-
iatrault, 2006). In a conventional yielding system such as a BRB frame, the R-factor is limited by the
ductility capacity of the system, which is defined as a ratio of the total possible deformation to the yield
deformation (Christopoulos & Filiatrault, 2006):
 max
  = ----------  R  1 (Eq. 3.192)
y
where   is the displacement ductility ratio,  max is the maximum inelastic deformation,  y is the defor-
mation at yield, and R is the response modification coefficient. For a yielding system, the initial stiffness is
directly related to the ductility since the same mechanism that defines the initial stiffness (e.g. the steel
cross-sectional area) also defines the yield force, yield deformation and maximum deformation. Therefore,
the ductility capacity is fairly consistent for a given type system and it depends on the geometry of that sys-

tem and the construction material. For a SCED brace, the initial stiffness and the maximum deformation
of the system are de-coupled. A ‘ductility capacity’ may be defined for the system as the ratio of the total
possible deformation to the activation deformation; however, this measure does not relate to any physical
‘ductility’. The total possible deformation of a SCED brace is dependant on the linear elongation capacity
of the tendons, which for a given tendon type is primarily dependant on the length of the brace. In con-
trast, the activation deformation is dependant on the initial stiffness of the brace and the activation force,
neither of which is directly related to the elongation capacity of the tendons. Furthermore, the initial stiff-
ness may be increased completely independently of the elongation capacity by simply increasing the sizes of
the axial members. For these reasons, the R-factor of a SCED brace does not have a physical meaning and
may not be consistent in all cases. This suggests that the use of force-based design methods for SCED-
braced frames is not ideal.
One alternative to the use of typical force-based design methods for building design is the use of a dis-
placement-based method such as the Direct Displacement-Based Design (DDBD) approach (Priestley et
al., 2007). This type of approach avoids the dependance on the initial stiffness that is typical of force-based
approaches. It does this by defining a structure’s dynamic properties based on the secant stiffness of an
equivalent SDOF structure at a target displacement. The DDBD method requires curves that define the
damping provided by the system as a function of the target displacement. It may be possible to derive such
curves for use with DDBD by using the results of single-degree-of-freedom (SDOF) study of self-centering
systems subject to earthquake excitations that was previously conducted by Christopoulos et al. (2002a).
This study provides ductility and energy-dissipation spectra for self-centering systems. These could poten-
tially be used to estimate the displacement and damping of an equivalent SDOF structure for use with the
DDBD approach. A displacement-based design approach to self-centering systems may be the most appro-
priate way to design a SCED-braced frame, and this warrants further study; however, it is outside the scope
of this thesis.
The numerical study by Tremblay et al. (2008) utilized a design approach for SCED-braced frames that
approximates a displacement-based design by starting with a reference BRB frame design. Each SCED
brace in that study was designed such that the axial force in the brace at 2% storey drift was equal to the
axial force in the equivalent BRB at the same drift. This resulted in the SCED and the BRB designs having
equal secant stiffness at the design drift level. Although the SCED-braced frames had higher initial stiffness
and less than half of the energy dissipation capacity of the equivalent BRB frames, this approach resulted in
SCED frame drifts that were less than those of the BRB frames under dynamic earthquake excitations

(Tremblay et al., 2008). If the initial stiffness had been used to determine the period of the structure and to
subsequently define the design loads, the SCED frame would have likely been over-designed relative to the
BRB and would have experienced lower deformations.
Although it is not ideal due to the questionable applicability of R-factor design to SCED braces, force-
based design of SCED-braced frames is possible. This may be done by using the same R-factor as that spec-
ified for BRB frames. These force-based SCED brace designs will result in higher design forces than a com-
parable BRB design due to the high initial stiffness of the SCED brace. The numerical study conducted by
Choi et al. (2008) used this approach. As previously shown in Table 3.18, the SCED braces designed for
this study had approximately twice the initial stiffness of the equivalent BRB frames. As the table also
shows, an equivalent fully designed T-SCED frame would have similar initial stiffness. Since the elastic
period is inversely proportional to the square root of the initial stiffness, the change in stiffness is not
directly proportional to the change in the design base shear. Even though the initial stiffness of the SCED
braces were approximately twice as large as those of the BRBs, the period of the SCED structure was only
15% less than that of the BRB building, and the design base shear for the SCED-braced frame was
3400kN as opposed to 3000kN for the BRB frame, a 13% increase. Like the Tremblay et al. (2008) design
method, the Choi et al. (2008) resulted in maximum drifts were similar for both types of frames.
3.7.3 THE EFFECT OF NON-STRUCTURAL COMPONENTS
Examples of non-structural components include interior partitions, surface finishes, cladding, and
mechanical services. The effect of these non-structural components on the seismic response of a building
are not typically considered during the design of a structure; however, they may provide some force resis-
tance and energy dissipation. If the energy dissipation capacity parameter  of the SCED braces in a frame
(previously described in Section 3.2.1) is equal to 1.0, then the combined hysteretic behaviour of all these
non-structural elements may potentially provide enough force resistance to prevent a SCED building from
fully self-centering. The effect of the non-structural components is equivalent to the effect of adding low-
capacity hysteretic dampers in parallel with the SCED-braced frames. To avoid this problem and to ensure
that the SCED braces provide full self-centering behaviour, the energy dissipation capacity parameter of
the braces is typically designed to be between 0.80 and 0.95. This maximizes the available energy dissipa-
tion while providing some extra restoring force capacity to self-center the non-structural components.
Additionally, a recent study by Eatherton and Hajjar (2011) used an SDOF study to determine that
even if the energy dissipation capacity parameter  of a self-centering system is as large as 1.5, that system

will still be able to prevent any significant residual drifts. This study seems to indicate that the presence of
non-structural components will not adversely affect the behaviour of self-centering structure.
3.7.4 THE ULTIMATE LIMIT STATE OF SCED BRACES
In order to retain the SCED brace prototypes for use in future studies, none of the tests in this thesis
directly investigate the ultimate limit state response of SCED braces; however, this section will discuss the
design implications associated with the achievement of the ultimate limit state.
As discussed in Sections 3.2 and 3.3, once a SCED or T-SCED brace has activated, all of the force is
transferred through the tendons and the friction interfaces. The tendons are designed to resist the maxi-
mum expected load and are effectively tested up to that maximum during installation before they are
released to the design lock-off pretension load (see Section 3.1.5). This minimizes the chances that a ten-
don will break during operation; however, even if all the tendons were to fail, a SCED or T-SCED brace
would revert to a simple ductile steel cross-brace. Since the friction interface slot length is designed for the
maximum expected brace deformation, if the brace attempts to exceed that deformation, the friction bolts
bear on the ends of the slots. The brace load would then carried through the inner and outer members
directly via bearing in the friction interfaces. Therefore, even though the primary load bearing elements of
the SCED and T-SCED braces consist of new advanced materials, the ultimate response is still governed by
traditional steel members.
Once bearing occurs in the friction interface slots, there are three possible ultimate limit states. First,
the interface bolts could fail. This limit state is not desirable, but it may be avoided by building steel blocks
into the friction interface which would allow direct bearing of the inner plate when the bolts reach the ends
of the slots. Second, the friction interface plates could yield in shear. Third, the brace connections could
yield. The preferred ultimate limit state for the brace would be the yielding of the connections. This yield
force may be easily controlled, especially at the inner member connection side. The achievement of this
connection yielding limit state may be ensured by capacity-designing the friction interface to prevent yield-
ing or failure prior to yield of the connection plate. Specifically designing the brace connection plate to
yield in this way would add a full ductile yielding mechanism on top of the underlying SCED brace mech-
anism. This would provide a level of performance at the ultimate limit state that greatly exceeds the perfor-
mance that is provided by traditional yielding systems such as the BRB.

3.7.5 SAMPLE SCED-FRAME DESIGN METHOD
This section will present a sample force-based SCED-braced frame design method. A starting point for
the SCED brace demand loads may be determined by using an equivalent lateral force procedure (ASCE,
2005; NRC, 2005). These loads may then be refined through the use of a modal response spectrum analy-
sis or dynamic time history analysis. As mentioned previously, previous studies have used the same
response modification coefficient (R-factor) that is specified for buckling-restrained braces (Choi et al.,
2008; Tremblay et al., 2008). The fundamental design choice once the brace demand forces are known is
whether the braces should be designed so that the SCED brace activation load is greater than the demand
load or so that the SCED brace load at a target drift (such as 2.0% drift) is greater than the demand load.
Neither of these two methods is typically preferred. The SCED braced frames that were designed by Choi
et al. (2008) equated the activation load to the demand, whereas the SCED braced frames designed by
Tremblay et al. (2008) designed the SCED braces such that they had the same axial load at 2.0% drift as a
comparable buckling-restrained braced frame. The SCED braces that will be discussed in Chapter 7 were
designed based on their activation load. The activation load method is significantly simpler because it does
not require the designer to know the post-activation stiffness of the brace before determining the target
activation load; however, there are some significant advantages to designing based on a brace load at a tar-
get drift level. For example, designing for the load at a target drift allows the capacity design loads of the
other structural members to be defined up-front. This would be particularly useful for the retrofit of an
existing structure. In addition, controlling the maximum force in the brace during the initial design may
help to control building accelerations.
Once the SCED brace design loads are determined, the braces themselves may be designed. SCED
brace design is fundamentally an iterative procedure; however, a suggested procedure for designing a
SCED brace for a target activation load which minimizes the amount of iteration is provided below:
1. Determine the braced-frame geometry.
2. Determine the SCED brace demand load using code methods (equivalent
lateral force procedure, modal response spectrum analysis, etc.).
3. Determine the SCED brace target activation load by equating it to the demand
load.
4. Select a desired energy dissipation capacity parameter (  ) value. This value is

typically chosen to be between 0.80 and 0.95 to maximize the energy

dissipation while allowing some reserve self-centering capacity to account for

the hysteretic response of the building’s non-structural components.
5. Using Equations 3.15 and 3.16, determine the target internal friction interface
total slip force F and the target total tendon pretension force P p0 .
6. Assuming a tendon length based on the frame geometry, an effective cyclic

Young’s Modulus E p , a conservative tendon elongation capacity and nominal
strength (see Section 3.1.5), and assuming that the tendon is pretensioned
between 25% and 40% of its nominal strength, determine the total tendon
cross-sectional area required to achieve the target tendon pretension force P p0 .
Select a tendon diameter and determine how many tendons will be required.
Recall that T-SCED braces require double the amount of tendons.
7. Confirm that the tendons that are selected have sufficient elongation capacity to
accommodate the maximum anticipated brace deformation. Note that the
critical case for this evaluation is when the brace is in tension. Determine if
there is a need for a external friction fuse to increase the elongation capacity of
the SCED brace system. If no external fuse is used, include a safety factor on the
estimated drift demand that is used to calculate the suitability of the tendons.
8. Assuming a given amount of friction slip force per normal force bolt (as
discussed in Section 3.1.2), determine how many bolts will be required to
achieve the target internal friction interface total slip force F . Divide these bolts
among two or four separate friction interfaces located at one end or both ends
of the brace.
9. Design the brace cross-section. Layout the tendons and allow sufficient
clearance between tendon anchors. Select steel section sizes for the inner and
outer members (and the intermediate member for a T-SCED brace). These
sections must be able to resist the full axial load of the brace at the maximum
brace deformation and must be able to geometrically contain the tendons. The
approximate maximum axial load may now be calculated based on the member
stiffnesses, the tendons pretension, the total friction slip force and the tendon
stiffness. Ensure that the end plate geometry allows the plates to contact the

appropriate axial members. This is especially critical for the T-SCED brace as
described in Section 3.3.
10. Design the connections at either end of the brace for the full axial load of the
brace at the maximum brace deformation.
11. Determine the anticipated length tolerance for the inner, outer and
intermediate members.
12. Use the SCED Mechanics Simulator or the method described in Section 3.5.2
to determine the effective initial stiffness of the SCED brace and to confirm the
expected brace behaviour. This new effective initial stiffness may modify the
brace demands that were determined in Step 2. If so, redesign the brace for the
modified demand load.
13. If necessary, design an external friction fuse to act in series with the SCED
brace.
14. [Optional] Conduct time history analyses of the SCED braced frame using the
braces’ effective initial stiffness, post-activation stiffness, activation load, and
energy dissipation capacity parameter to construct effective hysteretic models
for each brace (using the methods discussed in Section 3.6).
To achieve the demand load at a target drift level instead of at the brace activation point, the procedure
would be similar to the one above except that more iterations would be required because the brace activa-
tion load can then only be determined after the tendon area has been selected.
Use of larger diameter tendons may result in more compact brace designs; however, each brace should
have at least four tendons so that the loads on the end plates are balanced and to provide redundancy.
3.8 CONCLUSIONS
The function and behaviour of each of the components of a SCED braces were described, including the
axial members, internal friction damper, external friction fuse, tendons, tendon anchors, end plates and
connections. This discussion provided the background necessary for a discussion of the mechanics of
SCED and T-SCED braces.

The mechanical behaviour of the SCED and T-SCED braces has been completely characterized
through a combination of a simplified step-wise solution for the hysteretic response for both types of
braces, a full step-wise solution for the hysteretic response of the SCED brace, and a newly created software
tool that is able to model the detailed behaviour of any SCED or T-SCED brace. Using these methods, a
SCED brace designer may easily determine either the approximate response of the brace for use in initial
design iterations, or the detailed response subject to varying member length tolerances, member stiffnesses
and the friction damper slip forces at either end of the brace. In addition, an two approximate methods
have been proposed which allow estimation of the effective initial stiffness of a SCED brace including the
consideration of the member length tolerance.
At the system level, methods previously exist for modelling the behaviour of SCED braced frames using
simplified behavioural models using the nonlinear analysis programs Ruaumoko and SAP2000. To extend
the use of SCED brace models, a new element type was developed for use in the popular nonlinear struc-
tural analysis software OpenSees that includes the behaviour of an external friction fuse in series with the
SCED brace.
Some key design considerations for the design of SCED-braced frames were presented along with a
sample design procedure which minimized the amount of iteration necessary in design. The physical
design of SCED braces presents a number of additional unique challenges that were not reflected in this
chapter, especially related to the design of the internal friction dampers and the use of the aramid tendons;
however, the mechanics and design methods presented in this chapter provide context that will be neces-
sary for the discussion of the design and behaviour of the individual experimental SCED brace prototypes
that will be presented in the following three chapters.

CHAPTER 4: DESIGN AND TESTING OF A PROTOTYPE HIGH-
CAPACITY (HC-SCED) BRACE
Prior to this work, the largest capacity SCED brace that had been designed and tested had an axial
strength of approximately 750kN at maximum elongation (Christopoulos et al., 2008). Competing high-
performance cross-bracing systems for buildings are available with higher axial force capacities. For exam-
ple, buckling-restrained braces (BRBs) are available with maximum axial yield forces in the range of
approximately 1000kips (~4000kN). For SCED braces to be useful for use in a similar range of building
contexts as BRBs, constructible SCED brace designs of such high capacity must also be available. To this
end, one of the main goals of this thesis was to design, build, and test a high-capacity SCED (HC-SCED)
brace with a target axial strength of 4000kN, giving it a capacity similar to the largest available buckling
restrained braces.
To achieve designs with such high capacities, it was necessary to make significant improvements to the
original SCED design to ensure that the load paths were capable of handling such high loads while still
retaining a compact form. The final design of the HC-SCED brace resulted in excellent seismic perfor-
mance; however, the pretensioned tendons in the brace experienced several unexpected premature failures
that will be discussed in detail in the following sections.
4.1 HIGH-CAPACITY SCED (HC-SCED) BRACE DESIGN AND INNOVATIONS

Although the high-capacity SCED (HC-SCED) brace would be tested in an axial load frame, it was
designed using the same assumed frame geometry that was used for the second SCED prototype test
(Christopoulos et al., 2008). This frame geometry was also similar to the storey geometry that was used for
the SCED-braced buildings that were designed for the analytical study conducted by Choi et al. (2009),
and is shown in Figure 4.1. A minimum connection length of 1500mm was assumed for each end of the
117
Chapter 4: Design and Testing of a Prototype High-Capacity (HC-SCED) Brace 118
brace. For this braced frame geometry, 1.0% storey drift is equivalent to 34.6mm of elongation in the
brace.
Assumed
Connection
0mm
Length al 975
1500mm gon 0mm
3750 al Dia 675
mm Tot eng
th
nL
/Te ndo
ube
il. T
Ava
θ = 22.6°
9000mm
Figure 4.1: High-Capacity SCED Design Frame Geometry
In the analytical study by Choi et al. (2009), SCED-framed buildings with heights between two and
twelve storeys were designed, modelled and subjected to multiple suites of earthquakes at different levels of
seismic hazard. The HC-SCED brace was designed to act as a first storey brace in the twelve storey build-
ing frame that was designed for that study. The brace that was modelled in that study had an activation
load of 2135kN while the HC-SCED brace was designed to have an activation load of 1950kN. The full
list of the initial target design parameters that governed the brace design process are shown in Table 4.1. As
with any SCED brace design, the determination of these design parameters was an iterative process. The
activation load was equal to the tendon pretension plus the internal friction damper slip force. The target
pretension in the tendons determined the number of tendons which were required. This, in turn, governed
the post-activation stiffness, which dictated the activation force that was necessary to reach the 4000 kN
target ultimate load at the peak brace deformation. In addition, the initial stiffness depended on the size of
the inner and outer members, which was, in turn, dependant on the arrangement of the tendons and the
clearance required for their anchors. For the development of this design, the tendon length that was chosen
was less than the length that was available according to Figure 4.1. This length was limited by the 7.7m
maximum height of the 12MN testing machine that would be used to test the brace. These design param-
eters were later modified sightly to accommodate practical tendon tensioning issues as will be discussed in
future sections.
The final design of the high-capacity (HC-SCED) brace is shown schematically in Figure 4.2. Full
technical design drawings of the brace may be found in Appendix B. Both the inner and outer members

Table 4.1: HC-SCED Target Design Parameters

Parameter Value
SCED Ultimate Axial Load P u 4000 kN
Storey Drift  T 2.0%
SCED Elongation  u 70 mm
Tendon Design Length L p 5770 mm
Tendon Pretension Force P p0 1018 kN
Total Internal Friction Damper Slip Force F it 925 kN
SCED Activation Load P a 1950 kN
Peak Tendon Strain  pu 1.8%
Energy Dissipation Capacity Parameter  0.95
Estimated Initial Stiffness k 1 830 kN/mm
Estimated Post-Activation Stiffness k a 30 kN/mm
were built-up steel sections: the inner member was a WWF450x228, and the outer member was a custom
built up 508mm (20”) square tube section with 15.9mm (5/8”) thick walls. Two sides of the outer member
tube had 31.9mm (1 1/4”) over-hangs to allow for welding of the built-up section. The pretension force in
the brace was provided by twelve 22mm nominal diameter aramid tendons. These tendons were larger
than the tendons that had been previously used for all of the previous SCED brace prototypes, which used
17mm nominal diameter tendons. Larger tendons were used in order to reduce the number of tendons and
anchors that were required and, hence, to reduce the necessary cross-sectional size of the steel members.
The tendons were anchored to 120mm (5”) thick steel end plates at either end of the brace using alumi-
num spike-and-barrel tendon anchors. To provide the energy dissipation, four internal friction dampers
were used, with six bolts in each providing the normal force on the friction interfaces. The friction in the
internal friction dampers was provided by stainless steel plates welded to the friction interface plate surface
and Carlisle NF-916 friction pads recessed into steel angles that clamped that interface plate. This HC-
SCED brace was also equipped with an experimental external friction fuse in series with the main brace
which is shown on the left side of the brace in Figure 4.2. This external fuse used an experimental steel-on-
stainless-steel interface to provide the friction. The external connections for the brace were perpendicular
to the longitudinal axis of the brace to permit the connection to the 12MN axial load frame.

7.7m
Internal ly
Friction Damper e mb
e Ass
Friction Interface
p let
Plate (welded to
Inner Section) Com
External
Friction Fuse
ly
Interface B
emb
Tendon Ass
Inner Member er
Outer Member e mb
te rM
Friction Ou
Interface
Angle
ly
Section A Internal Friction
e mb
Internal Friction Damper
Interface Angle Ass
m ber
Me
er
Inn
Plastic Guide
Friction Interface es
Plate (with Stainless P lat
d
Steel Surface)
& En
ly
mb
sse
nA
do
Ten
Section B Tendon
Tendon Anchor
End Plate
Figure 4.2: High-Capacity SCED Schematic
The most obvious difference between this HC-SCED brace and previous SCED brace designs, besides
its scale, is that the inner member was a WWF-Section (I-shaped section) instead of a tube. This change
allowed a more direct transfer of load between the connection and the inner member. To fabricate a con-
nection between the inner member and the applied brace load (through the external friction fuse), a
720mm length of the web of the WWF inner member was cut out. In place of this removed portion of the
web, a 50mm (2”) thick steel plate was welded to the remaining web and the top and bottom flanges of the
member. That thick steel plate extended past the end of the member and through a slot cut in the end
plate, making it available for use as an end connection for the inner member. This connection design
allowed the axial load to be transferred directly into the flanges and web of the WWF section. Previous
SCED designs relied on less-direct transfer of the load through slotted plate connections in the steel tube
that formed the inner member for those designs. This arrangement resulted in considerable shear lag,

reducing the efficiency of the connection. The use of an I-section for the inner member also allowed more
direct transfer of the load from the internal friction damper interface plates to the inner member.
To ensure that the axial members were the same length, the members were first cut roughly to length
and then both ends of the inner and outer member were machined to length within 1/1000 of an inch
(0.025mm). Such a precise machining process was not used for previous SCED brace prototypes. This
tight length tolerance was achieved by inserting the inner member into the outer member, temporarily
tack-welding them together, and then machining the faces of both members simultaneously. In addition,
the surfaces of the end plates were also machined to be flat and smooth. All of this was done because it was
believed that tight length tolerances were important to proper functioning of the brace; however, as
described in Section 3.7, these tight tolerances may not be necessary for future SCED brace designs. This is
good news because the machining process was extremely expensive.
The HC-SCED used an internal friction damper design that was similar to previous SCED braces,
with the internal friction damper interface plate welded to the inner member and surfaced with stainless
steel, which was then sandwiched between two friction angles that were welded to the outer member sur-
face. Previous SCED designs experienced significant flexing of the angle connection to the outer member.
Since the outer member is typically slotted to permit the presence and movement of the friction damper
interface plates, the tube wall does not generally have much stiffness on its own. The eccentric location of
the friction interface bolts with respect to the tube surface has, in the past, caused significant flexing of the
outer tube member wall during brace loading as shown in Figure 4.3. Previous designs had triangular angle
stiffeners as shown in the figure; however, these did not prevent outer tube wall flexing because they did
not transfer the load to the outer tube walls. Note that in previous designs, the friction interface was con-
tinuous, so that the flexing of the inner member wall was not significant. To mitigate this problem in the
HC-SCED design, the friction angles have each been stiffened with larger triangular steel plates that trans-
fer the vertical load caused by the eccentric bolt loading directly to the outer tube walls (as shown in Figure
4.2).
To guide the inner member within the outer member, ultra-high-molecular-weight (UHMW) polyeth-
ylene plastic pads were screwed to the inner member. The pads filled up the extra gaps in order to reduce
the friction between the two members during activation of the mechanism.
The cross-sectional geometry of the brace was constrained by the requirement for twelve tendons and
the need to arrange these tendons to fit within the cross-section of the outer member. In addition, practical
requirements related to the tendon tensioning process were also considered. For example, it was necessary

Section Friction Friction

Friction Interface Interface
Plate Line of Force Interface Angle
Angle Stiffener from Friction Pads Plate Angle Stiffener
Eccentricity Outer
Friction Interface Member
Angle
Inner
Member
Friction Plate
Effective Moment Section Continuity
on Outer Member
due to Eccentricity Flexing of Outer
Member Wall
Flexing of Inner
Member Wall
if Friction Plate
not Continuous
Figure 4.3: Flexing of Member Walls in Earlier Designs due to Eccentric Bolt Loading
to consider the clearances required for the hydraulic jack assembly. In addition, some clearance was
required between adjacent anchor nuts to allow the installation of these nuts during tendon pretensioning.
The brace cross-section along with the clearances for the tendon anchors is shown in Figure 4.4. To accom-
modate these tendons in the smallest possible cross-sectional area while being able to resist the total brace
compression load, a WWF450x228 section was chosen for the inner member. Originally, the HC-SCED
brace was designed to use a large HSS508x508x16 tube for the outer member; however, due to difficulties
that were encountered with the procurement of this size of section in such a small quantity, a custom-
designed built-up section was used instead. The design drawing shown in Appendix B does not reflect this
change, but Figures 4.2 and 4.4 show this as-built condition. The outer built-up tube section has 16mm
overlaps at the section corners to facilitate welding.
As previously mentioned, earlier SCED designs included an external friction fuse to increase the defor-
mation capacity of the brace beyond the limit of the tendon. Similarly, such a fuse was also incorporated
into the design of the HC-SCED brace; however, the Carlisle friction pads that were used previously to
provide the friction could not be arranged into a configuration to accommodate a slip load of 4000kN
because too many friction pads would be required. This limit on the number of friction pads is caused by
the pads’ axial strength, which limits the amount of normal force that may be applied on a given area. Due
to this limit, an alternate friction interface was required. Therefore, an experimental, high-capacity, steel-
on-stainless-steel friction interface using sixteen 1-1/2” normal force bolts was designed and tested for use

540mm
Outer Member
15mm 15mm
Inner Member
13mm
End Plate
68mm 95mm
11mm
508mm 450mm
54mm
Tendon Anchor
19mm Nut
Tendon Anchor
Connection
13mm 50mm Plate
450mm
508mm
Figure 4.4: High-Capacity SCED Cross-Section with Clearances
with the HC-SCED brace. Before testing the full brace with the fuse, preliminary tests of a prototype sub-
component version of this interface were performed which will be described in the following section.
The inner and outer members for the HC-SCED brace were fabricated by Constructions Beauce-Atlas
in Sainte-Marie de Beauce, Quebec. All of the other brace components except for the tendons were fabri-
cated by LCL-Bridge in Lachine, Quebec. Final assembly of the HC-SCED brace was performed by LCL-
Bridge under supervision. Tendon pretensioning was conducted with support of the staff at LCL-Bridge.
Some photographs of the brace elements and the assembly process are shown in Figure 4.5.
4.2 HIGH-STRENGTH EXTERNAL FRICTION FUSE SUBCOMPONENT DESIGN AND TESTING
4.2.1 EXTERNAL FRICTION FUSE SUBCOMPONENT SPECIMEN AND CALIBRATION
Prior to the full test of the HC-SCED brace with the full-sized external friction fuse, a subcomponent
version of the external fuse was tested in a smaller load frame to determine whether the steel-on-stainless-
steel interface was feasible. This test fuse had the same geometry as the full-sized friction fuse with the same
plate thicknesses, bolt spacings and bolt sizes, except that it consisted of only one-quarter of the full-sized
interface, with only 4 bolts out of 16. This test fuse was detailed to fit into the test load frame as shown in

Figure 4.5: High Capacity SCED Assembly Photos
Figure 4.6. Full design drawings of the friction fuse subcomponent test may be found in Appendix C. A
photo of the test setup in-place is shown in Figure 4.7.

Outer Assembly
(Yoke)
Stainless Steel
Friction Plate
1-1/2” A325 Bolts

with washers
Inner Assembly
(Tongue)
50 160
160
Grip
Stainless Steel 6
(typ.)
700
1250 max
1110 min
800
Total
Deformation
String
Potentiometer
160
Grip
138 226
Figure 4.6: External Friction Fuse Subcomponent Test Specimen
The test results from the external fuse subcomponent tests are summarized in Tables 4.2 and 4.3. The
steel surface arrived in a clean mill scale condition and was not conditioned in any way prior to these tests.
The four 1-1/2” bolts in the fuse were torqued by hand using a 25x torque multiplier with a hand torque
wrench. Calibration of the test interface slip load was performed using the input torque measured by the
torque wrench combined with direct bolt length measurements as previously described in Section 3.1.2.
Tests F-1 to F-3 were carried out to determine the coefficient of friction of the interface and achieve the

Figure 4.7: External Friction Fuse Subcomponent Test Specimen Photos
target slip force of the four bolt fuse and are shown in Figure 4.8. In three calibration steps, the fuse
reached the design slip force of 1000kN. The hysteretic behaviour of the fuse was good and did not have a
significant difference in force between the point when slip was initiated (static friction) and the point when
the fuse was slipping (dynamic friction) as shown in the figure.
Fuse Calibration Tests (F-1, F-2, F-3)

1000
800
600 Test F-3
400
Test F-1 Test F-2
Axial 200
Force 0
(kN)
-200
-400
-600
-800
-1000
-5 -4 -3 -2 -1 0 1 2 3 4 5
Fuse Relative Displacement (mm)
Figure 4.8: External Friction Fuse Subcomponent Calibration Tests

Table 4.2: External Friction Fuse Subcomponent Testing Summary

Y-M-D mm/s mm
Disp. Disp
Test # Date Cycles Test Comments
Rate Ampl.
F-1 09-11-10 0.01 +/-4.5 1 Initial Test, evaluate slip load
F-2 09-11-10 0.01 +/-2.0 1 Increase torque to target 1000kN slip
F-3 09-11-11 0.01 +/-4.5 1 Increase torque to target 1000kN slip
F-4 09-11-11 0.05 +20.1 1/2 Increase displacement rate. Observed gouge in yoke plate,
stopped test early to repair
F-5 09-11-11 0.05 -25.0 1/2 Other half of F-4. Compr. Side only
F-6 09-11-11 0.1 +/-25 1 Increase displ. rate. Another small gouge repaired
F-7 09-11-11 0.5 +/-62 1 Increase displ. rate/ampl. Compression displ. Limited to 55
b/c of head stroke limit
F-8P 09-11-11 0.5 +20 NA Recenter fuse in preparation for full cycle
F-8 09-11-11 0.5 +/-65 1 Full Cycle. Forks constrict movement in compr.
F-9 09-11-11 1.0 +68 1/2 Increase displ. rate. Bearing in tension
F-10 09-11-11 1.0 +/-63 1 Same as F-9 but full cycle.
F-11 09-11-11 0.25 +/-1.0 1 Test of automatic loading protocol
F-12 09-11-11 2.0 +/-55 1 Increase displ. rate. Trip load limit
F-13 09-11-11 0.25 +43 1/2 Recenter fuse
F-14 09-11-12 1.0 +/-49 1 Similar to F-10 but cold start (cooled from prev day)
F-15 09-11-12 1.0 +/-49 1 Reduce torque to target 1000kN slip (dynamic)
F-16 09-11-12 2.8 +/-44 9 Increasing displ. cyclic test. Constant maximum velocity in
each cycle
F-17 09-11-12 5.6 +/-44 9 Increase maximum velocity
4.2.2 VELOCITY-DEPENDANCE TESTS
The next set of tests, F-4 to F-12 were performed to investigate the velocity dependence of the friction
fuse. The displacement speed in consecutive tests was increased from 0.01mm/s to 2mm/s. In addition, the
maximum displacement for each test was increased from 5mm up to 63mm, which is close to the displace-
ment capacity of the fuse. The effect of the increased velocity may be seen in Figure 4.9. This figure shows
that as the displacement velocity increased, the slip force of the fuse also increased from the original target
load of 1000kN to a new stable slip load of 1200kN; however, this increase in slip force was not necessarily
caused by the increased velocity. The displacement magnitude also increased with each of these tests, which

Table 4.3: External Friction Fuse Subcomponent Test Result Summary

kNm kN mm/s mm kN kN kN °C °C
Est.
Mean Sustained Sustained Mean Peak
Total Disp. Disp Temp. Temp.
Test # Bolt Tension Compr. Friction Compr.
Normal Rate Ampl. Before After
Torque Slip Load Slip Load Coeff. Load
Force
F-1 2.03 1200 0.01 +/-4.5 870 825 0.353 - - -
F-2 2.20 1650 0.01 +/-2.0 940 850 0.271 - - -
F-3 2.49 1835 0.01 +/-4.5 1020 990 0.274 - - -
F-4 2.49 1835 0.05 +20.1 1020 920 0.274 - - -
F-5 2.49 1835 0.05 -25.0 1080 980 0.281 - - -
F-6 2.49 1835 0.1 +/-25 1160 1060 0.302 - - -
F-7 2.49 1835 0.5 +/-62 1280 1160 0.332 1380 - -
F-8P 2.49 1835 0.5 +20 1250 - 0.341 - - -
F-8 2.49 1835 0.5 +/-65 1210 1170 0.324 1550 - -
F-9 2.49 1835 1.0 +68 1200 1140 0.319 - - -
F-10 2.49 1835 1.0 +/-63 1180 1120 0.313 1550 - -
F-11 2.49 1835 0.25 +/-1.0 - - - - - -
F-12 2.49 1835 2.0 +/-55 1170 1110 0.311 1550 - -
F-13 2.49 1835 0.25 +43 - - - - - 43
F-14 2.2 1610 1.0 +/-49 1170 1100 0.309 1400 22 28
F-15 2.2 1610 1.0 +/-49 1050 980 0.315 1350 26 31
F-16 2.2 1610 2.8 +/-44 1060 980 0.317 1350 31 47
F-17 2.2 1610 5.6 +/-44 1080 980 0.320 1400 47 79
F-18 2.2 1610 8.4 +/-44 1080 990 0.321 1400 79 79
F-19 2.2 1610 11.2 +/-44 1080 980 0.320 1400 79 86
F-20 2.2 1610 14.0 +/-44 1060 990 0.318 1400 86 104
could have been the cause of the slip force increase. As will be discussed in later sections, tests of the full-
scale external friction fuse with the HC-SCED brace seem to show that the friction interface must be con-
ditioned by imposing a few cycles of loading before it can reach its stable slip load.
In addition to the increase in slip load, the hysteretic response of these tests was unbalanced, with the
axial load steadily increasing when the inner tongue plate moves into the outer yoke plate but staying con-
stant when the tongue plate moves back out. This behaviour was caused by the yoke plates, which con-
strained the tongue plate since the fabrication of the fuse resulted in the outer yoke plates being closer
together near the bottom than they were at the top (with the top and bottom oriented as per the test setup

shown in Figure 4.7). This problem was not observed in the full scale tests that because the geometry was
better controlled; however, another possible way to mitigate this problem in a real structure would be to
construct a piece of steel that is identical to the tongue plate with the same sized stainless steel plates on
each side and insert that assembly near the brace’s gusset plate to enforce the geometry of the yoke. Lastly,
the faster tests showed a more pronounced peak in the hysteretic response when the load direction
reversed. These small spikes were caused by the difference between the static and dynamic friction coeffi-
cients for the interface.
Increasing Slip Load

2000 due to Constrained 0.01mm/s
Yoke Geometry (Test F-3)
1500 0.05mm/s
(Tests F-4, F-5) 1200kN
1000
0.50mm/s 1000kN
0.10mm/s (Tests F-8) (Target)
500 (Tests F-6)
Axial 1.0mm/s Tension
(Tests F-10)
Force 0
(kN) Compression
-500 2.0mm/s
(Tests F-12) 1000kN
(Target)
-1000
1200kN
-1500 Static Friction Peaks
-2000 In Out
-80 -60 -40 -20 0 20 40 60 80

Figure 4.9: Effect of Velocity in Friction Fuse Subcomponent Tests
4.2.3 TEMPERATURE DEPENDANCE TESTS
Following the previous set of tests, the temperature of the friction interface was measured using a digital
temperature probe. All temperature readings for the tests are shown in Table 4.3. The next morning, after
the interface had returned to room temperature, the interface was retested at the same speed as the previous
test to see if there was any effect of the increased temperature on the response of the fuse. A comparison of
the hysteretic response of the warm and cold tests are shown in Figure 4.10. Although the two tests (Tests
F-13 and F-14) had different maximum displacements, the figure shows that the temperature difference of
approximately 20°C had no effect on the hysteretic response of the test fuse.

2000 1.0mm/s, 43˚C

(Test F-13)
1500
1200kN
1000 1000kN
1.0mm/s, 22˚C
(Target)
500 (Test F-14)
Axial
Force 0
1.0mm/s, 26˚C
(kN) Reduced Torque
-500 1000kN
(Test F-15)
(Target)
-1000
1200kN
-1500
-2000
-80 -60 -40 -20 0 20 40 60 80
Figure 4.10: Effect of Temperature in Friction Fuse Subcomponent Tests
and Recalibration of Fuse Slip Force
4.2.4 HIGHER VELOCITY DYNAMIC TESTS
To accommodate the increase in friction slip force that was caused either by increased velocity or sur-
face conditioning, the friction fuse normal force bolts were re-torqued to reduce the normal force on the
interface and bring the slip force back to the target value of 1000kN. The recalibrated hysteretic response
of the fuse (Test F-15) is shown in Figure 4.10.
To further investigate the effects of velocity and temperature on the behaviour of the test fuse, further
dynamic tests were conducted using progressively higher deformation velocities (Tests F-16 to F-20). Dur-
ing these tests, the temperature also increased progressively as a result of the friction. These final tests are
shown in Figure 4.11. The displacement history for each of these tests was an increasing sine function with
increasing amplitude over nine cycles as shown in the figure for Test F-16. The time axis of the displace-
ment history was linearly scaled for each of the tests by dividing the original time length by a factor of 2, 3,
4, and 5, respectively, thereby increasing the maximum velocity for each test up to the limit that the load-
ing frame could sustain. These tests show very little variation in the hysteretic response as the displacement
velocity and friction interface temperature both increased. The spreading effect on the left side each hyster-
esis due to the constriction of the inner tongue plate caused by the outer yolk plate remained unchanged. A
slight velocity effect may be seen in these hystereses whereby the high velocity portion of the hysteresis near
zero displacement shows a slight dip in force when compared to the low velocity portions of the hystereses
near the maximum displacement. This effect is the most noticeable in Test F-20.
Based on these higher velocity external friction fuse subcomponent tests, it was anticipated that the full-
scale external friction fuse in series with the HC-SCED brace was likely to perform well; however, at even

Displacement History
Test F−16
50
Test F−16: max 2.8mm/s, 47˚C

Fuse
Displacement 0 1000
(mm)
Axial
Force 0
-50 (kN)
0 200 400
Time (s) -1000
Test F−17: max 5.6mm/s, 68˚C Test F−18: max 8.4mm/s, 79˚C
1000
Axial
Force 0
(kN)
-1000
Test F−19: max 11.2mm/s, 86˚C Test F−20: max 14.0mm/s, 104˚C
High Velocity
1000
Low
Axial Velocity
Force 0
(kN)
-1000
-50 0 50 -50 0 50
Fuse Displacement (mm) Fuse Displacement (mm)
Figure 4.11: Friction Fuse Subcomponent Sine Displacement History Tests
higher velocities encountered during the full-scale brace tests, the full-sized external friction fuse exhibited
significant stick-slip behaviour. This behaviour limits the applicability of this experimental fuse and will be
discussed in detail in a future section.
4.2.5 TEST ISSUES
During the Test F-4, it was observed that there was a gouge forming in the inner friction surface of the
outer yoke. A photograph showing this damage is shown in Figure 4.12. It is believed that this gouge in the
steel was formed due to an imperfection on the inner friction surface of the outer yoke, which caught and
aggravated by the movement of the inner tongue plate. A grinder was used to remove the protrusion that

was formed by this gouge before the tests continued. This type of problem was not encountered during the
full-scale external fuse tests for the HC-SCED brace.
Figure 4.12: Gouge in Friction Surface of Outer Yoke Plate
4.3 HC-SCED TENDON PRETENSIONING

The strain capacity of the tendon assemblies was a constant issue during the testing of the HC-SCED
brace. The issues surrounding the use of these tendons and the problems that were encountered will be dis-
cussed in more detail in a future section; however, as a result of these issues, two full sets of twelve tendons
were used for the HC-SCED tests. Both sets of tendons used the same tendon size and identical anchors;
however, slightly different target SCED design parameters were used for the second set of tests due to the
selection of a more conservative tendon tensioning profile for those tests as will be discussed below.
4.3.1 ASSUMED TENDON PROPERTIES
The tendons that were used for the HC-SCED were 22mm nominal diameter Technora aramid ten-
dons supplied by Linear Composites Ltd. The tendon anchors were fabricated by DRB which also assem-
bled the tendons with the anchors prior to shipping. Based on previous test data shown previously in
Section 3.1.5, the tendon properties that were assumed for the design of the HC-SCED test specimen are
shown in Table 4.4. The assumed tendon strain capacity of 1.8% in the table was selected to be a conserva-
tive value based on earlier tests.

Table 4.4: Assumed Tendon Properties - 22mm Tendons

Property Value
Nominal Diameter D 22 mm
Cross-Sectional Area A p 153 mm2
Initial Young’s Modulus E p0 44 GPa
Effective Cyclic Young’s Modulus E p 102 GPa
Initial Tendon Length (not including anchor) L p0 5880 mm
Service Tendon Length L p 6040 mm
Nominal Ultimate Tendon Stress  pu 1960 MPa
Nominal Ultimate Tendon Strength P piu 300 kN
Assumed Tendon Strain Capacity  pu 1.8%
4.3.2 PRETENSIONING METHOD
The tensioning of the first set of tendons for the HC-SCED brace was performed at LCL Bridge. The
complete tendon tensioning profile for the first set of tendons is shown in Figure 4.13. As the figure shows,
if the tensioning process adheres to a 1.8% maximum strain capacity that was used in the design of the test
specimen, and if the service pretension in the tendons is 79kN after the tendons are pre-stretched to
285kN, then the maximum deformation range of the tendon will be +/- 75mm which is greater than the
length of the internal friction damper interface slot. According to these calculations, it was decided that the
tendons would be tensioned to 95% their nominal breaking load of 285kN before the force was released
back to the service pretension of 85kN, which was the target tendon pretension for the brace design.
The pretensioning of the tendons was conducted in the criss-cross order shown in Figure 4.14 for both
stages of pretensioning. For the first set of tendons, the tendons were pretensioned in two phases as
described by the following procedure:
1. For each tendon in order, install the tendon in the brace and tension it up to
285kN and then release to 85kN. Tighten the anchor nut. Measure the differ-
ence in tendon elongation between the maximum force and the lock-off force.
This difference must be at least 70mm.

Tendon Forces calculated Seismic

Stress based on: Deformation Range Maximum Load
σp (MPa) Ap = 153mm2 εp = 1.32% εpa = 4.23%
Approximate Elongations ΔL p
≈ 78 mm σp = 1862 MPa
calculated based on ΔLpa ≈ 249 mm
Initial Length: Pp ≈ 285kN
Lp0 = 5880mm
Final Length Assumed: Ep ≈ 102 GPa
Lp = 6040mm
εp = 0.50%
ΔLp ≈ 29 mm Target Lock-Off Load
εpa = 2.91%
σp ≈ 516 MPa
Ep0 ≈ 44GPa
ΔLpa ≈ 171 mm
Pp = 79 kN
εp0
Absolute Tendon Strain εpa
εpu
1.82%
Figure 4.13: Tendon Tensioning Plot - First Set of Tendons
2. Go over each tendon again to ensure that all tendons have a lock-off force of
85kN because the tensioning of each tendon in stage 1 may have affected the
load in the other tendons.
The tendons could not be pretensioned in more stages than this. They have to be tensioned to the maxi-
mum axial force in the first step because otherwise the tendon anchor would not have been able to reach
the outside of the endplate to be locked into place with the anchor nut.
9 5 7 11
1 3
4 2
12 8 6 10
Figure 4.14: Tendon Tensioning Order

The tendon tensioning setup is shown in Figure 4.15. The tendons were pretensioned using a hydraulic
jack bearing against the end plate of the HC-SCED brace. The jack operated on the tendons via a tension-
ing rod that was threaded into the open end of the tendon barrel (see threads in Figure 4.33). A two-stage
tensioning chair was required to pretension the tendons because the total elongation of the tendon during
the process was longer than the stroke capacity of the jack. A temporary locking nut was used to temporar-
ily hold the tendon position at the halfway point to allow the stroke of the jack to be reset. The load in the
tendon during pretensioning was monitored using a 400kN load cell in series with the jack.
Figure 4.15: Tendon Pretensioning Setup
The tensioning of the second set of 12 tendons in the HC-SCED brace was performed using a slightly
modified method in the laboratory at Ecole Polytechnique. In an attempt to mitigate the tendon issues
that had been encountered during tests using the first set of tendons, the tensioning profile of the tendons
was modified to increase the margin of safety. Instead of going to the maximum load of 95% of the nomi-
nal breaking load of 300kN, to a pretension value of 285kN, the maximum load was limited to 85% of the
nominal breaking load, which corresponds to 255kN. In addition, the lock-off load for the brace was also
reduced from 85kN to 75kN. The complete revised tendon tensioning profile for the second set of ten-
dons is shown in Figure 4.16. This revised tendon tensioning profile necessitated some minor modifica-
tions to the target SCED response parameters as shown in Table 4.5. New values in the table are only

shown for parameters that had changed. Since there was a lower target SCED pretension load for the sec-
ond set of tendons, the activation load and ultimate load of the brace were reduced by reducing the force in
the internal friction damper normal force bolts. These changes increased the buffer between the maximum
load expected in the tendons and the load at which the tendon anchors could fail.
Tendon Forces calculated Seismic

σp (MPa) Ap = 153mm2 εp = 1.18% εpa = 3.79%
Approximate Elongations ΔLp ≈ 69 mm σp = 1667 MPa
Initial Length: Pp ≈ 255kN
Lp0 = 5880mm
Lp = 6040mm
εp = 0.49%
εpa = 2.61%
σp ≈ 490 MPa
Ep0 ≈ 44GPa
ΔLpa ≈ 153 mm
Pp = 75 kN
εp0
εpu
1.67%
Figure 4.16: Tendon Tensioning Plot - Second Set of Tendons
4.4 HC-SCED TESTING

The complete set of tests that were performed on the HC-SCED brace using both the first and second
sets of tendons are summarized in Tables 4.6 to 4.9. Tables 4.6 and 4.8 show the test displacement rates,
displacement amplitudes, number of cycles as well as comments that briefly describe the purpose of each
test and significant observations. Tables 4.7 and 4.9 summarize the resulting behaviour of the HC-SCED
brace including brace activation force, external friction fuse slip force, brace stiffnesses, and the energy dis-
sipation capacity parameter (  ). Note that although the initial stiffness leg of the SCED brace response is
actually multilinear as previously described in Section 3.5, the initial stiffness given in these tables is the
highest measured stiffness, not the effective initial stiffness. Effective initial stiffness values will be given for
selected tests in later sections.

Table 4.5: HC-SCED Revised Target Design Parameters

Parameter Old Value New Value
SCED Ultimate Axial Load P u 4000 kN 3800kN
Tendon Pretension Force P p0 1018 kN 900kN
Total Internal Friction Damper Slip Force F it 925 kN 850kN
SCED Activation Load P a 1950 kN 1750kN
Peak Tendon Strain  pu 1.80% 1.67%
Energy Dissipation Capacity Parameter  0.95 0.97
Estimated Post-Activation Stiffness k a 30 kN/mm
The tests that were performed using the first set of tendons were predominantly used to work out some
of the issues relating to the behaviour of the tendons and external fuse, and will be discussed towards the
end of the chapter. The results that represented the major contributions from this portion of the thesis
work were obtained from the tests that used the second set of tendons. The results from these tests will be
discussed in more detail in the following sections.
4.4.1 LOAD FRAME SETUP AND INSTRUMENTATION
The test setup and instrumentation summary is shown in Figure 4.17. The HC-SCED brace was tested
in the 12MN load frame at Ecole Polytechnique in Montreal. The load frame is equipped with a device to
measure the movement of the loading head and a load cell to measure the total force being applied. Two
string potentiometers were used to provide an additional measurement of the total deformation of the
brace. Two additional string potentiometers were mounted to the base of the frame extending to the inner
tongue plate of the external friction fuse to measure the total deformation of the fuse. Digital encoders
were mounted on the outer member with probes extending to the internal friction damper friction inter-
face plate to measure the relative movement of the inner and outer members. Eight strain gauges were used
to measure the strains in each member (inner and outer). The locations of these strain gauges are shown in
Figure 4.17. A photograph of the test setup is shown in Figure 4.18.

Table 4.6: HC-SCED Brace Load Frame Testing Summary - First Set of Tendons
Y-M-D mm/s mm
Disp. Disp
Rate Ampl.
ATTEMPT TENDON-ONLY CHARACTERIZATION (BUT EXT. FUSE SLIPS)
CT-1 10-04-13 0.5 +55 1/2 Tendon Only - Fuse Slipped Prematurely (T)
CT-2 10-04-13 0.5 -70 1/4 Tendon Only - Large Ext. Fuse Slip (C)
CT-3 10-04-13 0.5 -55 1/2 Increased bolt torque, Ext. Fuse still slips
CT-4 10-04-14 0.5 +70 1/4 Smooth Ext. Fuse slip.
EXTERNAL FRICTION FUSE SLIP TESTS
FE-1 10-04-23 0.5 +/-20 6 Slip force increases over all the cycles until surface is condi-
tioned and value remains constant
FE-2 10-04-23 0.5 +/-20 2 Increase Torque, higher slip
FE-3 10-04-23 0.5 +/-20 3 Increase Torque, higher slip
FE-4 10-04-23 0.5 +/-20 2 Different bolts used to condition the surface around them
FE-5a 10-04-23 0.5 +/-20 3 Different bolts used to condition the surface around them
FE-5b 10-04-23 1.0 +/-35 2 Different bolts used to condition the surface around them
CONTINUE TENDON-ONLY CHARACTERIZATION (TENDON ANCHORS FAIL)
CT-5 10-04-26 0.5 +69 1/2 External Fuse holds, but two tendon anchors fail
CT-6 10-04-28 0.5 +9 1/2 Step in response indicating tendon is stuck in between inner
member and end plate
TEST INTERNAL FRICTION DAMPER CALIBRATION
FI-1 10-04-29 0.5 +12 1/2 Internal Damper Calibration 1/4 total force
FI-2 10-04-29 0.5 +12 1/2 Internal Damper Calibration 1/2 total force
FI-3 10-04-29 0.1 -3.5 1/4 Attempt to close tube gap
FI-4 10-04-29 0.5 +12 1/2 Internal Calibration full total force
FI-5 10-04-29 0.1 -5.5 1/2 Attempt to close tube gap
REMOVE INTERNAL DAMPER FORCE AND TEST TENDONS TO DESIGN ELONGATION
CT-7 10-04-29 0.5 +69 1/4 Another tendon anchor fails. End Test Set.
4.4.2 SCED BRACE AXIAL FORCE CALCULATION
In general, for the HC-SCED tests, the axial force was simply measured using the load cell that is built
into the 12MN load frame; however, the force in the brace was also calculated using a combination of
strain gauge data and digital encoder data from the brace. This additional calculation was performed to

Table 4.7: HC-SCED Brace Load Frame Test Result Summary - First Set of Tendons
kN kN kN kN kN/mm kN/mm
Est. Est. Estimated
External Estimated
External Internal Activation Post-
Fuse Slip Initial Estimated
Test # Fuse Damper Force Activation
Normal Normal Pa
Force Stiffness
Stiffness  Parameter
F Et k1
Force Force ka
- + - + - + - +
ATTEMPT TENDON-ONLY CHARACTERIZATION (BUT EXT. FUSE SLIPS)
CT-1 Lock 1130 2560 751 30 0.07
CT-2 Lock -1020 2710 748 29
CT-3 Lock -1020 2520 746 29 0.04
CT-4 110 854
EXTERNAL FRICTION FUSE SLIP TESTS
FE-1 711 765 730 700
FE-2 800 810 760 690
FE-3 889 970 780 660
FE-4 889 1035 740 690
FE-5a 889 825 790 700
FE-5b 889 855 760 640
FE-6 952 795 770 660
FE-7 921 805 720 720
CONTINUE TENDON-ONLY CHARACTERIZATION (TENDON ANCHORS FAIL)
CT-5 4762 1150 830 30
CT-6 4762 997 26
TEST INTERNAL FRICTION DAMPER CALIBRATION
FI-1 4762 1553 1230 28 0.45
FI-2 4762 3142 1486 30 0.76
FI-3 4762 3142
FI-4 4762 6243 1832 33 1.07
FI-5 4762 6243
REMOVE INTERNAL DAMPER FORCE AND TEST TENDONS TO DESIGN ELONGATION
CT-7 4762 971 26 0.07
confirm that it could be used to determine the axial force SCED brace tests where a load cell in series with
the brace was not present.

Table 4.8: HC-SCED Brace Load Frame Testing Summary - Second Set of Tendons
Y-M-D mm/s mm
Disp. Disp
Rate Ampl.
INTERNAL FRICTION DAMPER CALIBRATION (TENDON ANCHOR FAILS)
SC1-A 10-07-20 0.5 +/-35 1 Tendon only test.
SC2-A 10-07-20 0.5 +/-15 1 Internal Calibration partial normal force
SC3-A 10-07-20 0.5 +/-15 1 Internal Calibration full normal force
SC4-A 10-07-21 0.5 +/-63 10 On second max deformation cycle, single tendon anchor
fails. Good response otherwise.
CONTINUE WITHOUT REMOVED TENDON - CHARACTERIZATION
SC5-A 10-07-21 0.5 +/-32 1 Single cycle characterization. Brace still functions well.
SIMULATED WIND LOADING
SC6-A 10-07-21 sine +/-1.1 500 Period = 1.25s Amplitude = 0.5 a
SC6-B 10-07-21 sine +/-1.1 500 Period = 0.5s Amplitude = 0.5 a
SC6-C 10-07-21 sine +/-2.1 1000 Period = 1.25s Amplitude = 0.95 a
SC6-D 10-07-21 sine +/-2.1 1000 Period = 0.5s Amplitude = 0.95 a
SC6-E 10-07-21 sine +/-1.1 500 Period = 1.25s Amplitude = 0.5 a
SC6-F 10-07-21 sine +/-1.1 500 Period = 0.5s Amplitude = 0.5 a
RE-CHARACTERIZATION CYCLES
SC7-A 10-07-21 0.5 +/-32 1 Single cycle characterization. No degradation due to wind.
SC8-A 10-07-22 0.5 +/-32 1 Repeated characterization after one day
DYNAMIC EARTHQUAKE LOADINGS
SC9-A 10-07-22 EQ 40 EQ 2-1 LA09 DBE (Tension Max) Velocity Limited to
70mm/s
SC10-A 10-07-22 EQ 44 EQ 2-2 LA18 DBE Lim. (Tension Max) Velocity Limited to
70mm/s, Deformation Limited to 1.25% Drift
SC11-A 10-07-22 EQ 40 EQ 2-1 LA09 DBE (Compression Max) Velocity Limited to
70mm/s
EXTERNAL FRICTION FUSE CALIBRATION
SC12-A 10-07-22 0.2 +/-5 1 Target half ext. fuse slip. Good Response
SC12-B 10-07-22 0.2 +3.5 1/4 Re-Target ext. fuse slip. Big Slip
SC12-C 10-07-22 0.2 -2.8 1/4 Same as previous but in compression. Big Slip
SC12-D 10-07-22 sine +/-5 5 0.2Hz Sine, much better response
SC13-A 10-08-03 sine +/-50 1 Increase Torque 0.15Hz Sine, very choppy fuse response
SC13-B 10-08-04 sine -50 1/2 Increase Torque 0.15Hz Sine, still choppy

Table 4.8: HC-SCED Brace Load Frame Testing Summary - Second Set of Tendons
Y-M-D mm/s mm
Disp. Disp
Rate Ampl.
SC13-C 10-08-04 1.0 +50 1/2 Increase Torque. No continuous Slip, Premature slip at
2450kN
SC13-D 10-08-04 1.0 -51 1/2 Premature slip at 2750kN
SC13-E 10-08-04 5.0 +/-59 1 Premature slip at 2600kN
SC13-F 10-08-04 5.0 -60 1/2 Reduced torque. Did not push far enough
SC13-G 10-08-04 5.0 -68 1/2 Same as previous, push further. Good slip.
SC13-H 10-08-04 5.0 +45 1/2 Same as previous but in tension. Good slip.
SC13-I 10-08-04 10.0 -58 1/2 Faster, but hit tendon elongation limit.
SC13-J 10-08-04 5.0 -58 1/2 Slower, but hit tendon limit again.
SC13-K 10-08-04 5.0 +/-65 1 Release ext. fuse bolt 6. Full Cycle. Good slip.
SC13-L 10-08-04 10.0 +/-80 1 Limit Tripped in Compression. Good Slip
SC13-M 10-08-04 10.0 -56 1/2 Lowered ext. fuse bolt torque. Good Slip
SC13-N 10-08-04 20.0 +/-70 1 Increased Speed. Very Choppy Slip
SC13-O 10-08-04 35.0 +/-70 1 Increased Speed. Still Very Choppy Slip
SC13-P 10-08-04 50.0 +/-70 1 Increased Speed. Choppy. Tripped Force Limit
SC13-Q 10-08-04 15.0 -65 1/2 Recenter at slower speed. Still Choppy
SC13-R 10-08-04 10.0 +/-70 1 Reduce Speed, slip is smooth again.
DYNAMIC EARTHQUAKE LOADINGS AND RE-CENTER
SC14-A 10-08-05 EQ 109 EQ 3-1 LA32 MCE Lim. (Compression Max) Velocity Lim-
ited to 10mm/s
SC14-B 10-08-05 10.0 +50 1/2 Recenter Ext. Fuse
SC14-C 10-08-05 7.0 -60 1/2 Recenter Ext. Fuse Sine excitation used by mistake
SC14-D 10-08-05 10.0 -64 1/2 Re-check torque to previous value. Recenter Ext. Fuse.
Choppy Slip
SC14-E 10-08-05 5.0 +/-56 1 Slower Full Cycle. Great slip response.
DYNAMIC PROTOCOL (TENDON ANCHOR FAILS)
SC15-A 10-08-05 5.0 +/-87 19 Dynamic Protocol with varying amplitude. Tendon Limit
Tripped
SC15-B 10-08-05 5.0 +/-105 3 Rest of Dynamic Protocol. Tendon anchor failure
TENDON TESTS
SC16-A 10-08-05 5.0 +/-15 1 Characterization. Step Response due to jammed tendon
SC17-A 10-08-05 1.0 -69 1/2 Try to fail additional tendon anchors. Step Response
SC17-B 10-08-05 1.0 +71 1/2 Try to fail additional tendon anchors. Some ext. fuse slip.
SC17-C 10-08-05 1.0 +70 1/2 Try to fail additional tendon anchors.

Table 4.9: HC-SCED Brace Load Frame Test Result Summary - Second Set of Tendons
Est. Est. Estimated
External Estimated
Normal Normal Pa
Force Stiffness
F Et k1
Force Force ka
- + - + - + - +
INTERNAL FRICTION DAMPER CALIBRATION (TENDON ANCHOR FAILS)
SC1-A Lock -1000 1120 738 790 30 32 0.13 0.13
SC2-A Lock 2937 -1618 1640 730 807 30 34 0.78 0.78
SC3-A Lock 3994 -1760 1780 730 764 30 34 0.88 0.84
SC4-A Lock 3994 -1760 1780 800 780 29 33 0.91 0.88
CONTINUE WITHOUT REMOVED TENDON - CHARACTERIZATION
SC5-A Lock 3994 -1670 1655 746 774 29 33 0.99 0.96
SC6-A Lock 3994 826 826
SC6-B Lock 3994 837 837
SC6-C Lock 3994 813 813
SC6-D Lock 3994 830 830
SC6-E Lock 3994 836 836
SC6-F Lock 3994 836 836
RE-CHARACTERIZATION CYCLES
SC7-A Lock 3994 -1714 1791 760 783 27 26 0.99 0.93
SC8-A Lock 3994 -1734 1724 720 757 28 32 1.00 0.99
SC9-A Lock 3994 -1677 1706 778 778 29 31 0.98 0.96
SC10-A Lock 3994 -1643 1641 795 795 27 31 0.95 0.98
SC11-A Lock 3994 -1641 1669 840 840 29 31 0.99 0.93
EXTERNAL FRICTION FUSE CALIBRATION
SC12-A 1277 3994 1100 784 784
SC12-B 1577 3994 1490 725 725
SC12-C 1577 3994 1540 770 770
SC12-D 1577 3994 1600 680 680
SC13-A 2880 3994 -1702 1722 2500 739 680 29 31 0.97 0.90
SC13-B 3245 3994 -1662 1722 2850 844 704 30 31 1.04 0.90
SC13-C 3570 3994 1722 830 31 1.00
SC13-D 3570 3994 -1725 711 29 1.12

Table 4.9: HC-SCED Brace Load Frame Test Result Summary - Second Set of Tendons
Est. Est. Estimated
External Estimated
Normal Normal Pa
Force Stiffness
F Et k1
Force Force ka
- + - + - + - +
SC13-E 3570 3994 -1720 1710 3180 735 700 29 29 1.20 1.12
SC13-F 3407 3994 -1670 1673 732 670 29 30 1.07 0.94
SC13-G 3407 3994 -1690 1730 3155 723 644 29 30 1.18 0.98
SC13-H 3407 3994 1750 3100 750 31 1.22
SC13-I 3407 3994 -1754 740 28 1.09
SC13-J 3407 3994 -1692 740 29 1.11
SC13-K 3245 3994 -1700 1700 3100 719 690 29 29 1.19
SC13-L 3245 3994 -1670 1693 3070 760 680 28 30 1.12
SC13-M 2943 3994 -1634 3190 780 29
SC13-N 2943 3994 -1672 1650 3100 700 640 28 29
SC13-O 2943 3994 -1625 1635 3050 680 800 28 30
SC13-P 2943 3994 -1610 1616 2980 630 740 28 29 1.05
SC13-Q 2943 3994 -1554 3180 810 29
SC13-R 2943 3994 -1600 1650 3040 700 800 29 31 1.08
DYNAMIC EARTHQUAKE LOADINGS AND RE-CENTER
SC14-A 2943 3994 -1660 1670 3200 690 770 28 31 1.02 0.95
SC14-B 2943 3994 1607 3140 680 30
SC14-C 2943 3994 -1607 3320 700 28
SC14-D 2943 3994 -1580 1620 3010 790 645 29 31 0.93
SC14-E 2943 3994 -1600 1633 2900 730 745 28 30
DYNAMIC PROTOCOL (TENDON ANCHOR FAILS)
SC15-A 2943 3994 -1600 1600 3000 680 720 28 29 0.98 1.01
SC15-B 2943 3994 -1490 1490 3100 640 630 29 29
TENDON TESTS
SC16-A 2943 3994 -1540 1540 26 29 0.87 0.87
SC17-A 2943 3994 -1513 1500 25 1.06
SC17-B 2943 3994 -1480 1530 2880 814 23 27 0.95
SC17-C 2943 1997 1058 25 0.50

CL
Upper Connection
Load Frame Length 400
Upper Head
2250
175
(typ.)
Linear Strain Gauges

One on front and back of
inner and outer members Brace Total
1700 (16 total) Member Test
Length Length
Internal Friction 6200 7670
Damper Deformation
Digital Encoders
(2 total)
2250 Total Brace Elongation
String Potentiometers
(2 total)
External Friction
Load Frame Fuse Deformation
Base String Potentiometers
(2 total)
Lower Connection
Length 1070
Figure 4.17: HC-SCED Test Setup and Instrumentation
The total force in a SCED brace is equal to the sum of the forces in the three main axial elements: the
inner member, the outer member and the tendon. The inner and outer member forces were calculated
using experimental data from the strain gauges on the members combined with the member cross-sectional
area as follows:
P ir = A i E s  i (Eq. 4.1)
P or = A o E s  o (Eq. 4.2)
where P ir and P or are the axial forces in the inner and outer members relative to the initial axial force
caused by the pretensioned tendons, A i and A o are the cross-sectional areas of the inner and outer mem-
bers, E s is the Young’s modulus of steel, and  i and  o are the average axial strain in the inner and outer
members measured using the strain gauges. The tendon force was calculated by using the experimental
data from the internal friction damper deformation digital encoders for each brace, which measure the rel-
ative movement of the inner and outer members. In other tests, this data may come from string potentiom-
eters or linear potentiometers instead of encoders:

Figure 4.18: HC-SCED Test Setup Photo
 rel E p A pt
P pr = ------------------------- (Eq. 4.3)
Lp

where P pr is the axial force in the tendons relative to the initial axial force caused by the pretension,  rel
is the absolute value of the relative movement of the two members which is effectively equal to the tendon
elongation  p , E p is the effective Young’s modulus of the aramid tendon, A pt is the nominal cross-sec-
tional area of all of the brace tendon fibres, and L p is the effective length of the tendons. The total axial
force applied to the brace, P a , was calculated, as previously discussed, by summing the axial forces of the
two members and the tendon:
P a = P ir + P or + P pr (Eq. 4.4)
Two examples of this calculation compared to the force recorded from the load cell will be presented in
future sections.
4.4.3 INTERNAL FRICTION DAMPER CALIBRATION
Prior to any protocol or earthquake tests, the HC-SCED was calibrated in place in the load frame. The
first step in this calibration was to test the brace without any of the internal friction damper normal force
bolts tightened. The maximum deformation for this test was limited to +/- 35mm since this was deemed
sufficient to calibrate the response. After this, the internal friction dampers were calibrated in two steps to
achieve the target activation load. The resulting hystereses from these calibration tests are shown in Figure
4.19. The hysteretic response for the tendon-only test (Test SC1-A) exhibited a small amount of energy
dissipation due to friction between the moving parts of the HC-SCED brace that were in contact.
3000
HC-SCED Internal Friction
Damper Calibration
2000
1000
Axial
Force 0
(kN)
Tendon Only
-1000 Test SC1-A
Partial Internal Friction
-2000 Test SC2-A
Full Internal Friction
Test SC3-A
-3000
-40 -30 -20 -10 0 10 20 30 40
Total Axial Deformation (mm)
Figure 4.19: HC-SCED Internal Friction Damper Calibration

Following the calibration of the friction interfaces, a full quasi-static cyclic test was performed. This test
consisted of progressively increasing deformation cycles up to a maximum of +/-63.5mm at a constant
deformation velocity of 0.5mm/s. The hysteretic response of the HC-SCED brace for this quasi-static test
(Test SC4-A) is shown in Figure 4.20. All of the cycles of this test up to the maximum deformation level
showed a stable flag-shaped hysteretic response with full self-centering after every cycle; however, one of the
tendon anchors failed during the first half of the second cycle at the maximum deformation (63.5mm).
The tendon was able to be removed safely from the brace and the tests were able to continue. The activa-
tion load and energy dissipation of the brace was slightly changed due to the loss of one tendon; however,
the fundamental behaviour of the brace was not changed since the entire tendon was removed, leaving no
debris inside the brace. This anchor failure will be discussed in greater detail in a future section.
Quasi Static Protocol Test SC4-A (0.5mm/s)

4000
Tendon Anchor
Failure (second cycle)
3000
2000
1000
Axial
Force 0
(kN) Shimming of Bottom
-1000 End Plate to Remove
Tendon Debris
-2000
-3000
-4000
-60 -40 -20 0 20 40 60
Figure 4.20: Calibrated HC-SCED Characterization and Tendon Anchor Failure
4.4.4 WIND RESPONSE SIMULATION
The next stage of the testing was to simulate a wind loading history on the brace to determine the effect
of thousands of cycles of deformation below the brace activation load. Tests SC6-A to SC6-F were used to
simulate the effect of years of wind loading on the HC-SCED before it is subjected to an earthquake. The
wind loading was simulated using the displacement-controlled sinusoidal deformation histories described
in Table 4.8. The deformation was applied at two different amplitudes using two different displacement
periods. Two thousand cycles were applied at half the brace deformation at activation ( 0.5 a ) and two

thousand cycles were applied at 95% of the brace deformation at activation ( 0.95 a ). Half of these cycles
were applied at a period of 1.25s which is the elastic period of the twelve-storey building from Choi et al.
(2009) which the HC-SCED design was sized for. The other half of the cycles were applied at a shorter
period of 0.5s to consider the case where the HC-SCED brace is used in a shorter or stiffer building. The
hysteretic response of the brace resulting from all of these tests are shown in Figure 4.21. The width of the
hysteretic loops in this figure are exaggerated by the fact that the deformation axis is only +/-3mm. This
figure shows that there was no significant change in the hysteretic response of the brace due to the 4000
total wind simulation cycles.
2000
Wind Loading Simulation Test SC6
1500
1000
500
Axial
Force 0
(kN)
-500
-1000
-1500
-2000
-3 -2 -1 0 1 2 3
Figure 4.21: Wind Loading Simulation Results
After the wind simulation tests, some characterization tests were performed to see if the brace hysteresis
changed due to the wind cycles (Tests SC7-A and SC8-A). One of these characterization tests is compared
to the HC-SCED brace response prior to the wind simulation tests in Figure 4.22. The results of these
tests are almost identical. The activation force after the wind simulation tests increased by approximately
50kN or approximately 3%. This small increase may have occurred due to a slight increase in internal fric-
tion between moving parts of the brace.

3000
Effect of Wind Loading
2000
1000
Axial
Force 0
(kN)
-1000
Before Wind Cycles
(Test SC5-A)
-2000 After Wind Cycles
(Test SC8-A)
-3000
-30 -20 -10 0 10 20 30
Figure 4.22: Effect of Wind Loading Comparison
4.4.5 SIMULATION OF SCED BRACE RESPONSE AND ALTERNATE CALCULATION OF HC-SCED

AXIAL FORCE
Using the hysteretic properties for the HC-SCED brace from Test SC8-A, the brace response was mod-
elled using the Mechanics Simulator previously described in Section 3.4. The inputs for the Mechanics
Simulator model are shown in Table 4.10. This table also includes inputs for a model that will be described
in a future section. A difference in length between the inner and outer members of 1.5mm was included in
the model. This length difference was not measured, but it provided a good fit between the simulation and
the data. The resulting brace response comparison is shown in Figure 4.23. The top plot in the figure
shows a comparison between the full hysteretic response of the brace from the Mechanics Simulator, from
the test with the axial force from the load cell, and from the test with the axial force calculated using the
method described in Section 4.4.2. The figure shows that the model well-predicts the overall hysteretic
response of the brace. In addition, the calculation of the brace force using the strain gauges and the encod-
ers provides a good approximation of the total force in the brace for this test.
The Mechanics Simulation model also makes a good estimate of the effective initial stiffness k 1 eff of
the brace. The measured effective initial stiffness, calculated by finding the secant stiffness between the
positive and negative activation points, was 749 kN/mm. The estimated effective initial stiffness from the
model was 784 kN/mm. Note that these are both significantly smaller than the theoretical initial stiffness
estimate that would be obtained by adding the axial stiffnesses of the inner and outer members which

Table 4.10: Mechanics Simulator Model Inputs

Test SC8-A Test SC16-A
Parameter
Value Value
Young’s Modulus Steel E s 200000 MPa 200000 MPa
Young’s Modulus Tendon E p 102000 MPa 102000 MPa
Inner Member Length L i 6200.0 mm 6206.5 mm
Inner Member Area A i 30345 mm2 30345 mm2

Inner Member Initial Position 0 mm 0 mm
Outer Member Length L o 6201.5 mm 6201.5 mm
Outer Member Area A o 33930 mm2 33930 mm2

Outer Member Initial Position 0 mm 0 mm
Internal Damper Friction End 1 F I1 710 kN 525 kN
Tendon Area A pt 1836 mm2 1836 mm2
Total Tendon Pretension P p0 870 kN 870 kN
Connection Part 1 Length 400 mm 400 mm

Connection Part 1 Area 24000 mm 2
24000 mm2
Connection Part 2 Area 20000 mm2 20000 mm2
Bearing Stiffness 100000 kN/mm 100000 kN/mm
Friction Interface Stiffness 100000 kN/mm 100000 kN/mm
Initial Pretensioning Force Step 0.1 kN 0.1 kN
Analysis Deformation Step 0.01 mm 0.005 mm
would be 2073 kN/mm. This large initial stiffness difference existed even though great care was taken to
ensure that the inner and outer sections would be the same length.
The lower plots in the figure compare the hysteretic response of the individual brace elements to the
predicted response from the Mechanics Simulator model. The measured force in the inner and outer mem-
bers was calculated using the strain gauges on those elements and their cross-sectional areas. Since the ini-
tial division of the pretensioning load between the inner and outer members was not known, only the
relative values for these member forces are useful. Therefore, the measured test hysteresis in the lower plots

3000
Test SC8-A T
2500 Axial Force
Calculated
2000 using Brace
1500 Axial Force Member Force
from Load Frame and Tendon
1000 Elongation
Brace
Axial 500
Force 0
0
(kN) -500 Model Model
Hysteresis Hysteresis
-1000
-1500
-2000
-2500 C
-3000
-50 -25 0 25 +/-50 -25 0 25 50
Inner Outer
Member Member
1500
Test SC8-A
1000
T
500
0 T
Axial 0
Force -500 Shifted
Test Hysteresis Model
(kN) -1000 0
Hysteresis
-1500
-2000 C
-2500 C
-50 -25 0 25 +/-50 -25 0 25 50
0 -870 0
Tendon
2500 Init. - 0 870
Test SC8-A
2000 0 0 0
C T 710 146
1500 710 2470
Axial 1614
1000 Tens. - T
Force
0 710 146 146
(kN) 500
710 -2214 146 2360
0
1504
-500
Comp. - C
-50 -25 0 25 50 710 -1650 146
Figure 4.23: Simulation of Brace Response and Calculation of Brace Element Forces for Test SC8-A
are vertically shifted so that they align with the predicted hystereses. The results show that the outer mem-
ber hysteresis is wider than the inner member hysteresis. This is an indication that the internal friction
damper force at the two ends of the brace was unbalanced. It was only after testing that this imbalance was
discovered. The cause for this imbalance was that the internal friction interface angles at one end of the
brace were much larger and stiffer than those at the other end of the brace (see previous Figure 4.2). When

the bolts in the interface with the larger angles were tensioned, more of the bolt force was used up bending
the angle to allow the friction interfaces to be in full contact. This meant that less of that bolt normal force
was available to be applied to the interface. In the test, this problem was worse that it would be in a real
braced frame because the larger angles were solidly welded at their ends to two large stiff connection plates
which held them solidly in place. Regardless, this is an issue that should be considered in future brace
designs. The resulting unbalance in the friction interface forces is reflected in the input to the Mechanics
Simulator model. The measured force in the tendons was calculated using the elongation of the tendons as
measured by the digital encoders. The individual brace element behaviour is well-predicted by the
Mechanics model. SCED brace diagrams showing the forces in each member and in each friction interface
are shown in the lower right corner of the figure. These demonstrate the effect that the internal friction
interface force imbalance has on the forces in the inner and outer members.
4.4.6 DYNAMIC EARTHQUAKE RESPONSE - DESIGN-LEVEL EARTHQUAKES
Three simulated earthquake deformation responses were compiled to test the HC-SCED brace’s
response to fully dynamic loading. These records were created by using the Ruaumoko analysis results for
the first storey brace from Choi’s 12-storey SCED braced frame building 2D model (Choi et al., 2009). In
that study, the 12 storey building was subjected to a suite of earthquake records for the Los Angeles area
that originated from the SAC steel project (Somerville et al., 1997). Three of the brace deformation histo-
ries that resulted from these models were used to provide an approximate deformation demand for the real
HC-SCED brace. The deformation history was extracted in terms of storey drift and that drift was con-
verted into a deformation in millimeters based on the design frame geometry previously shown in Figure
4.1. The resulting deformation histories from two earthquake records were chosen from the suite of
design-basis (DBE) seismic excitations and one record was chosen from the suite of maximum-considered
(MCE) excitations that would result in external friction fuse slip. The first DBE deformation history is the
response of the first storey brace in the model to LA09 1992 Landers, and the second is the response to
LA18 1994 Northridge. The response of the real HC-SCED brace to these forced deformations was esti-
mated using an OpenSEES model. The brace deformation records and the resulting response estimates for
these records are shown in Figure 4.24.
Prior to testing the deformation records on the HC-SCED brace, all of the proposed records were
tested in the empty 12MN load frame to ensure that the feedback displacement equaled the command dis-
placement. These tests showed that due to pump limitations, the load frame had a velocity limit of approx-

Test SC9-A and SC11-A Input Deformation

SAC Record LA09 1992 Landers
80
4000 OpenSees Prediction
60 2% Drift
40 2000
Axial 20 Axial
Def. 0 Force 0
(mm) -20 (kN)
-40 Velocity-Limited -2000
-60 Original 2%
-4000 Drift
-80
0 20 40 60 80 100 120 -80 -40 0 40 80
Test SC10-A Input Deformation
SAC Record LA18 1994 Northridge
80 4000 OpenSees Prediction
60 Original 2% Drift
40 Velocity and Amplitude Limited 2000
Axial 20 Axial
Def. 0 Force 0
(mm) -20 (kN)
-40 -2000
2%
-60
-4000 Drift
-80
0 10 20 30 40 50 60 70 80 90 -80 -40 0 40 80
Time (s) Axial Deformation (mm)
Figure 4.24: Input Deformation Records for DBE Earthquake Tests
imately 70mm/s. The consequence of this was that for portions of the records where the command velocity
was greater than 70mm/s, the load frame could not achieve the full command displacement. In order to
achieve the target displacements during the actual tests, each displacement record was modified to limit the
maximum velocity to 70mm/s. This was done by using a MATLAB script to evaluate each timestep in the
record, check whether the absolute velocity for that timestep was greater than 70mm/s and then adding
additional timesteps as necessary, using linear interpolation to determine the corresponding deformation
values between the original timesteps. Using this method, all of the deformations, velocities and frequen-
cies remained unchanged for the parts of the records that had velocities less than 70mm/s. The parts of the
record that had velocities greater than 70mm/s had their timescales dilated just enough to meet the limita-
tions of the load frame, decreasing the frequencies, but leaving the deformations unchanged. This way, the
target brace deformations for the record were unaffected at the expense of moderately decreased velocities
and frequencies only where required. This resulted in the most realistic records possible within the con-
straints of the load frame. In addition to the velocity limitation, for the second earthquake (LA-18, Test
SC10-A) the maximum amplitude was also limited to avoid over-straining the tendons. The difference
between the original records and the limited records is shown in Figure 4.24.

The results of these dynamic earthquake tests are shown in Figure 4.25. The shapes of the resulting hys-
tereses are similar to the predicted hysteresis shown in Figure 4.24. In each of the plots in Figure 4.25, a
single hysteretic cycle from one of the static tests is shown for comparison (Test SC8-A). The post-activa-
tion stiffness portion of that static hysteresis has a thickness due to some minor stick-slip behaviour of the
internal friction dampers when loaded slowly. The earthquake test hystereses from Tests SC9-A to SC11-A
all follow the static hysteresis very closely, showing that there is no significant difference in the behaviour of
the brace for static versus dynamic loading cases. In addition, in contrast to previous tests, when the HC-
SCED was subjected to these earthquake tests, it experienced many varied cycles of activation and did not
show any signs of response degradation or anomalous behaviour. To further prove this point, the first
earthquake test (Test SC9-A) was run twice to see if there was any change in response between the two runs
(the second run was Test SC11-A). As the top plot in Figure 4.25 shows, the hysteretic response of the
brace is nearly identical for both tests.
4.4.7 DYNAMIC EARTHQUAKE RESPONSE - MAXIMUM-CONSIDERED EARTHQUAKE
Testing the HC-SCED brace using larger MCE-level earthquake required calibration of the external
friction fuse. This process revealed some issues with the external fuse behaviour that will be discussed in a
future section; however, the use of a limited-velocity deformation history allowed the MCE earthquake test
to be conducted. The deformation history that was chosen for this test was that of the first storey brace in
the 12 storey model subjected to LA32 which is the simulated record Elysian Park 1. The lower magnitude
DBE earthquakes responses that were used previously were limited to a maximum velocity of 70mm/s;
however, based on the results of the external friction fuse calibration, this MCE record was limited to a
maximum velocity of 10mm/s. The resulting deformation history for the MCE test and the OpenSees pre-
diction of the HC-SCED brace hysteretic response are shown in Figure 4.26. Notice that at the end of this
MCE record, the residual deformation is not equal to the zero. This is because there was a residual drift
caused by the slip of the external fuse. There is also a residual force in the brace at the end of the record that
was caused by the residual lean of the building combined with P-Delta effects in the model.
This deformation history was applied to the HC-SCED brace in the load frame and the resulting hys-
teretic response is shown in Figure 4.27 (Test SC14-A). Even though the deformation velocity was reduced
for this test, when the external friction fuse activated there was a sharp drop in the force, which was then
immediately recovered. When the recovered force reached the slip load again, there was a smooth slip with-
out any stick-slip behaviour. Towards the end of the slip range there was an additional drop in force; how-

Dynamic Earthquake Tests

3000
2000
Dynamic Test
LA09 Trial 1
1000 (Test SC9-A)
Dynamic Test
Axial LA09 Trial 2
Force 0 (Test SC11-A)
(kN) Static Test
(Test SC8-A)
-1000
-2000
-3000
-40 -30 -20 -10 0 10 20 30 40
3000
2000
1000 Dynamic Test

LA18
Axial (Test SC10-A)
Force 0
(kN)
-1000
-2000
Static Test
-3000 (Test SC8-A)
-40 -30 -20 -10 0 10 20 30 40
Figure 4.25: Dynamic Earthquake Response - Design-Level Earthquake Records
ever, this drop was the result of an input deformation reversal of less than 1mm at the 59s point as shown
in Figure 4.26. This large force drop demonstrates how the external fuse is very sensitive to deformation
due to the high stiffness of the system. Besides these two unexpected external fuse force drops that were
encountered, the HC-SCED brace performed well during this test. It was able to accommodate the defor-

Test SC14-A Input Deformation

SAC Record LA32 Simulated Elysian Park 1
150
4000 OpenSees Prediction
100 2% Drift 2000
Axial 50 Velocity-Limited Axial
Def. 0 Force 0
(mm) (kN)
50 -2000
-100 2%
Original Small Deformation Drop -4000
-150 Drift
0 20 40 60 80 100 120 -150 -75 0 75 150
Figure 4.26: Input Deformation Records for MCE Earthquake Tests
mation demand caused by a maximum credible earthquake, which exceeds the deformation capacity of the
SCED brace itself. It was also able to limit the axial load in the brace at those high deformation levels, as it
was designed to do.
Tests SC14-A - LA32 Simulated Elysian Park 1

3000
2000
1000
Axial
Force 0
(kN)
-1000
-2000
Force Drop Due to
Small Deformation
Drop < 1mm -3000 Force Drop at External Fuse Slip
-120 -100 -80 -60 -40 -20 0 20 40 60
Figure 4.27: Dynamic Earthquake Response - Maximum Considered Earthquake Record
4.4.8 DYNAMIC PROTOCOL TESTS
The last major goal for the HC-SCED tests was to attempt to satisfy the test protocols that are used to
qualify the behaviour of buckling-restrained braces. The test protocol that was used was a combination of

two different code protocols, the ASCE 7-05 general protocol from Chapter 18 (ASCE, 2005) and the
AISC 341-05 buckling-restrained brace protocol from Appendix T (AISC, 2005b).
The ASCE 7-05 test protocol specifies three test sequences:
1. The first sequence consists of 2000 cycles of wind load at the natural frequency
of the building. This sequence is not required if the design wind load is less than
the activation/yield load of the system being tested.
2. The second sequence consists of five cycles at the MCE earthquake level defor-
mation at the effective natural period of the design building corresponding to
that deformation level.
3. The last sequence is required if the force-deformation properties of the brace are
sensitive to the frequency of loading. If so, the second sequence must be
repeated at the first mode period of the structure and at 0.4 times the first mode
period.
The wind loading sequence was applied to the HC-SCED brace twice, at two different frequencies as
described in Section 4.4.4.
The AISC 341-05 test protocol for buckling restrained braces specifies a test sequence that relies on two
parameters:  bm is the deformation corresponding to the design storey drift and  by is the deformation
corresponding to first yielding of the specimen. It does not specify loading rate. Where  b is the control
deformation, the full AISC 341-05 protocol is:
1. 2 cycles with amplitude  b =  by
2. 2 cycles with amplitude  b = 0.5 bm
6. Additional cycles at 1.5 bm such that the cumulative deformation of the brace
is equal to at least 200 by .
The design drift (  bm ) for the first storey SCED brace in the twelve storey building from the study by
Choi et al. (2009) was approximately 1.3% drift. This was determined by finding the mean plus one stan-

dard deviation of the first storey drift for all twenty DBE records in that study. The first mode natural
period of that twelve storey prototype building was 2.1s. The yield drift (  by ) for use with the AISC pro-
tocol was determined using the activation drift of the real HC-SCED brace.
A complete ideal protocol that combined the requirements of both the ASCE and AISC protocols was
formulated with the parameters shown in Table 4.11. The period for each cycle was determined by finding
the ratio of the initial linear stiffness of the brace hysteresis at each drift level divided by the secant stiffness
of the brace and multiplying the square root of that ratio by the first mode period of the prototype build-
ing model:
k1
T i = T 1 ----------
- (Eq. 4.5)
k i sec
Table 4.11: Combined ASCE/AISC Deformation Protocol

Maximum Reduced Equivalent Limited
Cycles Amplitude Drift Deformation Period
Velocity Drift Amplitude Velocity
2  by 0.037% 1.3 mm 1.25 s 6.5 mm/s 0.028% 0.76 by 5 mm/s
2 0.5 bm 0.65% 22.8 mm 4.6 s 31.1 mm/s 0.5% 0.38 bm 5 mm/s
5 1.0 bm 1.30% 45.5 mm 5.8 s 49.4 mm/s 1.0% 0.77 bm 5 mm/s
5 1.5 bm 1.95% 68.3 mm 6.4 s 66.5 mm/s 1.5% 1.15 bm 5 mm/s
2 2.0 bm 2.60% 91.0 mm 7.4 s 76.8 mm/s 2.0% 1.54 bm 5 mm/s
2 2.5 bm 3.25% 113.8 mm 8.3 s 85.8 mm/s 2.5% 1.92 bm 5 mm/s
1 3.0 bm 3.90% 136.5 mm 9.1 s 94.0 mm/s 3.0% 2.30 bm 5 mm/s
2 1.5 bm 1.95% 68.3 mm 6.4 s 66.5 mm/s 1.5% 1.15 bm 5 mm/s
5 5.0 by 0.19% 6.5 mm 1.25 s 32.7 mm/s 0.15% 3.85 by 5 mm/s
5 5.0 by 0.19% 6.5 mm 0.5 s 81.7 mm/s 0.15% 3.85 by 5 mm/s
However, based on the results of the previous HC-SCED brace tests, it was clear that this protocol would
have to be limited in both velocity and deformation amplitude. The maximum velocity of the protocol was
limited to 5mm/s to ensure smooth slippage of the external friction fuses, causing the protocol to look
more like a ‘saw-tooth’ than a sinusoidal excitation. The deformation was likewise limited to 105mm to
prevent over-straining the tendons and to safely keep the deformation within the range already achieved

during the MCE earthquake excitation Test SC14-A. The entire protocol was scaled by a factor of 77% to
stay within this 105mm limit. The scaled protocol parameters and the reduced equivalent drift levels in
terms of the design storey drift  bm are shown in Table 4.11. Note that the reduced drift values still satisfy
the number of cycles at the magnitudes required to satisfy the AISC protocol. It was originally intended
that following this test, larger cycles of deformation would be attempted. The full, scaled and limited pro-
tocol along with the predicted HC-SCED brace response are shown in Figure 4.28.
Test SC15-A&B Input Deformation

Full Brace Protocol Velocity Limited and
150
Amplitude Scaled 4000 OpenSees Prediction
100 2% Drift 2000
Axial 50 Axial
Def. 0 Force 0
(mm) (kN)
50 -2000
-100 2%
Original -4000
-150 Drift
0 100 200 300 400 500 600 -150 -75 0 75 150
Figure 4.28: Input Deformation Records for Dynamic Protocol Test
This protocol was applied to the brace in two tests (Tests SC-15A&B), giving the resulting hysteretic
response shown in Figure 4.29. During these tests, on the return stroke of the final 105mm amplitude
cycle, one of the tendon anchors failed. With the exception of the anchor failure, the rest of the test showed
that the HC-SCED brace performed as-expected, providing full self-centering capability prior to external
friction fuse activation. The external friction fuse slipped easily and provided a predominantly smooth
response at the relatively slow velocity of 5mm/s. Only one small drop in force was encountered on the
compression side at approximately -50mm deformation. This force drop was due to a small fuse slip simi-
lar to those previously described in Section 4.4.7.
The brace did not technically satisfy the ASCE protocol because the tests were pseudo-dynamic at a
velocity of 5mm/s maximum and the deformation history was not applied at the natural frequency of the
system for all cycles; however, in terms of cycle amplitude, the HC-SCED brace did satisfy the ASCE
requirement for five cycles at the MCE earthquake level deformation. In ASCE, the MCE earthquake level
drift is 1.5 times the design level drift, or 1.5 bm . During tests SC15-A&B, the HC-SCED brace resisted
two cycles at 1.54 bm , two cycles at 1.92 bm , and one cycle at 2.30 bm . As previously discussed, the

Tests SC15-A&B - Full Brace Protocol

3000 Tendon Anchor Failure
2000
1000
Axial
Force 0
(kN) Before External
Friction Fuse
-1000 Activation
After External Friction
Fuse Activation
-2000
First External Friction

-3000 Fuse Activation - Small Force Drop
-100 -80 -60 -40 -20 0 20 40 60 80 100
Figure 4.29: Dynamic Displacement Protocol Response
combined reduced protocol that was applied to the HC-SCED brace in Tests SC15-A&B completely satis-
fied the AISC buckling-restrained brace protocol.
4.5 EXTERNAL FRICTION FUSE ISSUES

Although the external friction fuse showed promising results during the subcomponent tests described
in Section 4.2, the full scale fuse that was included in series with the HC-SCED brace did not perform well
enough to be considered entirely successful. As was the case for the subcomponent fuse, the full-sized fuse
required conditioning prior to calibration, a process whereby multiple displacement cycles were applied to
the fuse with different combinations of bolts tightened. At the higher velocities required by the protocol
and earthquake simulation tests, the external fuse exhibited a severe stick-slip type behaviour which was
accompanied by a significant drop in force axial resistance. These issues that were encountered during the
testing of the external friction fuse will be discussed in detail in the following sections.
4.5.1 EXTERNAL FRICTION FUSE INITIAL CALIBRATION AND CONDITIONING
The first tests that were conducted on the HC-SCED brace (Tests CT-1 to CT-4 in Tables 4.6 and 4.7)
were intended to determine the tendon-only response of the brace itself, without the internal friction
dampers activated. So, the normal force bolts in the internal friction dampers were left snug tight and half

cycles of deformation were used to try to characterize the response of the brace due to the tendon only. All
of the bolts in the external friction fuse were tightened using 2200 Nm of torque; it was anticipated that
the normal force provided by this amount of bolt torque would be more than enough to lock the external
friction fuse and prevent it from slipping; however, during the tendon-only tests, the external friction fuses
slipped prematurely, preventing the tendons from reaching their full elongation.
Measurements of the external fuse interface geometry confirmed that it was built as designed and that
there was not any geometric effect that was causing the premature slip behaviour. To investigate the prema-
ture slip, the external fuse bolts were removed and added back to the interface two at a time to see how
each set of two bolts contributed to the total slip force (Tests FE-1 to FE-7). The results of these tests are
shown in Figure 4.30. The plots in this figure show the displacement of the fuse itself measured using the
string potentiometers versus the axial force in the fuse. The first such test (Test FE-1), with only bolts 14
and 15 torqued, had the peculiar looking hysteretic response shown in the top-left of the figure. During
this test, there appeared to be a non-zero post-slip stiffness in the fuse for the first couple of deformation
cycles until a maximum value was reached and the slip stiffness returned to zero for all subsequent cycles.
After inspection of the interface following the test, it was apparent that the non-zero stiffness phenomenon
was being caused by a progressive ‘conditioning’ of the friction surface. The slip force in the interface was
low to begin with due to a smooth initial surface condition. Slipping the interface under normal load
caused the surface to progressively roughen, increasing the required slip force until a steady-state rough
condition had been reached. Once this steady-state surface condition had been achieved, the interface
behaved as a normal friction interface, with constant, zero-stiffness slip behaviour.
Various bolt combinations were tested as shown in Figure 4.30. New combinations as shown by Tests
FE-1, FE-4, FE-6 and FE-7 tended to exhibit the conditioning phenomenon. Only bolts 6, 7, and 9-16
were conditioned during these tests because, based on the slip force results, only these bolts would be nec-
essary to reach the design slip load for the external friction fuse.
4.5.2 VELOCITY DEPENDANCE OF THE EXTERNAL FRICTION FUSE
In preparation for the third dynamic earthquake excitation that would push the HC-SCED brace to a
maximum-considered (MCE) level of response, the external friction fuse of the brace had to be fully cali-
brated. This operation was performed in Tests SC12-A to SC12-D and SC13-A to SC13-R as shown in
Tables 4.8 and 4.9. As described in the previous section, with the first set of tendons, the external fuse had
only previously been tested at a quarter of the total slip load for surface conditioning. So, the first tests of

1500
Test FE−1 Test FE−2
1000
500
Axial
Force 0
(kN)
-500
-1000 Stable Slip Surface
Response Conditioning
-1500
1500
1000
500
Axial
Force 0
(kN) Surface
-500 Conditioning
-1000
-1500
1500
1000
500
Axial
Force 0
(kN) Surface
-500 Conditioning
-1000
-1500
-40 -20 0 20 40
1500 Test FE−7 External Friction Fuse Displacement (mm)
1000
500
Axial
Force 0 13 14 15 16
(kN) Surface
-500 Conditioning 9 10 11 12
5 6 7 8
-1000
1 2 3 4
-1500
-40 -20 0 20 40
External Friction Fuse Displacement (mm)
Figure 4.30: External Friction Fuse Surface Conditioning Tests
the full calibration process were performed first at approximately one-quarter of the target slip. These tests
(SC12-A to SC12-D) were performed at slow velocity so that the slip could be monitored closely; however,
when the velocity was increased in tests SC13-A and SC13-B, as shown on the left side of Figure 4.31, the
external friction fuse slip behaviour became choppy, exhibiting a large stick-slip response with a significant
difference between the static friction slip force when the fuse starts to slip and the average dynamic friction
slip force when the fuse is activated. In Test SC13-B, when the static friction was overcome and the fuse

started to slip the force in the brace dropped by approximately 1000kN. This type of behaviour is not
acceptable for a usable friction slip device.
3000
Test SC13-B Test SC13-N
20 mm/s
2000 0.15Hz sine
Test SC13-L
Test SC13-A 10 mm/s
1000 0.15Hz sine
Axial
Force 0
(kN)
-1000
-2000
Test SC13-M
-3000 Stick-Slip Response 10 mm/s
-50 -25 0 25 50 -80 -40 0 40 80
Total Axial Deformation (mm) Total Axial Deformation (mm)
Figure 4.31: External Friction Fuse Calibration
Continuing the calibration, the torque in the external friction fuse normal bolts was increased to reach
the target slip force. This target slip force for the external friction fuse was lowered from the original
4000kN to 3000kN it account for the previous loss of one tendon and to provide an increased margin of
safety to prevent further tendon anchor failures. The tests were then repeated, but this time at slower veloc-
ity. For tests that had velocities in the range of 5 to 10mm/s, the slip behaviour stabilized and did not
exhibit the stick-slip behaviour as shown on the right side of Figure 4.31 for Tests SC13-L and SC13-M;
however, for faster tests as shown in Figure 4.32, the stick slip behaviour returned.
Test SC13-Q Test SC13-N Test SC13-O Test SC13-P Test SC13-R
15mm/s 20mm/s 35mm/s 50mm/s 10mm/s
3000
2000
Axial 1000
Force 0
(kN)
-1000
-2000
-3000 80mm
-40 0 40 0 40 0 40 0 40 0 40
Figure 4.32: Effect of Displacement Velocity on External Friction Fuse Response

It seems that this slip-stick behaviour in the external friction fuses may not have been seen in the sub-
component tests discussed previously in Section 4.2.1 because the maximum velocity used in those tests
was 14mm/s. It also seems likely that there is an interaction between the external friction fuse and the
behaviour of the HC-SCED brace itself when the full assembly is tested using a displacement-controlled
system. Since the brace itself is so stiff when the internal friction dampers are not activated, any small drop
in displacement results in a large drop in force. When the external fuse slips, its stiffness drops to zero,
allowing the brace to relax slightly, resulting in a drop in force and stopping slip of the external friction
fuse. The brace force then recovers until the external friction fuse can slip again, at which point the brace
force relaxes again slightly and the cycle repeats. The degree of relaxation of the brace when external fuse
slip occurs may be related to the momentum of the brace center of mass, making the effect velocity depen-
dant. This behaviour represents a kind of externally triggered stick-slip response in the external fuse. Below
a certain threshold velocity, around 10mm/s in this system, the brace does not have enough kinetic energy
to cause significant relaxation upon slip, resulting in little force drop and producing a smooth slip behav-
iour.
4.6 TENDON ANCHOR PERFORMANCE ISSUES

The first set of tendons did not achieve their assumed strain capacity during testing due to the prema-
ture slip of the tendon anchors. The reason for these failures was not understood at the time; however, it
was suspected that the tendons with the anchors were not properly assembled prior to shipping. Therefore,
after the first set of tests, a second set of tendons was ordered and all of the original tendons were replaced
in the HC-SCED brace. During the next set of tests, the tendon strain capacity limit was reduced and the
test protocols were modified accordingly. Unfortunately, during the testing using this second set, the ten-
dons continued to experience premature anchor failures. Although these failures do not affect the confir-
mation of the HC-SCED mechanism and behaviour, the design of the tendon anchors remains a
significant issue in the utilization of SCED braces.
4.6.1 ANCHOR SPIKES DISLODGED DURING SHIPPING
The first set of tendons were shipped directly to LCL Bridge in Lachine, QC prior to pretensioning.
Upon inspection of the tendons at LCL, it was found that the spikes inside the tendon anchor barrels had
dislodged from the ends of the anchors in four of the tendons (see Figure 4.33). Since there were extra ten-
dons, it was possible to discard these four tendons so that they would not be used in the HC-SCED brace.

The other tendons seemed to be satisfactory and undamaged; however, they were handled carefully during
brace assembly to prevent any of the other spikes from dislodging.
Figure 4.33: Dislodged Tendon Anchor Spike - First Set of Tendons
4.6.2 PROBLEMS ENCOUNTERED DURING THE TENDON TENSIONING PROCESS
Two main problems were encountered during the pretensioning of the first set of tendons. First, the
threads on the inside and outside of the tendon anchors, on the tensioning rod, and on the steel anchor
nuts were not standard sizes and were not consistent with each other. This caused numerous problems dur-
ing fitting and tensioning. Second, the aluminum anchors were sensitive to damage during handling. The
anchor threads were also easily damaged during the pretensioning operation when the anchors were pulled
through the holes in the end plate. In addition, since the anchor nuts were made of steel, the interior
thread of the nuts tended to catch and damage the softer exterior threads on the anchor. This meant that it
was necessary to use a hammer and punch to forcefully rotate the nut onto the anchor to lock it in place.
After the first phase of testing was halted, the HC-SCED brace was removed from the load frame and
the first set of tendons was removed from the brace. Following the anchor failures in the first phase, sugges-
tions were sent to DRB for improving the fabrication and shipping of the tendons assemblies. These
included using standard threading geometry, checking the fit between parts prior to shipping, and chang-
ing the material of the anchors from aluminum to steel to reduce the chance for anchor damage during
shipping and handling. The most important suggestion was that the tendons should be partially preten-
sioned to at least 100kN prior to shipping to set the anchors more firmly onto the tendons and thereby
prevent the spikes from falling out of the barrels during shipping. In response to this request, DRB shipped
thirteen new tendons that were pretensioned to 100kN so that all the tendons in the brace could be

replaced and the testing of the brace could continue. None of the other suggested changes were made to
the anchors design due to time constraints.
Due to the presence of large connection plates that were not present during the initial pretensioning
process, the second set of tendons had to be pretensioned from the opposite end of the brace. Since it was
then necessary to avoid the connection plate for the inner member for four of these tendons, a new, longer
tension rod was fabricated. Fitting the parts together remained a significant problem. This highlighted the
fact that there was a problem with quality control or standardization in the machining of the anchor
threads. In addition, since the anchors for the second set of tendons were still made out of aluminum, there
were similar anchor damage problems encountered during installation. The anchor threads were prone to
damage during the tensioning process, caused both by the end plate and by the steel anchor nuts.
4.6.3 TENDON ANCHOR FAILURE DURING THE FIRST ATTEMPT AT TENDON-ONLY

CHARACTERIZATION
Once the external friction fuse was properly conditioned, the first major test that was performed on the
HC-SCED brace was a tendon-only characterization without the internal friction dampers activated. This
test (Test CT-5) was designed to test the brace up to its full range of deformation. The full design deforma-
tion capacity of the HC-SCED brace was 70mm, so the test was set up to deform the brace +/- 70mm
using a ramp function at 0.5mm/s.
During that first test, the tendon anchors for tendons 10 and 12 (see Figure 4.14) failed within seconds
of each other at a tendon deformation of approximately 65mm in tension (see the plot in Figure 4.34). As
Figure 4.34 shows, each tendon break is accompanied by a drop in the total axial force since the capacity of
the broken tendon is lost. After the test, it was determined that the anchor failed due to slippage of the ten-
don fibres inside the tendon-anchor assembly. When the tendon assembly failed, the anchor spike was
ejected from the anchor barrel, releasing the tendon which then sprung upwards into the top of the brace,
with the anchor barrel falling freely to the ground. The width of the hysteresis from Test C-T5 shown in
Figure 4.34 is a result of the drop in the tendon force, not any friction in the system.
Following this tendon anchor failure, some of the broken tendon debris became lodged between the
inner W-section and the end plate at the top of the brace. This caused the inner section to be effectively
longer than the outer section, which caused a step in the hysteretic response of subsequent tests, including
Test C-T6 shown in Figure 4.34. This step response was previously introduced and explained in Section
3.5.

4000
3500 Test C-T5

Tendon Only Test 1st Tendon Anchor Failure
3000 2nd Tendon Anchor Failure
Axial 2500 Test C-T6
Force Test After
(kN) Anchor Failure
2000
1500
1000
Step in response due to Tendon Debris
500 from 1st and 2nd tendon anchor failures
(tendon is obstructing top of inner member)
0 10 20 30 40 50 60 70 80
Figure 4.34: Failure of Tendon Anchor during Tendon-Only Test
4.6.4 STEP-RESPONSE OBSERVED DUE TO TENDON DEBRIS
Since one tendon was stuck inside the brace following the first tendon-only test, causing the hysteretic
response to exhibit the step behaviour, the brace had to be disassembled and re-equipped with new tendons
before the critical behavioural tests could continue; however, before doing this, the internal friction
damper was calibrated to ensure that the target slip values could be achieved. The friction calibration was
performed in stages, increasing the friction slip force for the internal friction dampers at every stage. The
25% friction hysteresis was achieved by tightening only the lower friction damper bolts to half of their tar-
get normal force, the 50% hysteresis was achieved by tightening the lower bolts to their full target normal
force, and the 100% hysteresis was achieved by tightening all the bolts in both the upper and lower friction
dampers to their full target normal force. The results of this progressive calibration are shown in Figure
4.35. In addition to showing the widening of the flag hysteresis due to increasing internal damper friction,
these calibration hystereses clearly show the step behaviour caused by the broken tendon stuck in between
the inner member and the top end plate.
4.6.5 TESTING THE REMAINING TENDONS FROM THE FIRST SET
Before the disassembly of the HC-SCED brace to remove the broken tendons, an additional test was
performed to determine if any more tendon anchors would fail before the target maximum brace deforma-
tion. This test (Test CT-7) was similar to the tendon-only Test CT-5, and is shown in comparison to that
test in Figure 4.36. During this test one additional tendon anchor failed, but at the top of the brace this

HC-SCED Internal Friction Tests

(10 of 12 Tendons Only)
2500
Step behaviour due to inner
member obstruction at top
2000
100% Friction
Test FI-4
1500
Axial 50% Friction
Force Test FI-2
(kN)
1000 25% Friction
Test FI-1
500
0 5 10 15 20
Figure 4.35: Internal Friction Damper Calibration with Step Behaviour
time (tendon number 4, see Figure 4.14). This caused the body of the tendon to spring down, creating an
obstruction at the bottom of the brace between the outer section and the end plate as shown in the photo
in Figure 4.37. This tendon debris had to be removed from the inside of the brace by first shimming the
end plate so that the brace was safe to work on, and then cutting the tendon out of the brace using a utility
knife. These operations are reflected in the loading curve shown in Figure 4.36.
4000
3500
Step in response due 1st Tendon Break
to Tendon Debris Test C-T5
3000 from 1st and 2nd tendon 2nd Tendon Break
Axial anchor failures (tendon
2500 is obstructing top of
Force 3rd Tendon Break
inner member)
(kN) 2000 (Test C-T7)
Test C-T7
1500
1000 Shimming of Bottom End Plate

to Remove Tendon Debris from
500 3rd Break (at bottom)
0 10 20 30 40 50 60 70 80
Figure 4.36: Third Tendon Anchor Failure

Figure 4.37: Third Tendon Anchor Failure Photos
Once this debris was removed, the gap between the outer section and the end-plate could be closed
again. Since additional anchor failures would have made it difficult to safely remove the other tendons
from the brace during disassembly, the testing was halted at this point. When this last broken tendon was
removed from the brace, it could clearly be seen that the tendon fibres that were within the barrel of the
tendon anchors were not significantly damaged when the anchor failed (as shown in the photo in Figure
4.37). This suggested that the anchor failures occurred due to a slippage of the tendons between the anchor
spike and the barrel not due to a failure of individual anchor fibres themselves.
4.6.6 FAILURE OF TENDON ANCHOR DURING QUASI-STATIC TEST
Following the tendon anchor failures during the first set of tests, the tendons were removed from the
brace and an entirely new set of tendons was installed as previously described. The internal friction fuse
was calibrated as described in Section 4.4.3 and a quasi-static test was preformed on the brace. During this
quasi-static test, one of the tendon anchors failed during the first half of the second cycle at the maximum
deformation (63.5mm). This caused the body of the tendon to spring down, again creating an obstruction
at the bottom of the brace between the outer section and the end plate as shown in Figure 4.38 (similar to
the previous phase test shown in Figure 4.37). This tendon debris had to again be removed from the brace
by first shimming the end plate and then cutting the tendon out with a utility knife (as reflected in the
loading curve shown in Figure 4.20). Once this debris was removed, the gap between the outer section and
the end plate could be closed again. Following this failure, the decision was made to limit the deformation

of the tendons in all further tests to 44mm. These tendon anchors failed in the same way as the anchors
that failed from the first set of tendons: the tendon fibres slipped out of the anchor barrel without any sig-
nificant damage to the fibres themselves. Only some minor fibre residue was left in the barrel of the anchor
and on the surface of the anchor spike as shown in Figure 4.38.
Figure 4.38: Tendon Anchor Failure Photos
Since these tendons were partially pretensioned before they were shipped and there were no significant
problems with the seating of the anchor spikes observed during assembly, the failure of the tendon anchors
was likely not due to the shipping of the tendons. Therefore, the tendon anchor failures were likely due to
either: (1) a problem with the anchor geometry which prevented the anchor from properly holding on to

the tendon fibres, (2) a problem with assembly of the tendons, or (3) a problem related to the loading his-
tory and pretensioning process. The tendons are particularly sensitive with respect to assembly, and the
fibres may become damaged if the spike is pressed too strongly into the barrel during assembly.
After the tendon anchor failed and the anchor was removed, a characterization test was performed to
determine the new behaviour of the brace with only eleven tendons. The results of this single cycle test are
shown in Figure 4.39. As the figure shows, the removal of one tendon caused the activation force to drop
by approximately 100kN. Since the width of the hysteresis remained constant because the internal friction
damper slip force did not change, the energy dissipation capacity parameter (  ) increased. In addition,
since one tendon was removed, the post-activation stiffness k a also decreased. The measured values for
these brace parameters are shown in Table 4.9. The modified behaviour of the brace with only eleven ten-
dons was determined to be satisfactory to continue the remainder of the tests without replacing all of the
tendons again.
Tendon Anchor Failure Comparison

3000
2000
1000
Axial
Force 0
(kN) Before Anchor Failure
(Test SC4-A)
-1000 After Anchor Failure
(Test SC5-A)
-2000
-3000
-30 -20 -10 0 10 20 30
Figure 4.39: Changes in HC-SCED Hysteresis due to Tendon Anchor Failure
4.6.7 SIMULATION OF STEP BEHAVIOUR
After the tendon anchor failed during the Protocol test described in Section 4.4.8, the tendon itself
remained jammed between the inner member and the upper end plate. A single cycle test (Test SC16-A)
was performed to evaluate the new behaviour of the brace and is shown in Figure 4.40. In this figure, the
new hysteretic response is compared to the prior response from one of the cycles from the protocol Tests

SC15-A&B. As this figure shows, the step behaviour that is characteristic of a mismatched member length
is present due to the tendon debris inside the brace.
2500
2000 Step Behaviour
1500 due to Tendon Debris
in between Member
1000 and End Plate
500
Axial
Force 0
(kN)
-500
Before Tendon
-1000 Anchor Failure
(Test SC15-A)
-1500 After Tendon
Anchor Failure
-2000 (Test SC16-A)
-2500
-25 -20 -15 -10 -5 0 5 10 15 20 25
Figure 4.40: Effect of Tendon Debris on Hysteretic Response
This step behaviour in Test SC16-A was modelled using the Mechanics Simulator model as shown in
Figure 4.41. Similarly to the previous Figure 4.23, this figure compares the recorded test hystereses for the
full brace and the individual elements to the hystereses predicted by the Mechanics Simulator model. The
inputs for this model were previously presented in Table 4.10. At the top right of the figure, it also shows
the brace hysteresis as calculated using the individual brace element forces as described in Section 4.4.2.
This calculated response does not do as good of a job matching the load cell response for this test as it did
for Test SC8-A. This is because the tendon force during the test was calculated indirectly using the relative
movement of the inner and outer members. This method cannot capture the fact that the tendon actually
has a flat portion in its hysteresis where the increase in brace deformation does not increase the force in the
tendon (as shown in the model prediction for the tendon response). The tendon force does not increase
during that flat portion because of the differing member lengths. During that portion, the tendon was
bearing on the same member at both ends, while the other shorter member shifted within the longer mem-
ber. This shift is also responsible for the flat step in the overall hysteretic behaviour of the brace (as previ-
ously discussed in Section 3.5).

2000
Test SC16-A
Axial Force
1500 Axial Force Calculated
from Load using Brace
1000 Frame Model Member Force Model
Hysteresis and Tendon Hysteresis
Brace 500
Elongation
Axial
Force 0
(kN)
-500
-1000
-1500
-2000
-20 -10 0 10 +/-20 -10 0 10 20
Inner Outer
Member Member
1500
Test SC16-A
1000
500
Axial 0
Shifted
Force Test Hysteresis
-500
(kN)
Model
-1000 Hysteresis
-1500
-2000
-20 -10 0 10 +/-20 -10 0 10 20
Tendon
1500
Test SC16-A
1000
Axial
Force 500
(kN)
0
-500
-20 -10 0 10 20
Figure 4.41: Simulation of Brace Response and Calculation of Brace Element Forces for Test SC16-A
The figure shows that the Mechanics Simulator model again well-predicts the response of the HC-
SCED brace, including the step behaviour. It also has a similar effective initial stiffness k 1 eff , with a value
of 422kN/mm from the model versus a value of 397kN/mm calculated based on the test data.

4.6.8 ULTIMATE TESTING OF REMAINING TENDON ANCHORS
After the tendon debris became lodged within the brace as described above, it was not possible to con-
tinue with any further protocol tests. Therefore, the HC-SCED brace was tested one last time in compres-
sion and tension up to the internal friction damper deformation limit to see if any more tendon anchors
would fail (Tests SC17-A&C). The hysteretic responses for these tests are shown in Figure 4.42. To prevent
the external friction fuse from slipping, for the second test half of the internal friction damper normal force
bolts were released to increase the deformation in the brace at which the external fuse would slip. No more
tendon anchors failed during these tests. This suggests that the remaining tendon assemblies were able to
accommodate their design deformation capacity.
3000
Test SC17-C
Half Internal
2000
Friction
1000
Axial
Force 0
(kN)
-1000
-2000 Test SC17-A

Full Internal
-3000 Friction
-60 -40 -20 0 20 40 60
Tendon Deformation (mm)
Figure 4.42: Attempt to Fail Tendon Anchors
4.6.9 TENDON PERFORMANCE SUMMARY
By the end of the HC-SCED testing program, it was clear that there is an unresolved tendon anchor
reliability issue. Table 4.12 provides a summary of the tendon loading histories and anchor failure loads for
the first and second set of tendons. As this table shows, both sets of tendon anchors failed at similar tendon
deformations and tendon forces.
The 22mm tendons that were used for the HC-SCED brace were not used previously in SCED
designs. The anchors for the 22mm tendons were fabricated by DRB in Korea and not by the tendon sup-
plier Linear Composites Ltd. As previously mentioned, the potential causes for the tendon anchor failures

Table 4.12: Tendon Anchor Performance Comparison

First Set Second Set
of Tendons of Tendons
Tendons Pretensioned Prior to Shipping No Yes
Maximum Pretension Load during 285 kN 255kN
Installation
Tendon Lock-off Service Load 85 kN 75 kN
Approximate Tendon Elongation at 65mm 59mm
Anchor Failure
Approximate Tendon Load at Anchor 233 kN 255 kN
Failure
have been narrowed to (1) a problem with the anchor geometry which prevents the anchor from properly
holding on to the tendon fibres, or (2) a problem with assembly of the tendons, or (3) a problem related to
the loading history and pretensioning process. In addition, as previously mentioned, using a stronger,
harder, and more ductile material for the anchors such as steel could possibly improve their performance. It
is clear that a comprehensive testing program must be performed to thoroughly evaluate the behaviour of
the tendon anchors; however, such a testing program lies outside the scope of this thesis.
4.7 CONCLUSIONS
The high-capacity SCED (HC-SCED) brace test program proved that it is possible to build a function-
ing SCED brace with a high axial load capacity comparable to other competing available bracing systems
such as the buckling restrained brace (BRB). During the tests, the HC-SCED brace exhibited a stable and
smooth response when subjected to both static and dynamic loads including simulated earthquake defor-
mation histories.
In addition to the tests of the brace itself, a new type of high-capacity external friction fuse consisting of
a mild steel on stainless steel friction interface was tested. This fuse was designed to double the deforma-
tion capacity of the combined system at the expense of limited residual axial deformation in the system.
The fuse only performed well within a constrained range of velocities, within which it exhibited a predict-
able resistance and stable response. At higher velocities, the external friction fuse exhibited a severe stick-
slip response which may limit the applicability of this type of system; however, such large braces still have
the option to use NAO friction pads in the external fuse such as the one that was designed by Christopou-
los et al. (2008) instead of the steel-on-stainless-steel friction interface. Since this option would be quite
expensive for a high-capacity brace, it would be worthwhile in the future to continue exploration of the

type of high-capacity fuse described in this chapter using different friction surface materials. To eliminate
the need for an external friction fuse in order to accommodate extreme earthquake events, an enhanced-
elongation telescoping SCED brace (T-SCED) has been designed as part of this thesis and will be
described in the following chapter.
The combined HC-SCED brace with the external friction fuse was tested under a limited-velocity
qualification protocol that was designed to satisfy the requirements of both the ASCE and AISC qualifica-
tion protocols for buckling-restrained braces. With the exception of the dynamic requirements of the
ASCE protocol, the HC-SCED system satisfied both qualification protocols.
Despite the success of the HC-SCED brace design, serious lingering questions remain regarding the
behaviour and reliability of the tendon anchors. Multiple tendon anchor failures were experienced during
the testing of the HC-SCED brace at force levels that were significantly lower than their nominal design
capacity. Comprehensive tendon-anchor assembly tests must be conducted in the future to resolve this
issue and provide confidence in the use of the tendons. Although the HC-SCED brace did not fully reach
the target design load of 4000kN, these tests showed that such an axial force capacity is possible once the
tendon issues are solved.

CHAPTER 5: DESIGN AND TESTING OF A PROTOTYPE
TELESCOPING (T-SCED) BRACE
The deformation range over which traditional SCED braces can exhibit self-centering response is lim-
ited by the length of the brace members and by the material that is selected for the pretensioned tendons.
For typical building storey geometries, these design constraints typically result in an ability to self-center
under storey drifts of ±1.5% to ±2.0% times the storey height. To accommodate larger storey drifts, SCED
braces such as the HC-SCED brace described in the previous chapter must be equipped with an external
friction fuse in series with the main brace. The inclusion of this fuse allows the system to deform as much
as necessary to accommodate extreme earthquake scenarios. The use of an external fuse means that, for
drift demands greater than the self-centering deformation range of the brace, the SCED frame system
experiences some residual drift; however, this residual drift would only occur under maximum considered
earthquakes (MCE level, with a 2% probability of exceedence in 50 years) in high seismic zones. In addi-
tion, these residual drifts would be much smaller for a SCED with an external fuse than they would be for
yielding- or friction-type systems. For moderate and low seismic zones, and for more frequent but moder-
ate-level earthquakes in high seismic zones, previous SCED braces, including the HC-SCED, have been
able to provide full self-centering capability with no residual drifts.
To enhance the self-centering capability of SCED braces for use in high seismic applications, the next
major goal to be accomplished in this thesis was to extend the SCED concept so that it would be able to
sustain much higher deformations. This was done by developing a telescoping SCED brace concept that is
able to double the brace deformation over which full self-centering response can be achieved. This tele-
scoping SCED configuration, or T-SCED brace, was designed, built at full scale, and tested statically and
dynamically in a full scale, single-storey steel frame. This chapter describes this enhanced T-SCED brace
and presents results from the laboratory testing program that confirmed its behaviour.
177
Chapter 5: Design and Testing of a Prototype Telescoping (T-SCED) Brace 178
5.1 TELESCOPING SCED (T-SCED) DESIGN AND CONSTRUCTION

The mechanics of the telescoping SCED (T-SCED) brace concept were previously described in detail
in Section 3.3. As discussed previously, this system relies on the basic mechanics of the original SCED sys-
tem, except that it adds an extra element: one or more new 'floating' intermediate steel members that are
parallel to the existing inner and outer members. This new intermediate member allows the addition of a
second set of tendons. In this novel configuration, the two sets of tendons and the intermediate member
are effectively in series, allowing the two sets of tendons to share the brace deformation equally between
them. Therefore, if the strain capacity of each individual set of tendons is the same as it was for the original
SCED, the T-SCED will have double the total elongation capacity.
5.1.1 T-SCED DESIGN
The T-SCED brace prototype was designed to be tested in a full-scale vertical steel frame in the struc-
tural laboratory at the University of Toronto. The use of a realistic frame allowed the tests to include the
effect of the rotation of the joints at the ends of the brace caused by the frame drift. The design of this test
frame will be described in the next section. The steel test frame had a bay width of 7.32m (24 ft.) and a
height of 2.74m (9 ft.) which was determined based on the maximum available space in the laboratory.
This frame size constraint in turn determined the available space for the T-SCED brace and the connection
geometry.
Based on the size of the steel test frame, target design parameters were formulated for the prototype T-
SCED brace as shown in Table 5.1. The top priority for this prototype design was to prove that the T-
SCED concept was feasible and could provide comparable performance to previous designs without the
need for an external friction fuse. Therefore, the axial force capacity of the brace was chosen to be within a
similar range as earlier traditional designs, with an activation force comparable to that of the braced-frame
SCED prototype from Christopoulos et al. (2008); however, the novel telescoping configuration doubled
the target storey drift level compared to previous designs from 2.0% to 4.0%.
A schematic showing the layout of the major elements of the brace is shown in Fig. 5.1. Full technical
design drawings of the brace may be found in Appendix D. In contrast to the HC-SCED design, all of the
members in the T-SCED brace were commonly available steel sections: the outer member was an
HSS508x305x12.7, the intermediate member was an HSS356x254x9.5 and the inner member was a
W200x59. To try to avoid the tendon problems that were encountered during the tests of the HC-SCED
brace, the pretension force in the T-SCED brace was provided by two sets of six 17mm nominal diameter

Table 5.1: T-SCED Target Design Parameters

Parameter Value
SCED Ultimate Axial Load P u 820 kN
Tendon Pretension Force P p0 210 kN
Total Internal Friction Damper Slip Force F it 190 kN
SCED Activation Load P a 400 kN
Peak Tendon Strain  pu 1.6%
Energy Dissipation Capacity Parameter  0.95
Estimated Post-Activation Stiffness k a 4.2 kN/mm
aramid tendons. These tendons were identical to those that were used in earlier SCED prototype tests prior
to the HC-SCED as described by Christopoulos et al. (2008). The tendons were anchored to two 102mm
(4”) thick steel dual end plates at either end of the brace using aluminum spike-and-barrel tendon anchors.
These two nested end plates (an inner end plate and an outer end plate) at either end of the brace were
required to provide the mechanical workings of the T-SCED brace concept as previously described in Sec-
tion 3.3.
To provide energy dissipation, four internal friction dampers were used that had two 3/4” normal force
bolts each. Similar to the HC-SCED brace, the friction in the internal friction dampers was provided by
stainless steel plates welded to the friction interface plate surface and Carlisle NF-916 friction pads recessed
into steel friction interface angles that clamped the interface plate. The friction interface plate was welded
to the inner W-section member and the interface angles were welded to the outer member. The intermedi-
ate member ‘floated’ and did not interact with the internal friction dampers at all. As previously men-
tioned, since the T-SCED design provides double the elongation capacity of the traditional SCED design,
it is able to accommodate extreme earthquake scenarios without the need for an external friction fuse.
Like the HC-SCED brace, the inner member of the T-SCED brace was an I-shaped section instead of a
tube to allow a more direct transfer of load between the connection and the inner member. The ends of the

Friction Interface Angle/

Outer HSS Member
End Connection
Outer HSS Member

Friction Interface Angle
Full Brace Angle Stiffener
Inner End Plate

Tendon
Anchor
Intermediate HSS Member

Outer
End Plate
Middle of Brace
Tendon
Intermediate HSS Member

End Stiffener Plastic Pad
Inner W-Section Member
Interior of Brace
Inner End Plate
Tendon
Friction Interface Plate
Inner W-Section Stainless Steel Friction Plate
Member End Connection
Figure 5.1: T-SCED Brace Schematic
inner, intermediate, and outer members were also machined to achieve a tight length tolerance as was done
for the HC-SCED brace. The connection detail for the inner member was also similar, with a 445mm
(17.5”) long section of the web being cut out to accommodate a 25mm (1”) thick steel plate welded to the
remaining web and the top and bottom flanges of the member. That thick steel plate extended past the end
of the member and through a slot cut in the inner end plate, making it available for use as an end connec-
tion for the inner member. The details for the internal friction dampers were also similar to the HC-
SCED, with triangular steel plates used to stiffen the friction interface angles and transfer the vertical load
caused by the eccentric bolt loading to the outer tube walls.
The key design challenge in the development of the T-SCED brace was to size and arrange the inner
and outer end plates such that when the brace is at rest and fully closed, the inner end plate abuts the inner
and intermediate members and the outer end plate abuts the intermediate and outer members. This type
of arrangement is required for the T-SCED mechanism concept to work properly. The cross-section of the
prototype T-SCED brace, along with the layout of the tendons and the inner and outer end plates, is
shown in Figure 5.2.

A 508mm
Internal
Friction
Damper
Inner End Plate

Outer End Plate
8mm 50mm 80mm

9mm
Tendon
298mm 29mm Anchor
Nut
13mm Tendon
13
mm Anchor
Outer Member
13 Intermediate
mm
Member
Inner
Connection Member
Plate
521mm
Section A
Figure 5.2: T-SCED Cross Section with Clearances
The inner, intermediate and outer members for the T-SCED brace were fabricated by Constructions
Beauce-Atlas in Sainte-Marie de Beauce, Quebec. All of the other brace components except for the tendons
were fabricated by LCL-Bridge in Lachine, Quebec. Final assembly of the T-SCED brace was performed at
LCL-Bridge under supervision of the author. Tendon pretensioning was performed by the author directly
(with the support of the staff at LCL-Bridge). Some photographs of the brace elements and the assembly
process are shown in Figure 5.3.
5.1.2 TENDON ASSEMBLY
Unlike the fabrication process for the HC-SCED brace, where the tendons arrived from Korea pre-
assembled, for the T-SCED brace the tendons were all individually assembled on site prior to pretension-
ing. Performing the tendon assembly in-house allowed complete quality control over the assembly proce-

Figure 5.3: T-SCED Assembly Photos

dure. This was done to mitigate the assembly quality control and tendon shipping issues which may have
contributed to the tendon anchor failures during the HC-SCED brace tests. The tendons themselves were
sourced from DRB in Korea; however, for this test, the tendon anchors for the T-SCED brace were sourced
directly from Linear Composites Ltd. instead of being fabricated by DRB as they were for the HC-SCED
brace. The tendon assembly was conducted using instructions from the manufacturer of the tendons and
tendon anchors, Linear Composites Ltd. (2011). An image from this document depicting the tendon
assembly procedure is shown in Figure 5.4.
Figure 5.4: Tendon Assembly Procedure (from Linear Composites Ltd., 2011)
As instructed in this document, the tendons were first cut to length using a utility knife. Then a section
of the plastic sheath was removed from the end of the tendon, using a utility knife to cut most of the way
through the sheath and bending the tendon at the cut location to break the sheath the remainder of the
way through. This method prevents any of the fibres from being inadvertently cut. The spike was then
located with care at the centre of the fibres, ensuring that the fibres were evenly distributed all the way

around the spike. The anchor body was then pulled up over the spike and the tendon body was pulled
down hard to set the spike in place. To finish the installation, a punch and hammer was used to gently tap
on the head of the spike to ensure that the anchor assembly was secure. Some photos of this process are
Figure 5.5: Tendon Assembly Photos
5.1.3 TENDON TENSIONING
As previously mentioned, twelve 17mm nominal diameter Technora aramid tendons were used to pro-
vide the pretensioning force for the T-SCED brace. These tendons were smaller than the 22mm tendons

that were used for the HC-SCED brace, but they had the similar properties shown in Table 5.4. The
assumed strain capacity of the tendons was lowered based on the results from the HC-SCED tests to
reduce the possibility of premature anchor failure.
Table 5.2: Assumed Tendon Properties - 17mm Tendons

Property Value
Nominal Diameter D 17 mm
Cross-Sectional Area A p 76 mm2
Initial Young’s Modulus E p0 44 GPa
Effective Cyclic Young’s Modulus E p 102 GPa
Initial Tendon Length (not including anchor) L p0 5115 mm
Service Tendon Length L p 5232 mm
Nominal Ultimate Tendon Stress  pu 1960 MPa
Nominal Ultimate Tendon Strength P piu 149 kN
Assumed Tendon Strain Capacity  pu 1.6%
The tensioning of the tendons for the T-SCED brace was conducted at LCL Bridge. The tensioning
profile for the tendons is shown in Figure 5.6. For this brace, the tendons were taken to 77% of their nom-
inal breaking load and then released to a service load of 34 kN. The tendon tensioning procedure for the
T-SCED brace was similar to the procedure that was used for the HC-SCED brace previously described in
Section 4.3. The tendon tensioning order is shown in Figure 5.7.
5.2 TEST FRAME DESIGN

A full-scale reusable steel test frame was designed and fabricated to test the T-SCED brace. This test
frame utilized existing W360x347 reusable stiffened test columns from a previous test program (Gray,
2012). The frame was designed to be freestanding and was actuated relative to the strong wall in the struc-
tural laboratory at the University of Toronto using two dynamic 1000kN actuators. The full design draw-
ings for the test frame elements may be found in Appendix E. The full test setup is shown in Figures 5.8
and 5.9.
The test setup consisted of a single storey steel braced frame with a storey height of a 2.74m (9 ft) and a
bay width of 7.32m (24ft). The actuators that displaced the top of the frame were connected to the middle

Tendon Forces calculated Service

σp (MPa) Ap = 76mm2 εp = 1.04% εpa = 3.42%
Approximate Elongations ΔLp ≈ 51 mm σp = 1504 MPa
Initial Length: Pp ≈ 114 kN
Lp0 = 4787mm
Lp = 4904mm
εp = 0.43%
εpa = 2.38%
σp ≈ 443 MPa
Ep0 ≈ 44GPa
ΔLpa ≈ 114 mm
Pp = 34 kN
εp0
εpu
1.47%
Figure 5.6: T-SCED Tendon Tensioning Plot
5 7
3 9
1 11 12 2
10 4
8 6
Figure 5.7: T-SCED Tendon Tensioning Order
of the W460x128 beam at one end and anchored to the strong wall at the other end. The bases of the col-
umns were connected to the strong floor via two identical true pin connections. The beam to column con-
nection consisted of a double angle header with slotted holes in the beam to accommodate rotation. The
brace was connected to the frame via a straight bolted connection at the top (south end). At the bottom
(the north end), the brace was originally connected to the frame using a straight bolted gusset plate con-
nection (as shown in Figure 5.3); however, during loading, this connection caused some bending and shear
in the brace, as will be discussed in a future section. To fix this problem, the frame was retrofit by replacing
the lower gusset connection with a true pin connection which is shown in Figure 5.8. Beam and column
lateral restraints were constructed to restrain the motion of the frame out of plane.

Actuator Wall Mounts 7.32m

Reusable Column
North
Strong Wall 1000kN
Dynamic
Actuators Actuator Beam
Connection
Beam
Restraints
Upper Gusset
Brace
T-SCED
2.74m
Pinned Column
Bases
Lower Gusset
with Pin
Connection Strong Floor
Column Restraints
Figure 5.8: T-SCED Test Frame
Figure 5.9: T-SCED Test Frame Photo

5.3 TEST INSTRUMENTATION

The instrumentation of the test frame and the T-SCED brace is shown in Figure 5.10. Strain gauges
were applied to both columns at two locations each to measure the axial force, shear and moment. Six
strain gauges were applied at each of the two locations on each column. Six strain gauges were applied to
the beam to measure the axial force in the beam due to the actuators. Four strain gauges were applied to
each member in the T-SCED at two locations on each member. These gauges measured the axial force,
shear and moment in the brace members. In addition to the strain gauges, linear variable differential trans-
formers (LVDTs) were used to measure displacements and deformations of the brace and frame. One
LVDT was used to measure the total deformation of the brace from end to end. Four additional LVDTs
measured the movements of each of the end plates (inner and outer) relative to the outer member at each
end of the brace. Two LVDTs at the base of each column measured the movement of the steel pin relative
to the column itself in the vertical and horizontal directions. An instrumentation column was used to pro-
vide a tall absolute reference point equivalent to the strong floor. This instrumentation column was
mounted with two LVDTs: one to measure the horizontal in-plane movement of the beam, and one to
measure the horizontal out-of-plane movement of the beam.
5.4 BARE TEST FRAME TESTS

Before the T-SCED brace was inserted into the test frame, the steel test frame itself was tested. The first
goal of these tests was to determine how to properly control the two actuators together. Two control
schemes were tested: a master/slave scheme where the measured displacement of one actuator controlled
the command displacement to the other, and an independent command scheme where the same control
displacement signal was sent to both actuators. The independent command scheme resulted in better syn-
chronization between the force and displacement of the two actuators. Following selection of the control
scheme, the actuator response was tuned by applying a square wave excitation and manually adjusting the
control gains to optimize the response.
In addition to actuator control adjustment, these bare frame tests were used to adjust the lateral support
and get an estimate of the friction inherent in the frame itself without the brace installed. This hysteretic
response of the frame without the brace is shown in Figure 5.11. The test shown in this figure is a full
dynamic protocol displacement history up to 4% storey drift. The displacement of the two actuators (not
shown in the figure) were within 0.5mm of each other throughout this test; however, as shown in the fig-
ure, the force in the two actuators differs by up to 30kN. There was some concern at the time that this

Strain Gauge Locations
64 64
6 Beam 102
Strain
Gauges 102
2007
12 South
Column
Strain Gauges
12 North 64 64
Column
Strain Gauges 89
89 Same 89
locations as 1676
89 south column 89
24 T-SCED 1715
Strain Gauges 89
57 57
(8 each member) 991
64 76
1676
152
All Dimensions in mm
Instrument Locations
Instrumentation
Column
Beam In-Plane Beam
Movement LVDT Out-of-Plane
Movement
LVDT
End Plate LVDTs: one for Inner

End Plate, one for Outer End Plate
(at Both Ends of Brace) Horizontal
Horizontal and Vertical
and Vertical LVDTs to
LVDTs to measure
measure movement
movement of pin relative
of pin relative to column
Total Elongation
to column LVDT
Figure 5.10: T-SCED Test Instrumentation Summary
force difference indicated that the control scheme was not working well; however, during the tests of the
frame with the T-SCED brace the force difference between the two actuators did not increase. The total
force response from the two actuators gives an estimate of the friction response of the frame. This test
showed that there was approximately 20kN of friction force inherent to the test frame.

Bare Frame Test - Dynamic Protocol - Test BF-11

40
30 West
Actuator
20
10
Actuator
Force 0
(kN)
-10
-20
East
-30 Actuator Total Force
Both Actuators
-40
-100 -80 -60 -40 -20 0 20 40 60 80 100
Frame Drift (mm)
Figure 5.11: Bare Frame Hysteretic Response with no T-SCED Brace
5.5 T-SCED BRACED-FRAME TESTS

The T-SCED braced frame tests were conducted in the structural laboratory at the University of
Toronto. The testing program is summarized in Table 5.3, including the test input displacement parame-
ters and comments describing the tests and relevant observations. The resulting T-SCED brace hysteresis
parameters for these tests are shown in Table 5.4, including the measured brace activation force, initial and
post activation stiffnesses and the energy dissipation capacity parameter (  ). Note that although the initial
stiffness leg of the SCED brace response is actually multilinear as described in Section 3.5, the initial stiff-
ness given in these tables is the highest measured stiffness, not the effective initial stiffness. The activation
force in the table is dependent on the initial zeroing of the force measurements before the test, therefore an
average value is provided in cases where both positive and negative activation forces were measured. This
average value is the most useful estimate of the activation force, since this value is relatively stable across all
of the tests except when the brace behaviour changes for a known reason such as the removal of internal
friction damper normal force bolts. The  values are only provided for tests where the average activation
force is calculated. The full test program and results will be described in detail in the following sections.
5.5.1 T-SCED BRACE AXIAL FORCE CALCULATION
The force in the brace was estimated using three different methods. The first method used the input
actuator force from the load cells in the actuator and projected that force into the brace direction; however,

Table 5.3: T-SCED Brace Load Frame Testing Summary

Y-M-D mm/s mm
Disp. Disp
Rate Ampl.
TENDON-ONLY CHARACTERIZATION
TO-1 11-11-08 0.5 +10 1/2 Good response, but brace def. lags behind frame def.
TO-2 11-11-08 0.5 +/-10 1 Same problem, LDVTs may be loosely attached
TO-3 11-11-11 0.5 +/-10 1 Beam LVDT better, brace LVDT still loose
TO-4 11-11-15 0.5 +/-10 1 Def. lag caused by pin and beam connection movement
TO-5 11-11-15 0.5 +/-10 1 Calibrate amplitude for brace target. Good response.
TO-6 11-11-15 2.0 +/-10 1 Increased velocity, good response
TO-7 11-11-15 5.0 +/-10 1 Increased velocity, good response
INTERNAL FRICTION DAMPER CALIBRATION
TI-1 11-11-16 5.0 +/-10 1 Half friction calibration
TI-2 11-11-18 5.0 +/-10 1 Full friction calibration
QUASI-STATIC BRACE PROTOCOL - FIRST ATTEMPT
TPA-1 11-11-21 5.0 +/-1.0 1  by cycle
TPA-2 11-11-21 5.0 +/-1.0 1  by cycle
TPA-3 11-11-21 5.0 +/-18 1 0.5 bm cycle - significant frame hysteresis
TPA-4 11-11-21 5.0 +/-18 1 0.5 bm - loosened lateral support - still column shear
TPA-5 11-11-22 5.0 +/-18 1 0.5 bm - loosened lower gusset bolts
QUASI-STATIC BRACE PROTOCOL - SECOND ATTEMPT WITH RETROFITTED LOWER GUSSET
TPB-1 11-12-02 5.0 +/-1.0 1  by cycle
TPB-2 11-12-02 5.0 +/-18 1 First 0.5 bm cycle - elastic force reduce, good response
TPB-3 11-12-06 5.0 +/-18 1 Second 0.5 bm cycle
TPB-4 11-12-06 5.0 +/-36 1 First 1.0 bm cycle
TPB-5 11-12-06 5.0 +/-36 1 Second 1.0 bm cycle
TPB-6 11-12-06 5.0 +/-54 1 First 1.5 bm cycle - hear some pings, stop test
W-1 11-12-12 sine +/-1.0 500 1.25s period
W-2 11-12-12 sine +/-1.0 500 0.60s period
TPC-1 11-12-13 5.0 +/-18 1 Recharacterization of T-SCED response after wind
DYNAMIC BRACE PROTOCOL
DPA-1 11-12-13 sine +/-18 2 0.5 bm cycle, 2.3s period


Y-M-D mm/s mm
Disp. Disp
Rate Ampl.
DPA-2 11-12-13 sine +/-27 5 0.75 bm cycle, 3.1s period
>THIRTY-THREE HYBRID TESTS CONDUCTED BY KAMMULA
DPB-1 12-01-18 sine +54 1/2 1.5 bm half cycle, 3.6s period
DPB-2 12-01-18 sine +/-54 1 1.5 bm cycle, 3.6s period - tripped disp. difference limit
DPB-3 12-01-18 sine +/-54 1 1.5 bm cycle, 3.6s period - inner friction plate jamming on
end plate slot
DPB-4 12-01-18 5.0 +54 1/2 1.5 bm slow half cycle
DPB-5 12-01-18 sine +71 1/2 2.0 bm half cycle, 3.9s period
DPB-6 12-01-20 5.0 -38.9 1/2 Repair end plate slot, install end plate guides, test repair with
half cycle, lost data
DPB-7 12-01-20 5.0 -38.9 1/2 Repeat DPB-6
DPB-8 12-01-20 sine +/-54 1 1.5 bm cycle, 3.6s period
DPB-9 12-01-23 sine -71 1/2 2.0 bm half cycle, 3.9s period
DPB-10 12-01-23 sine +/-71 1 2.0 bm cycle, 3.9s period
DPB-11 12-01-23 sine +/-54 3 1.5 bm cycles, 3.6s period
DPC-1 12-01-23 sine +/-71 10 Full Protocol up to 2.0 bm . Problem with T-SCED End
plate LVDT readings - recalibrate after test.
>ADDITIONAL SIX HYBRID TESTS CONDUCTED BY KAMMULA
CONTINUE DYNAMIC BRACE PROTOCOLS
DPC-2 12-01-26 sine +/-71 1 2.0 bm cycle, 3.9s period
DPD-1 12-01-26 sine +89 1/2 2.5 bm half cycle, 4.1s period
DPD-2 12-01-26 sine -89 1/2 2.5 bm half cycle, 4.1s period
DPD-3 12-01-26 sine +/-89 1 2.5 bm cycle, 4.1s period
DPD-4 12-01-26 sine +100 1/2 2.75 bm half cycle, 4.2s period. Tripped force limit due to
reaching end of slot in internal friction damper
DPD-5 12-01-26 sine -100 1/2 2.75 bm half cycle, 4.2s period. Tripped disp. diff. limit
DPD-6 12-01-26 sine -100 1/2 2.75 bm half cycle, 4.2s period.
DPD-7 12-01-26 sine -107 1/2 3.0 bm half cycle, 4.3s period. Tripped disp. diff. limit


Y-M-D mm/s mm
Disp. Disp
Rate Ampl.
DPD-8 12-01-26 sine -107 1/2 3.0 bm half cycle, 6.1s period.
DPD-9 12-01-27 sine +/-100 1 Remove bolts from friction damper to allow greater sliding
range. Rerun 2.75 bm cycle, 6.0s period.
DPD-10 12-01-27 sine +107 1/2 3.0 bm half cycle, 6.1s period.
DPD-11 12-01-27 sine +/-107 1 3.0 bm cycle, 6.1s period.
RECALIBRATE INTERNAL FRICTION DAMPER TO COMPENSATE FOR REMOVED BOLTS
TI-3 12-01-27 man. +10 1/2 Recalibrate internal friction damper - manual control
TI-4 12-01-27 man. +/-6 1 Recalibrate internal friction damper - manual control
TI-5 12-01-27 man. +/-6 1 Recalibrate internal friction damper - manual control
DPE-1 12-01-31 sine +89 1 2.5 bm half cycle, 5.8s period
DPE-2 12-01-31 sine +/-107 1 3.0 bm cycle, 6.1s period.
DPF-1 12-01-31 sine +/-107 7 Full Protocol up to 3.0 bm , one cycle at each level
EQR-1 12-01-31 EQ 20 Simulated Cascadia earthquake response at full speed
EQJ-1 12-01-31 EQ 77 LA32 velocity limited (same as HC-SCED Test SC14-A)
DPF-2 12-01-31 sine +/-74 6 Reduced Speed Full Protocol up to 3.0 bm , one cycle at
each level. Tendon anchor failure at 74mm
ATTEMPT TO FAIL ADDITIONAL TENDONS
TB-1 12-01-31 man. -109 Slowly, manually take brace to 3.9% drift, no tendons fail
this method over-estimated the axial force in the brace because it included all of the force that was lost
through friction between the frame members and between the frame and the lateral support system.
Hence, this method provides an upper bound on the response and was only used to ensure that the force
calculated using the other two methods was reasonable.
The second calculation method was similar to that used for the HC-SCED brace as described in Sec-
tion 4.4.2, whereby the axial force in the SCED brace was calculated using the strain gauge data from the
inner and outer member and the measured elongation of the tendons. Since the T-SCED has an extra

Table 5.4: T-SCED Brace Load Frame Test Result Summary

kN kN kN/mm kN/mm
Est. Estimated
Estimated
Internal Post-
Activation Force Initial Estimated
Test # Damper Activation
Normal
Pa Stiffness
k1
Force ka
- + AVG - + - + - +
TENDON-ONLY CHARACTERIZATION
TO-1 210 1145 9.1
TO-2 -220 259 240 231 199 7.6 7.9 0.15 0.09
TO-3 -252 232 242 220 207 4.8 5.6 0.16 0.09
TO-4 -249 230 240 221 201 4.5 5.4 0.12 0.08
TO-5 -255 231 243 219 216 4.9 5.5 0.16 0.09
TO-6 -249 233 241 227 230 4.8 5.3 0.17 0.11
TO-7 -253 228 241 217 217 4.9 5.0 0.16 0.10
INTERNAL FRICTION DAMPER CALIBRATION
TI-1 327 -340 318 329 265 227 5.8 4.9 0.67 0.65
TI-2 645 -421 418 420 232 197 6.5 5.0 0.96 0.94
QUASI-STATIC BRACE PROTOCOL - FIRST ATTEMPT
TPA-1 645 230 218
TPA-2 645 204 222
TPA-3 645 -407 452 430 222 212 5.2 4.6 1.00 0.96
TPA-4 645 -393 483 438 232 246 5.3 4.8 0.97 0.94
TPA-5 645 -419 429 424 298 222 5.7 4.5 0.98 0.98
QUASI-STATIC BRACE PROTOCOL - SECOND ATTEMPT WITH RETROFITTED LOWER
GUSSET
TPB-1 645 246 205
TPB-2 645 -389 440 415 278 236 5.9 5.1 0.98 0.95
TPB-3 645 -371 471 421 268 231 5.3 4.9 0.94 0.91
TPB-4 645 -365 467 416 266 236 5.2 4.8 1.02 0.93
TPB-5 645 -362 474 418 270 244 5.3 4.8 1.00 0.92
TPB-6 645 -359 272 4.9
W-1 645 286 249
W-2 645 414 307
TPC-1 645 -457 450 463 316 211 3.9 4.6 1.01 0.95


kN kN kN/mm kN/mm
Est. Estimated
Estimated
Internal Post-
Normal
Pa Stiffness
k1
Force ka
- + AVG - + - + - +
DPA-1 645 -389 495 442 360 190 3.8 5.3 0.95 0.92
DPA-2 645 -522 310 416 370 203 4.0 5.1 0.96 0.92
>THIRTY THREE HYBRID TESTS CONDUCTED BY KAMMULA
DPB-1 645 427 460 5.1
DPB-2 645 -409 430 420 268 4.4 5.4 1.01
DPB-3 645 -507 344 426 298 209 5.3 5.4 1.16 0.96
DPB-4 645 387 205 4.8
DPB-5 645 372 460 5.0
DPB-6 645
DPB-7 645 -416 673 4.9
DPB-8 645 -429 412 421 175 180 4.8 5.1 1.16 1.04
DPB-9 645 -415 670 5.1
DPB-10 645 -434 410 422 196 190 5.0 5.2 1.12 1.05
DPB-11 645 -439 397 418 210 233 5.0 5.0 1.04 0.98
DPC-1 645 -443 410 427 326 260 5.0 5.1 0.96 0.92
DPC-2 645 -439 390 415 229 192 5.5 4.9 1.01 0.92
DPD-1 645 370 194 5.0
DPD-2 645 -276 400 5.3
DPD-3 645 -474 333 404 185 321 5.2 4.9 1.06 0.91
DPD-4 645 312 191 5.0
DPD-5 645 -340 342 5.2
DPD-6 645 -362 519 5.3
DPD-7 645 -371 384 5.4


kN kN kN/mm kN/mm
Est. Estimated
Estimated
Internal Post-
Normal
Pa Stiffness
k1
Force ka
- + AVG - + - + - +
DPD-8 645 -356 500 5.4
DPD-9 480 -348 320 334 332 170 5.4 5.1 0.98 0.88
DPD-10 480 312 178 4.9
DPD-11 480 -231 420 326 308 293 5.3 4.9 1.00 0.92
RECALIBRATE INTERNAL FRICTION DAMPER TO COMPENSATE FOR REMOVED BOLTS
TI-3 ? 373 320 6.1
TI-4 ? -370 368 369 240 560 6.5 6.6 0.97 0.89
TI-5 ? -400 376 388 226 260 5.8 6.3 1.04 1.00
DPE-1 ? -405 401 403 225 235 5.2 4.9 1.17 1.12
DPE-2 ? -438 335 387 200 210 5.3 4.9 1.21 1.12
DPF-1 ? -404 375 390 320 215 5.2 4.8 1.10 1.05
EQR-1 ? -355 348 352 200 190 5.7 6.0 0.93 0.84
EQJ-1 ? -357 370 364 217 231 5.5 4.9 1.10 0.89
DPF-2 ? -377 373 375 234 190 5.4 4.9 1.09 1.00
ATTEMPT TO FAIL ADDITIONAL TENDONS
TB-1 ?
intermediate member and two separate sets of tendons, the axial force calculation had to be modified to
include these extra elements as follows. The force in the intermediate member was calculated using experi-
mental data from the strain gauges on the members and the member cross-sectional area:
P mr = A m E s  m (Eq. 5.1)
where P mr is the axial force in the intermediate member relative to the initial axial force caused by the pre-
tensioned tendons, A m is the cross-sectional area of the intermediate member, E s is the Young’s modulus
of steel, and  m is the average axial strain in the intermediate member measured using the strain gauges.

The tendon force for each set of tendons was calculated by using the experimental data from the LVDTs
that measured the movements of the inner and outer end plates at each end of the brace relative to the
outer member:
  EPi1 +  EPi2 E p A pt
P pir = ------------------------------------------------- (Eq. 5.2)
Lp
  EPo1 +  EPo2 E p A pt
P por = -------------------------------------------------- (Eq. 5.3)
Lp
where P pir and P por are the axial force in the inner and outer tendons, respectively, relative to the initial
axial force caused by the pretension,  EPi1 and  EPi2 are the displacement of the inner end plates relative
to the outer member at ends 1 and 2 of the brace,  EPo1 and  EPo2 are the displacement of the outer end
plates relative to the outer member at ends 1 and 2 of the brace. The sum of  EPi1 and  EPi2 is effectively
equal to the inner tendon elongation  pi and the sum of  EPo1 and  EPo2 is effectively equal to the outer
tendon elongation  po . The parameter E p is the effective Young’s modulus of the aramid tendon, A pt is
the total nominal cross-sectional area of the tendon fibres in each set of tendons, and L p is the effective
length of the tendons. The total axial force applied to the brace, P a , was calculated by summing the axial
forces of the three members and the two sets of tendons:
P a = P ir + P mr + P or + P pir + P por (Eq. 5.4)
This calculation method was moderately successful; however, some problems with LVDT calibrations in
some of the tests affected the quality of the estimate it provided of the T-SCED brace force as will be dis-
cussed in future sections.
The third method that was used to estimate the brace axial force used the strain gauges in the south col-
umn. The data from these strain gauges was averaged to get an estimate of the axial force in the column
and then this force was projected into the brace direction. The only source of friction between the column
and the brace was from the rotation of the beam-column joint, which was low because it was a snug-tight
bolted connection. This brace force estimation method provided the best results in terms of stiffness and
strength; however, there was some movement in the beam column joint when the displacement direction
reversed and this tended to cause a fluctuation in the force in the column due to high frequency impact
vibrations. This effect was worst when the frame was displaced at higher velocities and will be shown and
discussed in future sections. Wherever the T-SCED brace axial force calculation method is not explicitly
identified in figures showing the test results, the south column force calculation method has been used.

5.5.2 BRACE CALIBRATION
The T-SCED brace was first tested without the internal friction damper activated, meaning that the
normal force bolts were installed but left loose. Then, the friction slip force of the internal friction mecha-
nism in the brace was progressively increased in two steps to reach the desired slip force. This whole opera-
tion was performed while the brace was in-place in the test frame so that the hysteretic response of the
brace at different friction force levels could be observed by displacing the test frame a small amount in
either direction. As was done for the HC-SCED brace, the axial force in the friction damper normal force
bolts was applied using a combination of bolt torque and direct bolt length measurements.
The resulting brace hysteresis at the varying levels of friction slip force is shown in Figure 5.12. The
response labelled 'No Internal Fuse Friction' represents the hysteretic response of the brace when the axial
load in the normal force bolts was zero (Test TO-6). This bilinear elastic response represents the behaviour
of the inner, intermediate and outer members with the tendons only. The load at which this bilinear curve
changes stiffness is equivalent to the total pretension force for half of the tendons. This bilinear curve has
some non-zero thickness due to the small inherent friction between the moving parts of the brace mecha-
nism. After half of the normal force bolts were tightened to their design axial force, the brace was tested
again. This is shown in the figure as 'Partial Internal Fuse Friction' (Test TI-1). The added friction wid-
ened the hysteresis around the bilinear response by an amount on each side that is equal to the friction slip
force in the internal friction interfaces, thereby increasing the brace activation force. Then, all of the bolts
were tightened to their final level, increasing the activation load of the brace up to its full target design
level, which is shown as 'Full Design Internal Fuse Friction' in the figure (Test TI-2).
Tests TO-6, TI-1, TI-2

500
400
Full Design Internal Fuse Friction
Partial Internal Fuse Friction
300
No Internal Fuse Friction
200
Axial 100
Force 0
(kN)
-100
-200
-300
-400
-500
-15 -10 -5 0 5 10 15
Figure 5.12: T-SCED Brace Progressive Calibration

5.5.3 T-SCED BRACE TEST PROTOCOLS
The primary test protocol that was used for the T-SCED brace was the same combined ASCE 7/AISC
341 test protocol that was used for the testing of the HC-SCED brace previously described in Section
4.4.8. The main differences for the T-SCED test were that the storey height was different, and that the
brace was assumed to be a second-storey brace in a four-storey-tall braced frame for the determination of
the dynamic periods for each cycle. Based on the models from Choi et al. (2008), this building was esti-
mated to have a 0.6 second fundamental period. The design drift  bm remained equal to the value that
was used for the HC-SCED tests, 1.3%. The complete combined protocol parameters are shown in Table
5.5. In addition to the cycles that were required by the combined protocol, extended qualification protocol
stages were constructed to take the test frame with the T-SCED brace up to the target drift capacity of
3.9%. These additional supplemental cycles are also shown in Table 5.5.
Table 5.5: Combined ASCE/AISC Deformation Protocol

Maximum
Cycles Amplitude Drift Deformation Period
Velocity
2  by 0.037% 1.0 mm 0.6 s 10.5 mm/s
2 0.5 bm 0.65% 17.8 mm 2.3 s 47.8 mm/s
5 1.0 bm 1.30% 35.7 mm 3.1 s 72.6 mm/s
5 1.5 bm 1.95% 53.5 mm 3.6 s 94.6 mm/s
2 2.0 bm 2.60% 71.3 mm 3.9 s 115.6 mm/s

SUPPLEMENTARY CYCLES
2 2.5 bm 3.25% 89.1 mm 4.1 s 135.5 mm/s
1 3.0 bm 3.90% 107.0 mm 4.3 s 155.6 mm/s
In general, the protocol cycles were applied at full speed using sine wave displacement histories at the
appropriate periods given in the table. Each set of cycles was applied independently such that by the end of
the testing program the entire protocol was satisfied. Three tests were conducted where multiple amplitude
cycles were applied in the same test (Tests DPC-1, DPF-1 and DPF-2). These will be described in future
sections.

5.5.4 WIND RESPONSE SIMULATION
Similarly to the HC-SCED testing program as discussed in Section 4.4.4, some low-level dynamic tests
representing the effect of wind loading on the frame were performed before pushing the braced frame to
higher drift levels. Like those previous wind tests, the wind response of the brace was simulated by applying
sinusoidal cycles at two different periods. The first period of 0.6s was the first mode period of the four-sto-
rey prototype building discussed previously in 5.5.3. The second period was the same as the first mode
period of the twelve-storey prototype building used for the HC-SCED tests. These cycles were applied at
an amplitude of  by , with 500 cycles at each period. The hystereses resulting from these tests (Tests W-1
and W-2) are shown in Figure 5.13. The slower 1.25s period cycles were applied first. The deformation
amplitude for these cycles started at the target of 1.0mm in the brace, but that amplitude reduced over 500
cycles. This occurred due to wear in the test frame system which caused the applied deformation to transfer
less efficiently to the brace; however, this only made a difference of approximately half a millimeter in the
positive direction only. By the time the 0.6s period wind simulation was applied, this wear had settled into
a steady-state condition.
400
Tests W-1 & W-2
300
200
100 Test W-1

Brace
Axial 1.25s Period
0
Force
(kN)
-100 Test W-2
0.6s Period
-200
-300
-400
-2.0 -1.5 -1.0 -0.5 0 0.5 1.0 1.5 2.0
Figure 5.13: Wind Simulation Response
A comparison of the brace response before and after the wind simulation tests is shown in Figure 5.14.
This figure shows that there was a moderate increase in the internal friction within the T-SCED brace as a

result of the wind simulation tests; however, Table 5.4 shows that after a few more cycles, from Tests DPA-
1 and DPA-2, the activation force in the brace returned back to the level it was at during Test TPB-3.
600
400
200
Brace
Axial
0
Force
(kN)
-200
Before Wind
(Test TPB-3)
-400
After Wind
(Test TPC-1)
-600
-20 -15 -10 -5 0 5 10 15 20
Figure 5.14: Effect of Wind Simulation on Hysteretic Response
5.5.5 DEFORMATION-LIMITED DYNAMIC PROTOCOL TEST
The lower levels of the dynamic protocol were applied to the frame dynamically to satisfy the AISC/
ASCE protocol requirements up to the level of 2.0 bm (Tests DPB-7 to DPB-11). In addition, a full
dynamic protocol test up to 2.0 bm was performed with one cycle at each deformation level as shown in
Figure 5.15. The resulting hysteresis for this test is shown using both the axial force in the brace calculated
using the axial force from the column and also using the axial force calculated using the brace member
strain gauges and LVDTs. The column force calculation on the left side of the plot shows force vibrations
due to slip and bearing of the beam-to-column connection. This effect is not reflected in the real hysteretic
response of the brace itself and appears only in the column axial force measurement. This phenomenon
will be discussed in more detail in a future section. The post-activation stiffness for the column-axial-force-
derived hysteresis is approximately equal in both the positive and negative directions as expected. The right
hysteresis in the plot, assembled using the brace instrument calculation is included because it does not
show any vibration in the force due to the beam connection movement, proving that the beam connection
slip and impact forces did not have an effect on the behaviour of the brace itself as stated above. The post-

activation stiffness of the brace using this calculation is significantly different in the positive and negative
directions because of a miscalibration of the LDVTs that measured the end plate movements relative to the
outer member. This is clear because the tendon hystereses as shown in the plot at the bottom of the figure
have a different stiffness in the positive and negative directions; this is not possible. The LVDTs were reca-
librated and offset to reduce this error in the future axial force calculations.
Full Protocol to 2.0Δbm (Test DPC-1)

1000
Equal Stiffness Lines (typ.)
800
600
Column
400 Axial
Calculation
Brace 200
Axial
Force 0
(kN) -200
-400 Brace
Tendon/Member
-600 Calculation
-800
-1000
-80 -60 -40 -20 0 20 +/-40 -20 0 20 40 60 80
500
Equal Stiffness Lines (typ.)
400
Outer Tendons
Tendon 300
Force 200
(kN) Inner
100 Tendons
-100
-80 -60 -40 -20 0 20 40 60 80
Figure 5.15: Deformation Limited Test Protocol Results up to 2.0 bm
5.5.6 FULL-DEFORMATION DYNAMIC PROTOCOL TESTS
Following the successful completion of the dynamic protocol up to the 2.0 bm level, the protocol was
continued into the higher supplemental amplitudes up to 3.0 bm as shown in Table 5.5 (Tests DPC-2 to

DPD-8). For these high amplitude tests, the simultaneous control of the two actuators became difficult,
with the difference in the displacement between the two actuators exceeding the prescribed limit. To elim-
inate the problem, following test DPD-7, all displacement histories with amplitude greater than or equal
to 2.5 bm were applied at 70% of the original speed, resulting in longer periods for those tests. For some
of these tests, two of the friction bolts were removed to prevent interference at the end of their slots due to
a fabrication error, slightly reducing the size of the hysteresis. This interference did not affect the comple-
tion of the brace protocol tests and will be discussed in more detail in a future section.
Up to Test DPD-11 all of the cycles from the combined deformation protocol described in Section
5.5.3 and shown in Table 5.5 were complete except that the 3.0 bm level cycle was completed with a
reduced activation load due to removal of those two internal friction damper normal force bolts. To com-
plete that cycles at the full design force at the maximum drift level, the T-SCED brace’s remaining internal
friction damper normal force bolts were recalibrated to achieve the full design force at the maximum drift
level. This was done by increasing the torque in the remaining six bolts until a small-amplitude manually-
controlled test showed that the activation force in the brace was back at the design-level (Tests TI-3 to TI-
5).
The completed full activation force 3.0 bm cycles were applied in Test DPE-2 and the resulting hyster-
etic response is shown in Figure 5.16. This figure compares the detailed response of the real T-SCED brace
to the simulated response of the brace using the SCED Brace Mechanics Simulator previously introduced
in Section 3.4. The inputs for the T-SCED brace mechanical simulation are shown in Table 5.6. The
material properties, element areas, and connection geometry were determined using known brace proper-
ties. The tendon pretension and friction forces were calibrated through comparison to the measured ele-
ment and brace hystereses.
The top axis in Figure 5.16 shows the overall hysteretic response of the brace. The left hysteresis com-
pares the brace axial load calculated using the actuator force to the simulated response. It shows that this
calculation overestimates the response as discussed previously due to the inclusion of friction in the frame
and between the frame and the lateral supports. The middle hysteresis uses the brace axial load that was
calculated using the projected axial load from the south column. As previously mentioned and described in
detail in a future section, this data shows a vibration in the force response due to the slip and impact of the
beam connection to the column. Despite these local effects, this response gives the best overall character-
ization of the envelope of the hysteretic response of the brace. It also compares well to the predicted hyster-
esis determined using the Mechanics Simulator model. Note that the test hysteresis has been shifted

1000
Test DPE-2
800
Axial Force
600 from Actuators
400 Model
Brace Hysteresis
200 Model
Axial Hysteresis
Force 0
Model Axial Force
(kN) Hysteresis
-200 Calculated
-400
using Brace
Member Force
-600 and Tendon
Axial Force Elongation
-800 from
Column
-1000
-120 -60 0 +/-60 0 +/-60 0 60 120
Inner Outer Intermediate
Member Member Member
400
Test DPE-2
200
0
Axial Shifted
Force -200 Test
Hysteresis
(kN) -400
-600 Model
-800
Hysteresis
-120 -60 0 +/-60 0 +/-60 0 60 120
Inner Outer
Tendon Tendon
800
Test DPE-2
600
Axial 400
Force
(kN) 200
-200
-120 -60 0 +60 +/-120 -60 0 60 120
Figure 5.16: Comparison of Test DPE-2 (recalibrated brace) to Mechanics Model Results
downwards in this figure to account for a zero force offset in the strain gauge data in the column. The right
hysteresis uses the axial force as calculated from the brace instruments (LVDTs measuring the end plate
movements and strain gauges on the inner, intermediate and outer members). This hysteresis does not
agree well with the model hysteresis; it under-predicts the width of the flag hysteresis on the tension side
and over-predicts the width on the compression side. This effect was caused by friction induced by the rub-

Table 5.6: Test DPE-2 Simulation Inputs

Parameter Value
Young’s Modulus Steel E s 200000 MPa
Young’s Modulus Tendon E p 102000 MPa
Inner Member Length L i 5030.0 mm
Inner Member Area A i 7560 mm2

Inner Member Initial Position 0 mm
Intermediate Member Length L m 5029.3 mm
Intermediate Member Area A m 9970 mm2

Intermediate Member Initial Position 0 mm
Outer Member Length L o 5029.0 mm
Outer Member Area A o 19584 mm2

Outer Member Initial Position 0 mm
Internal Damper Friction End 1 F I1 85 kN
Internal Damper Friction End 2 F I2 85 kN
End Plate Friction End 1 30 kN

End Plate Friction End 2 30 kN
Tendon Area of Each Set A pt 500 mm
Total Tendon Pretension for Each Set P p0 160 kN
Connection Part 1 Length 480 mm

Connection Part 1 Area 11600 mm2
Connection Part 2 Length 280 mm
Connection Part 2 Area 4516 mm2
Bearing Stiffness 100000 kN/mm
Friction Interface Stiffness 100000 kN/mm
Initial Pretensioning Force Step 0.1 kN
Analysis Deformation Step 0.02 mm
bing together of the inner and outer end plates at each end. This changes the values of the parameters for
the axial force calculation as will be discussed in more detail in a future section; however, as the hysteresis
derived from the axial force in the column shows, this calculation error does not affect the true hysteretic
behaviour of the brace. The effect of this end plate friction was included in the Mechanics Simulator
model.

The tendon hysteresis were well modeled by the Mechanics Simulator model; however, there still seems
to be some nonlinearity in the LDVT data for the outer tendon on the compression side. The test hystere-
ses for the tendons were also shifted because the recorded values are relative to the initial pretension.
The measured effective initial stiffness k 1 eff of the brace that was calculated by finding the secant stiff-
ness between the tension and compression activation points and was approximately equal to 285 kN/mm.
The Mechanics Simulator model predicted a value of 340 kN/mm with a member length difference of
0.7mm between the inner and intermediate members and 0.3mm between the intermediate and outer
members; this value is reasonably close to the measured effective initial stiffness. The values for the mem-
ber lengths were not directly measured, but were chosen such that the shapes of the individual element hys-
teresis matched the test data. Similar to the results for the HC-SCED discussed previously in Section 4.4.5,
these effective initial stiffnesses are much lower than the theoretical initial stiffness for this brace which is
1476 kN/mm. Since this T-SCED brace was fabricated carefully to ensure that the members were the same
length, it is unlikely that better fabrication practices could be followed during mass production. Therefore,
it is clearly important that a designer consider the realistic effective initial stiffness for the determination of
earthquake loads and for drift limit calculations.
In addition to completing the entire test protocol up to 3.0 bm using individual cycles, Test DPF-1
compiled all of the protocol levels from Table 5.5 together in a single dynamic test with a single cycle at
each level. The speed of this test was limited to 115mm/s to avoid actuator control problems. The results
of this test are shown in Figures 5.17 and 5.18. In Figure 5.18, the thick grey line shows the total applied
brace deformation with respect to time. The deformation demand in this test was applied in the form of a
sinusoid with increasing amplitude at each successive cycle. The maximum brace deformation in this test
was 100mm which is equivalent to 3.9% storey drift or three times the design drift. Figure 5.17 shows the
full hysteretic behaviour of the T-SCED brace under the applied sinusoidal deformation. The hysteresis on
the left uses the projected column axial force and the hysteresis on the right uses the calculated hysteresis
from the brace elements. In each case, the other hysteresis is shown behind in grey. The brace hysteresis
from the column axial force is stable and symmetric at all levels of the protocol, but shows the force vibra-
tion due to the beam connection slip. The calculated hysteresis has a smooth response, showing again that
the brace response was not affected by the beam connection slip vibrations; however, the calculated hyster-
esis is unbalanced due to the miscalculation resulting from the friction between the end plates. Overall, this
response to the protocol shows that the T-SCED brace behaves robustly and symmetrically, with full self-
centering capability up to almost 4.0% drift.

Full Protocol to 3.0Δbm (Single Cycles) - Test DPF-1

Brace Axial Deformation (Equivalent % Drift)
-4.0% -2.0% 0 2.0% 4.0%
1000
800
600
400 Axial Force

from Column
Brace 200
Axial
Force 0
True Balanced Unbalanced Hysteresis
(kN) -200 Hysteresis due to Calculation
-400 Axial Force

Calculated
-600 using Brace
Member Force
-800 Force Vibration
and Tendon
Smooth Brace
due to Beam Response Elongation
Connection Slip
-1000
-120 -80 -40 0 +40/-80 +80/-40 0 40 80 120
Figure 5.17: Full Test Protocol Hysteretic Results up to 3.0 bm
Tension Compression
Outer Member P Outer Member P
South Outer South Outer
Outer Tendons North Outer Outer Tendons North Outer
South Inner Intermediate Member South Inner Intermediate Member
Inner Tendons North Inner Inner Tendons North Inner
Inner Member Inner Member
Test DPF-1 North Inner End Plate

100
Brace Tension North Outer End Plate
End Plate moves
Out relative
50 to outer tube
Axial
Deform. /
Gap 0
Opening
(mm)
-50 Brace Compression
End Plate moves
In relative South Outer End Plate
to outer tube South Inner End Plate
-100 Total Brace Deformation
0 5 10 15 20 25
Time (s)
Figure 5.18: Full Test Protocol Tendon Deformation Results

Figure 5.18 depicts the end plate movement response for Test DPF-1 to demonstrate that the telescop-
ing mechanism of the T-SCED brace was functioning as it was designed to. With reference to the brace
mechanics diagram shown at the top of the figure, where the left and right sides of the figure are respec-
tively the south (upper) and north (lower) brace ends in the test: when the brace is in tension, the south
inner end plate moves two units out with respect to the outer member with a magnitude equal to the
applied deformation, and the south outer end plate moves one unit out with respect to the outer member.
This is identical to the behaviour shown in the plot in the lower part of the figure. Meanwhile, the north
inner end plate moves one unit in with respect to the outer member, and the north outer end plate does
not move at all with respect to the outer member. In compression, the situation is reversed north-to-south.
In addition, if the intermediate member is properly floating, equilibrium dictates that the deformation
demand in both sets of tendons should be equal. For this to be true, the two unit movement should be
twice the one unit movement. Figure 5.18 shows this to be true, confirming that the T-SCED brace
behaved as intended. This result was expected because any momentary difference in the tendon forces
between the two sets of tendons will self-equilibrate since the two sets are in series.
5.5.7 DYNAMIC EARTHQUAKE RESPONSE
In addition to the static and dynamic protocol testing, two dynamic earthquake deformation records
were applied to the frame. The hysteretic behaviour of the T-SCED brace subject to these excitations is
shown in Figure 5.19. The first drift history that was used (Test EQR-1) was taken from the dynamic
response of one storey in a twelve storey simulated SCED braced frame from Tremblay et al. (2008). The
excitation earthquake that was used for this model was a simulated Cascadia subduction zone earthquake
that was created for a study performed by Tremblay and Atkinson (2001). This record had a long duration
of 100s and produced a large number of cycles of activation in the brace. The T-SCED brace showed no
degradation in the response and did not sustain any residual drift. For a yielding or friction braced frame
system, this type of earthquake would likely result in a progressive P-Delta mechanism, biasing the
response in one direction and likely resulting in significant residual drift, if not collapse.
The second drift history shown in the figure (Test EQJ-1) is the same as Test SC14-A that was applied
to the HC-SCED as described in Section 4.4.7, which was taken from the dynamic response of the first
storey in a twelve storey simulated braced frame as modelled by Choi et al. (2008). It is a simulated maxi-
mum considered earthquake hazard level (2% probability of exceedence in 50 years) record for Los Ange-
les. The demand from this drift history approaches 3% drift in the brace. Even at this extreme level of

Test EQR-1 Test EQJ-1

Cascadia Subduction Earthquake SAC Project LA-32 Elysian Park 1
Simulated Drift Record Simulated Drift Record
0.8% 3%
2%
Brace 0.4%
1%
Axial
0 0
Def.
-1%
(% Drift) -0.4%
-2%
-0.8% -3%
0 20 40 60 80 0 5 10 15 20 25
Time (s) Time (s)
500 1000
750
250 500
Brace 250
Axial
0 0
Force
(kN) -250
-250 -500
-750
-500 -1000
-1.0% -0.5% 0 0.5% 1.0% -4.0% -2.0% 0 2.0% 4.0%
Brace Axial Deformation (% Drift Equivalent)
Brace Earthquake Response Comparison

SAC Project LA-32 Elysian Park 1
1000
T-SCED HC-SCED
750
Test EQJ-1 Test SC14-A
3000
500
2000
T-SCED 250 No Residual Residual
HC-SCED
Brace Drift 1000 Brace
Drift
Axial 0 0 Axial
Force -1000 Force
-250
(kN) -2000 (kN)
-500
-3000
-750 External Fuse Slip
-1000
-4.0% -2.0% 0 +/-2.0% 0 2.0%
Brace Axial Deformation (% Drift Equivalent)
Figure 5.19: Simulated Earthquake Drift History Hysteretic Results and Comparison to HC-SCED Results
demand, the T-SCED brace fully self-centers the frame, avoiding any residual drift. The results for this
deformation history for both the T-SCED and the HC-SCED are compared in the lower plot of Figure
5.19 with the same horizontal axis scaling but with differently-scaled vertical axes such that the activation
loads of the two braces line up in the plot. This figure shows the fundamental difference between a conven-

tional SCED such as the HC-SCED and the T-SCED. For the high levels of drift expected in high-seismic
areas under maximum considered excitations, the conventional HC-SCED required an external friction
fuse to extend its deformation capacity at the expense of residual drifts at the end of the excitation. The T-
SCED does not have this limitation; it was able to accommodate extreme levels of drift with no residual
drift. One caveat to this is that the T-SCED does not have a force limitation mechanism like the one pro-
vided by the external friction fuse in the HC-SCED. Forces in the T-SCED will continue to increase all the
way up to 4.0% drift. This must be accounted for in the capacity design of the other elements in the sys-
tem; however, as will be seen in Chapter 7, since the T-SCED has two sets of tendons in series, it generally
has a significantly lower post-activation stiffness than a comparable original SCED. This means that a T-
SCED designed for 4.0% drift will have a similar maximum axial load to a comparable original SCED
designed for 2.0% drift.
5.6 LOWER CONNECTION ROTATIONAL STIFFNESS AND RETROFIT

The first T-SCED brace tests that were performed following the brace calibration were quasi-static pro-
tocol tests starting with the smallest cycles from Table 5.5, applied to the frame at a constant velocity of
5.0mm/s in both directions. The 0.5 bm cycles from these tests showed peculiar stiffnesses and a signifi-
cant amount of moment in the columns, beam and brace as measured by the strain gauges. The hysteretic
and moment results for one of these tests (Test TPA-5) are shown in Figure 5.20. In the hysteretic plot
shown on the left side of this figure, it can be seen that the brace hysteresis resulting from the brace force
calculation using the strain gauges and LVDTs attached to the T-SCED brace elements has significantly
different stiffnesses than the hysteresis resulting from the brace force calculation using the south column
axial force. This suggested that there was another source of stiffness in the frame besides the brace itself.
When the moments in the frame elements were calculated using strain gauge data, it was found that there
were significant moments in the north column and in the outer member of the T-SCED brace as shown in
the figure. This indicated that there was moment transfer occurring through the lower gusset connection
between the brace and the north column. Loosening the bolts connecting the brace to the lower gusset did
not significantly reduce this problem.
To correct this problem, the lower gusset connection was replaced by a true pin connection as shown at
the top of Figure 5.20. This pin eliminated the connection’s capability to transfer moment between the
brace and the north column. This retrofit eliminated the stiffness differences as shown by the hysteretic

Original Gusset
Large Vertical Separation
North South
Retrofit Pin
Original Gusset (Test TPA-5) Retrofit Pin Connection (Test TPB-2)

600 600
Differing Similar
Stiffness Stiffness
400 400
Brace Brace
200 Instrument 200 Instrument
Brace Calculation Calculation
Axial
Force 0 0
(kN) South South
-200 Column -200 Column
Calculation Calculation
-400 -400
-600 -600
-25 -20 -15 -10 -5 0 5 10 15 20 25 -25 -20 -15 -10 -5 0 5 10 15 20 25
Brace Axial Deformation (mm) Brace Axial Deformation (mm)
100
Original Gusset (Test TPA-5)
50 North Column Bottom
Section
Moment 0
(kNm)
-50 T-SCED Outer Member
-100
100
Retrofit Pin Connection (Test TPB-2)
50 North Column Bottom
Section
Moment 0
(kNm)
-50 T-SCED Outer Member
-100
0 2 4 6 8 10 12 14 16 18
Time (s)
Figure 5.20: Effect of Lower Gusset Connection Retrofit

response of a test that was performed after the retrofit (Test TPB-2) on the right side of the figure. It also
drastically reduced the moment in those sections as shown.
The moment transfer into the brace through the lower gusset connection was significant for the proto-
type T-SCED brace because there was a large vertical spacing between the rows of bolts in the connection
as shown in Figure 5.20. There was no significant moment transfer through the more compact upper gus-
set connection. To avoid requiring a true pin connection in future designs, the lower gusset connection
could be improved by rotating the entire connection by ninety degrees about the parallel axis of the brace.
This would reduce the moment stiffness of the connection and thereby reduce the moment transfer into
the brace.
5.7 MOVING PART INTERFERENCE ISSUES AND RETROFITS

The primary challenges that were encountered during the T-SCED brace testing program related to
interferences between the moving parts in the system. Although allowances were made during the design of
the brace to account for this movement, it was clear from the results of the tests that future designs should
be improved in this regard. In particular, the out-of-plane movement of the inner and outer end plates
should be better controlled by the use of added and improved guides and more plastic pads to fill the gaps.
In addition, construction quality and dimension tolerances need to be more strictly enforced in the future
to ensure that internal friction damper bolts do not prematurely interfere with the ends of their slots. In
particular, the locations of the friction interface angles with respect to the friction interface plates requires
special attention during fabrication. This is unlikely to be a problem in practice. These and other issues
related to moving parts in the brace and frame will be discussed in sections below.
5.7.1 THE EFFECT OF BEAM CONNECTION SLIP
When dynamic protocol tests were first conducted at higher deformation amplitudes, the higher veloc-
ities of these dynamic tests caused the brace axial force that was calculated using the column axial force to
show a significant vibration on the compression side of the hysteresis. This effect is shown in Figure 5.21.
This vibration occurred due to impact forces experienced in the south-end beam connection. A schematic
of this connection is shown in the figure. When the brace was in compression, the resultant force from the
beam on the connection is upwards, overcoming the self-weight of the beam and brace and lifting the
beam upwards. This lift caused the bolts to impact the slots in connection, in turn causing the resulting
vertical vibration in the column which manifested as a high-frequency force variation. A similar effect

occurred when the compression force came back towards zero: the applied force on the connection no lon-
ger overcame the self-weight and the beam sat back down. This effect was not observed until quite late in
the testing program and so was not repaired; however, as the test results have shown, this effect did not
have a negative impact on the behaviour of the brace itself. Cyclic loading on this connection caused the
slotted holes to elongate vertically so that by the end of the testing program the connection appeared as
1000
Effect of End Plate Slot Jam
800
600
400
Brace 200 End Plate Slot Fixed
Test DPB-8 Beam Connection Slip
Axial
Force 0 Snug-tight Bolts
25mm (not shown) Column
(kN) Force Vibration due to
-200 Beam Connection Slip Beam
(beam sits back down)
Slotted Beam
-400 Hole 27mm
Force Vibration due to
-600 Beam Connection Slip Connection
(overcame the self-weight) Angle Hole
27mm
-800 End Plate Slot Jammed
Stiffness Increase in Compression
Test DPB-3
-1000 Upper
-50 -40 -30 -20 -10 0 10 20 30 40 50 Gusset Connection
Figure 5.21: Effect of Beam Connection Slip and Internal Friction Fuse Plate Jam on the Outer End Plate
5.7.2 END PLATE SLOT INTERFERENCE AND INSTALLATION OF END PLATE GUIDES
Figure 5.21 also shows that the T-SCED brace had an approximately 40% higher post-activation stiff-
ness in compression than it did in tension (approximately 6.9 kN/mm in compression versus approxi-
mately 5.0kN/mm in tension). This increased compression stiffness was due to one of the internal friction
fuse interface plates getting caught on the edge of the slot in the outer end plate at the north end of the
brace. The location of this jam is shown in Figure 5.23. The fuse interface plate edge was creating a small
gouge on the inner surface of the end plate slot, preventing that interface plate from properly moving
within the slot.
To fix this problem, the inner surface of the end plate slot was ground down to remove the burr and
make the edge of the slot smooth again. In addition, the reason that the friction damper interface plate was

Figure 5.22: Elongated Slotted Holes in Beam South Connection
grinding against the edge of the end plate slot in the first place was that over many consecutive cycles of
brace loading, the end plate had migrated across the face of the outer member. The end plate was not pre-
viously restrained from moving out-of-plane with respect to the frame. To pull the end plates back to their
original position and to prevent them from moving again, end plate guides were installed on both outer
end plates as shown in Figure 5.24. These guides were welded to the outer end plates at either end of the
brace. They extended from the end plates to the face of the outer member where steel bolts were used to
push a plastic-surfaced bearing plate. This bearing plate provided a reaction surface to push against in order
to move the end plate.
The repairs to the end plate and installation of the plate guides eliminated the increased stiffness
response as shown in Figure 5.21 which shows the hysteretic response of Test DPB-8 which was performed
after the repairs. One unintended consequence of this repair was that after the outer end plates were moved
back into their original places by the plate guides, they were pushed into contact with the inner end plates.
This produced an additional friction between the two end plates which, while it did not prevent the proper
functioning of the T-SCED brace, produced a visible effect on the hysteretic response of the individual ele-
ment in later tests as will be discussed in a future section.
To prevent the end plate slot problems from occurring in future designs, the slot could be eliminated by
stopping the internal friction interface plate short of the ends of the main longitudinal members. The end

Figure 5.23: Location of Outer End Plate Gouged Edge
plate friction could be eliminated by providing better guidance for the movement of the end plates in
future designs.
5.7.3 INTERNAL FRICTION DAMPER BOLT BEARING AT HIGH DRIFTS
When the 2.75 bm drift level was attempted in tension (Test DPD-4), a large increase of stiffness was
observed before the maximum amplitude was reached as shown at the top of Figure 5.25. This stiffness
increase occurred because the normal force bolts in the south end internal friction dampers were impacting
the ends of the slots in the friction damper interface plate as illustrated at the bottom of Figure 5.25. It is
suspected that this collision occurred because the friction damper interface angles at the south end of the
brace were mislocated towards the north end of the brace during fabrication.

Figure 5.24: End Plate Guides
To avoid this internal friction damper slot collision, the two normal force bolts that were colliding with
the slot ends in the south friction dampers were removed from the brace. This reduced the internal friction
damper slip force, but allowed all of the tension-direction protocol cycles up to 3.0 bm to be completed.
The friction interfaces were then recalibrated by increasing the normal force in the remaining bolts. This
problem may be easily avoided in the future by giving special attention to the relative locations of the fric-
tion interface angles and the friction interface plate during fabrication.
5.7.4 THE EFFECT OF FRICTION BETWEEN THE INNER AND OUTER END PLATES
As previously mentioned in Section 5.5.6, after the installation of the end plate guides, an error was
observed in the axial force calculation that used the brace’s individual member forces and tendon elonga-
tions. This error resulted in hysteretic plots where the width of the hysteresis was under-predicted on the

Internal Friction Damper

Normal Force Bolt Bearing (Test DPD-4)
1000
Recovers Initial Stiffness

800
Bolt Bearing at End of Slot
600
Brace
Axial 400
Force
(kN)
200
-200
-20 0 20 40 60 80 100
Beam
Interior Friction Upper Gusset

Damper
Interface Plate
Interior Friction
Damper
Interface Angle
Inner Member
Connection
Inner Member Plate
(within)
Slotted Hole in Plate
Friction Pads in Angle
When brace is in tension, End of slot collides with

interface plate moves to the Normal Force Bolt
right within the interface angles
Figure 5.25: Bolt Bearing in Internal Friction Damper

tension side and over-predicted on the compression side. To illustrate the cause of this error, the hysteretic
response of the brace and the brace elements during Test DPD-11 along with the output from a Mechanics
Simulator model of the test is shown in Figure 5.26. This test was a full amplitude 3.0 bm dynamic test
that was performed after the installation of the end plate guides, but before the internal friction damper
force was recalibrated to account for the two removed bolts. The inputs to the mechanics simulator for this
test were the same as those used for Test DPE-2 as shown in Table 5.6 except that the internal fuse friction
at End 1 was reduced to 25 kN to account for the two removed normal force bolts.The relative values of
the internal damper friction at each end of the T-SCED brace were determined by using the relative values
of the flag widths of the observed hystereses of the inner and outer members.
The resulting comparison for this test is similar to the previous one discussed in Section 5.5.6 and
shown in Figure 5.16: the hysteresis using the projected actuator force overestimates the actual hysteresis
because of frame friction; the hysteresis using the projected column axial force provides a good measure of
the true brace response; and the calculated hysteresis is skinny on the tension side and fat on the compres-
sion side. As mentioned previously, this type of result in the calculated hystereses only occurred in tests that
came after the recentering of the end plates and the installation of the end plate guides as previously
described in the Section 5.7.2. The overall response predicted by this calculation cannot be correct because
the hysteresis that was calculated using the axial force in the column does not show any similar strange
behaviour. Therefore, it must have been caused by the calculation method.
The calculated axial force in the brace was found by summing the axial forces in all five individual axial
elements: the inner, intermediate and outer members, and the inner and outer tendons. The axial forces in
all of these elements are shown in the lower plots in Figure 5.26 for both Test DPD-11 and for the model
hysteresis determined using the Mechanics Simulator. In a perfect brace, the inner and outer member hys-
tereses should have the same flag widths on the tension and compression sides; however, for this test it can
be seen that the tension side has a smaller flag width than the compression side for both members. This
indicates that there is some other friction elsewhere in the system. This friction came from contact between
the inner and outer end plates at each end of the brace. It was not present in the system before the end
plate position was adjusted and the end plate guides were installed. It was caused by the end plate guides,
which pulled the outer end plate into contact with the inner end plate. When this end plate friction, which
was determined to be approximately 30kN at each end, was added to the Mechanics Simulator model, the
model response for the inner and outer members matched the test data well as shown in the figure. Note
that the model hystereses are shifted in these plots as well since the model values are absolute relative to

1000
Test DPD-11 Axial Force
800 from
Axial Force Column
600 from Actuators
400 Model
Brace Hysteresis Model Model
200
Axial Hysteresis Hysteresis
Force 0
(kN)
-200
-400
-600 Axial Force Calculated

using Brace Member
-800 Force and Tendon
Force Vibration due to Beam Connection Slip Elongation
-1000
-120 -60 0 +/-60 0 +/-60 0 60 120
Inner Outer Intermediate
Member Member Member
400
Test DPD-11
200
0
Axial Shifted
-200 Test
Force Hysteresis
(kN) -400
-600
Model
-800 Hysteresis
-120 -60 0 +/-60 0 +/-60 0 60 120
Inner Outer
Tendon Tendon
800
Test DPD-11
600
Axial 400
Force
(kN) 200
-200
-120 -60 0 +60 +/-120 -60 0 60 120
Figure 5.26: Comparison of Test DPD-11 (with interfering bolts removed) to Mechanics Model Results
zero force, whereas the recorded test numbers are relative to the initial precompression in the members pro-
vided by the tendons.
When the brace was disassembled after all of the tests were complete, damage was observed on the end
plate edges, confirming that they were indeed in contact during the testing as shown in Figure 5.27. The
friction between the end plates should have caused the intermediate member hysteresis to have some width

as shown by the model hysteresis in the figure; however, this was not captured by the instrumentation,
which resulted in an error in the total force calculation.
Figure 5.27: End Plate Edge Damage due to Friction Between Inner and Outer End Plates
5.8 HYBRID TESTING PROGRAM (KAMMULA, 2012)

Since the T-SCED brace is essentially a damage-less system up to significant levels of frame drift, the
system may be tested many times without concern for failure or collapse. This makes it ideal for the cali-
bration and testing of large-scale hybrid modelling techniques. Therefore, in parallel with the prototype T-
SCED frame tests, a program consisting of forty-five total hybrid tests was conducted by Kummula
(2012). These tests were interspersed throughout the main testing program as shown in Table 5.3.
5.9 TENDON ANCHOR FAILURE DUE TO REPEATED LOADING CYCLES

After all of the goals of the testing program were complete, including the completion of the full test
protocol and the simulated earthquake drift histories, the full protocol loading from Test DPF-1 was
repeated (previously described in Section 5.5.6). During this test (Test DPF-2), which had the hysteretic
behaviour shown in Figure 5.28, one of the tendons anchors failed. This anchor did not fail the same way
as the anchors that failed during the HC-SCED tests, i.e. by losing the spike and releasing the tendon. The
anchor in this test failed by net section fracture through the cross-section of the threaded part of the anchor
barrel as shown in the photos in Figure 5.29. The point of fracture was located just inside of the location of
the anchor nut. It seems like there may have been a nick in the threads of the anchor, resulting in the initi-

ation of a crack that eventually propagated through the entire cross-section of the anchor after repeated
cyclic loading.
Full Protocol with

Tendon Anchor Failure
1000
Test DPF-2
800
600
400
Brace 200
Axial
0
Force
(kN) -200
-400
-600
-800
Tendon Anchor Failure
-1000
-100 -50 0 50 100
Figure 5.28: Full Protocol Results with Tendon Anchor Rupture
It is important to note that this seemingly fatigue-related anchor failure occurred after twenty-seven
quasi-static tests, forty-five hybrid simulation tests, and twenty-eight dynamic tests, or one hundred tests
in total. These are not loading conditions that any type of brace would experience during the service life of
a structure. If desired, the fatigue life of these anchors could be significantly increased by making them out
of a more ductile material like steel instead of aluminum.
After the failure of the tendon anchor, the tendon that was attached to that anchor was stuck in the
brace. This made it impossible to continue testing with that set of tendons. To see if any of the other ten-
dons were also vulnerable, the T-SCED brace was manually and slowly deformed in compression up to the
maximum compression deformation (Test TB-1). No other tendon anchors failed during this process.
When the load was released following this test, the end plates did not fully close onto the body of the
brace. The inner end plate at the south end of the brace where the anchor failed was found to be rotated
and stuck within the intermediate member. This happened because when the tendon anchor broke, the
inner end plate became unseated from top of the intermediate member at the top (see Figure 5.2 for refer-
ence). The plate rotated inwards at the top until the inner edge rested on the face of the inner member
(which was further inside at the time because the brace was in compression). Due to the high forces pro-

Figure 5.29: Photos of Tendon Anchor Rupture
vided by the remaining tendons, this situation could not be corrected and the testing program was
stopped.
5.10 CONCLUSIONS
The telescoping self-centering energy-dissipative (T-SCED) brace extends the self-centering deforma-
tion range of the original SCED brace by allowing it to deform twice as far as previous designs. The T-
SCED brace prototype was capable of accommodating 3.9% drift in the test frame, applied dynamically.
This is equivalent to over 100mm of axial brace deformation. This level of drift capacity is enough to resist
extreme earthquakes in high-seismic zones without the need for an external friction fuse to extend the
deformation capacity of the brace.

The brace easily satisfied both the ASCE and AISC test protocols for buckling-restrained braces,
accommodating storey drift up to the level of 2.0 bm at full speed. It then went further, accommodating
drifts up to 3.0 bm while providing full self-centering capability.
Based on the results of these tests some minor modifications should be made to the brace design. Partic-
ularly, increased guidance of the end plates and the floating intermediate section should be provided in
future designs to increase the robustness of the mechanism. In addition, a redesign of the outer member
brace connection by rotating the axis of the brace by 90 degrees would potentially avoid the need for a true
pin connection at that end of the brace.

CHAPTER 6: SHAKE TABLE TESTING OF A SCED BRACED
FRAME
In parallel with the study that was conducted to increase the axial force capacity of SCED braces, a
shake table study of a one-third scale, three-storey SCED-braced frame was conducted in the structural
laboratory at École Polytechnique in Montréal. The two primary goals of the study were (1) to evaluate the
dynamic response of a multi-storey SCED-braced frame, subjected to multiple types of ground motions
from different seismic environments, and (2) to assess the ability of computer models of differing complex-
ity, including commercial software that is widely used in practice, to accurately predict the seismic response
of a multi-storey SCED-braced frame. Additionally, a previous numerical study found that SCED-braced
frames may experience high peak floor accelerations when experiencing high-velocity stiffness transitions
and it was proposed in that study that this effect is caused by sharp transitions between the elastic and post-
activation stiffnesses in the brace hysteresis model (Tremblay et al., 2008). The acceleration response of the
real SCED-braced frame was carefully examined to further investigate the cause of these high accelerations
in the models.
6.1 DESIGN STRUCTURE

The prototype structure for the shake table test was designed by Dr. Hyungjoon Kim from the Univer-
sity of Seoul (Kim, 2009a). Kim’s report is attached to this thesis in Appendix F. His design results are
summarized below.
6.1.1 LOCATION AND PROTOTYPE DESIGN USING ASCE 7-05 (KIM, 2009A)
The prototype building is a three-storey steel office building located on a Class D site in Los Angeles,
California. The building was designed and detailed according to ASCE 7-05: Minimum Design Loads for
Buildings and Other Structures (ASCE, 2005), AISC 360-05: Load and Resistance Factor Design Specifi-
224
Chapter 6: Shake Table Testing of a SCED Braced Frame 225
cation for Structural Steel Buildings (AISC, 2005a), and AISC 341-05: Seismic Provisions for Structural
Steel Buildings (AISC, 2005b).
The building has three bays in the N-S direction totalling 27.43m in length and six bays in the E-W
direction totalling 36.58m in length (Figure 6.1). For the lateral load-resisting system, it uses four single-
bay perimeter SCED-braced frames in the N-S direction and two four-bay perimeter moment-resisting
frames in the E-W direction. The total seismic weight for the building is 12732kN and the design ground
motion parameters for the site were: S DS = 1.4 for short periods and S D1 = 0.8 for a 1.0s period. Addi-
tional design loads and seismic data are shown in Figure 6.1.
PLAN
N 6@6.096m = 36.58m
8.53m
Studied Gravity loads:
Roof: Dead = 4.07 kPa
SCED- Live = 0.96 kPa
Braced Floor: Dead = 4.31 kPa
Frame 10.36m Live = 2.4 kPa
Exterior walls = 1.5 kPa
Area
Tributary Seismic Load Data:
to Studied 8.53m Los Angeles, CA
Frame SS = 2.1 g; S1 = 0.80 g
Occupancy Category II
Seismic Use Group I
Gravity Moment- Importance Factor = 1.0
Column SCED FRAME ELEVATION Resisting Site Class D
(typ.) Frame Fa = 1.0; Fv = 1.5
27.43m
(typ.) SDS = 1.4 g; SD1 = 0.80 g
Seismic Design Category E
Steel:
3@3.96m Fy = 345 MPa
=11.89m
24.9˚
Studied Frame SCED Braces (typ.)
Figure 6.1: Prototype Building Geometry (after Kim, 2009a)
Since previous studies have shown that SCED-braced frames have seismic drift and force demands sim-
ilar to buckling-restrained braces (BRBs) (Tremblay et al., 2008; Christopoulos et al., 2002a; Choi et al.,
2009), the SCED-braced frames were designed using the same seismic design parameters that are used for
buckling-restrained braces in ASCE 7-05 (ASCE, 2005): response modification coefficient R = 7.0 ,
deflection amplification factor C d = 5.5 , and overstrength factor  0 = 2.0 . The moment-resisting

frames were designed using the seismic design parameters for special steel moment-resisting frames:
R = 8.0 , C d = 5.5 , and  0 = 3.0 .
Both of these lateral force-resisting systems were designed using the Equivalent Lateral Force Procedure
in ASCE 7. For the SCED-braced frame direction, the design period T was taken as 0.44 s ( T = C u T a ,
0.75
with C u = 1.4 and T a = 0.0488  11.887 = 0.31 s ). This resulted in a seismic response coefficient
C s = 0.2 and a corresponding seismic base shear of C s W = 2546 kN . In-plane torsion was omitted for
simplicity. Therefore, the base shear for a single braced frame was then equal to 2546  4 = 637 kN .
The resulting SCED brace properties for each storey of the studied frame are shown in Table 6.1. In
this table, the target SCED values were determined by projecting the storey shear force into the diagonal
direction and then dividing this value by a resistance factor  = 0.9 . The selected SCED properties
shown in the table (the right-most four columns) are the result of full-scale, realistic SCED braces that
were designed to meet the target values. At one end, each SCED brace is designed to connect to the frame
members using a friction bolted connection which acts as an external friction fuse. These external friction
fuses are designed to activate at a storey drift angle of 2% and allow the frame to accommodate storey drifts
up to 4% without structural damage. The resulting section properties for the other members in the SCED
braced frame, moment-resisting frame and gravity frame of the prototype building are shown in Table 6.2.
Table 6.1: Prototype SCED Frame Properties (after Kim, 2009a)

kN kN kN/mm kN kN
Target SCED
Storey Initial Post-Activation SCED External
Storey SCED Activation
Shear Stiffness Stiffness Ratio Fuse Activation
Force Force
V xi PT k1 r Pa F Et
3 309 379 128 0.024 373 589
2 527 646 188 0.028 654 1011
1 637 780 235 0.026 841 1269
6.1.2 SCALING OF PROTOTYPE FOR DESIGN OF TEST STRUCTURE (KIM, 2009A)
Based on the prototype building design, one of the SCED-braced frames in the building was scaled
down to one-third scale in order to satisfy the physical limitations of the shaking table at École Polytech-
nique. All of the length measurements were scaled by 1/3 and the elastic modulus scaling factor was kept at

Table 6.2: Prototype Building Section Properties (after Kim, 2009a)
mm mm mm mm mm2 106xmm4
Location Section Depth Width Flange Web Area 2nd Mom. Area
d b tf tw A Ix , Iy
SCED Braced Frame
Ext. Col. W200x52 206 204 12.6 7.9 6650 52.7, 17.8
Int. Col. W200x71 216 206 17.4 10.2 9100 76.6, 25.4
Beam W200x59 (2nd, 3rd Fl.) 210 205 14.2 9.1 7560 61.1, 20.4
Beam W200x52 (Roof ) 206 204 12.6 7.9 6650 52.7, 17.8
Special Moment-Resisting Frame
Columns W460x193 489 283 30.5 17.0 24600 1020, 115
Beam nd 612 227 19.6 11.9 15900 985, 39.3
W610x125 (2 Fl.)
Beam W610x101 (3nd Fl.) 603 228 14.9 10.5 13000 764, 29.5
Beam W530x74 (Roof ) 529 166 13.6 9.7 9520 411, 10.4
GravityLoad-Carrying System
Columns W200x52 206 204 12.6 7.9 6650 52.7, 17.8
Beam (N-S) W410x46 (2nd, 3rd Fl.) 403 140 11.2 7.0 5890 156, 5.14
Beam (N-S) W360x39 (Roof ) 353 128 10.7 6.5 4980 102, 3.75
Beam (E-W) nd rd 450 152 10.8 7.6 6630 212, 6.34
W460x52 (2 , 3 Fl.)
Beam (E-W) W410x39 (Roof ) 399 140 8.8 6.4 4990 127, 4.04
one since steel stiffness is the same regardless of size. The resulting scaling factors for all physical quantities
are shown in Table 6.3.
Note that this particular scaling regime resulted in a time compression, but left the accelerations
unscaled. The scaling factor for forces of 1/9 resulted in a reduced total building seismic weight W of 1415
kN which is equivalent to 354 kN for a single SCED braced frame. Using the same proportion, the base
shear for the test braced frame was also reduced to 0.2W = 70.7 kN , which resulted in the new scaled-
down target SCED design forces shown in Table 6.4. Due to limited availability of structural shapes and
tendons for the SCED braces, only the first-storey SCED initial stiffness requirements could be closely sat-
isfied in the test frame. Stiffness properties for the other two braces exceeded the required values, which led
to a relatively stiffer test frame compared to the prototype frame. The external friction fuse activation
loads, P E , were adjusted accordingly such that activation of the fuses in the test frame occurred at 2% sto-
rey drift as intended.

Table 6.3: Scaling Factors for the Test Structure (after Kim, 2009a)
Quantity Description Scale Factor
Length SL 0.3333
Elastic Modulus SE 1.0000

2
Force SF = SE SL 0.1111
Mass SM 0.1113
Acceleration Sa = SF  SM 0.9983  1.0000
Time ST = SM SL  SF 0.5774
2
Area SA = SL 0.1111
3
Section Modulus SZ = SL 0.0370
4
Moment of Inertia SI = SL 0.0123
Table 6.4: Test Frame SCED Properties (after Kim, 2009a)

kN kN kN/mm kN kN
Target SCED
Storey Initial Post-Activation SCED External
Storey SCED Activation
Shear Stiffness Stiffness Ratio Fuse Activation
Force Force
V xi PT k1 r Pa F Et
3 35.4 44 78.8 0.043 44 123
2 58.7 72 78.8 0.043 72 150
1 70.7 87 78.8 0.043 87 164
The weight of one-quarter of the scaled-down building (354 kN) was simulated in the tests using a
mass simulation system that was connected in series with the braced frame (which will be discussed in fur-
ther detail in Section 6.3). The columns in the test frame were designed to simulate the approximate com-
bined flexural stiffness of all building columns that would be tributary to the braced frame studied. This
was done to reproduce the stiffening effect of columns being continuous over the entire building height.
This required that each column of the braced frame have a stiffness equivalent to one eighth of the total
flexural stiffness of all building columns. This equivalent stiffness included the flexural stiffness of the
braced frame columns in the strong direction, the moment frame columns in the weak direction and the
gravity columns (see the tributary area shown in Figure 6.1). This gave a target moment of inertia of
6 4
I x req = 2.85  10 mm after applying the appropriate scaling factor. The selected column section was an

6 4
HSS152x76x8.0 oriented in the weak axis which had a moment of inertia of I y sel = 2.91  10 mm .
The test frame beams were selected during the test frame design (which will be discussed in Section 6.3).
6.2 SCED BRACE DESIGN (KIM, 2009A)

The small scale SCED braces that were used for the shake table tests were designed by Kim (2009a) and
were manufactured by Dongil Rubber Belt (DRB) in Korea. The same SCED brace design geometry was
used to achieve the target design parameters for all three storeys (as given in Table 6.4). The detailed SCED
design drawings may be found in Appendix G and a schematic of the brace specimen is shown in Figure
6.2. These SCED braces that were designed and fabricated for the test frame utilized a single tendon to
provide the restoring force and four single-bolt friction dampers to provide the energy dissipation. The
outer member of each brace was an HSS100x100x4.0 steel tube and the inner member was an
HSS75x75x4.5 steel tube. These are standard Korean tube sizes that are not available in North America.
The tendon was made of a straight-fibre, 17mm nominal diameter Technora aramid rope with aluminum
spike-and-barrel anchors. The internal friction damper interfaces consisted of stainless steel on one side
and Carlisle Industrial Brake and Friction NF-916 non-asbestos organic (NAO) friction pads on the other.
The normal force for the friction interfaces was provided by ½” steel bolts. An external friction fuse was
included for each brace which used a similar stainless steel on NAO friction pad interface using a similar
NF-336 friction material and larger ¾” steel bolts. The different activation forces for each storey of the test
frame in Table 6.4 were achieved by using different tendon pretension forces and friction interface slip
forces. The slip force in the friction interfaces was adjusted by varying the normal force on the interface
that is provided by the steel bolts.
6.3 TEST FRAME DESIGN AND SETUP

Using the initial prototype design and the scaling factors from Table 6.3 as the starting point, the test
frame was designed for the shake table test. The test frame was designed to be mounted to the single degree
of freedom shake table at École Polytechnique in Montréal. A summary of the test setup is shown in Fig-
ures 6.3 and 6.4. Full fabrication drawings of the SCED braces and the test frame and may be found in
Appendices G and H, respectively.
The multi-storey SCED-braced frame was attached to the shake table via true pin connections at the
base of each column. The columns in the frame were designed as described in the previous section. The
HSS beams and beam connections were designed to have a high axial stiffness and a low rotational stiffness

2.6m
ly
External Plan se mb
Friction Fuse As
te
Gusset Connection Gusset Connection m ple
B A Internal
Friction Dampers Co bly
s sem
erA
e mb
End Plate
Elevation
te rM ly
Tendon Anchor
Ou semb
Friction Friction Interface As
Plate (welded to er
Interface
Inner Section)
External
e mb
Friction Fuse
rM es
Inner Member Interface e lat
Outer Member Inn nd
P
Tendon & E
b ly
sem
Washer Plates
As
Friction Gusset Connection/ on
Interface nd
Angle
Friction Angle Te
Section A Section B
Internal Friction Damper External Friction Fuse
Figure 6.2: Small-scale SCED brace for Shake Table Test
Mass Simulation
Frame Connection Mass Simulation
Frame
Beam Assembly
Lateral Restraint
(connected to larger Columns with
frame - not shown) Rockers at
Both Ends
Steel Test Frame
Beam Connection
4.0m Horizontal Steel

Base Frame
SCED Brace Low Friction

Roller Bearings
Pinned Column Base
Base Frame
Connection
Base Connection Beam 2.8m

Shake Table Surface
Figure 6.3: Test Structure Overview
at their ends. This simulates a shear tab connection which typically allows for some rotation at the ends of
beams in a braced frame. The gusset plates were connected to the HSS column with a through plate con-

Figure 6.4: Test Structure Photographs
nection. The lateral support for the entire test was provided by a steel lateral support frame that surrounds
the shake table, which is not shown in Figure 6.3 for clarity, but is visible in Figure 6.4. The lateral restraint
supports shown in Figure 6.3 served as an interface between the larger lateral support frame and the test
frame, ensuring that the test frame could not move out-of-plane. Teflon pads were used to accommodate
sliding of the test frame within the lateral restraint supports while limiting the friction force that could
develop between the test specimen and the lateral supports.
The full required seismic mass could not be accommodated on the shaking table itself due to payload
limitations, therefore, a mass simulation frame was erected in series with the test frame beside the shake
table. The design of this mass simulation frame that is used for shake table tests at Ecole Polytechnique is
described by Tremblay et al. (2009). The mass simulation frame consisted of steel plates that simulated the
horizontal mass and gravity load of one-quarter of the scaled-down prototype building. These plates were
supported between the storey levels by steel HSS columns with friction-free rockers at their upper and
lower ends. At the base of the mass simulation frame, the columns were mounted on a horizontal steel
frame that sat on low-friction roller bearings. That base frame was horizontally connected to steel members
that were attached directly to the shake table surface, forcing the base of the mass simulation frame to move
with the base of the test frame. Using this arrangement, the mass simulation frame was able to impose con-

sistent P-Delta lateral forces to the test frame. The mass simulation frame was connected to the test frame
at every storey using stiff HSS beams with end connections that were designed with a tight-fit clevis and
pretensioned bolts to prevent slackness and horizontal impact forces.
The full instrumentation plan for the shake table test is summarized in Figure 6.5. To measure axial
force, moment and shear in the test frame columns, strain gauges were affixed to both sides of the left col-
umn at the top and bottom of all three storeys and to both sides of the right column at the top and bottom
of the first storey only. Each SCED brace had four strain gauges mounted on each tube section (as will be
shown on Figure 6.7), but only two gauges on each tube were recorded during the shake table test due to a
limited number of available instrumentation channels; this gave a total of four recorded strain gauges on
each brace. The brace axial deformations were measured using three instruments on each brace, one string
potentiometer to measure the total end-to-end deformation and two linear potentiometers to measure the
internal friction damper deformation (equivalent to the relative movement of the inner and outer mem-
bers) and exterior friction fuse deformation. String potentiometers were used to measure the absolute lat-
eral deflections of each storey and of the table itself. Lateral accelerations were measured using
accelerometers that were located at every level of the SCED braced frame, on the shake table surface and at
every level of the mass simulation system. Each of the mass simulation frame’s connection beams was
installed in series with a 222 kN load cell to directly measure the applied horizontal inertia forces and,
thereby, directly measure storey shears. Two additional load cells were located between the base of the mass
simulation frame and the shake table to measure the friction between the base of the mass simulation frame
and the ground and the inertia of the base of the mass simulation frame.
6.4 PRELIMINARY BRACE TESTS

Four identical SCED braces were manufactured by Dongil Rubber Belt (DRB) of Korea. One brace
was calibrated in Korea for each of the three storeys of the test structure (for values see Table 6.4). These
will be referred to as the first, second and third storey braces. In the interest of redundancy, an extra brace
was also fabricated in case there was a problem with one of the three calibrated braces.
Since SCED braces are not generally damaged as a result of testing, the braces for the shake table test
were pre-tested and calibrated before they were erected in the shake table test frame. This testing and cali-
bration was done using the 12MN load frame at École Polytechnique. The tests showed that the behaviour
of the SCED braces did not meet the specified load capacities and that two out of the four braces had fab-
rication defects that needed to be repaired.

Column
Strain
Gauge
Locations Instrumentation Locations
(left column shown,
right column gauges String Pots @ Uniaxial Strain Gauges on SCED Braces
only on first storey) Every Storey and Table 4 on inner member, 4 on outer member
Fixed to Outside Support For shake table test, used only
38 (4 total) 4 for each tube (2 inner, 2 outer) (12 total)
38
a External Fuse a
279 381
Linear Pot - Every Storey
(3 total)
102 Tube Relative Movement
Linear Pot - Every Storey
381 (3 total)
102 381 a a
Total SCED
Deformation String Pot
Horizontal Every Storey
Accelerometer (3 total)
279 381 (typ.)
381 a a
279
222kN Load Cell
Every Storey
102
381
a
Locations of Uniaxial Vertical Strain Gauges 2x222kN Load Cell

(both sides of column) Mass Simulator Base
(24 total on left column to Shake Table
8 total on right column)
Figure 6.5: Test Structure Instrumentation
6.4.1 TEST SETUP
The complete test setup for the SCED brace axial load frame tests is shown in Figure 6.6. The brace
was attached to the load frame via specially designed gusset plates at each end. The test instrumentation is
summarized in Figure 6.7. A load cell was installed in series with the brace at the top in order to record pre-
cise reading of the axial load in the brace. This extra load measurement was necessary because the load
frame’s internal load cell had a maximum capacity of 12MN, meaning that it would have a low resolution
over the range of the SCED brace’s maximum axial capacity of 165kN. Strain gauges were installed on
both the inner and outer members of the SCED brace to provide an estimate of the load in each tube.
String potentiometers were installed to measure the internal friction damper activation displacement, the
external friction fuse activation displacement and the total brace deformation. A photo of the full test setup
is shown in Figure 6.8.

Load Frame Head

Head (not shown) Top of Brace Detail
Connection
Plate
SW20 Load Cell (50kN)
Connected to plates on
either side using
1 1/4”-12 threaded rods Brace
Upper Connection
Gusset
Lock-up
Mechanism Lock-up
(optional) Mechanism
Two Nuts per Plate
Threaded Rod
SCED Brace Specimen 1/2” Threaded SCED Brace

Rod Body
Bottom of Brace Detail
SCED Brace
Body
Lock-up
Mechanism
Lock-up Plate
Mechanism
Base of Load Frame (optional) Brace
External
Friction Connection
Lower Fuse
Gusset
Figure 6.6: SCED Brace Calibration Load Frame Test Setup
6.4.2 SCED BRACE RESPONSE ISSUES AS-SHIPPED
There were two primary issues with the response of the braces as they were received from Korea. The
first issue was that none of the SCED braces had the activation load and internal friction damper forces
that were specified. Both the tendon pretension forces and the internal friction damper slip forces were sig-
nificantly lower than expected. This meant that all of the braces had to be recalibrated; the tendons had to
be re-tensioned and the internal friction fuses had to be re-torqued to achieve the target slip force. These
operations will be discussed in detail in the following section.
The second issue with the SCED braces as-received was that in two of the braces the inner and outer
members had slightly different lengths. This was because they were built using normal construction toler-
ances instead of the tighter tolerances that were used for the HC-SCED brace and T-SCED braces

External Instrument Locations
38 30
Load Frame Head
Load Cell
50kN Load Cell Plates
93 & Rods
30 191
Front Surface
Side Surface
Total Main Total
Brace Test Test
Length Length Length
2566 2845 3036
Tube Relative Movement

String Potentiometer
Total Brace Elongation


Load Frame Base
Strain Gauge Locations on Inner and Outer Tubes (8 Gauges total per brace)
Strain Gauge x2
Front Surface
(one on each side of the inner
2022 and outer members)
125 1717 180
CL
1322 700
HSS 100x100x4.0 50x25 Inner Section 440
Access holes
Side Surface (each side)
1172 850
CL
Strain Gauge x2
(one on each side of the inner
and outer members)
Figure 6.7: SCED Brace Load Frame Test Instrumentation

Figure 6.8: Photo of SCED Brace Press Test Setup
described in Chapters 4 and 5. As discussed previously in Section 3.5, in a SCED brace, when one member
is longer than the other by an amount greater than the linear deformation of the members under the initial
post-tensioning of the tendon, a step is created in the hysteretic response as shown for the third storey
brace in Figure 6.9. The difference in length between the two members causes this step in the hysteresis
because all of the tendon pretension force is taken by the longer tube, causing the shorter tube to ‘float’
back and forth on every cycle. During this floating process, the shorter section starts bearing on the end
plate at one end of the brace and then shifts to bear on the end plate at the other end, at which point it may
again take axial force. This defect is more likely to cause the step behaviour in shorter braces because the

members have less absolute deformation under the tendon pretension force, decreasing the length toler-
ance between the two members. Although this step behaviour does not necessarily adversely effect the
behaviour of the brace, it was decided that the shake table SCED braces should be retrofitted to eliminate
the step behaviour and achieve the most ideal brace behaviour possible. The relative member length caused
the step behaviour for the extra brace and for the third storey brace. It was the inner member that was the
shorter one for both of those braces. Those two braces were retrofit to fix this problem by welding thin
steel shims to the inside surface of one of the end plates. These shims provided a new bearing surface for
the inner member as shown in Figure 6.10.
Test 3−1 − Third Storey SCED Brace Interior Damper Activation Test
100
80 Step behavior caused by

difference in length
60 between the inner and
outer steel members
40
20
Force
0
(kN)
−20
−40 Unsymmetric hysteretic

response due to excessive
−60 mechanical friction between
moving parts
−80
−100
−8 −6 −4 −2 0 2 4 6 8
Brace Deformation (mm)
Figure 6.9: Step Response and Unsymmetric hysteresis in the Third Storey SCED brace
Another issue that was encountered that related to manufacturing quality control can be seen in Figure
6.9: the hysteretic response of the third storey brace is unsymmetric. This dissymmetry was caused by a
much higher amount of inherent friction between the moving parts of the brace mechanism when it was
activating in compression than when it was activating in tension. This problem was only encountered with
the third storey brace. It was not obvious from an external inspection of the brace where exactly this extra
friction was occurring. Therefore, to avoid having to repair the brace, the extra brace was used in place of
the third storey brace in the shake table tests.

Figure 6.10: End Plate Retrofit (Third Storey SCED Brace)
6.4.3 LOAD FRAME TESTING PROGRAM AND BRACE RECALIBRATION
The complete load frame testing program including the test input parameters for the SCED braces is
summarized in Tables 6.5 to 6.8. These tables provide loading magnitude and rate, internal friction
damper calibration, external friction fuse calibration, and the resulting SCED brace hysteretic behaviour.
The appropriate external friction fuse calibrations were determined by using the extra brace to test the
calibration scenarios for all three storeys (Tests E-1 to E-8, see Tables 6.5 and 6.6). In order to test the
external fuse without activating the internal friction fuses, lock-up mechanisms were designed and installed
at either end of the brace. These lock-up mechanisms prevented the end plates from separating from the

Table 6.5: Extra SCED Brace Load Frame Testing Summary

Y-M-D mm/s mm
Lock Disp. Disp
Up? Rate Ampl.
E-1 09-10-28 Y 0.01 -3.6 1/2 Problems with total length string pot
E-2 09-10-28 Y 0.01 -4.1 1/2 Discovered lock-up doesn’t work in tension
E-3 09-10-29 Y 0.01 -4.2 1/4 Problems with bolt tension measurement and
torque (washers are damaged)
E-4 09-10-30 Y 0.01 -4.0 1/4 Revised bolt length measurements used
E-5 09-11-02 Y 0.01 -4.7 1/4 Used new A490 bolts for external fuse
E-6 09-11-02 Y 0.01 -8.8 1/4 Simulate third storey external fuse slip
E-7 09-11-03 Y 0.01 -7.8 1/4 Simulate second storey external fuse slip
E-8 09-11-03 Y 0.01 -7.9 1/4 Simulate first storey external fuse slip
E-9 09-11-03 N 0.01 -6.5 1/2 Determine brace activation force in compression
- step first seen in hysteretic response
E-10 09-11-04 N 0.01 +4.0 1/2 Activation in tension. Step behaviour observed
E-11 09-11-04 N 0.05 -1.7 1/4 Shift short tube (go to end of step)
E-12 09-11-04 N 0.05 +/-4.5 1 Full Activation cycle. Step in both directions.
Replace total length string pot
RETENSIONED TENDON TO 46KN AND SHIMMED INNER MEMBER TO REMOVE GAP
EN-1 09-11-20 N 0.05 +/-4.5 1 No internal damper - tendon response only
EN-2 09-11-20 N 0.05 +/-4.7 1 Partially torque internal damper bolts
EN-3 09-11-20 N 0.05 +/-4.7 1 Fully torque internal damper bolts
RETENSIONED TENDON TO 26KN TO REPLACE 3RD STOREY BRACE IN SHAKE TABLE TEST
EN-4 09-11-26 N 0.05 +/-4.1 1 No internal damper - tendon response only
EN-4 09-11-26 N 0.05 +3.0 1/2 Torque internal damper bolts - too high
EN-4 09-11-26 N 0.05 +/-4.1 1 Torque internal damper bolts - too high
EN-4 09-11-26 N 0.05 +/-4.1 1 Fully torque internal damper bolts
inner and outer members when the brace was tested in compression, and they are shown in Figure 6.6.
Since these devices only prevented SCED brace activation in compression, in these tests the entire brace
assembly was deformed until external friction fuse slip occurred in compression and then brought back to
zero load. The normal force in the external friction fuse bolts was estimated using two different methods:
torque in the bolt was measured using a torque wrench and the elongation of the bolt was directly mea-
sured using a dial gauge attached to a C-shaped frame. This was the same method that was used to cali-
brate the internal friction damper interfaces for the HC-SCED and the T-SCED prototypes. The results of
these three tests are shown in Figure 6.11.

Table 6.6: Extra SCED Brace Load Frame Test Result Summary
Nm Nm kN kN kN/mm kN/mm
Estimated
External Internal External Estimated
Activation Post-
Fuse Damper Fuse Slip Initial Estimated
Test # Force Activation
Bolt Bolt
Pa
Force Stiffness
Torque Torque F Et k1
ka
- + - + - + - +
E-1 271 ? -45 56
E-2 406 ? -64 52
E-3 541 ? -81 60
E-4 609 ? -90 68
E-5 501 ? -105 64
E-6 528 ? -125 73
E-7 636 ? -140 73
E-8 650 ? -157 69
E-9 650 ? -65 53 4.0 1.11
E-10 541 ? 64 72 4.0 1.03
E-11 541 ? 35
E-12 541 ? -64 64 32 66 4.6 3.8 1.02 0.89
RETENSIONED TENDON TO 46KN AND SHIMMED INNER MEMBER TO REMOVE GAP
EN-1 541 0 -46 46 56 54 4.3 4.1 0.24 0.09
EN-2 541 81 -77 77 51 49 4.0 3.8 0.97 0.90
EN-3 541 101 -86 88 47 41 3.9 4.4 1.09 0.99
RETENSIONED TENDON TO 26KN TO REPLACE 3RD STOREY BRACE IN SHAKE TABLE TEST
EN-4 541 0 -28 29 45 40 3.8 4.3 0.36 0.14
EN-4 541 68 51 46 4.5 0.98
EN-4 541 60 -50 48 45 51 4.1 4.0 1.10 0.94
EN-4 541 54 -46 45 45 40 4.1 4.4 1.09 0.91
In the case of Tests E-7 and E-8, the slip load that was targeted was set to be approximately 5-10kN
below the actual target slip load to account for the anticipated effect of increased velocity during the actual
test. The stick-slip load fluctuations that are present on the slip plateau of the Test E-7 and E-8 curves were
caused by the slow loading rate of these tests. The apparent post-yield stiffnesses of these response curves
are actually transitional stiffnesses between partial slip and full slip. This transition is likely caused by a
conditioning of the surface similar to the phenomenon witnessed during the calibration of the external

Table 6.7: First to Third Storey SCED Brace Load Frame Testing Summary
Y-M-D mm/s mm
Lock Disp. Disp
Up? Rate Ampl.
FIRST STOREY BRACE
1-1 09-11-05 N 0.05 +/-4.6 1 Brace properties lower than expected
1-2 09-11-06 Y 0.05 -5.7 1/4 Test external friction fuse activation
1-3 09-11-06 Y 0.05 -8.2 1/4 Readjust external friction fuse torque
RETENSION TENDON TO 52KN
1-N1 09-11-20 N 0.05 +/-4.1 1 No internal damper - tendon response only
1-N2 09-11-20 N 0.05 +/-4.1 1 Fully torque internal damper bolts
SECOND STOREY BRACE
2-1 09-11-09 N 0.05 +/-5.0 1 Brace properties lower than expected
2-N1 09-11-23 N 0.05 +/-4.1 1 No internal damper - tendon response only
2-N2 09-11-23 N 0.05 +4.1 1/2 Partially torque internal damper bolts
2-N3 09-11-23 N 0.05 +/-4.1 1 Partially torque internal damper bolts
2-N4 09-11-23 N 0.05 +/-4.1 1 Fully torque internal damper bolts
THIRD STOREY BRACE
3-1 09-11-16 N 0.05 +/-6.2 1 Step behaviour observed
3-N1 09-11-24 N 0.05 +/-4.1 1 No internal damper - tendon response only -
unbalanced inherent friction
3-N2 09-11-24 N 0.05 +1.3 1/2 Partially torque internal damper bolts
3-N3 09-11-24 N 0.05 +/-4.1 1 Partially torque internal damper bolts - unbal-
ance response
3-N4 09-11-24 N 0.05 +/-4.1 1 Response still unbalanced - decided to substitute
extra brace for third storey brace
friction fuse for the HC-SCED as described in Section 4.5.1. As the displacement of the fuse increases, the
curve will eventually flatten to zero slope.
The detailed hysteretic response of the extra brace as it was shipped is shown in Figure 6.12 (Test E-12).
In addition to the overall brace hysteretic response, the relative force in each member is also shown in the

Table 6.8: First to Third Storey SCED Brace Load Frame Test Result Summary
Nm Nm kN kN kN/mm kN/mm
Estimated
External Internal External Estimated
Activation Post-
Fuse Damper Fuse Slip Initial Estimated
Test # Force Activation
Bolt Bolt
Pa
Force Stiffness
Torque Torque F Et k1
ka
- + - + - + - +
FIRST STOREY BRACE
1-1 650 ? -60 63 36 46 4.1 4.6 1.05 1.03
1-2 650 ? -126 96
1-3 704 ? -156 130
1-N1 704 0 -52 57 49 46 4.1 4.4 0.25 0.26
1-N2 704 95 -85 91 43 42 4.2 4.3 0.99 0.91
SECOND STOREY BRACE
2-1 528 ? -55 56 49 54 4.2 4.6 0.98 0.91
2-2 528 ? -115 80
2-3 609 ? -140 115
2-N1 609 0 -41 44 53 54 4.3 4.2 0.10 0.05
2-N2 609 61 61 55 4.3 0.66
2-N3 609 74 -65 68 53 47 4.3 4.1 0.83 0.78
2-N4 609 84 -68 72 50 50 4.3 4.3 0.90 0.82
THIRD STOREY BRACE
3-1 528 ? -40 33 42 43 3.9 3.9 1.18 0.76
3-2 528 ? -114 93
3-3 650 ? -160 120
3-N1 528 0 -23 20 45 45 3.8 3.6 0.52 0.20
3-N2 528 54 33 57 4.4 0.88
3-N3 528 70 -41 39 54 59 4.0 4.0 1.29 1.00
3-N4 528 61 -39 37 48 43 3.7 4.1 1.26 0.95
figure. These force values were determined using an average of the 4 strain gauge values for each member
and the tube cross-sectional area. It must be noted that these internal member forces are not absolute, they

Tests E-6 to E-8 Extra SCED Brace Exterior Fuse Tests

50
−50
Load
Cell
Force
(kN) −100 Test E-6, Equivalent Third Story
Target 123kN, Max 125kN
−150
Test E-7, Equivalent Second Story
Test E-8, Equivalent First Story
−200
−8 −6 −4 −2 0 2
External Fuse Displacement (mm)
Figure 6.11: Extra Brace External Fuse Tests (all slip levels)
are relative to the initial tube precompression that is present due to the tendon pretensioning prior to the
start of the test.
Test E-12 Extra SCED Brace Full Hysteresis Test

100
80 Total Hysteresis
Step behavior caused by 64kN

60
difference in length
between the inner and
40 outer steel members
20
Axial
Force 0
(kN)
−20 Outer Member Only
Inner Member Only
−40
−60 -64kN
−80
−100
−4 −3 −2 −1 0 1 2 3 4
Brace Deformation (mm)
Figure 6.12: Full Initial Hysteretic Behaviour of the Extra Brace

This figure shows that in addition to the step behavior, the activation load of the extra brace (64kN) is
significantly lower than the design value of 87kN (shown in Table 6.4). Since the energy dissipation capac-
ity parameter (  ) of this brace is between 0.9 (tension) and 1.1 (compression), it was clear that both the
internal friction slip force and the tendon pretension force were too low. It was not clear at this point
whether the lower-than-anticipated activation force was caused by a mislabelling of the brace prior to ship-
ping, miscalibration prior to shipping, or whether the process of shipping the braces had an effect on the
behavior. In addition, it was not clear whether the step behaviour would be present in all of the braces, or
just in the extra brace. Due to this uncertainty, the other three SCED braces (first, second and third storey)
were tested to determine their activation forces were and whether their inner and outer members were the
same length.
The key tests for the remainder of the SCED brace load frame testing program are summarized in Fig-
ure 6.13. The first column of test plots in the figure were conducted to evaluate the behaviour of all four
SCED braces as-shipped. The third storey and extra brace (Tests 3-1 and E-12), clearly show the hysteretic
step behaviour described earlier (see Figures 6.9 and 6.12), whereas the first and second storey braces did
not exhibit any such behaviour. In addition, all of the braces had activation forces that were below the tar-
get values from Table 6.4. In retrospect, it seems possible that the low activation forces may have been par-
tially caused by relaxation of the tendons due to creep in the aramid fibres as previously discussed in
Section 3.1.5. This relaxation may have reduced the pretension in the tendons, reducing the activation
load; however, it would not cause the low internal damper slip forces that were also observed. Since none of
the braces exhibited the specified activation load or internal friction damper slip force, all of the braces
were disassembled and the tendons were retensioned to the correct pretension load. In addition, the third
storey brace and the extra brace were repaired to remove the step behaviour (as discussed previously in Sec-
tion 6.4.2 and shown in Figure 6.10). To recalibrate the tendon pretension, the tendons were re-tensioned
using a jack to apply the load and a load cell to measure it as shown in Figure 6.14. The internal damper
slip force was recalibrated using a torque wrench and checked using load frame tests.
Following the SCED brace repairs and tendon retensioning, the braces were tested again to confirm the
tendon calibration. These tests are shown in the second column of test plots in Figure 6.13. The smaller,
thin hystereses in these plots show the response of the brace without any internal friction damper activa-
tion. The width of this thin flag-shaped hysteresis represents two times the inherent friction caused by fric-
tion between the moving parts of the SCED brace. The magnitude of this inherent friction is shown on the
figure at either end of the tendon-only hysteresis. The larger, full flag-shaped hystereses in this second col-

Preliminary SCED Brace Tests Summary

Initial Recalibrated
Hystereses Hystereses
Estimated Activation Force
Total Hysteresis
100 1-N1/1-N2 91 Tendon-Only Hysteresis
1-1 63 (No Internal Friction Fuse)
8
50 Estimated Inherent Friction
First
Storey 0
Brace First Storey
-50 7
-60
Shake
-100 -85
Table
100 Test
2-1 2N-1/2N-4 72 Frame
56
50 1
Second
Storey 0
Brace Second Storey Third
Storey
-50 2
-55
Axial -68
-100
Force
(kN) 100
3-1 3-N1/3N-4 Do Not
EN-4/EN-7
Use 45
50 33 37 2
Third 2
Storey 0
Brace 6 5
-50 -40 -39 Unbalanced -46
Hysteresis
-100
-8 -4 0 4 8
100 E-12 EN-1/EN-3 88
Axial Deformation (mm)
64
2
50
Extra
Brace 0
Replace Third Storey
Brace with Recalibrated
-50 6 Extra Brace
-64
-100 -86
-8 -4 0 4 8 -8 -4 0 4 8
Axial Deformation (mm)
Figure 6.13: Preliminary SCED Brace Tests Hysteresis Summary
umn of test plots show the total recalibrated hysteretic response of the brace with the internal friction
dampers calibrated and active. The activation force of each brace in each direction is shown on the plots at
the activation point on each hysteresis.
As shown in Figure 6.13, the third-storey re-calibrated brace exhibited an unsymmetric hysteretic
behaviour, with much more inherent friction in the compression direction than in the tension direction.
This was caused by a failure in quality control during the manufacturing of the brace which resulted in an
excessive amount of inherent friction between the moving parts. This third storey brace was, therefore,

Figure 6.14: Photo of Tendon Tensioning Setup
deemed unusable since it could not be repaired easily, and the extra brace was recalibrated to be used in the
place of the third storey brace in the shake table tests. From this point forward, the extra brace will be
referred to as the third storey brace.
6.4.4 SIMULATION OF THE SCED BRACE RESPONSE USING THE SCED MECHANICS SIMULATOR
To further investigate the detailed response of the small-scale SCED braces used for the shake table
tests, two of the SCED braces that were tested during the preliminary brace tests were modelled using the
SCED Mechanics Simulator that was introduced in Section 3.4.
The recalibrated second storey brace was modelled first because it is had the most ideal hysteretic
response and it therefore represented a best-case scenario for these small-scale SCED specimens. The
Mechanics Simulator model results for this brace were compared to the test results from Test 2N-4 which
is one of the key tests shown in Figure 6.13. The inputs for the SCED Mechanics Simulator for this brace
are shown in Table 6.9 and the test and model results are compared in Figure 6.15. As was the case for the
Mechanics Simulator comparisons in Chapters 4 and 5, the measured test hystereses were vertically shifted
to account for the fact that the measured values do not include the effect of the initial tendon pretension.
The tendon pretension force and damper friction for this specimen were known because they were directly
applied during the recalibration of this brace the laboratory. The length of the members for this analysis
were selected so that the model result provided a good fit to the test data. The best fit was obtained
between the test and model results when the inner member was approximately 0.35mm shorter than the

outer member. This demonstrates that even though this second storey brace had the best response out of all
the SCED braces, the small member length difference still affected the hysteretic response and the effective
initial stiffness.
Table 6.9: Mechanics Simulator Model Inputs

Test 2N-4 Test E-12
Parameter Second Storey Extra Brace
Brace Value Value
Inner Member Length L i 1999.65 mm 1999 mm
Inner Member Area A i 1217 mm2 1217 mm2

Outer Member Length L o 2000 mm 2000 mm
Outer Member Area A o 1495 mm2 1495 mm2

Tendon Area A pt 76 mm2 76 mm2
Total Tendon Pretension P p0 40 kN 33 kN

Connection Part 2 Area 1200 mm 2
1200 mm2
Bearing Stiffness 1000 kN/mm 1000 kN/mm
Friction Interface Stiffness 30 kN/mm 30 kN/mm
Previous SCED Mechanics Simulator comparisons in Chapters 4 and 5 assumed that the friction
damper interfaces at either end of the brace had practically infinite stiffness; however, for this small-scale
brace, the flexibility of the friction interfaces changed the hysteretic response. In particular, this flexibility
significantly affected the return branch stiffness highlighted in the figure. A value of 30kN/mm was

100 Full Tendon

Test 2N-4
Hysteresis
75
Model
50 Hysteresis
Model Shifted
Brace 25 Hysteresis Test Hysteresis
Axial
Force 0
(kN) Axial Force
-25
from Load
Frame
-50
-75
-100
-5.0 -2.5 0 2.5 +/-5.0 -2.5 0 2.5 5.0
75 Inner Outer
Test 2N-4
Member Member
50 Return Branch Less
Stiff Due to Friction Damper
25
Flexibility
Axial 0
Force
-25
(kN) Effect of Length
Difference
-50 (Zero Stiffness Portion)
Model
-75 Hysteresis Friction Damper
Shifted Flexibility not Equal
Test Hysteresis at Both Ends (not modelled)
-100
-5.0 -2.5 0 2.5 +/-5.0 -2.5 0 2.5 5.0
Figure 6.15: Simulation of Brace Response and Calculation of Brace Element Forces for Test 2N-4
selected for the friction interface stiffness in order to fit the inner member hysteretic shape. As the figure
shows, this return branch stiffness drop does not appear to be present in the outer member hysteresis as
predicted by the Simulator model. The reason for this may be that the friction interfaces at either end of
the brace do not have equal stiffnesses, an effect which cannot be captured by the model. As the figure
shows using the above brace modelling parameters for the member lengths and friction interface stiffness,
the Mechanics Simulator model seems to provide a reasonably good estimate of the full hysteretic response
of the real SCED brace.
The second brace that was modelled using the Mechanics Simulator was the extra brace. This brace was
modelled in its original state before it was recalibrated (Test E-12). This represents a more challenging
modelling scenario because this brace had unequal internal damper friction at each end and it exhibited the

step behaviour caused by unequal member lengths as previously shown in Figure 6.12. The input parame-
ters for the SCED Mechanics Simulator model of this brace are shown in Table 6.9 and the comparison
between the model response and the test hysteresis is shown in Figure 6.16. Similarly to the second storey
brace model, this model included a flexible friction interface with a stiffness of 30kN/mm. The tendon
pretension force and the damper friction were calculated based on the hysteretic response of the brace
using the activation force and the energy dissipation capacity parameter using the simplified equations pro-
vided in Chapter 3. The division of the internal friction damper force between the two ends was deter-
mined based on the relative widths of the inner and outer member hystereses (as previously discussed in
Section 4.4.5). The member lengths were again selected such that they provided a good fit for the test hys-
teresis data. As the figure shows, using these parameters the Mechanics Simulator model provided a good
approximation of the response of the real system, including the step behaviour and the hysteretic response
of the individual brace elements.
6.5 SHAKE TABLE TEST MODELING

Prior to the shake table testing, the full shake table test structure was modeled using the structural seis-
mic behaviour modelling software OpenSees (McKenna et al., 2000). The purpose of this model was to
evaluate the ability of computer models to replicate the behaviour of a real multi-storey SCED braced
frame. This component of the project was critical because the previous numerical building models that had
been used to evaluate the behaviour of buildings with SCED-braced frames did not have any multiple-
degree-of-freedom dynamic tests as a basis of comparison. The only physical tests that were done up to that
point were single degree of freedom brace tests. Following testing of the shake table frame, an additional
computer model of the frame was constructed using the commercial structural engineering analysis pack-
age SAP2000 (Computers and Structures Inc., 2009). The purpose of this second modelling phase was to
assess the capability of practicing structural engineers to accurately model the behaviour of SCED braces
using pre-existing commercial tools. All models were built in 2D since the single shake table test frame was
only tested in-plane.
6.5.1 SCED BRACE SIMPLIFIED MODELING
As mentioned previously in Section 3.6.3. prior to this test project, there was no material model avail-
able in OpenSees that was able to adequately represent the behaviour of the SCED brace. A custom mate-
rial model was developed for this purpose and is now included in the main OpenSees distribution. This

100 Full Tendon

Test E-12
Hysteresis
75
Model Shifted
Model Hysteresis Test Hysteresis
50 Hysteresis
Brace 25
Axial
Force 0
(kN) Axial Force
-25
from Load
Frame
-50
Step due to
-75 Member Length
Difference
-100
-5.0 -2.5 0 2.5 +/-5.0 -2.5 0 2.5 5.0
75 Inner Outer
Test E-12
Member Member
50
25
Axial 0
Force
-25
(kN)
-50 Model
Hysteresis
-75 Shifted
Test Hysteresis
-100
-5.0 -2.5 0 2.5 +/-5.0 -2.5 0 2.5 5.0
Figure 6.16: Simulation of Brace Response and Calculation of Brace Element Forces for Test E-12
material was used to model the SCED braces in OpenSEES. For details about this material model, refer to
Section 3.6.3.
Like OpenSees, SAP2000 does not have any built-in model to represent a SCED brace. To model the
SCED braces in SAP2000 a method was used that has been previously developed by Le Bec et al. (2010)
and was previously summarized in Section 3.6.2. In the SAP2000 models, there was no external friction
fuse modelled in series with any of the SCED braces.
6.5.2 OPENSEES TEST FRAME MODEL
Two related OpenSees models were built to represent the full test structure setup (see Figure 6.17). The
first model included the test structure in addition to elements that model the behaviour, mass and stiffness

of the shake table itself. For this model, the input acceleration was applied at the shake table actuator loca-
tion. This model was used to perform system identification, and to get an initial estimate of the loads on
the shake table components such as the actuator and the vertical components. The second model was used
to analyze the response of the test frame after the tests were complete. This was done by removing the ele-
ments representing the shake table from the original model. Then, the actual acceleration that was
recorded at the base of the test specimen from each shake table test was applied at the bottom of the struc-
ture. Hence, the method of applying the base acceleration and the exclusion of the shake table elements
were the only things that were changed between the first and second models.
a) Model Before Tests b) Model After Tests

(including Shake Table) (without Shake Table)
191 2212 441 Lumped 191 2212 441 1524
Rigid Weight (typ.)
Column 0.6kN 1.4kN 119.5kN 0.6kN 1.4kN 119.5kN
Offsets
(typ.) 1321
279
SCED Braces 0.9kN 121.1kN 0.9kN 121.1kN

279 1.9kN Rotational 1.9kN 1321
Pin (typ.)
102 Node (typ.)
0.8kN 2.1kN 121.1kN 0.8kN 2.1kN 121.1kN
102
y- and rot- y- and rot-
Restraint Restraint 1321
x- and y-
5.0kN Slave Node to S1x,S2y S1x 5.0kN Restraint 12.6kN
M1 in x and y 12.6kN 4.5kN
Fixed S1xy 4.5kN
Restraint 687
S1x S1x S1x
M1 67.1kN M2,S1x
Master 609 y
Node 1 y-Restraint (typ.) Master
Node 2 x
138 138
1778 1284 1284 1524
Figure 6.17: SCED shake table test structure OpenSEES models

(a) with the shake table - horizontal accelerations applied at the fixed restraint on the left, and
(b) without the shake table - horizontal accelerations applied at the bases of the columns
6.5.3 OPENSEES ASSUMPTIONS / SIMPLIFICATIONS
In the pre-test model, high axial stiffness elements were used to model the shake table itself. Horizontal
and vertical constraints were applied to these elements to force them to act as a rigid body. The lower verti-
cal members of the shake table were assigned an axial stiffness equivalent to the vertical stiffness of the
shake table's bearing supports. The member that connects the shake table to the fixed restraint on the left
was given an axial stiffness equivalent to the horizontal series stiffness of the entire shake table and the actu-
ator. The mass of the shake table was lumped at its centre of gravity.

In both models, the mass of the frame and mass simulation system were lumped at the closest nodes.
The mass simulation system was modeled as a single multi-storey leaning column in the model, consisting
of truss elements that were assigned an axial stiffness equivalent to the combined axial stiffnesses of all the
pin-ended columns in each storey of the real mass-simulation system. This system was connected to the
main frame in the model by truss elements with stiffness values equivalent to the actual HSS beams that
connected the mass simulation system to the test frame (no diaphragm constraints were used).
Vertical rigid end offsets were assigned to the test frame columns to simulate the physical dimensions of
the gusset plates in the corners. The beams were modeled using three elements in series. The left and right
elements were stiff beam members with full-length rigid offsets representing the gusset plates. The center
element represented the beam member itself with fibre-section lumped plasticity hinges at either end to
model the behaviour of the flexible horizontal splice plate end connections.
Inherent damping in the models was modeled using Rayleigh damping in Modes 1 and 3. Different
amounts of damping were tested to check the sensitivity of the model, which will be described further in
Section 6.7.7.
6.5.4 SAP2000 TEST FRAME MODEL
The purpose of the SAP2000 model was to compare the more sophisticated OpenSees model to one
that is more likely to be used in a design setting. This model had the same dimensions, linear frame prop-
erties, rigid end offsets and mass as the OpenSees model. However, the simplified SAP2000 model used
pinned ends for the beams since practicing engineers are unlikely to model all beam-column joints using
plastic hinges. This was not expected to create a significant difference in the results since the beam column
joints of the shake table specimen were designed to simulate simple shear connections (not moment-resist-
ing) and, hence, have a low rotational stiffness. The nonlinear time history analysis was performed using
the nonlinear modal time history analysis method using Ritz vectors instead of a full direct integration
method. This Ritz-modal analysis method was found to be orders of magnitude faster for this type of struc-
ture. The drawback of using the Ritz vector method in SAP2000 was that P-Delta effects could not be con-
sidered in the analysis. It was also not possible to create a stable model in SAP2000 that included the
response of the external friction fuse in series with the brace. Constant inherent damping of 1% of critical
was provided in all modes.

6.6 SELECTION AND ADAPTATION OF GROUND MOTIONS

Four groups of ground motions were used for the shake table tests, with each group representing a dif-
ferent phase of the testing program. The selected records with their properties and applied scaling factors
are shown in Table 6.10. The scaling for each phase is described in the following sections. All of the records
are scaled in time by a factor of 0.577 as described in Section 6.1.2. The selection of the records for each
phase is described below.
Table 6.10: Earthquake Record Properties and Scaling

SAC DBE
Total
R ATC MCE
No. Record Year Name Station M %EQ Scale
km Scale Scale
Fact.
Fact. Fact.
1-1 SAC LA50 1984 Morgan Hill gil3 6.2 15 2.35 1.00 30% 0.71
1-4 SAC LA53 1966 Parkfield cs08 6.1 8.0 2.92 1.00 100% 2.92
1-5 SAC LA55 1986 N. Palm Springs plma 6.0 9.6 2.75 1.00 100% 2.75
2-1 ATC N09B 1994 Northridge Rinaldi 6.7 10.9 0.69 1.33 50% 0.46
2-3 ATC N01B 1979 Imperial El C. 6 6.5 27.5 0.9 1.33 90% 1.08
Valley
2-4 KOBE 01B 1995 Kobe Takatori 6.9 13.1 1.00 1.00 90% 0.90
3-1 SUBMod- 2003 Tokachi Oki 84km 8.0 84 1.00 1.00 300% 3.00
F01B Station
4-1 SAC LA-50 1984 Morgan Hill gil3 6.2 15 2.35 1.00 100% 2.35
4-2 MTL 02 SIM Sim. Montreal N/A 7.0 42 1.00 1.00 250% 2.50
No. 2
4-3 CVX31 SIM Sim. Charlevoix N/A 7.0 26 1.00 1.00 110% 1.10
No. 31
Phase 1 consisted of three different records for Los Angeles taken from the suite of records of the SAC
joint venture project (Somerville et al., 1997) for a probability of exceedence of 50% in 50 years. The SAC
project records are site appropriate and scaled using a single scaling factor on the acceleration of each
record, given in Table 6.10 as the "SAC/ATC Scale Fact."
Prior to testing, all 20 records from the SAC 50% in 50 year suite for Los Angeles were evaluated using
the pre-test OpenSees model. Three of these were selected for the shake table tests on the basis of including
some diversity in the frequency content of the different records and, consequentially, in the characteristics

of the frame response (e.g. having records that produce both a one-sided versus two-sided hysteretic
response). In addition, one of these was scaled by 30% and 50% to include consideration of even lower
ground motion intensities.
Phase 2A consisted of two records for Los Angeles that have a probability of exceedance of 10% in 50
years. These records were from the online PEER NGA database and scaled as per the FEMA-P695 specifi-
cation (Federal Emergency Management Agency [FEMA], 2009). Similarly to Phase 1, the entire FEMA-
P695 suite of records was evaluated using the pre-test model and two records were selected. These records
were further scaled to achieve a maximum drift of approximately 2% in the first storey SCED brace.
Phase 2B consisted of a single large magnitude record to simulate an event with a probability of
exceedence of 2% in 50 years. For a significantly large record and to avoid repeating a previous record, the
1995 Kobe Takatori record was used and it was scaled to achieve approximately 4% drift in the first storey
brace in order to test external friction fuse activation.
Since self-centering systems are particularly useful for withstanding long duration earthquakes, Phase 3
consisted of a subduction type earthquake record. For this single test, the historical 2003 Tokachi Oki sub-
duction record was used and scaled up to match the 10% in 50 year spectrum for Los Angeles.
Records for Phase 4 of the project were selected to represent the demands of eastern North American
earthquakes, which are typically energy rich in the higher frequency ranges. One of the records from Phase
1 was re-run to represent a 10% in 50 year probability of exceedance in Montreal, Quebec. The last two
records were simulated records created by Atkinson (2009). The first represents the demand from a 2% in
50 year probability of exceedance event for Montreal and the second is representative of the demand from
a 2% in 50 year probability of exceedance event for Charlevoix, Quebec. To compensate for the difference
between the seismic demand for Montreal/Charlevoix and the seismic demand for Los Angeles that the test
frame was designed for, these records were scaled-up by a ratio of the design spectrum value for Los Angeles
to the design spectrum value for each site at the design period.
6.7 SHAKE TABLE TEST RESULTS AND DISCUSSION

A summary of the SCED brace hysteretic response parameters for all three storeys for all of the shake
table tests is shown in Table 6.11. In addition, a summary of the maximum system response quantities of
storey acceleration, storey shear, storey drift and column axial force for all of the shake table tests is shown
in Table 6.12. This table also compares those experimental response quantities to those that were predicted
using both the OpenSees and SAP2000 models that were described in Section 6.5. The OpenSees model

results in this table were attained using the after-test model shown in Figure 6.17. Inherent damping in the
models was considered by providing 1% of critical damping. Rayleigh damping was used for the OpenSees
model and constant damping was used for the SAP2000 model. Values between 1% and 3% were tested in
the models, but 1% was selected because it gave the best match for the storey drift and shear values for the
moderate to high acceleration earthquakes. The effect of inherent Rayleigh damping in the model will be
discussed in greater detail a later section.
Table 6.11: Experimental Shake Table SCED Brace Response

kN kN/mm kN/mm
Estimated
Estimated Estimated
Post-Activation Estimated
EQ Activation Force Initial Stiffness
Pa k1
ka
S1 S2 S3 S1 S2 S3 S1 S2 S3 S1 S2 S3
1-1 - - - 70 82 97 - - -
1-2 - - - 55 65 70 - - -
1-3 73 65 44 55 68 62 4.8 4.6 5.1 0.85 0.87 1.02
1-4 72 65 47 57 67 64 4.7 4.4 5.3 0.85 0.86 0.90
1-5 72 64 46 60 68 63 4.6 4.6 4.3 0.85 0.88 0.93
2-1 70 65 47 63 69 66 4.8 4.4 4.4 0.83 0.83 0.87
2-3 69 63 47 60 66 72 4.4 4.7 4.5 0.88 0.86 0.93
2-4 BEFORE EXTERNAL FUSE ACTIVATION (AT 175KN):
2-4A 73 62 48 51 74 64 4.1 4.5 4.5 0.83 0.88 0.91
2-4 AFTER EXTERNAL FUSE ACTIVATION:
2-4B 50 58 48 57 76 64 3.7 4.5 4.5 0.96 0.79 0.91
3-1 72 65 48 65 69 69 4.3 4.6 4.2 0.83 0.84 0.93
4-1 53 56 48 54 74 66 3.7 4.3 4.6 0.90 0.75 0.91
4-2 52 56 - 52 72 59 4.1 - - 0.87 0.70 -
4-3 51 55 - 57 70 60 3.7 4.1 - 0.89 0.78 -
A close-up photo showing the first-storey SCED brace mechanism during activation of the brace is
shown in Figure 6.18. This photo shows the lower end of the brace in tension activation. For this brace,
the inner member is connected to the frame on the lower-left and the outer member is connected to the
frame at the upper right (not shown). In the photo, the end plate is separated from the outer element, elon-
gating the tendon within.

Table 6.12: Experimental Results vs. Model Results

Experimental Result (OpenSees Result with 1% Damping) [SAP2000 Result with 1% Damping]
g g kN % kN
EQ PGA Accelerations Storey Shear Drift Column Axial
S1 S2 S3 S1 S2 S3 S1 S2 S3 L R
1-1 0.16 0.19 0.15 0.15 39 35 16 0.08 0.05 0.05 22 38
(0.32) (0.30) (0.37) (66) (58) (44) (0.13) (0.09) (0.08) (47) (73)
[0.31] [0.28] [0.37] [67] [59] [44] [0.14] [0.12] [0.10] [44] [79]
1-2 0.27 0.28 0.22 0.25 66 50 30 0.16 0.11 0.09 33 60
(0.41) (0.48) (0.41) (68) (66) (49) (0.19) (0.19) (0.15) (51) (84)
[0.39] [0.44] [0.41] [69] [67] [48] [0.19] [0.21] [0.16] [52] [86]
1-3 0.59 0.50 0.38 0.41 92 70 44 0.61 0.30 0.17 48 85
(0.50) (0.68) (0.44) (86) (71) (52) (0.51) (0.28) (0.22) (55) (91)
[0.56] [0.70] [0.44] [85] [70] [53] [0.48] [0.29] [0.25] [57] [92]
1-4 0.69 0.51 0.41 0.42 95 72 48 0.67 0.35 0.20 48 89
(0.84) (0.94) (0.48) (104) (76) (52) (0.83) (0.45) (0.36) (61) (103)
[0.86] [0.96] [0.50] [101] [76] [60] [0.80] [0.45] [0.36] [63] [102]
1-5 0.52 0.50 0.62 0.95 129 100 71 1.31 0.87 0.63 66 116
(0.71) (0.75) (0.61) (114) (89) (72) (1.09) (0.67) (0.61) (67) (118)
[0.72] [0.76] [0.61] [111] [98] [73] [1.07] [0.71] [0.59] [68] [118]
2-1 0.22 0.47 0.38 0.43 102 79 46 0.87 0.46 0.29 56 91
(0.83) (0.60) (0.48) (95) (87) (57) (0.70) (0.46) (0.39) (57) (98)
[0.81] [0.54] [0.45] [93] [87] [54] [0.69] [0.49] [0.37] [60] [100]
2-3 0.35 0.48 0.46 0.45 138 91 50 1.60 0.82 0.27 68 119
(0.73) (0.48) (0.44) (128) (93) (52) (1.25) (0.77) (0.31) (68) (115)
[0.71] [0.54] [0.44] [126] [89] [52] [1.23] [0.76] [0.32] [67] [111]
2-4 0.54 0.87 0.71 0.87 192 139 97 3.96 2.00 1.39 102 171
(1.07) (1.04) (1.02) (179) (157) (122) (3.26) (1.99) (1.63) (115) (194)
3-1 0.28 0.39 0.40 0.43 108 86 45 1.00 0.64 0.21 52 99
(0.76) (0.53) (0.41) (109) (82) (48) (0.91) (0.59) (0.25) (59) (113)
[0.76] [0.52] [0.40] [107] [84] [48] [0.89] [0.61] [0.25] [61] [113]
4-1 0.55 0.60 0.38 0.39 92 72 46 1.09 0.59 0.21 77 87
(0.59) (0.58) (0.41) (88) (65) (48) (0.93) (0.41) (0.20) (55) (85)
4-2 0.51 0.50 0.32 0.32 73 58 39 0.74 0.15 0.12 66 75
(0.58) (0.44) (0.40) (78) (60) (45) (0.71) (0.27) (0.15) (50) (75)
4-3 0.36 0.42 0.34 0.37 79 60 38 0.87 0.28 0.11 59 75
(0.60) (0.40) (0.40) (75) (60) (46) (0.67) (0.28) (0.11) (49) (78)
Sample time history drift and hysteretic responses are shown in Figure 6.19. The results in this figure
demonstrate the diversity of the shake table excitations that were chosen and the consistent response of the
SCED braces that were obtained for all of these records. One sample record is shown from each phase of
the testing program. Recall that the time dimension of the input excitation is scaled by 0.577 for all of

Figure 6.18: First Storey Brace Mechanism
these tests. Peak drift values for these records may be found in Table 6.12. The results in Figure 6.19 are
shown without comparison to the model results in the interest of clarity; individual test/model compari-
sons will follow. Test 1-4A represents a 50% in 50 year hazard level excitation with a moderate duration. In
this record, the SCED braces at all three storeys activate. The first storey in particular experiences approxi-
mately ten cycles of activation. Test 2-1A represents a 10% in 50 year hazard level excitation with higher
maximum drift and a balanced response in both directions. Under this record, the braces experience more
drift than in test 1-4A; however, the response is characterized by a couple of large cycles rather than a large
number of smaller cycles. Test 2-4A represents a 2% in 50 year hazard level excitation with high drifts,
external fuse activation in the first storey in both directions, and good propagation of the drift demand
into the second and third storey braces above. The hysteretic response of the first brace shows that the
external fuse activated in both directions, resulting in a small residual drift of less than 0.5%. The behavior
of the test frame subjected to this record will be discussed in further detail in a future section. Test 3-1A
represents a long-duration subduction-type earthquake. As these plots show, when subjected to this long
earthquake, the first storey brace undergoes approximately twenty cycles of activation. Even though the
brace activates so many times, the shape of the hysteretic response remains consistent and shows no signif-
icant degradation. A yielding system subjected to a similar history of demand may experience a progres-
sively increasing drift in one direction due to P-Delta effects, whereas the SCED brace experiences no
residual drift during consecutive activation cycles or at the end of the excitation. Finally, Test 4-3A repre-
sents a simulated east-coast earthquake response characterized by higher frequency ground accelerations
with higher drifts in the one direction than the other. Again, the self-centering capability of the SCED
braces prevents this excitation that is strongly-biased in one direction from causing residual drift in the
frame.

Shake Table Tests: Sample Drift and Hysteretic Responses

Test 1-4A - 100% SAC-LA53 (1966 Parkfield) First Second Third
1% Story Story Story
Story 3 150
0
Drift 100
-1%
Story 50
Story 2 1% SCED
0
Drift Force 0
-1%
1% (kN) -50
Story 1
0 -100
Drift
-1%
-150
-2% -1% 0 1% 0 1% 0 1%
Test 2-1A - 50% ATC-N09B (1994 Northridge)
1% 150
Story 3
0
Drift 100
-1%
Story 50
Story 2 1% SCED
0
Drift Force 0
-1%
1% (kN) -50
Story 1
0 -100
Drift
-1%
-150
-2% -1% 0 1% 0 1% 0 1%
Test 2-4A - 90% KOBE-01B (1995 Kobe)
4% 250
Story 3
0
Drift
-4% 100
Story
Story 2 4% SCED 50
0
Drift Force 0
-4% -50
4% (kN)
Story 1
0 -100
Drift
-4%
-250
-4% -2% 0 2% 0 2% 0 2%
Test 3-1A - 300% SUBMod-F01B (2003 Tokachi Subduction)
1% 150
Story 3
0
Drift 100
-1%
Story 50
Story 2 1% SCED
0
Drift Force 0
-1%
1% (kN) -50
Story 1
0 -100
Drift
-1%
-150
-2% -1% 0 1% 0 1% 0 1%
Test 4-3A - 110% CVX31 (Simulated Charlevoix No. 31)
1% 150
Story 3
0
Drift 100
-1%
Story 50
Story 2 1% SCED
0
Drift Force 0
-1%
1% (kN) -50
Story 1
0 -100
Drift
-1%
-150
0 5 10 15 20 -2% -1% 0 1% 0 1% 0 1%
Time (s) Story Drift
Figure 6.19: Sample Shake Table Test Results - Drift Time Histories and SCED Brace Hystereses
6.7.1 SCED BRACE AXIAL FORCE CALCULATION
Most of the brace axial forces that are given in Table 6.11 were calculated using a combination of strain
gauge data and linear potentiometer data from the SCED braces. The method used to calculate the axial

force using this method was previously described in Section 4.4.2. To check the accuracy of this SCED
axial force calculation, the results of this approximation for one record were compared to an alternate axial
force measurement that was calculated using the storey shears. The storey shear was determined using the
load cells that connect the test frame to the mass simulation frame (as shown in Figure 6.5). To calculate
the axial force in the brace, the estimated column shear that was calculated using the strain gauges on the
columns was subtracted from the storey shear and then the resulting value was projected into the brace
direction:
 V xi – V cLi – V cRi 
P ai = ------------------------------------------
- (Eq. 6.1)
cos   b 
where V xi is the storey shear at storey i , V cLi and V cRi are the shear in the left and right columns at that
storey, and  b is the brace angle. The storey shear V xi was determined by adding up the load in the load
cells from the current storey and all the storeys above it. The shear in the columns was calculated by using
the strain gauges on the columns. Since the second and third storeys only had strain gauges on the left col-
umn, the right column shear force was assumed to be the same as the left column shear force for those sto-
reys.
The comparison of the brace hystereses for each storey that resulted from the two different methods
that were used to calculate the axial force in the SCED braces is shown in Figure 6.20. Both methods pro-
vided a similar force reading; however, the measurement that was calculated using the forces in the brace
elements was less noisy, therefore it was used for all tables and plots except for the first storey brace for Test
2-4 as will be discussed in the following section.
6.7.2 SCED BRACE ACTIVATION FORCE IN SHAKE TABLE TEST VS. IN LOAD FRAME
Using Tables 6.6, 6.8 and 6.11, the SCED brace properties from the individual load frame tests of each
brace are compared with the estimated SCED brace properties from the shake table test in Table 6.13. For
the load frame tests, the values in this table are taken from the last calibration test for each brace. For the
shake table tests, the values in the table are estimated representative values for each brace from Table 6.11.
The activation forces for the shake table testing of the second and third storey braces are similar to their
activation forces that were determined during their respective load frame tests; however, the activation
force for the first storey brace seems to be approximately 16kN lower for the shake table test than it was in
the load frame tests. Since the  parameter also dropped significantly between the load frame tests and the
shake table tests, it seems like this loss of activation force was due to a loss in the internal friction damper

Test 2-3A - 90% ATC-N01B (1979 Imperial Valley)

Brace Axial Force Calculation Comparison
First Storey SCED Second Storey SCED Third Storey SCED
150 100 60
Axial
100 from 40
50 Brace
Elements
50 Axial 20
SCED
Axial from
0 0 Storey 0
Force Shear
(kN)
−50 −20
−50
−100 −40
−150 −100 −60

-20 -10 0 10 20 -10 -5 0 5 10 -4 -2 0 2 4
SCED Axial Deformation (mm)
Figure 6.20: Comparison between SCED Brace Axial Force Calculation Methods
Table 6.13: Comparison of Shake Table Test and Load Frame Test SCED Response
kN kN kN/mm kN/mm
Estimated
Estimated Estimated Estimated
Post- Estimated
Activation External Initial
Brace Force Fuse Slip Stiffness
Activation 
Stiffness Parameter
Pa F Et k1
ka
First Storey - Load Frame Test 88 156 43 4.3 0.95
First Storey - Shake Table Test (Before 2-4) 71 175 60 4.6 0.85
First Storey - Shake Table Test (After 2-4) 51 175 60 4.0 0.90
Second Storey - Load Frame Test 70 140 50 4.3 0.86
Second Storey - Shake Table Test (Before 2-4) 64 - 60 4.5 0.87
Second Storey - Shake Table Test (After 2-4) 56 - 60 4.5 0.79
Third Storey - Load Frame Test 45 114 42 4.3 1.0
Third Storey - Shake Table Test 47 - 60 4.5 0.93
slip force. This may have occurred during handling of the braces to position them in the braced frame.
Instead of adjusting the friction damper load while the brace was in the frame, the brace was left as-is.
The first storey brace was the only one that experienced an external friction fuse slip during any of the
tests. This slip occurred during Test 2-4. The axial force at which this slip occurred is higher than the cali-
brated force from the load frame test, but similar to the intended design value of 164kN from Table 6.4.
Recall that the external fuses were intentionally calibrated to be lower than the design value in order to

account for velocity effects as described previously in Section 6.4.3. After the external slip occurred, the
brace experienced a change in properties due to a hyper-elongation of the tendons past their original pre-
tension load as previously discussed in 3.1.5. This result was surprising because the first storey brace was
tested up to the maximum elongation including activation of the external friction fuse in Korea before they
were sent to Montreal as shown in Figure 6.21 (Kim, 2009b); however, there is evidence that the drop in
activation force was due to nonlinear tendon behaviour, which points to hyper-elongation being the cause.
The brace axial force that is calculated using the brace element elongations as previously described in Sec-
tion 6.7.1 diverges from the axial force calculated using the storey shear as shown in Figure 6.22. This
divergence occurred because the pretension in the real brace dropped when the tendon stiffness dropped.
The tendon stiffness dropped because the fibres had not previously been tensioned up to that level of axial
force; however, for the left plot, the axial force was still being calculated based on an assumed linear behav-
iour. This affects the apparent stiffness of the brace, distorting the hysteretic shape. In addition, when the
activation force dropped, the  parameter increased, suggesting that the internal friction damper slip force
remained constant when the tendon pretension force dropped. Since the axial force calculated using the
brace elements was no longer correct due to the tendon nonlinearity, the axial force value was calculated
using the storey shears for Test 2-4 in Tables 6.11 and 6.12.
Figure 6.21: Korean Test Result - First Storey SCED Brace Response (from Kim, 2009b)

Test 2-4A - 90% KOBE-01B (1995 Kobe)

Axial Force from Brace Elements Axial Force from Storey Shear
200
150
100
SCED 50
Axial
Force 0
(kN) -50
-100
-150
-200
-50 -25 0 25 50 -50 -25 0 25 50
SCED Axial Deformation (mm) SCED Axial Deformation (mm)
Figure 6.22: Comparison of Axial Force Calculation Method for First Storey SCED Brace Test 2-4
Even though there was no significant external friction fuse activation for the second storey brace during
Test 2-4, there also seemed to be a modest, approximately 8kN drop in brace activation force as shown in
Table 6.11. It seems that this drop may not be due to hyper-elongation of the tendon, though, because it
was accompanied by a drop in the  parameter, suggesting that the internal friction damper slip force
dropped. This may have been due to the internal friction damper normal force bolts bearing on the ends of
their slots, causing them to turn and release some of their pretension.
In general, the axial stiffness values in Table 6.13 are slightly higher for the shake table tests than they
were for the preliminary load frame tests. This is likely because the load frame test had a more flexible gus-
set plate connection than the shake table test (since both measurements included the stiffness of the con-
nection).
6.7.3 SCED PARAMETERS USED FOR COMPUTER MODELLING
The SCED brace elements that were used in the OpenSees and SAP2000 models were assigned param-
eters such that the hysteretic behaviour of those elements matched the measured shake table test results
from Table 6.13. For Test 2-4, the values from before the external friction fuse activation were used for the
SCED model because it was not practical to change the SCED properties halfway through the excitation.
For the tests that followed Test 2-4 (Tests 4-1, 4-2 and 4-3), the reduced properties that resulted from Test
2-4 for the first and second storey braces were used.

6.7.4 OVERALL COMPARISON OF EXPERIMENTAL AND MODEL RESULTS
As can be seen in Table 6.12, the OpenSees and SAP2000 models, despite their differing levels of
sophistication, resulted in similar good-quality predictions for the behavior of all the SCED braces in the
multi-storey frame. Figure 6.23 shows a comparison of some of the detailed results of these two different
models for two selected test earthquakes. As this figure shows, the two models provided almost identical
predictions for both the hysteretic response and the time history responses for top floor displacement and
column axial force. In addition, both models provide an accurate representation of the response of the test
frame, suggesting that the simplified SAP2000 model would be useful to practicing engineers for designing
SCED-braced frames.
A more comprehensive comparison of the two models results with the shake table test results is shown
in Figure 6.24. In this figure, the maximum values of floor acceleration, storey shear, storey drift, and col-
umn axial force are given for each test model relative to the recorded response of the shake table test struc-
ture. Each test model (either OpenSees or SAP2000) is plotted at a location on the x-axis corresponding to
the maximum drift that was recorded for that test in the real structure. This figure shows that the storey
shear, storey drift, and column axial loads are all well-predicted by the model except at the lowest maxi-
mum drift amplitudes and that the OpenSees and SAP models provide similar predictions for each test. In
most cases, modelled drifts were within 15-20% of the test result values and storey shear values were within
10%; however, the accelerations varied by up to 50-75%, but they were always overestimated by the
model. This makes the model predictions over-conservative with respect to the accelerations. These results
are similar to those found in a study conducted by Wolff and Constantinou (2004), where they found that
in seismically isolated structures, accelerations were poorly predicted while the other response quantities
were predicted well. The reasons for the large discrepancy between the model and test results for accelera-
tion will be discussed in more detail in a future section.
In general, the values of the model results divided by the test results as shown in Figure 6.24 decrease as
the maximum test drift increases. This trend may be attributed to the inherent damping in the system
which was modelled using a Rayleigh damping as will be discussed in a future section.
6.7.5 HYSTERETIC RESPONSE, DRIFT RESULTS, AND COMPARISON TO MODEL PREDICTIONS
The shake table tests demonstrated that the hysteretic response of the SCED braces was both stable and
repeatable under a large number of cycles of deformation (over more than twelve strong ground motions).
Using the measured hystereses, the total activation force for each brace was determined after each test, and

Test 1-3A - 100% SAC-LA50 Test 2-3A - 90% ATC-N01B

(1984 Morgan Hill) (1979 Imperial Valley)
Open- SAP- Open- SAP-
Sees 2000 Sees 2000
Test Model Model Test Model Model
100 150
100
50
First Story 50
SCED
0 0
Axial Force
(kN) -50
-50
-100
-100 0.5% -150 1.5%

-0.5% 0 0 0 0.5% -1.5% 0 0 0 1.5%
First Story Drift (%h1) First Story Drift (%h1)
5mm 10mm
Test (typ.)
Top Floor
0 0
Displacement
0 0
(mm) 0 0
OpenSees (typ.)
SAP2000 (typ.)
25kN 25kN
Left Column 0 0
Axial Force 0 0
(kN) 0 0
18 19 20 21 22 13 14 15 16 20
Time (s) Time (s)
Figure 6.23: Comparison of SAP2000 and OpenSees model results
these results showed that there was no appreciable drop in activation force as a result of internal friction
fuse degradation. Some sample drift responses and hysteretic plots for individual shake table tests are
shown with their corresponding model results in Figure 6.25.
The first results shown in Figure 6.25 are for the first storey of the test frame subjected to the long dura-
tion Tokachi subduction earthquake (Test 3-1A). This type of earthquake has the potential to result in pro-
gressive collapse for traditional hysteretic yielding or friction systems. The SCED braces, however, did not
experience any significant residual drift. The drift plot shows that the drift histories and peak drift values
are similar between the experiment and the OpenSees and SAP2000 models. The accompanying hysteresis
plot on the right shows that the real SCED brace was able to resist dozens of activation cycles without a

2.5
Maximum Acceleration Maximum Story Shear
2.0
Model Value
1.5
Test Value SAP2000
1.0
OpenSEES
0.5
2.5
Maximum Drift Maximum Column Axial Force
2.0
Model Value
1.5
Test Value
1.0
0.5
0 1% 2% 3% 4% 0 1% 2% 3% 4%
Maximum Drift in Test Maximum Drift in Test
Figure 6.24: Comparison of Test Results with both Computer Models
change in the response. The experimental hysteretic response is also well matched by the prediction of the
analytical model.
The second set of results in Figure 6.25 shows the response of the frame subjected to the largest ampli-
tude record that was used: the 1995 Kobe earthquake. The time history of the drift response for the first
storey is shown along with the hysteretic response of the SCED braces at all three storeys. Under this
record, the first storey experienced drifts on the order of 4.0%, which activated the external friction fuse.
This activation is indicated by the horizontal (zero stiffness) portions of the first storey hysteresis. Since it
was not possible to model the external friction fuse in SAP2000, only the OpenSees model results are
shown for this record. As the plots show, the first storey brace experienced considerable external fuse move-
ment in both directions, producing a residual drift of approximately 0.3% in the experimental results and
approximately 0.7% in the OpenSees model results as shown in the drift time history on the left side of the
figure. The model predicts this response very well and the lower two hysteresis plots show that the model
also predicted the response of the second and third storey braces well. As previously discussed in Section
6.7.2, during this excitation, the first storey brace experienced a drop in its activation force after the peak
cycle.
Since the residual drift that remained in the first storey SCED brace after Test 2-4 was not much larger
than the AISC erection tolerance for columns of 1/500 (AISC, 2005c), the frame was not adjusted and
later tests were conducted using the new position as the reference point for zero drift.

Test 3-1A - 300% SUBMod-F01B (2003 Tokachi Oki Subduction Earthquake) Open- SAP-
Sees 2000
1% Rayleigh Damping Test Model Model
150
2.0%
100
1.0% 0.998%
First 0.910% SCED 1 50
Story Axial
0 0
Drift Force
(%h1) (kN) -50
-1.0%
-100
-2.0% 1.0%
-150
0 10 20 30 40 50 -1.0% 0 0 0 1.0%
Time (s) First Story Drift (%h1 - same as left)
Test 2-4A - 90% KOBE-01B (1995 Kobe)
1% Rayleigh Damping OpenSees
Test Model
4.0% 200
3.0% OpenSees 150
Model 100
2.0%
First SCED 1
50
Story 1.0% Axial
0 0
Drift Force
-1.0% -50
(%h1) Test (kN)
-2.0% -100
-3.0% -3.261% -150
-3.960% 2.0%
-4.0% -200
0 5 10 15 -4.0% -2.0% 0 2.0% 0 2.0% 4.0%
Time (s) First Story Drift (%h1 - same as left)
OpenSees OpenSees
Test Model Test Model
200 200
150 150
100 100
SCED 2 SCED 3
50 50
Axial Axial
0 0
Force Force
-50 -50
(kN) (kN)
-100 -100
-150 -150
2.0% 2.0%
-200 -200
-2.0% -1.0% 0 1.0% 0 1.0% 2.0% -2.0% -1.0% 0 1.0% 0 1.0% 2.0%
Second Story Drift (%h2 ) Third Story Drift (%h3 )
Figure 6.25: Sample Drift and Hysterestic Responses (Experiment vs. Model)
6.7.6 ACCELERATION RESULTS AND COMPARISON TO MODEL PREDICTIONS
One of the goals of this shake table study was to assess the acceleration response of multi-storey SCED
braced frames and to determine whether the high accelerations that were observed in previous numerical
studies (Choi et al., 2008; Tremblay et al., 2008) were caused by a real phenomenon or were entirely or
partially numerical.
Self-centering systems are unique in that they experience a low-to-high stiffness transition at high-
velocity as they pass through zero deformation. Typically, when a self-centering system returns towards zero

deformation, it does so at a relatively low stiffness. Near zero deformation, the initial high stiffness of the
system is recovered before it can activate in the opposite direction. Since the system has momentum as it
passes through zero, this transition tends to occur at high-velocity, in contrast to when the system is chang-
ing directions at peak deformation when the velocity tends to be lower.
Because the stiffness changes in a SCED braced frame tend to occur at high velocity, the forces in the
braces also change quickly, easily causing large imbalances between the forces in braces in adjacent storeys
if they do not change stiffness at the exact same time. This imbalance in the SCED forces above and below
a given storey causes the presence of a floor acceleration which is required to satisfy dynamic horizontal
equilibrium as shown in Figure 6.26. In the figure, the force caused by that floor acceleration, F inertia , is
shown. Shears in the columns and the damping force are typically small compared to brace forces in braced
frames, and acceleration is therefore mostly caused by the brace force difference. Figure 6.26 also shows the
reason why high velocity stiffness transitions tend to produce larger force differences and, therefore, higher
accelerations in self-centering systems. This simplified figure assumes that SCED braces of two adjacent
storeys have a similar hysteretic shape. If it is assumed that the time lag between the responses of the braces
in two adjacent storeys is approximately constant, then less distance will be travelled around the hysteresis
in that amount of time when the velocity is low than when it is high. Accordingly, the distance between the
two SCEDs on the hysteresis will also be higher when the velocity is high. Larger distances mean larger
force imbalances when the stiffness changes, which in turn result in higher floor accelerations. Inertial floor
accelerations caused by the braces are however limited by the maximum difference between the forces in
the two SCED braces, equal to the sum of the highest axial load that is possible for each brace.
Dynamic Equilibrium of First Floor Effect of Velocity on Force Imbalance

Low Velocity (near peaks) High Velocity (near zero)
Vleftcol,2 Vrightcol,2 P Story A P
Imbalance
FSCED,2 Story B
Fdamping Finertia Story A
FSCED,1 δ δ
*Equal Time Lag Story B Imbalance
Vleftcol,1 Vrightcol,1 Between Points
on each Plot
Figure 6.26: First Floor Dynamic Equilibrium and the Effect of Velocity on the
Force Imbalance between Adjacent Storeys
The effect of low-to-high stiffness transitions at high velocities was observed and quantified by Wiebe
and Christopoulos (2010) using a two degree-of-freedom numerical model but it has never been studied or

observed experimentally. An additional study by the same authors (Wiebe & Christopoulos, 2011) found
that introducing rounded corners to provide a gradual stiffness transition to the hysteretic response of the
nonlinear element mitigates the acceleration spikes in the same two degree-of-freedom system model.
As shown in Table 6.12, the absolute storey accelerations that were recorded during testing were gener-
ally well-controlled and in the range of approximately 0.5g for all but the largest earthquake records (with
the notable exceptions of records 1-5 and 4-1 which are actually the same record for comparison, and the
2% in 50 year record number 2-4). Even those largest records showed accelerations under 1.0g in the worst
cases. However, as shown in Figure 6.24, the OpenSees and SAP2000 models tended to significantly over-
estimate the magnitude of these floor accelerations in most cases.
The acceleration responses of the test frame and the model were compared and investigated in detail to
determine the cause of this over-estimation. Figure 6.27 shows a comparison of the test versus model accel-
erations for the first floor of the frame subjected to 50% of the Northridge earthquake. Since the OpenSees
and SAP2000 model results were similar, only the OpenSees results are shown here for clarity. In general,
the time traces of the accelerations have a similar shape, except that the model response has a series of
spikes superimposed on the underlying acceleration pattern. To show the cause of these acceleration spikes,
the first floor acceleration response is magnified and compared to the forces that are present in the SCED
braces above and below the first floor level. The hysteretic responses of those SCED braces are also shown
at the top right of the figure.
Before the acceleration spike is present (prior to time A in the figure), the SCED braces both above and
below the first storey are returning towards zero brace deformation at low stiffness (post-activation stiffness
k a ). At time A, the first storey SCED mechanism closes (in both the test and the model), causing the stiff-
ness of the brace to revert to the higher initial stiffness k 1 . This causes the first storey brace to accumulate
force quickly; however, due to the inertia of the mass of the first floor, there is a time lag between the clo-
sure of the first and second storey brace mechanisms. The second storey brace is not able to close until time
B at which point the first storey brace has already activated. Between time A and time B, the force in the
second storey brace does not accumulate as quickly as the force in the first storey brace. This sudden, short
duration force imbalance results in a high acceleration between time A and time B that works to counteract
the force imbalance and pull the floor back into equilibrium. The difference between the axial force in the
first and second storey SCED braces is shown in the plot at the bottom left. When this force is projected
into the horizontal direction and divided by the floor mass, the resulting calculated acceleration is almost
identical to the measured acceleration as shown by the dashed lines in the first storey acceleration plot.

Test 2-1A - 50% ATC-N09B (1994 Northridge)

1% Rayleigh Damping
Test Brace Hystereses
100
First Story Floor Acceleration
1.0g
SCED 50
1st Story 0.5g
Force 2nd Story
Accel. A
0 (kN) 0
(Model)
B
-0.5g -50/100
-0.825g D D
-1.0g C B C
-100/50
1.0g
1st Story 0.5g SCED 0 A
Accel.
0 Force 1st Story
(Test) (kN) -50
-0.5g -0.468g
-1.0g -100
0 2 2.5 4 6 8 10 12 14 -10 -5 0 5 10
Time (s) SCED Deformation (mm)
0.5g 200
Model
1st Story 2nd Story SCED Activates
0 2nd Story 100
Accel.
SCED
Force 0
-0.5g Test 2nd Story SCED Closes
(kN) -100 (OpenSees)
*Dashed Lines Represent
SCED Force Difference / Mass 2nd Story SCED Closes (Test)
-1.0g -200
50 A BC D 200 A BC D
SCED 1st Story 100

0 SCED
Force 0
Difference Force 1st Story SCED Activates
-50 (kN) -100 1st Story SCED Closes
(kN)
-100 -200
2.0 2.1 2.2 2.3 2.4 2.5 2.0 2.1 2.2 2.3 2.4 2.5
Time (s) Time (s)
Figure 6.27: Sample Acceleration Spike Response
This is also the case for the acceleration of the second and third floors (not shown). This shows that the
acceleration in the real frame is mostly caused by this SCED force imbalance as suggested previously. For
the OpenSees model results, the measured and calculated accelerations are close but do not exactly match
up at the peaks. This difference between the two values is caused by shear force in the columns resulting
from the dynamic behaviour of the frame. At that particular time in the test, there was less shear force in
the test columns than there was in the model columns.
At time D, the first storey brace reverses direction and changes stiffness again. This change represents
the type of low-velocity stiffness transition that hysteretic-type systems generally experience. As the first
storey SCED force plot shows, even though the brace stiffness after the change at time D is the same as it
was after the stiffness change at time A, the force in the brace changes much more slowly because the veloc-
ity is at time D is lower. Accordingly, the force imbalance at time D is lower, as is the acceleration in both
the test and the model.

The peak acceleration is much lower in the test than in the model because the test braces provide
smoother transitions in time between the low and high stiffness. This gradual transition reduces the differ-
ence between the forces in the SCEDs above and below the first floor, thereby reducing the necessary coun-
teracting acceleration. Therefore, while the shake table tests have proved that the accelerations caused by
high velocity stiffness transitions exist and are significant, in practice, real self-centering systems have a
somewhat gradual stiffness transition which mitigates the severity of these accelerations. The consequence
of this is that models of self-centering systems that use sharp stiffness transitions will tend to significantly
over-estimate accelerations. Therefore, an engineer should exercise caution when designing secondary and
non-structural components of a building using the acceleration results from sharp-transition SCED brace
models since they are likely to be overly-conservative; however, it is important to note that since these over-
estimates are caused by overly sharp stiffness transitions, the accelerations provided by such models are
upper-bound estimates of the accelerations and may therefore still be useful in design.
To get a better estimate of the accelerations, a more realistic gradual transition could be modelled using
the method described by Wiebe and Christopoulos (2011). These results also suggest that, while high mag-
nitude acceleration spike can exist in self-centering systems, self-centering systems that are explicitly
designed to change stiffness gradually may significantly reduce the accelerations experienced by the system.
6.7.7 STOREY SHEAR RESULTS AND THE EFFECT OF INHERENT DAMPING
Similar to the storey drift results, the experimental storey shear results in Table 6.12 are in good agree-
ment with the predictions of the models. As mentioned previously, the inherent damping value of 1% of
critical for the models was selected since it results in the best predictions for the storey drifts and shears.
This inherent damping models all of the sources of damping that are not explicitly provided to the system
by the SCED braces; these sources include friction inside frame component connections, friction between
the test frame and the lateral support, and air resistance. However, this value of 1% for inherent damping
does not seem to be appropriate for all excitation amplitudes.
Towards the end of the time history in some of the models, when the model is responding predomi-
nantly in a damped free vibration mode, this low inherent damping value causes the storey shear to take an
unrealistically long time to damp out. Figure 6.28 shows the base shear response of the test frame subjected
to the 1979 Imperial Valley earthquake. Again, only the OpenSees model results are shown here for clarity.
The left side of the figure shows the response of the experiment compared to the response of the model
using 1% inherent damping, and the right side shows the same experimental data and the model using 2%

inherent damping instead. The experimental data shows that the free vibration response at the end of the
record quickly damps out in reality. The model under 1%, however, exhibits a significant vibration that is
not observed in the experimental results. Increasing the inherent damping to 2% reduces this effect, but
also reduces the peak storey shear.
Test 2-3A - 90% ATC-N01B (1979 Imperial Valley)

First Story Shear (Base Shear)
1% Rayleigh Damping 2% Rayleigh Damping
150
Test Reduction in Free
100 Vibration Amplitude
Model
with Higher Damping
50
First
Story 0
Shear -50
(kN)
-100 -127.5kN -123.3kN
-150 -137.6kN -137.6kN
0 5 10 15 20 0 5 10 15 20
Time (s) Time (s)
Figure 6.28: Reduction of Free Vibration Base Shear Amplitude with Added Inherent Damping
These results indicate that a damping model with low inherent damping for large cycles and signifi-
cantly higher inherent damping for the small cycles of response would best represent the damping behav-
iour that was observed during the shake table tests. The discrepancy between the two levels of response
with respect to inherent damping is likely occurring because the inherent damping in the shake table frame
is predominantly caused by friction. However, in the models, the inherent damping is modeled using the
Rayleigh damping model, which is proportional to the velocity. Friction causes a relatively greater amount
of inherent damping in the low magnitude cycles, when the velocity is low, than in the high cycles when
the velocity is higher. This trend is especially apparent in Figure 6.24 which shows that the model
responses decrease relative to the test responses as the maximum drift increases. This occurs because the
amount of viscous damping in the model effectively increases at higher drifts, which are associated with
higher velocities, whereas the amount of friction damping in the test does not increase for higher drifts.
These results emphasize the fact that inherent damping assumptions in dynamic models are critical to the
behaviour of the model. This is the true for all types of dynamic systems, not only SCED braced frames. It
is also important to note that the experimental SCED braced frame that was constructed for the shake
table test does not necessarily replicate realistic conditions for inherent damping in a structure since the
damping is likely primarily caused by friction between the frame and the lateral support.

6.8 CONCLUSIONS
The shake table tests that were conducted on a multi-storey SCED braced frame successfully demon-
strated that the SCED bracing system performed as designed when subject to a fully dynamic base excita-
tion and that the companion OpenSees computer model was capable of predicting the results of the tests.
The SAP 2000 model was able to predict the results equally well, with the limitation that it could not
model the external friction fuse behavior.
The braces were able to withstand twelve consecutive earthquakes without damage or loss of friction
fuse capacity. In addition, the external friction fuses that are in series with the braces worked as expected,
extending the drift capacity of the braces from 2% to 4% of storey height, while controlling the forces
imposed by the SCED braces on the adjacent members.
The peak drift results, peak storey shear results, and peak column axial loads were well-predicted by
both the OpenSees and the simplified SAP2000 numerical models of the structure. The time history
results for the storey shear in the model, however, showed an under-damped free-vibration-type response
following the strong motion part of the response. This may be due to the fact that the Rayleigh damping
model underestimates damping for low amplitude motions. This is an area of research that seems to war-
rant further study.
The storey accelerations in SCED-braced frames were found to be high due to the high-velocity stiff-
ness transitions that are characteristic of self-centering systems. These accelerations in the test and the
model were determined to be mostly caused by the force imbalances between the SCED braces in adjacent
storeys. The storey accelerations were generally over-estimated in the model due to the presence of sharp
transitions in the brace forces predicted by the SCED brace model. As suggested by Wiebe and Christo-
poulos (2011), using a more gradual stiffness transition in the model would allow it to more accurately pre-
dict the true floor accelerations.
Based on the results of these tests, self-centering energy-dissipative (SCED) braces may be considered
an effective stand-alone lateral force resisting system for multi-storey steel frames. They provide a high-per-
formance, fully self-centering response under design-level seismic hazards with a probability of exceedance
of 10% in 50 years and an increased performance level with reduced residual drifts when subjected to max-
imum-credible-level events with a probability of exceedance of 2% in 50 years. Additionally, design engi-
neers may feel confident that numerical models of SCED-braced systems that utilize either the

"SelfCentering" uniaxial material in OpenSees or the simulated SCED model in SAP2000 will be robust
and will provide meaningful predictions of the seismic response of buildings incorporating SCED braces.
There were, however, some problems with the quality control of the braces as-manufactured. In two of
the braces, the tubes were not the same length, requiring retrofit on-site. In addition, some of the tendons
were either not tensioned properly before the tubes were sent, or the tendons lost some of their pretension
in shipping. Both of these are serious issues that must be addressed in future projects.

CHAPTER 7: SCED FRAME BUILDING MODEL COMPARISONS
The SCED Mechanics Simulator that was described in Chapter 3 has provided a better detailed under-
standing of SCED and T-SCED mechanics. The test results from the HC-SCED and T-SCED braces
(Chapters 4 and 5) showed that the Mechanics Simulator was able to accurately predict the response of real
braces by including consideration of construction tolerances. In Chapter 6, the shake table tests proved
that numerical models of multi-storey SCED-braced frames provide a good prediction of real dynamic
SCED-braced frame behaviour.
To leverage the numerical work that has been described in those previous chapters, an existing six-storey
SCED-braced frame design was updated and extended to comparatively evaluate the seismic response of
different SCED frames using more realistic brace properties. For this purpose, new SCED braces were
designed and their hysteretic behaviour was evaluated using the SCED Mechanics Simulator. The resulting
hysteretic behaviour was then modelled using the nonlinear structural analysis package OpenSees to obtain
the dynamic response of the structures. Three types of SCED braces were designed: (1) the traditional
SCED configuration with two axial members (O-SCED), (2) the telescoping configuration with three
axial members (T-SCED), and (3) the telescoping configuration combined with viscous damping in paral-
lel (V-SCED).
This study had four primary goals. Since the original building model used assumed, not fully designed,
properties for the O-SCED braces, the first goal was to evaluate the behaviour of the existing model versus
the behaviour of a model that uses realistic, fully-designed and modelled, O-SCED braces. Next, since the
Mechanics Simulator has shown that there can be a significant difference between the theoretical and real-
istic initial stiffness of the SCED braces, the second goal was to study the effect that this difference makes
to the global behaviour of the structure. The third goal was to evaluate the behaviour of the braced frame
equipped with T-SCEDs instead of O-SCEDs. The last goal was to evaluate the benefit of adding viscous
dampers to every storey in parallel with the T-SCED braces to reduce the building’s storey accelerations.
274
Chapter 7: SCED Frame Building Model Comparisons 275
7.1 SIX-STOREY BUILDING DESIGN (CHOI ET AL., 2008)

The six-storey SCED-braced frame design for this numerical study was taken from the existing previous
study that was conducted by Choi et al. (2008). The buildings in this existing study were designed using
ASCE 7-05 Minimum Design Loads for Buildings and Other Structures (ASCE, 2005), AISC 360-05 Load
and Resistance Factor Design Specification for Structural Steel Buildings (AISC, 2005a), and AISC 341s1-05
Seismic Provisions for Structural Steel Buildings (AISC, 2005b).
This prototype building was designed for normal occupancy on class D soil in downtown Los Angeles,
California. The design was done using the modal response spectrum analysis procedure. The SCED braces
themselves were designed using the same response modification coefficient R = 7 , overstrength factor
 0 = 2 , and deflection amplification factor C d = 5.5 as those prescribed for buckling-restrained braced
frames in ASCE 7-05. All columns and beams were steel W-Sections. Concrete floor slabs acted as rigid
diaphragms at every storey. The total effective seismic weight of the structure W was 32 100kN. The full
design details may be found in Choi et al. (2008).
The plan and elevation of the six-storey building are shown in Figure 7.1. The building lateral force
resisting system consisted of SCED-braced frames in the North-South direction and special moment-
resisting frames (SMRFs) in the East-West direction. For the current study, only the SCED frame response
will be considered, meaning that the SCED frames were analyzed in 2D and the contribution of the
orthogonal SMRFs was neglected.
7.2 PREVIOUS STUDIES INTO COMBINING VISCOUS DAMPING WITH SELF-CENTERING

SYSTEMS
The use of viscous damping in structural systems is attractive because the forces in viscous dampers are
generally out-of-phase with the maximum forces elsewhere in a structural system. This is because the force
in a viscous damper is velocity-dependant, whereas the force in a structural member or in a friction or
yielding damper is displacement-dependant. Therefore, if a viscous damper is used in place of a friction or
yielding damper, in addition to reducing accelerations in a system, the viscous damper should also reduce
the forces in a system, if both systems provide the same amount of energy dissipation. To take advantage of
these properties of viscous dampers, this study will investigate their potential use by using viscous damping
to either fully or partially replace the friction damping in a SCED brace.
The use of viscous dampers as an energy dissipation mechanism for use with SCED braces was always
envisioned by its inventors (U.S. Patent No. US2008/0016794, 2008; Christopoulos et al., 2008); how-

Plan
N 5@9.144m = 45.72m
G1 G2 G2 G2 G1
C1 C2 C5 C5 C2 C1
G4 G6 G6 G6 G4
G3 G3 G3 G3 G3
C3 C4 C4 C4 C4 C3
3@9.144m
G5 G6 G6 G6 G5 = 27.43m
G3 G3 G3 G3 G3
Studied C3 C4 C4 C4 C4 C3
SCED- Gravity
Braced G4 G6 G6 G6 G4 Column
Frame G1 G2 G2 G2 G1 (typ.)
C1 C2 C5 C5 C2 C1
Column Girder Moment-Resisting Area Tributary to

Numbers Numbers Frame (typ.) Studied Frame
SCED Frame Elevation Gravity loads:

Roof: Dead = 4.07 kPa
9.144m 9.144m 9.144m Live = 0.96 kPa
Floor: Dead = 4.31 kPa
Live = 2.4 kPa
Column Seismic Load Data:

2.134m Splice Los Angeles, CA
SS = 2.1 g; S1 = 0.80 g
1.524m Occupancy Category II
6@3.658m Seismic Use Group I
= 21.95m Importance Factor = 1.0
Site Class D
Fa = 1.0; Fv = 1.5
SCED Shear SDS = 1.4 g; SD1 = 0.80 g
Brace G4 Typ. G5 Typ. G4 Typ. Connection Seismic Design Category E
(typ.)
21.8˚ Steel:
C1 C3 C3 C1
Fy = 345 MPa
Figure 7.1: Six-Storey SCED Building Design (after Choi et al., 2008)
ever, viscous damping has never been tested in a physical SCED system. There have been a few numerical
studies that have been conducted into the use of viscous damping with other self-centering systems, which
are summarized below.
Kim (2007), conducted a numerical study of a fully-designed, six-storey self-centering moment frame
building that used viscous dampers for energy dissipation. This frame was compared both to a similar self-
centering moment frame that used a friction mechanism to provide energy dissipation and a traditional
yielding moment frame. The frames were subjected to harmonic loading as well as a suite of dynamic
earthquake excitations. The study found that the frame with the viscous damping experienced smaller
drifts and accelerations than the other two frame types and was able to completely eliminate residual drifts.

Kam et al. (2010) conducted a dynamic single-degree of freedom (SDOF) study of self-centering sys-
tems with either friction (hysteretic) damping, viscous damping, or a combination of the two. Combina-
tions of the two types of damping were labelled “Advanced Flag-Shaped (AFS) systems”. They found that
the combined AFS systems tended to have the best response in terms of maximum drift, residual drift and
maximum acceleration without a significant increase in base shear. They showed that the response was
especially good if the peak viscous damper force was limited by a slipping friction device in series with the
damper.
A similar SDOF study which included both self-centering systems and bilinear elastoplastic systems
with added viscous damping was conducted by Karavasilis and Seo (2010). This study included consider-
ation of non-structural components and the effect of varying the strength ratio (equivalent to the response
modification coefficient R ) in design. They confirmed the results of the previous study. They also found
that increasing the strength ratio decreases accelerations, and that added damping is better at mitigating
accelerations in self-centering systems than in elastoplastic systems.
7.3 BUILDING MODELLING

A summary diagram of the 2D building model is shown in Figure 7.2. The structure was modeled and
analyzed using the structural seismic behaviour modelling software OpenSees (McKenna et al., 2000).
Similarly to the shake table model described in Section 6.5, the SCED braces were modelled using the
OpenSees material model that was developed as previously described in Section 3.6.3. Viscous dampers
were modelled using the plain Viscous material in OpenSees. All columns were considered to be continu-
ous with fixed column splices and pinned column bases. The column sections were modelled using a
lumped plasticity model with hinges at either end of each column. The moment-curvature response and
axial moment interactions of the column hinges were considered implicitly by using fibre sections at the
hinge locations. The nonlinear steel material that was used for these fibre elements was assumed to have a
post-yield stiffness equal to 2% of the initial stiffness. Beams were considered to be pin-ended and were
modelled using truss elements. All beam and column ends were assigned fixed end offsets to account for
beam and column depths at the connections.
To account for P-Delta effects in the model, four leaning columns were added. Each leaning column in
the model represented one type of gravity column in the building as shown in Figure 7.2. They each had
the combined properties of all the gravity columns of the same type. For example, the C4 leaning column
had four times the stiffness and area of column type C4 in the building because there are four C4 columns

Column Tributary Areas

C1 C2 C5
C3
C4
SCED
Braced C3
Frame
C1 C2 C5
Moment-Resisting Area Tributary to

Frame (typ.) Each Column
2D Model Elevation
Horizontal Translation
slaved to Master Leaning Columns
SCED Frame Storey Node (gravity nodes at each node)
* 3.658m
* 3.658m
*Master
Nodes
(with * 3.658m
lumped
mass) * 3.658m
* 3.658m
G5 Typ.
* 3.658m
C3 C3 2xC1 4xC4 2xC2 2xC5
9.144m
Figure 7.2: Six-Storey SCED Building OpenSees Model
in the area tributary to the frame as shown. Vertical loads were applied to each storey of each leaning col-
umn according to the amount of floor area that is tributary to each of the columns as shown in the figure.
For the model’s mass matrix, the storey mass for half of the building which includes the mass that is sup-
ported by all the gravity columns, was lumped together and assigned to the master node on the frame (see
Figure 7.2). A rigid diaphragm constraint was modelled at each floor to simulate the horizontal constraints
applied by the floor slabs.
Inherent damping was modeled using 3% of critical Rayleigh damping in Modes 1 and 2. This is the
same amount of damping that was assumed in the study by Choi et al. (2008).

7.4 DESIGN OF SCED BRACES

Previous numerical studies of SCED frame behaviour used brace properties and stiffnesses that were
based on previous experience up to that point (Choi et al., 2008; Tremblay et al., 2008); however, those
designs did not have completely realistic initial or post-activation stiffnesses. In particular, there was no
understanding at the time of the effect of construction length tolerances on the initial stiffness of SCED
braces. These previous designs also relied on high limits for the tendon deformation capacity. In addition
to these limitations of previous designs, there have been no equivalent studies to determine the behaviour
of a multi-storey building equipped with T-SCEDs or SCEDs combined with viscous damping.
Therefore, based on the results presented in Chapters 4 and 5, and using the SCED Mechanics Simula-
tor described in Section 3.4 and the design method outlines in Section 3.7, new realistic SCED braces were
designed. Both original SCEDs (O-SCEDs) with two axial members and new telescoping SCEDs (T-
SCEDs) with three axial members were designed for this study. In addition, the amount of internal friction
damper force in the T-SCED designs was adjusted in two ways to accommodate the addition of viscous
damping: either half of the friction was removed, or all of it was removed. This resulted in four design cases
that had to be modelled using the Mechanics Simulator for each floor.
The target design parameters for the SCED braces were chosen by matching the activation load of the
new braces to the activation load of original SCED braces in the Choi et al. (2008) report. As discussed
previously in Section 3.7.2, an alternative design could have targeted the same axial load at a given target
drift level instead of the same activation load. This may have led to more uniform accelerations and lower
drifts in the newly designed frames, and should be investigated as a design option in future studies. In
addition, it is important to note that the process of matching the activation loads results in SCED and T-
SCED designs that have different initial stiffness than the original Choi et al. (2008) SCED braces. If these
new initial stiffness were considered in the design, it would have resulted in modified design loads, and
hence different target activation loads for the braces. To facilitate comparison between the different
designs, this effect was not considered within the scope of this study.
The SCEDs that were designed in Choi et al. (2008) will hereafter be referred to as the C-SCEDs. For
the new SCED designs, it was assumed that larger tendons up to 31mm nominal diameter could be used
in order to minimize the cross-section size. Although this large-sized tendon has not been previously tested,
a future tendon test program should be able to qualify their use in future SCED braces. To provide a con-
servative design based on the past performance of the tendons, the tendon strain capacity was assumed to
be 1.7%. Adequate steel sections for the inner, intermediate and outer members were chosen based on the

axial load requirements and the geometric requirements that were dictated by the required number and size
of the tendons.
The resulting SCED brace design cross-sections are shown in Figure 7.3. The outer members for the T-
SCEDs were designed to use custom built-up square hollow sections. This was necessary due to the large
cross-sectional area required to accommodate the tendons. All of the other T-SCED members and the
members that were used for the O-SCEDs are available from commercial steel manufacturers. In addition,
since high capacity T-SCEDs would result in unreasonably large cross-sections, T-SCED designs for the
first two storeys use two braces on each side of the building instead of one.
O-SCEDs
Outer HSS457x457x12.7 HSS457x457x12.7 HSS406x406x19 HSS406x406x19 HSS406x406x19
Inner WWF400x157 WWF400x157 W310x158 W310x158 W310x107

Tendons 6-31mm 6-31mm 4-31mm 4-31mm 6-22mm
T-SCEDs
Outer (Built-up) HSS609x457x12.7 HSS609x457x12.7 HSS712x457x12.7 HSS712x457x12.7 HSS609x457x12.7
Intermediate HSS406x406x19 HSS406x406x19 HSS406x406x19 HSS406x406x19 HSS406x406x19
2x 2x
Inner W310x107 W310x107 W310x158 W310x158 W310x107

Tendons 2 x 6-22mm 2 x 6-22mm 2 x 4-31mm 2 x 4-31mm 2 x 6-22mm
Storeys 1 & 2 Storey 3 Storey 4 Storey 5 Storey 6
Figure 7.3: Six-Storey SCED Building OpenSees Model
In order to determine a best-estimate of the hysteretic response of each brace, they were all indepen-
dently modelled using the SCED Mechanics Simulator. The general inputs for these models that are com-
mon to all of the brace designs are shown in Table 7.1. The diagonal length available for the braces in the
prototype building is 9.85m. Based on previous designs and connection geometries, especially the T-SCED
frame design from Chapter 5, the available length for the brace tube, and therefore the tendons, was deter-
mined to be 7.0m for the T-SCED braces. For the O-SCED braces, the connections would be larger in
order to accommodate the external friction fuse; so, for those braces, the length of tubes and tendons was
assumed to be 6.5m. To take construction tolerances into account, it was assumed that the length of each

steel member could vary from the length of the other members by up to 1.5mm. This is equivalent to a
variation of 1/32 of an inch in either direction, which is the tolerance for columns with both ends finished
for contact bearing in the AISC Code of Standard Practice for Steel Buildings and Bridges (AISC, 2005c).
Table 7.1: Mechanics Simulator Model Inputs for Designed SCEDs

T-SCED
Parameter O-SCED
V-SCEDs
Inner Member Length L i 6500.0 mm 7000.0 mm

+/- 1.5mm +/- 1.5mm
Inner Member Area A i Table 7.2 Table 7.2

Intermediate Member Length L m - 7000.0 mm
+/- 1.5mm
Intermediate Member Area A m - Table 7.2
Intermediate Member Initial Position - 0 mm

Outer Member Length L o 6500.0 mm 7000.0 mm
+/- 1.5mm +/- 1.5mm
Outer Member Area A o Table 7.2 Table 7.2

Internal Damper Friction End 1 F I1 Table 7.2 Table 7.2
Internal Damper Friction End 2 F I2 Table 7.2 Table 7.2
Tendon Area A pt Table 7.2 Table 7.2
Total Tendon Pretension P p0 Table 7.2 Table 7.2

Connection Part 1 Area Table 7.2 Table 7.2
The specific Mechanics Simulator inputs for each storey’s brace for each design are shown in Table 7.2.
The first two design types shown in this table are for the O-SCED and T-SCED braces, which are designed
to have the same activation loads as the C-SCED (Choi et al., 2008) braces. The tendon pretension force
( P p0 ) and internal damper friction ( F I1 and F I2 ) in these braces were varied in order to attain the correct

activation force and an adequate energy dissipation capacity parameter  . The tendon pretension force
was permitted to be in the range of 25% to 40% of the nominal tendon capacity. Since each 3/4” friction
damper normal force bolt can generally provide a friction force of approximately 25kN, internal damper
friction F I1 and F I2 were kept in multiples of 50kN whereever possible so that only the number of bolts
at each end is varied, not the axial force on each bolt. The last two design types (the V-SCEDs) in the table
are the same as the second design type (T-SCED), but with lower internal damper friction F I1 and F I2 to
accommodate the addition of viscous dampers. The connection stiffnesses were determined based on past
experience. Higher capacity braces require stronger, and therefore stiffer, connections.
The resulting hysteretic properties of each type of SCED brace that was modelled are shown in Table
7.3. Each line in the table is the result of multiple analyses conducted with the Mechanics Simulator to
take into account multiple member length scenarios. The final value for each parameter was chosen by
selecting a characteristic value from all of those scenarios. For the realistic initial stiffness, the lowest value
from all the scenarios was chosen to provide the greatest possible difference between the realistic modelled
value and the theoretical value that would have been assumed in the past. For those models that use the
theoretical initial stiffness, this stiffness was calculated as the sum of the stiffnesses of all of the axial ele-
ments in parallel. The external fuse slip force was determined by finding the force in the brace at the ten-
don elongation limit. The activation forces listed are for slow loading, they do not include the viscous
damping force; however the maximum anticipated brace loads including viscous damping are included in
the right-most two columns of the table. The first of these two columns is the anticipated maximum load
if the brace and viscous damper are deformed together up to 2% storey drift at the effective angular fre-
quency associated with the first mode (which will be described in the following section). The second col-
umn provides the force that results from deformation occurring at the effective angular frequency
associated with the second mode. Note that is the combined force of both the viscous damper and the
SCED brace. The determination of the viscous damping constants will be described in detail in the follow-
ing section.
Table 7.3 shows that the post-activation stiffness k a of the C-SCED braces is significantly smaller than
that of the comparable newly-designed O-SCED braces. For example, the first storey O-SCED has 60%
more post-activation stiffness than the first storey C-SCED and the sixth storey O-SCED has 170% more
stiffness than the sixth storey C-SCED. Choi et al. (2008) does not provide the geometries for the axial
members or tendons for the C-SCED designs; however, there are three likely causes for this difference.
First, the C-SCED designs assumed a tendon elongation capacity of 2.1% whereas the O-SCED assumed

Table 7.2: SCED Brace Property Summary
Member Area Internal Damper

Tendon Tendon
Brace/ Friction
Ai , Am , Ao Area Pretension Connection
Storey F I1 , F I2 Area
(mm2) A pt P p0
(kN) (mm2)
(mm2) (kN)
Inner Interm. Outer End 1 End 2
O-SCED (REALISTIC INITIAL STIFFNESS)
S1/S2 20100 - 20700 450 450 1830 1053 15
S3 20100 - 20700 400 400 1830 894 15
S4 20100 - 26770 350 350 1220 787 12.5
S5 20100 - 26770 250 250 1220 617 12.5
S6 13600 - 26770 200 150 912 398 10
T-SCED (REALISTIC INITIAL STIFFNESS)
S1/S2 13600 26770 26431 250 250 912 557 15
S3 13600 26770 26431 200 150 912 398 15
S4 20100 26770 29050 300 300 1220 851 12.5
S5 20100 26770 29050 250 250 1220 617 12.5
S6 13600 26770 26431 200 150 912 398 10
V-SCED - NO FRICTION
S1/S2 13600 26770 26431 0 0 912 557 15
S3 13600 26770 26431 0 0 912 398 15
S4 20100 26770 29050 0 0 1220 851 12.5
S5 20100 26770 29050 0 0 1220 617 12.5
S6 13600 26770 26431 0 0 912 398 10
V-SCED - 50% FRICTION DAMPING
S1/S2 13600 26770 26431 150 110 912 557 15
S3 13600 26770 26431 100 75 912 398 15
S4 20100 26770 29050 150 150 1220 851 12.5
S5 20100 26770 29050 150 100 1220 617 12.5
S6 13600 26770 26431 100 75 912 398 10
1.7%. This lead to a higher possible initial tendon pretension strain in the C-SCED, leading to less
required tendon area and lower tendon stiffness. Second, the C-SCED designs assumed a lower tendon
material Young’s modulus than the O-SCED braces (93 GPa instead of 102GPa for the O-SCED), leading
to a lower tendon stiffness. Third, the C-SCED braces used higher energy dissipation capacity parameters
in their design which increased the proportion of the necessary activation load that was resisted by the fric-

Table 7.3: Complete SCED Brace Hysteretic Parameters

Post- Energy Viscous Maximum Maximum
External
Activation Initial Activation Dissipation Damping Axial Force Axial Force
Brace/ Fuse Slip
Force P a Stiffness k 1 Stiffness Capacity Constant 2% Drift 2% Drift
Storey Force P s
(kN) (kN/mm) ka Parameter CL (Mode 1) (Mode 2)
 (kN) (kN) (kN)
(kN/mm) (kNs/mm)
CHOI ET AL. SCED (C-SCED)
S1/S2 1964 494.7 17.2 1.0 3061 - 3132 -
S3 1703 468.4 14.9 1.0 2662 - 2715 -
S4 1481 411.4 13.6 1.0 2355 - 2404 -
S5 1141 361.2 9.5 1.0 1753 - 1786 -
S6 721 221.2 5.2 1.0 1056 - 1074 -
O-SCED (THEORETICAL INITIAL STIFFNESS)
S1/S2 1979 885.0 27.2 0.834 3891 - 3826 -
S3 1720 885.0 27.1 0.839 3790 - 3560 -
S4 1515 914.6 18.5 0.856 2740 - 2771 -
S5 1145 914.6 18.5 0.786 2540 - 2401 -
S6 760 766.3 13.9 0.808 1880 - 1704 -
O-SCED (REALISTIC INITIAL STIFFNESS)
S1/S2 1979 645.0 27.2 0.834 3891 - 3826 -
S3 1720 623.0 27.1 0.839 3790 - 3560 -
S4 1515 586.0 18.5 0.856 2740 - 2771 -
S5 1145 534.5 18.5 0.786 2540 - 2401 -
S6 760 388.7 13.9 0.808 1880 - 1704 -
T-SCED (THEORETICAL INITIAL STIFFNESS)
S1/S2 2 x 1070 2 x 1166.4 2 x 6.5 0.900 - - 3023 -
S3 2 x 760 2 x 1166.4 2 x 6.5 0.875 - - 2403 -
S4 1470 1161.4 8.6 0.790 - - 2054 -
S5 1130 1161.4 8.7 0.845 - - 1721 -
S6 760 1166.4 6.5 0.875 - - 1201 -
T-SCED (REALISTIC INITIAL STIFFNESS)
S1/S2 2 x 1070 2 x 337.0 2 x 6.5 0.900 - - 3023 -
S3 2 x 760 2 x 280.0 2 x 6.5 0.875 - - 2403 -
S4 1470 387.0 8.6 0.790 - - 2054 -
S5 1130 348.0 8.7 0.845 - - 1721 -
S6 760 280.0 6.5 0.875 - - 1201 -

Table 7.3: Complete SCED Brace Hysteretic Parameters

Post- Energy Viscous Maximum Maximum
External
Activation Initial Activation Dissipation Damping Axial Force Axial Force
Brace/ Fuse Slip
Force P a Stiffness k 1 Stiffness Capacity Constant 2% Drift 2% Drift
Storey Force P s
(kN) (kN/mm) ka Parameter CL (Mode 1) (Mode 2)
 (kN) (kN) (kN)
(kN/mm) (kNs/mm)
V-SCED - NO FRICTION, 1ST MODE VISCOUS DAMPING
S1/S2 2 x 575 2 x 228.4 2 x 6.5 0.01 - 9.03 2682 5235
S3 2 x 411 2 x 185.1 2 x 6.6 0.01 - 6.41 2084 3790
S4 870 278.7 8.7 0.01 - 4.96 1777 3139
S5 633 234.0 8.6 0.01 - 4.25 1463 2599
S6 411 185.1 6.6 0.01 - 3.21 1043 1898
V-SCED - 50% FRICTION DAMPING, 1ST MODE VISCOUS DAMPING
S1/S2 2 x 823 2 x 288.8 2 x 6.5 0.58 - 4.52 2729 3829
S3 2 x 588 2 x 239.8 2 x 6.6 0.56 - 3.21 2176 2853
S4 1183 339.6 8.7 0.49 - 2.48 1867 2427
S5 886 298.3 8.6 0.54 - 2.13 1540 1993
S6 588 239.8 6.6 0.56 - 1.61 1089 1428
V-SCED - NO FRICTION, 2ND MODE VISCOUS DAMPING
S1/S2 2 x 575 2 x 228.4 2 x 6.5 0.01 - 2.84 2117 2684
S3 2 x 411 2 x 185.1 2 x 6.6 0.01 - 2.01 1760 2083
S4 870 278.7 8.7 0.01 - 1.56 1499 1778
S5 633 234.0 8.6 0.01 - 1.33 1245 1461
S6 411 185.1 6.6 0.01 - 1.01 881 1043
V-SCED - 50% FRICTION DAMPING, 2ND MODE VISCOUS DAMPING
S1/S2 2 x 823 2 x 288.8 2 x 6.5 0.58 - 1.42 2550 2729
S3 2 x 588 2 x 239.8 2 x 6.6 0.56 - 1.01 2083 2177
S4 1183 339.6 8.7 0.49 - 0.78 1783 1867
S5 886 298.3 8.6 0.54 - 0.67 1477 1541
S6 588 239.8 6.6 0.56 - 0.51 1042 1090
tion elements. This again reduced the necessary tendon area, which led to lower stiffness. A combination of
these three factors likely contributed to the post-activation stiffness difference between the two designs.
The energy dissipation capacity parameter  for the newly designed O-SCED and T-SCED braces
were generally designed to be within a range of 80% to 90%. This was done to provide realistic designs for
the buildings because real SCED braces would be designed with such  values to account for non-struc-
tural component hysteretic behaviour as previously described in Section 3.7.3. The  values are not con-

stant for all the braces because the internal friction damping slip force was applied in multiples of 50kN as
described above.
7.4.1 DESIGN OF SCEDS WITH VISCOUS DAMPERS
As discussed, all of the SCED braces that were used for the six-storey model were fully designed such
that realistic hysteretic properties could be determined for use in the models. Those SCEDs were sized
based on a target activation force which was determined by Choi et al. (2008) using a modal response spec-
trum analysis with a response modification coefficient R equal to 7. To design a SCED brace with viscous
dampers, a rational method has been developed that will be presented in this section. Physically, these vis-
cous dampers may either be incorporated into the design of the brace itself, or may be installed in separate
braced-frame bays. Regardless, viscous dampers that are on the same storey as a SCED brace will act paral-
lel to that brace as if the devices were combined. For the purposes of this numerical study, the viscous
dampers were assumed to be coincident and parallel with the SCED braces and the parallel combination of
the two were considered to be a single V-SCED (viscous SCED) brace.
Since this numerical study is the first of its kind, only linear viscous dampers were considered in order
to reduce the number of design parameters. Hence, the goal of the design of the V-SCED viscous dampers
was to determine a set of linear viscous damping constants ( C L ) that results in reduced storey accelera-
tions, potentially reduced storey drifts, and does not significantly increase the building’s base shear. Select-
ing linear viscous damping constants that are too low would result in higher drifts and accelerations, while
selecting damping constants that are too high will cause the dampers to experience higher axial loads which
increase the capacity design forces and the base shear.
A traditional approach for designing viscous dampers for a structure would entail the determination of
the effective stiffness and mass properties of the structure which would then be used in conjunction with a
design response spectrum to determine the amount of viscous damping required to achieve a target dis-
placement or force level; however, since the goal of this pilot study was to compare different combinations
of SCED brace friction damping with added viscous damping, this simple method was not deemed to be
appropriate. Instead of explicitly attempting to reduce target response quantities, this study compares dif-
ferent SCED/viscous combinations with the aim of assessing the response quantity reductions that are
caused by the different combinations of friction and viscous damping.
Therefore, the selection of the linear viscous damping constant for the V-SCED braces is based on an
energy-dissipation equivalence between the new viscous dampers and a baseline SCED-braced frame

design. It may be summarized as follows. As a starting point, the building should be fully designed using
O-SCED or T-SCED dampers. For the designs presented in Table 7.3, the realistic initial stiffness T-
SCED design was used as this starting point. Then, a target design drift level for the structure must be
selected. This target drift determines the total amount of energy that is dissipated per loading cycle for the
starting-point T-SCED designs. Since T-SCED braces tend to spread the drift demand well across the
height of a building (as will be seen), the same drift demand may potentially be used for the T-SCEDs at
every storey. Then, either a portion of that energy-dissipation or all of it is removed from the T-SCED
braces and replaced with a viscous damping to dissipate the energy instead. The energy-dissipation capacity
of the T-SCED braces is removed by reducing the internal friction damper slip force or eliminating the
internal friction dampers all together. As previously mentioned, for this numerical study either half of the
energy dissipation was removed from the T-SCEDs or all of it was. These reduced energy T-SCEDs were
then remodeled using the SCED Mechanics Simulator to attain realistic initial stiffness values as provided
in Table 7.3. To replace the removed T-SCED energy dissipation with viscous damping, an equivalent lin-
ear viscous damping constant C L must be selected for each brace. Since viscous damper energy is velocity
dependant, the damping constant cannot be determined without the selection of a viscous damper loading
frequency. Therefore, an effective frequency is determined based on the modal properties of the building
and the effective stiffness of each T-SCED in the starting-point structure. Using this effective frequency
and the target energy dissipation, the viscous damping constant for each V-SCED brace may be calculated.
Since the goal of adding viscous damping to the SCED-braced frames is to reduce the accelerations, a
more traditional design approach may have been used instead of the energy-dissipation equivalence
method described above. This would have required an equivalent single-degree-of-freedom (SODF) linear-
ization of the structure and the use of response spectra to determine the amount of damping that would be
required to achieve a target maximum acceleration; however, this method would not provide a clear, direct
method to distribute the viscous damping among the storeys of the building. The advantage of using the
energy-dissipation equivalency is that it permits better control of the damping that is present in each storey.
By designing viscous dampers to provide the same amount of energy dissipation at a given drift level and
modal period, the accelerations are expected to decrease because the viscous damping is out-of-phase with
the damping and restoring forces provided by the SCED braces. This has the potential to provide a more
efficient energy damping mechanism for the same base shear.
Having outlined the full procedure, the details of each step in the viscous damper design process will be
described below. For all of the V-SCEDs in this study, the target drift was set to 2.0%, which was the storey

drift limit for this structure according to ASCE 7 (ASCE, 2005). Based on the associated design-level brace
deformation (  d ), the energy dissipated under a single cycle of hysteretic response for a SCED brace
( E flag ) is depicted in Figure 7.4. It is simply equal to the area within the hysteresis. This energy was calcu-
lated using the areas shown in the bottom right plot with reference to the parameters shown in the bottom
left plot of the figure:
Pa + Pd
E flag = 2 ---  a P a +   d –  a   -----------------  – ---   1 –   a    1 –  P a 
1 1
(Eq. 7.1)
2  2  2
 1 –   + P d P a
–
–   d –  a –  1 –   a   -------------------------------------------------
a
 2 
P d – P a + P d
–   d –   d –  a    --------------------------------
 2 
which simplifies to:
E flag = 2  P a  d – P d  a  (Eq. 7.2)
Also, since:
P
 a = -----a (Eq. 7.3)
k1
Pd = Pa + ka  d – a  (Eq. 7.4)
Combining equations 7.2, 7.3 and 7.4:
k P
E flag = 2P a  1 – ----a-    d – -----a  (Eq. 7.5)
 k1   k1 
where  is the energy dissipation capacity parameter, P a is the activation force, k 1 is the initial stiffness,
k a is the post-activation stiffness and  d is the brace elongation at the design-level drift.
The energy dissipated by a viscous damper is dependant on the both the maximum deformation of the
damper and the velocity of that deformation. For the purposes of design, it was assumed that the deforma-
tion history demand for the viscous dampers could be represented by an effective sinusoidal loading:
  t  =  d sin  t  (Eq. 7.6)
where  d is the elongation at the design-level drift and  is the angular frequency of the deformation his-
tory. For a sinusoidal excitation, the energy dissipated by the viscous damper may be determined as follows

T-SCED Brace Hysteresis Linear Viscous Damper Hysteresis

Force
Pd Hysteretic Force
Hysteretic Energy
Energy Pa βPa
Eflag ksec,d Ev δdCLω
1
δd δd
δa Displacement Displacement
Hysteresis Outline Hysteretic Energy Calculation

Force [δd ,Pd ] Force
Add total area
under backbone
ka
[δa ,Pa ] 1
k1 [δd -βδa ,Pd -βPa ] Subtract area

1 underneath
hysteresis
[(1-β)δa ,(1-β)Pa ]
[0,0] Displacement Displacement
Figure 7.4: Energy Dissipated by the SCED versus a Linear Viscous Damper
(adapted from Christopoulos & Filiatrault, 2006). The sinusoidal deformation history from Equation 7.6
may be differentiated to find the velocity:
·
  t  =  d  cos  t  (Eq. 7.7)
The force in a linear viscous damper P v is equal to the velocity multiplied by the linear damping constant
CL :
·
Pv  t  = CL   t  (Eq. 7.8)
Combining equations 7.7 and 7.8:
P v  t  = C L  d  cos  t  (Eq. 7.9)
Therefore, the peak force in the viscous damper is equal to:
P v max =  d C L  (Eq. 7.10)

as shown in the top-right plot of Figure 7.4. The energy dissipated by one cycle of this viscous response as
shown in the figure may be found by calculating the area within the ellipse bounded by   d on the hori-
zontal axis and   d C L  on the vertical axis:
2
E v = C L  d (Eq. 7.11)
where C L is the linear viscous damping constant,  d is the elongation at the design-level drift and  is the
angular frequency. This viscous damping energy will be equated to the flag hysteresis energy that was
removed from the T-SCED brace designs; however, this cannot be done without knowing the angular fre-
quency  .
Since the dynamic behaviour of the resulting V-SCED designs is not known, the viscous dampers were
designed by assuming that the dynamic behaviour of the V-SCED frame would be similar to the dynamic
behaviour of the starting-point realistic initial stiffness T-SCED frame. The only sinusoidal frequencies
that are defined for that starting-point structure are those that come from the linear modal analysis; how-
ever, the modal analysis uses only the initial stiffnesses of all the model elements, and, hence, does not take
into account the nonlinearity of the T-SCED braces. This nonlinear behaviour causes the stiffness and,
therefore, the frequency, to decrease when the brace is activated. The viscous damper design must take this
effect into account. Therefore, instead of using the high initial stiffness, which will generally overestimate
the frequency, or the lower activated stiffness, which will generally underestimate the frequency, an effec-
tive stiffness is calculated by finding the secant stiffness of the hysteresis at the design deformation as
shown in Figure 7.4:
P
k sec d = ----d- (Eq. 7.12)
d
This effective secant stiffness may then be used to find an effective frequency by relating it to the frequency
of one of the building’s natural modes of vibration. This is done using a similar method to the one that was
used to find the effective period at different design drift levels for the HC-SCED test protocol as described
in Section 4.4.8:
2 k sec d
 eff = ------ -----------
- (Eq. 7.13)
Ti k1
where  eff is the effective angular frequency, T i is the building period for mode i , k sec d is the secant
stiffness of the SCED brace at the design drift (as shown in the top-left plot of Figure 7.4), and k 1 is the

initial stiffness of the SCED brace. Although all of the V-SCED brace designs in a structure should be
based on the same modal period, the effective frequencies used for design will still be different for each
brace because the effective secant brace stiffnesses will be different. Therefore, these effective frequencies
therefore somewhat represent the effective dynamic response of each individual brace; however, in practice,
the design effective frequencies were similar for all of the braces in each V-SCED building design.
It seems obvious that the dynamic behaviour of the starting-point T-SCED structure would be best rep-
resented by calculating the effective frequency (Equation 7.13) by using the first mode frequency of the
structure; however, especially for taller buildings, higher modes may have a significant effect on the
dynamic behaviour of the lateral load resisting system. To check the effect of the higher modes on the T-
SCED structure, the participating mass ratios for each mode were calculated using methods that may be
found in the dynamics textbook by Chopra (2000). These ratios are shown in Table 7.4. As this table
shows, the first mode only accounts for 74% of the mass, suggesting that the second mode, which accounts
for almost 19% of the mass has a significant effect on the dynamic response of the structure. The first and
second modes together account for 92% of the mass in the modal analysis.
Table 7.4: Participating Mass Ratios for the

Starting-Point T-SCED Structure
Participating
Mode Period (s)
Mass Ratio
1 0.858 73.8%
2 0.273 18.5%
3 0.158 4.3%
4 0.113 2.0%
5 0.089 0.9%
6 0.069 0.5%
TOTAL 100%
Since the behaviour of the viscous dampers is velocity dependant, the effect of the second mode may be
investigated by observing the deformation velocities that the starting-point T-SCED braces experience
under nonlinear dynamic analyses and comparing these to the velocities that are assumed during the
damper design. The maximum velocity that each brace is expected to experience may be calculated by
modifying Equation 7.7:

·
 max =  d  eff (Eq. 7.14)
Using the effective angular frequencies calculated using Equation 7.13 for both the first and second mode
periods, the maximum expected velocities may then be determined. These are given in Table 7.5 for each
storey of the starting-point T-SCED design. In the table, these expected velocities are compared to the
interstorey velocities that were experienced by the full nonlinear realistic initial stiffness T-SCED model.
The results shown for the model are the mean of the peak interstorey velocities for a suite of twenty design-
based seismic hazard level earthquakes (DBE). The earthquake records and full dynamic results will be dis-
cussed in later sections; however, the DBE earthquakes produced a mean peak drift of approximately
2.0%, which was the design drift that was assumed for the V-SCED design. This table shows that the
interstorey velocities that result from the dynamic model are closer to the velocities that are based on the
second mode effective frequency than they are to those based on the first mode effective frequencies. This
seems to indicate that the higher modes may have a significant effect on the velocities experienced by the
T-SCED braces.
Table 7.5: Velocity Comparison for the Starting-Point T-SCED Structure

Expected Expected Mean of the
Velocity based Velocity based Peak Interstorey
on First Mode on Second Mode Velocities for all
Storey
Effective Effective DBE-level
Frequencies Frequencies Earthquakes
(mm/s) (mm/s) (mm/s)
1 127 404 392
2 127 404 449
3 124 394 567
4 138 438 565
5 133 423 576
6 124 394 574
Since the higher mode response of the structure does seem to have an effect on the behaviour of the
braces, the separate V-SCED buildings will be designed using both the first mode effective frequency and
the second mode effective frequency. Knowing these effective angular frequencies, the linear viscous damp-
ing constant ( C L ) may be calculated by rearranging Equation 7.11 and equating the viscous damping
energy to the amount of energy-dissipation capacity that was removed from the starting point T-SCED
braces. As mentioned previously, for this study either half of the energy dissipation capacity was removed

from the T-SCED braces or all of the energy dissipation capacity was removed. This energy dissipation
capacity was removed by reducing or eliminating the internal friction damper slip force. For the two differ-
ent scenarios given above, the amount of energy dissipation removed ( E flag d ) that will be replaced by vis-
cous damping is equal to:
E flag
E flag d = E flag -or- E flag d = ---------
- (Eq. 7.15)
2
where E flag is calculated using Equation 7.5. Equating this energy that was removed from the T-SCED to
the viscous energy at a given the effective angular frequency provides an equation which solves for the
required linear viscous damping constant C L :
E v = E flag d (Eq. 7.16)
E flag d
C L = ------------------2- (Eq. 7.17)
eff  d
The resulting viscous damping constants for the different designs are shown in Table 7.3. For each of the
two reduced energy dissipation T-SCED brace design cases that were used (no internal friction damping
and 50% internal friction damping), a separate set of viscous dampers was designed for the effective fre-
quency that was determined using the building’s first mode period and the effective frequency that was
determined using the second mode period. This resulted in four different design cases that combined T-
SCEDs with viscous dampers (referred to as V-SCED cases). To refer to these cases in shorthand, the two
designs with no friction will be referred to as V-SCED (0-M1) and V-SCED (0-M2) for the first and sec-
ond mode effective frequency designs, respectively. Similarly, the two designs with half of the internal fric-
tion damping remaining will be referred to as V-SCED (50-M1) and V-SCED (50-M2).
The anticipated maximum axial forces for the V-SCED designs using both the first and second mode
effective frequencies at a design drift of 2% are shown in the two right-most columns of Table 7.3. As the
shaded cells in the table show, when the first mode designs are subjected to a deformation history at the
first mode effective frequency they have approximately the same anticipated maximum axial force as the
second mode designs subjected to the second mode effective frequency. These values represent effective
design loads for the braces that result from the viscous damper design method presented above. These
loads are all lower than the maximum anticipated axial loads of the starting-point T-SCED designs. The
true maximum axial loads that the V-SCED braces will experience during seismic loading depends on the
dynamic behaviour of the structure subject to the earthquake accelerations and on the maximum lateral

deformations of the frame. The earthquake analysis base shear results that will be presented in Section 7.8
will provide a effective metric for these dynamic loads in the V-SCED braces and other structural elements
so that they may be compared to the starting-point T-SCED design.
The V-SCEDs that were designed using the first building mode required larger viscous damping con-
stants then those that were designed using the second mode. This is because the first mode period is longer
than the second mode period, meaning that it also has a slower peak velocity. Therefore, to attain the same
amount of energy dissipation in one cycle of response at the given design drift, the first mode design needs
a much higher viscous damping constant C L than the second mode design. In the dynamic analyses, it was
observed that the two designs encountered similar interstorey velocities during earthquake loading. There-
fore, the first mode designs have larger forces than the second mode designs and dissipate more energy.
7.5 BUILDING PERIODS AND PUSHOVER ANALYSES

The elastic periods of the first three modes of the different design structures are shown in Table 7.6.
The first and second mode periods of the T-SCED brace structure with realistic initial stiffnesses were the
periods that were used for the design of the viscous dampers as described in the previous section. The table
shows that the first, second, and third modes were similar across all of the analyses except for the theoreti-
cal initial stiffness O-SCED and T-SCED, which had significantly higher initial stiffness than the other
models because they did not take into account the effect of construction tolerances.
Table 7.6: Analysis Model Periods

Mode 1 Mode 2 Mode 3
Analysis
T 1 (s) T 2 (s) T 3 (s)
Choi et al. SCED (C-SCED) 0.90 0.30 0.17
O-SCED (Theoretical) 0.76 0.22 0.12
O-SCED (Realistic) 0.83 0.26 0.14
T-SCED (Theoretical) 0.68 0.18 0.09
T-SCED (Realistic) 0.86 0.27 0.16
V-SCED (No Friction) 0.89 0.29 0.17
V-SCED (50% Friction) 0.96 0.32 0.19
Single direction pushover analyses were conducted for each design and the results are shown in Figure
7.5. These force-controlled pushovers were conducted using an ASCE-7 force distribution over the height
(ASCE, 2005). The pushovers show that all of the full friction designs transition between the initial build-

ing stiffness and the post activation stiffness at approximately the same level of base shear. This shows that
all of those analyses have the same activation base shear which was the design intention. In addition, the T-
SCED with no internal friction damping activates at approximately half of that base shear and the half-
friction T-SCED activates at a base shear that is halfway between those two.
4000
Pushover Analysis Comparison
O-SCED T-SCED
3500 Theoretical Theoretical
O-SCED T-SCED
3000 True
True
2500
C-SCED
Base (Choi et al. 2008)
Shear 2000
(kN) T-SCED Theoretical
O-SCED Theoretical
1500 T-SCED O-SCED True
Half-Friction T-SCED True
C-SCED
1000 T-SCED
No Friction
T-SCED
500 Half-Friction
T-SCED
No Friction
0 200 400 600 800 0 10 20 30 40
Top Floor Displacement (mm)
Figure 7.5: Building Model Pushover Analysis Comparison
The initial pushover stiffnesses may be compared by observing the inset figure at the lower right. The
O-SCED and T-SCED models that used the theoretical initial stiffness that was derived from the sum of
the stiffness of all the axial elements had the highest initial stiffness out of all the analysis models. The the-
oretical T-SCED had the highest initial pushover stiffness because it includes a third intermediate member
which significantly increases the theoretical stiffness relative to comparable O-SCEDs. The initial pushover
stiffnesses of the O-SCED and T-SCED models that use the realistic initial stiffness were significantly less
than the theorectical models and similar to each other. The C-SCED model had a somewhat lower initial
stiffness than the newly designed realistic O-SCED and T-SCED models, but had a similar magnitude.
The no-friction and half-friction T-SCED models had lower initial stiffness because the effect of the axial
member length tolerance on the stiffness was greater when the activation force is lower.

The pushover figure shows that the original Choi et al. SCED designs significantly underestimated
both the initial and post-activation stiffness when compared to fully designed O-SCEDs. The post-activa-
tion stiffness of the Choi et al. design was actually similar to the fully designed T-SCED model. Post-acti-
vation stiffnesses did not change between the realistic and theoretical stiffness models, which is reflected in
the pushover curves. The original Choi et al. SCED and the O-SCEDs were equipped with external fric-
tion fuses, and therefore has a second stiffness transition when those fuses slip. The stiffness of the individ-
ual fuses is zero when they slip; however, the overall stiffness of the pushover becomes negative at that point
because of the influence of P-Delta effects.
7.6 BUILDING DAMPING

To assess the amount of damping that was added by the viscous dampers, the percentage of critical
damping  for each structure was determined using a log decrement method (Chopra, 2000):
ui
 = -------- ln  ---------- 
1
(Eq. 7.18)
2j  u i + j 
where u i is the maximum system deformation for cycle i , and u i + j is the maximum system deformation
for cycle i + j . The decrement was found using a low amplitude free-vibration analysis for each building
model. The model was first laterally deformed using forces that were applied with relative values at each
storey according to the vertical distribution of seismic loads from the ASCE 7 equivalent lateral force pro-
cedure (ASCE, 2005). The lateral forces were increased using a slow ramp function until a total lateral load
of 100kN was achieved. Then, the forces were released and the free vibration response of the building
model was observed. The lateral deformation of the top of the building was used to calculate the damping
in Equation 7.18.
The resulting percentage of critical damping for each building model is shown in Table 7.7. These val-
ues reflect the damping that is present in the system when the SCED braces are in their linear initial stiff-
ness range. Therefore, they do not reflect the damping provided by the friction elements in the SCED and
T-SCED braces. Accordingly, the damping in all of the structures that do not contain any viscous dampers
is equal to the 3% inherent Rayleigh damping that was assumed to be present in the structures as previ-
ously described in Section 7.3. The percentage of critical damping provided by the viscous dampers varies
in relation to the viscous damping constants of the dampers (see previous Table 7.3); however, the damp-
ing constants are not directly proportional to the percentage of critical damping because the percentage
damping is also dependant on the stiffness of the structure, which varies for the different model buildings.

Table 7.7: Percentage of Critical Damping in Each Building Model

(values do not include the effect of friction damping from the SCED braces)
Percentage Damping
of Critical Provided by
Analysis Model Damping Viscous
Total  Dampers
Choi et al. SCED (C-SCED) 3.0% -
O-SCED (Theoretical) 3.0% -
O-SCED (Realistic) 3.0% -
T-SCED (Theoretical) 3.0% -
T-SCED (Realistic) 3.0% -
V-SCED - No Friction, 1st Mode Damping 7.3% 4.3%
V-SCED - 50% Friction, 1st Mode Damping 4.7% 1.7%
V-SCED - No Friction, 2nd Mode Damping 4.5% 1.5%
V-SCED - 50% Friction, 2nd Mode Damping 3.8% 0.8%
It was not possible to determine a percentage damping due to the SCED brace friction damping using
this free-vibration method because the effective damping provided by the SCED brace is amplitude-depen-
dant.
7.7 TIME HISTORY GROUND MOTIONS

The time history analyses were performed using a suite of sixty total ground motions that were taken
from the SAC joint venture project (Somerville et al., 1997). The SAC records are calibrated for the Los
Angeles area and split into three different hazard levels. Twenty records (LA41 to LA60) represent the fre-
quently occurring earthquake (FOE) seismic hazard level with a probability of exceedence of 50% in 50
years. An additional twenty records (LA01 to LA20) represent the design basis earthquake (DBE) seismic
hazard level with a probability of exceedence of 10% in 50 years. The final twenty records (LA21 to LA40)
represent the maximum considered earthquake (MCE) seismic hazard level with a probability of
exceedence of 2% in 50 years. Some details about the individual records contained in the FCE, DBE and
MCE suites are shown in Tables 7.8, 7.9 and 7.10, respectively. Figure 7.6 shows the pseudo-spectral accel-
eration plots for all the records along with the mean spectrum of all of those records for each hazard level
and the code design spectra from ASCE 7 (ASCE, 2005).

Table 7.8: Frequently Considered (FOE) Hazard Level Earthquake Records

(50% Probability of Exceedence in 50 years)
Dist. PGA
Name Record Magnitude
(km) (g)
LA41/42 Coyote Lake, 1979 5.7 8.8 0.59/0.33
LA43/44 Imperial Valley, 1979 6.5 1.2 0.14/0.11
LA45/46 Kern, 1952 7.7 107 0.14/0.16
LA47/48 Landers, 1992 7.3 64 0.34/0.31
LA49/50 Morgan Hill, 1984 6.2 15 0.32/0.55
LA51/52 Parkfield, 1966 6.1 3.7 0.78/0.63
LA53/54 Parkfield, 1966 6.1 8.0 0.69/0.79
LA55/56 N. Palm Springs, 1986 6.0 9.6 0.52/0.38
LA57/58 San Fernando, 1971 6.5 1.0 0.25/0.23
LA59/60 Whittier, 1987 6.0 17 0.77/0.48
Table 7.9: Design-Basis (DBE) Hazard Level Earthquake Records

Dist. PGA
(km) (g)
LA01/02 Imperial Valley, 1940, El Centro 6.9 10.0 0.46/0.68
LA03/04 Imperial Valley, 1979, Array #05 6.5 4.1 0.39/0.49
LA05/06 Imperial Valley, 1979, Array #06 6.5 1.2 0.30/0.23
LA07/08 Landers, 1992, Barstow 7.3 36.0 0.42/0.43
LA09/10 Landers, 1992, Yermo 7.3 25.0 0.52/0.36
LA11/12 Loma Prieta, 1989, Gilroy 7.0 12.0 0.67/0.97
LA13/14 Northridge, 1994, Newhall 6.7 6.7 0.68/0.66
LA15/16 Northridge, 1994, Rinadi RS 6.7 7.5 0.53/0.58
LA17/18 Northridge, 1994, Sylmar 6.7 6.4 0.57/0.82
LA19/20 North Palm Springs, 1986 6.0 6.7 1.02/0.99
7.8 TIME HISTORY ANALYSIS RESULTS

The sixty different earthquake records from each of the three seismic hazard levels described above were
applied to each of the different building designs that were previously shown in Table 7.3. The results of
these analyses and the trends that were observed from these results are described below.

Table 7.10: Maximum Considered (MCE) Hazard Level Earthquake Records

Dist. PGA
(km) (g)
LA21/22 Kobe, 1995 6.9 3.4 1.28/0.92
LA23/24 Loma Prieta, 1989 7.0 3.5 0.42/0.47
LA25/26 Northridge, 1994 6.7 7.5 0.87/0.94
LA27/28 Northridge, 1994 6.7 6.4 0.93/1.33
LA29/30 Tabas, 1974 7.4 1.2 0.81/0.99
LA31/32 Elysian Park (Simulated) 7.1 17.5 1.30/1.19
LA37/38 Palos Verdes (Simulated) 7.1 1.5 0.71/0.78
LA39/40 Palos Verdes (Simulated) 7.1 1.5 0.50/0.63
Frequently-Occurring Design-Basis Maximum-Considered

Earthquakes (FOE) Earthquakes (DBE) Earthquakes (MCE)
50% in 50 year Hazard 10% in 50 year Hazard 2% in 50 year Hazard
5.0
4.5
4.0
3.5
Pseudo- 3.0
Spectral
Acceleration 2.5 Individual ASCE 7
(g) 2.0 Spectra Code
Spectrum
1.5 Mean
1.0 Spectrum
0.5
0 1 2 3 4 0 1 2 3 4 0 1 2 3 4
Period (s)
Figure 7.6: Spectral Acceleration Plots for the Considered Earthquake Records
7.8.1 INDIVIDUAL RECORD RESPONSE AND HYSTERESES
The differences between the different building designs may best be seen by comparing sample hysteretic
responses for the SCED braces in each different design. Sample hystereses for the first storey brace in each
of the nine different designs subject to a typical maximum considered earthquake (MCE-LA21) are shown
in Figure 7.7. As mentioned previously, all of the O-SCEDs and T-SCEDs and the Choi et al. SCED had
the same activation force. The Choi et al. SCED and the other O-SCEDs were equipped with external fric-

tion fuses, which limited the force in the braces and resulted in residual drifts under high-amplitude seis-
mic loadings such as this one. The O-SCEDs had higher post-activation stiffness than the Choi et al.
SCED because the O-SCEDs were fully designed using realistic properties as described in Section 7.4. The
T-SCEDs had a lower post-activation stiffness because the tendons in those braces were effectively twice as
long due to the telescoping mechanism. The figure also shows that there was very little difference between
the hysteretic response of the O-SCEDs and T-SCEDs that used the realistic initial stiffness versus those
that used the theoretical initial stiffness. The four V-SCED designs near the bottom of the figure demon-
strate the characteristic hysteretic response of a T-SCED brace in parallel with a viscous damper. It can be
seen in this sample comparison that, for this record, the V-SCED braces that used viscous dampers that
were designed using the second mode period have less energy dissipation than the first mode period
designs. This is because the second mode designs use smaller linear viscous damping constants as previ-
ously described in Section 7.4.1.
The response of the V-SCED braces is shown in greater detail in Figure 7.8. This figure shows that each
total hysteretic plot for the V-SCED (the left column of plots in the figure) represents a superposition of
two different hysteretic behaviours: the one of the SCED brace and the one of the viscous damper (the
centre and left columns of the figure, respectively). The two top V-SCEDs with no internal friction damp-
ing have a SCED hysteretic response that is effectively a bilinear elastic curve. This is a flag hysteresis with
an energy dissipation capacity parameter (  ) value of zero. The hystereses shown behind these plots in grey
are the behaviour of the full-friction starting-point T-SCED brace. The activation load of the bilinear elas-
tic curve is lower than the T-SCED because the activation force is equal to the sum of the tendon preten-
sion and the internal friction damper slip force. Without the internal damper force, the activation force
drops. The two bottom V-SCEDs with 50% internal friction damping have a SCED hysteretic flag width
that is half that of the starting-point T-SCED. The viscous damper hystereses that are shown in the right
column of plots show the typical behaviour of a viscous damper subjected to an earthquake-derived defor-
mation history. As discussed earlier, the first mode period viscous designs resulted in higher viscous damp-
ing constants and, therefore, experience higher forces and also dissipate more energy than the second mode
period designs; however, the first mode designs also result in greater floor shears.
While the previous figures showed the response of individual braces under a characteristic maximum
credible earthquake (MCE) seismic hazard level record, the plots in Figure 7.9 show the effect of varying
seismic hazard on selected designs. The earthquakes that were selected for these plots were LA51 for the
frequently occurring earthquake (FOE) hazard level, LA01 for the design basis earthquake (DBE) hazard

First Storey SCED Brace Hysteretic Response Comparison

MCE-Level Earthquake LA21 - 1995 Kobe M6.9
O-SCED O-SCED
C-SCED (Choi et al. 2008) Theoretical Stiffness Realistic Stiffness
5000
3000
Brace
Axial 1000
0
Force -1000
(kN)
-3000
-5000
-150 -100 -50 0 50 100 150
Brace Axial T-SCED T-SCED
Deformation (mm) Theoretical Stiffness Realistic Stiffness
5000
3000
Brace
Axial 1000
0
Force -1000
(kN)
-3000
-5000
V-SCED No Friction V-SCED No Friction

with Mode 1 Viscous with Mode 2 Viscous
5000
3000
Brace
Axial 1000
0
Force -1000
(kN)
-3000
-5000
V-SCED 50% Friction V-SCED 50% Friction

with Mode 1 Viscous with Mode 2 Viscous
First-Storey 5000
SCEDBrace 3000
Brace
Axial 1000
0
Force -1000
(kN)
-3000
-5000
-150 -100 -50 0 50 100 150 -150 -100 -50 0 50 100 150
Figure 7.7: Hysteretic Response Comparison for First Storey Brace (MCE Earthquake LA21)
level, and LA21 for the MCE hazard level (see Tables 7.8, 7.9 and 7.10). The external friction fuses in the
C-SCED and the O-SCED typically only activate during MCE level excitations. The O-SCEDs and T-
SCEDs have the same backbone shape regardless of the hazard level. Only the maximum deformation
changes, the activation load always stays the same. The V-SCEDs, on the other hand, have lower effective
axial loads for lower seismic hazard levels because those lower levels produce lower drift velocities and the

First Storey V-SCED with Viscous Damping

Total Hysteresis = SCED with Friction + Viscous Damper

V-SCED No Friction with Viscous from 1st Mode Period
5000
3000
1000
0
-1000
= Starting-
+
-3000 Point
T-SCED
-5000 Hysteresis
V-SCED No Friction with Viscous from 2nd Mode Period

5000
3000
1000
0
-1000
= +
-3000
Brace -5000
Axial
Force V-SCED 50% Friction with Viscous from 1st Mode Period
5000
(kN)
3000
1000
0
-1000
= +
-3000
-5000
V-SCED 50% Friction with Viscous from 2nd Mode Period

5000
3000
1000
0
-1000
= +
-3000
-5000
-150 -100 -50 0 50 100 150 -150 -100 -50 0 50 100 150 -150 -100 -50 0 50 100 150
Figure 7.8: Hysteresis Breakdown for the First Storey Braces in the V-SCED designs (MCE Earthquake LA21)
viscous component of the force is velocity dependant. The V-SCEDs also provide proportionally less
damping at the lower hazard levels for the same reason.
In addition to these hysteretic comparisons, the time histories for displacement, accelerations and base
shear for the different building designs are shown in Figures 7.10 and 7.11. This comparison is shown for
the same MCE earthquake LA21. The displacement plots in Figure 7.10 show that all of the models have
similar time traces for top floor displacement, with the no friction V-SCED designed using the second

First Storey SCED - Seismic Hazard Response Comparison

FOE-Level Earthquake LA51 - 1966 Parkfield M6.1
DBE-Level Earthquake LA01 - 1940 Imperial Valley M6.9
C-SCED (Choi et al. 2009)
5000
FOE DBE MCE
3000
1000
0
-1000
-3000
-5000
O-SCED Realistic Stiffness

5000
FOE DBE MCE
3000
1000
0
-1000
-3000
-5000
T-SCED Rralistic Stiffness

5000
FOE DBE MCE
3000
1000
0
-1000
-3000
Brace -5000
Axial
Force V-SCED No Friction with Viscous from 2nd Mode Period
5000
(kN) FOE DBE MCE
3000
1000
0
-1000
-3000
-5000
V-SCED 50% Friction with Viscous from 2nd Mode Period

5000
FOE DBE MCE
3000
1000
0
-1000
-3000
-5000
-150 -100 -50 0 50 100 150 -150 -100 -50 0 50 100 150 -150 -100 -50 0 50 100 150
Figure 7.9: Comparison of SCED Brace Response under Different Seismic Hazard Levels
mode period (V-SCED (0-M2)) experiencing the greatest displacement. The first floor displacement plot
(equivalent to first storey drift) shows that the C-SCED and the two O-SCED models have significant

residual drift at the end of the record. This residual drift is visible in the brace hysteresis as a zero stiffness
external fuse slip and a horizontal shift of the flag-shaped hysteresis as previously shown for these braces
subjected to this record in Figure 7.7. All of the models have similar first storey displacement time traces
except for the two V-SCEDs that were designed using the first mode period (V-SCED (0-M1) and V-
SCED (50-M1)). For these two V-SCED designs, the peak drift at around 8.5s on the horizontal axis is
significantly less than it was for the other designs.
Sample Time History Results - Displacments

500mm
0 C-SCED
0 O-SCED (Theor.)
0 O-SCED (Realistic)
0 T-SCED (Theor.)
Top Floor
Displacement 0 T-SCED (Realistic)
(mm)
0 V-SCED (0-M1)
0 V-SCED (50-M1)
0 V-SCED (0-M2)
0 V-SCED (50-M2)
100mm
0 C-SCED
0 O-SCED (Theor.)
0 T-SCED (Theor.)
First Floor
Displacement 0 T-SCED (Realistic)
(mm)
0 V-SCED (0-M1)
0 V-SCED (50-M1)
0 V-SCED (0-M2)
0 V-SCED (50-M2)
4 6 8 10 12 14 16 18 20 22 24
Time (s)
Figure 7.10: Time History Comparison of First and Top Floor Displacements (MCE Earthquake LA21)

Sample Time History Results - Accelerations and Base Shear

1.0g
0 C-SCED
0 O-SCED (Theor.)
Second 0 T-SCED (Theor.)

Floor 0 T-SCED (Realistic)
Acceleration
(g) 0 V-SCED (0-M1)
0 V-SCED (50-M1)
0 V-SCED (0-M2)
0 V-SCED (50-M2)
2000kN
0 C-SCED
0 O-SCED (Theor.)
0 T-SCED (Theor.)
Base
Shear 0 T-SCED (Realistic)
(kN)
0 V-SCED (0-M1)
0 V-SCED (50-M1)
0 V-SCED (0-M2)
0 V-SCED (50-M2)
4 6 8 10 12 14 16 18 20 22 24
Time (s)
Figure 7.11: Time History Comparison of Second Floor Acceleration and Base Shear (MCE Earthquake LA21)
The acceleration time histories in Figure 7.11 show a significant difference between the behaviour of
the V-SCEDs and the other SCEDs. The V-SCEDs had lower acceleration frequencies had smaller peak
accelerations. As one would expect, as the viscous damping increased, the accelerations decreased: the first
mode period V-SCED designs (M1) had a lower acceleration response than the second mode period
designs (M2). The base shear time histories shown in the same figure are similar for all of the designs

except for the first mode period V-SCED designs, which had significantly higher base shears due to the
high viscous damping forces. All of the V-SCED designs also had smoother base shear time-histories than
the other SCED designs.
7.8.2 VISCOUSLY DAMPED SCED MODELLING COMPARISON
To confirm the anticipated behaviour of the V-SCED buildings under seismic excitations, the V-SCED
responses were compared to the hysteretic behaviour predicted by the SCED Mechanics Simulator. The
Mechanics Simulator simulates the effect of viscous damping on a SCED hysteresis using a specified linear
viscous damping constant applied at a specified sinusoidal frequency. This comparison is shown in Figure
7.12. The top two plots show a comparison between the OpenSees response and the Mechanics Simulator
predictions for the first storey V-SCED braces designed using the first mode effective frequency. These
results are for the DBE-level LA01 record. The grey lines in the figure show the Mechanics Simulator enve-
lope prediction for the brace deformed at the first mode effective frequency (dotted line) and the second
mode effective frequency (solid line). For these first mode designs, the OpenSees model response generally
lies somewhere in between the first and second mode predictions from the Mechanics Simulator model.
The first mode prediction is a lower bound on the response because the structure cannot oscillate any
slower than its first mode. The real velocities are generally higher; however, they are not so high that the
dynamic response approaches the second mode predicted response. This is because the high damping con-
stants that were used for these two designs effectively limited the response of the frame, reducing the veloc-
ities.
The two plots in the centre row of the figure show the responses of the first storey V-SCED braces that
were designed using the second mode effective frequency. These plots show that the second mode designs
have maximum forces that are close to the Mechanics Simulator predictions for the second mode period.
The first mode period represents a lower bound prediction of the seismic response. These results suggest
that the second mode design method seems to produce more predictable results under dynamic earthquake
excitations. This is because the second mode effective frequency is associated with velocities that are closer
to the velocities experienced by the nonlinear model when subjected to DBE-level earthquakes (as previ-
ously discussed in Section 7.4.1 and shown in Table 7.5).
The bottom two plots in the figure show the same second mode period designs under an MCE excita-
tion (LA21). Under this increased-magnitude record, the second mode period Mechanics Simulator model
still provides a good prediction of the dynamic response of the V-SCED braces.

First Storey SCED - Comparison with Mechanics Simulator Prediction

DBE-Level Earthquake LA01 - 1940 Imperial Valley M6.9
3000 V-SCED No Friction with V-SCED 50% Friction with

Mode 1 Visc. Mode 1 Visc.
2000
1000
Brace
Axial Mechanics
0 Simulator @
Force Mode 2 Effective
(kN) -1000 Frequency
Mechanics
-2000 Simulator @
Mode 1 Effective
Frequency
-3000
OpenSees Result
3000 V-SCED No Friction with V-SCED 50% Friction with

Mode 2 Visc. Mode 2 Visc.
2000
1000
Brace
Axial 0
Force
(kN) -1000
-2000
-3000
-80 -60 -40 -20 0 20 40 60 80 -80 -60 -40 -20 0 20 40 60 80


5000
V-SCED No Friction with V-SCED 50% Friction with
4000 Mode 2 Visc. Mode 2 Visc.
3000
Mechanics
2000 Simulator @
Brace Mode 2 Effective
1000 Frequency
Axial 0
Force
(kN) -1000
-2000
-3000 Mechanics
Simulator @
-4000 Mode 1 Effective
Frequency
-5000 OpenSees Result
-150 -100 -50 0 50 100 150 -150 -100 -50 0 50 100 150
Figure 7.12: Time History Comparison of Second Floor Acceleration and Base Shear (MCE Earthquake LA21)
7.8.3 RESPONSE SUMMARY
The median, mean (  ) and mean plus one standard deviation (  +  ) results for peak acceleration,
peak storey drift, peak residual storey drift and peak base shear are shown for all of the different designs in

Table 7.11. More detailed analyses with individual design comparisons will be presented in the following
sections; however, some broad trends in these results are immediately apparent.
Table 7.11: Maximum Response Quantities - Median [Mean  ] (Mean + One Standard Deviation  +  )
Maximum Peak Residual Maximum Base Shear
Peak Drift (%)
Acceleration (g) Drift (%) (kN)
Analysis FOE DBE MCE FOE DBE MCE FOE DBE MCE FOE DBE MCE
C-SCED 0.77 1.25 1.58 0.82 1.87 5.65 0.01 0.04 2.37 2160 2680 3330
[0.80] [1.23] [1.59] [0.97] [2.09] [6.73] [0.05] [0.29] [4.35] [2230] [2650] [3350]
(1.13) (1.48) (1.78) (1.53) (2.86) (10.9) (0.18) (0.73) (8.84) (2510) (2900) (3800)
O-SCED 0.97 1.50 2.03 0.75 2.03 4.80 0.01 0.01 2.00 2300 3180 3860
(Theor.) [0.97] [1.42] [1.97] [0.87] [1.99] [6.08] [0.04] [0.16] [3.16] [2390] [3150] [3940]
(1.33) (1.70) (2.21) (1.52) (2.68) (9.73) (0.16) (0.39) (6.44) (2750) (3530) (4320)
O-SCED 0.82 1.43 1.96 0.73 1.99 5.31 0.01 0.02 2.31 2280 3250 3850
(Real.) [0.94] [1.38] [1.89] [0.92] [2.03] [6.31] [0.05] [0.17] [3.34] [2400] [3150] [3930]
(1.32) (1.67) (2.12) (1.61) (2.72) (10.1) (0.21) (0.42) (6.87) (2740) (3550) (4290)
T-SCED 1.08 1.64 1.97 0.57 1.79 4.88 0.01 0.01 0.02 2260 2620 3670
(Theor.) [1.10] [1.58] [1.96] [0.79] [1.88] [4.96] [0.01] [0.01] [0.04] [2310] [2630] [3710]
(1.48) (1.86) (2.24) (1.40) (2.57) (7.11) (0.02) (0.01) (0.11) (2550) (2850) (4340)
T-SCED 0.83 1.39 1.82 0.79 2.00 5.33 0.01 0.01 0.03 2280 2610 3750
(Real.) [0.93] [1.37] [1.78] [0.94] [2.06] [5.34] [0.02] [0.01] [0.06] [2310] [2610] [3700]
(1.30) (1.64) (2.03) (1.57) (2.78) (7.67) (0.03) (0.02) (0.15) (2540) (2830) (4320)
V-SCEDa 0.36 0.62 0.99 0.55 1.61 3.45 0.01 0.01 0.01 2010 2940 4970
(0-M1) [0.43] [0.64] [1.00] [0.66] [1.65] [3.57] [0.01] [0.01] [0.04] [2100] [2920] [4870]
(0.65) (0.85) (1.29) (0.95) (2.19) (4.90) (0.01) (0.01) (0.10) (2740) (3490) (5940)
V-SCED 0.52 1.00 1.26 0.58 1.59 4.15 0.01 0.01 0.01 2020 2830 4310
(50-M1) [0.59] [0.97] [1.32] [0.70] [1.70] [4.10] [0.01] [0.01] [0.02] [2170] [2780] [4320]
(0.87) (1.21) (1.56) (1.06) (2.34) (5.73) (0.02) (0.01) (0.06) (2620) (3210) (5100)
V-SCED 0.62 1.05 1.29 0.91 2.43 4.82 0.01 0.01 0.01 1610 2400 3670
(0-M2) [0.63] [1.00] [1.31] [1.00] [2.54] [5.19] [0.01] [0.01] [0.02] [1715] [2360] [3590]
(0.88) (1.20) (1.53) (1.49) (3.50) (7.11) (0.02) (0.01) (0.06) (2100) (2710) (4300)
V-SCED 0.83 1.26 1.52 0.83 1.96 4.95 0.01 0.01 0.02 1840 2410 3610
(50-M2) [0.79] [1.20] [1.52] [0.92] [2.20] [5.05] [0.02] [0.01] [0.03] [1950] [2410] [3570]
(1.11) (1.44) (1.71) (1.43) (3.02) (7.01) (0.03) (0.03) (0.08) (2260) (2710) (4190)
a. For V-SCEDs the first number within the parentheses represent the percentage of the original T-SCED friction. The second num-
ber within the parentheses indicates which mode was used to calculate the effective period for the damping constant calculation.
First, the C-SCED (Choi et al., 2008) and the more realistic initial stiffness O-SCED designs resulted
in similar accelerations and peak drifts; however, the maximum base shear of the O-SCED showed a mod-
est increase over the C-SCED, while the O-SCED had a lower residual drift at the MCE level. It will be
seen that these differences were due to the increased external fuse slip force in the O-SCED.

Second, the use of the theoretical versus the realistic initial stiffness in the O-SCED and T-SCED
braces did not seem to have a significant effect on the dynamic response. The realistic stiffness generally
resulted in slightly increased drifts and slightly reduced accelerations. This effect was greater for the T-
SCED than for the O-SCED because the difference between the realistic and theoretical stiffnesses was
greater for the T-SCED.
Next, the accelerations of the V-SCEDs were significantly lower than the accelerations of the other
SCEDs. This shows that adding viscous dampers to SCEDs produces the desired effect of reducing acceler-
ations. These accelerations were reduced at the expense of increased base shear in the first mode effective
frequency designs. For the second mode design with no internal friction, accelerations were reduced at the
cost of modestly increased drifts at the DBE level. The only V-SCED design that improved relative to the
T-SCED design in all categories was the second mode period design with 50% internal friction damping.
Compared to the T-SCED, this design produced a modest decrease in accelerations, similar drifts and
slightly reduced base shears.
In the peak drift MCE results, the mean peak values for each design tended to be within the range of
4.0% to 5.7%. Mean plus standard deviation values for these MCE drifts are even higher, up to 10% in the
O-SCEDs and 7.7% in the T-SCEDs. O-SCED braces that were equipped with external friction fuses
could technically accommodate unlimited drift, provided that the slots in the friction fuse are detailed to
be long enough; however, the T-SCEDs that were designed for this study had deformation capacities
between 4.5% and 5.0%. Although real T-SCEDs are not likely to survive much higher drifts, the ultimate
limit state response of the T-SCED braces was not studied and is outside the scope of this thesis. It is antic-
ipated that the failure mechanism of T-SCEDs past 5% drift will consist of a combination of steel yielding
in the internal friction dampers and tendon rupture. These mechanisms would change the dynamic behav-
iour of the brace, lowering the activation force and changing the hysteretic shape. In any case, if steel is
assumed to have a post-yield stiffness of 2% of its initial stiffness, then the stiffness of the brace failure
mechanism would likely be similar to or greater than the activated stiffness of the brace tendons. There-
fore, for the purposes of simplicity for the dynamic comparison in this study, the drift levels above 5.0%
were assumed to be valid for the estimation of the mean peak drift. The number of earthquake records that
caused peak drifts greater than 5.0% (termed n 5% ) for each design under each seismic hazard level is given
in Table 7.12. This table shows that for the non-viscous designs, approximately half of the MCE earth-
quakes caused these high drifts. The addition of viscous damping was able to reduce the MCE drifts to
varying degrees depending on the viscous damping constant. First mode effective stiffness designs had

higher viscous damping constants and, therefore, lower MCE drifts and less earthquakes causing drifts
greater than 5%, but also once again sustained higher base shears.
Table 7.12: Number of Earthquake Records that Caused

Maximum Drifts Greater than 5%
Number of EQs with
Maximum Drift > 5% n 5%
Analysis FOE DBE MCE
Choi et al. SCED 0 0 11
O-SCED (Theor.) 0 0 9
O-SCED (Real.) 0 0 11
T-SCED (Theor.) 0 0 10
T-SCED (Real.) 0 0 12
V-SCED (0-M1) 0 0 3
V-SCED (50-M1) 0 0 6
V-SCED (0-M2) 0 0 8
V-SCED (50-M2) 0 0 9
7.8.4 COMPARISON OF TIME HISTORY RESULTS TO V-SCED MODELS WITH NO FRICTION
Using the measured damping values for the two no-friction V-SCED models (as previously presented in
Section 7.6), the mean results for the lateral displacement of those models were compared to the estimated
lateral displacement values based on the design displacement response spectrum. Since the effect of the
SCED brace internal friction on the percentage of critical damping could not easily be determined, this
comparison was only performed for the two V-SCED models that did not include any friction (models 0-
M1 and 0-M2).
The design displacement response spectra were determined using the ASCE code design spectrum pre-
viously shown in Figure 7.6 as a starting point. The acceleration spectrum values from that design spec-
trum were first adjusted to account for the different values of damping using the Eurocode 8 method
(CEN, 2004). Then, the design displacement spectra were calculated based on those modified acceleration
spectra for each level of damping. The building model periods were previously presented in Table 7.6.
The comparison results are shown in Table 7.13. As this table shows, the design response spectrum dis-
placement values significantly underestimate the mean measured displacement values by 40% to 70%.

This suggests that using the response spectrum method assuming a linear elastic system plus viscous damp-
ing does not provide a good estimate of dynamic V-SCED behaviour.
Table 7.13: Comparison of Model Results with Response Spectrum Predictions

for the V-SCED Models with No Friction
Percentage Linear Model Mean Adjusted Model Mean Adjusted

of Critical Period Lateral Roof Spectral Lateral Roof Spectral
Analysis Damping Displacement Displacement Displacement Displacement
T1
DBE DBE MCE MCE
 (s) (mm) (mm) (mm) (mm)
V-SCED (0-M1) 7.3% 0.89 274 159 626 239
V-SCED (0-M2) 4.5% 0.89 406 181 901 272
7.8.5 C-SCED VERSUS REALISTIC DESIGNED O-SCED
A first detailed comparison is carried out between the C-SCED designs that were assumed for the
building design in the Choi et al. study (2008) and the redesigned O-SCEDs that take into account realis-
tic stiffnesses and construction tolerances. This comparison was important in order to determine how sim-
ilar the results from models that were previously built and analyzed using the older assumed SCED
properties were to the results from models built using more realistic designs. The most fundamental differ-
ence between these two designs was the post-activation stiffness of the tendons. This resulted in a signifi-
cantly higher external friction fuse slip force in the more realistic O-SCED braces. The results from the
two models are compared in Figure 7.13.
The upper plots in the figure show that the C-SCED designs and the equivalent O-SCED designs had
similarly shaped mean displacements over the height of the frames. The magnitudes were also similar
except at the MCE level, where the C-SCED designs had approximately 8% higher mean and mean plus
standard deviation displacements. Of course, the more important parameter related to the performance of
the braces themselves is the storey drift. The second row of plots from the top shows that the two designs
had similar drift magnitudes, with the C-SCED drifts generally being a little higher except at the FOE
level. Neither design experienced much concentration of the drift except at the MCE level, where the drifts
tended to be concentrated at the lower storeys.
Neither design experienced any significant residual drift at the FOE level. At the DBE level, the drift
plots look relatively chaotic because only some of the earthquakes caused the external friction fuses to slip.
At that design-basis level, even the mean plus standard deviation of the drift was lower than 0.5%, indicat-

C-SCED (Choi et al. 2008) vs. O-SCED (Realistic Initial Stiffness)

Results for Mean (μ) and Mean plus One Standard Deviation (μ+σ)
6
FOE DBE MCE
4
Storey
3
Level
2
0 50 100 150 200 0 200 400 600 0 500 1000 1500

Displacment (mm) Displacment (mm) Displacment (mm)
6
FOE DBE MCE
5
C-SCED O-SCED
4 μ, μ+σ μ, μ+σ
Storey
3
Level
2
0 0.5 1.0 1.5 2.0 0 1 2 3 4 0 3 6 9 12

Drift (%) Drift (%) Drift (%)
6
FOE DBE MCE
4
Storey
3
Level
2
0 0.1 0.2 0.3 0 0.2 0.4 0.6 0.8 0 2 4 6 8 10

Residual Drift (%) Residual Drift (%) Residual Drift (%)
6
FOE DBE MCE
O-SCED
5 μ, μ+σ
Storey 4 C-SCED
μ, μ+σ
Level
3
1
0.5 1.0 1.5 2.0 2.5 0 0.5 1.0 1.5 2.0 2.5 0 0.5 1.0 1.5 2.0 2.5
Acceleration (g) Acceleration (g) Acceleration (g)
C-SCED O-SCED C-SCED O-SCED C-SCED O-SCED
Max. Base 2230 2400 μ 2650 3150 μ 3350 3930 μ
Shear (kN) 2510 2740 μ+σ 2900 3550 μ+σ 3800 4290 μ+σ
Figure 7.13: Result Comparison - Choi et al. SCED versus O-SCED

ing that the building would likely still be usable after an earthquake at that hazard level. The MCE-level
results show a significant concentration of the residual drift in the lower storeys. The C-SCED designs
experienced approximately 35% more residual drift than the fully designed O-SCED brace designs, due to
its lower external friction fuse activation force.
The O-SCED design experienced significantly higher accelerations than the C-SCED design at all
three hazard levels because of its higher, more realistic stiffness and higher external friction fuse slip load.
This was also reflected in the maximum base shear results that are shown at the bottom of the figure. The
external friction fuse slip force had a particularly significant effect on these parameters since it effectively
caps-off the base shear and accelerations. The lower fuse slip force in the C-SCEDs also caused increased
residual drifts, especially at the MCE-level. Based on these results, it seems that the previous analytical full
SCED building studies were conservative in terms of drift and residual drift, which are the primary damage
indices for building structures; however, they may have underestimated the acceleration and base shear lev-
els that realistic SCED braces would produce. Balancing this is the fact that, as discussed in Section 6.7.6,
analytical models of SCED braces, including realistic ones, tend to overestimate the accelerations that will
be present in a physical system.
7.8.6 THEORETICAL STIFFNESS VERSUS REALISTIC STIFFNESS
As discussed earlier, due to construction tolerances in the lengths of the axial members in a SCED
brace, there can be a significant difference between the theoretical and realistic initial stiffness of the brace.
This difference between the theoretical and realistic stiffness tends to be greater for the T-SCED braces
than for the O-SCED braces because the T-SCED has three axial members instead of only two (refer back
to Table 7.3 for stiffness values). Typically, the O-SCEDs had a realistic initial stiffness of approximately
50% to 70% of the theoretical initial stiffness and the T-SCEDs had a realistic initial stiffness of approxi-
mately 25% to 35% of the theoretical initial stiffness. A comparison of the theoretical stiffness versus real-
istic stiffness results for the O-SCED brace design is shown in Figure 7.14.
As the figure shows, the results for the theoretical and realistic initial stiffness O-SCED designs both
have similar peak drifts, residual drifts, accelerations and base shears. The theoretical initial stiffness design
resulted in sightly larger accelerations at all hazard levels due to the higher stiffness. That higher stiffness
also resulted in slightly lower peak and residual drifts. The differences between the two designs was greatest
at the highest seismic hazard level (MCE). These results suggest that the initial stiffness does not have a sig-
nificant effect on the dynamic response of this six-storey O-SCED frame. This may be caused by the rela-

O-SCED Theoretical vs. Realistic Initial Stiffness

6
FOE DBE MCE
Theor.
5 μ, μ+σ
4 Realistic
Storey μ, μ+σ
3
Level
2
0 0.5 1.0 1.5 2.0 0 1 2 3 4 0 3 6 9 12

6
FOE DBE MCE
4
Storey
3
Level
2
0 0.1 0.2 0.3 0 0.2 0.4 0.6 0.8 0 2 4 6 8 10

6
FOE DBE MCE
Theoretical
5 μ, μ+σ
Storey 4 Realistic
μ, μ+σ
Level
3
1
0.5 1.0 1.5 2.0 2.5 0 0.5 1.0 1.5 2.0 2.5 0 0.5 1.0 1.5 2.0 2.5
Theoretical Realistic Theoretical Realistic Theoretical Realistic
Max. Base 2390 2400 μ 3150 3150 μ 3940 3930 μ
Shear (kN) 2750 2740 μ+σ 3530 3550 μ+σ 4320 4290 μ+σ
Figure 7.14: Result Comparison - Theoretical versus Realistic Initial Stiffness of O-SCEDs
tively high initial stiffness of the O-SCED frames. This high stiffness for both the theoretical and realistic
braces causes them to behave more like a rigid-plastic system than an elasto-plastic system. Therefore, even
modest changes in the initial stiffness result in an effectively rigid-plastic system, making the deformation
demand relatively insensitive to the initial stiffness. This result indicates that it is possible that the practice
of designing high stiffness frames using seismic design loads which are heavily dependant on the building
period may be questionable for mid-rise frames. This is an area of research that may warrants study.

The differences between the theoretical and realistic initial stiffness in the SCED designs made a larger
difference for the T-SCED braces as shown in Figure 7.15. As was the case for the O-SCED, the T-SCED
had higher accelerations for the theoretical stiffness design but lower drifts. For the MCE-level earth-
quakes, the mean accelerations of the theoretical stiffness design was the same as the mean plus standard
deviation acceleration of the realistic stiffness design. The base shear results were not appreciably different.
T-SCED Theoretical vs. Realistic Initial Stiffness

6
FOE DBE MCE
Theor.
5 μ, μ+σ
4 Realistic
Storey μ, μ+σ
3
Level
2
0 0.5 1.0 1.5 2.0 0 1 2 3 4 0 3 6 9 12

6
FOE DBE MCE
5 Theoretical
μ, μ+σ
Storey 4 Realistic
Level μ, μ+σ
3
1
0.5 1.0 1.5 2.0 2.5 0 0.5 1.0 1.5 2.0 2.5 0 0.5 1.0 1.5 2.0 2.5
Theoretical Realistic Theoretical Realistic Theoretical Realistic
Max. Base 2310 2310 μ 2630 2610 μ 3710 3700 μ
Shear (kN) 2550 2540 μ+σ 2850 2830 μ+σ 4340 4320 μ+σ
Figure 7.15: Result Comparison - Theoretical versus Realistic Initial Stiffness of T-SCEDs
Based on these comparisons, it seems that if the acceleration results are important in a study, then it is
important to model the initial stiffness of SCED braces accurately, especially for T-SCEDs at high seismic
hazard levels; however, it is important to note that the accelerations that are attained from OpenSees mod-
els may not fully represent the accelerations of real systems. The more gradual stiffness transitions that are
present in real SCED braces have a significant effect on the accelerations as discussed previously in Section
6.7.6. One important caveat is that, as mentioned previously, the initial stiffness of the braces would effect

the design forces in the system. That effect is not considered in this study. In addition, the initial stiffness
may have a larger effect on the dynamic behaviour of short period structures.
7.8.7 REALISTIC DESIGNED O-SCED VERSUS T-SCED
Since the T-SCED permits a full self-centering capability up to 4% to 5% drift, the dynamics of a sys-
tem using the T-SCED should be significantly different than the O-SCED system that used external fric-
tion fuses to reach high drift levels. There have been no previous numerical studies that examine the
behaviour of T-SCED braces in a frame. The T-SCED and O-SCED model comparison is shown in Fig-
ure 7.16.
The drift plots in this figure show that since there is no zero-stiffness external fuse activation, the T-
SCED design did a much better job of spreading the drift demand across all the storeys than the O-SCED
did. In fact, even at the MCE-level, the T-SCED building had a nearly uniform distribution of drift with
respect to height. This is in contrast to the O-SCED, which had a good distribution of drift under the
DBE-level earthquakes when the external friction fuse is unlikely to slip, but experienced significant con-
centration of drift in the lower storeys under the MCE-level earthquakes. The additional consequence of
this was that the T-SCED had significantly lower peak drift at the MCE-level. Of course, the T-SCED also
did not experience any residual drift, whereas at the MCE-level, the O-SCED experienced significant
residual drift; however, if the model had included any yielding that may be associated with the ultimate
limit state in the T-SCED brace above 5% drift, there would be some limited residual drift that was not
captured in these analyses.
The two different buildings had similar peak accelerations, except at the MCE-level where the O-
SCED had slightly higher accelerations. This was probably the result of the higher post-activation stiffness
of the O-SCED braces. The O-SCED structures also experienced slightly higher base shear at all hazard
levels.
Overall, the T-SCED design provided a better response than the O-SCED in terms of peak drift, accel-
eration and base shear. The main benefit though, of course, is that the T-SCED did not experience any
residual drifts due to its ability to fully self-centre even at large drift levels.
7.8.8 T-SCED VERSUS VISCOUSLY DAMPED V-SCED
As discussed in Section 7.8.3, the addition of viscous dampers to the T-SCEDs had the effect of reduc-
ing the accelerations as intended. The viscous dampers were designed using either the first or second mode

O-SCED vs. T-SCED (Realistic Initial Stiffness)

6
FOE DBE MCE
T-SCED
μ, μ+σ
5
O-SCED
4 μ, μ+σ
Storey
3
Level
2
0 0.5 1.0 1.5 2.0 0 1 2 3 4 0 3 6 9 12

6
FOE DBE MCE
4
Storey
3
Level
2
0 0.1 0.2 0.3 0 0.2 0.4 0.6 0.8 0 2 4 6 8 10

6
FOE DBE MCE
O-SCED
5 μ, μ+σ
T-SCED
Storey 4 μ, μ+σ
Level
3
1
0.5 1.0 1.5 2.0 2.5 0 0.5 1.0 1.5 2.0 2.5 0 0.5 1.0 1.5 2.0 2.5
O-SCED T-SCED O-SCED T-SCED O-SCED T-SCED
Max. Base 2400 2310 μ 3150 2610 μ 3930 3700 μ
Shear (kN) 2740 2540 μ+σ 3550 2830 μ+σ 4290 4320 μ+σ
Figure 7.16: Result Comparison - O-SCED versus T-SCED (Realistic Initial Stiffness)
effective frequencies to determine a maximum design velocity which, in turn, determined the viscous
damping constant. A comparison of the different V-SCED results to the T-SCED baseline results is shown
in Table 7.14. In general, the first mode effective frequency designs (V-SCED (0-M1) and (50-M1))
resulted in the greatest reductions in peak drifts (approximately 20-35% lower) and accelerations (approxi-
mately 30% lower for the half-friction designs and 50% lower for the no-friction designs) relative to the
regular T-SCED design; however, these first mode V-SCED designs also resulted in up to 15% higher base
shears for the no-friction design and up to 30% higher base shears for the half-friction design. These base

shear increases would also be associated with increased capacity design loads in the rest of the building. The
second mode effective frequency designs resulted in more modest reductions in acceleration and no
increase in base shear: accelerations were reduced approximately 10-15% for the half-friction designs and
25-30% for the no-friction design. The 50% internal friction damping second mode design (50-M2),
showed similar peak drifts compared to the T-SCED, whereas the design that had no internal friction
damping (0-M2) showed up to a 20% increase in peak drift at the DBE level. The second mode V-SCED
design results over the building height are compared to the T-SCED design results in Figure 7.17.
Table 7.14: Response Quantities Relative to T-SCED (Realistic Stiffness) - Median Values
Energy
Mean
Dissipation Maximum Maximum Base Shear
Activation Peak Drift (%)
Analysis Capacity Acceleration (g) (kN)
Force P a
Parameter
(kN)  FOE DBE MCE FOE DBE MCE FOE DBE MCE
V-SCED
55% 0.01 43% 45% 54% 70% 81% 65% 88% 113% 133%
(0-M1)
V-SCED
78% 0.55 63% 72% 69% 73% 80% 78% 89% 108% 115%
(50-M1)
V-SCED
55% 0.01 75% 76% 71% 115% 122% 90% 71% 92% 98%
(0-M2)
V-SCED
78% 0.55 100% 91% 84% 105% 98% 93% 81% 92% 96%
(50-M2)
As this figure shows, for the no internal friction damping second mode effective frequency design (0-
M2), the DBE-level drift results show a concentration of drift at the lower floors, whereas the MCE-level
results had a more regular of drift demands along the height. This resulted in an increased drift of up to
20% for the DBE hazard level, but actually resulted in a decrease in drift by 10% at the MCE hazard level.
These drifts resulted from the fact that the removal of the internal friction damping lowers the effective
activation force of the brace considerably when the velocities are low, thus increasing the drift. When veloc-
ities are higher, like under the MCE hazard level, the effective activation force increases enough to actually
cause a decrease in the drift. The acceleration results for this brace design were approximately 25% lower
than the T-SCED results, with the mean plus standard deviation of the acceleration for the V-SCED lower
than the mean of the T-SCED in all cases. These results were achieved in tandem with a small reduction in
the total base shear.

T-SCED vs. V-SCED (No Friction, 2nd Mode Effective Frequency)

6
FOE DBE MCE
5
T-SCED V-SCED
4 μ, μ+σ (0-M2)
Storey μ, μ+σ
3
Level
2
0 0.5 1.0 1.5 2.0 0 1 2 3 4 0 3 6 9 12

6
FOE DBE MCE
5 T-SCED
μ, μ+σ
Storey 4
Level V-SCED
3 (0-M2)
μ, μ+σ
2
1
0.5 1.0 1.5 2.0 2.5 0 0.5 1.0 1.5 2.0 2.5 0 0.5 1.0 1.5 2.0 2.5
T-SCED V-SCED T-SCED V-SCED T-SCED V-SCED
Max. Base 2310 1715 μ 2610 2360 μ 3700 3590 μ
Shear (kN) 2540 2100 μ+σ 2830 2710 μ+σ 4320 4300 μ+σ
T-SCED vs. V-SCED (50% Friction Damping, 2nd Mode Effective Frequency)
6
FOE DBE MCE
5
T-SCED
4 μ, μ+σ
Storey V-SCED
3 (50-M2)
Level μ, μ+σ
2
0 0.5 1.0 1.5 2.0 0 1 2 3 4 0 3 6 9 12

6
FOE DBE MCE
5 T-SCED
V-SCED
μ, μ+σ
Storey 4 (50-M2)
Level μ, μ+σ
3
1
0.5 1.0 1.5 2.0 2.5 0 0.5 1.0 1.5 2.0 2.5 0 0.5 1.0 1.5 2.0 2.5
T-SCED V-SCED T-SCED V-SCED T-SCED V-SCED
Max. Base 2310 1950 μ 2610 2410 μ 3700 3570 μ
Shear (kN) 2540 2260 μ+σ 2830 2710 μ+σ 4320 4190 μ+σ
Figure 7.17: Result Comparison - T-SCEDs versus Second Mode Period V-SCEDs

The results in the figure for the 50% internal friction damping design (50-M2) show a much better dis-
tribution of drift demands over the height, resulting in less concentration and lower drifts which were on
par with the T-SCED results. The accelerations were only about 10% smaller than for the T-SCED design
which is not quite as severe as the benefit provided by the no internal friction damping design. Again, the
base shears also show a modest reduction. Therefore, it seems that this design provides comparable, but
slightly better performance to the T-SCED. This result is justified by the fact that this second mode V-
SCED configuration was designed for maximum velocities that were similar to those that it experiences
under DBE-level earthquakes as previously discussed in Section 7.4.1.
The V-SCED braces that were designed using the first mode effective frequency of the building resulted
in viscous damping constants that were significantly higher than for the second mode designs. The results
from the first and second mode designs are compared in Figure 7.18. In this figure, the starting-point T-
SCED results are also shown in grey. As this figure and Table 7.14 show, for both the no internal damper
friction and 50% internal damper friction, the first mode effective period designs experienced significantly
less drifts and accelerations than the second mode designs; however, the base shears were up to 35% higher
for the MCE level for the no friction design (0-M1), which may necessitate redesign of the rest of the
building to accommodate the resulting increase in the capacity design forces. For the 50% internal friction
damping first mode design, V-SCED (50-M1), the accelerations were reduced by about 30% across the
board and drifts were reduced by at least 20%, while the base shear only increased by 10% to 15%. There-
fore, for a very modest increase in base shear, the first mode 50% friction design resulted in significant
behavioural improvements.
These results suggest three main conclusions about the V-SCED behaviour. First, designing using the
second mode effective frequency of a six storey building seems to provide viscous damping constants that
provide the best match between the response of the resulting V-SCED building and the starting-point T-
SCED building. This shows that higher modes influence the behaviour of the viscously damped SCEDs. It
may be possible that for taller buildings, other higher modes should also be considered when selecting the
damping constants if a comparable response to the starting point building is desired. Second, the 50%
internal damping friction design seems to provide the best balance between controlling drifts and reducing
accelerations. Third, if increased capacity design forces around 10-15% are tolerable, the highest level of
performance may be provided by designing using the first mode period with 50% friction and 50% viscous
damping.

V-SCED (No Friction) 1st Mode vs. 2nd Mode Effective Frequency
6
FOE DBE MCE
T-SCED
5
V-SCED
4 (0-M2)
Storey μ, μ+σ
V-SCED
3
Level (0-M1)
μ, μ+σ
2
0 0.5 1.0 1.5 2.0 0 1 2 3 4 0 3 6 9 12

6
FOE DBE V-SCED MCE
(0-M2)
μ, μ+σ
5
V-SCED
Storey 4 (0-M1)
μ, μ+σ
Level
3
2 T-SCED
1
0.5 1.0 1.5 2.0 2.5 0 0.5 1.0 1.5 2.0 2.5 0 0.5 1.0 1.5 2.0 2.5
V (0-M1) V (0-M2) V (0-M1) V (0-M2) V (0-M1) V (0-M2)
Max. Base 2100 1715 μ 2920 2360 μ 4870 3590 μ
Shear (kN) 2740 2100 μ+σ 3490 2710 μ+σ 5940 4300 μ+σ
V-SCED (50% Friction Damping) 1st Mode vs. 2nd Mode Effective Frequency
6
FOE DBE T-SCED
MCE
4
Storey V-SCED
3 (50-M2)
Level V-SCED μ, μ+σ
2 (50-M1)
μ, μ+σ
1
0 0.5 1.0 1.5 2.0 0 1 2 3 4 0 3 6 9 12

6
FOE DBE MCE
5 V-SCED
(50-M2)
4
μ, μ+σ
Storey
Level
3 V-SCED
(50-M1) T-SCED
2 μ, μ+σ
1
0.5 1.0 1.5 2.0 2.5 0 0.5 1.0 1.5 2.0 2.5 0 0.5 1.0 1.5 2.0 2.5
V (50-M1) V (50-M2) V (50-M1) V (50-M2) V (50-M1) V (50-M2)
Max. Base 2170 1950 μ 2780 2410 μ 4320 3570 μ
Shear (kN) 2620 2260 μ+σ 3210 2710 μ+σ 5100 4190 μ+σ
Figure 7.18: Result Comparison - V-SCEDs First Mode Period Design versus Second Mode Period Design

7.9 CONCLUSIONS
The results from the analyses of various designs of hysteretically and viscously damped SCED systems
provided insight into: (1) the behaviour of multi-degree-of-freedom (MDOF) SCED frame structures with
realistic stiffness properties, (2) the behaviour of MDOF frames equipped with traditional O-SCED braces
versus enhanced-elongation T-SCED braces, and (3) the behaviour of MDOF SCED-braced structures
with viscous damping.
The analyses that compared the behaviour of a frame using assumed SCED properties versus more real-
istic SCED properties showed that, due to differing post-activation stiffness and external friction fuse acti-
vation forces, the realistic SCEDs had higher accelerations and base shears, but lower peak drifts and
residual drifts. This suggests that previous studies were conservative with respect to peak and residual drifts,
but perhaps underestimated accelerations and base shears. The models also assessed the effect of construc-
tion tolerances, which cause lower effective initial stiffness in the braces. These results showed that, for the
six-storey building modelled in this case study, the initial stiffness did not have a significant effect on peak
drifts or base shears; however, for the T-SCED, which had a realistic initial stiffness of between 25% and
35% of the theoretical initial stiffness, it did have a significant effect on the accelerations, with the realistic
stiffness design experiencing significantly lower accelerations. These results indicate that using the theoret-
ical initial stiffness in building models will likely result in conservative estimates of the maximum response
quantities; however, the results did not consider the effect that the SCED brace initial stiffness has on the
design loads for the structure.
Comparing the O-SCED and T-SCED designs determined that, on top of eliminating residual drifts,
the T-SCED provided a better response than the O-SCED in terms of peak drift, acceleration and base
shears.
The results of the models that included viscous damping in parallel with the SCED-braced frame con-
firmed the results of the previous studies by Kim (2007) which studied MDOF self-centering moment
frames with viscous damping only, and Kam et al. (2010) which studied SDOF self-centering systems with
a combination of hysteretic and viscous damping (both previously described in Section 7.2). In the current
analysis study, an MDOF SCED-braced frame was designed using both viscous only (no friction) and 50%
friction with viscous. The best match between the response of the T-SCED building and the V-SCED
building was achieved by using 50% friction damping with viscous damping constants that were deter-
mined using the second mode effective frequency of the structure; however, the best dynamic response was
achieved by using 50% friction, but increasing the viscous damping constant by using the first mode effec-

tive frequency for design. This first mode design resulted in a modest 15% increased base shear but it expe-
rienced significant performance improvements, decreasing accelerations by 30% and drifts by 20%.
The numerical study described here points to some avenues for further study. Particularly, the study
should be extended to taller and shorter buildings to determine whether the suggested viscous damper
design would be applicable to a wider range of structures. For taller structures consideration of higher
modes than the second mode may be necessary to develop a good V-SCED design. Since all of the viscous
dampers in the study used linear viscous damping constants, nonlinear viscous dampers may be investi-
gated in the future to determine whether the behavioural improvements that were provided by the viscous
dampers could be improved upon further. In addition, upon better understanding of the failure mecha-
nism of T-SCED braces determined through testing, analyses of T-SCED buildings with appropriate mod-
elling of the ultimate limit state could be conducted.

CHAPTER 8: CONCLUSIONS
This goal of this thesis was to improve upon the design and use of the Self-Centering Energy-Dissipa-
tive (SCED) brace, which was conceived by Tremblay and Christpoulos (U.S. Patent No. US2008/
0016794, 2008) and later confirmed with two separate prototypes by Christopoulos et al. (2008). The
improvements that were made to the SCED in this thesis were both physical and analytical. Physically, the
SCED concept was extended in two ways: first, by permitting it to resist much higher axial capacities
through the design of a high-capacity SCED; and second, by doubling the previous deformation capacity
through the design of a telescoping SCED (T-SCED). A third physical improvement to the use of SCED
braces was the confirmation of the dynamic behaviour of a multi-storey SCED frame using a shake table
test. Analytically, the numerical models of that same multi-storey SCED frame showed that the models
were able to predict the dynamic response of SCED braced frames well. Additionally, the full step-wise
hysteretic behaviour of the SCED brace was derived and a software tool (the SCED Mechanics Simulator)
was created in order to simulate the detailed behaviour of SCED and T-SCED braces. Then, to take
advantage of the experimental and analytical work in the rest of the thesis, a six-storey case study building
model was created that facilitated the investigation of the effect of the SCED brace effective initial stiffness,
the behaviour of SCED versus T-SCED braces, and the benefit of adding viscous damping to SCED
braces.
8.1 SUMMARY OF THE RESEARCH

In Chapter 2, the background and motivation for the work was presented. This included a discussion of
performance-based design concepts and the effects and consequences of residual drifts. Previously-existing
self-centering systems were described, including rocking walls, self-centering moment frames, and self-cen-
tering braces. In the context of this background work, the concept of the Self-Centering Energy-Dissipa-
324
Chapter 8: Conclusions 325
tive (SCED) brace was introduced and the results of previous building models of SCED-braced frames
were discussed.
In Chapter 3, the mechanics and hysteretic behaviour of the SCED brace were fully characterized. The
function and behaviour of each component of the SCED brace was described and explained. Then, the full
step-wise hysteretic behaviour of the SCED brace was derived, including the effect of differing inner and
outer member stiffness which causes multi-linear stiffness transitions. The concept for the new T-SCED
brace was introduced and a simplified step-wise hysteretic behaviour was derived for it. To fully character-
ize the hysteretic response of the T-SCED brace, a new software tool called the SCED Mechanics Simula-
tor was created. This tool was programmed in JavaScript and HTML5 and uses a nonlinear incremental
stiffness method analysis to determine the full, detailed hysteretic behaviour of SCED and T-SCED braces.
It allows the inner, intermediate and outer members to have different lengths so that it can simulate the
effect of member length construction tolerances on the hysteretic behaviour of the brace. This allowed an
investigation into the effective initial stiffness of SCED braces and the development of a hand calculation
method for determining the effective initial stiffness without the need for modelling. The investigation
highlighted the fact that the realistic initial stiffness of SCED and T-SCED braces is significantly lower
than the theoretical stiffness (which is equal to the sum of the axial member stiffnesses). Using the new
effective initial stiffness determined with the hand calculation method or the Mechanics Simulator, SCED
braces may be effectively modelled using simplified flag-shaped hysteretic models within nonlinear analysis
programs such as Ruaumoko or SAP2000. For the nonlinear analysis program OpenSees, a new flag-
shaped element was developed, which includes the behaviour of an external friction fuse in series with a
SCED brace. The chapter concluded by presenting a sample SCED-braced frame design method that min-
imized the amount of iteration necessary in design.
In Chapter 4, a new high-capacity SCED (HC-SCED) brace concept was presented, and the results
from the axial load frame tests of a prototype HC-SCED brace were described. During these tests, the HC-
SCED brace exhibited a stable and smooth response when subjected to both static and dynamic loads
including simulated earthquake deformation histories. In addition to the tests of the HC-SCED brace
itself, a new type of high-capacity external friction fuse consisting of a mild-steel-on-stainless-steel friction
interface was tested. This fuse only performed well within a constrained range of velocities, within which it
exhibited a predictable slip load and a stable response. At higher velocities, the external friction fuse exhib-
ited a severe stick-slip response which may limit the applicability of this type of system. The HC-SCED
brace together with the external friction fuse satisfied the requirements of both the ASCE and AISC quali-

fication protocols for buckling-restrained braces, with the exception of the dynamic requirements of the
ASCE protocol.
In Chapter 5, a prototype of the new double-elongation telescoping T-SCED brace concept was pre-
sented. The T-SCED brace prototype was tested in a full-scale vertical steel test frame. It was capable of
providing full self-centering behaviour while accommodating 3.9% drift in the test frame, applied dynam-
ically. This level of drift capacity is sufficient to resist extreme earthquakes in high-seismic zones without
the need for an external friction fuse to extend the deformation capacity of the brace. The brace satisfied
both the ASCE and AISC test protocols for buckling-restrained braces.
In Chapter 6, results from shake table tests of a three-storey one-third scale SCED-braced frame were
presented. The shake table tests successfully demonstrated that the SCED-braced frame performed as
designed when subject to a fully dynamic earthquake base excitation. The small-scale SCED braces used
for this test were equipped with external friction fuses that extended the drift capacity of the braces from
2% to 4% of the storey height, while controlling the forces imposed by the SCED braces on the adjacent
members. A companion OpenSees computer model was capable of predicting the results of the tests and a
SAP2000 model of the tests was able to predict the results equally well, with the limitation that it could
not model the external friction fuse behavior. As was the case in previous numerical studies of SCED-
braced frames (Choi et al., 2008; Tremblay et al., 2008), the storey accelerations in the model were found
to be high due to the presence of sharp stiffness transitions in the SCED brace numerical model. The test
results showed that, while high-velocity stiffness transitions did cause high accelerations in the real test, the
models tended to significantly overestimate this effect.
In Chapter 7, a numerical model of a case-study six-storey SCED-braced frame was presented. One of
the reasons that this study was conducted was to investigate of the effect of the initial stiffness of the brace
on the frame response. This was important because prior to this thesis, the initial stiffness of a SCED brace
was assumed to be equal to the theoretical initial stiffness, which was defined as the parallel stiffness of the
main axial members. This study showed that, while the use of the lower effective initial stiffness would
have an effect on the design loads for the frame because it would change the building fundamental period,
using the lower effective initial stiffness did not have a significant effect on the dynamic response of the
frame. This case study model also compared the response of the frame equipped with original SCED
braces versus T-SCED braces. The T-SCED braces were found to provide a better response than the origi-
nal SCEDs in terms of peak drift, acceleration and base shears. This study also included an investigation
into the possibility of using viscous dampers with SCED braces. A method was proposed for determining

the viscous damping constants of the dampers. It was found that if a modest 15% increase in the design
base shear was acceptable, then the addition of viscous damping could result in significant performance
improvements, decreasing accelerations by 30% and drifts by 20%.
8.2 PRIMARY CONTRIBUTIONS OF THE THESIS

This thesis has significantly improved the design and use of Self-Centering Energy-Dissipative Braces.
The primary contributions of the thesis are summarized below:
1. The development of a full step-wise closed-form hysteretic response for the

SCED brace including the effect of different inner and outer member stiffness
and non-simultaneous gap openings.
2. The creation of the SCED Mechanics Simulator, which allows the detailed
simulation of SCED and T-SCED brace hysteretic behaviour, including the
effect of member length construction tolerances and added parallel viscous
damping.
3. The determination that the initial stiffness of SCED and T-SCED braces is
significantly lower than previously thought. The theoretical initial stiffness
determined using the parallel stiffness of the axial members does not provide a
good estimate of the realistic initial stiffness of a SCED brace, even if the axial
members have identical lengths.
4. The development of a simple hand-calculation method and a rapid estimation

method to estimate the initial effective stiffness of a SCED brace given a
specified member length tolerance. Limitation: This hand-calculation may only
be used for SCED braces. T-SCED brace mechanics are too complex for the
development of a simplified equation and must be analyzed using the SCED
mechanics simulator.
5. The creation of a new self-centering hysteresis for OpenSees, which includes the
effect of an external friction fuse in series with the self-centering element.
6. The development, testing and validation of a new high-capacity HC-SCED

brace, which proved that high axial capacity SCED braces are feasible up to at
least 3000kN. This significantly enhances the range of application for SCED

braces in building structures. The main elements of the SCED brace did not
exhibit any significant velocity-dependant effects (with the exception of the
experimental end fuse).
7. The development of a new type of high-capacity external friction fuse

consisting of a mild steel on stainless steel friction interface. Limitation:
Although the performance of this fuse was not adequate and exhibited a stick-
slip behaviour at higher velocities, the resulting behaviour suggests that the
concept could be improved in order to develop a viable high capacity fuse in the
future.
8. The development, testing and validation of a new enhanced-elongation

telescoping T-SCED brace, which eliminated the necessity for an external
friction fuse in series with a SCED brace. This is the first self-centering brace
that can provide self-centering behaviour up to the maximum-considered
earthquake seismic hazard level. The main elements of the T-SCED brace did
not exhibit any significant velocity-dependant effects.
9. The confirmation that real multi-storey SCED-braced frames behave as-

designed when subjected to earthquake accelerations at their base.
10. The confirmation that the modelling methods for multi-storey SCED-braced
frames using OpenSees and SAP2000 provide good predictions for the dynamic
behaviour of the frame. Limitation: The SAP2000 model did not include the
external friction fuses at the ends of the braces.
11. The confirmation of the source of the high accelerations in SCED-braced

frames using dynamic testing. These are due to differences between the SCED
brace axial forces in adjacent storeys caused by high velocity stiffness transitions.
These stiffness transitions are sharper in the numerical brace models, causing an
overestimation of the accelerations in the models compared to the tests;
however, these overestimated values provide a true upper-bound on the
acceleration response of the system.

12. The contributions resulting from the six-storey case-study model are provided
below. Limitation: These results are valid for the single building height studied.
Further modelling is required to confirm that these results are general to all
SCED-braced frames.
a) The determination that the use of the lower effective initial stiffness
instead of the theoretical initial stiffness of SCED and T-SCED braces
may not have significant effect on the dynamic response of a mid-rise
SCED-braced frame.
b) The determination that the T-SCED brace may provide increased seis-
mic performance relative to the original SCED brace.
c) The development of a method that may be used to design viscous damp-

ers in parallel with SCED braces.
d) The determination that viscous damping in parallel with a SCED brace

may increase acceleration and drift performance.
The contributions made by this thesis bring the state of SCED braces to a point where they are almost
ready to be used in real structures. The only remaining problems to be solved are related to the perfor-
mance of the tendon anchors as discussed below.
8.3 SUGGESTIONS FOR FUTURE RESEARCH
8.3.1 TENDON AND TENDON ANCHOR BEHAVIOUR
The performance issues surrounding the tendon and tendon anchor behaviour remain to be solved. A
comprehensive testing program is necessary to properly evaluate the performance of the tendons and
anchors. This testing program should consider different sizes of tendons and anchors, varied rate of load-
ing, load dwell near the maximum deformation, cyclic loading, long-term loading and the use of an alter-
nate tendon anchor material. Steel anchors should be tested to see if they provide more reliable
performance than the aluminum anchors. In addition, larger tendons should be tested because the use of
larger tendons would reduce the total number of tendons required in a SCED brace, thereby resulting in
potentially more compact designs for high-capacity SCED braces.

In addition, as discussed in Section 3.1.5, creep/relaxation of the aramid tendons has the potential to
effect the tendon pretension in the medium- to long-term. It is not known how the aramid fibre creep
behaviour is effected by the tendon pretensioning method or the tendon anchor assemblies. A comprehen-
sive test program should be conducted using accelerated creep testing to answer these questions.
8.3.2 LONG-TERM BEHAVIOUR OF THE FRICTION DEVICES
While the behaviour of the internal friction dampers has been shown to be stable, reliable and repeat-
able in the short-term, a thorough study should be conducted in to investigate the long-term behaviour of
the friction interfaces. This study should particularly focus on bolt-relaxation and friction material creep to
determine how the friction interface behaviour may be expected to change over time.
8.3.3 THE EFFECT OF MEMBER LENGTH CONSTRUCTION TOLERANCES
It has been shown in this thesis that the effective initial stiffness of SCED braces is lower than previ-
ously assumed, even for braces with very precise length tolerances; however, the effect of this lower initial
stiffness on the design loads for SCED-braced frames has not been evaluated. In addition, although the six-
storey case study in Chapter 7 did not show a significant effect of the initial stiffness on the dynamic
response of the structure, the effect could possibly be significant for shorter-period structures. A design
study should be conducted in the future to assess the dynamic behaviour of properly-designed buildings
incorporating SCED braces with realistic initial stiffnesses. Since the case study also showed that changing
the initial brace stiffness of the SCED braced frame did not significantly effect the dynamic response, a
study should also be conducted to assess the effect of the initial brace stiffness on the dynamic behaviour of
low- to mid-rise nonlinear braced frames.
8.3.4 HIGH-CAPACITY EXTERNAL FUSE DEVELOPMENT
The high-capacity external fuse prototype that was tested along with the HC-SCED brace in Chapter 4
exhibited a stick-slip response during high velocity loading. This response made the fuse undesirable for
use in real structures; however, the behaviour of the fuse may be improved if different materials are used for
the friction interfaces. Further tests could be conducted in the future to investigate this possibility.

8.3.5 RAYLEIGH DAMPING
The shake table test results in Chapter 6 indicated that a damping model with low inherent damping
for large cycles and significantly higher inherent damping for the small cycles of response would best repre-
sent the damping behaviour that was observed during the shake table tests. The Rayleigh damping model
that was used for the shake table models did not allow the models to match the test results at both high and
low amplitudes because the Rayleigh damping model provides damping forces that are proportional to the
velocity. A future study could investigate the use of an alternative damping formulation, perhaps based on
friction at the nodes to better predict the test results at all amplitudes.
8.3.6 MORE EXTENSIVE NONLINEAR MODELLING
The six-storey case study presented in Chapter 7 represents a pilot study into the effect of the initial
brace stiffness, the O-SCED versus T-SCED behaviour, and the effect of added viscous damping. This
study should be extended to taller and shorter buildings to determine whether the conclusions drawn from
the pilot study are applicable to a wider range of structures. In addition, since all of the viscous dampers in
the study used linear viscous damping constants, nonlinear viscous dampers may be investigated in the
future to determine whether the behavioural improvements that were provided by the viscous dampers
could be improved upon further.
8.4 CURRENT STATE OF SCED BRACE TECHNOLOGY

As previously mentioned, there are still questions surrounding the behaviour of the Technora tendons
and the friction interfaces that lie outside the scope of this thesis. Studies need to be conducted to investi-
gate the behaviour of the tendon anchors and the tendons’ susceptibility to creep. These studies will pro-
vide (1) a reliable tendon anchor design and (2) new load and deformation limits, if necessary, to account
for creep/relaxation. In addition, a research program should be conducted to comprehensively and statisti-
cally evaluate: the variability in the tendon properties, reliability of the material, acceptable pretensioning
and lock-off loads for the tendons, and develop methods for quality control for the manufacture and
assembly of the tendon anchors. For the friction interfaces, a research program should be conducted to
study the long-term behaviour to determine how the response of the friction interfaces change over time
due to material degradation, bolt relaxation and creep. Once these issues are addressed, the SCED brace
will be ready to be widely used in real buildings; however, there are some additional challenges in fabrica-
tion and engineering that will be associated with this full-scale application.

On the fabrication side, it is clear that the SCED brace cannot be constructed by any arbitrary steel fab-
ricator. It is an engineered device that requires a diverse range of expertise and precise quality control. The
manufacture of a SCED brace involves heavy steel fabrication, high-precision machining, tendon preten-
sioning, and the ability to conduct robust friction damper calibrations. Whoever fabricates the brace in
large quantities should also go through a value-engineering process to reduce the cost of the brace.
Although the prototypes developed in this thesis were expensive to construct, there is certainly room for
improvement with regards to the brace cost. This process will naturally include consideration of appropri-
ate construction tolerances, but may also include refining the friction damper design, re-evaluating the end
plate thickness, and potentially modifying the design or material for the tendon anchors. Prior to the
installation of SCED braces in a structure, the brace fabricator will also have to conduct production testing
of multiple identical braces.
On the building design and engineering side, initial adoption of the SCED brace will be a challenge
because it is not yet explicitly included in any design codes. For the first few SCED brace projects, this will
necessitate highly-skilled design engineers who will be able to perform nonlinear dynamic structural analy-
ses and guide these projects through an external peer review process. In addition, the ultimate limit state
behaviour of the brace must be properly engineered to ensure an yielding steel back-up system as discussed
in Section 3.7.4. This will provide engineers and owners with an increased level of confidence in the sys-
tem. For these initial designs, the force-based code methods may be used as described in Chapter 3; how-
ever, these designs should be evaluated using nonlinear time-history analyses and modified accordingly. It
is clear from the shake table tests that the inherent damping assumptions that are used for these nonlinear
time history analyses must be carefully chosen and justified.
The above challenges are similar to those that would be associated with the adoption of any new earth-
quake-resistant technology. Overall, the SCED brace provides a level of seismic performance that is unpar-
alleled by any other bracing element for structures. It is the only bracing system that exists for buildings
that can accommodate such high levels of deformation under such high levels of loading without sustain-
ing any damage to the structural system. For these reasons, this technology has great potential for further
development and application in real structures.

References 333
REFERENCES
American Institute of Steel Construction [AISC] (2005a). Load and Resistance Factor Design Specification
for Structural Steel Buildings, ANSI/AISC 360-05 Including Supplement No.1. Chicago: American
Institute of Steel Construction, Inc.
American Institute of Steel Construction [AISC] (2005b). Seismic Provisions for Structural Steel Buildings,
ANSI/AISC 341s1-05 Including Supplement No.1. Chicago: American Institute of Steel Construction,
Inc.
American Institute of Steel Construction [AISC] (2005c). Code of Standard Practice for Steel Buildings and
Bridges, ANSI/AISC 303-05. Chicago: American Institute of Steel Construction, Inc.
American Society of Civil Engineers [ASCE] (2000). FEMA 356 - Prestandard and Commentary for the
Seismic Rehabilitation of Buildings. Reston, VA: American Society of Civil Engineers.
American Society of Civil Engineers [ASCE] (2005). Minimum Design Loads for Buildings and Other
Structures, ANSI/SEI7-05 Including Supplement No.1. Chicago: American Society of Civil Engineers.
Applied Technology Council [ATC] (2011a). Seismic Performance Assessment of Buildings: Volume 1 -
Methodology (ATC-58-1: 75% Draft). Redwood City, CA: Applied Technology Council for the Federal
Emergency Management Agency.
Applied Technology Council [ATC] (2011b). Seismic Performance Assessment of Buildings: Volume 2 -
Implementation Guide (ATC-58-1: 75% Draft). Redwood City, CA: Applied Technology Council for
the Federal Emergency Management Agency.
Atkinson, G. (2009). Earthquake Time Histories Compatible with the 2005 NBCC Uniform Hazard
Spectrum. Can. J. Civil Engineering. 36(6), 991-1000.
Beck, J.L., & Skinner, R.I. (1974). The seismic response of a reinforced concrete bridge pier designed to
step. Earthquake Engineering & Structural Dynamics. 2(4), 343-358.

References 334
Building Seismic Safety Council [BSSC] (1997). FEMA 273 - NEHRP Guidelines for the Seismic
Rehabilitation of Buildings. Washington, D.C.: Building Seismic Safety Council for the Federal
Emergency Management Agency.
Canadian Institute of Steel Construction [CISC] (2003). Handbook of Steel Construction. Toronto,
Ontario: CISC.
Carlisle Industrial Brake and Friction (2012). Friction Material Reference Sheets. Retrieved from http://
www.carlislebrake.com/dry-friction/friction-material-reference-sheets/
Carr, A. J. (2005a). Ruaumoko Program for Inelastic Dynamic Analysis [Computer Software].
Christchurch, NZ.
Carr, A. J. (2005b). Ruaumoko Program for Inelastic Dynamic Analysis - User Manual. Christchurch, NZ:
Department of Civil Engineering, University of Canterbury, New Zealand.
Choi, H., Erochko, J., Christopoulos, C., & Tremblay, R. (2008). Comparison of the Seismic Response of
Steel Buildings Incorporating Self-Centering Energy-Dissipative Dampers, Buckling Restrained Braces and
Moment Resisting Frames. (Report No. 05-2008). Toronto: Dept. of Civ. Eng., University of Toronto,
ON.
Choi, H., Erochko, J., Christopoulos, C., & Tremblay, R. (2009). Seismic response of 2D and 3D
buildings incorporating buckling-restrained and self-centering bracing systems. In F. Mazzolani, J.
Ricles, R. Sause (Ed.), Behaviour of Steel Structures in Seismic Areas 2009. Leiden NL: CRC Press/
Balkema.
Chopra, A.K. (2000). Dynamics of Structures: Theory and Applications to Earthquake Engineering (2nd
Edition). Upper Saddle River, NJ: Prentice-Hall.
Chou, C-C., & Chen, Y-C. (2012). Development and Seismic Performance of Steel Dual-Core Self-
Centering Braces. Proc. 15th World Conf. Earthquake Engng, Lisbon, Portugal.
Christopoulos, C., Filiatrault, A., & Folz, B. (2002a). Seismic response of self-centring hysteretic SDOF
systems. Earthquake Engineering & Structural Dynamics. 31(5), 1131-1150.
Christopoulos, C., Filiatrault, A., Uang, C-M., & Folz, B. (2002b). Posttensioned Energy Dissipating
Connections for Moment-Resisting Steel Frames. J. Struct. Eng. 128(9), 1111-1120.

References 335
Christopoulos, C., Pampanin, S., & Priestley, M.J.N. (2003). Performance Based Seismic Response of
Frame Structures Including Residual Deformations. Part I: Single-Degree of Freedom Systems. Journal
of Earthquake Engineering, 7(1), 97-118.
Christopoulos, C., & Pampanin, S. (2004). Towards performance-based seismic design of MDOF
structures with explicit consideration of residual deformations, ISET Journal of Earthquake Technology,
41(1), 53–73.
Christopoulos, C., & Filiatrault, A. (2006). Principles of Passive Supplemental Damping and Seismic
Isolation. Pavia, Italy: IUSS Press.
Christopoulos, C., Tremblay, R., Kim, H.-J., & Lacerte, M. (2008). Self-Centering Energy Dissipative
Bracing System for the Seismic Resistance of Structure: Development and Validation. Journal of
Structural Engineering, 134(1), 96-107.
Coglan, J. (2012). Sylvester: Vector and Matrix math for Javascript (Version 0.1.3) [Computer Software].
London, UK. Available from http://sylvester.jcoglan.com/
Computers and Structures Inc. (2009). SAP2000 [Computer Software]. Berkeley, CA.
Cornell, C. A. & Krawinkler, H. (2000). Progress and challenges in seismic performance assessment. PEER
Center News, 3(2).
Deierlein, G., Krawinkler, H., Ma, X., Takeuchi, T., Eatherton, M., Kasai, K., Hajjar, J., & Midorikawa,
M. (2011). Earthquake resilient steel braced frames with controlled rocking and energy dissipating
fuses. Steel Construction, 4(3), 171-175.
Della Corte, G., D’Aniello, M., Landolfo, R., & Mazzolani, F.M. (2011). Review of steel buckling-
restrained braces. Steel Construction, 4(2), 85-93.
Eatherton, M.R., Hajjar, J.F. (2011). Residual Drifts of Self-Centering Systems Including Effects of
Ambient Building Resistance. Earthquake Spectra, 27(3), 719-744.
Erdik M., Kamer, Y., Demircioglu, M., Sesetyan, K. (2012). 23 October 2011 Van (Turkey) earthquake.
Nat Hazards. 64(1), 651-665.

References 336
Erochko, J., Christopoulos, C., & Tremblay, R. (2011). Residual Drift Response of SMRFs and BRB
Frames in Steel Buildings Designed according to ASCE 7-05. J. Struct. Eng. 137 (5), 589-599.
European Committee for Standardization [CEN] (2004). Eurocode 8 Design of structures for earthquake
resistance, Part 1: General rules, seismic actions and rules for buildings. EN 2004-1-1, Brussels: CEN.
Fahnestock, L.A., Ricles, J.M., & Sause, R. (2007). Experimental Evaluation of a Large-Scale Buckling-
Restrained Braced Frame. J. Struct. Eng. 133 (9), 1205-1214.
Federal Emergency Management Agency [FEMA] (2009). Quantification of Building Seismic Performance
Factors, FEMA P695. Washington DC: Federal Emergency Management Agency.
Filiatrault, A., Tremblay, R., & Kar, R. (2000). Performance Evaluation of Friction Spring Seismic
Damper. J. Struct. Eng. 126 (4), 491-499.
Giannopoulos, I.P., & Burgoyne, C.J. (2012a). Accelerated and Real-Time Creep and Creep-Rupture
Results for Aramid Fibers. Journal of Applied Polymer Science. 125 (5), 3856-3870.
Giannopoulos, I.P., & Burgoyne, C.J. (2012b). Creep and Strength Retention of Aramid Fibers. Journal of
Applied Polymer Science. 126 (1), 91-103.
Google Inc. (2006). ExplorerCanvas [Computer Software]. Mountain View, CA. Available from: http://
excanvas.sourceforge.net/
Gray, M.G. (2012). Cast Steel Yielding Brace System for Concentrically Braced Frames, Doctoral Thesis,
University of Toronto, Toronto, Ontario.
Hill, A. (2008). LiveValidation (Version 1.3) [Computer Software]. London, UK. Available from: http://
livevalidation.com/
Housner, G.W. (1963). The behavior of inverted pendulum structures during earthquakes. Bulletin of the
Seismological Society of America, 53(2), 403-417.
JQuery Foundation (2012a). jQuery (Version 1.7.1) [Computer Software]. Hingham, MA. Available
from: http://jquery.com/
JQuery Foundation (2012b). jQuery UI (Version 1.8.18) [Computer Software]. Hingham, MA. Available
from: http://jqueryui.com/

References 337
Kaiser, A., Holden, C., Beavan, J., Beetham, D., Benites, R., Celentano, A., … Zhao, J. (2012). The Mw
6.2 Christchurch earthquake of February 2011: preliminary report. New Zealand Journal of Geology and
Geophysics. 55(1), 67-90.
Kam, W-Y., Pampanin, S., Palermo, A., & Carr, A-J. (2010). Self-centering structural systems with
combination of hysteretic and viscous energy dissipations. Earthquake Engineering and Structural
Dynamics, 39(10), 1083-1108.
Kammula, V. (2012). Application of Hybrid Simulation to Fragility Assessment of Self-Centering Energy

Dissipative (SCED) Bracing System, Masters Thesis, University of Toronto, Toronto, Ontario.
Kar, R., Rainer, J.H., & Lefrancois, A.C. (1996). Dynamic Properties of a Circuit Breaker with Friction-
Based Seismic Dampers. Earthquake Spectra, 12(2), 297-314.
Karavasilis T.L., & Seo, C-Y. (2010). Seismic structural and non-structural performance evaluation of
highly damped self-centering and conventional systems. Engineering Structures. 33(8), 2248-2258.
Kawashima K., MacRae, G.A., Hoshikuma, J. & Nagaya, K. (1998) Residual Displacement Response
Spectrum. Journal of Structural Engineering, 124(5), 523-530.
Kelly, J.M., & Tsztoo, D.F. (1977). Earthquake simulation testing of stepping frame with energy-
absorbing devices. Bulletin of the New Zealand National Society for Earthquake Engineering, 10(4), 196-
207.
Kim, H-J., Christopoulos, C., & Tremblay, R. (2004) Experimental Characterization of Bolt-Stressed Non-
Asbestos Organic (NAO) Material-to-Steel Interfaces. (Report). Toronto: Dept. of Civ. Eng., University of
Toronto, ON.
Kim, H-J. (2007). Self-Centering Steel Moment-Resisting Frames with Energy Dissipating Systems, Doctoral
Thesis, University of Toronto, Toronto, Ontario, Canada.
Kim, H-J., & Christopoulos, C. (2008). Friction Damped Posttensioned Self-Centering Steel Moment-
Resisting Frames. J. Struct. Eng. 134 (11), 1768-1779.
Kim, H-J. (2009a). Design of Shake Table Test Specimen. (Department of Civil Engineering Report).
Toronto: Dept. of Civ. Eng., University of Toronto, ON.

References 338
Kim, H-J. (2009b). Results of Preliminary Test. (Department of Civil Engineering Report). Toronto: Dept.
of Civ. Eng., University of Toronto, ON.
Kurama, Y., Pessiki, S., Sause, R., & Lu, L-W. (1999). Seismic Behaviour and Design of Unbonded Post-
Tensioned Precast Concrete Walls. PCI Journal, 44(3), 72-89.
Laursen, O. (2012). Flot: Attractive JavaScript plotting for JQuery (Version 0.7) [Computer Software]
Aalborg, Denmark. Available from: http://www.flotcharts.org/
Le Bec, A., Tremblay, R., Erochko, J., & Christopoulos, C. (2010). Modélisation du Comportement d’une
Diagonale de Contreventement SCED avec le Logiciel d’Analyse Structurale SAP2000. (Rapport No. GRS
SR10-02). Montreal: École Polytechnique de Montréal, Département des génies civil, géologique et des
mines, (Publication No. 2010-05). Toronto: University of Toronto Department of Civil Engineering.
Linear Composites Ltd. (n.d.). PARAFIL: The Ultimate Synthetic Rope. Retrieved from http://
linearcomposites.net/?pageid=Parafil.xml
Linear Composites Ltd. (2011). Parafil Rope: End Terminations and Fitting Instructions. (Technical Note
TN3, Issue 2). West Yorkshire, UK: Linear Composites Ltd.
MacRae G. A., & Kawashima K. (1997). Post-Earthquake Residual Displacements of Bilinear Oscillators.
Earthquake Engineering and Structural Dynamics, 26(7), 701-716.
Maffei, J., Telleen, K., & Nakayama, Y. (2008). Probability-Based Seismic Assessment of Buildings,
Considering Post-Earthquake Safety. Earthquake Spectra, 24(3), 667-699.
Magdy, T.E-S., Sause, R., Pessiki, S., & Lu, L-W. (1999). Seismic Behaviour and Design of Unbonded
Post-Tensioned Precast Concrete Frames. PCI Journal, 44(3), 54-71.
McCormick, J., Aburano, H., Ikenaga, M., & Nakashima, M. (2008). Permissible Residual Deformation
Levels for Building Structures Considering Both Safety and Human Elements. Proc. 14th World Conf.
Earthquake Engng, Beijing, China. Paper No. 05-06-0071.
McKenna, F., Fenves, G. L., Scott, M. H., & Jeremic, B. (2000). Open System for Earthquake
Engineering Simulation (OpenSees) [Computer Software]. Pacific Earthquake Engineering Research
Center, University of California, Berkeley, CA. Available from: http://opensees.berkeley.edu/

References 339
Miller, D., Fahnestock, L.A., & Eatherton, M. (2012). Behaviour of Self-Centering Buckling-Restrained
Braces. Proc. Behaviour of Steel Structures in Seismic Areas 2012. Santiago, Chile.
National Research Council of Canada [NRC] (2005). National Building Code of Canada. Ottawa, ON:
National Research Council of Canada.
Nims, D.K., Richter, P.J., & Bachman, R.E. (1993). The Use of the Energy Dissipating Restraint for
Seismic Hazard Mitigation. Earthquake Spectra, 9(3), 467-489.
Oberg, E., Jones, F. D., Horton, H. L., Ryffel, H. H., McCauley, C. J. (Ed.), Heald, R. M. (Ed.), &
Hussain, M. I. (Ed.). (2008). Machinery’s Handbook (28th Ed.). New York: Industrial Press.
Palermo, A., Pampanin, S., Buchanan, A., & Newcombe, M. (2005). Seismic Design of Multi-Storey
Buildings using Laminated Veneer Lumber (LVL). In Proceedings, 2005 NZSEE Conference, 11-13
March 2005, Wairakei, NZ. Wellington: New Zealand Society for Earthquake Engineering.
Pampanin, S., Christopoulos, C., & Priestley, M. J. N. (2003). Performance-based Seismic Response of
Frame Structures including Residual Deformations. Part II: Multi-degree of Freedom Systems. Journal
of Earthquake Engineering. 7(1), 119-147.
Pettinga, D., Christopoulos, C., Pampanin ,S., & Priestley, N. (2007). Effectiveness of simple approaches
in mitigating residual deformations in buildings. Earthquake Engineering and Structural Dynamics.
36(12), 1763-1783.
Priestley, M.J.N., Calvi, G.M., & Kowalsky, M.J. (2007). Displacement-Based Seismic Design of Structures.
Pavia, Italy: IUSS Press.
Priestley, M.J.N., Sritharan, S., Conley, J. R. & Pampanin, S. (1999). Preliminary Results and Conclusions
from the PRESSS Five-story Precast Concrete Test-building. PCI Journal, 44(6), 42-67.
Ramirez C.M., & Miranda, E. (2012). Significance of residual drifts in building earthquake loss
estimation. Earthquake Engineering and Structural Dynamics. 41(11), 1477-1493.
Ricles, J.M., Sause, R., Garlock, M., & Zhao, C. (2001). Posttensioned Seismic-Resistant Connections for
Steel Frames. J. Struct. Eng. 127(2), 113-121.

References 340
Rojas, P., Ricles, J.M., & Sause, R. (2005). Seismic Performance of Post-tensioned Steel Moment Resisting
Frames With Friction Devices. J. Struct. Eng. 131(4), 529-540.
Rosenblueth E, & Meli R. (1986). The 1985 Mexico earthquake: causes and effects in Mexico City.
Concrete International, 8(5), 23–34.
Ruiz-García, J., & Miranda, E. (2006a). Residual displacement ratios for assessment of existing structures.
Earthquake Engineering & Structural Dynamics, 35(3), 315-336.
Ruiz-Garcia J, & Miranda E. (2006b). Evaluation of residual drift demands in regular multi-story frames
for performance-based seismic assessment. Earthquake Engineering & Structural Dynamics, 35(13),
1609-1629.
Sause, R., Ricles, J.M., Roke, D.A., Chancellor, N.B., & Gonner, N.P. (2010). Large-Scale Experimental
Studies of Damage-Free Self-Centering Concentrically-Braced Frame under Seismic Loading. In ASCE
Structures Congress 2010.
SEAOC (1995). Performance Based Seismic Engineering of Buildings, Vision 2000 Report, Volumes I and II,
Structural Engineers Association of California: Sacramento, California.
Skinner, R.I., Robinson, W.H., & McVerry, G.H. (1991). Seismic isolation in New Zealand. Nuclear
Engineering and Design, 127(3), 281-289.
Somerville, P., Smith, H., Puriyamurthala, S., & Sun. J. (1997). Development of Ground Motion Time
Histories for Phase 2 of the FEMA/SAC Steel Project (SAC Joint Venture Report SAC/BD 97/04).
California: SAC Joint Venture.
Teijin Aramid (n.d. A). Comparing Aramids. Retrieved from http://www.teijinaramid.com/aramids/

comparing-aramids/
Teijin Aramid (n.d. B). Technora gold filament yarn. Retrieved from http://www.teijinaramid.com/aramids/
technora/technora-filament-yarn-yellow/
Teijin Aramid (n.d. C). Technora:.High Tenacity Aramid Fiber. [Brochure] Arnhem, Netherlands: Author.

References 341
Toranzo, L.A., Restrepo, J.I., Mander, J.B., & Carr, A.J. (2009). Shake-Table Tests of Confined-Masonry
Rocking Walls with Supplementary Hysteretic Damping. Journal of Earthquake Engineering, 13(6),
882-898.
Tremblay, R. & Atkinson, G.-M. (2001). Comparative Study of the Inelastic Seismic Demand of Eastern
and Western Sites. Earthquake Spectra, 17(2), 333-358.
Tremblay, R., Bolduc, P., Neville, R., & DeVall, R. (2006). Seismic testing and performance of buckling-
restrained bracing systems. Canadian Journal of Civil Engineering, 33(2), 183-198.
Tremblay, R., & Christopoulos, C. (2008). U.S. Patent No. US2008/0016794. Washington, DC: U.S.
Patent Office.
Tremblay, R., Lacerte, M., & Christopoulos, C. (2008). Seismic Response of Multistory Buildings with
Self-Centering Energy Dissipative Steel Braces. Journal of Structural Engineering, 134(1), 108-120.
Tremblay R., Léger P., Rogers C., Bouaanani N., Massicotte B., Khaled A., & Lamarche C-P. (2009).
Experimental testing of large scale structural models and components using innovative shake table,
dynamic, real-time hybrid simulation and multi-directional loading techniques. Proc., 3rd International
Conf. on Advances in Exp. Struct. Eng., San Francisco, USA.
Tsopelas, T., Constantinou, M.C. (1994). Experimental and Analytical Study of a System Consisting of
Sliding Bearings and Fluid Restoring Force/Damping Devices (NCEER-94-0014). Buffalo, NY: Dept. of
Civ. Eng., State University of New York at Buffalo.
Uma, S.R., Pampanin, S., & Christopoulos, C., (2010). Development of Probabilistic Framework for
Performance-Based Seismic Assessment of Structures Considering Residual Deformations. Journal of
Earthquake Engineering, 14(7), 1092 - 1111.
Wiebe L., & Christopoulos C. (2010). Characterizing acceleration spikes due to stiffness changes in
nonlinear systems. Earthquake Engineering and Structural Dynamics, 39(14),1653-1670.
Wiebe, L. & Christopoulos, C. (2011). Using Bézier curves to model gradual stiffness transitions in
nonlinear elements: Application to self-centering systems. Earthquake Engineering & Structural
Dynamics, 40(14), 1535-1552.

References 342
Wiebe, L., Christopoulos, C., Tremblay, R., & Leclerc, M. (2012a). Mechanisms to limit higher mode
effects in a controlled rocking steel frame. 1: Concept, modelling, and low-amplitude shake table
testing, Earthquake Engineering & Structural Dynamics, [accepted].
Wiebe, L., Christopoulos, C., Tremblay, R., & Leclerc, M. (2012b). Mechanisms to limit higher mode
effects in a controlled rocking steel frame. 2: Large-amplitude shake table testing, Earthquake
Engineering & Structural Dynamics [accepted].
Wilkinson, S., Grant, D., Williams, E., Paganoni, S., Fraser, S., Boon, D., … Free, M. (2013).
Observations and implications of damage from the magnitude Mw 6.3 Christchurch, New Zealand
earthquake of 22 February 2011. Bull Earthquake Engineering. 11(1), 107-140.
Wolff, E.D., Constantinou, M.C. (2004). Experimental Study of Seismic Isolation Systems with Emphasis on
Secondary System Response and Verification of Accuracy of Dynamic Response History Analysis Methods
(MCEER-04-0001). Buffalo, NY: Dept. of Civ. Eng., State University of New York at Buffalo.
Zhu, S., & Zhang, Y. (2008). Seismic Analysis of Concentrically Braced Frame Systems with Self-
Centering Friction Damping Braces. J. Struct. Eng. 134(1), 121-131.
Zienkiewicz, O.C. & Taylor R.L. (2000). The Finite Element Method - Volume 1: The Basis. Woburn, MA:
Butterworth-Heinemann.

APPENDIX A: SCED MECHANICS SIMULATOR SOURCE CODE
The JavaScript file below contains all of the primary calculations that are related to the computation of
the SCED brace behaviour in the SCED Mechanics Simulator. All of the other source code for the pro-
gram is associated with the user interface, and is therefore omitted.
// SCEDhandler.js -- Contains Functions to compute SCED Brace Calculations

// This is a Webworker enabled script page.
onmessage = function(event) {
// Call SCED Calculations here
importScripts('sylvester.js'); // Matrix Math Package

importScripts('globalelements.js'); // Global constants and helper functions
CalculateSCED(event.data, true);
};
// JSON function interface for data transfer

function resp_data(iSteps, iLeft_Loc, iRight_Loc, iCur_Length, iAxial, iElongation, iShift, iStiffness, istep_pret,
istep_end, iblMiddle, iOverallDef, iOverallForce, iFriction_Act, iBIN) {
this.Steps = iSteps;
this.Left_Loc = iLeft_Loc;
this.Right_Loc = iRight_Loc;
this.CurLength = iCur_Length;
this.Axial = iAxial;
this.Elongation = iElongation;
this.Shift = iShift;
this.Stiffness = iStiffness;
this.step_pret = istep_pret;
this.step_end = istep_end;
this.blMiddle = iblMiddle;
this.Overall_Def = iOverallDef;
this.Overall_Force = iOverallForce;
this.Friction_Act = iFriction_Act;
this.BIN = iBIN;
}
function SCEDMatrix(k_in, k_mid, k_out, k_t1, k_t2, k_f1, k_f2, k_conn, k_big, k_small, tend1_conn, tend2_conn,
k_ep1,
k_ep2) {
// k_in, k_mid, k_out - section stiffnesses -- supply k_mid = 0 if no middle section

// k_t1, k_t2 - tendon stiffnesses
// k_f1, k_f2 - friction interface stiffnesses -- supply 0 if interface is activating
// k_conn - external connection stiffness
// k_big, k_small -- stiffnesses to apply for tendon gap connections
// tend1_conn, tend2_conn -- for each tendon, a four number array that represents in binary
343
Appendix A: SCED Mechanics Simulator Source Code 344
// inleft, inright, outleft, outright (1 or 0 to show if connected or not)

// supply 0 for tend2_conn if no middle section
// k_ep1, k_ep2 - stiffness of friction between inner and outer end plates
// NODES:
// 0 - inner section left (rigid connection)
// 1 - inner section right
// 2 - inner tendon left
// 3 - inner tendon right
// 4 - outer section left
// 5 - outer section right / spring connection left
// 6 - spring connection right (point of load/def application)
// -------------------------------
// 7 - middle section left
// 8 - middle section right
// 9 - outer tendon left
// 10 - outer tendon right
// ELEMENTS
// 1 - inner section (N 0&1)-
// 2 - left friction interface (N 0&4)-
// 3 - right friction interface (N 1&5)-
// 4 - outer section (N 4&5)-
// 5 - inner tendon (N 2&3)-
// 6 - gap inner-inner left (N 0&2)-
// 7 - gap inner-inner right (N 1&3)-
// 8 - gap inner-outer left (N 2&4)- or inner-mid left (N 2&7)-
// 9 - gap inner-outer right (N 3&5)- or inner-mid right (N 3&8)-
// 10 - Spring Connection (N 5&6)-
// ------------------------------------
// 11 - middle section (N 7&8)-
// 12 - outer tendon (N 9&10)-
// 13 - gap mid-outer left (N 7&9)-
// 14 - gap mid-outer right (N 8&10)-
// 15 - gap outer-outer left (N 4&9)-
// 16 - gap outer-outer right (N 5&10)
// 17 - left end plate friction (N 2&9)
// 18 - right end plate friction (N 3&10)
if (tend1_conn[0] == 1) { var k_t1_il = k_big; }

else { var k_t1_il = k_small; }
if (tend1_conn[1] == 1) { var k_t1_ir = k_big; }
else { var k_t1_ir = k_small; }
if (tend1_conn[2] == 1) { var k_t1_ol = k_big; }
else { var k_t1_ol = k_small; }
if (tend1_conn[3] == 1) { var k_t1_or = k_big; }
else { var k_t1_or = k_small; }
if (k_mid != 0) {
if (tend2_conn[0] == 1) { var k_t2_il = k_big; }
else { var k_t2_il = k_small; }
if (tend2_conn[1] == 1) { var k_t2_ir = k_big; }
else { var k_t2_ir = k_small; }
if (tend2_conn[2] == 1) { var k_t2_ol = k_big; }
else { var k_t2_ol = k_small; }
if (tend2_conn[3] == 1) { var k_t2_or = k_big; }
else { var k_t2_or = k_small; }
}
if (k_mid == 0) {
var M = $M([
// 0 1 2 3 4
5 6
[k_in+k_f1+k_t1_il, -k_in, -k_t1_il, 0, -k_f1,
0, 0], //0
[-k_in, k_in+k_f2+k_t1_ir, 0, -k_t1_ir, 0,
-k_f2, 0], //1
[-k_t1_il, 0, k_t1+k_t1_il+k_t1_ol, -k_t1, -k_t1_ol,
0, 0], //2
[0, -k_t1_ir, -k_t1, k_t1+k_t1_ir+k_t1_or, 0,
-k_t1_or, 0], //3
[-k_f1, 0, -k_t1_ol, 0, k_f1+k_out+k_t1_ol,
-k_out, 0], //4
[0, -k_f2, 0, -k_t1_or, -k_out,
k_f2+k_out+k_t1_or+k_conn, -k_conn], //5
[0, 0, 0, 0, 0,

-k_conn, k_conn], //6

]);
} else {
var M = $M([
// 0 1 2 3 4
5 6 7 8 9
10
[k_in+k_f1+k_t1_il, -k_in, -k_t1_il, 0, -k_f1,
0, 0, 0, 0, 0,
0], //0
[-k_in, k_in+k_f2+k_t1_ir, 0, -k_t1_ir, 0,
-k_f2, 0, 0, 0, 0,
0], //1
[-k_t1_il, 0, k_t1+k_t1_il+k_t1_ol+k_ep1, -k_t1, 0,
0, 0, -k_t1_ol, 0, -k_ep1,
0], //2
[0, -k_t1_ir, -k_t1, k_t1+k_t1_ir+k_t1_or+k_ep2, 0,
0, 0, 0, -k_t1_or, 0,
-k_ep2], //3
[-k_f1, 0, 0, 0, k_f1+k_out+k_t2_ol,
-k_out, 0, 0, 0, -k_t2_ol,
0], //4
[0, -k_f2, 0, 0, -k_out,
k_f2+k_out+k_conn+k_t2_or, -k_conn, 0, 0, 0,
-k_t2_or], //5
[0, 0, 0, 0, 0,
-k_conn, k_conn, 0, 0, 0,
0], //6
[0, 0, -k_t1_ol, 0, 0,
0, 0, k_t1_ol+k_mid+k_t2_il, -k_mid, -k_t2_il,
0], //7
[0, 0, 0, -k_t1_or, 0,
0, 0, -k_mid, k_t1_or+k_mid+k_t2_ir, 0,
-k_t2_ir], //8
[0, 0, -k_ep1, 0, -k_t2_ol,
0, 0, -k_t2_il, 0,
k_t2+k_t2_il+k_t2_ol+k_ep1, -k_t2], //9
[0, 0, 0, -k_ep2, 0,
-k_t2_or, 0, 0, -k_t2_ir, -k_t2,
k_t2+k_t2_ir+k_t2_or+k_ep2], //10
]);
}
return M;
}
function FEFMatrixAll(FA, blMid, TI) {

// Add Fixed End Forces in Axial Elements
// FA is vector of all axial forces in elements
// TI is increase in Tendon Force
if (blMid == 0) {
var V = $M([
[-FA[INNER]-FA[F_LEFT]-FA[T1_IL]-FA[V_LEFT]-FA[V_EXT]], // 0
[FA[INNER]-FA[F_RIGHT]+FA[T1_IR]-FA[V_RIGHT]], // 1
[-FA[TENDON]+FA[T1_IL]+FA[T1_OL]-TI], // 2
[FA[TENDON]-FA[T1_IR]-FA[T1_OR]+TI], // 3
[-FA[OUTER]+FA[F_LEFT]-FA[T1_OL]], // 4
[FA[OUTER]+FA[F_RIGHT]-FA[CONNECT]+FA[T1_OR]+FA[V_RIGHT]+FA[V_LEFT]], // 5
[FA[CONNECT]+FA[V_EXT]], // 6
]);
} else {
var V = $M([
[-FA[INNER]-FA[F_LEFT]-FA[T1_IL]-FA[V_LEFT]-FA[V_EXT]], // 0
[FA[INNER]-FA[F_RIGHT]+FA[T1_IR]-FA[V_RIGHT]], // 1
[-FA[TENDON]+FA[T1_IL]+FA[T1_OL]-TI-FA[EP_LEFT]], // 2
[FA[TENDON]-FA[T1_IR]-FA[T1_OR]+TI-FA[EP_RIGHT]], // 3
[-FA[OUTER]+FA[F_LEFT]-FA[T2_OL]], // 4
[FA[OUTER]+FA[F_RIGHT]-FA[CONNECT]+FA[T2_OR]+FA[V_RIGHT]+FA[V_LEFT]], // 5
[FA[CONNECT]+FA[V_EXT]], // 6
[-FA[MID]-FA[T1_OL]-FA[T2_IL]], // 7
[FA[MID]+FA[T1_OR]+FA[T2_IR]], // 8
[-FA[TENDON2]+FA[T2_IL]+FA[T2_OL]-TI+FA[EP_LEFT]], // 9
[FA[TENDON2]-FA[T2_IR]-FA[T2_OR]+TI+FA[EP_RIGHT]], // 10
]);

return V;
}
function ForceMatrix(A_Force, blMid) {

// Create Force Matrix
// for normal SCED - 7 rows
// for T-SCED - 11 rows
// A_force is the applied force at node 6
if (blMid) { var V = Matrix.Zero(11,1);}

else { var V = Matrix.Zero(7,1);}
V.elements[6][0] = A_Force;
return V;
}
function FindAxial(NodeA, NodeB, DispVect, AStiff, FixedEndForce) {
// Function to Determine axial force of asingle element

// NodeA and NodeB - Element Nodes
// DispVect - Output Displacement Vector
// AStiff - Current Axial Stiffness of Element (Scalar)
// FixedEndForce - Axial Force in Element from Previous Step
Def = $M([[DispVect.elements[NodeA][0]],[DispVect.elements[NodeB][0]],]);
FEF = $M([[-FixedEndForce],[FixedEndForce],]);
Stiff = $M([[AStiff, -AStiff],[-AStiff, AStiff],]);
Force = (Stiff.multiply(Def)).add(FEF);
return Force.elements[1][0];
}
function CalculateSCED(BIN, asWorker) {
// Check Data Fidelity

if (arraymin(BIN.Init_Pos, BIN.test_set) != 0) {
if (asWorker) {
postMessage({messageType: "ErrorMsg", workdata: 'One of the tubes must have an initial position of 0'}
);
} else {
alert('One of the tubes must have an initial position of 0');
setTimeout(Cancel(),2000);
}
return;
}
for (var i = 0; i <= BIN.test_set.length; i++) {
if ((BIN.Init_Pos[BIN.test_set[i]]+BIN.Sect_Len[BIN.test_set[i]]) > arraymax(BIN.Sect_Len, BIN.test_set)) {
if (asWorker) {
postMessage({messageType: "ErrorMsg", workdata: 'Short tubes must remain within the length of the
longest tube'}
);
} else {
alert('Short tubes must remain within the length of the longest tube');
setTimeout(Cancel(),2000);
}
return;
}
}
//---------------------------------------------------------
// Calculate the pretensioning behaviour.
// NOTE: Left side is connected to the outer section
//---------------------------------------------------------
var C_RLOC = 100; // Connection Location Relative to Right end of outer section
// First Step -- s = step

var s = 0;
// Initialize Output Variables

var Left_Loc = [];

var Right_Loc = [];
var Cur_Length = [];
var Axial = [];
var Elongation = [];
var Shift = [];
var Stiffness = [];
var Steps = [];
//Effective tendon locations (not same as true tendon locations -- used only for
// determining stiffness matrix
var TendLoc = [];
TendLoc[TENDON] = [];
TendLoc[TENDON2] = [];
// Won't be used in Preliminary Tensioning Analysis

var Overall_Def = [];
var Overall_Force = [];
var Friction_Act = [];
var d_F = [];
Steps[0] = 0;
// Initialize Locations and Lengths of elements

BIN.Sect_Len[CONNECT] = 100; // Connection Length Dummy Value for use with effective stiffness
Left_Loc[s] = [];
Left_Loc[s][INNER] = BIN.Init_Pos[INNER]; // Inner Section
Left_Loc[s][MID] = BIN.Init_Pos[MID]; // Middle Section
Left_Loc[s][OUTER] = BIN.Init_Pos[OUTER]; // Outer Section
if (BIN.blMiddle) {
var testset1 = [INNER, MID];
} else {
var testset1 = [INNER, OUTER];
}
Left_Loc[s][TENDON] = arraymin(Left_Loc[s], testset1); // Tendon
Left_Loc[s][CONNECT] = BIN.Init_Pos[OUTER] + BIN.Sect_Len[OUTER];
Right_Loc[s] = [];
Right_Loc[s][INNER] = BIN.Sect_Len[INNER]+BIN.Init_Pos[INNER];
Right_Loc[s][MID] = BIN.Sect_Len[MID]+BIN.Init_Pos[MID];
Right_Loc[s][OUTER] = BIN.Sect_Len[OUTER]+BIN.Init_Pos[OUTER];
Right_Loc[s][TENDON] = arraymax(Right_Loc[s], testset1);
Right_Loc[s][CONNECT] = Left_Loc[s][CONNECT]+BIN.Sect_Len[CONNECT];
var testset2 = [MID, OUTER];
if (BIN.blMiddle) {
Left_Loc[s][TENDON2] = arraymin(Left_Loc[s], testset2); // Tendon
Right_Loc[s][TENDON2] = arraymax(Right_Loc[s], testset2); // Tendon
}
TendLoc[TENDON][LEFT] = Left_Loc[s][TENDON];
TendLoc[TENDON][RIGHT] = Right_Loc[s][TENDON];
TendLoc[TENDON2][LEFT] = Left_Loc[s][TENDON2];
TendLoc[TENDON2][RIGHT] = Right_Loc[s][TENDON2];
Cur_Length[s] = [];
Axial[s] = [0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0];
Elongation[s] = [];
Shift[s] = [];
Stiffness = [];
for (var i = INNER; i <= CONNECT; i++) {

Cur_Length[s][i] = Right_Loc[s][i] - Left_Loc[s][i];
if (i != TENDON && i != TENDON2) {
// Steel Sections Start out with no load or elongation
Axial[s][i] = 0;
Elongation[s][i] = 0;
Shift[s][i] = 0;
} else {
// The tendon has an initial load and elongation
Axial[s][i] = 0;
Elongation[s][i] = 0;
BIN.Sect_Len[i] = Cur_Length[s][i];
}
if (i != CONNECT) {
// Initialize Stiffness of all steel sections and tendon
Stiffness[i] = BIN.Esp[i]*BIN.Sect_Area[i] / BIN.Sect_Len[i];
} else {

// Exterior Connection must have an effective stiffness of

// elements in series.
var StiffSum = 0;
for(var k = 0; k<=2; k++) {
if(isNaN(BIN.Conn_Len[k])||isNaN(BIN.Conn_Area[k])||BIN.Conn_Len[k]==0||BIN.Conn_Area[k]==0) {
continue;
}
StiffSum = StiffSum + 1/(BIN.Esp[INNER]*BIN.Conn_Area[k] / BIN.Conn_Len[k]);
}
Stiffness[i] = 1/(StiffSum);
}
}
if (asWorker) {
postMessage({messageType: "TextA", workdata: "Calculating Pretension"});
postMessage({messageType: "ProgressA", workdata: 5});
} else {
$( "#progressbarmain" ).progressbar("value", 5);
$( "#progmessA" ).prop("innerHTML", "Calculating Pretension");
}
//////////////////////////////////////////////////////////////
// Apply Pretensioning to Brace
// (as progressive fixed end force in both tendons)
//////////////////////////////////////////////////////////////
var AddedForce = 0;
var FS = BIN.def_step;
var M_FEF = [];
var M_Force = [];
var M_SCED = [];
var M_SCED_MOD = [0];
var M_SCED_INV = [];
var M_REACT = [];
var M_Disp = [];
var RESTRAINT = 10E12;
var BIG_N = 10E7;
var SMALL_N = 10E-5;
// values determining which tubes each tendon abuts at each end

var gap_t1 = [];
var gap_t2 = [];
while (Axial[s][TENDON] < BIN.TendPreten) {
s = s + 1; //next step
Steps[s] = s;
Left_Loc[s] = [];
Right_Loc[s] = [];
Cur_Length[s] = [];
Axial[s] = [];
Elongation[s] = [];
Shift[s] = [];
// Below are set to zero each step to avoid undefined values

Overall_Def[s] = 0;
Overall_Force[s] = 0;
Friction_Act[s] = [];
Friction_Act[s][LEFT] = 0;
Friction_Act[s][RIGHT] = 0;
Friction_Act[s][LEFT_EP] = 0;
Friction_Act[s][RIGHT_EP] = 0;
// Determine which elements are in contact and find stiffness matrix

if (BIN.blMiddle) {
if (Left_Loc[s-1][INNER] <= Left_Loc[s-1][TENDON]) { gap_t1[0] = 1; }
else {gap_t1[0] = 0;}
if (Right_Loc[s-1][INNER] >= Right_Loc[s-1][TENDON]) { gap_t1[1] = 1; }
else {gap_t1[1] = 0;}
if (Left_Loc[s-1][MID] <= Left_Loc[s-1][TENDON]) { gap_t1[2] = 1; }
else {gap_t1[2] = 0;}
if (Right_Loc[s-1][MID] >= Right_Loc[s-1][TENDON]) { gap_t1[3] = 1; }
else {gap_t1[3] = 0;}
if (Left_Loc[s-1][MID] <= Left_Loc[s-1][TENDON2]) { gap_t2[0] = 1; }
else {gap_t2[0] = 0;}
if (Right_Loc[s-1][MID] >= Right_Loc[s-1][TENDON2]) { gap_t2[1] = 1; }

else {gap_t2[1] = 0;}

if (Left_Loc[s-1][OUTER] <= Left_Loc[s-1][TENDON2]) { gap_t2[2] = 1; }
else {gap_t2[2] = 0;}
if (Right_Loc[s-1][OUTER] >= Right_Loc[s-1][TENDON2]) { gap_t2[3] = 1; }
else {gap_t2[3] = 0;}
M_SCED = SCEDMatrix(Stiffness[INNER], Stiffness[MID], Stiffness[OUTER], Stiffness[TENDON],

Stiffness[TENDON2], 0, 0, Stiffness[CONNECT], BIN.Bear_Stiff, SMALL_N, gap_t1, gap_t2,0,0);
} else {
else {gap_t1[0] = 0;}
else {gap_t1[1] = 0;}
if (Left_Loc[s-1][OUTER] <= Left_Loc[s-1][TENDON]) { gap_t1[2] = 1; }
else {gap_t1[2] = 0;}
if (Right_Loc[s-1][OUTER] >= Right_Loc[s-1][TENDON]) { gap_t1[3] = 1; }
else {gap_t1[3] = 0;}
M_SCED = SCEDMatrix(Stiffness[INNER], 0, Stiffness[OUTER], Stiffness[TENDON], 0, 0, 0,

Stiffness[CONNECT], BIN.Bear_Stiff, SMALL_N, gap_t1, 0,0,0);
// Fixed end Forces provided by tendons

if (BIN.blMiddle) {
M_FEF = FEFMatrixAll(Axial[s-1], 1, FS);
} else {
M_FEF = FEFMatrixAll(Axial[s-1], 0, FS);
// Force Matrix is a column of zeros

M_Force = ForceMatrix(0, BIN.blMiddle);
// Multiply Restrained DOF diagonal values by big num

M_SCED_MOD = M_SCED.dup();
M_SCED_MOD.elements[0][0] *= RESTRAINT;
// Find Inverse of Stiffness Matrix

M_SCED_INV = M_SCED_MOD.inverse();
// Solve for Displacements

M_Disp = M_SCED_INV.multiply(M_Force.subtract(M_FEF));
// Update Element Positions and Forces

Left_Loc[s][INNER] = Left_Loc[s-1][INNER] + M_Disp.elements[0][0];
Right_Loc[s][INNER] = Right_Loc[s-1][INNER] + M_Disp.elements[1][0];
Left_Loc[s][OUTER] = Left_Loc[s-1][OUTER] + M_Disp.elements[4][0];
Right_Loc[s][OUTER] = Right_Loc[s-1][OUTER] + M_Disp.elements[5][0];
Cur_Length[s][INNER] = Right_Loc[s][INNER] - Left_Loc[s][INNER];

Cur_Length[s][OUTER] = Right_Loc[s][OUTER] - Left_Loc[s][OUTER];
Elongation[s][INNER] = (Cur_Length[s][INNER] - Cur_Length[s-1][INNER]);

Elongation[s][OUTER] = (Cur_Length[s][OUTER] - Cur_Length[s-1][OUTER]);
Axial[s][INNER] = FindAxial(0,1,M_Disp,Stiffness[INNER],Axial[s-1][INNER]);
Axial[s][OUTER] = FindAxial(4,5,M_Disp,Stiffness[OUTER],Axial[s-1][OUTER]);
if (BIN.blMiddle) {
Left_Loc[s][MID] = Left_Loc[s-1][MID] + M_Disp.elements[7][0];
Right_Loc[s][MID] = Right_Loc[s-1][MID] + M_Disp.elements[8][0];
Cur_Length[s][MID] = Right_Loc[s][MID] - Left_Loc[s][MID];

Elongation[s][MID] = (Cur_Length[s][MID] - Cur_Length[s-1][MID]);
Axial[s][MID] = FindAxial(7,8,M_Disp,Stiffness[MID],Axial[s-1][MID]);
}
Left_Loc[s][TENDON] = Left_Loc[s-1][TENDON] + M_Disp.elements[2][0]; // Tendon

Right_Loc[s][TENDON] = Right_Loc[s-1][TENDON] + M_Disp.elements[3][0];
if (BIN.blMiddle) {
Left_Loc[s][TENDON2] = Left_Loc[s-1][TENDON2] + M_Disp.elements[9][0];
Right_Loc[s][TENDON2] = Right_Loc[s-1][TENDON2] + M_Disp.elements[10][0];
}

Axial[s][TENDON] = FindAxial(2,3,M_Disp,Stiffness[TENDON],Axial[s-1][TENDON] + FS);

Elongation[s][TENDON] = Axial[s][TENDON] * BIN.Sect_Len[TENDON] / (BIN.Ep * BIN.Sect_Area[TENDON]);
if (BIN.blMiddle) {
Axial[s][TENDON2] = FindAxial(9,10,M_Disp,Stiffness[TENDON2],Axial[s-1][TENDON2] + FS);
Elongation[s][TENDON2] = Axial[s][TENDON2] * BIN.Sect_Len[TENDON2] / (BIN.Ep * BIN.Sect_Area[TENDON2]);
}
// Connection
Left_Loc[s][CONNECT] = Right_Loc[s][OUTER];
Right_Loc[s][CONNECT] = Left_Loc[s][CONNECT] + BIN.Sect_Len[CONNECT];
Cur_Length[s][CONNECT] = BIN.Sect_Len[CONNECT];
Elongation[s][CONNECT] = 0;
Axial[s][CONNECT] = 0;
// Friction and Viscous (Not Activated)

Axial[s][F_LEFT] = 0;
Axial[s][F_RIGHT] = 0;
Axial[s][EP_LEFT] = 0;
Axial[s][EP_RIGHT] = 0;
Axial[s][V_LEFT] = 0;
Axial[s][V_RIGHT] = 0;
Axial[s][V_EXT] = 0;
// Connections for Tendon to Tubes

if (BIN.blMiddle) {
Axial[s][T1_IL] = FindAxial(0,2,M_Disp,BIN.Bear_Stiff*gap_t1[0],Axial[s-1][T1_IL]);
Axial[s][T1_IR] = FindAxial(3,1,M_Disp,BIN.Bear_Stiff*gap_t1[1],Axial[s-1][T1_IR]);
Axial[s][T1_OL] = FindAxial(7,2,M_Disp,BIN.Bear_Stiff*gap_t1[2],Axial[s-1][T1_OL]);
Axial[s][T1_OR] = FindAxial(3,8,M_Disp,BIN.Bear_Stiff*gap_t1[3],Axial[s-1][T1_OR]);
} else {
}
M_Disp_PREV = M_Disp.dup();
// Mark this step as the end of Post-Tensioning

var step_pret = s;
//---------------------------------------------------------
// Calculate the full hysteretic behaviour.
// NOTE: Left side is connected to the outer section
// Deformation is applied to outer section
//---------------------------------------------------------
// These are defined above before Tensioning Analysis

Overall_Def[s] = 0;
Overall_Force[s] = 0;
Friction_Act[s][RIGHT] = 0; // If == 1 Positive Force Activated
// If == -1 Negative Force Activated
// If == 0 Not Activated (High Stiffness)
Axial[s][F_LEFT] = 0;
Axial[s][F_RIGHT] = 0;
var Def_Lim;
var FS_Large;

var FS_Small;
var FS; // DISPLACEMENT CONTROLLED 0.1mm per step
var FS_prev;
// Variables for Viscous Force Determination

var t_prev = 0; // Current time
var t_curr;
var dt; // time step
var omega = 2 * Math.PI * BIN.ViscFreq; // angular frequency
var c_damp = BIN.ViscConst;
if (omega == 0) {
c_damp = 0;
omega = 1;
}
var A_curr; // current amplitude
var bl_shift; // phase shift
var v_int; // velocities
var v_ext;
var def_pct;
var def_pct_use;
var a_sin;
var countdown; // wait a certain number of steps when switching directions;
var cdnum = 4;
var FrDef = [];

var FrFEF = [];
var FrStiff = [];
var FrForce = [];
// Friction fuse Stiffness

Stiffness[F_LEFT] = BIN.Interface_Stiff;
Stiffness[F_RIGHT] = BIN.Interface_Stiff;
Stiffness[EP_LEFT] = BIN.Interface_Stiff;
Stiffness[EP_RIGHT] = BIN.Interface_Stiff;
for (var i = 1; i <= 4; i++) { // should go to 4
if (i == 1 || i == 5) {
Def_Lim = BIN.Def_Lim_A;
A_curr = BIN.Def_Lim_A;
bl_shift = false;
countdown = -1;
FS_Large = BIN.def_stepH * nsign(BIN.Def_Lim_A);

FS_Small = BIN.def_stepHsm * nsign(BIN.Def_Lim_A);
if (asWorker) {
postMessage({messageType: "TextA", workdata: "First Stage"});
} else {
$( "#progmessA" ).prop("innerHTML", "First Stage");
}
} else if (i == 2 || i == 6) {
Def_Lim = 0;
A_curr = BIN.Def_Lim_A;
bl_shift = true;
countdown = cdnum;
FS_Large = BIN.def_stepH * -nsign(BIN.Def_Lim_A);

FS_Small = BIN.def_stepHsm * -nsign(BIN.Def_Lim_A);
if (asWorker) {
postMessage({messageType: "TextA", workdata: "Second Stage"});
} else {
$( "#progmessA" ).prop("innerHTML", "Second Stage");
}
} else if (i == 3 || i == 7) {
Def_Lim = BIN.Def_Lim_B;
A_curr = BIN.Def_Lim_B;
bl_shift = false;
FS_Large = BIN.def_stepH * nsign(BIN.Def_Lim_B);

FS_Small = BIN.def_stepHsm * nsign(BIN.Def_Lim_B);

if (asWorker) {
postMessage({messageType: "TextA", workdata: "Third Stage"});
} else {
$( "#progmessA" ).prop("innerHTML", "Third Stage");
}
} else if (i == 4 || i == 8) {
Def_Lim = 0;
A_curr = BIN.Def_Lim_B;
bl_shift = true;
countdown = cdnum;
FS_Large = BIN.def_stepH * -nsign(BIN.Def_Lim_B);

FS_Small = BIN.def_stepHsm * -nsign(BIN.Def_Lim_B);
if (asWorker) {
postMessage({messageType: "TextA", workdata: "Fourth Stage"});
} else {
$( "#progmessA" ).prop("innerHTML", "Fourth Stage");
}
}
while ((Overall_Def[s]*nsign(FS_Large)) < (Def_Lim*nsign(FS_Large))) {
s = s + 1; //next step
// Add new step to Arrays
Steps[s] = s;
Left_Loc[s] = [];
Right_Loc[s] = [];
Cur_Length[s] = [];
Axial[s] = [];
Elongation[s] = [];
Shift[s] = [];
d_F[s] = [];
// Increase Overall Deformation by one Step

// Step Size Depending on whether friction devices have activated
if (Math.abs(Overall_Def[s-1]) > 4.0) {
FS = FS_Large;
}
else {
FS = FS_Small;
}
Overall_Def[s] = Overall_Def[s-1] + FS;
// Determine whether friction fuses are activating

if (Friction_Act[s-1][LEFT] == 0) {kf_left = Stiffness[F_LEFT];}
else {kf_left = SMALL_N;}
if (Friction_Act[s-1][RIGHT] == 0) {kf_right = Stiffness[F_RIGHT];}
else {kf_right = SMALL_N;}
// Determine timestep for viscous damping

// 1 - Find current time
def_pct = Overall_Def[s-1]/A_curr;
if (def_pct > 1) {def_pct_use = 1;} else {def_pct_use = def_pct;}
if (bl_shift == true) {
a_sin = Math.asin(-def_pct_use);
t_curr = (a_sin+Math.PI)/omega;
} else {
a_sin = Math.asin(def_pct_use);
t_curr = a_sin/omega;
}
// 2 - Find current timestep
dt = t_curr - t_prev;
// 3 - Find velocity for the interior and parallel dampers
if (dt == 0 || countdown > 0 ) {
v_int = 0; v_ext = 0;
} else {
v_int = ((Right_Loc[s-1][OUTER] - Left_Loc[s-1][INNER]) - (Right_Loc[s-2][OUTER] -

Left_Loc[s-2][INNER])) / dt;
v_ext = (Overall_Def[s-1] - Overall_Def[s-2]) / dt;
}
// 4 - Axial Force
if (BIN.blInterior == true) {
Axial[s-1][V_LEFT] = BIN.ViscConst*v_int;
Axial[s-1][V_RIGHT] = 0;
Axial[s-1][V_EXT] = 0;
} else {
Axial[s-1][V_LEFT] = 0;
Axial[s-1][V_RIGHT] = 0;
Axial[s-1][V_EXT] = BIN.ViscConst*v_ext;
}
t_prev = t_curr;
countdown = countdown - 1;
// Determine which elements are in contact and find stiffness matrix

if (BIN.blMiddle) {
if (Friction_Act[s-1][LEFT_EP] == 0) {kfep_left = Stiffness[EP_LEFT];}

else {kfep_left = SMALL_N;}
if (Friction_Act[s-1][RIGHT_EP] == 0) {kfep_right = Stiffness[EP_RIGHT];}
else {kfep_right = SMALL_N;}

else {gap_t1[0] = 0;}
else {gap_t1[1] = 0;}
if (Left_Loc[s-1][MID] <= Left_Loc[s-1][TENDON]) { gap_t1[2] = 1; }
else {gap_t1[2] = 0;}
if (Right_Loc[s-1][MID] >= Right_Loc[s-1][TENDON]) { gap_t1[3] = 1; }
else {gap_t1[3] = 0;}
if (Left_Loc[s-1][MID] <= Left_Loc[s-1][TENDON2]) { gap_t2[0] = 1; }
else {gap_t2[0] = 0;}
if (Right_Loc[s-1][MID] >= Right_Loc[s-1][TENDON2]) { gap_t2[1] = 1; }
else {gap_t2[1] = 0;}
if (Left_Loc[s-1][OUTER] <= Left_Loc[s-1][TENDON2]) { gap_t2[2] = 1; }
else {gap_t2[2] = 0;}
if (Right_Loc[s-1][OUTER] >= Right_Loc[s-1][TENDON2]) { gap_t2[3] = 1; }
else {gap_t2[3] = 0;}
M_SCED = SCEDMatrix(Stiffness[INNER], Stiffness[MID], Stiffness[OUTER], Stiffness[TENDON],

Stiffness[TENDON2], kf_left, kf_right, Stiffness[CONNECT], BIN.Bear_Stiff, SMALL_N, gap_t1,
gap_t2, kfep_left, kfep_right);
} else {
else {gap_t1[0] = 0;}
else {gap_t1[1] = 0;}
if (Left_Loc[s-1][OUTER] <= Left_Loc[s-1][TENDON]) { gap_t1[2] = 1; }
else {gap_t1[2] = 0;}
if (Right_Loc[s-1][OUTER] >= Right_Loc[s-1][TENDON]) { gap_t1[3] = 1; }
else {gap_t1[3] = 0;}
M_SCED = SCEDMatrix(Stiffness[INNER], 0, Stiffness[OUTER], Stiffness[TENDON], 0, kf_left,

kf_right, Stiffness[CONNECT], BIN.Bear_Stiff, SMALL_N, gap_t1, 0, 0, 0);
// Fixed end Forces provided by tendons

if (BIN.blMiddle) {
M_FEF = FEFMatrixAll(Axial[s-1], 1, 0);
} else {
M_FEF = FEFMatrixAll(Axial[s-1], 0, 0);
}
// Multiply Restrained DOF diagonal values by big num (Payne and Irons Method)
M_SCED_MOD = M_SCED.dup();
// Force Matrix includes a Payne and Irons method force in sixth row
M_Force = ForceMatrix(FS*M_SCED_MOD.elements[6][6], BIN.blMiddle);

// Find Inverse of Stiffness Matrix

M_SCED_INV = M_SCED_MOD.inverse();
// Solve for Displacements

M_Disp = M_SCED_INV.multiply(M_Force.subtract(M_FEF));
// Update Element Positions and Forces

Left_Loc[s][INNER] = Left_Loc[s-1][INNER] + M_Disp.elements[0][0];
Right_Loc[s][INNER] = Right_Loc[s-1][INNER] + M_Disp.elements[1][0];
Left_Loc[s][OUTER] = Left_Loc[s-1][OUTER] + M_Disp.elements[4][0];
Right_Loc[s][OUTER] = Right_Loc[s-1][OUTER] + M_Disp.elements[5][0];
Cur_Length[s][INNER] = Right_Loc[s][INNER] - Left_Loc[s][INNER];

Cur_Length[s][OUTER] = Right_Loc[s][OUTER] - Left_Loc[s][OUTER];
Elongation[s][INNER] = (Cur_Length[s][INNER] - Cur_Length[s-1][INNER]);

Elongation[s][OUTER] = (Cur_Length[s][OUTER] - Cur_Length[s-1][OUTER]);
Axial[s][INNER] = FindAxial(0,1,M_Disp,Stiffness[INNER],Axial[s-1][INNER]);
Axial[s][OUTER] = FindAxial(4,5,M_Disp,Stiffness[OUTER],Axial[s-1][OUTER]);
if (BIN.blMiddle) {
Left_Loc[s][MID] = Left_Loc[s-1][MID] + M_Disp.elements[7][0];
Right_Loc[s][MID] = Right_Loc[s-1][MID] + M_Disp.elements[8][0];
Cur_Length[s][MID] = Right_Loc[s][MID] - Left_Loc[s][MID];

Elongation[s][MID] = (Cur_Length[s][MID] - Cur_Length[s-1][MID]);
Axial[s][MID] = FindAxial(7,8,M_Disp,Stiffness[MID],Axial[s-1][MID]);
}
Left_Loc[s][TENDON] = Left_Loc[s-1][TENDON] + M_Disp.elements[2][0]; // Tendon

Right_Loc[s][TENDON] = Right_Loc[s-1][TENDON] + M_Disp.elements[3][0];
if (BIN.blMiddle) {
Left_Loc[s][TENDON2] = Left_Loc[s-1][TENDON2] + M_Disp.elements[9][0];
Right_Loc[s][TENDON2] = Right_Loc[s-1][TENDON2] + M_Disp.elements[10][0];
}
Axial[s][TENDON] = FindAxial(2,3,M_Disp,Stiffness[TENDON],Axial[s-1][TENDON]);
Elongation[s][TENDON] = Axial[s][TENDON] * BIN.Sect_Len[TENDON] / (BIN.Ep * BIN.Sect_Area[TENDON]);
if (BIN.blMiddle) {
Axial[s][TENDON2] = FindAxial(9,10,M_Disp,Stiffness[TENDON2],Axial[s-1][TENDON2]);
Elongation[s][TENDON2] = Axial[s][TENDON2] * BIN.Sect_Len[TENDON2] / (BIN.Ep *
BIN.Sect_Area[TENDON2]);
}
// Find overall force by finding reaction at node 0

M_REACT = (M_SCED.multiply(M_Disp)).add(M_FEF);
Overall_Force[s] = -M_REACT.elements[0][0];
// Connection
Left_Loc[s][CONNECT] = Right_Loc[s][OUTER];
Right_Loc[s][CONNECT] = Right_Loc[s-1][CONNECT] + M_Disp.elements[6][0];
Cur_Length[s][CONNECT] = Right_Loc[s][CONNECT] - Left_Loc[s][CONNECT];
Elongation[s][CONNECT] = (Cur_Length[s][CONNECT] - Cur_Length[s-1][CONNECT]);;
Axial[s][CONNECT] = FindAxial(5,6,M_Disp,Stiffness[CONNECT],Axial[s-1][CONNECT]);
// Check if increased force in the friction interfaces causes

// them to slip (or stop slipping)
// Find Axial Force in Friction Fuses

Axial[s][F_LEFT] = FindAxial(0,4,M_Disp,kf_left,Axial[s-1][F_LEFT]);
Axial[s][F_RIGHT] = FindAxial(1,5,M_Disp,kf_right,Axial[s-1][F_RIGHT]);
if (BIN.blMiddle) {
Axial[s][EP_LEFT] = FindAxial(2,9,M_Disp,kfep_left,Axial[s-1][EP_LEFT]);
Axial[s][EP_RIGHT] = FindAxial(3,10,M_Disp,kfep_right,Axial[s-1][EP_RIGHT]);
} else {
}
if (nsign(FS*FS_prev) == -1) {
// IF Direction of loading has changed set friction activations
// to off at both ends so that brace must activate friction in

// other direction.
} else {
// Otherwise, check friction activation conditions

if (Axial[s][F_RIGHT] >= BIN.Friction[RIGHT] && nsign(FS) == 1) {
// Set activation flag
} else if (Axial[s][F_RIGHT] <= -BIN.Friction[RIGHT] && nsign(FS) == -1) {
Friction_Act[s][RIGHT] = -1;
} else {
}
if (Axial[s][F_LEFT] >= BIN.Friction[LEFT] && nsign(FS) == 1) {
} else if (Axial[s][F_LEFT] <= -BIN.Friction[LEFT] && nsign(FS) == -1) {
Friction_Act[s][LEFT] = -1;
} else {
}
if (BIN.blMiddle) {
if (Axial[s][EP_RIGHT] >= BIN.EPFriction[RIGHT] && nsign(FS) == 1) {

} else if (Axial[s][EP_RIGHT] <= -BIN.EPFriction[RIGHT] && nsign(FS) == -1) {
Friction_Act[s][RIGHT_EP] = -1;
} else {
}
if (Axial[s][EP_LEFT] >= BIN.EPFriction[LEFT] && nsign(FS) == 1) {
} else if (Axial[s][EP_LEFT] <= -BIN.EPFriction[LEFT] && nsign(FS) == -1) {
Friction_Act[s][LEFT_EP] = -1;
} else {
}
} else {
}
}
// Connections for Tendon to Tubes
if (BIN.blMiddle) {
} else {
}
M_Disp_PREV = M_Disp.dup();
// Record value of current F_Step

FS_prev = FS;
}
}
//Finish the set of viscous forces

// Mark current step as last step

var step_end = s;
// Return Full Data Set

var sentdata = new resp_data(Steps, Left_Loc, Right_Loc, Cur_Length, Axial, Elongation, Shift, Stiffness,
step_pret, step_end, BIN.blMiddle, Overall_Def, Overall_Force, Friction_Act, BIN);
if (asWorker) {
postMessage({messageType: "Hyst", workdata: sentdata});
} else {
$('#butCalculate').button("enable", true);
$('#butCancel').button("enable", false);
$('#CalculationMessages').slideUp();
$('#FullOutput').slideDown();
rd = sentdata;
$( "#plotrange" ).slider("option", "max", rd.step_end);
$( "#plotrange" ).slider("option", "value", rd.step_pret);
$("#ScaleIn").prop("value", 100);
refreshPlots();
}

APPENDIX B: HC-SCED DRAWINGS
357
Bolts MTS Upper Plate
MTS Connection Upper Base Plate (S3)
2" Bolts (@ EPM)
12 Required at Top Upper End Plate (S4)
Interior Friction Interfaces: Upper Friction Interface
Appendix B: HC-SCED Drawings
3/4" A325 Bolts

4x6 = 24 Required
Exterior Friction Interfaces: Side Width Front Width

1-1/2" A325 Bolts 508 894
Upper Conn Len 400
16 Required
MTS Connection Inner Tube (within) (S2)
2" Bolts (@ EPM)
8 Required at Bottom Outer Tube (S3)
Total Test Length 7670
Brace Member Length 6200
Lower Friction Interface

Lower End Plate (S4)
1070 Lower Conn Len
Lower Base Plate (S4)

Existing Connection Plate

MTS Base
Test Assembly Side Elevation Front Elevation
All dimensions in mm unless otherwise noted. Steel Material: ASTM A572 Grade 50 for outer tube, CSA G40.21 350W for all other steel.
Project Title REV Date
UNIVERSITYȱofȱTORONTO 4MN Self-Centering Energy Dissipative Brace 7 2009/07/21
Drawing Title Designer Drawing No.
DepartmentȱofȱCivilȱEngineering
Testing Assembly Summary J.Erochko S1
358
890 720
13
426
400
13
1610 Shape of Original
WWF Section without
50mm CONNECTION PLATE
SHAPE DETAIL 50mm Plate
SCALE 1 : 20
450
50 mm Plate in place of
WWF Web (see Shape

Detail Above) 2.0 thick stainless WWF450x228
25
steel plate (Typ)

R2 890 720 720 50
426
450
156
0T
115
232
94
y p.
25
10
16
426
400
R1 10
6.0 thick 1T Typ
stainless 440 540 y p. 12 540 SECTION A
450
steel plate Slot Slot 16mm Plates TYP Slot SCALE 1 : 20
both sides Typ
1610 2 Paint inner section at ends
6 only (200mm from end of WWF)
Plastic Pads Paint with colour that contrasts with
Thickness TBD the colour of the outer section (tube)
450
A B
Do not paint stainless steel surfaces.

16
Length of Inner and Outer
Section must be exactly the
same. Both ends should be
B faced and polished so that
A they are exactly
2000 500 500 1500 perpendicularflat and flat
(for compatibility with the
6200 (1/1000 inch tolerance) end plate)
7090
SECTION B
INNER SECTION SCALE 1 : 50 SCALE 1 : 20
UNIVERSITYȱofȱTORONTO 4MN Self-Centering Energy Dissipative Brace 7 2009/07/21
Inner Section Detail J.Erochko S2
359
508
0
R2
20
15.90
Outer Tube
508
Showing Slots Only
89
Paint outer section, angles and connection
plates. Paint with colour that contrasts with 13

120
127
the colour of the inner section (WWF)

1100 100 196
Do not paint inside of angles.
465
930 HSS508x508x15.9 80 D
203
80 C SECTION C
SCALE 1 : 20
130
10mm Stiffener 800
75
Plates (Typ) 10mm

guide 75 75 125 275
127
4xL127x89x13 4xL203x152x19 plates
140 400 22
508
T y p. 240 400 300 (Typ)
0
C D R2
465
10mm Cover 6
10 Plate Typ 20
20 680
680 20
in Tube
20 Slot

6
508
110 78 187 125
12 CL
50
800
152
685 Typ
6 15
193
203
Typ 19
1470 Slot 800 Slot y p.
T 190
6200 (1/1000 inch tolerance) 64
6730
SECTION D
SCALE 1 : 20
UNIVERSITYȱofȱTORONTO Large Self-Centering Energy Dissipative Brace 7 2009/07/21
Outer Section Detail J.Erochko S3
360
90 180 Typ
95
340
95
64
506
120mm Plate
110
100
426 Face and Polish

38.10 one side of each
38.10 63 100 Typ
50
end plate (the side
100
54 that will be in
506
430
62 but not set until 63 contact with the
Inner plate is installed inner and outer
sections.
68 Do not paint that
100 Typ.
Typ. 55
600
40 55
38 Ty
p. polished surface.
196 LOWER UPPER
147
END PLATE SCALE 1:20
100
530 720
LOWER BASE PLATE SCALE 1:20

Bolt Slot to Bolt

and Bolt Hole to Bolt Distance between Inner and Clamped interface
Tolerance = 1.5mm Tube Slot to Friction Plate Outer Sections = 13mm No tolerance
all around Typical Tolerance = 2mm each side All around E (as of right now)
(but plate is clamped by angles)
End Plate
Tolerance
= 1mm Friction Interface Welded in place after
every side Clamped Together E SECTION E
SCALE 1 : 20
COMPLETE BRACE WITH TOLERANCES SCALE 1:50
Plate Details and Brace Tolerances J.Erochko S4
361
15 175
15
196
30
15 181
70 10
1.70
22
154
30
193
0
87
R2
17 30
120
22 0
18
R2
19
35 SECTION G
13
DETAIL F
25
SCALE 1 : 5
25
80
G
76 100 130 40
190
13
127
65
10mm STIFFENER PLATES (8 of Each)
260 80 G 270 89 SCALE 1:5
930 460
ANGLE BOTTOM END SCALE 1:20

100
80
Two Required As Above, Two Required as Mirror Image
40
400
40 80 22
70
1.70 10mm GUIDE PLATE
20mm WASHER PLATE SCALE 1:10
SCALE 1:20
30
685
50
70 10
22
6
30
10mm COVER PLATE SCALE 1:20
17 30
39
22

DETAIL H 35 94 440 50
80 SECTION J R20
J SCALE 1 : 5 88 540 88
R11
70
100
414
19
120
57
40
203
22
60
86
30
716
460 80 240 152 584
J
1100 3.3mm FRICTION PAD SCALE 1:5
Carlisle Pad Material NF916 1.5mm THICK STAINLESS STEEL PLATES
ANGLE TOP END SCALE 1:20 96 Required (cut at LCL)
Two Required As Above, Two Required as Mirror Image SCALE 1:20
Angle and Plate Details J.Erochko S5
362
APPENDIX C: FRICTION SUBCOMPONENT TEST DRAWINGS
363
UNIVERSITYȱofȱTORONTO
4MN SCED Brace
Friction Interface Test
ÉCOLEȱPOLYTECHNIQUEȱMONTRÉAL
Départmentȱdesȱgéniesȱcivil,ȱgéologiqueȱetȱdesȱmines
Bill of Materials
Appendix C: Friction Subcomponent Test Drawings
Total
Part # Description Page Weight (kg) # Required Material
Weight (kg)
P1 Inner Plate PT1 49 1 49 350W
P2 Outer Plate PT1 32 2 64 350W
P3 Stainless Steel Friction Plate PT1 3 2 6 Stainless Steel
P4 Shim Plate PT1 1 2 2 350W
P5 Grip Plate PT1 31 1 31 350W
TOTAL 152

UNIVERSITYȱofȱTORONTO Project Title REV Date
DepartmentȱofȱCivilȱEngineering 4MN SCED Brace Friction Interface Test 2 2009/07/23
ÉCOLEȱPOLYTECHNIQUEȱMONTRÉAL Drawing Title Designer Drawing No.
Départmentȱdesȱgéniesȱcivil,ȱgéologiqueȱetȱdesȱmines Title and Bill of Materials J.Erochko JE05-BM1
364
Grip
Len
g th 1
60
Outer Assembly
A2 (dwg. JE05-A1)
Inner Assembly
A1 (dwg. JE05-A1)
TO
TAL
LEN
G TH

1 100
M IN,
1 250
M AX
1-1/2" A325 Bolts
with washers
4 Required
All dimensions in mm [inches] unless otherwise noted

Départmentȱdesȱgéniesȱcivil,ȱgéologiqueȱetȱdesȱmines Test Setup Summary J.Erochko JE05-TS1
365
43 40 60 40 43
100
100
63 100 63
50
6
214
240
37 40 60 40 37
100
SHIM PLATE
P4 - SCALE 1:5
94
p .
Ty
40
700
yp.
0T
202
R2
500
63 100 63
162
33 33
350
384
240
33 33
38
38
400
yp.
160

50
0T
R2
160
57 100 57
6
38.1
50
214
50
226
STAINLESS STEEL 160
160 226
226 OUTER PLATE FRICTION PLATE
P2 - SCALE 1:5 P3 - SCALE 1:5 GRIP PLATE
INNER PLATE P5 - SCALE 1:5
P1 - SCALE 1:5 All dimensions in mm [inches] unless otherwise noted
Départmentȱdesȱgéniesȱcivil,ȱgéologiqueȱetȱdesȱmines Parts J.Erochko JE05-PT1
366
6 214 6
62
6
P3
P2 P2
400
384
502
P3 P3 44 44
38 6 6 38
6
P4 P4
Behind
100
6
6 214 6
6
138
6 P4 to P5
Do first
P1 P5
310

300
25
P5
160
P1
OUTER
INNER ASSEMBLY ASSEMBLY
160 50
A1 - SCALE 1:5 A2 - SCALE 1:5
Départmentȱdesȱgéniesȱcivil,ȱgéologiqueȱetȱdesȱmines Assemblies J.Erochko JE05-A1
367
LVDT LVDT

Départmentȱdesȱgéniesȱcivil,ȱgéologiqueȱetȱdesȱmines Instrumentation Summary J.Erochko JE05-IN1
368
APPENDIX D: T-SCED DRAWINGS
369
UNIVERSITY of TORONTO
Department of Civil Engineering
Telescoping Self-Centering
Energy Dissipative (SCED)
ÉCOLE POLYTECHNIQUE MONTRÉAL
Brace Départment des génies civil, géologique et des mines
Appendix D: T-SCED Drawings
Bill of Materials
Total
Weight (kg)
S1 Outer Section PT1 761 1 761 ASTM A500
S2 Intermediate Section PT2 427 1 427 ASTM A500
S3 Inner Section PT3 291 1 291 ASTM A500
AL1 Short Angle A PT4 12 2 24 350W
AL2 Short Angle B PT4 12 2 24 350W
AL3 Long Angle A PT5 22 2 44 350W
AL4 Long Angle B PT5 22 2 44 350W
P1 Angle Stiffener PT6 1.5 16 24 300W
P2 Washer Plate PT6 2 8 16 300W
F1 Friction Pad PT6 0 32 0 Carlisle NF-916 Material
or Equivalent
P3 Inner Section Connection Plate PT7 38 1 38 350W
P4 Intermediate Section Stiffener PT7 2 8 16 300W

P5 Inner Section Friction Plate PT7 10 4 40 300W
P6 Stainless Steel Friction Surface PT7 1 8 8 Stainless Steel
P7 Outer End Plate PT8 44 2 88 300W
P8a Inner End Plate with Slot PT8 50 1 50 300W
P8a Inner End Plate without Slot PT8 50 1 50 300W
P9 Cover Plate PT6 2 300W
P10 Intermediate Cover Plate PT6 2 300W
P11 Stop Plate PT6 2 300W
PP1 Inner Assembly Plastic Spacer PT10 4 UHMW Plastic
PP2 Intermediate Assembly Plastic Spacer PT10 4 UHMW Plastic
TOTAL 1945
UNIVERSITY of TORONTO Project Title REV Date

Department of Civil Engineering Telescoping SCED Brace 4 2011/03/14
ÉCOLE POLYTECHNIQUE MONTRÉAL Drawing Title Designer Drawing No.
Départment des génies civil, géologique et des mines Title and Bill of Materials J.Erochko JE15-BM1
370
39" 23"
R7/1 16"
6" R 7/
Slot both sides of tube (typ.)
PLAN
7/8"
7/8"
198"
ELEVATION
20"
1/2"
12"

SECTION
3D VIEW
OUTER SECTION HSS 20"x12"x1/2" (HSS508x305x12.7)

S1 - SCALE 1:50

Départment des génies civil, géologique et des mines Parts - Outer Section J.Erochko JE15-PT1
371
39" 21"
"
R7/16
"
R7/16
Slot both sides of tube (typ.)
PLAN
7/8"
7/8"
198"
ELEVATION
14"
3/8"
10"
SECTION

3D VIEW
INTERMEDIATE SECTION HSS 14"x10"x3/8" (HSS356x254x9.5)

S2 - SCALE 1:50

Départment des génies civil, géologique et des mines Parts - Intermediate Section J.Erochko JE15-PT2
372
17 1/2"
Cut out Web

PLAN
7 5/32"
198"
ELEVATION
210
14.2
9/16"
8 9/32"
9.1
11/32"
205

8 1/16"
3D VIEW
SECTION
INNER SECTION W8x40 (W200x59)

S3 - SCALE 1:50

Départment des génies civil, géologique et des mines Parts - Inner Section J.Erochko JE15-PT3
373
1/2"
1/8"
1 9/32"
5" 7 5/16" 3 1/8" 12 15/16"

1 3/16"
Extents
of Recess
Recess
Slot and
Recess on Back 7/8"
1/32" btw
Face of Angle
2 15/16"
Hole typ.
6"
3/8" Slotted
Hole
"
R7/16 R1
of Slotted
Recess
Slot and
/4"
1/32" btw
1 1/4" Extents
4 1/8"
2 11/16"
Extents
1 3/16"
2 3/4"
of Recess
1 5/8"
1 3/16"
1/2"
28 3/8"
1/8"
1 1/32"
SHORT ANGLE A L6"x3 1/2"x1/2" (L152x89x12.7)

AL1 - SCALE 1:5 3 1/2"
AL1 Section - SCALE 1:2

AL1 3D Views - SCALE 1:10

Départment des génies civil, géologique et des mines Parts - Short Angles J.Erochko JE15-PT4A
374
For Recess
12 15/16" 3 1/8" 7 5/16" 5"
Detail see
AL1
2 15/16"
Hole typ.
3/8" Slotted
6"
1/2"
4 1/8"
2 11/16"
3 1/2"
1 5/8"
28 3/8"
SHORT ANGLE B L6"x3 1/2"x1/2" (L152x89x12.7)
AL2 - SCALE 1:5

AL2 3D Views - SCALE 1:10

Départment des génies civil, géologique et des mines Parts - Short Angles J.Erochko JE15-PT4B
375
3 1/2" 3" 3" 3" 21 7/16" 3 1/8" 12 15/16"
1 1/8" typ. Recess on Back 7/8"

Face of Angle
2 3/4"
Hole typ.
2 15/16"
Hole typ.
1/2" Slotted
3/8" Slotted
" R1
. /16
/4"
R7
t yp
4 1/8"
1 5/8"
"
/16
2 3/4"
R9
2 11/16"
2 3/4"
1 3/16"
50"
1/2"

For Recess
Detail see
AL1
LONG ANGLE A L6"x3 1/2"x1/2" (L152x89x12.7)
6"
AL3 - SCALE 1:5

1/2"
3 1/2"

Départment des génies civil, géologique et des mines Parts - Long Angles J.Erochko JE15-PT5A
376
12 15/16" 3 1/8" 21 7/16" 3" 3" 3" 3 1/2"
7/8" 1 1/8" typ.
2 3/4"
Hole typ.
2 15/16"
Hole typ.
3/8" Slotted
1/2" Slotted
R1
6"
/4
"
R7/1
1 5/8"
4 1/8"
p.
" ty
/16
R9
2 3/4"
2 11/16"
2 3/4"
1 3/16"
50"
For Recess
Detail see
AL1

6"
LONG ANGLE B L6"x3 1/2"x1/2" (L152x89x12.7)

AL4 - SCALE 1:5
1/2"
3 1/2"

Départment des génies civil, géologique et des mines Parts - Long Angles J.Erochko JE15-PT5B
377
2" 6"
1 11/16" 3 1/8" 1 11/16"
3 1/8"
3 5/8"
1 5/8"
1"
3 1/4"
7/8"
1"
1 5/8"
1"
1/2"
2 1/4" 4 3/4"
3/4"
1/2"
8" 6 1/2"
ANGLE STIFFENER WASHER PLATE

P1 - SCALE 1:2 P2 - SCALE 1:2

2"
2"
1 3/16"
1 1/4"
R1
/4
"
5/8"
2 3/4" 5" 5"
1/2"
17"
3/8"
1/4"
STOP PLATE
FRICTION PAD COVER PLATE INTERMEDIATE P11 - SCALE 1:5
F1 - SCALE 1:2 P9 - SCALE 1:5 COVER PLATE
P10 - SCALE 1:5

Départment des génies civil, géologique et des mines Parts - Angle Assemblies J.Erochko JE15-PT6
378
2 1/4" 3" 3" 3" 32 3/4"
2"
1 1/
8" t
yp.
3"
7"
2"
1"
44"
INNER SECTION CONNECTION PLATE

P3 - SCALE 1:5
3 15/16" 3 15/16"
11 1/8" Slot 3 13/16" 3 13/16"
2 1/2"
14 1/2" 3 1/2" 11 1/8" Slot

2 3/8"
7 1/4"
2"
4 3/4"
4 3/4"

Round all Vertical Edges of P5
2 3/8"
2 1/2"
with grinder Below Dashed Line
2mm
5/8"
18" 19" 18 3/4"
1/2"
INTERMEDIATE SECTION STIFFENER INNER SECTION FRICTION PLATE STAINLESS STEEL FRICTION SURFACE
P4 - SCALE 1:5 P5 - SCALE 1:10 P6 - SCALE 1:10

Départment des génies civil, géologique et des mines Parts - Intermediate and Inner Parts J.Erochko JE15-PT7
379
2" 2"
1 3/4" 13" 1 3/4"
1 5/8" 2 1/2" 2 1/4" 2 1/4" 2 1/2" 1 5/8"
2 3/8"
2 11/16"
/8"
21
3 1/2"
2 3/8"
11 3/4"
10 1/8"
2 3/8"
2 1/8
"
7 1/16" Slotted Hole
3 1/2"
R1
"
10 3/8"
"
/1 6
2 11/16"
R9
2 3/8"
7/8" Chamfer Typical 1 1/8"
4"
4"
20 1/2" 12 3/4"
OUTER END PLATE INNER END PLATE WITH SLOT

P7 - SCALE 1:5 P8a - SCALE 1:5

P7 3D View P8a 3D View
SCALE 1:10 SCALE 1:10

Départment des génies civil, géologique et des mines Parts - End Plates J.Erochko JE15-PT8
380
1 5/8" 2 1/2" 4 1/2" 2 1/2" 1 5/8"
2 11/16"
2 3/8"
10 1/8"
2 3/8"
2 1/8
"
2 11/16"
4"
12 3/4"
INNER END PLATE WITHOUT SLOT

P8b - SCALE 1:5

P8b 3D View
SCALE 1:10

Départment des génies civil, géologique et des mines Parts - End Plates J.Erochko JE15-PT9
381
12"
8"
3/8"
12"
12"
3/8"
INNER ASSEMBLY PLASTIC SPACER

PP1 - SCALE 1:5 INTERMEDIATE ASSEMBLY PLASTIC
SPACER
PP2 - SCALE 1:5

Départment des génies civil, géologique et des mines Parts - Plastic Spacers J.Erochko JE15-PT10
382
S3
P6
8 1/16"
P6
1"
5/8"
P3 P11 P6 P5 P6 P5
5/8"
25/32"
P6 P6 2mm
5" 6" 2mm
8 7/8"
4 3/4"
7 1/4"
11 3/8"
11 3/8"
P11 P5 S3 P5
3/8"
3/8"
3/16"
7/16"
22 25/32"
P3 1/4"
7"
3/16" 3/8" 3/8"
8 9/32"
P5
P11 P6 17 1/2" P5 2mm
26 1/2" 19" P6 19"
2mm
198"
224 1/2"
25/32"
5/8"
P6 P6 INNER SECTION ASSEMBLY P6

A1 - SCALE 1:25
P5 Plastic Pad P5
Top and Bottom
PP1 Plastic Pad
S3 Top and Bottom
P3 Screwed to Steel S3
Beyond Section
1"

11/32"
P5
P6 P6
P5
P11 A1 3D VIEW
P5
P3 SCALE 1:25
yp ical
42" T
P6 P6
8 1/16" P11 P6
P5
A1 SECTION - SCALE 1:10

Départment des génies civil, géologique et des mines Assemblies - Inner Section Assembly J.Erochko JE15-AM1
383
P10 to Stop P4
P4 Friction Plate if Slot
S2
2"
is Machined too long
2"
18"
14"
Typ. All P10
3/8"
2"
2"
P4 P4
5"
21"
S2
P4 P10 P4
1/2" typ.
3 1/4"
3 1/4"
1 3/4"
10"
18" 3/8" 18"
Typ. All P4 P10
3 1/4"
1 3/4"
3 1/2"
3 1/4"
198"
INTERMEDIATE SECTION ASSEMBLY

A2 - SCALE 1:25
2xP4
Behind
P10
PP2 Plastic Pad Top and Bottom
Top and Bottom
3/8"
S2 Screwed to Steel
Section
P4
P4 P4
PP2 Plastic Pad S2
Top and Bottom

Screwed to Steel
Section
3/8"
1/2"
A2 3D VIEW
P4 ical SCALE 1:25
Typ
A2 SECTION - SCALE 1:10 42"
P4

Départment des génies civil, géologique et des mines Assemblies - Interm. Section Assembly J.Erochko JE15-AM2
384
Distance btw. Angles Determined by
Position of Plate P5 During Brace Assembly
(Both Ends) S1 5/16"
See
1/4" To Angle and Outer Section
Note
P1 5/16" Typ. All P1 Stiffeners P1
See
Note
at Left
To Angle and Outer Section 1/4"

AL2 (AL1 Underside) Typ. All P1 Stiffeners
8 1/2"
P9 Top and Bottom AL3 (AL4 Underside)
20"
1/2"
AL4 (AL3 Underside)
8 1/2"
AL1 (AL2 Underside)
P1 P1 5/16"
5/16"
16 1/16" 16 1/16"
2 15/16" 5" 5" 2 15/16"
5"
6"
5"
6"
8 7/8"
2 7/8"
2 7/8"
8 7/8"
AL4 (AL3 Beyond)
S1
P9
P1 AL1 (AL2 Beyond) P1
24"
12"
4 3/8" 24" 24" 26"
1/2" typ.
P9
AL3 (AL4 Beyond)
P1 AL2 (AL1 Beyond) P1
198"
228 3/8"
OUTER SECTION ASSEMBLY
A3 - SCALE 1:25 P1 AL4
See Note Above
AL1/2/3/4 1/2"
AL3
S1
S1 AL2 AL1
P1
P9
P1

12"
24"
AL3
21 1/4"
AL4 P1
P1
A3 3D VIEW
SCALE 1:25
AL1 AL2 P1
AL1/2/3/4
8 1/2" 8"
A3 SECTION - SCALE 1:10

Départment des génies civil, géologique et des mines Assemblies - Outer Section Assembly J.Erochko JE15-AM3
385
APPENDIX E: VERTICAL TEST FRAME DRAWINGS
386
Vertical Testing Braced Frame
(for Telescoping SCED Brace)
Appendix E: Vertical Test Frame Drawings
UNIVERSITY of TORONTO
Department of Civil Engineering

ÉCOLE POLYTECHNIQUE MONTRÉAL
Départment des génies civil, géologique et des mines

Department of Civil Engineering Vertical Testing Braced Frame 4 2010/06/22
Départment des génies civil, géologique et des mines Title Page J.Erochko JE17-BM1
387
Part # Description Page Weight (kg) # Required Total Material
Bill of Materials Weight (kg)
(by Part) BP-P1 Base Pin Bottom Plate PT1 2598 2 5196 300W
BP-P2A Base Pin Holder (with Notch) PT1 201 2 402 300W
BP-P2B Base Pin Holder (without Notch) PT1 199 2 398 300W
BP-P3 Pin PT1 75 2 150 AISI 4340
BP-P3K Pin Key PT1 2 2 4 AISI 1018 Keystock
BP-P4 Base Pin Main Vertical Support PT2 272 4 1088 300W
BP-P5 Base Pin Secondary Vertical Support PT2 51 4 204 300W
BP-P6 Base Pin Spacer Plate PT2 9 2 18 300W
BP-P7 Base Pin Horizontal Stiffener A PT2 13 4 52 300W
BP-P8 Base Pin Horizontal Stiffener B PT2 16 4 64 300W
BP-P9 Base Pin Diagonal Stiffener PT2 4 4 16 300W
WM-S1 Wall Mount W Section PT3 753 2 1506 350W
WM-P1 Wall Mount Wall Plate PT3 1563 2 3126 300W
WM-P2 Wall Mount Large Stiffener PT4 57 16 912 300W
WM-P3 Wall Mount Small Stiffener PT4 10 12 120 300W
WM-P4 Actuator Support Front Plate PT4 4.5 6 27 300W
WM-P5 Actuator Support Side Plate PT4 3.5 12 42 300W
WM-P6 Actuator Support Bottom Plate PT4 5 6 30 300W

BM-P1 Beam Mount Actuator Plate PT5 181 2 362 300W
BM-P2 Beam Mount Beam Plate PT5 250 2 500 300W
BM-P3 Beam Mount Main Stiffnener PT5 196 2 392 300W
BM-P4 Beam Mount Minor Stiffener A PT5 4 8 32 300W
BM-P5 Beam Mount Minor Stiffener B PT5 9 4 36 300W
BM-P6 Beam Mount Main Back Stiffnener PT5 24 2 48 300W
BA-S1 Beam W Section PT6 883 1 883 350W
BA-P1 Beam Mid Doubler Plate PT6 16 2 32 300W
BA-P2 Beam End Doubler Plate PT6 2 4 8 300W
BA-L1 Beam Connection Angle PT6 13 4 52 350W
BA-P3 Lower Gusset Column Plate PT7 127 1 127 300W
BA-P4 Lower Gusset Brace Plate PT7 89 1 89 300W
BA-P5 Upper Gusset Brace Plate PT8 23 2 46 300W
BA-P6 Upper Gusset Beam Connection Pl. PT8 20 2 40 300W

LR-S1 Column Connection Tube PT9 52 2 104 A500
LR-S2 Beam Restraint Tube PT9 22 4 88 A500
LR-S3 Diagonal Brace Tube PT9 48.5 4 194 A500
LR-S4 Column Restraint W-Section PT9 33 4 132 350W
LR-P1 Beam Restraint Plate PT10 19 4 76 300W
LR-P2 Diagonal Brace Lower Plate PT10 28 4 112 300W
LR-P3 Diagonal Brace Upper Plate PT10 10 4 40 300W
LR-P4 Column Restraint Base Shim PT10 85 4 340 300W
LR-P5 Column Restraint Connection Plate PT10 10 4 40 300W
TOTAL 17000

Départment des génies civil, géologique et des mines Bill of Materials J.Erochko JE17-BM2
388
Bill of Materials Type Length Area Weight Material
(kg)
(by Type)
STEEL SECTIONS
W360x33 13.4 ft (4.1m) 135 350W
W360x382 13 ft (4.0m) 1500 350W
W460x128 22.5 ft (6.9m) 883 350W
L203x203x13 4.7 ft (1.4m) 52 350W
HSS356x254x9.5 3.9 ft (1.2m) 104 A500
HSS127x76x9.5 11 ft (3.4m) 88 A500
HSS102x102x6.4 37.3 ft (11.4m) 200 A500
STEEL PLATE
6" Plate 6 sqft (0.6 sqm) 716 300W
2" Plate 292 sqft (27.1 sqm) 10818 300W
1 1/2" Plate 69 sqft (6.4 sqm) 1917 300W

1" Plate 32 sqft (3.0 sqm) 598 300W
3/4" Plate 8 sqft (0.8 sqm) 112 300W
5/8" Plate 5 sqft (0.5 sqm) 54 300W
1/2" Plate 5 sqft (0.4 sqm) 41 300W
1/8" Plate 18 sqft (1.7 sqm) 42 300W
OTHER PARTS
6 1/2" Round Pin 2.9 ft (0.9m) 150 AISI 4340
2" Square Keystock 1 ft (0.3m) 4 AISI 1018
TOTAL 17000

Départment des génies civil, géologique et des mines Bill of Materials J.Erochko JE17-BM3
389
18 3/8" 168 1/8" Center of Actuator Stroke
78"
108"
140"
237"
gth
Len
Tube gth
ace L en
" Br Br ace
61 198 o t al
/2" T
1/2"
250

32"
128" 96"
48" 66 3/4" 288"
418 3/4"
ELEVATION - SCALE 1:50

Départment des génies civil, géologique et des mines Test Setup Summary J.Erochko JE17-TS2
390
72 1/2"
Clearance
7 1/16" Actuator
11 3/32" Beam Width
78"
20" Brace Width

23 15/32" Lateral Support Clearance
80"
180"
66"
97 17/32"

PLAN - SCALE 1:50

Départment des génies civil, géologique et des mines Test Setup Summary J.Erochko JE17-TS3
391
10" 12" 36" 12" 36" 12" 10"
O
ut
er
Ra 10 1/2"
Notch for Pin on One Pin Holder
12" 10"
d .5
1/4 only (out of two)
"
Hole
Tol. + Rad. 3 1/4
0.001 "
"/-0.0
36"
80"
. 00"
t yp
3"
3 3/4"
4 1/2"
10" 12"
/4"
R3
19/32"
1 1/2"
2"
16"
128"
33 1/4"
BASE PIN BOTTOM PLATE

BP-P1 - SCALE 1:50
28"
1-8 Tapped Hole 1 3/8" 1 3/8"
2 1/2" Deep
12"
2 1/2"
2"
7 3/4" 2" 2" 2"
3/4"
6"
6"
6 1/2"
Notch BASE PIN HOLDER

Length
Notch
Width
1 7/16"
BP-P2A (without Notch)
Tol. +0.000"/-0.006"
BP-P2B (with Notch) - SCALE 1:10
1 31/32" 1 7/16"
17 5/16"
R2
1 1/4"
3/3
6"
2
PIN
"
BP-P3 - SCALE 1:10

PIN KEY
BP-P3K - SCALE 1:5

Départment des génies civil, géologique et des mines Base Pin Parts J.Erochko JE17-PT1
392
6" 39" 6" 2"
p.
ty
6"
4" 1 1/2"
1/
1
20"
4" 4"
14"
14"
14"
14"
1 1/2"
1 1/2"
1 1/2"
51" 17 1/4"
1 1/2"
BASE PIN MAIN VERTICAL SUPPORT 1/2" BASE PIN SECONDARY VERTICAL SUPPORT
BP-P4 - SCALE 1:10 BP-P5 - SCALE 1:10
8"
2 1/2"
15/16" typ.
6 1/2"
1 1/2"
15/16" typ.
6 1/2" 1 1/2" 1"

8"
39"
1"
2"

BASE PIN DIAGONAL BASE PIN HORIZONTAL STIFFENER A
STIFFENER PLATE BP-P7 - SCALE 1:10
BP-P9 - SCALE 1:10
7"
2 1/2"
15/16" typ.
5 1/4" 15/16" typ.
51"
1"
BASE PIN SPACER PLATE

BP-P6 - SCALE 1:10 BASE PIN HORIZONTAL STIFFENER B
BP-P8 - SCALE 1:10

Départment des génies civil, géologique et des mines Base Pin Parts J.Erochko JE17-PT2A
393
1 1/2"
22 1/4"
1 1/2"
1 1/2"
2"
67" 2"
1 1/2"

1 1/2" 1 1/2"
BASE PIN SUBASSEMBLY PLATE

BP-P10 - SCALE 1:10

Départment des génies civil, géologique et des mines Base Pin Parts J.Erochko JE17-PT2B
394
15" 48" 15"
typ.
62.5
15"
2 15/32"
typ.
62.5
2 15/32"
62.5 62.5 62.5 62.5
2 15/32" 2 15/32" 2 15/32" 2 15/32"
typ. typ. typ. typ.
48"
78"
11
/2"
typ.
62.5 typ
.
2 15/32"
9 5/8" 11 3/4" 11 3/4" 11 3/4" 11 3/4" 11 3/4" 9 5/8"

typ.
15"
62.5
2 15/32"
11 3/4"
15 31/32"
" typ.
1 3/4
5 7/8" 5 7/8"

2"
78"
1 7/8"
WALL MOUNT WALL PLATE
WM-P1 - SCALE 1:20
1 3/16"
16 3/8"
1 7/8"
78"
WALL MOUNT W-SECTION (W360x382)

WM-S1 - SCALE 1:20

Départment des génies civil, géologique et des mines Acutator Wall Mount Parts J.Erochko JE17-PT3
395
31" 5" 1 1/2"
5" 1 1/2"
3"
3"
6 1/2"
14 3/8"
6 1/2"
12 1/2"
3"
3"
28" 8" 1 1/2" 5" 1 1/2"
1 7/8"
1 1/2"
37 1/2" 1"
6 1/2"
WALL MOUNT LARGE STIFFENER WALL MOUNT SMALL STIFFENER

WM-P2 - SCALE 1:10 WM-P3 - SCALE 1:10
3" 3"
3"
6"

6"
6"
3"
1/
4"
1"
3/4"
3/4"
6" 6" 8 1/2"
ACTUATOR SUPPORT FRONT PLATE ACTUATOR SUPPORT SIDE PLATE ACTUATOR SUPPORT BOTTOM PLATE
WM-P4 - SCALE 1:5 WM-P5 - SCALE 1:5 WM-P6 - SCALE 1:5

Départment des génies civil, géologique et des mines Actuator Wall Mount Parts J.Erochko JE17-PT4
396
4 1/2" 4 1/2"
2"
22 7/8" 18" 4 1/2" 4 1/2" 4 1/2" 4 1/2" 4 1/2"
3 1/8"
3" 9" 9" 9" 3"
2 1/4"
2 3/4"
.
t yp
/4"
13
18"
15 1/2"
11 3/4"
2 3/4"
5 1/2"
15 1/2"
28" 11 3/4" 1
1/
3 1/8"
3 1/8"
2 1/4"
8"
t yp
.
2"
42 7/8"
2"
64 1/2"
BEAM MOUNT ACTUATOR PLATE
BM-P1 - SCALE 1:20 BEAM MOUNT BEAM PLATE

BM-P2 - SCALE 1:20
3" 37 3/8"
6"
3/4"
3/4"
36"
1"
42"
24"

10"
BEAM MOUNT MINOR STIFFENER B
BM-P5 - SCALE 1:10
13"
3/4"
4"
6"
4"
3/4" 4"
4"
1"
6" 20"
1 1/2"
40 3/8"
1 1/2"
BEAM MOUNT MAIN STIFFENER BEAM MOUNT MINOR STIFFENER A BEAM MOUNT MAIN BACK STIFFENER
BM-P3 - SCALE 1:20 BM-P4 - SCALE 1:10 BM-P6 - SCALE 1:20

Départment des génies civil, géologique et des mines Actuator Beam Connection Parts J.Erochko JE17-PT5
397
5 5/16"
Holes on one flange of beam only 11 4" typ.
/8"
typ
2 9/16"
. 3"
11 3/32"
3"
40"
4 1/2" Cut holes and slots in beam and
2 9/16"
4 1/2" doubler plates after welding

4 1/2" 4 1/2"
doubler plates to beam
3" to beginning of slot 4 1/2" 4 1/2" 4 1/2" See Assembly A4 on JE17-AM4 3" to beginning of slot
25/32"
5 1/2"
2 3/4"
5 1/2"
2 3/4"
4 1/2"
4 1/2"
R17/3 .
2" typ
. t yp
/8"
11
18 7/16"
1" Slot typ. 1" slot typ.
4 1/2"
4 1/2"
98 1/2" 9" 9" 9" 15/32"
270 1/2"
4 1/4"
2 11/32" 3" 2 5/8"
BEAM W-SECTION (W460x128) 3 3/4"

2 1/2"
BA-S1 - SCALE 1:30

4 1/2" 4 1/2"
4"
11
/16
4 1/2" 4 1/2" 4 1/2" 4 1/2" 4 1/2" 2 3/4"

" ty
1" slot
p.
4 1/2"
3" 9" 9" 9" 3" 1

3 3/4" 1/
2 1/4"
16
14"
3 1/4"
6"
"t
2 3/4"
yp
.
4 1/2"
4 1/2"
1 1/8
4"
" typ
.
5 1/2"
15 1/2"
15 1/2"
8" 8"
4 1/2"
2 1/2"
R1
2 3/4"
7/
32
2 1/4"
3 1/4"
"t

yp
.
64 1/2" 1/8" 7 1/2"
1/8"
BEAM MID DOUBLER PLATE 1/2"
BEAM END DOUBLER PLATE
1/2"
BA-P1 - SCALE 1:20 BA-P2 - SCALE 1:20
BEAM CONNECTION ANGLE (L203x203x13)

BA-L1 - SCALE 1:10

Départment des génies civil, géologique et des mines Beam Parts J.Erochko JE17-PT6
398
2"
12" 2"
14 3/8" 11 1/16"
3"
.
6"
t yp
/8"
11
7 29
/32"
6"
1
1/
31 1
8"
5
t yp
6"
.
/32"
29 15/32"
42"
36"
6"
18"
6"
3"
3"
3"
3"
5 9/
6"
16"
6 17/32"
/8"

18 5
3"
8" 17 7/16"
1"
16" 25 7/16"
1 1/2"
LOWER GUSSET COLUMN PLATE LOWER GUSSET BRACE PLATE
BA-P3 - SCALE 1:10 BA-P4 - SCALE 1:10

Départment des génies civil, géologique et des mines Gusset Plate Parts J.Erochko JE17-PT7
399
3"
3"
3"
3"
4"
3"
1 1/8" typ.
14 7
/8"
"
13 15/16"
13 15/16"
1/32
39 2
37 1/8"
5 7/32"
42 11/32"
5/8"
UPPER GUSSET BRACE PLATE
BA-P5 - SCALE 1:10
4" 4" 4" 4" 4" 2 5/16"

2 5/16"
4" 4" 4" 4" 4"
2 7/8"
4 7/8"

2"
.
typ
/8"
11
40"
44 5/8"
3/4"
UPPER GUSSET BEAM CONNECTION PLATE
BA-P6 - SCALE 1:10

Départment des génies civil, géologique et des mines Gusset Plate Parts J.Erochko JE17-PT8
400
10"
254
3"
76.2
9.5
3/8"
5"
127
9.5
14"
355.6
3/8"
9.5
3/8"
9.5
33"
3/8"
23 1/2"
BEAM RESTRAINT TUBE (HSS127x76x9.5)
LR-S2 - SCALE 1:10
COLUMN CONNECTION TUBE (HSS356x254x9.5)
LR-S1 - SCALE 1:10
2807.1
110 1/2"
5"
127

4"
101.6
8.5
11/32"
°
43.6 67 2526.5 237
23°
1 23/32" 99 15/32" 9 11/32"
5.8
350
6.4 7/32"
13 25/32"
1/4"
4"
101.6
1016
6.4
1/4"
40"
DIAGONAL BRACE TUBE (HSS102x102x6.4)
LR-S3 - SCALE 1:20 COLUMN RESTRAINT W-SECTION (W360x33)
LR-S4 - SCALE 1:20

Départment des génies civil, géologique et des mines Lateral Restraint Parts J.Erochko JE17-PT9
401
2"
4" 12" 4" 6" 2"
2 1/2"
3" 6" 3"
6"
ty p.
4"
3"
12"
6"
15"
20"
9"
.
ypt
.
8"
t yp
1/
1
8"
1/
1
1"
20"
9"
33"
DIAGONAL BRACE LOWER PLATE 2 1/2"

LR-P2 - SCALE 1:10
10"
3/8"
DIAGONAL BRACE UPPER PLATE
9"
LR-P3 - SCALE 1:10
2"
6" 2"
3"
2"
15"

16"
12"
3/8"
12"
.
BEAM RESTRAINT PLATE t yp
/8 "
LR-P1 - SCALE 1:10 11
2"
22"
2"
1/2"
COLUMN RESTRAINT BASE SHIM

LR-P4 - SCALE 1:10
10"
COLUMN RESTRAINT CONNECTION PLATE

LR-P5 - SCALE 1:10

Départment des génies civil, géologique et des mines Lateral Restraint Parts J.Erochko JE17-PT10
402
BP-P1
BP-P7
BP-P4 BP-P2A
BP-P2B BP-P2A
BP-P5 BP-P7
2 5/8"
BP-P7
BP-P10 BP-P1
BP-P6 P2 to P1 BP-P8 BP-P4
11 1/8"
typ.
8 5/8"
CP 1 1/4" 7 3/4"
BP-P9 BP-P9
8 5/8"
11 1/8"
5/8"
BP-P10
typ.
2 5/8"
BP-P4
BP-P5
BP-P4 BP-P2B
BP-P7
5 1/4"
tol -0"/+3/64" A- Make Continuous around top
7/8""
Corner with B. Groove in
BP-P2B Plates BP-P2A and BP-P2B
(BP-P2A beyond) BP-P2B
B - Make Continuous
around Corners with A 1 1/4" BP-P2A
BP-P4

BP-P7 BP-P4 BP-P7
1"
BP-P5 beyond
5/8" BP-P5
typ. 5/8"
BP-P8 typ.
32"
BP-P5 beyond BP-P8

20 3/4"
BP-P1 1 1/2" CP BP-P1

2"
1 1/2"
2"
1 1/2"
9 1/4"
BP-P10
BP-P9 3 7/8" 1" BP-P9

BP-P9 BP-P6 beyond
25 1/2" BP-P10
33 1/2"
BASE PIN HOLDER ASSEMBLY (2 Required)
64"
A1 - SCALE 1:25

Départment des génies civil, géologique et des mines Base Pin Holder Assembly J.Erochko JE17-AM1
403
BP-P3 BP-P3
6 1/2"
1 7/16"
7 23/32"
3 1/4"
3 3/4"
4 15/32"
BP-P3K
TACK
"
32
3/
6" BP-P3K
R2
1 7/32"
BP-P3K
BP-P3
BP-P3K
BP-P3

17 5/16"
BASE PIN ASSEMBLY (2 Required)

A1a - SCALE 1:5

Départment des génies civil, géologique et des mines Base Pin Assembly J.Erochko JE17-AM1a
404
WM-P1 WM-P2 (typ.)
1/2"
typ.
A2a (typ.)
1/2" typ.
WM-P2 (typ.)
A2a (typ.)
WM-P1 WM-P3 (typ.)
WM-S1 WM-P3 (typ.)
WM-S1
3/4"
typ.
3/4" WM-P3 (typ.) WM-P3 beyond

typ.
WM-P2 (typ.) WM-S1 A2a (typ.)
1/2" WM-P2
typ. WM-S1

WM-P1
3/4"
3/4"
16 3/8"
18 3/8"
2"
8"
39"
3 3/4" 11 3/4" 11 3/4" 11 3/4" 11 3/4" 11 3/4" 11 3/4" 3 3/4"
WM-P1
78"
ACTUATOR WALL MOUNT ASSEMBLY (2 Required)

A2 - SCALE 1:25

Départment des génies civil, géologique et des mines Actuator Wall Mount Assembly J.Erochko JE17-AM2
405
WM-P5
3/4" 6" 3/4" WM-P4
1/2" WM-P6 1/2"
WM-P5 WM-P5
1/2" WM-P5
4 1/4"
3/4"
3/4"
6"
WM-P4
1/2"
3/4"
1"
3/4"
3/4"
3/4"
WM-P6
WM-P5 WM-P4 Beyond

WM-P4 WM-P5
WM-P5
6"

6 3/4"
WM-P6
WM-P6
3/4"
8 1/2"
ACTUATOR SUPPORT SUB-ASSEMBLY (6 Required)

A2a - SCALE 1:5

Départment des génies civil, géologique et des mines Actuator Support Sub-Assembly J.Erochko JE17-AM2a
406
BM-P1 BM-P6
BM-P2 BM-P3
BM-P5
BM-P1
BM-P4
BM-P4
BM-P2
BM-P6
BM-P4 BM-P5 BM-P4
BM-P3
BM-P1
1/2" 18"
typ.
BM-P3
BM-P4
9"
3/4" BM-P5 BM-P1
BM-P3
BM-P6
1/2" BM-P5
typ.
BM-P5
44 7/8"
BM-P4
1/2"
33 7/8"
typ.
3/4" BM-P4

1" 3/4" BM-P4
BM-P2
BM-P2
2"
16 1/2" 18" 13 1/2" 16 1/2"
15 1/2"
64 1/2"
ACTUATOR BEAM CONNECTION ASSEMBLY (2 Required)

A3 - SCALE 1:20

Départment des génies civil, géologique et des mines Actuator Beam Connection Assembly J.Erochko JE17-AM3
407
BA-S1
BA-S1 110 1/2"

7 1/2"
95 1/2" 64 1/2" BA-P2 both sides of web
1/4"
typ. typ.
1/8" 1/8"
1/4"
7 1/2" BA-P2 both sides of web BA-P1 both sides of web 3"
typ.
1/8"
BA-S1
270 1/2"
Plans and Elevations - SCALE 1:30
BA-S1
BA-S1 and S2 BA-S1

BA-P1 both sides of web

doubler plates
on either side
18 7/16"
of beam web BA-P2 both sides of web
BA-P2 both sides of web
Section - SCALE 1:20

3D View - SCALE 1:50
BEAM ASSEMBLY (1 Required)

A4 - SCALE VARIES

Départment des génies civil, géologique et des mines Beam Assembly J.Erochko JE17-AM4
408
BA-P3
BA-P4
3/4"
BA-P3
BA-P4
BA-P3
BA-P4
BA-P4
42"

BA-P3
6"
16"
26 15/16"
LOWER GUSSET ASSEMBLY (1 Required)

A5 - SCALE 1:10

Départment des génies civil, géologique et des mines Lower Gusset Assembly J.Erochko JE17-AM5
409
BA-P5 below BA-P6
BA-P6
4 7/8"
44 5/8"
BA-P6
3/4"
9/32"
3/8"
BA-P5
14 11/16"
13 15/16"
BA-P5
42 11/32" 2"
3/8"
5/8"
3 7/8"
UPPER GUSSET ASSEMBLY A (1 Required)
A5a - SCALE 1:20
BA-P6
BA-P5 below
BA-P6
4 7/8"
44 5/8"

BA-P6
3/4"
9/32" 3/8"
14 11/16"
13 15/16"
BA-P5
42 11/32" 2"
3/8"
5/8" BA-P5
UPPER GUSSET ASSEMBLY B (1 Required) 3 7/8"
A5b - SCALE 1:20

Départment des génies civil, géologique et des mines Upper Gusset Assemblies J.Erochko JE17-AM5a
410
LR-P1
1/4" LR-S2
1/4"
LR-S2 LR-S2
LR-P1
LR-S2
LR-P1
LR-P1
33"

5 3/8" 12"
LATERAL RESTRAINT BEAM RESTRAINT ASSEMBLY (4 Required)

A6 - SCALE 1:10

Départment des génies civil, géologique et des mines Lateral Restraint Beam Restraint J.Erochko JE17-AM6
411
LR-P2
LR-S3 LR-P3
6"
5 /1
41
2"
5 /3
10
31
LR-P2 LR-S3 LR-P3
41
3/1
6"
1/3
4"
6" 3/8"
2"
5 /1
41
3/8"
6"
10"
10"
31
6"
LR-P3
3 /1
Plans and Elevations - SCALE 1:20
6"
3/8
"
1"
47
3/4
"
"
1/2
107
LR-S3

LR-P2
3D View - SCALE 1:50
LATERAL RESTRAINT DIAGONAL BRACE ASSEMBLY (4 Required)

A7 - SCALE VARIES

Départment des génies civil, géologique et des mines Diagonal Brace Assembly J.Erochko JE17-AM7
412
LR-P5
LR-S4
LR-S4
LR-P5
LR-S4
LR-S4

1/8" 1/8"
1/4" 1/4"
LR-P5 LR-P5
COLUMN RESTRAINT ASSEMBLY (4 Required)

A8 - SCALE 1:10

Départment des génies civil, géologique et des mines Column Restraint Assembly J.Erochko JE17-AM8
413
APPENDIX F: SHAKE TABLE TEST PROTOTYPE DESIGN REPORT
EXCERPT (KIM, 2009A)
414
DESIGN OF SHAKE TABLE TEST SPECIMEN
1. DESIGN OF PROTOTYPE BUILDING

Besides being considered as representative of typical low-rise steel
structures in the North America, three story structures are the simplest
structures that represent typical multi-story buildings consisting of a
ground floor, middle floors, and a top floor (Kusumastuti et al. 2005).
A three-story steel structure considered as the prototype building was
designed according to ASCE 7-05 Minimum Design Loads for
Buildings and Other Structures (ASCE/SEI 2005). The design of steel
members in this building followed AISC 360-05 Load and Resistance
Factor Design Specification for Structural Steel Buildings (AISC
2005a), and AISC 341-05 Seismic Provisions for Structural Steel
Buildings (AISC 2005b).
1.1. Preliminary Design

1.1.1. Description of Prototype Building
Figure 1-1 shows the building layout and dimensions of the building.
The building was assumed to be located on stiff soil (Site Class D as
per ASCE 7-05). In this report, one of exterior braced frames in the
North-South direction which two diagonal self-centering energy
dissipative (SCED) braces are placed for each story was studied.
1.1.2. Seismic Analysis Parameters

Design Response Spectrum Figure 1-1. Building layout of prototype building: (a) Plan view
Appendix F: Shake Table Test Prototype Design Report Excerpt (Kim, 2009a)
and, (b) Elevation

From the map given in ASCE 7-05, the mapped acceleration
parameters were determined to be SS = 2.1 and S1 = 0.8. Considering
the site coefficient Fa = 1.0 and Fv = 1.5 for Site Class D, the
maximum considered earthquake spectral response acceleration
parameters were calculated to SMS = 2.1 for short periods and SM1 =
1.2 for period of 1s. The design acceleration parameters are than
simply equal to 2/3 of the maximum considered earthquake spectral
response acceleration parameters, giving: SDS = 1.4 for short periods
and SD1 = 0.8 for period of 1s. The periods required for the design
response spectrum were also determined to T0 = 0.114 sec, TS = 0.571
sec, and TL = 8.0 sec. The design response spectrum obtained from the
parameters above is shown in Figure 1-2. Figure 1-2. Design Response Spectrum
Page 1
415
Occupancy Category, Importance Factor and Seismic Design In this report, the dead load on each story only is considered as the
Category effective seismic weight. The seismic masses on the roof and floor
level of the entire structure are, respectively, 0.416 kN·s2/mm (4084
Since the prototype building is assumed to be an office building, the
kN) and 0.441 kN·s2/mm (4324 kN). Considering tributary areas on
occupancy category is II according to ASCE 7-05, which results in the
the columns and beams, gravity loads in kN are illustrated in Figure
importance factor, I = 1.0. This building is assigned to Seismic Design
1-3.
Category E because S1 = 0.8 is larger than 0.75 according to ASCE 7-
05. 1.1.3. Design of SCED braced Frames
Seismic Response Parameters An initial design of members comprised of a seismic-force-resisting
system (SFRS) in north-south direction of the prototype building
The ASCE 7-05 does not yet specifically define the seismic response
follows the equivalent lateral force procedure according to ASCE 7-
parameters such as R, Cd, :0 for SCED braced frames. However,
05.
several previous researches presented that the seismic response of
SCED braced frames is comparable to that of buckling restrained Seismic Base Shear and Vertical Force Distribution
braced frames (BRB frames) in terms of story drifts and floor In order to determine seismic loads of the prototype building, the
accelerations. Based on this, the seismic design parameters of R = 7.0, fundamental period Ta has to be defined. The approximate
Cd = 5.5, and :0 = 2.0 are given for SCED braced frames and R = 8.0, fundamental period can be estimated using an equation given in
Cd = 5.5, and :0 = 3.0 for special moment-resisting frames. ASCE 7-05:
Gravity Load Description and Seismic Weight
The dead loading that were assumed at typical floor levels and the
roof level is given in Table 1-1. For office buildings the minimum
uniformly distributed live load on a typical floor is 2.4 kPa and 0.96
kPa on the flat roof. ASCE 7-05 allows live load reduction for the

design of columns and beams and the reduced live loads are also
presented in Table 1-1.
Table 1-1 Gravity Loading for Prototype Building

Reduced Live Load
Dead Load (kPa)
Floor
(kPa) For
For beam
column
Roof
4.07 0.76 0.96
Level
Floor Figure 1-3 Applied Gravity Loads
4.31 0.96 2.40
Level
Page 2
416
ܶܽ = ‫ = ݔ݄݊ ݐܥ‬0.0731 × 11.8860.75 = 0.47 sec Equation 1.1 Table 1-2 Lateral Seismic Forces per Studied One-Bay Frame
The upper limit on the period Ta is given by: Floor wi (kN) hi (m) Fxi (kN) Vxi (kN)
ܶ‫ = ݑ‬1.4ܶܽ = 0.66 sec Equation 1.2 Roof 1021 11.886 309 309
Floor 3 1081 7.924 218 527
The seismic response coefficient Cs for this building is calculated by:
ܵ‫ܵܦ‬ 1.4 Floor 2 1081 3.962 109 637
‫= ݏܥ‬ = = 0.2 Equation 1.3
ܴ/‫ܫ‬ 7/1
Sum 3183 637
Note that the calculated seismic response coefficient Cs is larger than The considered load combinations are as follow:
the lower limit, 0.01 and
(1.2 + 0.2ܵ‫ ܦ) ܵܦ‬+ ߩܳ‫ ܧ‬+ 0.5‫ܮ‬ Equation
0.5ܵ1 0.5×0.8
‫= ݏܥ‬ = = 0.06 Equation 1.4 1.9
ܴ/‫ܫ‬ 7/1
And
and is less than the upper limit: (0.9 െ 0.2ܵ‫ ܦ) ܵܦ‬+ ߩܳ‫ܧ‬ Equation 1.10
ܵ1 0.8
‫ܴܦ ܶ = ݏܥ‬/‫ܫ‬ = = 0.24 Equation 1.5 where U is equal to 1. Also the following two combinations related to
ܽ 0.47×7/1
the overstrength factor shall be considered.
The seismic base shear for the entire building is obtained by: (1.2 + 0.2ܵ‫ ܦ) ܵܦ‬+ π0 ܳ‫ ܧ‬+ 0.5‫ܮ‬ Equation 1.11
ܸ = ‫ = ܹ ݏܥ‬0.2 × 12732 = 2546 kN Equation 1.6 And
The seismic base shear of one bay SCED braced frame of interest is (0.9 െ 0.2ܵ‫ ܦ) ܵܦ‬+ π0 ܳ‫ܧ‬ Equation 1.12

2546/4 = 637 kN. The base shear is distributed along the building Equations 1.9 and 1.11 govern the design of columns, the
height using the equation: compression loads applied to the columns are summarized in Table
‫ܸ ݔݒܥ = ݔܨ‬ Equation 1.7 1-3.
where: Table 1-3 Applied Compression for Designing Columns

‫݇ݔ ݄ ݔ ݓ‬ Columns D.L. L.L. QE PU PUCD PUOV
‫= ݔݒܥ‬ ݇ Equation 1.8 3rd
σܰ
݅=1 ‫݅ ݄ ݅ ݓ‬ 53 10 0 248 331 83
Story
and the value of 1.0 is assigned to the factor k for structures having a Ext. 2nd
109 22 389 674 846 950
period less than 0.5 sec. Using these equation above the lateral Col. Story
1st
seismic force for each story in one bay braced frame was calculated 165 35 389 1208 1470 1040
Story
and summarized in Table 1-2. Figure 1-3 illustrates the applied the Int. 3rd
seismic loads in unit kN obtained from the ELF method which will be 117 22 144 248 432 472
Col. Story
used for preliminary member sizing.
Page 3
417
2nd Using the structural member sizes and SCED braces, the dynamic
241 50 144 674 1055 670
Story characteristics of the studied one-bay SCED braced frame are
1st evaluated and presented in Table 1-7.
265 77 679 1208 1639 1789
Story
In this table, PU is the applied compression in activating external fuses Table 1-5 Components of SCED Braces
in the SCED braces. The column compression determined by SCED Braces
Level
Equation 1.9 with the consideration of the activation of external fuses Int. Tube Ext. Tube Tendon Friction Fuse
is denoted as PUCD while the values in the next column, PUOV are 3rd CHS178x HSS203x203 4- 8-
Story 9.5 x6.4 3xI17mm LDFIs MDFIs
obtained from Equation 1.11 with the overstrength factor :o=2.0. The
governing compression forces in designing the exterior and interior 2nd CHS219x HSS254x254 3xI17mm 7- 14-
columns are 1040 and 1639 kN. Based on this, the members of Story 9.5 x6.4 1xI22mm LDFIs MDFIs
W200x71 (Cr = 1680 kN) and W200x52 (Cr = 1210 kN) were selected 1st CHS219x HSS254x254 2xI17mm 9- 17-
for the interior and exterior columns, respectively. The members of Story 13 x8.0 2xI22mm LDFIs MDFIs
W200x52 and W200x59 were chosen for the roof beam and the other LDFI: Double NF916 and stainless steel friction interface
floor beams, respectively. MDFI: Double NF780 and stainless steel friction interface
Table 1-4 Section Properties of Members in SCED Braced Frames Table 1-6 Structural Properties of SCED Braces
Structural Properties of SCED Braces Structural Properties of SCED Braces
Level d b tf tw A Ix / I y Level ki
TO (kN) FFR (kN)
(kN/mm) D FY (kN) FU (kN)
(mm) (mm) (mm) (mm) (mm2) (106 x mm4)
Ext. 3rd Story 184 184 128 0.024 373 589
W200x52 206 204 12.6 7.9 6650 52.7/17.8
Col. 2nd Story 322 322 188 0.028 654 1011
Int.
W200x71 216 206 17.4 10.2 9100 76.6/25.4 1st Story 414 414 235 0.026 841 1269

Col.
TO: Initial tensioning force of tendons FFR: Friction force
W200x59
210 205 14.2 9.1 7560 61.1/20.4 ki: Initial stiffness D: Post-yielding stiffness ratio
(2nd&3rd F)
Beams FY: Force at the initiation of gap-opening
W200x52 FU: Force at the activation of fuse elements
206 204 12.6 7.9 6650 52.7/17.8
(Roof)
1.1.4. Design of Moment-Resisting Frames
d: Depth of member, b: Width of member
tf: Thickness of flange, tw: Thickness of web, A: Section area An initial design of members of special moment-resisting
Ix: moment of inertia respect to strong axis, Iy: moment of inertia respect to weak axis frames(SMRFs) of the prototype building follows the equivalent
Under the assumption that most of seismic loads is resisted by SCED lateral force procedure according to ASCE 7-05.
braces, the components comprising SCED braces were designed and Seismic Base Shear and Vertical Force Distribution
are summarized in Table 1-5. The structural characteristics of SCED
braces installed in each level are presented in Table 1-6. In order to determine seismic loads of the prototype building, the
fundamental period Ta has to be defined. The approximate
fundamental period can be estimated with Ct = 0.0724 and x=0.8:
Page 4
418
Table 1-7 Dynamic Characteristics of SCED Braced Frame where:

Modes Mode 1 Mode 2 ‫݇ݔ ݄ ݔ ݓ‬
0.421 0.153 ‫= ݔݒܥ‬ ݇ Equation 1.20
Tm, sec σܰ
݅=1 ‫݅ ݄ ݅ ݓ‬
Floor wi, kN {Ii1} wiIi1 wiIi12 {Ii2} wiIi2 wiIi22
Roof 1021 1.000 1021 1021 1.000 1021 1021 and the value of 1.013 is assigned to the factor k for structures having
3F 1081 0.649 702 456 -0.975 -1054 1028 a period of 0.525 sec. Using these equation above the lateral seismic
2F 1081 0.287 310 89 -1.075 -1162 1250 force for each story in one bay braced frame was calculated and
Sum 3183 2033 1566 -1196 3299 summarized in Table 1-2. Figure 1-3 illustrates the applied the seismic
Wm, kN 2640 (83%) 433 (14%)
loads in unit kN obtained from the ELF method which will be used for
preliminary member sizing.
*m 1.298 -0.362
ܶܽ = ‫ = ݔ݄݊ ݐܥ‬0.0724 × 11.8860.8 = 0.525 sec Equation 1.13 Table 1-8 Lateral Seismic Forces per Moment-Resisting Frame
Floor wi (kN) hi (m) Fxi (kN) Vxi (kN)

The upper limit on the period Ta is given by:
ܶ‫ = ݑ‬1.4ܶܽ = 0.734 sec Equation 1.14 Roof 2042 11.886 543 543
Floor 3 2161 7.924 381 925
The seismic response coefficient Cs for this building is calculated by:
Floor 2 2161 3.962 189 1114
ܵ‫ܵܦ‬ 1.4
‫= ݏܥ‬
ܴ/‫ܫ‬
= = 0.175 Equation 1.15
8/1 Sum 6364 1114
Note that the calculated seismic response coefficient Cs is larger than The considered load combinations are as follow:
the lower limit, 0.01 and (1.2 + 0.2ܵ‫ ܦ) ܵܦ‬+ ߩܳ‫ ܧ‬+ ‫ܮ‬ Equation 1.21

0.5ܵ1 0.5×0.8
‫= ݏܥ‬ = = 0.05 Equation 1.16 And
ܴ/‫ܫ‬ 8/1
(0.9 െ 0.2ܵ‫ ܦ) ܵܦ‬+ ߩܳ‫ܧ‬ Equation 1.22

and is less than the upper limit:
where U is equal to 1. Based on the load cases and load combinations
ܵ‫ ܦ‬1 0.8
‫= ݏܥ‬ = = 0.19 Equation 1.17 above, the members of W460x193 was selected for the columns and
ܶܽ ܴ/‫ܫ‬ 0.525×8/1
the members of W610x125, W610x101, and W530x74 were,
The seismic base shear for the entire building is obtained by: respectively, chosen for the second floor beams, the third floor beams,
and the roof floor beams to satisfy the stiffness requirement of 2%
ܸ = ‫ = ܹ ݏܥ‬0.175 × 12732 = 2228 kN Equation 1.18 story drift limitation of ASCE 7-05(See Table 1-10). The section
properties of the columns and beams for SMRFs are presented in
The seismic base shear of one of SMRFs is 2228/2 = 1114 kN. The Table 1-9 and Figure 1-4.
base shear is distributed along the building height using the equation:
‫ܸ ݔݒܥ = ݔܨ‬ Equation 1.19
Page 5
419
Table 1-9 Section Properties of Members in SMRFs 1.1.5. Design of Gravity Load-Carrying Systems
Structural Properties of SCED Braces The columns in gravity load-carrying systems were sized by the load
Level d b tf tw A Ix / I y combination:
(mm) (mm) (mm) (mm) (mm2) (106 x mm4)
Columns, W460x193 489 283 30.5 17.0 24600 1020/115 1.2‫ ܦ‬+ 1.6‫ܮ‬ Equation 1.23
W610x125 The compression of a 1st story column due to dead loads is 731 kN
612 229 19.6 11.9 15900 985/39.3
(2nd F) and that due to reduced live load is 154 kN. The factored compression
W610x101 is 1136 kN. To carry the factored compression, W200x52 (A=6650
Beams 603 228 14.9 10.5 13000 764/29.5
(3rd F)
mm2, Cr=1210 kN from pp. 4-40 in CAN/CSA S16-01) was chosen as
W530x74
529 166 13.6 9.7 9520 411/10.4 the column members.
(Roof)
The factored uniformly distributed load, ߱ of the north-south direction
beams in 2nd and 3rd floors was calculated by:
߱ = (1.2 × 4.31 + 1.6 × 2.4) × 6.096Τ2 = 27.5 kNΤm
Equation 1.24
The corresponding factored moment is 251 kN·m such that W410x46
(Z=885,000 mm3, Mr=275 kN·m, ߜ = 10.5݉݉ = ‫ܮ‬/810) was
selected for these beam members. Similarly, the factored moment of
the roof beams in the same direction is 178 kN·m such that W360x39
(Z=662,000 mm3, Mr=206 kN·m, ߜ = 6.5݉݉ = ‫ܮ‬/1320)
Concentrated loads are applied to the center of beams system in the
east-west direction of the gravity load-carrying. The applied

Figure 1-4 Special Moment Resisting Frame concentrated factored-loads and factored moment to such beams in 2nd
and 3rd floors were calculated by:
Table 1-10 Story Drift of Moment-Resisting Frame ܲ = 27.5 × 8.534Τ2 = 117.3 kN Equation 1.25
‫ = ݎܯ‬117.3 × 6.096Τ4 = 179 kN ȉ m Equation 1.26
Floor ߜ݁ (mm) ο݁ (mm) ο݁ (%) ο (%)
In order to transfer the factored moment, W460x52 was chosen for
Roof 36.5 11.0 0.28 1.53 such beams. In the similar way, W410x39 was selected for the beams
Floor 3 25.5 14.2 0.36 1.98 in roof floors.
Floor 2 11.3 11.3 0.29 1.57 Table 1-11 summarizes section properties of members comprised of
the gravity load-carrying systems.
Sum 6364 1114
Ge: Elastic Floor Displacement, 'e: Elastic Story Drift
': Expected Story Drift
Page 6
420
Table 1-11 Members in Gravity Load-Carrying Systems Analysis Time Step 0.0005 second
Structural Properties of SCED Braces
Output Time Step 0.005 second
d b A Ix / I y
Member
(mm) (mm) (mm2) (106 x mm4) Floor Diaphragm “Master and Slave” node concept
Columns W200x52 206 204 6650 52.7/17.8 Steel Beam-Column elements with RAMBERG-
Beams 2nd & 3rd F W410x46 403 140 5890 156/5.14 Beam Modeling OSGOOD hysteresis Model (J = 10.0) which is similar
N-S dir. to elasto-plastic hysteresis
Roof W360x39 353 128 4980 102/3.75 Expected Strength (Fy=1.1*345MPa)
Beams 2nd & 3rd F W460x52 450 152 6630 212/6.34 Steel Beam-Column elements with RAMBERG-
E-W dir. Roof W410x39 399 140 4990 127/4.04 Column Modeling OSGOOD hysteresis Model (J = 10.0) which is similar
to elasto-plastic hysteresis
Expected Strength (Fy=1.1*345MPa)
1.2. Time History Analysis SCED Brace Flag-Shaped hysteresis spring + Elasto-perfect Plastic
1.2.1. Modeling Modeling spring + two Big “I” and Small “A” springs
An analytical model was established for nonlinear time history 1.2.2. Ground Motion Records
analysis using RUAUMOKO which is one of commercial nonlinear 20 strong ground motion records which were developed for the SAC
dynamic structural analysis software (Carr 2005). The unique flag- Phase II Project were used for the time-history analyses of the
shaped hysteresis of SCED braces was modeled using built-in prototype building (Somerville et al. 1997). The selected ground
hysteresis rule in RUAUMOKO and elasto-perfect plastic behavior motions are summarized in Figure 1-5. Also, mean spectrum obtained
spring elements were used for modeling fuse elements which were from the 20 ground motions is presented in this figure for the
directly connected with SCED braces in order to ductile cyclic comparison with the design based earthquake (DBE) level response
behavior after the self-centering limit and to prevent the abrupt failure spectrum. This shows that the mean spectrum agrees well with the
of tendons in an SCED brace. Also in order to avoid unlimited design spectrum for DBE although small differences are observed in

rotation at the connection between an SCED brace element and a fuse the acceleration sensitive region.
element, elastic springs with significantly large rotational stiffness and
negligibly small axial stiffness were added into the analysis model.
Table 1-12 summarizes some important assumptions used for the
analytical model.
Table 1-12 Description of Analytical Model

Item Description
Algorithm Newmark Constant Average Acceleration method

Constant Damping Model with 3 % damping ratio for
Damping Model
all modes
Second Order Large Displacement Effects
Effect P-D Effects using “Leaning Column” concept Figure 1-5 Elastic Response Spectrum of Selected Ground Motion
Page 7
421
APPENDIX G: SHAKE TABLE SCED DRAWINGS (KIM, 2009A)
422
PROJECT TITLE
SCED BRACED
FRAME SHAKE-
TABLE TEST @ UB
NOTE
Any changes are NOT allowed
without a permission of project
engineers.
All dimensions in these drawings

are in inch for lengths and degree
for angles, otherwise mentioned.
Ecole Polytechnique and
University of Toronto, Canada
have all rights regarding exclusive
world-wide patent of Self-
Centering Energy Dissipation
(SCED) dampers.
The design and manufacturing of
SCED dampers in the set of
drawings are applicable to only the
prototype test specimens.
This prototype testing program is

as a part of the Idea to Innovation
(I2I) program financially
supported by the Natural Science
and Engineering Research Council
of Canada.
Appendix G: Shake Table SCED Drawings (Kim, 2009a)

REV. DATA DESCRIPTION OF REASON NAME
CLIENT
-
DRAWING TITLE
SCED DAMPER SHAKE-
TABLE TESTS
3D VIEW DRAWING
DATA ISSUED APPROVED

MARCH, 17, 2009
PROJECT CODE CHECKED
-
SCALE DESIGNED
SHEET NO. DRAWN
SCED BRACE SPECIMEN TESTED AT EPM DRAWING NO. REV.
EPM-SCED-01 2
423
PROJECT TITLE
2566 SCED BRACED

FRAME SHAKE-
224 2222 120 TABLE TEST @ UB
24 32
B Tendon C
32
32
A
18
18
230
234
NOTE
A Connector engineers.
B C All dimensions in these drawings
160 2254 120
21, Typ. have all rights regarding exclusive
(SCED) dampers.
203
146
146
203

21, Typ. SCED dampers in the set of

DRAWINGS FOR SCED BRACE (I2I) program financially
of Canada.
203
93 18 93
6 6 6
6
49
6 6
6 6
Ext. Tube
100

Int. Tube
18
18
CLIENT
-
DRAWING TITLE
SCED DAMPER SHAKE-
TABLE TESTS
SCED BRACE DRAWINGS
100
MARCH, 17, 2009
6 6
6 PROJECT CODE CHECKED
49
6 -
6 SCALE DESIGNED
6 6
6
SHEET NO. DRAWN
DRAWING NO. REV.
SECTION A-A SECTION B-B SECTION C-C EPM-SCED-02 2

424
PROJECT TITLE
SCED BRACED
ET FRAME SHAKE-
EFA 2 TABLE TEST @ UB
EFA 1 IT (Inside of ET)
NOTE
IFP 1 IFP 2 engineers.

2022 Ecole Polytechnique and

125 1717 180 have all rights regarding exclusive
(SCED) dampers.
16
16
100
HSS 100X100X4.0

ET(EXTERIOR TUBE) as a part of the Idea to Innovation
of Canada.
2022
96 1830 96
75
HSS 75X75X4.5
12
IT(INTERIOR TUBE) NF-916 Pad 125

(35X15)
7

33
Stainless sheet, both faces 14

65
65
415 (87X64, t=2mm)

32
48 128 170 48 22 91
22 48 22
6
30
1.7mm Recess,Typ.
3
2 R7
35
57
35
2 2 CLIENT
87
2 2
87
EFA 1(Ext. Friction Angle 1, 4pcs.)
14
52
R1 Cut from L-65X65X6 -
52
1
50
DRAWING TITLE
45
87
NF-916 Pad 266 135 SCED DAMPER SHAKE-
20
2 (35X15) TABLE TESTS
2
Stainless sheet,both faces Stainless sheet,both faces SCED BRACE DRAWINGS
(87X64, t=2mm)
7
(220X96, t=2mm)
33
33

14
65
MARCH, 17, 2009

6
32
32
IFP 1(Int. Friction Plate 1, 2pcs.) IFP 2(Int. Friction Plate 2, 2pcs.) PROJECT CODE CHECKED
Thickness : 10 mm excluding thickness of stainless sheets Thickness : 10 mm excluding thickness of -
which are welded on the both faces stainless sheets which are welded on the both faces 30 1.7mm Recess,Typ. SCALE DESIGNED
EFA 2(Ext. Friction Angle 2, 4pcs.) SHEET NO. DRAWN
Cut from L-65X65X6 DRAWING NO. REV.
EPM-SCED-03 2
425
PROJECT TITLE
SCED BRACED
FA FRAME SHAKE-
BP TABLE TEST @ UB
NOTE
engineers.
GCA
have all rights regarding exclusive
100 world-wide patent of Self-
NF-916 Pad 40 80 40 (SCED) dampers.
4
(70X30) 10 The design and manufacturing of
54
5
104
22 This prototype testing program is
30
100
7
BP(Blocking Plate, 2pcs.) and Engineering Research Council
of Canada.
Thickness: 30mm
7
93
1.7mm Recess,Typ.
51
L-100X100X7 21
Length:120mm
29
57
100
7
100
51

GCA (Gusset plate 93
Connecting Angle, 4pcs.) FA(Fuse Angle, 4pcs.)

Cut from L-100X100X7
CLIENT
-
DRAWING TITLE
SCED DAMPER SHAKE-
TABLE TESTS
SCED BRACE DRAWINGS

MARCH, 17, 2009
-
SCALE DESIGNED
SHEET NO. DRAWN
DRAWING NO. REV.
EPM-SCED-04 2
426
PROJECT TITLE
SCED BRACED
ASSEMBLING PROCEDURE 6 FRAME SHAKE-
TABLE TEST @ UB
6
6 Step 1: Welding between IT and IFP 2
NOTE
engineers.

6
Step 2: Inserting IT into ET and welding between IT and IFP 1 have all rights regarding exclusive
(SCED) dampers.
Stressing SCED dampers in the set of
Side prototype test specimens.

Step 3: Placing tendon and BPs on the both end of ET an IT and Stressing the tendon and Engineering Research Council
of Canada.
6 6
6 6

Step 4: Placing EFA 1 and 2s, torquing friction bolts, and welding EFA 1 and 2 on the ET
6
6
CLIENT
6 -
DRAWING TITLE
6 SCED DAMPER SHAKE-
Step 5: Placing GCAs, and welding between GCA and EFA 2 TABLE TESTS
SCED BRACE DRAWINGS

MARCH, 17, 2009
-
SCALE DESIGNED
SHEET NO. DRAWN
DRAWING NO. REV.
Step 6: Placing FAs, and torquing friction bolts EPM-SCED-05 2

427
APPENDIX H: SHAKE TABLE FRAME DRAWINGS
428
SCED Braced Frame UNIVERSITYȱofȱTORONTO
Shake Table Test DepartmentȱofȱCivilȱEngineering
at Ecole Polytechnique Montreal

Frame and Connections ÉCOLEȱPOLYTECHNIQUEȱMONTRÉAL
Départmentȱdesȱgéniesȱcivil,ȱgéologiqueȱetȱdesȱmines
Bill of Materials
Total
Weight (kg)
CS1 Left Column Section PT1 98 1 98 G40.21 350W
CS2 Right Column Section PT1 97 1 97 G40.21 350W
Appendix H: Shake Table Frame Drawings
BS1 Beam Section PT1 47 3 141 G40.21 350W

CP1 Base Gusset and Pin Plate A PT2 16.5 1 16.5 350W
CP2 Base Pin Plate B PT2 7 1 7 350W
CP3 Pin Douber Plate PT2 0.5 2 1 350W
CP4 Base Gusset and Pin Reinforcing Plate PT2 4 1 4 350W
CP5 Beam Gusset Plate First Story PT3 32 1 32 350W
CP6 Beam Gusset Plate Second Story PT3 32 1 32 350W
CP7 Beam Gusset Plate Third Story PT4 22 1 22 350W
CP8 Column Connection Plate A PT4 6 2 12 350W
CP9 Column Connection Plate B PT4 6 1 6 350W
CP10 Column Connection Plate C PT5 10 3 30 350W
CP11 Column Connection Plate D PT5 5 3 15 350W
CP12 Mass Connection Plate PT5 3 3 9 350W

BP1 Beam Interface Plate PT5 2 12 24 350W
BP2 Beam Connection Plate PT5 5.5 6 33 350W
LS1 Lateral Restraint Main Section PT6 140 2 280 G40.21 350W
LS2 Lateral Restraint Secondary Section PT6 10 8 80 G40.21 350W
MS1 Mass Connection Beam Section PT7 26 3 78 G40.21 350W
MP1 Mass Connection Beam End Plate A PT7 5 3 15 350W
MP2 Mass Connection Beam End Plate B PT7 4.5 3 13.5 350W
MP2A Nut Plate PT7 1.5 3 4.5 350W
MP3 Mass Grip Plate PT7 30 3 90 350W
MP4 Loading Arm Bolt Holder PT8 0.5 12 6 350W
TOTAL 1146.5

DepartmentȱofȱCivilȱEngineering SCED Braced Frame Shake Table Test 4 2009/09/01
Départmentȱdesȱgéniesȱcivil,ȱgéologiqueȱetȱdesȱmines Title and Bill of Materials J.Erochko JE04-BM1
429
Mass Grip Assembly
MA2 (dwg. A6)
Loading Arm Bolt Holder
(to be welded at EPM)
MP4 (dwg. PT8)
Mass Connection Assembly Mass Simulation Frame at EPM
MA1 (dwg. A6)
Load Cell and Fittings
at EPM
Beam Assembly
BA1 (dwg. A5)
Lateral Restraint
Sections LS1, LS2

(dwg. PT6)
Beam Connection
Plate BP1
(dwg. PT5)
Column A Assembly
CA4 (dwg. A4)
COLOURS IN THIS DRAWING ARE

FOR CLARITY ONLY AND ARE NOT

MEANT TO REPRESENT PAINT COLOURS
OF BUILT ELEMENTS.
Column B Assembly
CA5 (dwg. A4) PAINT COLOUR SUMMARY:
All Braced Frame Elements:
Base Pins at EPM (Beams, Columns and Connecting Plates)
BRIGHT YELLOW.
Base Beam, Lateral Restraint and
Base Beam at EPM SCED Braces Mass Connection Assembly:
(from South Korea) NEUTRAL NON-GLOSS RED.
Shake Table Surface See Part Drawings for Painting Area Restrictions

Départmentȱdesȱgéniesȱcivil,ȱgéologiqueȱetȱdesȱmines Test Setup Summary J.Erochko JE04-TS1
430
1 3/4"
44 15/16" 45" 59"
1142.2 1143 1498.6 44.4
5/16"
6"
8
120.6
152.4
4 3/4"
8 1/16" 8 1/16" 22 1/16" 8 1/16" 3"
204.8 204.8 560.4 204
Typ.
Typ.
Typ.
76.2
67.5
17.4
67.5
11/16"
2 11/16"
2 11/16"
157"
3987.8
LEFT COLUMN SECTION HSS152x76x8.0 Slots in CS2 may need to be enlarged to accommodate
CS1 - SCALE 1:20 CP12 plate weld (see assemblies CA2, CA3 & CA4)
37 15/16" 59" 45" 1 3/4"

964.4 1498.6 1143 44.4
5/16"
6"
8
120.6
152.4
4 3/4"
8 1/16" 22 1/16" 8 1/16" 15 1/16" 3"

204.8 560.4 204.8 381.8 76.2
27
Typ.
17.4
Typ.
Typ.
67.5
62.7
62.7
11/16"
67.5
1 1/16"
2 7/16"
2 7/16"
2 11/16"
2 11/16"
157"
3987.8
RIGHT COLUMN SECTION HSS152x76x8.0

CS2 - SCALE 1:20 1 3/4"
44.4

5/16"
6"
8
120.6
152.4
4 3/4"
The slots in CS1, CS2 and BS1

3" are 1/16" larger than the plates
76.2
6 1/16" 6 1/16"
Typ.
29.4
that they accommodate.

1 3/16"
154 154
The width of these slots may be
74 1/8"
adjusted as necessary to ensure
1882.6 proper fit and welding.
Typ.
17.4
Typ.
29.4
11/16"
1 3/16"
BEAM SECTION HSS152x76x8.0

BS1 - SCALE 1:20
Départmentȱdesȱgéniesȱcivil,ȱgéologiqueȱetȱdesȱmines Column and Beam HSS Sections J.Erochko JE04-PT1
431
16" 3 1/2"
407 88.3
13/16"
20.6 T
209
yp.
/4"
8 1 .6
14
/4"
53 6
" 2 1/8"
24.9°
104
1/2
/8"
54
4 1 .8
339.7
190.1
7 1/2"
13 2. 6
13 3/8"
34
DO NOT PAINT
27
73
/8"
(both sides)
" p
ho
104
15 S
/8"
4 1 .8
0" .00 ne
00 0 hi
2. 0"/+ ac
2.000"
339.8
27
0 M
13 3/8"
0015"
5"
73
00 by
/8"
127
0. t Tol. -0.0
000"/+0. ne Shop
l . - cu Machi
31
be cut by
/4"
to
1 1 .8
e
60.3
H ol
2 3/8"
To o be
t /2"
e 1 1 .1
ol 38
H
24.9°
/4"
149.6
5 7/8"
2 1 .2
57
60.3
2 3/8"
4 1/4"
24.9°
108
1"
25.4
2 1/8" 14"
54 .3
355 1 15/16"
49.3 BASE PIN PLATE B
CP2 - SCALE 1:5
314.8
12 3/8"
111.1
4 3/8"

55.6
2 3/16"
6 13/16" 12 11/16" p
173 322.3 5" Sho
0" .001 ne P1
i
5/8"
19 1/2" 15.9 00 0 h C
2. 0"/+ ac to
495.3 M t
00 y ed u
0. 0 ut b e ld i s c 3 7/8" 4"
l . - e c W o le
5/8"
15.9
o b e h 98.5
4.8
b 101.6
3/16"
BASE GUSSET AND PIN PLATE A T o

t to ore
e f
(for Painting restrictions, see A1) ol 3 e
H CP b BASE GUSSET
CP1 - SCALE 1:5 PIN DOUBLER PLATE AND PIN
DO NOT PAINT REINFORCING
(FOR PIN PLATE A) PLATE
CP3 - SCALE 1:5 CP4 - SCALE 1:10
Départmentȱdesȱgéniesȱcivil,ȱgéologiqueȱetȱdesȱmines Column - Base Pin Support Parts J.Erochko JE04-PT2
432
16"
407.1
209
/4"
8 1 .6
13/16"
14
/4"
°
53 6
20.6 Typ. 24.9° 24.90
SAME DIMENSIONS
9"
AS CP5 BUT WITHOUT
228.6
27
31
/4"
1 1 .8
/2"
73
NOTCH
/8"
1 1 .1
38
5/8"
15.9
22"
27
558.8
73
/8"
/4"
2 1 .2
57
2"
24.90
24.90
50.8
°
°
CL Symmetry
178.6
7 1/16"
& Beam
4"
24.90°
24.90°
101.6
89.3
3 1/2"
/
.8
41
104 8"
13
1
342 /2"
.6
190.1
7 1/2"
/
.8
41
104 8"

2 1/8" 13 7/8" 3 1/2"
54 353.1 88.2
CL Column
19 1/2"
495.3
5/8"
15.9
BEAM GUSSET PLATE FIRST STORY BEAM GUSSET PLATE SECOND STORY
(for Painting restrictions, see A2) (for Painting restrictions, see A2)
CP5 - SCALE 1:5 All dimensions in mm [inches] unless otherwise noted
CP6 - SCALE 1:5
Départmentȱdesȱgéniesȱcivil,ȱgéologiqueȱetȱdesȱmines Column-Beam-Brace Connection Parts J.Erochko JE04-PT3
433
19 1/2"
495.3
2"
50.8
13
1
342 /2"
.6
2"
50.8
CL Beam
190.9
7 1/2"
8"
203.2
4"
24.9°
101.6
/
.8
89.3
3 1/2"
41
104 8"
5/8"
/8
15.9
73 "
27
15"
381
/
.8
/8
41
104 8"
73 "
27
9"
/4
228.6
190.1
21
7 1/2"
/4 24.9° 9 5/8"
53
5/8"
15.9
146 "
57. " 244.5
2 11
/2
38. "
1
" COLUMN CONNECTION PLATE A
/4
/
.6
8
13/16 Typ.
20.6
11
81
31. "
209 4"
FIRST AND THIRD STORIES
CP8 - SCALE 1:5
2 1/8" 13 7/8"
54 353.1
3 1/2"
2"
50.8
CL Column 88.2
4"
8"
101.6
19 1/2"
203.2
495.3

5/8"
15.9
5/8"
BEAM GUSSET PLATE THIRD STORY 15.9
2"
50.8
(for Painting restrictions, see A3)

CP7 - SCALE 1:5
9 5/8"
5/8"
15.9
244.5
COLUMN CONNECTION PLATE B

SECOND STORY
CP9 - SCALE 1:5
Départmentȱdesȱgéniesȱcivil,ȱgéologiqueȱetȱdesȱmines Column-Beam-Brace Connection Parts J.Erochko JE04-PT4
434
3 1/8" 15 7/8"
79.4 403.2 38.1 82.6 38.1
PAINT ONLY
ONE SIDE
OF PLATE
38.1
38.1
1 1/2"
DO NOT PAINT (see A1,A2,A3)
(both sides)
85.8
3 3/8"
19
57.2
3/4"
57.2
2 1/4"
190.5
7 1/2"
190.6
7 1/2"
57.2
57.2
2 1/4"
13/16
"
85.8
3 3/8"
20.6
Typ.
38.1
38.1
1 1/2"
1 1/2" y p.
38.1 6" T
/1 .6
13 20
6 1/4"
9.5
3/8"
19" 158.8
5/8"
15.9
482.6
COLUMN CONNECTION PLATE C BEAM INTERFACE PLATE

CP10 - SCALE 1:5 BP1 - SCALE 1:5
3 1/8" 6" 1 1/2"

79.4 152.4 PAINT ONLY 38.1
WELDED SIDE
DO NOT PAINT AND EDGES
38.1
38.1
1 1/2"
1 1/2"
(both sides) (see A1,A2,A3)
3 1/8"
85.8
3 3/8"
79.4
19
3/4"
57.2
57.2
2 1/4"
2 1/4"
DO NOT PAINT
190.6
190.6
7 1/2"
7 1/2"
(both sides)

57.2
57.2
2 1/4"
2 1/4"
4"
13/16 13/16"
85.8
"
101.6
3 3/8"
20.6 20.6 Typ
Typ. .
38.1
38.1
1 1/2"
1 1/2"
1 1/2"
38.1
38
1 1/2"
9 1/8" 4" 9 1/8"
5/8"
15.9
5/8"
15.9
231.8 101.6 231.8
COLUMN CONNECTION PLATE D MASS CONNECTION PLATE BEAM CONNECTION PLATE

CP11 - SCALE 1:5 CP12 - SCALE 1:5 BP2 - SCALE 1:5
Départmentȱdesȱgéniesȱcivil,ȱgéologiqueȱetȱdesȱmines Column-Beam Connection Parts J.Erochko JE04-PT5
435
44 3/8" 10" 35 1/4" 10" 44 3/8"
1126.8 254 896 254 1126.8
1/4" 7"
8"
6.4
177.8
203.2
1 7"
1/
4"
1 177.8
101.6
27 6"
8"
Ty
p. 203.2
144"
3657.6
LATERAL RESTRAINT MAIN SECTION

HSS203x203x6.4
LS1 - SCALE 1:20
3" 10" 3"

76 254 76
3/8"
4"
9.5
101.6
2"
50.8
1 2 1/2"
1/
1 63.6
27 6"

Ty
p. 4"
101.6
16"
406
LATERAL RESTRAINT SECONDARY SECTION

HSS102x102x9.5
LS2 - SCALE 1:10
Départmentȱdesȱgéniesȱcivil,ȱgéologiqueȱetȱdesȱmines Lateral Restraint Beams J.Erochko JE04-PT6
436
2 1/2"
63.6
3/8"
9.5
4"
101.6
4"
101.6
40 1/2"
1028.7
MASS CONNECTION BEAM SECTION HSS102x102x9.5

2" 3" 3" 2"
MS1 - SCALE 1:5 50.8 76.2 76.2 50.8
2
le
Ho
2 1/2" 2 1/2"
1
PAINT ONLY
2"
ed
63.5 63.5
50.8
1
pp /4"-1
ONE SIDE
Ta
38.1
1 1/2"
OF PLATE 2
5"
-1 l e
127
2"
(see A6) 4" o

50.8
63.5
1/ H
2 1/2"
1 ped
p
Ta
3"
5"
5"
76.2
127
127
2"
10"
254
50.8
38.1
1 1/2"
/16
"
63.5
1 5 33.3
2 1/2"
1 1/2" 1 1/2" 6"
2"
38.1 38.1 1 1/1 Typ.

50.8
27
5"
127

2"
50.8
5" 5" 3"

38.1
127 76.2
1 1/2"
127
38.1
38.1
1 1/2"
1 1/2"
63.5
2 1/2"
MASS CONNECTION BEAM MASS CONNECTION BEAM NUT PLATE

END PLATE A END PLATE B MP2A - SCALE 1:5 10"
254
MP1 - SCALE 1:5 MP2 - SCALE 1:5
MASS GRIP PLATE
(DO NOT PAINT)
All dimensions in mm [inches] unless otherwise noted MP3 - SCALE 1:5
Départmentȱdesȱgéniesȱcivil,ȱgéologiqueȱetȱdesȱmines Mass Connection - Parts J.Erochko JE04-PT7
437
Part MP4 is to be provided
and fabricated by
machine shop
31.8
1 1/4"
63.5
2 1/2"
31.8
1 1/4"
" 015" hop
00 S
1.0 "/+0.0 hine
0 c
00 Ma 1 1/4" 1 3/4"
l . - 0.0 t by
To be cu 31.8 44.4
o
le t
Ho
14
5/8"
15.9
Bevel in Preparation for
62.2 14 a 14mm Partial Penetration
3" Groove Weld to be done
76.2 in the lab.
LOADING ARM
BOLT HOLDER
MP4 - SCALE 1:2

Départmentȱdesȱgéniesȱcivil,ȱgéologiqueȱetȱdesȱmines Mass Connection - Rod Mount J.Erochko JE04-PT8
438
3D VIEW
SCALE 1:10
CL Symmetry
16" 3 1/2"
407 88.3 4"
101.6
3 5/8" 12 3/8"
2"
92.2 314.8 CP4 50.8
DO NOT PAINT CP4

(both sides) 1 11/16"
42.9
CP1
8
190.1
7 1/2"
DO NOT PAINT
(both sides) 6"
5" 152
339.8
13 3/8"
127
Identical-Sized
Concentric Holes CP1
drilled by machine

shop after welding CP3
149.6
5 7/8"
CP3
5"
Typ
127
4.8
CP3
4.8
3/16"
3/16"
6 13/16" 12 11/16" 4.8
3/16" 173 322.3
5/16"
4.8 19 1/2" 8
495.3
CL Symmetry
BASE PIN AND GUSSET ASSEMBLY
CA1 - SCALE 1:5
Départmentȱdesȱgéniesȱcivil,ȱgéologiqueȱetȱdesȱmines Base Pin and Gusset Assembly J.Erochko JE04-A1
439
16" 3 1/2"
407.1 88.2
CP5
CP5
190.1
7 1/2"
9"
228.6
6"
152 CP12
5/8" 2"
15.9 CP Grind Surface of Weld
50.8
Flush with CP5 so that
plate fits through slot in
CP12 Column Section CS2

ALTERNATIVELY:
Weld with 12mm fillet
CL Symmetry either side and enlarge
22"
178.6
558.8
7 1/16"
slot in CS2 column as

1 1/2"
4"
required.
101.6
38
DO NOT PAINT DO NOT PAINT 1 11/16"

(Left surface) 7/8" (BOTH SIDES) 42.9
22.1
6"
152
9"
228.6
190.1
7 1/2"

19 1/2" 5/16"
495.3 8
3D VIEW
SCALE 1:10
BEAM AND GUSSET CONNECTION ASSEMBLY A
CA2 - SCALE 1:5 CL Symmetry

Départmentȱdesȱgéniesȱcivil,ȱgéologiqueȱetȱdesȱmines Gusset Connection Assembly A J.Erochko JE04-A2
440
19 1/2" 4"
495.3 101.6
2"
5/8" 50.8 CP12
2"
15.9
50.8
DO NOT PAINT CP7
(Left surface)
CP12 DO NOT PAINT
(both sides)
1 1/2"
4"
190.9
7 1/2"
101.6
38
CP Grind Surface of Weld

Flush with CP5 so that 1 11/16"
7/8" plate fits through slot in 42.9
6"
22.1 Column Section CS2 152
ALTERNATIVELY:
Weld with 12mm fillet
either side and enlarge
slot in CS2 column as
9"
required.
228.6
190.1
7 1/2"
CP7
3D VIEW
SCALE 1:10
16" 3 1/2"
407.1 88.2 5/16"
8
CP9 BEAM AND GUSSET CONNECTION ASSEMBLY B
9 5/8" CA3 - SCALE 1:5
244.5 4" CL Symmetry
101.6
5/8" 2"
2"
50.8
15.9 50.8
CP9
CP Grind Surface of Weld

CP12 Flush with CP5 so that
plate fits through slot in CP12
Column Section CS2
4"
8"
101.6
203.2
1 1/2" ALTERNATIVELY:
38 Weld with 12mm fillet
DO NOT PAINT either side and enlarge
(Left surface) slot in CS2 column as 1 11/16"
7/8" 42.9
required.
2"
22.1
50.8
5/16"
8
10 1/2" 3D VIEW
266.6 SCALE 1:10
CL Symmetry
BEAM CONNECTION ASSEMBLY C
CA4 - SCALE 1:5
Départmentȱdesȱgéniesȱcivil,ȱgéologiqueȱetȱdesȱmines Gusset Connection Assembly B J.Erochko JE04-A3
441
52" 52" 52" 8
1320.8 1320.8 1320.8
2 3/8" 3" 49" 8 4" 4" CS1
11" 11"
54
76.2
5/8"
5/8"
5/8"
15.9
15.9
15.9
60.3 1244.6 101.6 Typ. 101.6 Typ.
2 1/8"
279.4 Typ. 279.4 Typ.
CL Col.
5/8"
5 3/8" 8 8
8 8 15.9
190.5
7 1/2"
136.5 8"
231.8
269.9
269.9
231.8
190.5
9 1/8"
9 1/8"
7 1/2"
10 5/8"
10 5/8"
203.3
314.9
12 3/8"
8
19"
CL Hole CL Hole
520.7
482.6
441.3
Hole to be
20 1/2"
17 3/8"
Drilled by CP11 CP8 CS1 CP8 CP11
machine shop 5/8" CP6
15.9
CL Plate CP11 CL Plate CP8
after assembly CS1 CL Hole
of CA5 CA1
CP11 CL Plate CP10 CP11
CA1 CP8 CP10 CP8
CP6
1/2"
12.8
CL
Symmetry
2"
8
50.8
8
8
8
5/16"
95.3
95.3
95.3
162 3/8"
5/16"
5/16"
5/16"
3 3/4"
3 3/4"
3 3/4"
4124.3
COLUMN A ASSEMBLY
CA5 - SCALE 1:20
MP4 Behind CP10 CP11
162 3/8" MP4 Behind MP4 Behind
CA2 4124.3 CA4
8
1/2"
12.7
5/16"
CL
Symmetry
CP10
8
8
13 3/8"
95.3
95.3
95.3
5/16"
5/16"
3 3/4"
3 3/4"
3 3/4"
339.7 CL Plate CA3 CL Plate

CS2
CP10 CA3
CP2
CL Hole CL Hole
CA2

CP10
CL Plate CA3
5 3/8" 8" CS2 CA4 CP11 CP10
136.5 203.2 8
19"
19"
CL Hole
441.3
482.6
520.7
441.3
482.6
520.7
17 3/8"
20 1/2"
2 3/8"
17 3/8"
20 1/2"
8 8
54
60.3 8
2 1/8"
CP2
5/8"
5/8"
15.9
15.9
231.8
269.9
9 1/8"
10 5/8"
190.5
7 1/2"
CL Col.
8 49" 8 CS2
3"
54
1244.6 8 8
2 1/8"
38
38
76.2
5/8"
5/8"
8
38
15.9
15.9
5/8"
52" 52" 52"
1 1/2"
1 1/2"
15.9
1 1/2"
1320.8 14 1320.8 14 1320.8 14 19

MP4
Hole to be
MP4x2 MP4 19
Drilled by MP4x2 MP4x2
machine shop COLUMN B ASSEMBLY (Welds at
(Welds at Machine Shop) (Welds at
after assembly Machine Shop) CA6 - SCALE 1:20 Machine Shop)
of CA6 All dimensions in mm [inches] unless otherwise noted
Départmentȱdesȱgéniesȱcivil,ȱgéologiqueȱetȱdesȱmines Column Assemblies J.Erochko JE04-A4
442
3D VIEW
SCALE 1:20
77 3/8"
1965.2
3 1/8" 6"
6"
79.4 Typ. 152.4 Typ.
152.4
3/4"
19.1 Typ.
CL
190.6 Typ.
7 1/2"
Symmetry
BP2 BS1
9 1/8"
231.8 Typ. BP2
BS1
3"
95.3 Typ.
76.2
3 3/4"
BP2 BP2 BP2
CL

Symmetry
8
74 1/8"
Typ.
5/16"
1882.6 BS1
80 3/8"
2041.4
BEAM ASSEMBLY
BA1 - SCALE 1:10

Départmentȱdesȱgéniesȱcivil,ȱgéologiqueȱetȱdesȱmines Beam Assembly J.Erochko JE04-A5
443
3D VIEW DO NOT PAINT
SCALE 1:10 outer surface of MP1
1" 1" 5"

25.4 25.4 43 1/2" 127
4"
1104.9 101.6
3" MP1
1"
76.2 MP4
25.4
8 8
10
5"
127
CL Symmetry
4"
3"
76.2
101.6
MP2A
1"
25.4
MP2A 19
1/2"
12.7
1/2"
12.7
63.5
2 1/2"
MP2 MP4 19
MP4
MP2 MS1
CL Symmetry MP1 CL Symmetry
1 1/2" 40 1/2" 1 1/2"
38.1 1028.7 38.1
MP2 MS1 MP1 14 MP4 (Welds at Machine Shop)

MP2A
19
CL Symmetry
19
MP4 (Welds at Machine Shop)
1/2"
12.7
1/2"
12.7
63.5
2 1/2"
14
MASS CONNECTION ASSEMBLY
CL MA1 - SCALE 1:10

CL
10"
254
10"
MP3 254 3D VIEW
SCALE 1:10
MASS GRIP ASSEMBLY
MA2 - SCALE 1:10
Départmentȱdesȱgéniesȱcivil,ȱgéologiqueȱetȱdesȱmines Mass Connection Assemblies J.Erochko JE04-A6
444
Uniaxial Strain Gauges on SCED Braces
String Pots 2 on inner tube, 2 on outer tube
Every Storey and Base either side of tube near end Accelerometers Every
LEFT COLUMN Fixed to Outside 12 Total Level and at
STRAIN GAUGES Support (4 Total) Shake Table Base
2" 50kip Load Cell A
2"
50.8 Every Storey
50.8
13"
9"
330.2
228.6
External Fuse LVDT
(Every Storey)
3 Total
Tube Relative Movement
2"
50.8
LVDT on either side of brace
(Every Storey) 6 Total
13"
330.2
13"
330.2
2"
50.8
9"
228.6
13"
330.2
13"
9"
330.2
228.6

2x50kip Load Cell A
Mass System Base
2"
to Shake Table
13"
50.7
330.2
Locations of Uniaxial
Vertical Strain Gauges
(both sides of column)
24 Total

Départmentȱdesȱgéniesȱcivil,ȱgéologiqueȱetȱdesȱmines Instrumentation Summary J.Erochko JE04-IN1
445

Erochko Jeffrey A 201306 PHD Thesis PDF

Uploaded by

Copyright:

Available Formats

Erochko Jeffrey A 201306 PHD Thesis PDF

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

Erochko Jeffrey A 201306 PHD Thesis PDF

Uploaded by

Copyright:

Available Formats

What is the research improving upon?

What is the research improving upon?

What were the goals of the three experimental studies?

What were the goals of the three experimental studies?

Improvements to the Design and Use of Post-Tensioned

Self-Centering Energy-Dissipative (SCED) Braces

A thesis submitted in conformity with the requirements

© Copyright by Jeffrey A. Erochko, 2013

erations while only causing a small increase in the base shear.

List of Figures..................................................................................................... xiv

Chapter 1: Introduction ........................................................................................1

Chapter 2: Background and Motivation ................................................................5

2.1 Performance-Based Design – 5

2.8 Impetus for the Research / Necessary Improvements – 31

3.1 The Components of a SCED Brace – 36

Chapter 4: Design and Testing of a Prototype High-Capacity (HC-SCED) Brace ..

4.1 High-Capacity SCED (HC-SCED) Brace Design and Innovations – 117

Chapter 5: Design and Testing of a Prototype Telescoping (T-SCED) Brace ....177

5.1 Telescoping SCED (T-SCED) Design and Construction – 178

5.1.1 T-SCED Design – 178

Chapter 6: Shake Table Testing of a SCED Braced Frame ................................224

6.1 Design Structure – 224

6.7 Shake Table Test Results and Discussion – 254

Chapter 7: SCED Frame Building Model Comparisons ....................................274

7.1 Six-Storey Building Design (Choi et al., 2008) – 275

Chapter 8: Conclusions .....................................................................................324

8.1 Summary of the Research – 324

8.3.6 More Extensive Nonlinear Modelling – 331

Appendix A: SCED Mechanics Simulator Source Code ....................................343

Appendix B: HC-SCED Drawings ....................................................................357

Appendix C: Friction Subcomponent Test Drawings ........................................363

Appendix D: T-SCED Drawings.......................................................................369

Appendix E: Vertical Test Frame Drawings.......................................................386

Appendix G: Shake Table SCED Drawings (Kim, 2009a).................................422

Appendix H: Shake Table Frame Drawings.......................................................428

Table 4.11: Combined ASCE/AISC Deformation Protocol – 158

Table 7.10: Maximum Considered (MCE) Hazard Level Earthquake Records

Figure 3.14: SCED Brace Detailed Hysteretic Response – 62

Figure 4.17: HC-SCED Test Setup and Instrumentation – 144

Figure 5.7: T-SCED Tendon Tensioning Order – 186

HN - height of a bolt nut

L p0 - initial unloaded pretensioning tendon length

Ps - external friction fuse slip force of a SCED brace

 - SCED brace energy dissipation capacity parameter

Yielding/Friction System Self-Centering System

Jeffrey Erochko, Department of Civil Engineering, University of Toronto

Jeffrey Erochko, Department of Civil Engineering, University of Toronto

Jeffrey Erochko, Department of Civil Engineering, University of Toronto

2.1 PERFORMANCE-BASED DESIGN

Jeffrey Erochko, Department of Civil Engineering, University of Toronto

Jeffrey Erochko, Department of Civil Engineering, University of Toronto

Jeffrey Erochko, Department of Civil Engineering, University of Toronto

2.2 RESIDUAL DRIFTS

2.2.1 THE IMPORTANCE OF RESIDUAL DRIFTS

Jeffrey Erochko, Department of Civil Engineering, University of Toronto

Jeffrey Erochko, Department of Civil Engineering, University of Toronto

6 Storey Models 12 Storey Models

2.2.2 RESIDUAL DRIFTS IN PERFORMANCE-BASED DESIGN

Jeffrey Erochko, Department of Civil Engineering, University of Toronto