SC Pre Conc Wall in Building For Low To Medium Seismicity PDF

http://researchspace.auckland.ac.
nz
ResearchSpace@Auckland
Copyright Statement
The digital copy of this thesis is protected by the Copyright Act 1994 (New
Zealand).
This thesis may be consulted by you, provided you comply with the
provisions of the Act and the following conditions of use:
x Any use you make of these documents or images must be for

research or private study purposes only, and you may not make
them available to any other person.
x Authors control the copyright of their thesis. You will recognise the
author's right to be identified as the author of this thesis, and due
acknowledgement will be made to the author where appropriate.
x You will obtain the author's permission before publishing any
material from their thesis.
To request permissions please use the Feedback form on our webpage.

http://researchspace.auckland.ac.nz/feedback
General copyright and disclaimer
In addition to the above conditions, authors give their consent for the
digital copy of their work to be used subject to the conditions specified on
the Library Thesis Consent Form and Deposit Licence.
Note : Masters Theses
The digital copy of a masters thesis is as submitted for examination and

contains no corrections. The print copy, usually available in the University
Library, may contain corrections made by hand, which have been
requested by the supervisor.
Self-centering Precast Concrete
Walls for Buildings in Regions with
Low to High Seismicity
By Richard S. Henry
A thesis submitted in partial fulfilment of the requirements
for the degree of Doctor of Philosophy
Supervised by Assoc. Prof. Jason Ingham and Prof. Sri Sritharan
Department of Civil and Environmental Engineering
The University of Auckland
June 2011
ABSTRACT
Self-centering precast concrete walls that utilise unbonded post-tensioning (PT) can provide
superior lateral load resistance to conventional concrete construction by minimising structural
damage and residual drifts. This thesis investigates the application of both simple self-
centering walls that are suitable for regions of low seismicity and more complex wall systems
that are suitable for regions with moderate to high seismicity.
A series of experimental tests confirmed the behaviour of individual unbonded PT concrete

walls that had no specific confinement reinforcement in the compression toe. Individual PT
walls are suitable for buildings located in regions of low seismicity, where the lateral drift
demand is small and so extensive damage to the wall toe can be avoided. The nominal flexural
strength can be used to easily design the wall, and an ultimate compressive strain in excess of
0.003 may be more suitable for PT walls. The results from both experimental tests and finite
element analyses were used to quantify the strains in the wall toe and develop refined equations
to accurately predict the unbonded tendon stresses and wall nominal flexural strength.
Energy dissipating O-Connectors were designed for use in a self-centering wall system that
consists of a Precast Wall with End Columns (PreWEC), and is suitable for regions of
moderate to high seismicity. A finite element model was developed that accurately captured
the cyclic behaviour of the PreWEC system that was observed during experimental testing.
The PreWEC model was used to investigate the influence of several important design
parameters, including the relative quantities of PT and energy dissipating connectors.
Current design procedures that are used to ensure that self-centering is achieved were shown to
be inaccurate. Analysis of the cyclic hysteresis behaviour of self-centering systems and
dynamic time-history analyses were used to demonstrate how self-centering should be defined.
The PreWEC system was used as an example to demonstrate that realistic residual drift limits
could be satisfied following an earthquake using a simple design check that was developed.
Lastly, the interaction between the self-centering wall and the surrounding structure was
investigated. Detailed finite element and time-history analysis of a prototype building
indicated the influence of the type of wall-to-floor connection that was incorporated into the
building, and concluded that it is not sufficient to model the wall system in isolation.
-i-
- ii -
ACKNOWLEDGMENTS
I would like to acknowledge the support of my supervisors, Assoc. Professor Jason Ingham and
Professor Sri Sritharan. Jason your continued enthusiasm, encouragement and support have
been central to both the completion of this thesis and my career development. Sri your input
into the direction and technical aspects of this project has been immense. My time in Iowa was
truly a life changing experience. I am grateful to have you two as both mentors and friends, it
has been an enjoyable journey and I look forward to working with you both in the future.
I would also like to thank several others who assisted or advised me with technical aspects of
this project. Nic Brooke for your help during the wall panel tests, Dr. Gavin Wight and
Dr. Liam Wotherspoon for your assistance with ABAQUS and Ruaumoko respectively, and
Dr. Quincy Ma and Len McSaveney for your general advice and support throughout this study.
This project would not have been possible without the financial contributions provided by
University of Auckland, Iowa State University, the Tertiary Education Commission, Research
and Education Advanced Network New Zealand Ltd., Fulbright New Zealand, Ministry or
Research Science and Technology, Golden Bay Cement Ltd. and Pacific Steel Group.
I also greatly appreciate the testing supplies that were donated by Stresscrete Ltd. Papakura,
Golden Bay Cement Ltd., VSL Australia Pty. Ltd., and Construction Techniques Group Ltd.
Additionally, I would like to acknowledge Hank Mooy, Tony Daligan, Mark Byrami, Sujith
Padiyara, and Mark Twiname for their assistance during the experimental testing.
I would like to acknowledge all the staff and post-graduate students at the University of
Auckland and Iowa State University how have assisted or supported me during this study, and
in particular Aaron Wilson, Tom Algie, and Sriram Aaleti, for your friendship while we were
all “living the dream”.
Thank-you to my family and friends who have supported and encouraged me throughout the
duration of this study, I promise I will try to come out from hiding now.
Last but not least, to my wife Weiwei. I would not have made it through if it were not for you,
thank-you for your endless encouragement and for never giving up on me. We can both thank
this thesis for providing us the opportunity to meet in the vast corn fields of Iowa!
- iii -
- iv -
TABLE OF CONTENTS
Abstract ...................................................................................................................................... i
Acknowledgments ...................................................................................................................iii
Table of Contents ...................................................................................................................... v
List of Figures ........................................................................................................................xiii
List of Tables ......................................................................................................................... xix
Notation ................................................................................................................................. xxi
Chapter 1 Introduction ............................................................................... 1

1.1 Overview........................................................................................................................ 1
1.2 Research Motivation and Objectives ............................................................................. 4
1.2.1 Objectives ............................................................................................................... 5
1.3 Thesis Outline ................................................................................................................ 6
Chapter 2 Literature Review ..................................................................... 9

2.1 Concrete Walls............................................................................................................... 9
2.2 Rocking Behaviour ...................................................................................................... 10
2.3 Self-centering Concrete Systems ................................................................................. 11
2.3.1 PRESSS ................................................................................................................ 11
2.3.2 Individual Walls ................................................................................................... 12
2.3.3 Wall Systems ........................................................................................................ 13
2.3.3.1 Jointed walls ..................................................................................................... 14
2.3.3.2 Hybrid walls...................................................................................................... 15
2.3.3.3 Hybrid coupling beams ..................................................................................... 17
2.3.3.4 Other sources of energy dissipation .................................................................. 18
2.3.3.5 Codification ...................................................................................................... 18
2.3.4 Building Response ................................................................................................ 19
2.3.4.1 PRESSS building .............................................................................................. 19
2.3.4.2 Mexico test........................................................................................................ 20
2.3.4.3 Australian test ................................................................................................... 20
2.3.4.4 DSDM ............................................................................................................... 21
2.3.4.5 Moment frames ................................................................................................. 22
2.4 Dynamic Behaviour ..................................................................................................... 22
2.4.1 Shake-table Experiments ...................................................................................... 22
2.4.2 Time-history Analysis .......................................................................................... 23
2.4.2.1 Hysteresis behaviour ......................................................................................... 23
2.4.2.2 Wall and building analysis ................................................................................ 25
2.5 Application of Systems ................................................................................................ 28
2.6 Discussions and Conclusions ....................................................................................... 29
-v-
Table of Contents
2.7 References ................................................................................................................... 30
Chapter 3 Testing of Post-tensioned Wall Panels.................................... 37

3.1 Nominal Flexural Strength .......................................................................................... 37
3.2 Construction Details .................................................................................................... 39
3.2.1 Wall Panel Specification ...................................................................................... 39
3.2.2 Panel Construction ............................................................................................... 42
3.2.3 Material Properties ............................................................................................... 44
3.2.3.1 Concrete............................................................................................................ 44
3.2.3.2 Prestressing steel .............................................................................................. 44
3.3 Testing Details............................................................................................................. 45
3.3.1 Test Setup............................................................................................................. 45
3.3.2 Test Procedure...................................................................................................... 47
3.4 Instrumentation............................................................................................................ 47
3.4.1 Strain Measurement ............................................................................................. 48
3.4.1.1 External gauges (Ext) ....................................................................................... 49
3.4.1.2 Strain gauges on reinforcement mesh (SGR) .................................................... 49
3.4.1.3 Surface mounted concrete strain gauges (SGS) ................................................ 49
3.4.1.4 Embedded concrete strain gauges (SGE) .......................................................... 49
3.4.1.5 Photogrammetry image tracking (Photo) ......................................................... 49
3.4.2 Photogrammetry Trials ........................................................................................ 51
3.5 Test Results ................................................................................................................. 53
3.5.1 General Observations ........................................................................................... 53
3.5.1.1 Phase I .............................................................................................................. 54
3.5.1.2 Phase II ............................................................................................................. 55
3.5.1.3 Phase III ............................................................................................................ 56
3.5.2 Force-Displacement Response ............................................................................. 56
3.5.3 Wall Deformation Components ........................................................................... 59
3.5.4 Neutral Axis Depth .............................................................................................. 60
3.5.5 Wall Slip .............................................................................................................. 62
3.5.6 Prestressing Force ................................................................................................ 63
3.5.7 Concrete Compressive Strain ............................................................................... 64
3.5.7.1 External displacement gauges .......................................................................... 64
3.5.7.2 Strain gauges on reinforcing steel .................................................................... 65
3.5.7.3 Concrete strain gauges...................................................................................... 65
3.5.7.4 Photogrammetry ............................................................................................... 66
3.5.7.5 Comparison of strain for wall C1 ..................................................................... 67
3.6 Discussion of Results .................................................................................................. 68
3.6.1 Nominal Flexural Strength ................................................................................... 68
3.6.2 Bedding Layer ...................................................................................................... 72
3.6.3 Rigid Body Assumption ....................................................................................... 73
- vi -
Table of Contents
3.6.4 Initial Stiffness...................................................................................................... 73

3.6.5 Shear Friction ....................................................................................................... 75
3.6.6 Yielding of Post-tensioning Steel ......................................................................... 75
3.7 Conclusions.................................................................................................................. 76
3.8 References .................................................................................................................... 77
Chapter 4 Analysis and Design of Post-tensioned Wall Panels .............. 79

4.1 Finite Element Modelling ............................................................................................ 80
4.1.1 2D Model .............................................................................................................. 80
4.1.1.1 Model description ............................................................................................. 80
4.1.1.2 Validation of the 2D FEM ................................................................................ 82
4.1.2 3D Model .............................................................................................................. 84
4.1.2.1 Model description ............................................................................................. 84
4.1.2.2 Validation of the 3D model .............................................................................. 86
4.2 Predicting Nominal Flexural Strength ......................................................................... 90
4.2.1 Data Set for Validation ......................................................................................... 91
4.2.2 Predicting Tendon Stress ...................................................................................... 93
4.2.2.1 Review of existing equations ............................................................................ 93
4.2.2.2 Comparison with current equations .................................................................. 97
4.2.2.3 Development of an improved equation ............................................................. 99
4.2.3 Predicting Flexural Strength ............................................................................... 104
4.2.4 Predicting Displacement ..................................................................................... 105
4.3 Conclusions................................................................................................................ 108
4.4 References .................................................................................................................. 109
Chapter 5 PreWEC Connector Design .................................................. 113

5.1 PreWEC Wall System................................................................................................ 113
5.2 PreWEC Connector Requirements ............................................................................ 115
5.3 Possible Types of Connector ..................................................................................... 117
5.4 Finite Element Modelling .......................................................................................... 119
5.4.1 Slotted Flexural Plate (SFP) ............................................................................... 120
5.4.2 Flexural Plates with Holes .................................................................................. 124
5.4.3 J-Shaped Flexural Plate (J-Connector) ............................................................... 126
5.4.4 Oval Shaped Flexural Plate (O-Connector)........................................................ 128
5.4.5 Summary of FEM Analyses ............................................................................... 130
5.4.6 Optimisation of O-Connector ............................................................................. 130
5.5 Experimental Validation ............................................................................................ 133
5.5.1 Test Setup ........................................................................................................... 133
5.5.2 Detailed FEM of Test Configuration.................................................................. 136
5.5.3 Experimental and Analytical Results ................................................................. 137
5.6 Large-scale PreWEC Test.......................................................................................... 140
- vii -
Table of Contents
5.6.1 O-Connector Design and Testing....................................................................... 141

5.6.2 PreWEC Test Results and Connector Performance ........................................... 144
5.7 Conclusions ............................................................................................................... 145
5.8 References ................................................................................................................. 147
Chapter 6 Finite Element Modelling of the PreWEC System ................. 149

6.1 Model Formulation .................................................................................................... 150
6.2 Confined Concrete ..................................................................................................... 151
6.2.1 Fundamental Equations ...................................................................................... 151
6.2.2 ABAQUS Concrete Damaged Plasticity Model ................................................ 154
6.2.3 FEM Trials ......................................................................................................... 154
6.2.3.1 Columns tests ................................................................................................. 154
6.2.3.2 Wall tests ........................................................................................................ 159
6.2.4 PreWEC Confined Toe Region .......................................................................... 164
6.3 Modelling of PreWEC ............................................................................................... 166
6.3.1 Details of PreWEC Test Specimen .................................................................... 166
6.3.2 Material Properties ............................................................................................. 169
6.3.2.1 Concrete and grout ......................................................................................... 169
6.3.2.2 Steel ................................................................................................................ 171
6.3.3 FEM Formulation............................................................................................... 172
6.3.4 Solution Control and Monotonic Trials ............................................................. 175
6.3.4.1 Implicit vs. explicit solver .............................................................................. 175
6.3.4.2 Explicit solver control parameters .................................................................. 176
6.3.4.3 Inclusion of wall reinforcing steel .................................................................. 178
6.3.4.4 Steel channels in the wall toe ......................................................................... 178
6.4 PreWEC Cyclic Response ......................................................................................... 179
6.4.1 General Behaviour ............................................................................................. 180
6.4.2 Concrete Cyclic Parameters ............................................................................... 180
6.4.2.1 Global response .............................................................................................. 181
6.4.2.2 Local response ................................................................................................ 185
6.4.3 Wall-to-Foundation Contact Properties ............................................................. 189
6.4.4 Recommendations .............................................................................................. 190
6.5 Modified PreWEC Designs ....................................................................................... 191
6.5.1 Number of Connectors ....................................................................................... 192
6.5.1.1 More connectors and less PT.......................................................................... 192
6.5.1.2 Fewer connectors and more PT ...................................................................... 194
6.5.1.3 More connectors and lower PT stress............................................................. 195
6.5.2 Elastic Tendons .................................................................................................. 196
6.5.3 Wall Thickness ................................................................................................... 197
6.6 Influence of Connectors ............................................................................................ 198
6.7 PreWEC Core Wall System ...................................................................................... 203
- viii -
Table of Contents
6.7.1 PreWEC Core Wall Concept .............................................................................. 203

6.7.2 FEM Analysis ..................................................................................................... 204
6.7.2.1 In-plane direction ............................................................................................ 204
6.7.2.2 Bidirectional loading....................................................................................... 206
6.7.3 Summary ............................................................................................................ 206
6.8 Conclusions................................................................................................................ 208
6.8.1 FEM Development ............................................................................................. 208
6.8.2 Investigation of PreWEC Behaviour .................................................................. 209
6.9 References .................................................................................................................. 210
Chapter 7 Analysis of Residual Drifts .................................................... 213

7.1 Background ................................................................................................................ 214
7.1.1 Static Self-centering ........................................................................................... 214
7.1.2 Current Design Procedures ................................................................................. 215
7.1.3 Dynamic Self-centering ...................................................................................... 216
7.2 Static Analysis of Flag-shape Response .................................................................... 218
7.2.1 Lumped Plasticity Model ................................................................................... 218
7.2.2 Ruaumoko Analysis ........................................................................................... 219
7.2.3 Summary ............................................................................................................ 223
7.3 Residual Drift Limits ................................................................................................. 223
7.4 Dynamic Analysis ...................................................................................................... 226
7.4.1 Model Development ........................................................................................... 227
7.4.1.1 Two-spring model ........................................................................................... 227
7.4.1.2 Seven-spring model ........................................................................................ 229
7.4.2 Displacement Based Design ............................................................................... 234
7.4.3 Parametric Matrix ............................................................................................... 235
7.4.3.1 Fundamental period / building height ............................................................. 238
7.4.3.2 Hysteretic energy dissipation .......................................................................... 241
7.4.3.3 Viscous damping............................................................................................. 241
7.4.4 Earthquake Records ............................................................................................ 242
7.4.4.1 Ground motion selection................................................................................. 242
7.4.4.2 Ground motion scaling.................................................................................... 244
7.4.5 Analysis Details .................................................................................................. 245
7.4.6 Analysis Results and Discussion ........................................................................ 246
7.4.6.1 Typical responses............................................................................................ 246
7.4.6.2 Wall set Ptest .................................................................................................... 247
7.4.6.3 Wall sets Pmore and Pless ................................................................................... 251
7.4.6.4 Pulse type ground motions.............................................................................. 252
7.4.6.5 Residual drift ratio .......................................................................................... 253
7.4.6.6 Assumed viscous damping.............................................................................. 257
7.5 Design Procedure ....................................................................................................... 259
- ix -
Table of Contents
7.6 Implications to Other Systems .................................................................................. 260

7.7 Conclusions ............................................................................................................... 261
7.8 References ................................................................................................................. 262
Chapter 8 Analysis of a Self-centering Building ..................................... 265

8.1 Introduction ............................................................................................................... 266
8.2 Wall-to-Floor Connection Details ............................................................................. 267
8.2.1 Cast-in-Place Floor ............................................................................................ 267
8.2.2 Precast Floor Units Parallel to the Wall ............................................................. 268
8.2.3 Precast Floor Units Bearing on Wall ................................................................. 271
8.2.4 Possible Connections for PreWEC Systems ...................................................... 272
8.3 Prototype Building .................................................................................................... 274
8.3.2 Precast Floor Units ............................................................................................. 275
8.4 Finite Element Modelling.......................................................................................... 276
8.4.1 Description of Model ......................................................................................... 276
8.4.1.1 Material properties ......................................................................................... 276
8.4.1.2 FEM construction ........................................................................................... 277
8.4.2.1 Connection modelling .................................................................................... 279
8.4.2.2 Results ............................................................................................................ 279
8.4.3 Isolated Floor ..................................................................................................... 283
8.4.3.1 Connection modelling .................................................................................... 283
8.4.3.2 Results ............................................................................................................ 284
8.4.4 Section Component Model................................................................................. 285
8.4.4.1 Description of model ...................................................................................... 285
8.4.4.2 Results ............................................................................................................ 286
8.5 Dynamic Analysis ..................................................................................................... 287
8.5.1 Description of Model ......................................................................................... 288
8.5.2 Earthquake Ground Motions .............................................................................. 290
8.5.3 Analysis Details ................................................................................................. 291
8.5.4 Results ................................................................................................................ 291
8.5.4.1 Interstorey drift ............................................................................................... 292
8.5.4.2 Residual interstorey drift ................................................................................ 293
8.5.4.3 Floor acceleration ........................................................................................... 294
8.5.4.4 Base shear ....................................................................................................... 295
8.6 Conclusions ............................................................................................................... 296
8.7 References ................................................................................................................. 298
Chapter 9 Conclusions ........................................................................... 299

9.1 Individual PT Walls – Chapters 3 & 4 ...................................................................... 300
-x-
Table of Contents
9.2 PreWEC System – Chapters 5 & 6 ............................................................................ 302

9.2.1 Connector Design ............................................................................................... 302
9.2.2 PreWEC FEM..................................................................................................... 302
9.3 System Behaviour – Chapters 7 & 8 .......................................................................... 303
9.3.1 Residual Drift Analysis ...................................................................................... 303
9.3.2 Building Analysis ............................................................................................... 304
9.4 Recommendations for Future Research ..................................................................... 305
APPENDIX A: Individual PT Wall Test Results .................................. 307
APPENDIX B: PreWEC Confinement Calculations ............................ 373
APPENDIX C: Ground Motion Scaling Graphs .................................. 377
APPENDIX D: Residual Drift Analysis Graphs ................................... 381
- xi -
Table of Contents
- xii -
LIST OF FIGURES
Figure 1.1 – Lateral load behaviour of an unbonded post-tensioned (PT) concrete wall ............ 3
Figure 1.2 – Hysteresis behaviour of traditional and self-centering systems ............................... 4
Figure 2.1 – PRESSS test building [2-20] .................................................................................. 12
Figure 2.2 – Lateral force-displacement response for Lehigh tests [2-22]................................. 13
Figure 2.3 – Jointed wall system [2-31] ..................................................................................... 14
Figure 2.4 – Moment-displacement response for PRESSS jointed wall system [2-20] ............. 15
Figure 2.5 – Hybrid PT wall concept [2-35] .............................................................................. 16
Figure 2.6 – Lateral force-drift response for hybrid wall test [2-37] ......................................... 17
Figure 2.7 – Shake-table test of DDSM precast building at San Diego ..................................... 21
Figure 2.8 – The Paramount building in San Francisco [2-67] .................................................. 28
Figure 2.9 – Victoria University building in Wellington, NZ [2-70] ......................................... 29
Figure 3.1 – Strain profile along the edge of an RC and PT wall .............................................. 39
Figure 3.2 – Phase I design drawings ......................................................................................... 41
Figure 3.3 – Phase II and III design drawings ............................................................................ 42
Figure 3.4 – Construction of wall panels for Phase I and III ..................................................... 43
Figure 3.5 – Measured stress-strain response of prestressing bar and strand samples ............... 45
Figure 3.6 − Wall test setup........................................................................................................ 46
Figure 3.7 – Typical instrumentation setup ................................................................................ 48
Figure 3.8 – Instrumentation used to measure concrete strains at the wall toe .......................... 50
Figure 3.9 – Photogrammetry calibration tests........................................................................... 52
Figure 3.10 – Results from photogrammetry calibration tests ................................................... 52
Figure 3.11 – Observed base crack opening in wall D1-i .......................................................... 53
Figure 3.12 – Observed damage in wall E-ii at the maximum lateral displacement .................. 54
Figure 3.13 – Condition of the bedding layer after removal of the wall .................................... 56
Figure 3.14 – Lateral force-displacement behaviour for walls A and B .................................... 57
Figure 3.15 – Lateral force-displacement behaviour for walls C1 and C2 ................................ 58
Figure 3.16 – Lateral force-displacement behaviour for walls D1 and D2 ................................ 58
Figure 3.17 – Lateral force-displacement behaviour for walls E and F ..................................... 59
Figure 3.18 – Calculated wall deformation components ............................................................ 60
Figure 3.19 – Calculated neutral axis (NA) depth for walls E and F ......................................... 61
Figure 3.20 – Comparison of neutral axis (NA) depth for bedding type and condition............. 62
Figure 3.21 – Measured wall and loading beam slip for test E-iv .............................................. 62
Figure 3.22 – Measured stresses in unbonded post-tensioning .................................................. 63
Figure 3.23 – Experimental strain measurements from external displacement gauges ............. 64
Figure 3.24 – Experimental strain measurements from reinforcement strain gauges ................ 65
Figure 3.25 – Experimental strain measurements from concrete strain gauges ......................... 66
Figure 3.26 – Experimental strain measurements from photogrammetry .................................. 67
Figure 3.27 – Comparison of measured concrete strains for different bedding types................ 68
Figure 3.28 – Photos of the wall toe condition at measured strains of 0.003 ............................. 69
- xiii -
List of Figures
Figure 3.29 – Measured lateral force-displacement response correlated with average measured
compressive strains ............................................................................................... 70
Figure 3.30 – Maximum measured compressive strains from embedded strain gauges............ 70
Figure 3.31 – Maximum measured compressive strains against axial stress ratio .................... 71
Figure 3.32 – Measured and calculated flexural deformation using Eq. 3.3 ............................. 74
Figure 3.33 – Measured and calculated shear deformation using Eq. 3.4 ................................. 74
Figure 4.1 – Concrete stress-strain response assumed during FEM .......................................... 81
Figure 4.2 – 2D FEM representation of test wall E ................................................................... 82
Figure 4.3 – Calculated lateral force-displacement response using 2D FEM ............................ 83
Figure 4.4 – Calculated compressive strains using 2D FEM ..................................................... 84
Figure 4.5 – 3D FEM representation of test wall E ................................................................... 85
Figure 4.6 – Displaced shape of the 3D FEM of test E-iii at maximum lateral displacement
(magnified 3 times) alongside the plotted stress profile ....................................... 86
Figure 4.7 – Calculated lateral force-displacement response using 3D FEM ............................ 87
Figure 4.8 – Calculated localised wall behaviour using the 3D FEM for test E-iii ................... 88
Figure 4.9 – Calculated localised wall behaviour using the 3D FEM for test F-iii ................... 89
Figure 4.10 – Flexural behaviour of an unbonded PT wall ....................................................... 91
Figure 4.11 – Equivalent wall representation of a beam............................................................ 96
Figure 4.12 – Predicted change in tendon stress using existing equations ................................ 98
Figure 4.13 – Predicted neutral axis (NA) depth at nominal flexural strength ........................ 100
Figure 4.14 – Parameters influencing base rotation ................................................................. 101
Figure 4.15 – Predicted base rotation at nominal flexural strength ......................................... 101
Figure 4.16 – Predicted tendon stress using Eq. 4.11 compared with FEM data .................... 102
Figure 4.17 – Predicted change in tendon stress using Eq. 4.11 for larger ultimate strains .... 103
Figure 4.18 – Predicted tendon stress using Eq. 4.11 compared with experimental data ........ 104
Figure 4.19 – Predicted base shear at nominal flexural strength using Eqns. 4.11 and 4.13 ... 105
Figure 4.20 – Lateral displacement due to rocking .................................................................. 106
Figure 4.21 – Predicted lateral displacement using Eq. 4.16 ................................................... 107
Figure 4.22 – Predicted lateral displacement for higher strain limits using Eq. 4.16 .............. 108
Figure 5.1 – PreWEC wall system [5-1] .................................................................................. 115
Figure 5.2 – Force-displacement design envelope established for PreWEC specimen [5-1] .. 116
Figure 5.3 – U-shaped flexural plate (UFP) [5-9] .................................................................... 117
Figure 5.4 – Typical restraints used during FEM analyses ...................................................... 120
Figure 5.5 – Displacement history used for FEM analyses ..................................................... 120
Figure 5.6 – Dimensions and FEM analysis for SFP-1 ........................................................... 121
Figure 5.9 – Dimensions and FEM analysis for H-1 ............................................................... 124
Figure 5.10 – Dimensions and FEM analysis for H-2 ............................................................. 125
Figure 5.11 – Dimensions and FEM analysis for H-3 ............................................................. 127
Figure 5.12 – Dimensions and FEM analysis for J-Connector ................................................ 128
Figure 5.13 – Dimensions and FEM analysis for O-Connector............................................... 129
Figure 5.14 – Dimensions and FEM analysis for optimised O-Connector .............................. 132
- xiv -
List of Figures
Figure 5.15 – Experimental test setup and instrumentation used for O-Connector tests ......... 134
Figure 5.16 – Displacement history used for connector Tests-A1 & A2 ................................. 135
Figure 5.17 – Measured stress-strain results for tensile coupons ............................................. 136
Figure 5.18 – FEM of the test rig and O-Connectors .............................................................. 136
Figure 5.19 – Measured force-displacement response of Test-A1 .......................................... 137
Figure 5.20 – Out-of-plane buckling of the connectors in Test-A1 at a relative vertical
displacement of 50 mm........................................................................................ 138
Figure 5.21 – Restraining plate used during Test-A2............................................................... 138
Figure 5.22 – Comparison of measured and calculated response of Test-A2 .......................... 139
Figure 5.23 – Comparison of measured and calculated strains for connector in Test-A2 ....... 140
Figure 5.24 – PreWEC specimen test setup ............................................................................. 141
Figure 5.25 – Sketch of O-Connector used for PreWEC test ................................................... 142
Figure 5.26 – Comparison of measured and calculated response of Test-B ............................ 143
Figure 5.27 – Failure of O-Connectors in Test-B..................................................................... 143
Figure 5.28 – Performance of PreWEC wall test specimen ..................................................... 144
Figure 5.29 – Measured responses for PreWEC test ................................................................ 145
Figure 5.30 – Failure of O-Connectors during PreWEC test ................................................... 145
Figure 6.1 – PreWEC wall system ........................................................................................... 150
Figure 6.2 – Confined concrete model proposed by Mander et al. [6-1] ................................. 152
Figure 6.3 – ABAQUS concrete damaged plasticity model [6-3]............................................ 154
Figure 6.4 – Concrete stress-strain input used for Column 7 FEM .......................................... 156
Figure 6.5 – Results for FEM analysis of concrete confinement effects on Column 7 ............ 157
Figure 6.6 – Concrete stress-strain input used for Wall 12 FEM ............................................. 160
Figure 6.7 – Results for FEM analysis of concrete confinement effects on Wall 12 ............... 161
Figure 6.8 – Details of the confined toe in the PreWEC test specimen ................................... 165
Figure 6.9 – Concrete compressive stress-strain definitions for PreWEC ............................... 166
Figure 6.10 – Drawing of PreWEC test setup .......................................................................... 168
Figure 6.11 – Material stress-strain definitions used for the PreWEC FEM............................ 170
Figure 6.12 – Illustrations of the PreWEC FEM assembly ...................................................... 174
Figure 6.13 – Lateral force-displacement response for monotonic trials of PreWEC FEM .... 177
Figure 6.14 – PreWEC FEM at 3% lateral drift (displacements magnified 2×) ...................... 181
Figure 6.15 – Comparison of measured and calculated lateral force-displacement responses for
the PreWEC FEM ................................................................................................ 183
Figure 6.16 – Comparison of measured and calculated hysteresis loop parameters for the
PreWEC FEM ...................................................................................................... 184
Figure 6.17 – Comparison of measured and calculated wall unbonded tendon stresses for the
PreWEC FEM ...................................................................................................... 186
Figure 6.18 – Comparison of measured and calculated neutral axis depth and base rotation
backbones for the PreWEC FEM ........................................................................ 187
Figure 6.19 – Comparison of measured and calculated concrete strain backbones for the
PreWEC FEM ...................................................................................................... 188
Figure 6.20 – Comparison of measured and calculated connector vertical displacement
backbone for PreWEC FEM ................................................................................ 189
- xv -
List of Figures
Figure 6.21 – Over-closure contact definition in ABAQUS [6-3] ........................................... 190

Figure 6.22 – Comparison of measured and calculated lateral force-displacement response for
PreWEC FEM with modified contact properties ................................................ 191
Figure 6.23 – Monotonic lateral force-displacement response of modified PreWEC FEM .... 193
Figure 6.24 – Lateral force-displacement response calculated for PreWEC-A & B ............... 194
Figure 6.25 – Local responses backbones calculated for PreWEC-A & B.............................. 194
Figure 6.26 – Lateral force-displacement and tendon stress response calculated for PreWEC-C
............................................................................................................................. 196
Figure 6.27 – Lateral force-displacement and tendon stress response calculated for PreWEC-D
............................................................................................................................. 197
Figure 6.28 – Lateral force-displacement response calculated for PreWEC-E ....................... 198
Figure 6.29 – Localised response backbones calculated for PreWEC-E ................................. 198
Figure 6.30 – Calculated lateral force-displacement response for PreWEC FEM with 28 and 16
connectors............................................................................................................ 199
Figure 6.31 – Calculated local response backbones for PreWEC FEM with 28 and 16
connectors............................................................................................................ 200
Figure 6.32 – Equivalent viscous damping for PreWEC FEMs with varied number of
connectors............................................................................................................ 201
Figure 6.33 – Equivalent viscous damping for each lateral drift level .................................... 202
Figure 6.34 – Comparison of equivalent viscous damping calculated using Eq. 6.14 and the
PreWEC FEM ..................................................................................................... 202
Figure 6.35 – Cross-section of the PreWEC core wall design ................................................. 204
Figure 6.36 – In-plane response of PreWEC core wall FEM at 3% lateral drift ..................... 205
Figure 6.37 – Lateral force-displacement response for PreWEC core wall FEM ................... 206
Figure 6.38 – Bidirectional response of PreWEC core wall FEM at 3% lateral drift .............. 207
Figure 7.1 – Idealised flag-shape hysteresis response typically assumed for self-centering
systems ................................................................................................................ 214
Figure 7.2 – Dynamic shake-down behaviour expected for seismic load resisting systems ... 217
Figure 7.3 – Two-spring lumped plasticity representation of a self-centering wall system .... 219
Figure 7.4 – Two spring analysis with idealised hysteresis and equal moment resistance ..... 220
Figure 7.5 – Two spring analysis with idealised hysteresis and unbalanced moment resistance
............................................................................................................................. 221
Figure 7.6 – Two spring analysis with realistic hysteresis and equal moment resistance ....... 222
Figure 7.7 – Two spring analysis with realistic hysteresis and unbalanced moment resistance
............................................................................................................................. 222
Figure 7.8 – Seismic performance objectives for buildings [7-20].......................................... 225
Figure 7.9 – Residual drift performance objectives for PreWEC ............................................ 226
Figure 7.10 – Two-spring lumped plasticity model of PreWEC ............................................. 228
Figure 7.11 – Ruaumoko seven spring lumped plasticity model for PreWEC ........................ 230
Figure 7.12 – Hysteresis rules used for PreWEC Ruaumoko model [7-16] ............................ 232
Figure 7.13 – Calculated cyclic response of PreWEC Ruaumoko model ............................... 233
Figure 7.14 – Calculated hysteresis loop parameters for PreWEC Ruaumoko model ............ 234
Figure 7.15 – Graphical representation of parametric model variations.................................. 236
- xvi -
List of Figures
Figure 7.16 – Base moment-rotation behaviour for all walls in the Ptest set ............................ 237
Figure 7.17 – 20% damped displacement response spectrum for Zone 4, soil type Sc, and
seismic hazard level EQ-III ................................................................................. 240
Figure 7.18 – Ruaumoko base moment-rotation behaviour for wall sets Pmore and Pless .......... 241
Figure 7.19 – Unscaled selected earthquake acceleration records ........................................... 243
Figure 7.20 – SEAOC 5% damped acceleration response spectrum for Zone 4 soil type Sc... 244
Figure 7.21 – Example responses from the dynamic analyses of PreWEC walls .................... 247
Figure 7.22 – Maximum drift results from the dynamic analyses of wall set Ptest ................... 249
Figure 7.23 – Residual drift results from the dynamic analyses of wall set Ptest ...................... 250
Figure 7.24 – Maximum drift results for wall set Pmore and Pless, hazard level EQ-III ............. 251
Figure 7.25 – Residual drift results for wall set Pmore and Pless, hazard level EQ-III ................ 252
Figure 7.26 – Effect of asymmetrical response on residual drift for all wall sets .................... 253
Figure 7.27 – Residual drift ratio for wall sets Ptest, Pmore and Pless .......................................... 254
Figure 7.28 – Influence of fundamental period on the residual drift ratio ............................... 255
Figure 7.29 – Influence of ground motion parameters on the residual drift ratio .................... 256
Figure 7.30 – Maximum drift results for wall set Ptest with 1% and 5% viscous damping, EQ-III
............................................................................................................................. 257
Figure 7.31 – Residual drift results for wall set Ptest with 1% and 5% viscous damping, EQ-III
............................................................................................................................. 258
Figure 7.32 – Residual drift ratio for wall set Ptest with 1% and 5% viscous damping ............ 259
Figure 8.1 – Lateral load behaviour of a PT wall ..................................................................... 266
Figure 8.2 − Cast-in-place floor ............................................................................................... 268
Figure 8.3 − Precast floor unit parallel to the wall ................................................................... 269
Figure 8.4 – Floor to shear wall connections detailed in PCI design handbook [8-8] ............. 269
Figure 8.5 – Possible slotted wall-to-floor connection to isolate the floor from the wall uplift
............................................................................................................................. 270
Figure 8.6 − Precast floor with insitu topping bearing on the wall .......................................... 271
Figure 8.7 – Floor to bearing wall connections detailed in PCI design handbook [8-8] .......... 272
Figure 8.8 – Possible isolating wall-to-floor connections suitable for PreWEC wall system .. 273
Figure 8.9 – Four-storey prototype building layout ................................................................. 274
Figure 8.10 – Material stress-strain definitions used in floor FEM ......................................... 277
Figure 8.11 – FEM assembly of 4-storey prototype building .................................................. 278
Figure 8.12 – Elevation view showing the calculated displaced shape of the prototype building
FEM with cast-in-place floor (displacements magnified 3 times) ...................... 280
Figure 8.13 – Calculated principal strain contours of the prototype building FEM with cast-in-
place floor at 3% lateral drift ............................................................................... 280
Figure 8.14 – Calculated principal strain contours of the top cast-in-place floor slab ............. 281
Figure 8.15 – Calculated reinforcing steel strains of the top cast-in-place floor slab .............. 282
Figure 8.16 – Calculated response of the prototype building FEM with cast-in-place floor ... 283
Figure 8.17 – Elevation view showing the calculated displaced shape of the prototype building
FEM with isolated floor diaphragms (displacements magnified 3 times) ........... 284
Figure 8.18 – Calculated response of the prototype building FEM with isolated floors.......... 285
Figure 8.19 – FEM representation of the wall and floor section .............................................. 286
- xvii -
List of Figures
Figure 8.20 – Calculated displaced shape of the floor section FEM at 3% wall drift
(displacements magnified 10 times) .................................................................... 286
Figure 8.21 – Calculated response of the floor section FEM................................................... 287
Figure 8.22 – Lumped plasticity model of four storey prototype building .............................. 288
Figure 8.23 – Cyclic behaviour of the prototype building using the lumped plasticity
Ruaumoko model ................................................................................................ 289
Figure 8.24 – Top lateral drift time history responses for GM-8, hazard level III .................. 292
Figure 8.25 – Maximum interstorey drift results for the prototype building ........................... 293
Figure 8.26 – Residual displacement results for the prototype building ................................. 294
Figure 8.27 – Peak floor acceleration results for the prototype building................................. 295
Figure 8.28 – Peak base shear results for the prototype building ............................................ 296

- xviii -
LIST OF TABLES
Table 3.1 – Details of test walls ................................................................................................. 40
Table 3.2 – Summary of prestressing steel properties ................................................................ 44
Table 3.3 – Details of instrumentation used to measure concrete strains .................................. 51
Table 4.1 – Details of FEM wall data set ................................................................................... 92
Table 5.1 – Summary of connector dimensions and capacities obtained from FEMs ............. 130
Table 5.2 – Effect of O-Connector dimensions on strength and strain demand....................... 131
Table 6.1 – Details of Column 7 tested by Mander et al. [6-4] ................................................ 155
Table 6.2 – Details of Wall 12 tested by from Mander et al. [6-4] .......................................... 159
Table 6.3 – Summary of PreWEC confined region properties ................................................. 165
Table 6.4 – Details of PreWEC test setup ................................................................................ 167
Table 6.5 – Summary of measured PreWEC material properties ............................................ 169
Table 6.6 – Summary of changes to the PreWEC design......................................................... 192
Table 6.7 – Details of PreWEC core wall ................................................................................ 203
Table 7.1 – PreWEC residual drift performance limits ............................................................ 226
Table 7.2 – Element properties for PreWEC Ruaumoko model .............................................. 231
Table 7.3 – Scale factors for full size PreWEC model ............................................................. 237
Table 7.4 – DDBD parameters for the five buildings designed with PreWEC ........................ 240
Table 7.5 – Summary of earthquake ground motions (GM) .................................................... 242
Table 7.6 – Scale factors for earthquake ground motions ........................................................ 245
Table 8.1 – Properties of cast-in-place the wall-to-floor rotational spring .............................. 289
Table 8.2 – Summary of ground motions used for the dynamic analyses ................................ 291
- xix -
List of Tables
- xx -
NOTATION
Acronyms:
ACI = American Concrete Institute
CIP = cast-in-place
DBE = design base earthquake
DDBD = direct displacement based design
DSDM = diaphragm seismic design methodology
ED = energy dissipation
FBD = force based design
FEM = finite element model
GM = ground motion
ISU = Iowa State University
ITG = Innovation Task Group
J-Connector = J-shaped flexural plate
LVDT = linear variable displacement transducer
MCE = maximum credible earthquake
NA = neutral axis
NCREE = National Center for Research on Earthquake Engineering
NZS = New Zealand Standards
O-Connector = Oval shaped flexural plate
PRESSS = Precast Seismic Structural Systems
PreWEC = precast wall with end columns
PT = post-tensioned or post-tensioning
RC = reinforced concrete
RD = residual drift
SDOF = single degree of freedom
SG = strain gauge
SFP = slotted flexural plate
UFP = U-shaped flexural plate
UoA = University of Auckland

- xxi -
Notation
Roman characters:
a = length of the equivalent rectangular stress block
Aideal = area of an ideal elasto-plastic hysteresis loop
Aloop = area enclosed by a single hysteresis loop
A ps = prestressing steel area
As = mild reinforcing steel area
Astrain = area equal to the strain energy for hysteresis cycle peak
b = beam thickness
bw = wall thickness
c = neutral axis depth at nominal flexural strength

d = lateral drift (%)
di = distance from extreme compression fibre to centroid of the ith prestressing
tendon
dn = lateral displacement at he corresponding to nominal flexural strength
d n , fl = flexural displacement at he corresponding to nominal flexural strength
d n , gap = gap opening displacement at he corresponding to nominal flexural strength
dr = residual displacement at the end of excitation

d r ,design = design residual displacement
d r ,max = maximum possible residual displacement
d rr = residual drift ratio (dr / dr, max)

dp = distance from extreme compression fibre to centroid of prestressing tendon
e = eccentricity of load parallel to axis measured from centroid of section

Ec = concrete modulus of elasticity
E ps = prestressing steel modulus of elasticity
E SEC ,c = concrete compression unloading secant stiffness (cyclic hysteresis parameter)
E SEC ,t = concrete tension unloading secant stiffness (cyclic hysteresis parameter)
fc = concrete compressive stress
f c' = concrete compressive strength
- xxii -
Notation
f cc' = confined concrete compressive strength
fl' = effective lateral confining stress
fm = masonry axial stress
f m' = masonry compressive strength
f ps = prestressing steel stress at nominal flexural strength
f pu = ultimate rupture stress of prestressing steel
f py = yield stress of prestressing steel
f se = initial stress in prestressing steel after losses
ft = concrete tensile stress
ft' = concrete tensile strength
f un = stress prior to unloading (cyclic hysteresis parameter)
fu = ultimate stress of mild reinforcing steel
fy = yield stress of mild reinforcing steel
GC = shear modulus of concrete
he = equivalent or effective wall height
hi = height to the ith story (DDBD parameter)
ke = confinement effectiveness ratio
Ke = effective stiffness of the SDOF representation (DDBD parameter)
lp = unbonded tendon length
lw = wall length
L = total beam length

L1 = sum of lengths of loaded spans containing tendon(s) considered
L2 = total length of tendon(s) between anchorages

me = effective mass at he for SDOF representation (DDBD parameter)
mi = lumped mass at the ith story (DDBD parameter)
M con = moment contribution provided by the energy dissipating connectors
Mn = nominal moment resistance
- xxiii -
Notation
MN = moment contribution provided by additional axial load (including self-weight)
M pt = moment contribution provided by the PT steel
M pr = probable moment resistance
Ms = moment contribution provided by energy dissipating reinforcing steel
N = axial load due to wall self-weight and dead and live

n = number of storeys
nc = number of connectors
Sa = spectral acceleration
Sd = spectral displacement
Ti = initial of fundamental period of the structure
Te = effective period of the SDOF representation (DDBD parameter)
Vb = design base shear of structure (DDBD parameter)
Vf = lateral force applied at he corresponding to nominal flexural strength
W e = effective weight at he for SDOF representation (DDBD parameter)
Greek characters:
 = factor to determine average stress in the equivalent rectangular stress block
o = overstrength factor of energy dissipating elements
 = factor to determine length of the equivalent rectangular stress block

 = energy dissipation ratio (for Chapter 6 only)
 f ps = change in tendon stress due to wall deformation
d = design lateral displacement at he for SDOF representation (DDBD parameter)
 Drift = design level interstorey drift (DDBD parameter)
i = design lateral displacement at the ith story (DDBD parameter)
 Flexure = measured wall lateral displacement due to panel flexural deformation
 Gap = measured wall lateral displacement due to gap opening
 Shear = measured wall lateral displacement due to panel shear deformation
c = concrete compressive strain
- xxiv -
Notation
 cc = concrete compressive strain at the confined concrete compressive strength
 co = concrete compressive strain at the unconfined concrete compressive strength
 cu = ultimate concrete compressive strain
 mu = ultimate masonry compressive strain
t = concrete tensile strain
 to = concrete tensile strain at the concrete tensile strength
 un = strain prior to unloading (cyclic hysteresis parameter)
o = strain at zero stress prior to tension load cycle (cyclic hysteresis parameter)
 = factor to represent the proportion of elastic damping acting at design level
displacement (DDBD parameter)
 = moment contribution or self-centering ratio
 = secant stiffness correction factor (for DDBD in Chapter 7)
 = displacement ductility
 = rotation at wall base
p = ratio of prestressed reinforcement
 el = elastic damping
 eq = equivalent viscous damping
 hyst = hysteretic damping
- xxv -
Notation
- xxvi -
Chapter 1
Introduction
1.1 OVERVIEW
Concrete walls are frequently used in building construction worldwide, and can be designed to
resist both vertical gravity loads and lateral forces generated by wind or earthquakes. When
designed correctly, concrete walls provide an efficient and economical lateral force resisting
system that limit lateral deflections due to their high in-plane stiffness. For this reason,
structural concrete walls are often used in buildings as the primary lateral force resisting
system and are considered essential in the design of high rise structures. The use of concrete
walls for low rise or warehouse building construction is also common, where the walls can
additionally function as the exterior facade of the structure. In accordance with the prevailing
ductile design philosophy, a reinforced concrete (RC) wall should resist lateral forces
generated by earthquakes through the formation of a flexural plastic hinge at the base of the
wall. A well detailed plastic hinge provides sufficient ductility without compromising the
strength of the wall. However, cracking and crushing of the concrete in the plastic hinge
region, and yielding of the longitudinal reinforcing steel, results in significant and often
irreparable damage to the wall following a large earthquake.
-1-
Chapter 1 Introduction
Prefabricated or precast concrete components are becoming increasingly common, where the
concrete members are cast at an offsite location and transported to the constructing site to be
erected and connected together. In New Zealand, the construction of a reinforced concrete
building would typically include extensive use of precast elements, including flooring units,
beams, columns, and wall panels. Precast concrete components have several advantages in
building construction, including:
 high quality finish and lower construction tolerances,
 improved architectural finishes,
 reduced construction time,
 reduced onsite labour,
 optimised use of materials,
 longer spans when prestressing is used,
 use of advanced technology.
Particularly common in precast concrete components, prestressing involves the application of

an initial compression force to the member using high strength steel bars or prestressing strand.
Depending on when the steel tendons are stressed, a prestressed concrete member is described
as either pre-tensioned or post-tensioned. For a pre-tensioned member, the prestressing
tendons are stressed before the concrete is poured around them, and the prestress force is
released onto the member when the concrete has gained sufficient strength. For a post-
tensioned member, the prestressing tendons are inserted into ducts that are cast into the
member, after which the tendons are stressed and anchoring at each end. Once the tendons are
stressed and anchored, the ducts in a post-tensioned member can be filled with grout or the
tendons can be left unbonded over their length. Prestressing concrete ensures that a material
with relatively high compressive strength, but substantially lower tensile strength, remains in
compression when subjected to service loads. This precompression prevents cracks from
opening up, which improves durability and reduces the deflection of the member, allowing for
longer spans and smaller member sizes to be used.
Self-centering precast concrete walls maximise the advantages of structural concrete walls by
using both precast and prestressing technology to achieve superior seismic performance. The
simplest form of a self-centering wall consists of a precast concrete panel with unbonded post-
tensioned tendons that are anchored between the top of the wall and the foundation. When
subjected to a lateral force the behaviour of a self-centering wall is characterised by a single
-2-
1.1 Overview
crack at the wall base, as shown in Figure 1.1, with the lateral strength provided by the
prestressing tendons. As uplift and rocking occurs at the wall base, the unbonded prestressing
tendons are elongated, with the strain evenly distributed along the length of the tendon. This
evenly distributed elongation allows the tendons to be designed to remain in their elastic state
to provide a restoring force that can re-center the wall back to its original position when the
lateral force is removed. As a result, a self-centering precast concrete wall can safely resist the
lateral forces that are generated during an earthquake with minimal structural damage
occurring, resulting in superior seismic performance when compared to the extensive damage
that would be expected for a monolithic reinforced concrete wall.
PT tendon
Wall
Foundation
Figure 1.1 – Lateral load behaviour of an unbonded post-tensioned (PT) concrete wall
A typical reinforced concrete member exhibits a degrading stiffness hysteresis behaviour, as

shown in Figure 1.2a, which provides significant ductility and energy dissipation during an
earthquake. In contrast, the response of a self-centering member with unbonded post-
tensioning is predominantly bi-linear elastic, as shown in Figure 1.2b. Systems with a bi-linear
elastic hysteresis behaviour typically lack the ability to dissipate the energy imparted to the
structure during a large earthquake, which may result in large lateral displacements and high
accelerations. To improve the performance of self-centering structures during large
earthquakes, walls with unbonded post-tensioning are often coupled with supplementary
energy dissipating elements. The combination of unbonded post-tensioning and energy
dissipating elements can be idealised by a flag-shape hysteresis behaviour, as shown in Figure
1.2c, where self-centering behaviour is retained with added hysteretic energy dissipation.
-3-
(a) Traditional RC (b) Bi-linear elastic (c) Flag-shaped
Figure 1.2 – Hysteresis behaviour of traditional and self-centering systems
1.2 RESEARCH MOTIVATION AND OBJECTIVES

Extensive research into the design of self-centering precast concrete structural components has
been completed over the last two decades, but implementation of these components into
building practice has been limited. In particular there are few examples of self-centering
precast concrete wall systems that have been implemented into constructed buildings.
Limitations of current self-centering wall systems that may have contributed to this lack of
implementation include:
 complex design procedures that prevent the use of self-centering systems for simple
building designs,
 a reduction in moment capacity when compared to traditional reinforced concrete walls,
which leads to larger and less economical walls,
 failure of the wall to provide a fully seismic resilient system that requires only minor
repair following a large earthquake,
 uncertainty in the behaviour of the complete building and the interaction between self-
centering components and other structural and non-structural elements.
The main objective of this study was to overcome these limitations and allow for widespread
implementation of self-centering precast concrete wall systems. This objective included
investigation of wall systems that are suitable for regions with low seismicity as well as wall
systems that are more suitable for regions of moderate to high seismicity.
An individual precast concrete wall panel with unbonded post-tensioning may be suitable for
use in regions of low seismicity and in particular for low-rise buildings. Previous research that
-4-
1.2 Research Motivation and Objectives
investigated individual post-tensioned walls focused on large lateral displacement demands

that would be required for regions of high seismicity, and so confinement reinforcement was
always placed in the wall toe where large compressive strains are developed. For limited
ductility design in regions of low seismicity, confinement reinforcement in the wall toe may
not be necessary due to the lower lateral drift demand, and the walls could be designed using
the nominal flexural strength at the ultimate limit state.
The use of self-centering wall system in regions with moderate to high seismicity requires
additional energy dissipating elements to be incorporated. The PreWEC wall system,
conceived by researchers at Iowa State University, consists of a precast concrete wall that is
connected to end columns using replaceable energy dissipating connectors. The PreWEC
system overcomes several of the limitations that are associated with previously developed self-
centering wall systems, and was considered the most suitable solution for regions with
moderate to high seismicity.
1.2.1 Objectives
To address the main objective of overcoming the aforementioned limitations, three phases of
study were conducted. The objectives of each of these three phases included:
 Solutions for regions with low seismicity:

o Investigate the behaviour of individual PT wall panels with no specific
confinement reinforcement in the wall toe;
o Develop simplified equations that would allow individual PT walls to be
designed within the scope of current concrete design standards;
 Solutions for regions with moderate to high seismicity:
o Investigate suitable energy dissipating connectors for the PreWEC wall system;
o Develop a detailed finite element model to investigate the behaviour of the
PreWEC wall system;
 System response:
o Investigate the residual drifts in self-centering systems and how appropriate
performance targets can be included during the design process;
o Extend the investigation of the wall system to consider the seismic resilience of
the entire building.
-5-
1.3 THESIS OUTLINE

This thesis is separated into chapters which each describe a distinct step in the research. First,
individual post-tensioned wall panels that are suitable for regions of low seismicity were
investigated, followed by analysis of the PreWEC wall system which is more suitable for
regions of moderate or high seismicity. Lastly, the expected response of the entire building
during an earthquake was considered, focusing on the self-centering behaviour and the
interaction between the wall and the surrounding structural elements.
This thesis is organised into the following chapters:
Literature review: Chapter 2 provides a review of previous research related to the

development of self-centering wall systems. As well as discussing alternative wall systems
that have been developed, a summary of experimental and analytical studies that have
investigated the seismic response of both the wall and building are presented.
Testing of post-tensioned wall panels: Chapter 3 presents the experimental results from the
testing of post-tensioned concrete wall panels that had no additional energy dissipating
elements. The main focus of the 32 tests completed was to investigate the strain behaviour in
the wall toe when no specific confinement reinforcement was used.
Analysis and design of post-tensioned wall panels: To supplement the experimental results
from Chapter 3, finite element analysis was also conducted to investigate the behaviour of
post-tensioned concrete walls. The validation of a 3D finite element model is outlined in
Chapter 4, followed by the development of a simple set of equations that can be used to predict
the wall response at the ultimate limit state.
PreWEC connector design: The PreWEC wall system is introduced in Chapter 5. The
design of a suitable energy dissipating connector for the PreWEC system is presented,
consisting of a series of finite element models. Chapter 5 concludes with a summary of a
companion experimental study that was conducted to validate both the connector performance
and the entire PreWEC wall concept.
Finite element modelling of the PreWEC system: Chapter 6 presents the development of a
finite element model that was used to calculate the response of a PreWEC wall subjected to the
same reverse cyclic loading that was used during the experimental test reported in Chapter 5.
-6-
1.3 Thesis Outline
The model was an extension of the single wall model that was developed in Chapter 4, and
included investigation of how to accurately capture the behaviour of the confined concrete in
the wall toe. After comparison with the experimental results, the PreWEC finite element
model was used to investigate the effect of several important design parameters, including the
amounts of energy dissipating connectors and prestressing steel.
Analysis of residual drifts: The procedures that are currently used to ensure that a self-
centering response is achieved are discussed in Chapter 7. The idealised flag-shaped hysteresis
behaviour is not accurate for real self-centering systems, and a series of time-history analyses
were performed to develop a design method that can be used to ensure that residual drift limits
are satisfied by a realistic self-centering building following an earthquake.
Analysis of a self-centering building: To ensure a satisfactory seismic response is achieved

for a building with a self-centering wall, the interaction between the wall and the surrounding
structural elements must be considered. Chapter 8 discusses the interaction between the wall
and the floor diaphragms, and several different connection details are presented. The results
from finite element and time-history analyses were used to discuss the effect of the connection
detail on the seismic response of a prototype building.
Conclusions: Chapter 9 summarises the main conclusions of the thesis and provides
recommendations for future research.
-7-
-8-
Chapter 2
Literature Review
An extensive review of literature related to self-centering precast concrete walls was

undertaken as part of this study. First, the performance of concrete walls and precast elements
during historical earthquakes is described and the motivation behind the development of self-
centering components is discussed. An examination of the development of self-centering wall
systems is then outlined, along with the current code provisions for such systems and a
summary of several experimental investigations that considered entire building response.
Following this, both experimental and analytical studies investigating the dynamic behaviour
of self-centering walls are detailed. Lastly, application of self-centering precast concrete
systems in a number of buildings constructed around the world is described.
2.1 CONCRETE WALLS

Concrete walls provide excellent seismic resistance and are often used as the primary lateral
force resisting system in a building. During historical earthquakes, structural walls were found
to exhibit superior seismic performance compared with other structural components [2-1].
Structural walls can be constructed from either cast-in-place or precast concrete. Precast
construction has additional benefits, including offsite fabrication, improved quality, and speed
-9-
Chapter 2 Literature Review
of construction. However, the application of precast technology in seismic regions was

initially limited due to a lack of research and subsequent limits imposed by design codes.
When investigating the performance of precast structures during previous earthquakes, precast
concrete structural wall systems are found to perform well. Several precast parking structures
suffered significant damage during the 1994 Northridge earthquake, but this damage was
caused by incompatibilities between the seismic and gravity resisting components as well as
inadequate floor diaphragms. No damage to the structural wall systems was observed [2-2].
Additionally, amongst the devastation of the 1995 Kobe earthquake, precast concrete buildings
that used structural wall systems performed well with minimal damage observed.
The use of precast concrete in seismic regions has been restricted by several limitations. The
poor performance of many precast frame structures in earthquakes, mainly due to poor
construction practice and inadequate connection details, led to the practice of “cast-in-place
emulation” [2-3, 2-4]. However, this reduced the economical advantage of precast
construction, with complicated joint details often requiring “wet joints”.
2.2 ROCKING BEHAVIOUR

Evidence suggests that the use of a rocking mechanism for seismic protection dates back to the
Greek era, where segmental construction of columns led to gap opening at the joints when
subjected to lateral shaking during an earthquake [2-5]. However, only in the last 50 years has
this rocking behaviour been investigated by modern structural engineers. Housner [2-6] is
credited as the first significant researcher to study and model rocking behaviour as he
attempted to explain why many tall slender structures, such as water tanks, had survived the
1960 Chilean earthquake. Predicting the response of a rocking block is complicated by the
reduction in velocity that occurs when impact takes place at the rocking interface, referred to as
contact or radiation damping. This reduction in velocity is poorly predicted when using
complex finite element software, and the use of mathematical expressions derived from the
equation of motion of a simplified model remains the most accurate method for predicting the
response of a rocking block [2-7, 2-8]. Despite the modelling complexities, rocking
mechanisms have been implemented into several structures as a form of base isolation for
seismic protection. Two examples include the South Rangitikei rail viaduct in New Zealand,
which was designed to allow for rocking of the piers on the pile cap [2-9] and the northern
- 10 -
2.3 Self-centering Concrete Systems
viaduct approach to the Lions Gate bridge in Vancouver Canada which was retrofitted to allow
for pier rocking, which resulted in reduced the base shear demand [2-10].
2.3 SELF-CENTERING CONCRETE SYSTEMS

2.3.1 PRESSS
To overcome limitations which restricted the use of precast construction in seismic regions, a
joint US-Japan research program entitled PRESSS (PREcast Seismic Structural Systems) was
initiated in the early 1990’s [2-11]. Because the practice of cast-in-place emulation was not
common in the US, research institutions and industry partners across the US focused on the
development of new technologies which would improving the seismic behaviour of structures
while retaining and maximising the benefits of precast construction. Following initial
experimental tests conducted at the laboratory of the National Institute of Standards and
Technology (NIST) [2-12] and an analytical investigation by Priestley and Tao [2-13], the use
of unbonded post-tensioning to tie precast components together became the highlight of the
US-PRESSS program. As well as tying the components together and providing moment
resistance, the unbonded tendons are designed to remain elastic during an earthquake so that
they provide a restoring force to bring the structure back to its original position. This leads to
the desired self-centering behaviour which was described in Chapter 1.
The PRESSS research team developed several self-centering prestressed frame connections
that were found during experimental testing to perform better than traditional cast-in-place
connections [2-14-16]. Additionally, researchers at Lehigh University began analytically
investigating the lateral force behaviour of unbonded post-tensioned concrete walls [2-17, 2-
18]. The final phase of the PRESSS research program involved the design, analysis and
experimental testing of a 60% scale prototype building, shown in Figure 2.1 [2-19, 2-20]. This
test building included many of the frame, wall, and floor diaphragm technologies that were
developed, including the jointed wall system which is discussed in more detail below.
- 11 -
Figure 2.1 – PRESSS test building [2-20]
2.3.2 Individual Walls

The simplest representation of a self-centering wall is a single precast concrete panel connected
to its foundation using unbonded post-tensioning. Researchers at Lehigh University conducted
a series of experimental tests on wall panels with unbonded post-tensioning to assess their
monotonic and cyclic lateral force behaviour [2-21, 2-22]. The test walls behaved as expected,
with flexural deformation concentrated at a single crack that opened up at the wall base.
Confinement reinforcement was provided in the wall toe to ensure a stable response when large
compressive strains developed and as a result the walls were able to maintain capacity up to
between 3% and 6% lateral drift. A typical measured lateral force-displacement response from
one of the tests is shown in Figure 2.2. Interestingly, the measured response does not follow
the idealised bilinear-elastic hysteresis rule, which is often assumed for members with only
unbonded reinforcement. Instead, the tests indicated a significant reduction in stiffness, and
small amounts of hysteretic energy dissipation. Perez et al. [2-21] primarily attributed this
behaviour to irrecoverable inelastic strains and spalling of cover concrete in the wall toe, as
well as yielding and subsequent loss of initial prestress in the unbonded tendons. Perez et al.
also calculated the behaviour of the test walls with reasonable accuracy using both simplified
equations and a fibre based analytical model [2-23].
- 12 -
Figure 2.2 – Lateral force-displacement response for Lehigh tests [2-22]
Self-centering prestressed concrete masonry walls have also been investigated extensively at
the University of Auckland. Laursen and Ingham began investigating the in-plane behaviour of
post-tensioned masonry walls, conducting a series of experimental tests to verify the expected
behaviour [2-24-26]. In general, masonry walls were found to behave in a similar manor to
precast concrete walls, with flexural deformation concentrated at a single crack which opened
up at the wall base. Following this, Wight et al. continued investigating the in-plane lateral
behaviour of post-tensioned masonry walls, conducting both pseudo-static and dynamic shake-
table tests [2-27, 2-28]. Wight also used finite element modelling to validate the performance
and refine the design procedures for such walls [2-29]. Wight’s study concluded with the
construction of New Zealand’s first post-tensioned concrete masonry house [2-30].
2.3.3 Wall Systems

Self-centering precast concrete walls utilising unbonded post-tensioning are excellent structural
elements, able to resist lateral forces with minimal structural damage resulting. However,
because the response is predominantly elastic the walls lack the ability to dissipate large
amounts of energy which may be imparted to the structure during an earthquake. This results
in large lateral displacements and high accelerations, which characterise poor seismic
performance. To improve the performance of the wall and structure during large earthquakes,
walls with unbonded post-tensioning are often coupled with supplementary energy dissipating
elements. Several self-centering wall systems that are suitable for application in areas with
- 13 -
high seismicity have been developed including jointed walls, hybrid walls, hybrid coupled
walls, and other variations.
2.3.3.1 Jointed walls

The jointed wall system, shown in Figure 2.3, was developed and tested during the PRESSS
research program discussed earlier. The system uses two or more precast concrete panels
positioned side by side, all of which are post-tensioned to the foundation using unbonded
tendons. Special shear connectors are placed along the vertical joints between the panels.
When subjected to a lateral force, the walls behave in the same manner as an individual wall
panel, with inelastic deformation concentrated at a single crack which opens up at the wall
base. As the walls rotate and the base crack causes uplift, a relative vertical displacement
occurs between adjoining panels, forcing the connectors to undergo large inelastic
deformations. Through this hysteretic action the connectors are designed to dissipate large
amounts of energy and are the primary energy dissipating source for the wall system. As with
individual walls, the post-tensioning tendons are designed to remain elastic and provide a
restoring force to self-center the structure.
Figure 2.3 – Jointed wall system [2-31]
During testing of the PRESSS building the jointed wall, which consisted of two precast
concrete panels and U-shaped flexural plate (UFP) connectors, displayed excellent seismic
- 14 -
resistance [2-20]. The measured base moment-lateral displacement response for the wall
direction of the building is reproduced in Figure 2.4. The self-centering behaviour from the
post-tensioning tendons and hysteretic energy dissipation from the UFP connectors can be
observed and the structure returned to negligible residual displacement at the completion of the
pseudo-dynamic testing. Lateral drifts were limited to 1.85% at the design level earthquake
loading and minimal damage was observed, with only minor spalling of the cover concrete in
the compressed toe of the wall panels. Design guidelines for structural systems developed and
validated during the PRESSS research program were published by Stanton and Nakaki [2-32].
Figure 2.4 – Moment-displacement response for PRESSS jointed wall system [2-20]
Following the PRESSS program, researchers at Iowa State University conducted extensive
analysis of the jointed wall system behaviour, which culminated in the production of refined
design procedures [2-31, 2-33]. A simplified analysis procedure was developed by Aaleti and
Sritharan [2-34] to predict the monotonic lateral force behaviour of unbonded post-tensioned
concrete walls. The procedure assumes rigid body rotation of the wall while ensuring
equilibrium and compatibility between the wall and the anchorages of the unbonded tendons.
2.3.3.2 Hybrid walls

Following the success of hybrid beam-column connections which were tested during the
PRESSS program, researchers began to develop a similar concept for precast concrete walls.
The hybrid wall system, shown in Figure 2.5, uses a combination of unbonded post-tensioning
and traditional mild steel reinforcing bars at the wall base. When subjected to a lateral force,
- 15 -
the wall behaves in the same manner as an individual unbonded post-tensioned precast
concrete panel, with a single crack opening up at the wall base and the unbonded tendons
elongating between their anchors. The mild steel reinforcing bars at the wall base undergo
tension and compression yielding, contributing to the moment resistance and providing a
source for energy dissipation during earthquake excitation. The bars are cast into the
foundation and then inserted into grouted ducts in the wall panel, but are typically debonded
over a short length to reduce the strain demand and prevent fracture of the mild steel when
subjected to the design level earthquake loading. By ensuring that the moment contribution
from the unbonded post-tensioning exceeds that of the mild steel, self-centering of the system
can be maintained.
Figure 2.5 – Hybrid PT wall concept [2-35]
Holden and Restrepo [2-36] tested a hybrid post-tensioned wall and compared the response to
that of a conventional precast concrete wall that followed the “cast-in-place emulation” design
philosophy. Both systems exhibited a stable response when subjected to reverse cyclic lateral
loading. The conventional wall dissipated energy through the formation of a plastic hinge and
subsequent nonlinear behaviour of the reinforcing steel and concrete, which resulted in
significant damage. In contrast, the hybrid wall system was able to achieved lateral drifts of
2.5% with only minimal structural damage resulting. It was observed that the energy
dissipation of the hybrid wall was significantly lower than the traditionally reinforced wall,
mainly because of the small number of energy dissipating mild steel bars used. Additionally,
the reinforcing steel in the hybrid wall panel was designed using strut and tie methods which
resulted in a significant reduction in the total amount of reinforcing steel used.
- 16 -
Restrepo and Rahman [2-37] tested a further three wall panels, two of which used the hybrid
wall concept with mild steel reinforcing bars at the wall-to-foundation interface. The half-scale
wall specimens were 4 m high and the two hybrid walls were designed with different amounts
of energy dissipating bars. The units were tested under pseudo-static reverse cyclic lateral
loading and demonstrated an excellent and stable response up to lateral drifts of 3-4%, as
shown in Figure 2.6. The hysteresis loops displayed the characteristic flag-shape, with
significant energy dissipation from the mild-steel bars while the unbonded post-tensioning
tendons self-centered the system, with negligible residual drift observed during testing.
Additionally, only minor repairable damage was observed in the wall toe, which included
spalling of the cover concrete. Confinement to the wall toe ensured that the concrete
maintained its strength when subjected to high strain demand, preventing any significant
crushing or loss of strength.
Figure 2.6 – Lateral force-drift response for hybrid wall test [2-37]
A research program is currently underway at the University of Notre Dame to further validate
the seismic performance of the hybrid wall system and develop appropriate design guidelines
for design code implementation [2-35].
2.3.3.3 Hybrid coupling beams

Because of architectural requirements for penetrations in structural walls to allow for windows
and doors, coupled wall systems are common in building construction where adjacent walls are
connected with short coupling beams [2-38]. These coupling beams are subjected to extremely
- 17 -
high ductility demand during earthquakes, which typically results in extensive damage and
makes them difficult to detail. Kurama et al. [2-39] have investigated the use of precast
concrete coupling beams with unbonded post-tensioning to reduce the damage experienced by
these members. The coupling beams are anchored to the adjacent walls using unbonded post-
tensioning and utilise steel angles for added energy dissipation. The concept has been
validated through experimental testing, with specimens demonstrating excellent strength,
stiffness, ductility, energy dissipation and self-centering [2-40].
2.3.3.4 Other sources of energy dissipation

Many variations of these proposed self-centering wall systems have also been investigated.
The use of friction dampers as energy dissipating elements for a jointed wall system was
analysed by Kurama [2-41]. Friction based energy dissipating elements have an advantage
over yielding steel connectors as they do not require replacement following a large earthquake.
However, friction elements require increased supervision to ensure that they are installed and
maintained correctly. Another jointed wall concept that is suitable for warehouse buildings
was developed at the University of Canterbury [2-42, 2-43]. In this case the silicon based seal
between the wall panels was relied on for energy dissipation. Yet another option for energy
dissipating elements is the use of viscous dampers. This has been investigated by Kurama [2-
44] as well as researchers at the University of Canterbury [2-45]. Although expensive, viscous
dampers are useful to limit displacements during earthquakes with a high velocity pulse.
2.3.3.5 Codification
In recognition of the potential use of self-centering concrete walls, the American Concrete
Institute (ACI) set up Innovation Task Group 5 (ITG-5) to develop appropriate design code
provisions. The task group has since released two documents. ITG-5.1 [2-46] outlines
acceptance criteria for special unbonded post-tensioned precast concrete walls based on
experimental validation. The document covers appropriate construction techniques, material
properties, design assumptions, and performance targets. Two separate experimental tests are
required to successfully validate a new wall system and confirm that it meets these
requirements. Additionally, ITG-5.2 [2-47] outlines specific design provision for unbonded
post-tensioned walls, with design procedures suitable for jointed and hybrid wall systems based
on experimental and analytical research. When developed according to these documents, as
well as ACI 318-08 [2-48], a wall system can be used in seismic design practice in the US.
- 18 -
Appropriate design provisions for the use of unbonded post-tensioned precast concrete
members have also been introduced into the New Zealand concrete design standard.
Appendix B of NZS 3101-06 [2-49] details additional, or modified, provisions that must be
accounted for when detailing jointed precast components. Additionally, more specific design
guidelines have been published by the New Zealand Concrete Society following a recent
seminar series [2-50].
2.3.4 Building Response

The primary advantage of self-centering systems is their ability to resist seismic forces with
minimal structural damage resulting. The advantage of the wall is lost when significant
damage is caused to other structural components, such as the gravity framing system and floor
diaphragms. Because of the unique behaviour of self-centering walls, their interaction with the
surrounding structure is also unique. For example, as the wall uplifts and rocks, there is a
displacement incompatibility between the wall and rigid floor diaphragms. Investigation into
the entire building response, including interaction between components, is limited due to the
costly nature of these large scale experiments. However, a small number of studies have
considered interaction between the wall system and floor diaphragms.
2.3.4.1 PRESSS building

As explained earlier, the PRESSS research program concluded with experimental testing of a
five storey precast concrete building, which included a self-centering jointed wall system, four
precast moment resisting frame systems and two precast floor systems [2-20]. During this test
the interactions between some of the structural elements was studied. Two different precast
flooring systems were used: pretopped double T’s placed parallel to the jointed wall and
hollow-core units with an in-situ topping spanning between the wall and the gravity frame. As
the main focus of the test was to validate the performance of the lateral force resisting systems,
the interactions between the wall system and the floors were isolated using special details.
These wall-to-floor connection details were not considered appropriate for use in actual design
and construction because of the high cost and difficulty of installation. The double T’s were
attached to the center of each wall panel with an expensive connector which allowed the lateral
inertia force to be transferred to the wall, while allowing the wall panels to rotate free of the
double T units. Additionally, the hollow-core units were seated on steel channels bolted to the
sides of the wall panels and inside faces of perimeter columns located in the plane of the walls.
- 19 -
The overall behaviour of the test building was satisfactory, with lateral inertia forces
transferred to either the wall or frame moment resisting components. The wall-to-floor
connections performed well during testing and successfully isolated the floor diaphragms from
any damage that might have been caused by the displacement incompatibility between the wall
and floor. However, interaction between structural components may explain the measured
moment-displacement response in the wall direction, shown previously in Figure 2.4. The
hysteresis behaviour deviates from the idealised “flag-shape” with some residual displacements
when the lateral force was released to zero. This could be caused by the behaviour of the
gravity framing system and connections between components which exhibited some damage.
2.3.4.2 Mexico test

Seismic testing was also conducted on a two storey precast concrete building at the National
University of Mexico [2-51]. Unlike the PRESSS test, the precast concrete test building was
constructed using cast-in-place emulation rather than self-centering technology. In this case,
the wall-to-floor connection was detailed with precast double T’s with an in-situ topping,
which provided a semi-rigid joint between the wall and floor. During the test, extensive
damage occurred at the wall-to-floor connection with significant concrete cracking and
crushing, and yielding of the reinforcing steel that tied the wall into the topping.
2.3.4.3 Australian test

As part of a study into the application of self-centering precast concrete walls for low-rise
construction, Wilson et al. [2-52] conducted a series of experimental tests to investigate the
behaviour of typical wall-to-floor connections. The wall-to-floor connection was deemed
critical to building performance because the rigid floor diaphragms would resist rocking of the
wall panels. The typical connection detail used short reinforcing bars to tie the wall to the
topping concrete of the precast flooring units. This detail was found to perform poorly, with
the reinforcing bars breaking out from the topping, which resulted in significant loss of
strength in the connection. This failure mode limited the lateral drift of the system
significantly. To improve the detail the reinforcing steel ties were cast deeper into the
hollowcore flooring, providing a more rigid connection. The performance of the connection
was improved substantially, with a stable response observed up to 1.5% lateral drift. However,
failure of the wall-to-floor connection was still the limiting factor and this is not considered a
desirable failure mechanism during an earthquake due to the brittle behaviour.
- 20 -
2.3.4.4 DSDM
Most recently, as part of a large collaborative project to establish a diaphragm seismic design
methodology (DSDM), shake-table testing was conducted on a prototype precast concrete
parking structure at the University of California at San Diego (UCSD) [2-53]. The test
building, shown in Figure 2.7, was constructed using precast column and beam elements, three
precast flooring systems and two hybrid self-centering walls. The floor diaphragms consisted
of hollow-core with in-situ topping, double T’s with in-situ topping, and pretopped double T’s.
Although this structure was primarily designed to observe the behaviour of the precast floor
diaphragms, the interaction between the wall and floor diaphragm needed to be considered. In
this case researchers decided to isolate the floors from the vertical uplift of the wall by using
slotted insert connectors designed by JVI Inc. These slotted connectors were expected to
transfer the horizontal inertia forces from the floors to the wall, while allowing the wall to
uplift independently from the floor. This prevented any damage occurring to the floor system
due to the wall uplift, allowing researchers to focus on the in-plane behaviour of the
diaphragms. Preliminary results confirmed that these connectors performed well under small
intensity earthquake loads, but failed when the building was subjected to maximum considered
input motions. Failure of the connector was attributed to the uplift of the wall panel exceeding
the available stroke of the slot. No damage was observed in the floor diaphragms adjacent to
the walls and the hybrid walls behaved as expected from the pseudo-static tests reported
earlier.
Figure 2.7 – Shake-table test of DDSM precast building at San Diego
- 21 -
2.3.4.5 Moment frames

Similar to wall systems, self-centering moment frames also experience displacement
incompatibility issues with the floor diaphragm. The rigid floor must accommodate gap
opening at the beams-to-column interfaces. Solutions to this problem are being investigated
[2-54-56], however they cannot be applied to the wall-to-floor connection since the
mechanisms are different.
2.4 DYNAMIC BEHAVIOUR

Understanding the dynamic behaviour of self-centering wall systems is critical to assessing
their seismic performance. A limited number of true dynamic tests have been carried out, in
addition to many studies that utilise nonlinear time history analysis to simulate the response of
self-centering walls to earthquake ground motions.
2.4.1 Shake-table Experiments

Wight et al. [2-27, 2-28] carried out a series of shake-table tests to evaluate the dynamic
performance of post-tensioned concrete masonry walls. Rectangular walls with and without
openings were subjected to a suite of historical earthquake records. All walls were observed to
perform satisfactorily, with a rocking dominated response and only minimal and repairable
damage occurred. The walls achieved large lateral displacement capacities up to 1.9% and
returned to negligible residual displacement at the conclusion of each test. When compared
with comparable walls tested using pseudo-static reverse cyclic loading, a similar force-
displacement response and observed damage was recorded for each of the dynamic tests.
Shake-table testing of several self-centering precast concrete walls was recently carried out by
Marriot et al. [2-45]. In total 16 tests were performed on four post-tensioned walls with
different energy dissipating schemes, including contact damping alone, hysteretic tension and
compression yielding bars, viscous dampers, and combinations of the above. Test results
showed that peak lateral displacements were below the targets for design level and maximum
expected earthquake excitation except for the wall relying only on contact damping. This
highlights the current uncertainty associated with the evaluation of contact or radiation
damping. Observed damage was insignificant for all test specimens.
- 22 -
2.4 Dynamic Behaviour
As described earlier, the three storey precast concrete building tested on the shake-table at San
Diego included two hybrid self-centering precast concrete walls [2-53]. Although the main
objective of the test was to observe the diaphragm response, behaviour of the walls was also
documented. Overall the hybrid walls provided excellent lateral resistance during earthquake
shaking but two problems associated with the walls arose during testing. First, failure of the
prestressing tendons was observed at the anchor. This was attributed to the strand and wedge
type anchorage used, the stressing technique, and the large number of load cycles. Second, one
of the mild-steel energy dissipating bars fractured during design level excitation. The exact
reason for this failure is not given, but because the bar was located out of view, inside the wall
panel, the failure was not identified prior to the subsequent test. This caused an undesirable
wall response with larger displacements than expected, which contributed to the failure of
several other components during the maximum considered earthquake.
Cheng and Lu [2-57] performed a series of shake-table tests on self-centering bridge columns
with unbonded tendons. The main objective of the tests was to investigate the contact or
radiation damping, and the measured damping during the test was higher than all of the current
proposed predictive equations.
2.4.2 Time-history Analysis

2.4.2.1 Hysteresis behaviour
Priestley and Tao [2-13] investigated the hysteretic behaviour of self-centering systems during
the conception of the prestressed frame with unbonded post-tensioning. Through analysis of
the connection mechanics, Priestley and Tao established that the force-displacement response
followed a flag-shape (originally referred to as bilinear degrading) hysteresis rule. A series of
single degree of freedom inelastic dynamic analyses were then conducted to compare this
behaviour with existing hysteresis types which included elastic, bilinear elastic and bilinear
elasto-plastic. Analyses were conducted using several earthquake records and a range of initial
periods from 0.4s to 2.0s. Based on the analysis results, Priestley and Tao concluded that the
ductility demand for structures with partially unbonded tendons would be no greater than that
of fully bonded tendons where prestress degrades as a consequence of inelastic strains.
Additionally, although a higher displacement ductility demand can be expected for short period
structures, the response of medium to long period structures with debonded tendons should not
exceed that of an equivalent monolithic reinforced concrete frame.
- 23 -
The seismic response of single-degree-of-freedom self-centering structures with flag-shaped

hysteretic behaviour was studied by Christopoulos et al. [2-58]. The investigation was focused
on steel self-centering systems but the flag-shape hysteresis response is characteristic of
concrete structures also. A comprehensive parametric study was conducted to compare the
seismic response of a system exhibiting flag shape hysteresis behaviour with the corresponding
response of a system having a bilinear elasto-plastic hysteresis behaviour that represented
traditional steel moment resisting frame connections. The properties of the flag-shape rule
were varied, including initial stiffness, post-yield stiffness and energy dissipation coefficient,
and natural periods ranging from 0.1s to 2s were selected to represent typical steel building
structures with different strength levels. A suite of 20 historical earthquakes were selected and
scaled to match the response spectrum for high seismic regions in California with a 10%
chance of exceedence in 50 years. The performance of the structures was evaluated based on
displacement ductility demand, maximum acceleration, absorbed energy and residual
displacement. Christopoulos et al. concluded that for any structure with a given initial period
and strength, an appropriate flag-shape system could be designed to match or exceed the
performance of a traditional elasto-plastic system. In general, it was found that for short period
structures, increasing the post-yield stiffness has a greater effect on reducing displacement
ductility demand, as opposed to increasing the energy dissipation coefficient. The opposite
was true of structures with long periods. Additionally, the flag-shape rule typically resulted in
higher accelerations when compared to elasto-plastic systems. Because of the idealised flag-
shape hysteresis, no residual displacement was predicted for all self-centering systems which
had energy dissipation coefficients equal to 1.0 or less.
In a further investigation, Christopoulos et al. [2-59] conducted a performance-based

assessment of structures that included self-centering systems. Inelastic dynamic analysis of
single-degree-of-freedom systems were conducted with three different hysteresis responses
considered, including bilinear elasto-plastic to represent a steel member, takeda degrading
stiffness to represent a reinforced concrete member, and flag-shape to represent a self-centering
system. The focus of the assessment was on residual drifts, which is a performance factor that
is often overlooked and that is one of the main advantages of self-centering systems. Four
prototype structures ranging from 4 to 20 stories were designed and subjected to the same suite
of historical earthquake records as used in the earlier study [2-58]. The later study concluded
that the residual displacements were a function of the hysteresis rule, the hysteretic parameters
- 24 -
including post-yield stiffness, and the intensity of the seismic input. With performance based
design objectives becoming increasingly common, residual displacements should be
considered an important design criterion that is critical for many traditional systems with fat
hysteresis loops. Self-centering systems were found to sustain similar displacement ductility
demands but negligible residual displacements. For this reason, Christopoulos et al. argued
that the advantage of self-centering systems can only be correctly evaluated when the
performance based assessment considers residual displacements. This study was also extended
to multi-degree-of-freedom systems in a companion paper by Pampanin et al. [2-60], with
results confirming the previous conclusions.
From experimental investigation, Chou and Hsu [2-61] found that the hysteresis behaviour of
concrete filled steel tubes with unbonded post-tensioning did not follow either the idealised bi-
linear elastic or flag-shape rules that are often assumed for self-centering members. Chou and
Hsu developed a more realistic hysteresis model that captured the stiffness degradation
observed during testing and compared dynamic time-history analysis results with a flag-shaped
hysteresis behaviour to assess the implications of this assumption. Their results indicated that
the maximum lateral displacements predicted by the two hysteresis models could vary by up to
200%. This highlights some concern as to the validity of previous analyses which assumed the
idealised flag-shape even when experimental observations indicate stiffness degradation.
2.4.2.2 Wall and building analysis

Kurama et al. [2-18] studied the seismic performance of unbonded post-tensioning precast
concrete walls with no additional hysteretic energy dissipation. Non-linear dynamic time-
history analysis was conducted using an analytical model that was developed using fiber
elements to represent the concrete wall panel, and truss elements to represent the post-
tensioning tendons. Gap opening in the model was approximated from distributed tensile
deformation in the wall panels, with the concrete tensile strength set to zero. The walls were
designed for a 6 storey prototype structure in a region of moderate seismicity. Analysis
indicated that walls with unbonded post-tensioning exhibited larger lateral displacements than
a comparable monolithic cast-in-place reinforced concrete wall. However, the residual drifts
were significantly lower for the walls with unbonded post-tensioning.
In a further study, Kurama et al. [2-62] extended the analysis of unbonded post-tensioned
walls, conducting a comprehensive parametric study using the fiber model. The effects of site
- 25 -
seismicity, soil type, initial stress and location of the tendons, assumed viscous damping, and
inelastic energy dissipation were investigated. Walls were subjected to a suite of historical and
artificial earthquake records that were scaled to design-level and survival-level intensities. The
study concluded that on average the equal displacement assumption provides reasonable
estimation of maximum design and survival level drifts for all sites except those with a soft soil
profile. Additionally, the amount of PT force, the initial stress and the location of the tendons
are the most critical parameters that affect the wall’s seismic behaviour. An increase in the
viscous damping from 3% to 5% resulted in reduced lateral displacements, which highlighted
the importance of additional damping in achieving a more desirable seismic response.
Researchers at Iowa State University have conducted extensive non-linear dynamic analysis of
the PRESSS five storey building. Rahman and Sritharan [2-63] analysed the wall direction of
the building response to evaluate the performance when both force-based design (FBD) and
direct displacement-based design (DDBD) procedures were used. The analysis used a lumped
plasticity model consisting of elastic frame elements and rotational springs. The jointed wall
system was represented by two vertical frame members with properties representing the elastic
stiffness of the panels, and by rotational springs at the foundation with bilinear-elastic
definitions to represent the wall behaviour when uplift occurs and the wall rocks. The UFP
connectors placed between the wall panels were represented by a spring connected to both the
wall members with rigid elements. Additionally, the contribution from the seismic and gravity
frame system was also modelled with elastic beam and column elements and rotational springs
at the beam-column joints. The model was first subjected to the same set of short duration
earthquake records used during testing of the PRESSS building, which successfully validated
the modelling procedure, with the results showing good correlation with the response measured
during testing. A further series of historical earthquake records were then used to evaluate the
performance of the designed buildings. Records were scaled to match four performance levels,
I, II, III and IV, representing different earthquake intensities, as outlined in the performance
based design guidelines published by the Structural Engineers Association of California [2-64].
The analysis results showed that the building designed using DDBD exhibited higher inter-
storey drifts than the building designed using FDB but that the maximum floor accelerations
where higher for the FBD building. Additionally, while the response of the DDBD building
was below all of the acceptance limits at each performance level, the building designed using
FBD exceeded the maximum floor acceleration limit for levels I, III and IV. Because all of the
- 26 -
performance criteria were meet, it was concluded that DDBD was suitable for the design of
buildings with self-centering walls. Additionally, DDBD also results in a lower design base
shear, which leads to a more economical seismic resisting solution.
Pennucci et al. [2-65] analysed a series of hybrid walls to assess the accuracy of developed
DDBD procedures for self-centering systems. A multi-spring model was adopted that
consisted of an elastic frame element for the wall and a bed of contact springs to capture the
gap opening at the wall base. Additionally, post-tensioning tendons and mild steel energy
dissipating bars were represented with elasto-plastic springs. This model was successfully
validated through comparison with experimental data from the hybrid wall tests performed by
Restrepo and Rahman [2-37]. Five hybrid walls were then designed using DDBD for
prototype buildings ranging from 4 to 20 stories and subjected to a series of earthquake records
that were scaled to match an arbitrary design spectrum. Pennucci et al. concluded that the
analysis results showed good correlation with the performance targets, indicating that the
design procedure was accurate and suitable for self-centering structures up to 20 stories high.
Additionally, the expressions developed to estimate the equivalent damping for flag-shape
hysteresis behaviour were suitable for the hybrid wall response, but would need to be validated
or redefined for systems with different hysteresis parameters.
Lastly, Erkmen and Schultz [2-66] recently conducted a series of dynamic analyses of precast
concrete walls with unbonded post-tensioning. Whereas previous researchers had designed
walls such that the unbonded tendons remained elastic up to the design level seismic response,
Erkmen and Schultz investigated the behaviour of walls where the tendons were allowed to
yield. They argued that without additional energy dissipating components, self-centering can
still be achieved even when the prestress is completely lost because of the self-weight of the
wall and additional gravity loads. A fiber element model was used, similar to that used by
Kurama et al. [2-18]. Five different walls with various prestressing tendon configuration were
designed for a 5 storey prototype structure, and each wall was subjected to 5 different scaled
historical earthquakes to assess its seismic performance. The time-history analysis results
indicated that although tendon location has a significant effect on wall stiffness and maximum
lateral displacements, it has negligible impact on the self-centering behaviour. Self-centering
was achieved for all walls, even when significant yielding of the post-tensioning tendons
occurred. However, Erkmen and Schultz identified that the behaviour of actual building
systems may differ and that the effect of gravity framing systems and additional energy
- 27 -
dissipating components needs to be considered when investigating the ability of the structure to
self-center.
2.5 APPLICATION OF SYSTEMS

Since the conception of self-centering precast concrete technology, several buildings have been
constructed that utilize such structural components. One of the first applications was the
Paramount building, shown in Figure 2.8, which is a 39 storey apartment building located in
San Francisco, California [2-67]. The Paramount building was designed using a self-centering
hybrid moment resisting frame which was originally developed and tested during the PRESSS
program and was constructed by Pankow Builders Ltd. Through the use of this innovative
system, the Paramount building was the tallest concrete structure, as well as the tallest precast,
prestressed concrete building constructed in seismic zone 4 (highest seismic zone in
California). Pankow Builders Ltd. have since constructed several other precast concrete
structures in California using hybrid moment resisting frames [2-68, 2-69].
Figure 2.8 – The Paramount building in San Francisco [2-67]
- 28 -
2.6 Discussions and Conclusions
In New Zealand there has also been interest in self-centering seismic resisting systems. This
has resulted in the recent construction of a 5 storey precast concrete building at Victoria
University in Wellington, shown in Figure 2.9a, that uses both frame and wall systems
designed using unbonded post-tensioning [2-70]. The frame system utilizes externally
mounted energy dissipating elements at the beam-column joint and the wall system consists of
two precast concrete wall panels post-tensioned to the foundation and connected together with
steel coupling beams, as shown in Figure 2.9b. As well as supporting the floor diaphragms, the
steel coupling beams were designed to act as energy dissipating components during an
earthquake. The option of using mild-steel reinforcing bars for energy dissipation was
considered during design but space limitation and reinforcement cutter at the base of the wall
panel would have made the insertion of these bars difficult within construction tolerances. To
the author’s knowledge this is the first application of a self-centering precast concrete wall
system in a multi-storey building located in a region of high seismicity.
(a) Artist impression (b) Details of wall system

Figure 2.9 – Victoria University building in Wellington, NZ [2-70]
2.6 DISCUSSIONS AND CONCLUSIONS

An extensive literature review was presented outlining the conception, development, testing
and analysis of self-centering precast concrete wall and building systems. Although the
development and validation of self-centering wall systems has been extensive, the
- 29 -
implementation into buildings is limited. Additionally, previous research has focused on the
application of self-centering wall systems for regions of high seismicity. Simplified design
procedures for unbonded post-tensioned precast concrete walls that would allow for their use in
low-rise buildings in low seismic regions need to be developed.
Testing and analysis of self-centering wall systems has often been performed in isolation and
the interaction between structural and non-structural components has typically not been
considered. In the few experimental studies that have dealt with the displacement
incompatibility between wall and floor systems, the focus of the study was on other aspects of
structural behaviour and so extensive investigation and analysis of the wall-to-floor interaction
was not completed. The seismic behaviour of a building with self-centering walls cannot be
understood unless the complete building system is analysed. Many studies conclude that the
wall system has the ability to self-center, but in reality the wall system will have to self-center
not just itself but also the entire building.
When performing dynamic time-history analysis, previous studies often assumed that the
behaviour of self-centering components follows the flag-shape hysteresis rule. However,
experimental studies have indicated that this idealised response is not realistic and that
significant stiffness degradation and residual drift is typically observed, especially for wall
systems where significant inelastic strains can be generated in the wall toe.
Furthermore, researchers typically comment on the self-centering ability from the results of
slow speed reverse cyclic experiments without considering the residual displacement following
dynamic excitation.
2.7 REFERENCES
[2-1] Fintel, M. (1995) Performance of buildings with shear walls in earthquakes of the last
thirty years. PCI Journal, 40(3), 62-80.
[2-2] Ghosh, S. K. (2001) Earthquake resistant precast buildings. MC magazine, winter

2001.
[2-3] Park, R. (1995) A Perspective on the seismic design of precast concrete structures in
New Zealand. PCI Journal, 40(3), 40-60.
- 30 -
2.7 References
[2-4] Ghosh, S. K. (2002) Seismic design provisions in U.S. codes and standards: A look
back and ahead. PCI Journal, 47(1), 94-99.
[2-5] Vayas, I., Dasiou, M. E., and Marinelli, A. (2007) Columns of Greek Temples under
seismic loading. Bautechnik, 84(6), 388-396.
[2-6] Housner, G. W. (1963) Behavior of inverted pendulum structures during earthquakes.

Seismological Society of America -- Bulletin, 53(2), 403-417.
[2-7] Ma, Q. T. and Butterworth, J. W. (2006) A generalised mathematical model for

controlled rocking systems. 4th International Conference on Earthquake Engineering,
Taipei, Taiwan, October 12-13.
[2-8] Ma, Q. T., Wight, G. D., Butterworth, J. W., and Ingham, J. M. (2006) Assessment of
current procedures for predicting the in-plane behaviour of controlled rocking walls
Eighth U.S. National Conference on Earthquake Engineering (8NCEE), San Francisco,
USA, April 18-22.
[2-9] Skinner, R. I., Robinson, W. H., and McVerry, G. H. (1993) An introduction to seismic
isolation. Wiley, Chichester, New York.
[2-10] Crippen, K. (2002) North Viaduct to Lions Gate Bridge: Vancouver engineers allow a
structure to rock gently on its foundations to help it withstand a major earthquake
earthquake. Canadian Consulting Engineer, 43(7), 34.
[2-11] Priestley, M. J. N. (1991) Overview of PRESSS research program. PCI Journal, 36(4),
50-57.
[2-12] Cheok, G. S. and Lew, H. S. (1991) Performance of precast concrete beam-to-column

connections subject to cyclic loading. PCI Journal, 36(3), 56-67.
[2-13] Priestley, M. J. N. and Tao, J. R. (1993) Seismic response of precast prestressed

concrete frames with partially debonded tendons. PCI Journal, 38(1), 58-69.
[2-14] Priestley, M. J. N. and MacRae, G. A. (1996) Seismic tests of precast beam-to-column

joint subassemblages with unbonded tendons. PCI Journal, 41(1), 64-80.
[2-15] Stanton, J., Stone, W. C., and Cheok, G. S. (1997) A hybrid reinforced precast frame
for seismic regions. PCI Journal, 42(2), 20-32.
[2-16] Stone, W. C., Cheok, G. S., and Stanton, J. (1995) Performance of hybrid moment-
resisting precast beam-column concrete connections subjected to cyclic loading. ACI
Structural Journal, 92(2), 229-249.
[2-17] Kurama, Y., Pessiki, S., Sause, R., and Lu, L.-W. (1999) Seismic behavior and design
of unbonded post-tensioned precast concrete walls. PCI Journal, 44(3), 72-89.
- 31 -
[2-18] Kurama, Y., Sause, R., Pessiki, S., and Lu, L.-W. (1999) Lateral load behavior and
seismic design of unbonded post-tensioned precast concrete walls. ACI Structural
Journal, 96(4), 622-632.
[2-19] Nakaki, S. D., Stanton, J. F., and Sritharan, S. (1999) An overview of the PRESSS five-
story precast test building. PCI Journal, 44(2), 26-39.
[2-20] Priestley, M. J. N., Sritharan, S., Conley, J. R., and Pampanin, S. (1999) Preliminary
results and conclusions from the PRESSS five-story precast concrete test building. PCI
Journal, 44(6), 42-67.
[2-21] Perez, F. J., Pessiki, S., and Sause, R. (2004) Experimental and analytical lateral load
response of unbonded post-tensioned precast concrete walls. ATLSS Report No. 04-11,
Lehigh University, Bethlehem, PA.
[2-22] Perez, F. J., Sause, R., and Lu, L. W. (2003) Lateral load tests of unbonded post-
tensioned precast concrete walls. ACI Special Publication, 211, 161-182.
[2-23] Perez, F. J., Sause, R., and Pessiki, S. (2007) Analytical and experimental lateral load
behavior of unbonded posttensioned precast concrete walls. Journal of Structural
Engineering, 133(11), 1531-1540.
[2-24] Laursen, P. T. and Ingham, J. M. (2001) Structural testing of single-storey post-

tensioned concrete masonry walls. The Masonry Society Journal, 19(1), 69-82.
[2-25] Laursen, P. T. and Ingham, J. M. (2004) Structural testing of enhanced post-tensioned

concrete masonry walls. ACI Structural Journal, 101(6), 852-862.
[2-26] Laursen, P. T. and Ingham, J. M. (2004) Structural testing of large-scale posttensioned

concrete masonry walls. Journal of Structural Engineering, 130(10), 1497-1505.
[2-27] Wight, G. D., Ingham, J. M., and Kowalsky, M. J. (2006) Shaketable testing of
rectangular post-tensioned concrete masonry walls. ACI Structural Journal, 103(4),
587-595.
[2-28] Wight, G. D., Kowalsky, M. J., and Ingham, J. M. (2007) Shake table testing of
posttensioned concrete masonry walls with openings. Journal of Structural
Engineering, 133(11), 1551-1559.
[2-29] Wight, G. D. and Ingham, J. M. (2008) Tendon stress in unbonded post-tensioned

masonry walls at nominal in-plane strength. Journal of Structural Engineering, 134(6),
947-960.
[2-30] Wight, G. D., Ingham, J. M., and Wilton, A. R. (2007) Innovative seismic design of a
post-tensioned concrete masonry house. Canadian Journal of Civil Engineering,
34(11), 1393-1402.
- 32 -
2.7 References
[2-31] Sritharan, S., Aaleti, S., and Thomas, D. J. Seismic analysis and design of precast
concrete jointed wall systems. ISU-ERI-Ames Report ERI-07404, Department of Civil,
Construction and Environmental Engineering, Iowa State University, Ames, IA, 2007.
[2-32] Stanton, J. F. and Nakaki, S. D. (2002) Design guilelines for precast concrete seismic
structural systems. PRESSS Report No. SM 02-02, University of Washington, Seattle,
WA.
[2-33] Thomas, D. J. and Sritharan, S. (2004) An evaluation of seismic design guidelines

proposed for precast jointed wall systems. ISU-ERI-Ames Report ERI-04643.
Department of Civil, Construction and Environmental Engineering, Iowa State
University, Ames, IA.
[2-34] Aaleti, S. and Sritharan, S. (2009) A simplified analysis method for characterizing
unbonded post-tensioned precast wall systems. Engineering Structures, 31(12), 2966-
2975.
[2-35] Smith, B. J. and Kurama, Y. C. (2009) Design of hybrid precast concrete walls for
seismic regions. Proceedings of the 2009 Structures Congress - Don't Mess with
Structural Engineers: Expanding Our Role, 1673-1682.
[2-36] Holden, T., Restrepo, J., and Mander, J. B. (2003) Seismic performance of precast
reinforced and prestressed concrete walls. Journal of Structural Engineering, 129(3),
286-296.
[2-37] Restrepo, J. I. and Rahman, A. (2007) Seismic performance of self-centering structural

walls incorporating energy dissipators. Journal of Structural Engineering, 133(11),
1560-1570.
[2-38] Paulay, T. and Priestley, M. J. N. (1992) Seismic design of reinforced concrete and
masonry buildings. John Wiley and Sons Inc., New York.
[2-39] Kurama, Y. C. and Shen, Q. (2004) Posttensioned hybrid coupled walls under lateral
loads. Journal of Structural Engineering, 130(2), 297-309.
[2-40] Kurama, Y. C., Weldon, B. D., and Shen, Q. (2006) Experimental evaluation of
posttensioned hybrid coupled wall subassemblages. Journal of Structural Engineering,
132(7), 1017-1029.
[2-41] Kurama, Y. C. (2001) Simplified seismic design approach for friction-damped

unbonded post-tensioned precast concrete walls. ACI Structural Journal, 98(5), 705-
716.
[2-42] Abdul Hamid, N. H. (2006) Seismic damage avoidance design of warehouse buildings
constructed using precast hollow core pannels. PhD Thesis. Dept. of Civil and
Environmental Engineering, University of Canterbury, New Zealand.
[2-43] Hamid, N. H. and Mander, J. B. (2010) Lateral seismic performance of multipanel

precast hollowcore walls. Journal of Structural Engineering, 136(7), 795-804.
- 33 -
[2-44] Kurama, Y. C. (2000) Seismic design of unbonded post-tensioned precast concrete

walls with supplemental viscous damping. ACI Structural Journal, 97(4), 648-658.
[2-45] Marriott, D., Pampanin, S., Bull, D., and Palermo, A. (2008) Dynamic testing of
precast, post-tensioned rocking wall systems with alternative dissipating solutions.
Bulletin of the New Zealand Society for Earthquake Engineering, 41(2), 90-103.
[2-46] ACI Innovation Task Group 5. (2007) Acceptance criteria for special unbonded post-
tensioned precast structural walls based on validation testing (ITG 5.1-07). American
Concrete Institute, Farmington Hills, MI.
[2-47] ACI Innovation Task Group 5. (2009) Requirements for Design of a Special Unbonded
Post-Tensioned Precast Shear Wall Satisfying ACI ITG-5.1 (ITG 5.2-09). American
Concrete Institute, Farmington Hills, MA.
[2-48] ACI 318-08. Building code requirements for structural concrete. American Concrete
Institute, Farmington Hills, MI, 2008.
[2-49] NZS 3101:2006. Concrete structures standard. Standards New Zealand, Wellington,
New Zealand.
[2-50] Pampanin, S., Cattanach, A., and Haverland, G. (2010) PRESSS design handbook :
seminar notes. Technical report, TR44, New Zealand Concrete Society, Auckland, N.Z.
[2-51] Rodriguez, M. E. and Blandon, J. J. (2005) Tests on a half-scale two-story seismic-

resisting precast concrete building. PCI Journal, 50(1), 94-114.
[2-52] Wilson, J. L., Robinson, A. J., and Balendra, T. (2008) Performance of precast concrete
load-bearing panel structures in regions of low to moderate seismicity. Engineering
Structures, 30(7), 1831-1841.
[2-53] Schoettler, M. J., Belleri, A., Dichuan, Z., Restrepo, J. I., and Fleischman, R. B. (2009)
Preliminary results of the shake-table testing for the development of a diaphragm
seismic design methodology. PCI Journal, 54(1), 100-124.
[2-54] Amaris, A., Pampanin, S., Bull, D., and Carr, A. (2008) Experimental performance of
hybrid frame systems with non-tearing floor connections. 14th World Conference in
Earthquake Engineering (14WCEE), Beijing, China, October 12-17.
[2-55] Garlock, M. E. M., Li, J., and Blaisdell, M. L. (2006) Collector beam interaction with
steel self-centering moment frames. 4th International Conference on Earthquake
Engineering, Taipei, Taiwan, October 12-13.
[2-56] Cheng, C. T., Chen, H. H., Lin, K. C., Chen, P. C., and Huang, S. J. (2008) Seismic
performance of a 3D precast/post-tensioned reinforced concrete sub-structure under bi-
axial loads. 14th World Conference in Earthquake Engineering (14WCEE), Beijing,
China, October 12-17.
- 34 -
2.7 References
[2-57] Cheng, C. T. and Lu, W. H. (2006) Shaking table tests of self-centering designed bridge
sub-structures. 4th International Conference on Earthquake Engineering, Taipei,
Taiwan, October 12-13.
[2-58] Christopoulos, C., Filiatrault, A., and Folz, B. (2002) Seismic response of self-centring
hysteretic SDOF systems. Earthquake Engineering and Structural Dynamics, 31(5),
1131-1150.
[2-59] Christopoulos, C., Pampanin, S., and 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.
[2-60] Pampanin, S., Christopoulos, C., and Priestley, M. J. N. (2003) Performance-based

seismic response of frame structures including residual deformations. Part II: Multi-
[2-61] Chou, C. C. and Hsu, C. P. (2008) Hysteretic model development and seismic response
of unbonded post-tensioned precast CFT segmental bridge columns. Earthquake
Engineering and Structural Dynamics, 37(6), 919-934.
[2-62] Kurama, Y. C., Sause, R., Pessiki, S., and Lu, L.-W. (2002) Seismic response
evaluation of unbonded post-tensioned precast walls. ACI Structural Journal, 99(5),
641-651.
[2-63] Rahman, M. A. and Sritharan, S. (2006) An evaluation of force-based design vs. direct
displacement-based design of jointed precast post-tensioned wall systems. Earthquake
Engineering and Engineering Vibration, 5(2), 285-296.
[2-64] Seismology Committee. Recomented lateral force requirements and commentary (blue
book). Structural Engineers Association of California (SEAOC), California, USA,
1999.
[2-65] Pennucci, D., Calvi, G. M., and Sullivan, T. J. (2009) Displacement-based design of
precast walls with additional dampers. Journal of Earthquake Engineering, 13(1
SUPPL. 1), 40-65.
[2-66] Erkmen, B. and Schultz, A. E. (2009) Self-centering behavior of unbonded, post-

tensioned precast concrete shear walls. Journal of Earthquake Engineering, 13(7),
1047-1064.
[2-67] Englekirk, R. E. (2002) Design-construction of the Paramount - A 39-story precast

prestressed concrete apartment building. PCI Journal, 47(4), 56-71.
[2-68] Ingham, J. M. (2000) A presentation of the PRESSS technology applied to a 39-story

building by Pankow Builders in California. Concrete 2000 Better, Faster, Smarter,
Wairakei, Taupo, New Zealand, 13-15 Oct, 27-32.
[2-69] Snapper, G. Defying Mother Nature. Precast Solutions Magazine, National Precast
Concrete Association (NPCA), 2005.
- 35 -
[2-70] Cattanach, A. and Pampanin, S. (2008) 21st century precast: the detailing and
manufacture of NZ's first multi-storey PRESSS-building. New Zealand Society for
Earthquake Engineering Conference 2008, Rotorua, New Zealand.

- 36 -
Chapter 3
Testing of Post-tensioned Wall Panels
As explained in Chapter 1, individual precast concrete wall panels with unbonded post-
tensioning (PT) and no additional energy dissipating devises are suitable for building
applications in regions of low seismicity. In particular, PT walls with only the minimum
required longitudinal and transverse reinforcing steel, and no additional confinement
reinforcement in the compression toe, are a simple construction alternative. However, minimal
experimental validation has been performed on such wall details. This chapter details a series
of experimental tests that were conducted to investigate the lateral force behaviour of wall
panels with unbonded post-tensioning. A major objective of this testing was to investigate the
compressive strains in the wall toe so that procedures for estimating wall nominal flexural
strength could be validated to assist in developing simplified design procedures that could be
used within the scope of current concrete design standards.
3.1 NOMINAL FLEXURAL STRENGTH

To estimate the flexural strength of a reinforced concrete member, the concrete compressive
stress profile is often approximated using an equivalent rectangular stress block which was first
- 37 -
Chapter 3 Testing of Post-tensioned Wall Panels
proposed by Whitney [3-1]. Calculation of member nominal flexural strength should then be
performed when the concrete at the extreme compression fibre of the section reaches its
maximum usable strain. As reported by Mattock et al. [3-2], historical tests indicated that a
strain of 0.003 provides a conservative approximation for the maximum compressive strain for
unconfined concrete and that at this state the concrete in a flexural member will usually not
show any visible cracking or spalling. For this reason, current design standards [3-3, 3-4]
recommend using a maximum usable concrete strain, εcu, of 0.003 at the extreme compression
fibre for nominal flexural strength calculations. However, Paulay and Priestley [3-5] argue that
this limit is only suitable for members subjected to a constant moment along the member’s
length, and that for members with a significant moment gradient the onset of crushing is
delayed beyond strains of 0.003 and sometimes occurs at strains as high as 0.006 to 0.008.
PT wall systems developed for use in regions with high seismicity [3-6, 3-7, 3-8, 3-9] do not
typically rely on the traditional definition of nominal flexural strength. The large design level
drifts result in extremely high concrete strain demands, and so confinement reinforcement is
used in the wall toe to ensure that a stable ductile response is achieved for compressive strains
in excess of 0.003. However, calculation of the nominal flexural strength of PT walls designed
with no additional confinement reinforcement must be performed in accordance with current
concrete design standards [3-3, 3-4] using a maximum usable concrete strain of 0.003.
Quantifying the concrete strains in a precast concrete wall with unbonded PT is complicated by
the moment and curvature profiles that develop. When a PT wall is subjected to a lateral force,
inelastic flexural deformation is concentrated at a single crack that opens up at the wall base.
The wall curvature, and thus strain at the extreme compression fibre of the section, increases
dramatically near the base of the wall, as shown in Figure 3.1a. In comparison the inelastic
deformation in a traditional reinforced concrete (RC) wall is distributed over a larger height of
the wall panel and so the concrete strains at the wall toe vary less rapidly, as shown in Figure
3.1b. As a result, measurement of concrete strains at the extreme compression fibre is more
difficult for an unbonded PT wall, because the location and length of the measurement device
is critical due to the dramatic increase in strains at the base of the wall. For this reason, the
measurement of these critical concrete strains was extensively investigated during the
experimental tests and several different techniques for strain measurement were evaluated.
- 38 -
3.2 Construction Details
εc εc
(a) PT wall (b) RC wall

Figure 3.1 – Strain profile along the edge of an RC and PT wall
3.2 CONSTRUCTION DETAILS

3.2.1 Wall Panel Specification
The tests were conducted in three phases using a total of eight wall panels. Instead of basing
the wall designs on a particular prototype building, the major parameters, including panel
dimensions, prestressing tendon configuration, and concrete strength were varied over a range
expected for typical low-rise construction details. The dimensions and details of the test walls
are shown in Table 3.1, where lw refers to the wall length, he the wall height and bw the wall
thickness (also shown on the wall drawings in Figure 3.2 and Figure 3.3). Phase I of the test
program consisted of two precast wall panels, labelled A and B. Wall A was 3.00 m high,
2.00 m long and 0.14 m thick and wall B was 3.30 m high, 1.00 m long and 0.14 m thick.
Phase II of the test program consisted of another four precast wall panels, labelled C1, C2, D1
and D2. Wall C1 was 3.00 m high, 1.00 m long and 0.12 m thick and wall D1 was 3.00 m
high, 2.00 m long and 0.12 m thick. Walls C2 and D2 were constructed by extracting
undamaged regions of panels C1 and D1, respectively. Because the damage was isolated to the
corner of the panel, a 0.40 m high strip was cut off the top and bottom of panels C1 and D1,
resulting in a new height of 2.20 m for walls C2 and D2. Phase III of the test program included
a final two precast wall panels, labelled E and F. Wall E was 3.00 m high, 1.00 m long and
0.12 m thick and wall F was 3.00 m high, 2.00 m long and 0.12 m thick.
- 39 -
Table 3.1 – Details of test walls
Wall lw he he/lw bw Ten Aps fse fc ' fc/ fc' Bedding

(m) (m) (m) # (mm2) (%fy) (MPa)
1
Phase I None /
1 A i 2.00 3.00 1.5 0.140 5* 883.5 60 32.9 0.062 Mortar
2 ii 2.00 3.00 1.5 0.140 5* 883.5 40 32.9 0.044 Mortar
3 iii 2.00 3.00 1.5 0.140 2* 353.4 30 32.9 0.016 Mortar
4 iv 2.00 3.00 1.5 0.140 2* 353.4 60 32.9 0.026 Mortar
5 B i 1.00 3.30 3.3 0.140 1* 176.7 40 31.7 0.020 Plaster
6 ii 1.00 3.30 3.3 0.140 2* 353.4 40 31.7 0.038 Plaster
7 iii 1.00 3.30 3.3 0.140 3* 530.1 40 31.7 0.056 Plaster
8 iv 1.00 3.30 3.3 0.140 3* 530.1 60 31.7 0.079 Plaster
Phase II
9 C1 i 1.00 3.00 3 0.120 2 293.1 50 25.8 0.074 Plaster
10 ii 1.00 3.00 3 0.120 2 293.1 50 27.9 0.074 Mortar
11 iii 1.00 3.00 3 0.120 3 439.65 75 25.8 0.166 Plaster
12 iv 1.00 3.00 3 0.120 3 439.65 75 27.9 0.166 Mortar
13 C2 i 1.00 2.20 2.2 0.120 2 293.1 50 28.1 0.068 Plaster
14 ii 1.00 2.20 2.2 0.120 2 293.1 50 28.1 0.068 Plaster
15 iii 1.00 2.20 2.2 0.120 3 439.65 75 28.1 0.154 Plaster
16 iv 1.00 2.20 2.2 0.120 3 439.65 75 28.1 0.154 Plaster
17 D1 i 2.00 3.00 1.5 0.120 3 439.65 50 26.0 0.055 Plaster
18 ii 2.00 3.00 1.5 0.120 3 439.65 50 26.0 0.055 Plaster
19 iii 2.00 3.00 1.5 0.120 5 732.75 70 26.0 0.129 Plaster
20 iv 2.00 3.00 1.5 0.120 5 732.75 70 26.0 0.129 Plaster
21 D2 i 2.00 2.20 1.1 0.120 3 439.65 30 28.1 0.031 Plaster
22 ii 2.00 2.20 1.1 0.120 3 439.65 50 28.1 0.051 Plaster
23 iii 2.00 2.20 1.1 0.120 5 732.75 50 28.1 0.086 Plaster
24 iv 2.00 2.20 1.1 0.120 5 732.75 70 28.1 0.120 Plaster
Phase III
25 E i 1.00 3.00 3 0.120 1 293.1 70 42.4 0.033 Plaster
26 ii 1.00 3.00 3 0.120 2 293.1 50 40.2 0.047 Plaster
27 iii 1.00 3.00 3 0.120 3 439.65 50 42.4 0.070 Plaster
28 iv 1.00 3.00 3 0.120 3 439.65 75 40.2 0.105 Plaster
29 F i 2.00 3.00 1.5 0.120 3 439.65 50 39.5 0.036 Plaster
30 ii 2.00 3.00 1.5 0.120 3 439.65 70 39.1 0.051 Plaster
31 iii 2.00 3.00 1.5 0.120 5 732.75 50 39.5 0.061 Plaster
32 iv 2.00 3.00 1.5 0.120 5 732.75 70 39.1 0.085 Plaster
1
: Phase I completed with the assistance of Nicholas Brooke
* 15 mm bar (fpy = 1050 MPa) all else was 15.2 mm strand (fpy = 1580 MPa)
Ten # = number of tendons, Aps = total area of prestressing and fse = initial prestressing stress

Steel mesh was used to provide nominal reinforcement, but no additional confinement
reinforcement was provided in the wall toe. Walls A and B were constructed with two layers
of reinforcing mesh with a 4 mm diameter, 150 mm spacing and a nominal yield stress of
475 MPa, and walls C1, C2, D1, D2, E and F were constructed with four layers of the same
mesh. Ducts for the post-tensioning tendons were cast into the wall panel, with five ducts in
- 40 -
walls A, D1, D2 and F and three ducts in walls B, C1, C2 and E. The 30 mm diameter plastic
or 25 mm diameter steel ducts were spaced at 406.4 mm centers in order to align with the
existing strong floor anchor points. Design drawings used for the construction of the wall
specimens are shown in Figure 3.2 and Figure 3.3.
The prestressing tendons used for walls A and B consisted of 15 mm diameter high strength
steel bars, and 15.2 mm diameter prestressing strand was used for walls C1, C2, D1, D2, E and
F. As described in detail below, the testing procedure allowed for four individual tests on each
wall panel. This allowed the prestressing configuration to be varied, resulting in a total of 32
unique test arrangements. The number of tendons used during each test and the initial
prestressing stress are also listed in Table 3.1. When fewer tendons were used than available
tendon ducts, the configuration was always kept symmetrical, and when three tendons were
used for the 2 m long walls with five ducts, the tendons were placed in the center three ducts.
(a) Wall A (b) Wall B

Figure 3.2 – Phase I design drawings
- 41 -
94 406 406 94 187 406 406 406 406 187

cross
section
bw = 120
lw = 1000 lw = 2000
194 300 287 100 400
Reid Eye Anchor

(2 per panel)
630
25 mm ducts
for tendons
665 mesh 665 mesh

elevation
he = 3000
he = 3000
(4 layers) (4 layers)
Reid Foot Anchor

(4 per panel)
300 mm long D25

bars, drill and tap
for M12 bolt
630
All dimensions in mm
(a) Walls C1 & E (b) Walls D1 & F

Figure 3.3 – Phase II and III design drawings
3.2.2 Panel Construction

The test walls were all constructed as prefabricated precast panels. Walls A and B were
constructed onsite at the University of Auckland (UoA) test hall and walls C1, C2, E and F
were constructed and cast offsite at Strescrete Papakura. The panels were cast using custom
made wood forms at the UoA and on a heated steel casting bed with wood side forms at the
Strescrete facilities. Once the forms were constructed, the first layers of steel mesh were
placed using plastic chairs to ensure a minimum of 30 mm cover. Following this the tendon
ducts were positioned before the top layers of mesh were added. Concrete was placed using an
overhead crane and bucket, and vibrated into the forms before being finished with a trowel.
The concrete was cured for one to two days before the forms were removed and the panels
lifted and transported to the testing laboratory. Figure 3.4 shows the completed panels prior to
the concrete being cast.
- 42 -
(a) Walls A and B
(b) Walls E and F

Figure 3.4 – Construction of wall panels for Phase I and III
Lifting inserts were placed in the panels during construction to allow the panels to be
manoeuvred once they were cast. As shown in Figure 3.3, the inserts included two edge
anchors at the top end of the panel and four face anchors. In addition, short D25 reinforcing
bars, tapped and threaded for an M12 bolt, were placed at the ends of the panels, as shown in
Figure 3.2, Figure 3.3, and Figure 3.4. The purpose of these bars was to provide an attachment
for the steel loading beam which is described in the test setup below.
- 43 -
3.2.3 Material Properties

3.2.3.1 Concrete
The compressive strength of the concrete used to cast the wall panels was determined from
concrete cylinders cast at the same time using the same concrete mix. Two or three cylinders
were tested on the same day as each wall test, and the average measured compressive strengths
are listed in Table 3.1 for each test.
3.2.3.2 Prestressing steel

The tensile stress-strain response of the prestressing steel was determined from experimental
testing. Three samples of both the 15 mm bar and 15.2 mm strand were tested in direct tension
until failure, with the recorded results presented in Figure 3.5. The stress was calculated by
dividing the measured force from a load cell by the nominal area of the bar or tendon, and the
strain was calculated by measuring displacements across a central 250 mm gauge length of the
bar or strand. The 15 mm bar was found to remain fully elastic up to a tensile stress of
~1000 MPa, with the recorded yield stress being between 1055 – 1120 MPa. The 15.2 mm
strand was found to remain fully elastic up to a tensile stress of ~1500 MPa, with the stress-
strain response closely matching the stress-strain curve provided by the manufacturer on the
strand certificate. Additionally, the strand samples were unloaded and reloaded several times
to 1550 MPa with less than 5% change in the modulus of elasticity observed. The strand
samples all failed at loads close to the yield stress of 1580 MPa due to fracture of one or more
wires directly at the anchor. Premature strand failure at anchorages was found by Welsh and
Kurama [3-10] to be a common problem in mono-stand systems. However the strand failure
was not considered critical to the wall tests because the tendons were not intended to be loaded
beyond yield. The calculated modulus of elasticity for each test is presented in Table 3.2, with
average values of 186.1 GPa and 199.6 GPa for the bar and strand respectively.
Table 3.2 – Summary of prestressing steel properties
Type Sample Elastic modulus (GPa)

Test Average
15.2 mm strand Manufacturer certificate 200.4
Test 1 197.2
Test 2 201.2 199.6
Test 3 200.5
15 mm bar Test 1 199.1
Test 2 182.1 186.1
Test 3 177.0
- 44 -
3.3 Testing Details
1200 2000
1000
1500
800
Stress (MPa)
Stress (MPa)
600 1000
400
Test 1
500
Test 1 Test 2
200
Test 2 Test 3
Test 3 Manufacturer
0 0
0 0.005 0.01 0.015 0 0.005 0.01 0.015
Strain (m/m) Strain (m/m)
(a) 15 mm bar (b) 15.2 mm strand

Figure 3.5 – Measured stress-strain response of prestressing bar and strand samples
3.3 TESTING DETAILS

3.3.1 Test Setup
Experimental testing was conducted using the setup shown in Figure 3.6. The precast concrete
panels were erected directly on the laboratory strong floor, without requiring a purpose-built
foundation. A bedding material was placed between the wall and foundation to achieve
uniform contact and this is described in more detail below. No additional restraints were
placed at the wall base, which meant that friction was relied upon to resist any slip at the wall-
to-foundation interface. A steel channel was placed at the top of the wall, providing a loading
beam to which an actuator was attached to apply the lateral force at the top of the wall. The
steel loading beam was initially bolted to the wall panel during construction. However during
testing, friction created by the post-tensioning was primarily relied upon to transfer the lateral
shear force between the loading beam and the wall panel. To isolate the in-plane response of
the wall panel, the top of the wall was restrained to prevent any out-of-plane displacements by
attaching two pinned struts between the loading beam and an adjacent load frame. The post-
tensioning tendons, located within ducts in the wall and floor, were anchored at the top of the
wall and underneath the strong floor. The measured unbonded length of the tendons was equal
to the wall height plus an additional 750 mm. The tendons were post-tensioned prior to testing
using a hollow-core jack at the top of the wall. Additionally, the top tendon anchorage utilised
- 45 -
a slotted washer to allow the barrel and wedges to be released and the tendon de-stressed
without exceeding the yield stress of the bar or tendon. Apart from the self-weight of the wall
and loading beam, no additional load was applied to the specimen to represent tributary gravity
loads. Instead, the PT force was relied on to achieve the required axial loads imposed on the
wall.
Lw
Plan bw
s
Loading
Tendon
beam
Reference Actuator
frame
Elevation Wall
panel he
Strongfloor
(a) Schematic representation (b) Wall C1 prior to testing

Figure 3.6 − Wall test setup
Initial trial tests on wall A were completed with no bedding material between the wall and the
concrete strong floor foundation. However, during these tests uniform contact was not
achieved at the wall-to-foundation interface, which resulted in an erratic force-displacement
response. Following this, tests A-i to A-iv were performed using a 5 mm thick cement mortar
bedding layer. Finally, the remaining walls, with the exception of wall C1, were tested using a
high strength, high stiffness, gypsum plaster placed between the wall and strong floor. This
provided an effective interface material, allowing uniform contact to be achieved with no
detrimental effect on wall behaviour. The use of a gypsum plaster bedding material also
conformed to the recent ACI guidelines specifying that the bedding material must be of higher
strength than the wall material [3-11]. Wall C1 was tested with a gypsum plaster bedding for
- 46 -
3.4 Instrumentation
the first two tests, before being seated on a cement mortar bedding when the wall was inverted
to allow for comparison. The bedding material used for each test is also listed in Table 3.1.
The Ceramical™ gypsum plaster that was used constitutes the same dry mix as Hydrostone™,
and when mixed at 32 parts water to 100 parts plaster (by weight) it had a compressive strength
of approximately 69 MPa, which is well above the compressive strength of the wall concrete.
3.3.2 Test Procedure

It is well established that the hysteretic behaviour of unbonded post-tensioned members is
generally bilinear-elastic. In addition, the main objective of this study was to investigate the
nominal flexural strength of PT walls when subjected to a lateral force, and so monotonic
lateral loading was applied to the top of the walls. The monotonic lateral force was applied in
a pseudo-static manner, pausing several times to examine and document the wall behaviour and
observed damage. Loading was terminated when either crushing occurred in the wall toe, or
when one of the prestressing tendons approached its yield stress. Because the walls were
loaded using a monotonic force, each panel was tested in both directions before being inverted
and retested, enabling each corner of the panel to be subjected to large compressive strains only
once. Therefore, four tests were performed on each panel using different tendon configurations
and/or levels of initial tendon stress, resulting in a total of 32 individual tests.
3.4 INSTRUMENTATION
The test walls were extensively instrumented to measure critical aspects of the wall’s lateral
load behaviour. The typical arrangement of instrumentation, excluding the specific
instrumentation for concrete strain measurement, is shown in Figure 3.7. The lateral force
from the actuator was measured using an integrated load cell. Additionally, load cells were
also used to measure the force in each prestressing tendon and the tendon stress was calculated
based upon the nominal bar or tendon area. During tests on walls A and B these load cells
were located at the top tendon anchorage, whereas they were placed at the bottom tendon
anchorage for the remaining tests, as shown in Figure 3.7. Additionally, displacement gauges
were used to capture the behaviour of the precast concrete panel, consisting of: lateral
displacement at the top of the wall panel, uplift occurring at the wall-foundation interface,
horizontal slip of the wall relative to the foundation and the loading beam, and deformation of
the panel itself. For all tests, the wall lateral displacement was measured at a height of 50 mm
- 47 -
below the top edge of the wall. For wall A, eleven displacement gauges were installed along
the base of the wall at 300 mm centers except for two gauges placed at each end, and nine
displacement gauges were installed at 200 mm centers for wall B with two gauges placed at
each end. For walls C1, C2 and E, seven displacement gauges were installed along the base of
the wall at 200 mm centers except the end gauges, and for walls D1, D2 and F, eleven gauges
were installed at 200 mm centers. The pattern of gauges to measure panel deformations was
typically located 150 mm from the ends of the wall panel and 50 mm in form the wall edges.
Reference frame
Strong wall
Horizontal disp.
Panel disp.
Slip disp.
Uplift disp.
Load cell
(b) Top of wall panel
(a) Schematic layout (c) Bottom of wall panel

Figure 3.7 – Typical instrumentation setup
3.4.1 Strain Measurement

Measurement of concrete strains at the extreme compressive fibre, or corner toe of the wall,
was performed using several measurement techniques that were refined over the three phases
of testing. Instrumentation included: displacement gauges (Ext), strain gauges (SGR, SGS, SGE)
and photogrammetry (Photo). Table 3.3 includes details of which measurement techniques
were used for each test.
- 48 -
3.4 Instrumentation
3.4.1.1 External gauges (Ext)

Walls A, B and C1 included externally mounted displacement gauges to measure strains in the
wall toe, connected to 4 mm steel rods that were drilled and epoxied into the end face. For
walls A and B, displacements were measured between the foundation and a point located
150 mm up the compression edge of the wall, and for wall C1, displacements were measured
between two points on the compression edge, at heights of 35 mm and 65 mm above the
foundation, as shown in Figure 3.8a.
3.4.1.2 Strain gauges on reinforcement mesh (SGR)

Based upon the assumption that the embedded steel mesh reinforcement would experience the
same strain as the surrounding concrete prior to any significant damage, strain gauges were
placed on the steel mesh in the corner toe of walls A and B. The gauges had a 5 mm gauge
length and were mounted on a wire of mesh that was 25 mm in from the end of the wall at
heights of 25, 75, 125, 175 and 225 mm from the base of the wall, as shown in Figure 3.8b.
3.4.1.3 Surface mounted concrete strain gauges (SGS)

Surface mounted strain gauges with a 30 mm gauge length were mounted on the end surface of
the wall panel as close as possible to the corner, for wall tests B to F. The arrangement of the
surface mounted strain gauges varied between tests but typically consisted of one or more
gauges centered at a height of 30 mm above the wall base, with additional gauges occasionally
placed higher up. Up to four strain gauges were used in the wall toe for each test, as seen in
Figure 3.8c, and the height of each gauge above the base of the wall is given in Table 3.3.
3.4.1.4 Embedded concrete strain gauges (SGE)

For walls E and F, embedded strain gauges were cast into the wall panel during construction.
The gauges consisted of a strain gauge with a 30 mm gauge length, surrounded by a polymer
case. Two gauges were cast into each corner of the wall, placed 20 mm in from the wall
compression edge and centered 25 mm up from the base of the wall, as shown in Figure 3.8d.
3.4.1.5 Photogrammetry image tracking (Photo)

To minimize the negative effects of traditional instrumentation, a non-contact strain
measurement technique was also investigated. For this purpose the use of photogrammetry
was trialled. Photogrammetry involves tracking the relative movement of targets from a series
of photographic images captured during a test. Although used previously for structural
measurements, photogrammetry has primarily been used to capture global displacements or to
- 49 -
monitor cracking [3-12, 3-13, 3-14]. Following the successful trials described below,
photogrammetry was used to supplement the other strain measurement techniques during the
testing of walls C2, D2, E and F. The strain measurement setup is shown in Figure 3.8c, where
up to 10 targets were placed on the side face of the wall panel.
Strain gauge
Gauge
length
(a) External displacement gauge (b) SG on reinforcement mesh
Strain gauge
(c) Photogrammetry and surface SG (d) Embedded SG
Figure 3.8 – Instrumentation used to measure concrete strains at the wall toe
- 50 -
3.4 Instrumentation
Table 3.3 – Details of instrumentation used to measure concrete strains
Wall SGR Ext SGS* SGE Photo

1 2 3 4
Phase I
1 A i Y Y
2 ii Y Y
3 iii Y Y
4 iv Y Y
5 B i Y Y 30 130 430 -
6 ii Y Y 30 130 430 -
7 iii Y Y 30 130 430 -
8 iv Y Y 30 130 430 -
Phase II
9 C1 i Y 50 100 - -
10 ii Y 50 100 - -
11 iii Y 50 100 - -
12 iv Y 50 100 - -
13 C2 i 30 30 30 30 Y
14 ii 30 30 30 30 Y
15 iii 30 30 30 30 Y
16 iv 30 30 30 30 Y
17 D1 i 30 80 230 -
18 ii 30 80 230 -
19 iii 30 80 230 -
20 iv 30 80 30 -
21 D2 i 30 30 30 30 Y
22 ii 30 30 30 30 Y
23 iii 30 30 30 30 Y
24 iv 30 30 - 30 Y
Phase III
25 E i 30 75 30 30 Y Y
26 ii 30 75 30 30 Y Y
27 iii 30 75 30 30 Y Y
28 iv 30 75 30 30 Y Y
29 F i 30 75 30 30 Y Y
30 ii 30 30 30 30 Y Y
31 iii 30 75 30 30 Y Y
32 iv 30 75 30 30 Y Y
* Columns for surface mounted strain gauges report gauge height above the wall base (mm)
3.4.2 Photogrammetry Trials

To develop a suitable photogrammetry setup and assess the accuracy for strain measurement, a
series of calibration tests were completed when crushing concrete test cylinders. The setup,
shown in Figure 3.9, used a 12.1 megapixel digital SLR camera to capture images of targets
placed on the side of the cylinder during a compression test. A software package called ImPro
[3-15] was used to process the images and calculate the relative displacements of the targets,
- 51 -
which were then converted to strain measurements. The results from two tests are plotted in
Figure 3.10 alongside a surface mounted strain gauge and externally mounted LVDT
displacement gauges. It can be seen that the photogrammetry technique captured the concrete
strain with sufficient accuracy up to failure of the cylinder and was able to calculate strains
with a resolution of approximately 50 microstrain.
(a) Test setup (b) Sample capture image

Figure 3.9 – Photogrammetry calibration tests
60 60
50 50
40 40
Stress (MPa)
Stress (MPa)
30 30
20 20
LVDT LVDT
10 10
Strain gauge Strain gauge
Photogrammetry Photogrammetry
0 0
0 1000 2000 3000 4000 0 1000 2000 3000 4000
Compressive microstrain Compressive microstrain
(a) Cylinder 1 (b) Cylinder 2

Figure 3.10 – Results from photogrammetry calibration tests
- 52 -
3.5 Test Results
3.5 TEST RESULTS

This section presents observations from the experimental tests, followed by a description of the
wall behaviour and discussion of critical results. Due to the large number of completed tests,
only a summary of the most relevant test results is presented in this section. Full results for
each test, including the measured and derived behaviour, as well as photos of the final damage
state, are presented in Appendix A.
3.5.1 General Observations

During all tests the walls typically behaved as expected. As the lateral force was applied at the
top of the wall, the behaviour was initially elastic with deformation concentrated in the wall
panel as elastic flexural and shear deformations. As the lateral force increased, a single crack
began to open up at the wall-to-foundation interface, as shown in Figure 3.11. No damage or
cracking was observed in the wall panel until minor crushing of the corner toe occurred, at
which stage loading was terminated. This damage was confined to a small area in the corner
toe, usually no higher or longer than the wall thickness, as shown in Figure 3.12 for test E-ii.
Figure 3.11 – Observed base crack opening in wall D1-i
- 53 -
(a) Side view of wall toe (b) End view of wall toe
Figure 3.12 – Observed damage in wall E-ii at the maximum lateral displacement
3.5.1.1 Phase I
As explained earlier, tests on wall A were initially conducted with no bedding material placed
between the wall and strong floor foundation. These tests resulted in an erratic force-
displacement response because uniform contact was not achieved at the wall-to-foundation
interface. Tests A-i to A-iv were then preformed with the wall seated on a cement mortar
bedding to improve the uniformity of the interface contact. However, some damage and
cracking occurred in the lower portion of the panel during the initial trials, and as a result the
behaviour of wall A was not satisfactory, with significant damage and large cracks opening in
the wall toe region. Additionally, the cement mortar was weaker than the wall concrete, which
led to extensive crushing of the bedding layer at the wall toe.
To improve the performance of the wall and in particular the bedding layer, tests on wall B
were performed using a high strength gypsum plaster bedding material. Gypsum plaster has a
higher strength than the wall concrete and so crushing of the bedding layer did not occur during
any of the four tests. The plaster bedding layer provided a uniform contact surface for the wall
panel, which ultimately lead to a more stable response and only minor crushing of the wall toe
occurred at large lateral drifts.
- 54 -
3.5 Test Results
3.5.1.2 Phase II
Tests C1-i and C1-iii were performed using a plaster bedding layer, before the same
configurations were repeated for tests C1-ii and C1-iv using a cement mortar bedding layer
when the wall was inverted. The plaster bedding layer performed well and remained intact and
uncrushed throughout the tests. In contrast, the cement mortar was weak, which led to
significant crushing of the bedding layer. Additionally, crushing of the mortar bedding
resulted in a non-uniform contact surface which caused higher levels of damage to the wall toe
when compared to the plaster equivalent.
Tests on wall C2 were preformed with a plaster bedding layer for all tests, which resulted in
satisfactory performance with only localised crushing in the wall toe at the conclusion of the
tests. However, the corners of the first end of the wall panel to be tested were damaged during
construction and repaired using the plaster bedding material. This caused problems with the
surface mounted strain gauges because the repaired corners spalled off during early stages of
the test. For this reason the tests were repeated when the wall was inverted, with more
satisfactory performance and more reliable strain measurements achieved.
The observed behaviour and damage for wall D1 was also dependent on the bedding layer.
After completing the first two tests and removing the wall, the plaster bedding layer was found
to be in a poor condition. More care was taken to construct the bedding when the walls were
inverted and the tests were repeated. When constructed correctly, the bedding layer preformed
well and provided uniform contact to the wall, as shown in Figure 3.13a. However, a poorly
constructed bedding layer (which is often a result of too much water in the mix) resulted in air
pockets and uneven contact below the wall, as shown in Figure 3.13b. The uneven contact
caused by the poor bedding layer resulted in high stress concentrations which induced crushing
of the bedding layer and in some cases caused more extensive crushing of the wall toe, as
observed during test D1-iii.
Fracture of one or more wires of the prestressing strand was observed during several tests
completed during Phase II. The failures always occurred at the anchor when stresses in the
tendon approached the idealised yield stress of 1580 MPa, which was unexpected because the
tendon strains were well below their expected failure strain. The failure was attributed to the
type of anchor used and the increased demand of repeated stressing, anchoring and de-stressing
as the tendons were reused for each test.
- 55 -
(a) Satisfactory bedding layer (b) Poor bedding layer

Figure 3.13 – Condition of the bedding layer after removal of the wall
3.5.1.3 Phase III

Walls E and F performed exceptionally well for all tests. The plaster bedding layer was well
constructed and provided full even contact throughout the tests. The observed damage was
minor and confined to a localised region in the wall toe.
3.5.2 Force-Displacement Response

Lateral force-displacement results for all 32 tests are plotted in Figure 3.14 to Figure 3.17. The
observed wall behaviour followed the typical response expected for unbonded PT members,
which includes a high initial stiffness followed by decompression at the tension toe of the wall
and opening of the base crack. Continued loading led to the base crack opening wider, until
either crushing occurred at the wall toe or the tendons approached their yield condition.
As explained earlier, the response of wall A was somewhat erratic due to issues with the
bedding layer. The initial stiffness of the wall varied considerably between the four tests, as
seen in Figure 3.14a, and this was attributed to the variable behaviour of the weak cement
mortar bedding layer. Additionally, damage to the wall toe occurred at relatively low lateral
displacements. For example, wall A-i reached just 12 mm lateral displacement (0.6% drift)
before large splitting cracks appeared on the end face of the wall and in comparison a similar
wall, F-iv, which was tested under more reliable conditions reached twice that displacement
with much less damage resulting.
- 56 -
3.5 Test Results
The measured responses for wall B tests were much more consistent, as seen in Figure 3.14b.
As the number of prestressing tendons used and initial prestress level increased, the wall
strength increased accordingly. For all four tests on wall B, loading was terminated due to
tendons reaching close to their yield state and only minor crushing was observed in the wall
toes. It is possible that the behaviour would continue to be stable for lateral displacements in
excess of those subjected to the wall during testing.
250 80
70
Lateral Force Resistance (kN)

200
60
50
150
40
100
30
i i
20
50 ii ii
iii 10 iii
iv iv
0 0
0 10 20 30 40 0 20 40 60 80
Top Lateral Displacement (mm) Top Lateral Displacement (mm)
(a) Wall A (b) Wall B

Figure 3.14 – Lateral force-displacement behaviour for walls A and B
A comparison between the repeated tests for wall C1 using plaster and mortar bedding material
are plotted in Figure 3.15a. In general the global force-displacement responses are not
significantly different. A slight reduction in initial stiffness is observed for the mortar bedding
in test C1-iv, but tests C1-i and C1-ii have almost identical initial stiffness. The most notable
difference in the responses was the ultimate displacements observed. In both cases the mortar
bedding layer resulted in crushing of the wall toe, and termination of the test, well before the
equivalent lateral displacement associated with the corresponding plaster bedding layer tests.
Figure 3.15b shows a comparison of the repeated tests for wall C2. Again, the force-
displacement responses do not differ significantly and observed differences are most likely
attributed to minor inconsistencies in the initial prestress level of the tendons. This is not
surprising, given that the tests were only repeated because successful strain measurements were
not obtained during the first two tests and differences in the plaster bedding layer condition
were considered to have negligible effect on wall performance.
- 57 -
100 140
120

80
100
60
80
60
40
i (Plaster) 40 i
20 ii (Mortar) ii
iii (Plaster) 20 iii
iv (Mortar) iv
0 0
0 20 40 60 80 0 10 20 30 40
(a) Wall C1 (b) Wall C2

Figure 3.15 – Lateral force-displacement behaviour for walls C1 and C2
The force-displacement responses of the repeated tests for wall D1 are shown in Figure 3.16a,
where tests i and iii were performed with an imperfect plaster bedding layer. Test i and ii
indicate no significant difference in performance, with similar initial and inelastic stiffnesses
and ultimate displacements. However, a significant reduction in strength was observed for
test iii when compared to test iv, which had a well constructed bedding layer. This was most
likely a result of loss of contact area in the wall toe due to trapped air pockets in the bedding
layer, as observed in Figure 3.13b.
350 450
400
300
350
250
300
200 250
150 200
150
100 i i
ii 100 ii
50 iii iii
50
iv iv
0 0
0 10 20 30 40 50 0 10 20 30 40
(a) Wall D1 (b) Wall D2

Figure 3.16 – Lateral force-displacement behaviour for walls D1 and D2
- 58 -
3.5 Test Results
The responses of wall D2 tests, shown in Figure 3.16b, highlight how the tendon’s initial
prestress level affects the lateral strength and displacement capacity. In tests with lower initial
prestress (i & iii) the ultimate strength achieved was comparable to their higher initial prestress
equivalents (ii & iv). However, the tendons were elongated further to reach this strength, so
the walls achieved larger lateral displacements before the tendons reached their yield state.
As described earlier, the response of walls E and F was consistent and reliable as a result of a
well constructed bedding layer. The force-displacement responses in Figure 3.17 confirm the
stable and dependable behaviour. As the number of prestressing tendons used and initial
prestress level increased, the walls exhibited increased strength and reduced displacement
capacity. The reduced displacement capacity is attributed to two reasons: First, higher initial
prestress levels leads to smaller elongation before the tendons reached their yield state.
Second, the increased prestress force leads to higher compressive demand on the wall toe,
which causes crushing to initiate at lower lateral displacements. Tests F-iii and F-iv were both
terminated when prestressing tendons fractured a wire at the anchor and caused a sudden loss
of lateral strength, as seen in Figure 3.17b.
100 350
300
80
250
60
200
150
40
i 100 i
20 ii ii
iii 50 iii
iv iv
0 0
0 20 40 60 80 0 10 20 30 40
(a) Wall E (b) Wall F

Figure 3.17 – Lateral force-displacement behaviour for walls E and F
3.5.3 Wall Deformation Components

When a PT wall is subjected to a lateral force, the measured lateral displacement is a function
of the gap opening (or rocking) at the base of the wall as well as elastic flexure and shear
deformations in the wall panel. To assess their contribution to the total lateral displacement,
- 59 -
each of these components was calculated from displacements measured during testing. The
displacement profile of the gauges that measured uplift at the wall base were used to calculate
the base rotation, which was multiplied by the wall height to estimate the lateral displacement
component due to gap opening. Additionally, the arrangement of gauges placed on the wall
panel was used to estimate the panel deformations due to flexure and shear using a procedure
reported by Sritharan [3-16]. The results of these calculated deformation components are
plotted in Figure 3.18 for tests C1-iii and F-ii. Additionally, the sum of these calculated
components is shown along with a dotted line which represents the actual measured lateral
deformation. For both tests the results confirm the observed behaviour of the walls. Initial
lateral displacements are a result of flexure and shear deformations of the wall panel, but as
loading increases and the base crack opens up, gap opening becomes the predominant source of
lateral displacement. At larger lateral displacements, gap opening contributes to over 95% of
the measured displacement. Additionally, the higher aspect ratio of panel C1 results in a larger
proportion of flexural deformation when compared to wall F.
60 35
 Sum  Sum
50 30
 Gap  Gap
 Flexure  Flexure
25
Deformation (mm)
Deformation (mm)
40  Shear  Shear
20
30
15
20
10
10
5
0 0
0 10 20 30 40 50 60 0 5 10 15 20 25 30 35
(a) Test C1-iii (b) Test F-ii

Figure 3.18 – Calculated wall deformation components
3.5.4 Neutral Axis Depth

The neutral axis (NA) depth, or contact length, at the base of the wall was calculated from the
profile of the displacement gauges that measured uplift along the wall base. Prior to
decompression at the tension toe, the NA depth is greater than the wall length. As the lateral
- 60 -
3.5 Test Results
force increases and the base crack starts to open up, the NA depth reduces until the maximum
concrete compressive stress is reached in the wall toe. The variation of the calculated NA
depth for all tests on walls E and F are shown in Figure 3.19. As the post-tensioning force in
the walls increased the NA depth became larger due to the increased axial load. Test E-iii
shows an unusual increase in the NA depth after the lateral displacement exceeded 40 mm.
This might explain the damage caused to the wall when the corner toe crushed significantly.
 
i i
//
//
700 ii ii
iii 1200 iii
600 iv iv
1000
500
NA (mm)
NA (mm)
800
400
600
300
400
200
100 200
0 0
0 20 40 60 80 0 10 20 30 40

Figure 3.19 – Calculated neutral axis (NA) depth for walls E and F
The calculated NA depth also provided insight into the behaviour of the bedding layer. A
comparison of the calculated NA depth for tests C1-iii and C1-iv is shown in Figure 3.20a.
The NA depth for the cement mortar bedding layer (iv) is larger than for the comparable test
using the plaster bedding layer (iii). This indicates that the low mortar strength controlled the
NA depth, rather than the strength of the wall concrete. Furthermore, the larger NA depth
explains the reduced wall strength observed in the lateral force-displacement response for the
mortar bedding layer (Figure 3.15a).
The NA depth was also found to vary depending on the condition of the plaster bedding layer.
Figure 3.20b compares the NA depth for test D1-iii, which had a poorly constructed bedding
layer, with test D1-iv, which had a good bedding layer. The air pockets observed in the
bedding layer of test D1-iii contributed to a reduced contact area which led to a larger NA
depth. Again, this explains the reason for the reduced lateral wall strength observed previously
in Figure 3.16a.
- 61 -
 
iii (Plaster) iii (Poor bedding)
//
//
700 iv (Mortar) iv (Good bedding)
1200
600
1000
500
NA (mm)
NA (mm)
800
400
600
300
400
200
100 200
0 0
0 10 20 30 40 50 60 0 5 10 15 20 25 30
(a) Wall C1 (b) Wall D1

Figure 3.20 – Comparison of neutral axis (NA) depth for bedding type and condition
3.5.5 Wall Slip

For all tests, no significant slip was observed at either the loading beam-to-wall or the wall-to-
foundation interfaces. The measured displacements across both interfaces are plotted in Figure
3.21 for test E-iv. The loading beam interface remained fixed and as a result negligible
displacement was recorded, with small variations attributed to sensitivity of the gauge.
Additionally, the displacement gauge measuring wall slip indicated some movement of up to
0.25 mm, but this was caused by wall rotation and not slip at the wall-to-foundation interface.
No sudden increases, indicating slipping, were measured by either of the displacement gauges.
0.1
-0.1
Slip (mm)
-0.2
-0.3
Wall
Loading beam
-0.4
0 10 20 30 40
Top Lateral Displacement (mm)
Figure 3.21 – Measured wall and loading beam slip for test E-iv
- 62 -
3.5 Test Results
3.5.6 Prestressing Force

As a PT wall is displaced laterally and gap opening occurs at the wall base, the post-tensioning
tendons are elongated, resulting in an increase in tendon force beyond their initial prestress.
The results for two tests are plotted in Figure 3.22, where the stress in the post-tensioning
tendons was calculated by dividing the measured force from the load cells by the nominal area
of the bar or strand. It can be seen that the increase in tendon stress varied for each tendon
depending on its position. The further the tendon was from the wall’s center of rotation
(neutral axis depth), the larger the uplift of the wall and thus the larger the increase in tendon
stress. Additionally, the stress in tendons located within the compression region of the wall toe
was observed to decrease due to shortening of the wall. Because the post-tensioning was left
unbonded, the increase in stress was kept within the elastic range. Figure 3.22b indicates some
softening of the increase in tendon stress for lateral displacements in excess of 50 mm, but this
was caused by crushing of the wall toe and subsequent shortening of the wall, and was not
attributed to yielding of the post-tensioning tendons.
1500 1500
Ten #1 Ten #1
1400
Ten #2 Ten #2
1400 Ten #3 1300 Ten #3
Ten #4
1200
Ten #5
Stress (MPa)
Stress (MPa)
1300
1100
1000
1200
900
1100 800
700
1000 600
0 5 10 15 0 10 20 30 40 50 60 70
(a) Test D2-iv (b) Test E-iii

Figure 3.22 – Measured stresses in unbonded post-tensioning
In most cases, upon release of the lateral force, the stress in the tendons returned to the initial
prestress force, with negligible losses. However, as confirmed by the tendon tensile tests, the
post-tensioning strand deviated from its elastic stiffness at approximately 1500 MPa, and in
tests where this stress was exceeded some loss of prestress was observed when the lateral force
was removed.
- 63 -
3.5.7 Concrete Compressive Strain

3.5.7.1 External displacement gauges
Measurements from the external displacement gauges mounted in the toe region of walls A and
B provided spurious results. The calculated strains never exceeded 0.003, even after crushing
of the toe was observed, and it was concluded that the 150 mm gauge length was too long to
capture the critical compressive strains in the wall toe.
The external displacement gauges used to measure strain for wall C1 were placed closer to the
critical region in the corner of the wall panel. However, measurements from these gauges
varied substantially and appeared unreliable. This is highlighted in Figure 3.23, where the
results from the two external displacement gauges (LVDTs) are plotted alongside a surface
mounted concrete strain gauge for tests C1-i and C1-iii. Although, the external gauges were
measuring strains over the same 30 mm gauge length and were placed 20 mm either side of the
surface strain gauge, the results showed poor correlation. Initially the average from the
LVDT’s was consistent with the strain gauge measurement for test C1-iii, but after 5 mm
lateral displacement one of the gauges started to deviate from its initial trend and the results
become wayward. It is possible that drilling holes on the end of the wall to insert rods to
which the external gauges were attached may have weakened the toe region of the wall and
caused localized strains to vary significantly. For this reason, external displacement gauges
were not used for strain measurement during subsequent tests.
5000 5000
Surface strain gauge
External LVDT
4000 4000
Compressive Microstrain
Ave. External LVDT
3000 3000
2000 2000
1000 Surface strain gauge 1000

External LVDT
Ave. External LVDT
0 0
0 10 20 30 40 50 60 0 5 10 15 20 25 30
(a) Test C1-i (a) Test C1-iii

Figure 3.23 – Experimental strain measurements from external displacement gauges
- 64 -
3.5 Test Results
3.5.7.2 Strain gauges on reinforcing steel

The strain gauges attached to the reinforcement mesh also provided unreliable measurements.
Recorded measurements from strain gauges attached to the reinforcing steel at heights of
25 mm and 50 mm above the base in the wall toe are plotted in Figure 3.24 for two separate
tests alongside readings from a surface mounted concrete strain gauge. For both tests the
measured strains from the strain gauges on the reinforcement mesh were much lower than
those of the surface mounted strain gauge and recorded measurements never exceeded 0.003,
even when crushing of the wall toe was observed. This indicates that the assumption of strain
compatibility between the concrete and the reinforcing steel cannot be relied on in the wall toe.
Based on these findings, the use of strain gauges attached to the reinforcement mesh was also
deemed unsuitable for this investigation and subsequently this measurement technique was not
used during the remaining tests.
4000 4000
Surface strain gauge Surface strain gauge
3500 3500
Reinforcement strain gauge Reinforcement strain gauge
3000 3000
2500 2500
2000 2000
1500 1500
1000 1000
500 500
0 0
0 20 40 60 80 100 120 0 10 20 30 40 50 60
(a) Test B-i (a) Test B-ii

Figure 3.24 – Experimental strain measurements from reinforcement strain gauges
3.5.7.3 Concrete strain gauges

A comparison between the embedded concrete strain gauges and the surface mounted strain
gauges which were centered at a height of 30 mm above the wall base is shown Figure 3.25 for
tests E-iv and F-iv. The surface gauges typically failed at strains of between 0.002 and 0.004
(2000 and 4000 microstrain), when debonding of the gauge occurred, whereas the embedded
gauges were more robust with measured strains reaching in excess of 0.01 (10000 microstrain).
All the strain gauges plotted in Figure 3.25 were located in close proximity, with the surface
mounted strain gauges distributed across the end face of the wall toe and the embedded gauges
- 65 -
located 20 mm inside from the end face. However, the measured strain magnitudes varied
dramatically between the different gauges. The scatter in measured strains resulted in a
calculated lateral displacement at the critical 0.003 strain limit that varied by up to 200%,
making it difficult to determine the point at which nominal flexural strength was reached. This
large spread in strain measurements was believed to have been contributed to by:
 variations in the concrete matrix, including aggregate position, air voids and
construction flaws,
 the small size of the region of interest (30 mm gauge length) resulting in the effects of
aggregate and air void locations being magnified,
 effectiveness of the contact between the wall and bedding material,
 micro cracks on the concrete surface,
 quality of the gauge placement.
5000 5000
4000 4000
3000 3000
2000 2000
1000 1000
Embedded strain gauge Embedded strain gauge
0 0
0 5 10 15 20 25 30 0 5 10 15 20
(b) Test E-iv (b) Test F-iv

Figure 3.25 – Experimental strain measurements from concrete strain gauges
The scatter in results from gauges located in close proximity emphasised the importance of
placing multiple gauges in the critical region. The results from earlier tests where only one
gauge was placed at this location are questionable and fail to provide sufficient information to
estimate the average strain at the extreme compression fibre.
3.5.7.4 Photogrammetry
Methods used for the photogrammetry measurement technique were improved as testing
advanced. A comparison between strains calculated from photogrammetry, a surface mounted
- 66 -
3.5 Test Results
strain gauge at the same location as the photo targets, and the average from the two embedded
concrete strain gauges are shown for tests E-iii and F-iii in Figure 3.26. As found from the
cylinder tests, photogrammetry provided accurate prediction of the concrete strains at a single
location, with the results initially correlating well with the surface mounted strain gauge.
Additionally, while the surface mounted strain gauge failed at strains of 0.003 or lower,
photogrammetry continued to provide measurements up to the conclusion of the tests (~0.007
for both tests). However, the location of the photogrammetry targets was not appropriate to
capture the average strain over the entire toe region. This is highlighted in test F-iii, where
although the photogrammetry results closely match the surface strain gauge readings, there is a
significant difference in readings when compared to the average measurement from the two
embedded strain gauges. Although photogrammetry was found to provide an accurate non-
contact technique for strain measurement at a specific location, the photo targets need to be
distributed across the end face of the wall toe so that an average strain measurement can be
accurately determined.
8000 8000
7000 7000
6000 Photogrammetry 6000 Photogrammetry
5000 5000
4000 4000
3000 3000
2000 2000
1000 1000
0 0
0 10 20 30 40 0 5 10 15 20 25 30
(a) Test E-iii (b) Test F-iii

Figure 3.26 – Experimental strain measurements from photogrammetry
3.5.7.5 Comparison of strain for wall C1

The measured concrete strains also confirmed the difference in observed behaviour between
the plaster and the mortar bedding material. A comparison of the measured strains from surface
mounted strain gauges during tests C1-iii and C1-iv is shown in Figure 3.27. It can be seen
that the measured strains were higher when the plaster bedding layer was used. This suggests
- 67 -
that the weak mortar bedding layer crushed before the concrete wall exhibited crushing, and
thus reduced the strain demand in the corner toe. Although the strain demand in the wall was
lower when the mortar bedding was used, crushing of the bedding layer was considered
undesirable as it ultimately led to premature failure and loss of strength in the wall toe.
5000
4000
3000
2000
1000
iii (Plaster)
iv (Mortar)
0
0 10 20 30 40 50 60
Figure 3.27 – Comparison of measured concrete strains for different bedding types
3.6 DISCUSSION OF RESULTS

This section contains a discussion of the experimental test results and the implications to PT
wall design, including nominal flexural strength, the influence of the bedding layer, suitability
of the rigid body assumption, prediction of the initial stiffness, shear friction at the wall-to-
floor interface, and yielding of the post-tensioning steel.
3.6.1 Nominal Flexural Strength

In all tests where reliable strain measurements were achieved, the wall exhibited a stable
response well beyond the code definition of nominal flexural strength, corresponding to a
measured strain of 0.003 at the extreme compression fibre. Additionally, no significant damage
or crushing was observed when the 0.003 strain limit was reached. This is highlighted in Figure
3.28 which shows the condition of the wall toe for two tests when one or more of the strain
gauges measured 0.003. Some minor cracking can be observed along the end edge for test C2-
iv, as well as debonding of the surface mounted strain gauges, which caused them to fail at
strains of between 0.002 and 0.004. Additionally, no visible damage is observed for test F-iv.
- 68 -
3.6 Discussion of Results
(a) Test C2-iv (b) Test F-iv

Figure 3.28 – Photos of the wall toe condition at measured strains of 0.003
Reliable measurement of the concrete compressive strains in excess of 0.003 at the wall toe
was only achieved with the robust embedded concrete strain gauges used during tests on walls
E and F. During these tests the wall behaviour was observed to remain stable to measured
concrete compressive strains levels that were well in excess of 0.003. This is highlighted in
Figure 3.29, which plots the lateral-force displacement response of two tests alongside the
average measured strain from the two embedded concrete strain gauges. In both tests the
average strain measurements are consistent until they peak at 0.009 and 0.010 respectively,
when crushing occurred in the wall toe. Additionally, in both tests the lateral force-
displacement response was stable until these peak strains were reached, and even then the walls
showed no significant loss of strength (the drop in strength for test F-ii is attributed to failure of
a prestressing tendon at the anchor, as described earlier). Interestingly, even without additional
confinement reinforcement, the wall behaviour was satisfactory to peak concrete compressive
strains well in excess of the 0.003 strain limit. This may have been due to unintentional
confinement provided by the foundation, as well as the steep curvature profile, which leads to
high strain demand in only a small region of the wall toe.
- 69 -
80 12000 300 12000
70

250 10000
10000
Compresive Microstrain
Compresive Microstrain
60
8000 200 8000
50
40 6000 150 6000
30
4000 100 4000
20
2000 50 2000
10 Force-displacement Force-displacement
0 0 0 0
0 10 20 30 40 50 60 70 0 5 10 15 20 25 30 35
(a) Test E-ii (b) Test F-iii

Figure 3.29 – Measured lateral force-displacement response correlated with average
measured compressive strains
To investigate the ultimate compressive strain limit in more detail, Figure 3.30 plots the peak
measured concrete strains from each embedded strain gauge for the eight tests on walls E and
F. The measured strains peaked at between 0.0045 to over 0.015 (4500 to over 15000
microstrain) before localized crushing occurred (measurements in excess of 0.015 are indicated
with an ×). These results indicate that a maximum concrete compressive strain of 0.003 may
be too conservative for PT walls and that a value of 0.005 or above may be more appropriate,
as shown in Figure 3.30.
15000
Maximum Compressive Microstrain
10000
0.005
5000
0.003
0
E-i E-ii E-iii E-iv F-i F-ii F-iii F-iv
Figure 3.30 – Maximum measured compressive strains from embedded strain gauges
- 70 -
As seen from Figure 3.30, tests with lower levels of prestress appeared to withstand higher
strain demand before crushing occurs. This may be due to the small compression zone in the
corner toe, with only a small region having strains in excess of 0.003. To investigate this
relationship further, the peak measured concrete strains from each wall test were re-plotted in
Figure 3.31 against the ratio of the wall axial stress, fc, to the concrete crushing strength, fc',
which is defined as the axial stress ratio. It can be seen that as the axial stress ratio increases,
the peak measured concrete strain decreases, and that for axial stress ratios of less than 0.07, an
ultimate concrete compressive strain of greater than 0.005 may be suitable. The formula
shown in Eq. 3.1, and plotted in Figure 3.31, could instead be used to predict the maximum
concrete compressive strain for the design of post-tensioned concrete walls.
15000
Eq. 3.1
Maximum Compressive Microstrain
10000
0.005
5000
0.003
0
0.02 0.04 0.06 0.08 0.1 0.12
Axial Stress Ratio (fc / fc ')
Figure 3.31 – Maximum measured compressive strains against axial stress ratio
 fc 
 cu  0.012  0.1    0.005
'  (3.1)
f
 c
where
f se A ps  N
fc  (3.2)
l w bw
where fse is the initial stress in prestressing steel after losses, Aps is the prestressing steel area,
N is the axial load due to wall self-weight and additional dead and live loads, lw is the wall
length and bw is the wall thickness.
- 71 -
3.6.2 Bedding Layer

The test results confirmed that the bedding layer is an important parameter that influences the
performance of the wall. Initial tests indicated that a bedding layer was essential to achieve
uniform contact and even stress distribution at the wall-to-foundation interface.
The use of a bedding material that is weaker than the concrete wall, such as cement mortar, has
some advantage as it crushes before the concrete wall toe crushes and so could potentially be
used to protect the wall and reduce the strain demand for small lateral displacements.
However, there are several major disadvantages of using a weak bedding material that were
observed during the experimental tests, including:
 the weak bedding layer crushed prematurely, which reduced the maximum lateral drift
capacity,
 crushing of the weak bedding layer led to unstable behaviour in the wall toe which
resulted in higher damage to the wall,
 the weak bedding layer reduced the initial stiffness of the wall,
 the quality and strength of the cement mortar bedding layer appeared to vary
significantly, which led to less reliable wall performance.
To minimise the influence of the bedding layer on wall performance, the bedding material
should be stronger and stiffer than the concrete wall. This eliminates the uncertainties
associated with a weak bedding layer and leads to a more dependable structural performance.
During the experimental tests, gypsum plaster was shown to provide an excellent bedding
material but it was important to ensure that the bedding layer was constructed so that uniform
contact was achieved below the wall. Observed damage was isolated to the wall toe where it
would be easily accessible for repair if necessary. However, plaster is not an ideal material for
use in real construction and was only used during the tests due to convenience of the short
curing time (24 hours). Recent guidelines published by ACI, covering the use of precast walls
with unbonded post-tensioning [3-11], recommend the use of high strength grout that is more
durable than the gypsum plaster used during the tests. Additionally, for seismic applications,
fibres should be added to the grout to prevent deterioration of the bedding layer during
repeated loading cycles.
- 72 -
3.6.3 Rigid Body Assumption

The results from the experimental tests confirmed that when a concrete wall with unbonded PT
is subjected to a lateral force, wall deformation is concentrated at the single crack that opens up
at the wall base. For the wall configurations tested, gap opening accounted for over 95% of the
total lateral displacement. The remaining lateral displacement was accounted for through
flexure and shear deformations of the concrete wall panel, which remained within the elastic
region.
For analysis purposes, the lateral load behaviour of an unbonded PT wall can be approximated
by ignoring the flexure and shear deformations of the panel and considering the wall as a rigid
body. The rigid body will rotate about a point located at the end of the contact length along the
wall base.
3.6.4 Initial Stiffness

The rigid body assumption is only reliable for large displacements and the elastic stiffness of
the wall prior to gap opening should be determined by calculating the elastic flexure and shear
deformation of the wall panel. The elastic response can be calculated using the uncracked
section properties and so the lateral-displacement components due to flexure (∆Flexure) and
shear (∆Shear) at the top of the wall can be represented by Eqns. 3.3 and 3.4.
F  he3
 Flexure  (3.3)
3E c I g
F  he
 Shear  (3.4)
G c l w bw
where F is the lateral force applied at height he, Ec is the concrete modulus of elasticity, Gc is
the shear modulus of concrete, and Ig is the second moment of inertia of the gross section.
The accuracy of using the uncracked or gross section properties to calculate the elastic flexure
and shear deformations is demonstrated in Figure 3.32 and Figure 3.33, which plot the
measured flexural and shear deformations from tests on walls E and F against Eqns. 3.3 and
3.4. The concrete modulus of elasticity was calculated using Ec  3200 f c'  6900 MPa , as
recommended by the New Zealand concrete design standard [3-4], and an assumed Poisson’s
- 73 -
ratio of 0.2 was used to calculate the shear modulus. In general a good correlation was
achieved which confirmed that the panel remained in an uncracked elastic state throughout the
test. The accuracy is especially good for low lateral forces, indicating that Eqns. 3.3 and 3.4
can be used to accurately predict the initial stiffness of the PT concrete wall.
3.5 1.6
3 1.4
Flexural deformation (mm)
Flexural deformation (mm)

1.2
2.5
1
2
0.8
1.5
0.6
1
0.4
0.5 Calculated 0.2 Calculated

Measured Measured
0 0
0 20 40 60 80 100 0 50 100 150 200 250 300 350
Lateral force (kN) Lateral force (kN)

Figure 3.32 – Measured and calculated flexural deformation using Eq. 3.3
0.3 0.6
0.25 0.5
Shear deformation (mm)
Shear deformation (mm)
0.2 0.4
0.15 0.3
0.1 0.2
0.05 0.1
Calculated Calculated
Measured Measured
0 0
0 20 40 60 80 100 0 50 100 150 200 250 300 350
Lateral force (kN) Lateral force (kN)

Figure 3.33 – Measured and calculated shear deformation using Eq. 3.4
- 74 -
3.6.5 Shear Friction

For all the walls tested, shear friction was sufficient to prevent any slip at the wall-to-
foundation interface. The highest friction demand occurred during test D2-iii. Wall D2 had a
low aspect ratio of 1.1 and test D2-iii was subjected to a high initial prestress using five
prestressing tendons, which resulted in a required coefficient of friction of at least 0.475 to
prevent sliding. Although thorough investigation of the friction coefficient at the interface was
not investigated, this wall configuration is close to the upper bound of what would be expected
for building applications.
Although shear friction at the wall-to-foundation interface was shown to be sufficient to

prevent any slip, Appendix B of NZS 3101:2006 [3-4] requires shear dowels to be placed
across the joint at the base of unbonded PT walls.
3.6.6 Yielding of Post-tensioning Steel

It is not desirable for the post-tensioning tendons to yield because this results in a loss in
prestress when the wall is unloaded. Both the material tests and the wall tests confirmed that
the prestressing strand that was used remained fully elastic up to approximately 1500 MPa.
This is equal to 95% of the idealised yield stress assumed during design practice. Therefore, to
ensure that yielding of the post-tensioning does not occur, stresses in the tendons should be
kept below 0.95fpy at the ultimate wall lateral displacement. This is consistent with
Appendix B of NZS 3101:2006 [3-4] and ACI ITG 5.1 [3-11], which recommend a more
conservative 0.9fpy limit for prestressing in PT walls.
When considering the design of PT walls, the wall tests reported here indicated that a lower
initial prestress, such as 0.5fpy, will typically lead to higher displacement capacity because the
prestressing tendons are allowed to elongate further before reaching their yield state.
Additionally, placing the tendons near the edges of the wall panel leads to a reduced
displacement capacity because the tendon elongation is greater than for a tendon located at the
center of the wall.
- 75 -
3.7 CONCLUSIONS
The behaviour of individual precast concrete wall panels with unbonded post-tensioning was
examined by extensive experimental testing. The wall panels behaved as was expected for a
member with unbonded PT, with deformation concentrated at a single crack that opened up at
the base of the wall and only minor localised damage occurring in the wall toe.
The bedding layer was found to be a critical parameter affecting the wall performance. From
the tests, it was concluded that it is most desirable to use a bedding material that is stronger
than the concrete wall, as a strong bedding layer eliminates uncertainties associated with
crushing and results in more dependable response and a higher lateral displacement capacity.
It is also important to ensure that the bedding layer is constructed correctly so that uniform
contact exists at the wall-to-foundation interface.
The post-tensioning tendons remained predominantly elastic, which meant that the walls self-
centered with minimal residual drift observed after unloading. It is recommended that the
stresses in prestressing tendons do not exceed 0.95fpy in order to ensure that the tendons remain
fully elastic.
Over 95% of the wall lateral displacement was due to gap opening at the wall base. As a
result, rigid body rotation can be assumed during analysis to predict wall lateral load behaviour
with sufficient accuracy. However, wall initial stiffness should be calculated from elastic
flexure and shear deformation using uncracked gross section properties.
Accurate and reliable measurement of concrete compression strains at the extreme compressive
fibre in PT walls is difficult due to the steep increase in strains in the vicinity of the wall toe.
Additionally, a large variation in the observed readings was obtained because localised
behaviour strongly influenced the gauge reading. It is recommended that an average from
several measurement devices at the wall toe is required to accurately quantify when nominal
flexural strength is reached.
Embedded concrete strain gauges performed the best out of the several alternative
instrumentation techniques used to measure concrete strains. The embedded gauges provided
consistent measurements up to strains in excess of 0.01. Additionally, a photogrammetry
image tracking technique was developed to provide a non-contact option for measuring
- 76 -
3.8 References
concrete strains. Preliminary results indicated that the technique can provide sufficient
accuracy and greater robustness when compared to traditional strain gauges. However, care
must be taken to ensure that the photogrammetry targets are spread throughout the critical
region so that an average measurement can be obtained.
The lateral load resistance of the wall was maintained well beyond the code defined ultimate
concrete compressive strain of 0.003, and minimal concrete damage was observed at this limit
state. Experimental strain measurements indicated that a higher compressive strain limit of
0.005 or greater may be more suitable for describing the nominal flexural strength of PT
concrete walls.
3.8 REFERENCES
[3-1] Whitney, C. S. (1937) Design of reinforced concrete members under flexure or
combined flexure and direct compression. American Concrete Institute -- Journal, 8(4),
483-498.
[3-2] Mattock, A. H., Kriz, L. B., and Hognestad, E. (1961) Rectangular concrete stress
distribution in ultimate strength design. American Concrete Institute -- Journal, 32(8),
875-928.
New Zealand.
[3-6] Kurama, Y. C. (2002) Hybrid post-tensioned precast concrete walls for use in seismic
regions. PCI Journal, 47(5), 36-59.
Journal, 44(6), 42-67.

1560-1570.
- 77 -
[3-9] Sritharan, S., Aaleti, S., Henry, R. S., Liu, K. Y., and Tsai, K. C. (2008) Introduction to
PreWEC and key results of a proof of concept test. M. J. Nigel Priestley Symposium,
North Tahoe, California, IUSS Press, Pavia, Italy, 95-106.
[3-10] Walsh, K. Q. and Kurama, Y. C. (2010) Behavior of unbonded post-tensioning

monostrand anchorage systems under monotonic tensile loading. PCI Journal, 55(1),
97-117.
[3-12] Lange, J. and Benning, W. (2006) Crack detection using photogrammetry. ECNDT
2006, Berlin, Germany.
[3-13] Maas, H. G. and Hampel, U. (2006) Programmetric techniques in civil engineering

material testing and structure monitoring. Photogrammetric Engineering and Remote
Sensing, 72(1), 39-45.
[3-14] Sonnenberg, A. M. C. and Al-Mahaidi, R. (2007) Investigation of dowel shear in RC

beams using photogrammetry. Magazine of Concrete Research, 59(9), 1-626.
[3-15] Yang, Y. S., Wu, C. L., Tu, W. H., Yang, C. M., Chen, A. C., and Loh, C. H. (2006)
Preliminary study on image-based measurement and ImPro package. 2006 NCREE
Research Programs and Accomplishments.
[3-16] Sritharan, S. (1998) Analysis of concrete bridge joints subjected to seismic actions.
PhD thesis. University of California, San Diego, CA.

- 78 -
Chapter 4
Analysis and Design of Post-tensioned
Wall Panels
Following the experimental investigation that was described in Chapter 3, this chapter
addresses aspects of the analysis and design of individual post-tensioned (PT) concrete walls.
The experimental tests highlighted the challenges associated with acquiring accurate and
reliable concrete strain measurements in the wall toe and it was therefore decided that
analytical techniques were required to assist in quantifying the compressive strains and the
ultimate limit state. For this purpose, detailed finite element modelling was used, starting with
a 2D representation before focusing on a 3D representation that provided improved accuracy.
An extensive parametric study was conducted to assess the accuracy of current design
equations and assist is developing refined design procedures.
As discussed in previous chapters, PT walls with no specific confinement reinforcement could

be designed in accordance with current force-based design standards for application in regions
with low seismicity. Design procedures that can predict the backbone response of an unbonded
PT concrete wall have been previously published [4-1, 4-2], but because confinement
reinforcement was placed on the wall toe, the nominal flexural strength was not considered.
- 79 -
Chapter 4 Analysis and Design of Post-tensioned Wall Panels
Additionally, the nominal flexural strength is not addressed by the design recommendations
published in Appendix B of NZS 3101:2006 [4-3] or ACI-ITG 5.2 [4-4]. This chapter
addresses the calculation of the nominal flexural strength of PT walls, which relies on
calculation of the stress in the unbonded PT tendons. Current equations for predicting the
unbonded tendon stresses at nominal flexural strength were found to be inaccurate, and so a
new equation was developed based on the mechanics of a rocking PT wall which accurately
predicted the unbonded tendon stresses, and thus the nominal flexural strength. Additionally,
an equation for predicting the displacement at nominal flexural strength was also developed.
4.1 FINITE ELEMENT MODELLING

Detailed finite element models (FEMs) of the test walls described in Chapter 3 were
constructed to supplement the experimental results and assist in understanding and quantifying
the concrete strains in the wall toe. For this study, finite element software ABAQUS [4-5] was
selected because of its detailed material models and sophisticated contact algorithms, that are
capable of modelling wall uplift. Modelling was initially conducted using a 2D FEM
representation, similar to that used Allen and Kurama [4-6] and Wight and Ingham [4-7].
However, the investigation later shifted to a more detailed 3D FEM representation that was
able to overcome several limitations that were identified with the 2D FEM.
4.1.1 2D Model
4.1.1.1 Model description
Concrete wall panels were modelled using 2D plane stress elements. The wall material was
modelled using the “concrete damaged plasticity model”, ignoring the embedded reinforcing
steel. The concrete compressive stress-strain behaviour was approximated by defining a
parabolic function based on the concrete crushing strength, fc', as described by Park and Paulay
[4-8]. However, the slope of the strain softening path was reduced to minimise convergence
problems during the analysis using a constant slope from peak stress to 0.1 fc' at a compressive
strain of 0.01. Trial runs indicated that the wall behaviour was not sensitive to the chosen
strain softening slope and that the model remained accurate up to and beyond compressive
strains of 0.003. The concrete tensile stress-strain behaviour was modelled in a similar manner
with an assumed maximum tensile stress of 0.62 f c' (MPa), which is consistent with
- 80 -
4.1 Finite Element Modelling
recommendations by Paulay and Priestley [4-9]. To minimise problems with solution

convergence, the tensile stress was capped at the peak stress with no strain softening, and trials
indicated that this capped tensile stress had minimal impact on the model accuracy. An
example of the calculated concrete stress-strain curve that was used in the FEM is shown in
Figure 4.1 alongside the measured response of a concrete test cylinder.
-45
-40 FEM model

Cylinder Test
-35
-30
Stress (MPa)
-25
-20
-15
-10
-5
5
0 0.002 0.004 0.006 0.008 0.01
Strain (m/m)
Figure 4.1 – Concrete stress-strain response assumed during FEM
The foundation was idealised using an analytically rigid surface, and the horizontal joint
between the wall and foundation was modelled using a contact interaction providing “hard
contact” under compressive stresses while allowing unrestricted “gap opening”. Additionally,
the friction mechanism for transferring shear forces was sufficient to prevent any slip occurring
during experimental testing, so a “no slip” condition was used to prevent horizontal wall
sliding during the analysis. The steel channel loading beam at the top of the wall was modelled
using rigid member which was constrained to the top of the wall with no slip or uplift allowed.
The unbonded tendons were modelled as truss elements and given nonlinear stress–strain
properties that were found from the material testing described previously in section 3.2.3.2.
The tendon lengths were equal to the unbonded length measured during testing. The top
tendon anchorage was modelled by coupling the horizontal and vertical degrees of freedom of
the top tendon node with a corresponding node on the loading beam. This constrained the
displacement of the top tendon anchorage to the displacement of the wall panel. The bottom
tendon anchorage was represented by restraining the horizontal and vertical degrees of freedom
of the bottom tendon node.
- 81 -
Loading was applied using a series of analysis steps. First, the tendons were prestressed by
specifying an initial stress in the truss elements and the top of the tendons were initially
restrained to prevent any displacement due to this initial stress condition. The tendon restraints
were removed during the first load step which allowed the prestress force to be transferred into
the wall as a precompression. Following the prestress step, the wall self-weight was applied as
a gravity load during the second load step. Lastly, the end node of the loading beam at the top
of the wall was subjected to a displacement controlled monotonic lateral loading history. The
lateral displacement was applied during a “static riks” analysis step using the implicit
(standard) solver available within ABAQUS.
The meshed assembly of the 2D FEM is shown in Figure 4.2. A rectangular mesh of 50 mm
square was used for the wall panel. A mesh sensitivity study found that further reducing the
mesh size had no significant effect on the calculated wall response but it did cause convergence
problems within the analysis.
Figure 4.2 – 2D FEM representation of test wall E
4.1.1.2 Validation of the 2D FEM

A comparison between the experimental test data reported in Chapter 3 and the 2D FEM
calculated lateral force-displacement response for tests E-iii and F-iv is shown in Figure 4.3.
- 82 -
In general the 2D FEM showed reasonable correlation with the experimental data but
overestimated both the strength and the initial stiffness. Additionally, the 2D analysis
terminated early when crushing of the wall toe was predicted, and for both analyses this
occurred well before crushing was observed during the experimental tests. It was believed that
these inaccuracies were contributed to by the idealised conditions used in the model.
Primarily, the wall-foundation interface was not well captured using a rigid foundation
element, leading to overestimation of the wall stiffness and a higher demand on the concrete in
the wall toe, which resulted to premature crushing being calculated.
90 350
80
300

70
250
60
50 200
40 150
30
100
20
Test 50 Test
10
2D FEM 2D FEM
0 0
0 10 20 30 40 50 60 0 5 10 15 20 25
(a) Test E-iii (b) Test F-iv

Figure 4.3 – Calculated lateral force-displacement response using 2D FEM
Strain in the wall toe region was calculated from the FEM by extracting the total vertical strain
from the element in the corner of the wall panel. The strains calculated by the 2D FEM for
tests E-iii and F-iv are shown in Figure 4.4, alongside the average measured strains from both
the surface and the embedded strain gauges at the same location as the corner element. It can
be seen that the 2D FEM significantly over-predicted the measured compressive strains. For
test E-iii, the strain calculated the 2D FEM led to an estimated lateral displacement at nominal
flexural strength, or strain equal to 0.003, that was 40% below the displacement observed from
the measured strains. This highlights the earlier finding that the 2D FEM inadequately
captured the flexibility of the wall-to-foundation interface, which led to a higher strain demand
and premature crushing of the wall toe.
- 83 -
4000 4000
3500 3500
3000 3000
2500 2500
2000 2000
1500 1500
1000 1000
2D FEM 2D FEM
500 Embeded gauge 500 Embeded gauge
Surface gauge Surface gauge
0 0
0 5 10 15 20 25 30 35 0 5 10 15
(a) Test E-iii (b) Test F-iv

Figure 4.4 – Calculated compressive strains using 2D FEM
4.1.2 3D Model
To improve the modelling accuracy at the wall-to-foundation interface, trials were conducted
using a flexible foundation block with the 2D FEM representation. However, properties such
as the width of the foundation block were difficult to define using 2D plane-stress elements. In
addition, the 2D representation failed to accurately capture the distribution of the wall
compression stresses into the foundation, as well as the lateral expansion of the wall toe. To
overcome the limitations and inaccuracies of the 2D representation, a 3D FEM was developed
to replicate the test setup with increased detail. In particular, the 3D FEM allowed the
introduction of a more realistic foundation that allowed for the compressive stress in the wall
toe to be distributed into the foundation in all directions.
4.1.2.1 Model description

The assembled 3D FEM is shown below in Figure 4.5. The concrete wall panel was modelled
with 3D solid brick elements using the same concrete damaged plasticity model that was
described for the 2D FEM. A rectangular mesh of approximately 50 mm square was used for
the wall, except that the element length along the base of the wall was reduced from 50 to
30 mm in the compressed wall toe. The reduced element size in the wall toe was chosen
because it best represented the location and length of the strain measurement instrumentation,
allowing for direct comparison with experimentally acquired results. Additionally, three
elements were used through the thickness of the wall panel. A mesh sensitivity study found
- 84 -
that further reducing the element size had no effect on the global response of the wall, and thus
a mesh size in the wall toe equivalent to the length and position of the strain measurement
instrumentation was suitable. The steel reinforcement mesh was also included in the 3D FEM
by modelling the bars discretely and placing them inside the wall as embedded elements.
Figure 4.5 – 3D FEM representation of test wall E
Instead of the idealised rigid foundation used for the 2D FEM, a section of the 460 mm thick
strong floor foundation that the test walls were seated on was included in the 3D FEM using
solid brick elements. The foundation elements used the same concrete damaged plasticity
model as the wall, based on its assumed compressive strength of 50 MPa and with an
approximately 100 mm mesh size. Additionally, the 5 mm thick gypsum plaster bedding layer
was included in the model using the concrete damaged plasticity model with an assumed
compressive strength of 70 MPa. Uplift was typically observed to occur at the wall-to-bedding
interface during testing and so the same contact interaction described earlier for the 2D FEM
was used between the wall and the bedding layer. The lower surface of the bedding layer was
constrained to the strong floor foundation with no uplift or slip was allowed at this interface.
- 85 -
The bottom tendon anchorages were represented by restraining the horizontal and vertical
degrees of freedom to the end of a load cell. The load cell was modelled as an elastic 3D
cylinder corresponding to the vertical stiffness of the load cells that were used during the
experimental tests. The top surface of the load cell was constrained to the bottom of the
foundation, as it was during testing.
The rigid loading beam, tendon prestressing, self-weight, and applied lateral loading were all
modelled in the same way as for the 2D model description in section 4.1.1.1. Additionally, as
the lateral displacement was applied, the wall was restrained from movement in the out-of-
plane direction, as it was during testing.
4.1.2.2 Validation of the 3D model

The typical behaviour of the 3D FEM is shown below in Figure 4.6. As with the experimental
tests, gap opening occurred at the wall-to-foundation interface and high stresses and strains
developed in a localised region of the wall toe. Additionally, high contact stresses were
observed in the foundation at the wall toe with minor flexibility and deformation of the
foundation occurring. The wall toe was also found to bulge or expand in the out-of-plane
direction, which was not able to be captured when 2D plane stress elements were used.
(a) Elevation view (b) Gap opening at the wall base

Figure 4.6 – Displaced shape of the 3D FEM of test E-iii at maximum lateral
displacement (magnified 3 times) alongside the plotted stress profile
- 86 -
Experimental results from experimental test walls E and F were used to validate the 3D FEM
because the most reliable strain measurements were achieved for these walls. The increased
detail of the 3D FEM led to improved estimation of the wall response. A comparison between
the measured and 3D FEM calculated lateral force-displacement responses for tests on walls E
and F is shown in Figure 4.7. The initial stiffness, decompression point and strength of the
walls were closely matched for all eight tests. The 3D FEM response did deviate from some of
the experimental curves during later stages because concrete crushing in the wall toe was not
accurately captured by the idealised concrete strain softening material definition. However,
this deviation occurred well beyond the region of interest with strains usually in excess of
twice the 0.003 strain limit. The improved accuracy when compared to the 2D FEM was
attributed to more accurate representation of the wall-to-foundation interface and to the
inclusion of 3D effects in the wall panel.
100 350
300
80
250
60
200
150
40
100
20
Test 50 Test
3D FEM 3D FEM
0 0
0 10 20 30 40 50 60 0 10 20 30 40

Figure 4.7 – Calculated lateral force-displacement response using 3D FEM
In addition to capturing the global force-displacement response, comparison of the localised

behaviour in regions of the PT wall was also used to confirm the accuracy of the 3D FEM. A
comparison of the measured and calculated unbonded tendon stress, compressive strains in the
wall toe, neutral axis depth, and base rotation are shown for wall test E-iii in Figure 4.8 and for
wall test F-iv in Figure 4.9. In general, the 3D FEM captured these localised behaviours well,
confirming the accuracy of the modelling technique.
- 87 -
Estimation of the unbonded tendon stresses is critical to the calculation of the wall strength and
the 3D FEM showed good correlation with the measured values. Some deviation was observed
in the later stages of loading, but as with the global response, this was attributed to crushing of
the wall toe during experimental tests which caused wall shortening and thus a softening in the
slope of the increase in tendon stress. Additionally, a unusually poor correlation was observed
between the measured and 3D FEM calculated stress for the first tendon in test F-iii, as shown
in Figure 4.9a.
1600 3000
Test
3D FEM 2500
1400
2000
Stress (MPa)
1200
1500
1000
1000
800 Embeded gauge

500
Surface gauge
3D FEM
600 0
0 10 20 30 40 50 60 70 0 5 10 15 20 25 30
(a) Unbonded tendon stress (b) Concrete strain
 0.025
Test
//
3D FEM
600 0.02
Base Rotation (rad)
500
0.015
NA (mm)
400
300 0.01
200
0.005
100 Test
3D FEM
0 0
0 10 20 30 40 50 60 70 0 10 20 30 40 50 60 70
(c) Neutral axis (NA) depth (d) Base rotation

Figure 4.8 – Calculated localised wall behaviour using the 3D FEM for test E-iii
- 88 -
1800 3000
Test
1600 3D FEM 2500
1400 2000
Stress (MPa)
1200 1500
1000 1000
Embeded gauge
800 500
Surface gauge
3D FEM
600 0
0 5 10 15 20 25 30 35 0 5 10 15
(a) Unbonded tendon stress (b) Concrete strain
 0.012
Test
//
3D FEM 0.01
1200
Base Rotation (rad)
1000 0.008
NA (mm)
800
0.006
600
0.004
400
0.002
200 Test
3D FEM
0 0
0 5 10 15 20 25 30 35 0 5 10 15 20 25 30 35
(c) Neutral axis (NA) depth (d) Base rotation

Figure 4.9 – Calculated localised wall behaviour using the 3D FEM for test F-iii
Strains in the critical toe region were calculated from the 3D FEM by extracting the total
vertical strain from the center element in the corner toe of the wall. This calculated
compressive strain showed good correlation with the average measured strains from both the
surface and embedded strain gauges that were used during the tests, especially within the
critical strain range of between 0 and 0.003. As shown in Figure 4.9b, there was a significant
difference between the average measurements from the embedded and surface strain gauges for
test F-iii. However, the 3D FEM calculated strain for test F-iii was in-between the average
- 89 -
measurements from the embedded and surface strain gauges, and thus correlated well with the
total average measured strains. This accuracy confirmed that although the measured strains
from individual strain gauges were scattered, the average strain was accurately captured by the
3D FEM and thus the 3D FEM can be used for accurate estimation of the strains at the extreme
compression fibre of the PT wall.
The calculated neutral axis depth and base rotation also correlated well with the values
calculated from the experimental measurements. This confirmed that the uplift, contact and
rotation at the base of the wall were also accurately captured by the 3D FEM.
4.2 PREDICTING NOMINAL FLEXURAL STRENGTH

To enable PT walls to be designed using current force-based design codes, accurate prediction
of the nominal flexural strength is required. The nominal flexural strength was not considered
during the development of self-centering wall systems that are designed for regions with high
seismicity [4-1, 4-2]. As stated previously, according to current concrete design standards [4-
10, 4-11], the wall is deemed to have reached nominal flexural strength when the strain in the
extreme compression fibre of the critical section reaches the code defined ultimate compressive
strain, εcu, of 0.003.
The in-plane behaviour of a PT wall at nominal flexural strength is shown in Figure 4.10. The
lateral strength, or moment resistance, of the wall relies on the unbonded prestressing tendons
and the axial load, which represents the self-weight of the wall and the surrounding structure.
For this reason, prediction of the stresses in the unbonded tendons is critical to accurate
estimation of the wall nominal flexural strength. Compared to members with bonded
reinforcement, where the stresses in reinforcement are found through strain compatibility with
the surrounding concrete, predicting stresses in unbonded tendons is more difficult. This is
because strain compatibility between the concrete and the unbonded tendons does not exist at
the section level, but instead the change in tendon stress is related to the wall deformation
between the two tendon anchorages. This phenomenon significantly complicates the
calculation of tendon stresses, and thus semi-empirical equations are favoured to simplify the
design of systems with unbonded PT.
- 90 -
4.2 Predicting Nominal Flexural Strength
lw
he Axial Load
PT force
∆pt
c
θ
Foundation d2
d1
Figure 4.10 – Flexural behaviour of an unbonded PT wall
4.2.1 Data Set for Validation

To investigate the accuracy of current equations for predicting the unbonded tendon stresses at
nominal flexural strength and develop refined design procedures for PT walls, a data set of
typical wall details was compiled and analysed using the 3D FEM described in section 4.1.2.
Instead of using the test data, the FEM was used for two reasons. First, as discussed
previously, a large variation in the experimental strain data was observed, whereas the 3D FEM
provided accurate and stable strain calculation. Second, the data set used was much more
comprehensive due to the ease in running multiple FEM analyses, compared with expensive,
time-consuming laboratory tests. The 3D FEM model used to analyse the data set was similar
to the 3D FEM that was validated with the experimental data in section 4.1.2. However, in
order to replicate actual construction, the load cells were removed from the 3D FEM and the
bottom of the tendons were instead anchored directly to the base of the foundation. The wall
models were loaded with a monotonically increasing lateral displacement. The wall behaviour
was recorded and the key response parameters were found when the concrete strains in the
corner element of the wall base reached 0.003, representing the theoretical nominal flexural
strength state.
- 91 -
Table 4.1 – Details of FEM wall data set
Wall lw he he/lw bw Ten# Aps fse fc ' fc / fc'

No. (m) (m) (m) (mm2) (%fy) (MPa)
1 1.00 4.50 4.5 0.120 2 293.1 50 30 0.068
2 1.00 4.50 4.5 0.120 3 439.65 70 30 0.139
3 1.00 4.50 4.5 0.120 6 879.3 50 30 0.197
4 1.00 4.50 4.5 0.120 3 439.65 70 50 0.083
5 1.00 4.50 4.5 0.120 6 879.3 70 50 0.164
6 1.00 3.50 3.5 0.140 2 293.1 50 35 0.050
7 1.00 3.50 3.5 0.140 3 439.65 50 35 0.073
8 1.00 3.50 3.5 0.140 3 439.65 70 35 0.102
9 1.00 3.50 3.5 0.140 6 879.3 50 35 0.144
10 1.00 3.50 3.5 0.140 6 879.3 70 35 0.201
11 1.00 3.00 3 0.120 2 293.1 50 40 0.050
12 1.00 3.00 3 0.120 3 439.65 70 40 0.103
13 1.00 3.00 3 0.120 6 879.3 70 40 0.204
14 1.00 3.00 3 0.120 3 439.65 70 50 0.083
15 1.00 3.00 3 0.120 6 879.3 70 50 0.164
16 2.00 4.50 2.25 0.120 3 439.65 50 40 0.039
17 2.00 4.50 2.25 0.120 5 732.75 70 40 0.083
18 2.00 4.50 2.25 0.120 10 1465.5 75 40 0.184
19 2.00 4.50 2.25 0.180 5 732.75 70 40 0.059
20 2.00 4.50 2.25 0.180 10 1465.5 75 40 0.123
21 2.00 3.00 1.5 0.120 3 439.65 50 35 0.043
22 2.00 3.00 1.5 0.120 5 732.75 50 35 0.071
23 2.00 3.00 1.5 0.120 5 732.75 70 35 0.099
24 2.00 3.00 1.5 0.120 10 1465.5 50 35 0.140
25 2.00 3.00 1.5 0.120 10 1465.5 75 35 0.209
26 2.00 2.00 1 0.120 3 439.65 50 39.5 0.038
27 2.00 2.00 1 0.120 5 732.75 50 39.5 0.062
28 2.00 2.00 1 0.120 5 732.75 70 39.5 0.083
29 2.00 2.00 1 0.120 10 1465.5 50 39.5 0.123
30 2.00 2.00 1 0.120 10 1465.5 75 39.5 0.184
#: number of tendons

Table 4.1 shows the matrix of 30 walls included in the parametric study. The wall matrix was
designed to vary major parameters within the expected range for PT concrete walls designed
for application in low-rise buildings. Six different wall height/length ratios were used, varying
from 1 to 4.5. Wall thickness was altered in several walls to analyse its effect. The concrete
strength was varied between 30 and 50 MPa, as this was considered the typical range for
standard precast elements. Most notably the number, location, and initial stress in the
prestressing tendons was varied. Tendon locations were the same as for the tested walls
reported in Chapter 3, with three ducts for walls with a length of 1 m and five ducts for walls
with a length of 2 m. Walls with 6 or 10 tendons were designed with two tendons at each duct
location. For all walls the unbonded tendon length, lp, was equal to the effective wall height,
- 92 -
he, plus an additional 0.500 m to account for the tendon anchorages at both the foundation and
the top of the wall. Apart from the wall self weight, no additional axial load was applied.
However, varying the number of tendons and their initial prestressing stress meant that the
initial axial stress on the wall was also varied. This resulted in an axial stress ratio (i.e. initial
stress/compressive strength) ranging from 0.038 (3.8%) to 0.21 (21%). Tendons were
distributed along the wall length, resulting in multiple tendon calculations for each wall. In
total the wall matrix used for the parametric study provided 114 unique tendon configurations,
giving a large data set to investigate the design equations.
4.2.2 Predicting Tendon Stress

The wall data set was used to assess the accuracy of current code equations, as well as recently
proposed equations for predicting unbonded tendon stresses in post-tensioned members at
nominal flexural strength. Following this, a revised equation was developed which was shown
to estimate tendon stresses with increased accuracy.
4.2.2.1 Review of existing equations

Both the NZ and US concrete design standards provide Eq. 4.1 and Eq. 4.2 to estimate the
tendon stresses, fps, in members designed with unbonded prestressing depending on the span-
to-depth ratio. These appear as Equations 19-8 and 19-9 in NZS 3106:2006 [4-11] and in US
customary units as Equations 18-4 and 18-5 in ACI 318-08 [4-10].
For span-to-depth ratio < 35,
f c'
f ps  f se  70  MPa  (4.1)
100  p
For span-to-depth ratio > 35,
f c'
f ps  f se  70  MPa  (4.2)
300  p
where fse is the initial stress in the prestressing steel after losses, fc' is the concrete compressive
strength, and ρp is the ratio of prestressed reinforcement.
Eqns. 4.1 and 4.2 were originally developed by Mattock et al. [4-12] based on a series of
prestressed beam tests, and later modified by Mojtahedi and Gamble [4-13] to account for the
- 93 -
span-to-depth ratio. However, numerous researchers have since shown Eqns. 4.1 and 4.2 to be
inaccurate when predicting unbonded tendon stresses in more recent prestressed concrete beam
tests [4-14-17]. Many of these researchers have presented alternative equations based on
experimental and analytical results. A review of the relevant literature published prior to 2002
was presented by the ‘Subcommittee on Stresses in Unbonded Tendons’ of the Joint ACI-
ASCE Committee 423, ‘Prestressed Concrete’ [4-16]. Acknowledging the inaccuracy of
Eqns. 4.1 & 4.2, the subcommittee recommended a modified version of the expression
presented by Naaman and Alkhairi [4-18] for finding the unbounded tendon stresses in
prestressed members at nominal flexural strength, as shown in Eq. 4.3.
dp L
f ps  f se   u E ps  cu   1 1 (4.3)
 c  L2
where Ωu is a bond reduction coefficient dependent on the span-to-depth ratio and the loading
type, Eps is the prestressing steel modulus of elasticity, dp is the distance from extreme
compression fibre to centroid of prestressing tendon, c is the neutral axis depth at nominal
flexural strength, L1 is the sum of lengths of loaded spans containing tendon(s) considered, and
L2 is the total length of tendon(s) between anchorages.
However, the inclusion of the neutral axis depth, c, in Eq. 4.3 causes some uncertainty because
the neutral axis depth is dependent on the stress in the prestressing tendons, and thus accurate
estimation will require several iterations or an initial assumption.
To eliminate the need for the aforementioned iteration, Ozkul et al. [4-17] developed a closed
form equation which did not rely on calculation of the neutral axis depth. Their equation,
reproduced in Eq. 4.4, was based on a simplified analytical model and developed from virtual
work principles instead of empirical data. Ozkul et al. validated Eq. 4.4 with experimental data
from 25 prestressed beam tests that were conducted as well as additional historical test data,
showing significant improvement when compared to Eqns. 4.1 & 4.2.
E ps  e  f c' b 
f ps  f se   k1   f py (4.4)
196  As f y  A ps f pu 
where e is the eccentricity measured from the centroid axis, β defines the equivalent
rectangular stress block length, b is the beam thickness, As is the mild reinforcing steel area, fy
- 94 -
is the yield stress of mild reinforcing steel, Aps is the prestressing steel area, fpu is the ultimate
stress of prestressing steel, and k1 is a constant as defined by Ozkul et al. [4-17].
Eq. 4.4 was found to be inappropriate for use on walls for several reasons. The equation
includes a parameter, e, defined as the eccentricity of the tendon. If the tendon is located in the
center of the section, which is a common occurrence with walls, the resulting eccentricity is
zero, leading to a zero stress increase with increasing lateral deformation. This may be
acceptable for beams, where the tendons are generally located near the bottom of the section,
but unbonded tendons located at the center of the walls have been found both analytically and
experimentally to significantly increase in stress, as shown in Chapter 3 and section 4.1. Also,
the equation was developed using virtual work methods with the assumption that all tendons
are located at the same eccentricity, and thus is unable to provide predictions for multiple
tendons distributed through the section, which again is a common occurrence in PT walls.
The aforementioned equations predominantly focused on simply supported prestressed

concrete beams, and application of the proposed approaches for predicting unbonded tendon
stresses in prestressed concrete walls was not given consideration. Although Eq. 4.1 and
Eq. 4.2 were developed based on prestressed beam tests, both NZS 3101 and ACI 318-08 state
that they should be used for all flexural members with unbonded prestressing. Because a wall
can be represented by a cantilever beam, as shown in Figure 4.11, the unbonded tendon stresses
in a wall can be determined by analysing the wall as one half of a beam that is subjected to a
point load at the mid-span. However, when using the equations primarily developed for
beams, inaccurate prediction of tendon stresses should be expected in PT walls for the
following reasons:
 A wall generally has a much lower height/length ratio than a typical beam length/depth
ratio, which reduces the elastic flexural deformation of the member and its contribution
to the change in tendon stress or strain.
 Walls with unbonded tendons experience significantly less flexural cracking along the
member length than beams and thus a larger percentage of wall lateral deformation is
provided by the concentrated cracking developed at the critical section.
 Walls are stressed to higher axial load levels than beams because walls are also
subjected to gravity loads from the structure as well as higher prestress levels.
 In walls the tendons are not concentrated near an edge but instead are distributed
symmetrically along the wall length or lumped at the center.
- 95 -
Figure 4.11 – Equivalent wall representation of a beam
Noting that the equations developed for beams are not applicable to the prediction of tendon
stresses in walls, Wight and Ingham [4-7] developed an equation to predict the unbonded
tendon stresses at nominal flexural strength in post-tensioned concrete masonry walls. Their
equation was based upon the mechanics of a rocking PT wall, which was shown previously in
Figure 4.10. Assuming that the wall rotates as a rigid body, and thus the elastic flexural and
shear deformation of the wall panel can be neglected at the ultimate limit state, the change in
tendon length, Δpt, is simply equal to the gap opening at the base of the wall at the location of
the tendon. Therefore,
 pt   d i  c 
(4.5)
where θ is the rotation at wall base, di is the distance from extreme compression fibre to
centroid of the ith prestressing tendon, and c is the neutral axis depth.
Wight and Ingham assumed that the tendon remains in the elastic state because this is a
requirement of prestressed wall systems to ensure self-centering behaviour. The total tendon
stress, fps, can then be represented by the initial prestress, fse, plus the increase due to the gap
opening displacement, Δpt, distributed over the entire unbonded tendon length, lp. Hence,
E ps
f ps  f se   d i  c  (4.6)
lp
Using experimental data from masonry walls, combined with finite element analysis results,
Wight and Ingham then showed that the base rotation at the point of nominal flexural strength,
- 96 -
θcu, was proportional to the wall aspect ratio (he/lw) and the inverse of the axial stress ratio
(fm / fm' ). By fitting a linear correlation Wight and Ingham defined θcu for masonry walls as,
 he  
 l  mu
 cu   w
f  (4.7)
30  m ' 
 fm 
where he is the effective wall height, lw is the wall length, εmu is the ultimate masonry
compressive strain, fm masonry axial stress, and fm' is the masonry compressive strength.
The neutral axis depth, c, can be calculated using the rectangular stress block definition. Given
that an accurate neutral axis depth is dependent on the unbonded tendon stress, the calculation
is iterative. Wight and Ingham avoided this iteration by ignoring the change in tendon stress
and calculating the neutral axis depth from the initial prestress load. Thus the total unbonded
tendon stress, fps, was expressed as follows,
  mu he f m' E ps   
f ps  f se    d i   f m l w   f py (4.8)
 30l w f m l p   f '
   m 
and
f se A ps  N
fm  (4.9)
l w bw
where α defines the equivalent rectangular stress block average stress, β defines the equivalent
rectangular stress block length, N is the axial load due to wall self-weight and additional dead
and live loads, and bw is the wall thickness.
Wight and Ingham showed that Eq. 4.8 was able to predict the unbonded tendon stresses in
masonry walls with much greater accuracy than several masonry design code provisions,
including Eq. 4.1, which is also included in the New Zealand masonry Design Standard [4-19].
4.2.2.2 Comparison with current equations

For each tendon in each of the 30 FEM walls in the data set described in section 4.2.1, Eq. 4.1
and Eq. 4.8 were used to predict the unbonded tendon stresses when εcu = 0.003 at the extreme
compression fibre. Figure 4.12 shows a comparison of the NZS/ACI and the Wight and
- 97 -
Ingham predictions against the FEM results, where the dashed line represents a perfect
correlation. The graphs only plot the predicted change in tendon stress, ∆fps, excluding the
known initial stress, fse, from the calculation. This allowed the critical part of the equation to
be investigated more closely. It can be seen in Figure 4.12a that the current NZS/ACI equation
resulted in poor prediction of the change in tendon stress with errors of up to 300%, equating to
errors of up to 30% when calculating the total tendon stress. This was not surprising given the
simplicity of the equation, its failure to represent the mechanism of wall behaviour, and the fact
that it was exclusively validated using unbonded prestressed beam tests. The Wight and
Ingham equation resulted in a better prediction, as seen in Figure 4.12b. However, Eq. 4.8
appeared to consistently underestimate the change in tendon stress by as much as 50%. While
underestimating the stresses would usually be considered conservative, in the case of self-
centering members with unbonded tendons, the higher actual stresses could lead to premature
yielding of the tendons, which must be avoided. On closer inspection, it was found that the
empirical definitions developed by Wight and Ingham to calculate the neutral axis depth and
wall base rotation at the nominal flexural strength were not accurate for the concrete wall data
set. As described earlier, when considering the wall as a rotating block, the calculation of these
parameters is essential in estimating the unbonded tendon stress.
500 500
400 400
 fps : Predicted (MPa)
300 300
200 200
100 100
0 0
-100 -100
-100 0 100 200 300 400 500 -100 0 100 200 300 400 500
 fps : FEM (MPa)  fps : FEM (MPa)
(a) NZS/ACI Eq. 4.1 (b) Wight and Ingham Eq. 4.8
Figure 4.12 – Predicted change in tendon stress using existing equations
Although the behavioural mechanism of masonry and concrete walls are the same, the
inaccuracy of the Wight and Ingham equation was not completely surprising. Masonry
- 98 -
construction consists of masonry blocks with “soft” mortar joints that would have a significant
influence on the wall behaviour, and especially the compressive strains that are generated.
Additionally, the Wight and Ingham equation was validated based on a matrix of concrete
masonry walls having typical construction details. These typical details for masonry walls
differ significantly from precast concrete walls. Most notably, concrete has a much higher
stiffness and compressive strength. Also, concrete walls are generally designed for use with
larger height/length ratios and larger initial axial stress than masonry walls.
The other two equations described previously in section 4.2.2.1 were found to be unsuitable for
predicting the unbonded tendon stresses in PT walls. First, Eq. 4.3 suggested by the ACI-
ASCE Committee [4-16] was not compared because it contained the neutral axis depth, c,
which is an unknown parameter. This means that Eq. 4.3 does not result in a simple direct
calculation because the neutral axis depth is a function of the tendon stress being calculated.
Additionally, a comparison is not shown for Eq. 4.4 published by Ozkul et al. [4-17], because
as described previously, the equation was not applicable for PT walls with multiple tendon
locations.
4.2.2.3 Development of an improved equation

A suitable equation for predicting the unbonded tendon stresses in PT concrete walls was
developed by improving the accuracy of the Wight and Ingham equation. In this context, the
empirical definitions used by Wight and Ingham to estimate the neutral axis depth and wall
base rotation at nominal flexural strength required revision.
First, the prediction of the neutral axis depth at the point of nominal flexural strength was
investigated. The Wight and Ingham method defined the neutral axis depth based on
equivalent rectangular stress block assumptions. However, the axial load was calculated based
on the initial stresses in the tendons, ignoring the contribution from the change in tendon stress.
This resulted in under prediction of the true neutral axis at nominal flexural strength. While
the calculation of the exact neutral axis depth would require an iterative process, a simple
approximation was sought in order to avoid this process. It was found that when plotting the
neutral axis depth predicted by the FEM at nominal flexural strength against the neutral axis
depth calculated from the initial stress, a simple linear fit could be used, as shown in Figure
4.13. It can be seen that a suitable approximation was obtained when the neutral axis depth
calculated based on the initial wall axial stress was multiplied by a factor of 1.36.
- 99 -
900
800 y = 1.36x
R2 = 0.95
700
600
NA : FEM (mm) 500
400
300
200
100
0
0 200 400 600 800
NA : Initial stress (mm)
Figure 4.13 – Predicted neutral axis (NA) depth at nominal flexural strength
The definition of the base rotation at nominal flexural strength was also investigated. Wight
and Ingham found that the base rotation was proportional to the wall height/length ratio and the
ultimate concrete strain, and inversely proportional to the axial load ratio. For this
investigation, the wall height in the above relationship was substituted for the unbonded tendon
length, lp, which was considered to have a greater influential on the ultimate base rotation. As
shown in Figure 4.14, results from the FEM data set of concrete walls confirmed that when the
tendon length/wall length and axial load ratios were varied, while all others parameters are kept
constant, a strong correlation was observed.
By combining the tendon length/wall length ratio and the inverse of the axial load ratio, a
relationship was established. As plotted in Figure 4.15, it was found that a simple linear fit
could be applied when the base rotation was plotted against the tendon length/wall length ratio
and the inverse of the axial load ratio, both transformed by a 0.7 power. By calibrating the
relationship at εcu = 0.003, the following definition for the base rotation at nominal flexural
strength, θcu, was developed:
0.7
 lp 
 cu  l w 
 cu  (4.10)
8  fc 
 f'
 c 
- 100 -
-3
x 10
6 0.012
5 0.01
Base Rotation (rad)
Base Rotation (rad)

4 0.008
3 0.006
2 0.004
l p / l w = 1.25
1 fc / fc' = 0.10 0.002 l p / l w = 1.75

fc / fc' = 0.17 l p / l w = 3.5
0 0
0 1 2 3 4 0 5 10 15 20 25 30 35
lp / lw Inverse Axial Stress Ratio (f c ' / fc )
(a) Tendon length (lp) / wall length (lw) (b) Axial stress ratio (fc / fc' )
Figure 4.14 – Parameters influencing base rotation
0.01
R2 = 0.91
0.008
Base Rotation (rad)
0.006
0.004
0.002
0
0 5 10 15 20 25 30
0 .7
 lp 
 l 
 w 
 fc 
 f'
 c 
Figure 4.15 – Predicted base rotation at nominal flexural strength
- 101 -
With these redefined definitions for the neutral axis depth and the base rotation at nominal
flexural strength, the original equation proposed by Wight and Ingham was revised as follows:
0.7
  cu E ps  l p f c'    
f ps  f se     d i  1.36 f c l w   f py (4.11)
 8l  l f    f ' 
 p  w c    
c 
where
f se A ps  N
fc  (4.12)
l w bw
The prestressing tendons were assumed to remain elastic during the equation development, and
so Eq. 4.11 is only valid up to the yield limit state. This is not of concern because the tendons
are designed to remain elastic so that they can provide self-centering capability to the wall.
The PT wall data set described in section 4.2.1 was again used to verify the accuracy of this
revised equation. Figure 4.16 compares Eq. 4.11 with the FEM at nominal flexural strength
(εcu = 0.003) for a) the change and b) the total tendon stress. It is observed that the revised
equation captured the tendon stress with good accuracy with a maximum error of 6% for the
total tendon stress. The unbonded tendon stresses were predicted with much better accuracy
than previous equations, especially that provided by the current NZS 3101:2006 and ACI 318-
08 concrete design codes, which was shown previously in Figure 4.12a.
500 1600
400
1400
fps : Predicted (MPa)
300
1200
200
1000
100
800
0
-100 600
-100 0 100 200 300 400 500 600 800 1000 1200 1400 1600
 fps : FEM (MPa) fps : FEM (MPa)
(a) Change in stress (b) Total stress

Figure 4.16 – Predicted tendon stress using Eq. 4.11 compared with FEM data
- 102 -
As found during the experimental testing reported in Chapter 3, the current code defined
ultimate concrete compressive strain of 0.003 may be too low for PT concrete walls. Instead a
strain limit of up to 0.005 may be more appropriate for defining the ultimate limit state.
Although development and calibration of Eq. 4.11 was performed at the current code defined
ultimate concrete strain of 0.003, other strain limits can be employed in the equation if desired.
To verify this, the unbonded tendon stresses were recalculated using Eq. 4.11 and compared
against the FEM calculated change in tendon stress at both εcu = 0.004 and εcu = 0.005. It can
be seen in Figure 4.17 that the equation remained accurate despite increasing the εcu value.
700 700
600 600
500 500

400 400
300 300
200 200
100 100
0 0
-100 -100
0 200 400 600 0 200 400 600

 fps : FEM (MPa)  fps : FEM (MPa)
(a) εcu = 0.004 (b) εcu = 0.005

Figure 4.17 – Predicted change in tendon stress using Eq. 4.11 for larger ultimate strains
In addition to comparison with the FEM data set, the accuracy of Eq. 4.11 was also validated
against the experimental test data that was reported in Chapter 3. As described in Chapter 3,
the most reliable and consistent concrete compressive strain measurements were achieved for
walls E and F that were listed in Table 3.1, and so the results from the eight tests completed on
walls E and F were used to assess the accuracy of Eq. 4.11. Because a large spread in
measured strains was observed between gauges, the average strain of all the gauges that
reached a measured compressive strain of 0.003 was used to determine the point of nominal
flexural strength. As seen in Figure 4.17, when compared with the experimental data obtained
from walls E and F, Eq. 11 predicted both the change and the total tendon stress with
acceptable accuracy. There were several outlying points that resulted in significant over
prediction of the tendon stresses, which are represented in Figure 4.17 by the solid dots. These
- 103 -
outliers occurred for test F-i and test F-ii and it was attributed to the strain measurements that
were achieved for these two walls. For both test F-i and test F-ii, only the two embedded strain
gauges provided consistent strain measurements that exceeded 0.003, this meant that the
average strain used to determine the point of nominal flexural strength was not as reliable as
the walls that included an average of the embedded and surface mounted strain gauges.
500 1600
400
1400
fps : Predicted (MPa)

300
1200
200
1000
100
800
0
-100 600
-100 0 100 200 300 400 500 600 800 1000 1200 1400 1600
fps : Experimental (MPa) fps : Experimental (MPa)
(a) Change in stress (b) Total stress

Figure 4.18 – Predicted tendon stress using Eq. 4.11 compared with experimental data
4.2.3 Predicting Flexural Strength

With the unbonded tendon stresses accurately estimated, calculation of the nominal flexural
strength can easily be performed. As shown earlier in Figure 4.10, a rectangular stress block
can be assumed at the wall toe, with moment resistance provided by the unbonded tendons and
additional axial load including self-weight. The depth of the rectangular stress block, a, can be
calculated without iteration from the total axial load using the calculated unbonded tendon
stresses from Eq. 4.11. The base shear of the PT wall at nominal flexural strength, Vf, can then
be calculated using Eqns. 4.13 and 4.14, where n represents the total number of tendons.
n
 a l a (4.13)
M n  V f he   f ps ,i A ps ,i  d i    N  w  
i 1  2  2 2
and

n
f ps ,i A ps ,i  N (4.14)
a i 1
f b '
c w
- 104 -
The nominal flexural strength of the 30 walls in the data set described in section 4.2.1 were
calculated using Eq. 4.13 and this was compared with the lateral wall strength extracted from
the 3D FEM when the strain at the extreme compression element reached 0.003. Figure 4.19
shows the wall strength calculated using Eq. 4.13 normalised by the lateral strength calculated
by the FEM. It can be seen that an accurate and slightly conservative estimate of the wall
strength was achieved, with predicted strengths generally below that of the FEM but within
10%, and a maximum over-prediction of 4% for a single outlying wall. The under-prediction
may also be due to the unintentional confinement provided by the foundation to the wall toe.
As well as increasing the strain capacity, this confinement effect may have increased the
compressive strength of the wall toe, leading to higher lateral strengths. However, the under-
prediction was not significant and is conservative for design purposes, which confirmed that
Eqns. 4.11 and 4.13 are suitable for calculating the unbonded tendon stresses and nominal
flexural strength of PT concrete walls.
1.2
0.8
Vf / Vf FEM
0.6
0.4
0.2
0
0 5 10 15 20 25 30
FEM wall configuration
Figure 4.19 – Predicted base shear at nominal flexural strength using Eqns. 4.11 and 4.13
4.2.4 Predicting Displacement

As well as predicting strength, the design process usually involves a check of the displacement
at the ultimate limit state. As observed during the experimental testing reported in Chapter 3,
the lateral displacement of a PT wall is a function of gap opening at the wall base as well as
elastic flexural and shear deformation of the wall panel. Displacements due to gap opening (or
rocking) dominate the behaviour of a PT wall, with rigid body rotation often assumed. The
displacement due to gap opening is obtained from the calculated rotation at the wall base
- 105 -
multiplied by the wall height. Figure 4.20 shows the displacement corresponding to nominal
flexural strength that is due to gap opening, dn,gap, normalised by the total lateral displacement
corresponding to nominal flexural strength, dn, for each of the 30 walls in the FEM data set that
was described in section 4.2.1. It can be seen that displacements due to gap opening accounted
for at least 75% of the total lateral displacement. Furthermore, it was observed that the
contribution of the displacement due to gap opening reduced as the axial load increased,
because the deformations of the wall panel become larger as the lateral strength increased.
1.2
1
dn, gap / dn: FEM
0.8
0.6
0.4
0.2
0
0 5 10 15 20 25 30
FEM wall configuration
Figure 4.20 – Lateral displacement due to rocking
Calculation of the lateral displacement corresponding to nominal flexural strength based only
on displacements due to gap opening is not sufficient and will result in under-prediction of the
total displacement, which is not conservative. To improve estimation of the lateral
displacement, the calculation needs to include the elastic deformation of the wall panel also.
As was shown in section 3.5.3, the experimental tests found that deformation due to shear are
small in PT walls, and can typically be ignored. Thus the panel deformation can be calculated
from the flexural deformation alone. A simplified non-iterative equation was proposed by
Wight [4-20] to calculate the panel flexural deformation corresponding to nominal flexural
strength, dn,fl. The expression, shown in Eq. 4.15, was calculated by integrating the curvature
profile at the decompression state of the wall. Furthermore, the lateral displacement due to gap
opening can be calculated by using Eq. 4.10 developed earlier to estimate the wall base rotation
when nominal flexural strength is reached. The lateral displacement corresponding to nominal
- 106 -
flexural strength can then be estimated by adding the calculated contributions from gap
opening and flexure, resulting in Eq. 4.16.
2 he2 f c
d n , fl  (4.15)
3 Ec l w
0.7
 lp 
he  cu  l w  2 he2 f c
d n  d n , gap  d n , fl   cu he  d n , fl     (4.16)
8  fc  3 Ec l w
 f'
 c 
The accuracy of Eq. 4.16 was also checked using the FEM data set. The lateral displacements
calculated using Eq. 4.16 are plotted in Figure 4.21 against the lateral displacements calculated
from the FEMs at εcu = 0.003, with a perfect correlation represented by the dashed line. Some
scatter in the results was observed due to the scatter in the calculated base rotation (see Figure
4.15), but overall a good correlation was obtained with maximum error of 24%. However,
under-prediction of displacements is not desired during design and so the lateral displacement
calculated using Eq. 4.16 should be multiplied by a factor of 1.2 (represented by the dotted
line) to ensure that the actual displacements are below code defined lateral drift limits.
40
35
30
dn: Predicted (mm)
25
20
15
10
0
0 10 20 30 40
dn: FEM (mm)
Figure 4.21 – Predicted lateral displacement using Eq. 4.16
The accuracy of Eq. 4.16 was also checked for higher ultimate concrete strain limits. The
lateral displacements calculated using Eq. 4.16 are plotted in Figure 4.22 against the lateral
displacements calculated from the FEMs at εcu = 0.004 and 0.005. For both strain limits the
- 107 -
correlation was still acceptable, which confirmed that Eq. 4.16 can also be applied if the
ultimate strain limit was increased for PT concrete walls.
50 60
50
40
dn: Predicted (mm)
dn: Predicted (mm)

40
30
30
20
20
10
10
0 0
0 10 20 30 40 50 0 10 20 30 40 50 60
dn: FEM (mm) dn: FEM (mm)
(a) εcu = 0.004 (b) εcu = 0.005

Figure 4.22 – Predicted lateral displacement for higher strain limits using Eq. 4.16
4.3 CONCLUSIONS
This chapter presents analytical modelling of individual PT concrete walls, as well as a simple
set of equations for predicting the wall response at the nominal flexural strength limit state.
The experimental testing reported in Chapter 3 highlighted many uncertainties in compression
strain measurement in the wall toe and so analytical finite element modelling was used to assist
in quantifying the compressive strains and thus the ultimate limit state.
Traditional 2D finite element modelling failed to capture the wall response with sufficient
accuracy due to the idealised representation of the toe and wall-to-foundation regions. The
rigid foundation caused over-estimation of the compressive strains in the wall toe, which led to
premature crushing. A 3D FEM was developed to overcome the limitations of the 2D model
and was able to capture both the global and local response of the walls with good correlation
when compared to the experimental results. The 3D FEM allowed for accurate calculation of
the average strain at a position close to the extreme compression fibre. This accuracy of the
FEM compensated for the complexity of interpreting experimental measurements of concrete
strains and allowed for accurate determination of strain based limit states.
- 108 -
4.4 References
Prediction of the nominal flexural strength of a PT wall requires accurate estimation of the
unbonded tendon stresses. Previously developed equations, including those in NZ and US
concrete design standards, were shown to poorly predict the unbonded tendon stresses in PT
concrete walls when compared to finite element analyses, because most of these equations
were developed empirically using data from unbonded post-tensioned beams.
A proposed equation for predicting the unbonded tendon stresses in post-tensioned concrete
walls was developed which provided greater versatility through the increased number of
variables, including wall aspect ratio, initial wall axial stress, unbonded length of the tendon,
and the distance from the compression force to each prestressing tendon. The developed
equation provided good estimation of unbonded tendon stress across a wide range of wall
parameters when compared with both finite element analyses and experimental test results.
Additionally, the proposed equation was found to be accurate for an ultimate concrete
compressive strain of 0.003 and 0.005. When the new equation for predicting unbonded
tendon stresses was used, accurate prediction of the nominal flexural strength was also
achieved for all walls analysed.
An equation for predicting the lateral displacement at the effective wall height was also
developed, which included components from gap opening at the wall base and flexural
deformation of the wall panel. The equation provided good estimation of the wall
displacement, when compared to the finite element analyses, for an ultimate compressive
concrete strain of between 0.003 and 0.005.
4.4 REFERENCES
[4-1] Aaleti, S. and Sritharan, S. (2009) A simplified analysis method for characterizing
unbonded post-tensioned precast wall systems. Engineering Structures, 31(12), 2966-
2975.
[4-2] Perez, F. J., Sause, R., and Pessiki, S. (2007) Analytical and experimental lateral load
behavior of unbonded posttensioned precast concrete walls. Journal of Structural
Engineering, 133(11), 1531-1540.
[4-3] NZS 3101:2006. Appendix B: Special provisions for the seismic design of ductile
jointed precast concrete structural systems. Concrete structures standard, Standards
New Zealand, Wellington, New Zealand.
- 109 -
[4-5] ABAQUS user's manual version 6.8. Dassault Systèmes Simulia Corp., 2008.
[4-6] Allen, M. G. and Kurama, Y. C. (2002) Design of rectangular openings in precast walls
under combined vertical and lateral loads. PCI Journal, 47(2), 58-79.
[4-7] Wight, G. D. and Ingham, J. M. (2008) Tendon stress in unbonded post-tensioned

masonry walls at nominal in-plane strength. Journal of Structural Engineering, 134(6),
947-960.
[4-8] Park, R. and Paulay, T. (1975) Reinforced concrete structures. John Wiley and Sons,
New York.
New Zealand.
[4-12] Mattock, A. H., Yamazaki, J., and Kattula, B. T. (1971) Comparative study of
prestressed concrete beams, with and without bond. Proc. ACI Journal, 68(2), 116-125.
[4-13] Mojtahedi, S. and Gamble, W. L. (1978) Ultimate steel stresses in unbonded

prestressed concrete. ASCE Structural Journal, 104(7), 1159-1165.
[4-14] Harajli, M. H. (2006) On the stress in unbonded tendons at ultimate: Critical

assessment and proposed changes. ACI Structural Journal, 103(6), 803-812.
[4-15] Naaman, A. E. and Alkhairi, F. M. (1991) Stress at ultimate in unbonded post-

tensioning tendons. Part 1. Evaluation of the state-of-the-art. ACI Structural Journal,
88(5), 641-651.
[4-16] Naaman, A. E., Burns, N., French, C., Gamble, W. L., and Mattock, A. H. (2002)
Stresses in unbonded prestressing tendons at ultimate: Recommendation. ACI
Structural Journal, 99(4), 518-529.
[4-17] Ozkul, O., Nassif, H., Tanchan, P., and Harajli, M. H. (2008) Rational approach for
predicting stress in beams with unbonded tendons. ACI Structural Journal, 105(3), 338-
347.
[4-18] Naaman, A. E. and Alkhairi, F. M. (1991) Stress at ultimate in unbonded post-

tensioning tendons. Part 2. Proposed methodology. ACI Structural Journal, 88(6), 683-
692.
- 110 -
4.4 References
[4-19] NZS 4230:2004. Design of reinforced concrete masonry structures. Standards New
Zealand, Wellington, New Zealand.
[4-20] Wight, G. D. (2006) Seismic performance of a post-tensioned concrete masonry wall

system. PhD thesis. Dept. of Civil and Environmental Engineering, University of
Auckland, New Zealand.
- 111 -
- 112 -
Chapter 5
PreWEC Connector Design
As described in Chapter 1 and Chapter 2, additional energy dissipating components are

required when precast concrete walls with unbonded tendons are constructed in regions with
high seismicity. This chapter introduces a recently developed self-centering wall system that
consists of a Precast Wall with End Columns (PreWEC) [5-1]. The PreWEC system
overcomes several of the limitations of current self-centering systems that were identified in
Chapter 1 and is suitable for application in regions with moderate to high seismicity. A special
energy dissipating connector is required for the PreWEC system and this chapter describes
extensive finite element modelling that was conducted to design a suitable connector. A
summary of a companion study conducted by Aaleti [5-2] is also presented to experimentally
validate the connector design and provide data from a large-scale PreWEC wall test that is used
to validate analytical models that are developed in Chapters 6 and 7.
5.1 PREWEC WALL SYSTEM

The main purpose of self-centering structural components is to offer an alternative building
solution that is more resilient to damage caused by earthquakes. However, minimising
- 113 -
Chapter 5 PreWEC Connector Design
structural damage alone is not sufficient for a self-centering wall system to be considered as a
viable alternative for buildings constructed in regions of high seismicity. As well as providing
superior seismic performance, self-centering concrete wall systems must be economically and
architecturally comparable to conventional reinforced concrete construction techniques.
Although the jointed wall and hybrid wall systems described in Chapter 2 performed well
during large-scale testing [5-3, 5-4], their implementation into real structures has been limited.
This lack of implementation is believed to be due to failure of the wall systems to meet the
aforementioned requirements. The main difficulty in achieving seismic resilience without
compromising economic or architectural factors is associated with the placement of energy
dissipating devices. The jointed wall system requires the wall to be divided into two or more
panels to allow for the placement of energy dissipating shear connectors along the vertical
joints. This division of the wall into several panels reduces the length of the lever arm between
the post-tensioning tendons and the compression block in the wall toe. This reduced lever arm
results in a reduction in moment resisting capacity when compared to a monolithic reinforced
concrete wall with similar dimensions, which significantly reduces the cost-effectiveness of the
jointed wall system. Alternatively, the hybrid wall system relies on mild steel reinforcing bars
placed across the wall-to-foundation interface to provide energy dissipation. With its
arrangement of post-tensioning tendons and mild steel reinforcing bars, a hybrid wall could be
designed to match the moment capacity of a traditional reinforced concrete wall. However,
due to their placement within the wall, the mild steel reinforcing bars cannot be easily replaced
after they are subjected to large inelastic strains and possible fatigue fracture during an
earthquake. Therefore, despite improving the seismic performance, a fully resilient building
cannot be designed with a hybrid self-centering precast wall. To rectify these deficiencies, a
new system consisting of a Precast Wall with End Columns (or PreWEC) was developed [5-1].
The PreWEC system, shown in Figure 5.1, consists of a single precast concrete wall panel with
two steel or concrete end columns that are each anchored to the foundation using unbonded
post-tensioning. Only minimum reinforcing steel is required within the wall panel, except for
additional confinement requirements in the corner toes where large compressive strains are
expected. The wall and columns are joined horizontally using special shear connectors along
the vertical joints to provide additional energy dissipation. As with previous self-centering
technology, when subjected to a lateral load the PreWEC system concentrates inelastic
deformation at a single crack that opens up at the base of the wall and columns. The post-
- 114 -
5.2 PreWEC Connector Requirements
tensioning tendons are unbonded to reduce the strain demand and are designed to remain
elastic up to the design level drift, providing a restoring force to self-center the structure.
Using this arrangement of components, the PreWEC system maximises the lever arm between
the post-tensioning tendons and the compression block in the wall toe and can be designed to
obtain a moment capacity equal to that of a comparable monolithic reinforced concrete wall.
Additionally, only minor structural damage is expected during large lateral drifts, and the
energy dissipating shear connectors can be easily replaced, so that performance based targets
for seismic resilience can be easily achieved.
Figure 5.1 – PreWEC wall system [5-1]
5.2 PREWEC CONNECTOR REQUIREMENTS

The shear connectors in the PreWEC system have two functions. First, by transferring forces
between the wall and column elements, the connectors contribute to the systems moment
capacity. Second, they are expected to undergo large inelastic deformations and act as the
primary source of energy dissipation in the system. During cyclic loading, the connectors are
subjected to relative vertical displacements at the wall-to-column joint due to uplift at the base
of the wall and column. The relative vertical displacements are much larger in one direction
- 115 -
due to differences in the levels of uplift that occur at the wall and column toes. This leads to
the connectors experiencing an unsymmetrical cyclic displacement history.
Considering the unique behaviour of the PreWEC system, an investigation into suitable shear
connectors was conducted. A PreWEC wall was analyzed for use in a six storey prototype
structure [5-1], which was subsequently used to determine the requirements for the shear
connectors. Based on a simplified design procedure described by Aaleti [5-2], a target force-
displacement envelope was established for the connector, as shown in Figure 5.2. The
connector was required to maintain a stable force-displacement response, maximize energy
dissipation, and be able to sustain relative vertical displacements of up to 60 mm with the peak
strains that were generated being limited to less than 0.10. This strain limit was chosen to
reflect a dependable limit for mild steel to prevent fracture due to low cycle fatigue when
subjected to repeated seismic cyclic deformations, as described in Priestley et al. [5-5]. Grade
50 steel has an ultimate tensile strain of ~0.18 for monotonic loading. Priestley et al. [5-5]
recommend that for seismic loading the strain softening portion of the stress-strain response
should be ignored, resulting in an effective ultimate strain limit of ~0.12 for Grade 50 steel
(yield strength of 345 MPa). Additionally, when subjected to reverse cyclic loading the sum of
the maximum tension and compression strains should not exceed the effective ultimate strain
limit. As explained earlier, loading of the connectors in the negative direction is limited for the
PreWEC system, thus the maximum strain was assumed to be 0.02 in the negative direction.
This resulted in a maximum allowable strain of 0.10 in the positive loading direction.
40
35
30
25
Force (kN)
20
15
10
0
0 10 20 30 40 50 60 70
Displacement (mm)
Figure 5.2 – Force-displacement design envelope established for PreWEC specimen [5-1]
- 116 -
5.3 Possible Types of Connector
5.3 POSSIBLE TYPES OF CONNECTOR

As part of the PRESSS program, which was described in Chapter 2 [5-6], a study into the
behaviour of various steel shear connectors was conducted by Shultz and Magana [5-7]. This
study investigated several connector types, including:
 notched shear plate (NSP)
 slotted flexural plate (SFP)
 inclined flat bar (IFB)
 X shaped axial plate (XAP)
 pinned tension strut (PTS)
 vertical joint friction (VJF)
 U-shaped flexural plate (UFP)
The experimental program examined the connectors’ behaviour under a reverse cyclic vertical
displacement history. The UFP, originally developed in the 1970’s [5-8], was found by Shultz
and Magana to be one of the most suitable connectors, maintaining a stable force-displacement
response up to large cyclic displacements and dissipating large amounts of energy. Following
the study by Shultz and Magana, the UFP connector was included in the jointed wall system of
the PRESSS test building, where it performed as expected [5-3]. Figure 5.3 shows two half
UFPs being tested, as well as the measured force-displacement response [5-9].
60
40
20
Force (kN)
-20
-40
-60
-80 -60 -40 -20 0 20 40 60 80
Displacement (mm)
(a) UFP test (b) Force-displacement response for UFP

Figure 5.3 – U-shaped flexural plate (UFP) [5-9]
- 117 -
Due to the similarities between the requirements of the connectors in the jointed wall and the
PreWEC systems, the findings by Shultz and Magana are relevant when considering suitable
types of connector. A list of possible connectors was collated, and connectors were assigned to
categories based on the mechanism of plastic deformation, including:
 direct shear mechanism

 direct tension/compression mechanism
 flexural mechanism
 friction mechanism
 a combination of the above
After studying the results from the experimental investigation by Shultz and Magana,
connectors that used a direct shear or tension/compression mechanism were deemed less
suitable. Although these connectors performed extremely well, they generated large strain
demands, which resulted in displacement capacities that were much less than the 60 mm
requirement that was selected for the PreWEC system.
In consideration of the above finding, flexural yielding was identified as the most desirable
mechanism for a PreWEC connector because the more indirect load path lowers the connector
strain demand and results in a larger connector displacement capacity. The U-shaped flexural
plate (UFP) used a rolling and flexural yielding mechanism that allowed it to accommodate
large vertical displacements with stable force-displacement behaviour. However, because of
the large inelastic strains developed when the UFP was manufactured, it was required to be
constructed from stainless steel. The use of stainless steel meant the drawbacks of the UFP
connector were that it was expensive and its behaviour was dependent on strain history due to
isotropic hardening. For these reasons, the use of stainless steel, and thus the UFP, was
avoided in this study and more economical connectors made from grade A50 mild steel were
investigated instead.
It was decided that the most suitable and economic connectors would be plate type connectors
that used flexural dominated yielding mechanisms. Detailed investigation into the performance
and design of such connectors was conducted, concentrating on flexural plate connectors with
various slot and hole configurations, as well as J and oval-shaped flexural plate connectors.
- 118 -

After selecting flexurally deforming plate connectors as the most suitable option for the
PreWEC system, a series of finite element analyses were conducted to investigate their
performance. Finite element models (FEM), developed using ABAQUS [5-10], were used to
evaluate the behaviour of different types of connector as well as different configurations. The
finite element modelling allowed the optimal connector design to be determined prior to an
experimental investigation.
All finite element models were constructed using 3D deformable elements, and a steel plate
material model was defined to simulate Grade A50 steel properties. An idealized bilinear
stress-strain material model was used based on an elastic modulus of 200 GPa, a yield stress of
345 MPa, an ultimate stress of 450 MPa, and an ultimate strain of 0.18. No failure criteria
were used in the steel material definition, so the stress-strain response showed no strength loss
beyond the monotonic load ultimate strain of 0.18. This did not affect the results of the
analysis because a maximum strain limit of 0.10 was used to determine the ultimate
displacement capacity for each connector. The steel material definition used a kinematic
hardening model to simulate the steel cyclic behaviour [5-11, 5-12]. Meshing of the plates was
completed using linear 3D stress elements with 8 nodes and 1 integration point per element,
and the mesh size was approximately 5 mm. The plates were appropriately partitioned to allow
structured meshing to be used, resulting in rectangular dominated elements. Mesh widths were
reduced around penetrations such as holes and slots to provide more realistic stress and strain
predictions in the critical regions.
The restraint conditions were idealized for the initial models to reduce computational time.
The end face of the plate on the left side used a boundary condition to fix the displacement in
all three degrees of freedom, while the end face on the right was coupled in all degrees of
freedom to a reference point which was used to control the loading, as shown in Figure 5.4.
The displacement controlled loading was applied by defining a displacement boundary
condition on the reference point, restraining movement in all degrees of freedom except the
vertical. The reversed cyclic loading history applied increasing vertical displacements on the
reference point in steps of 10 mm up to a maximum 60 mm displacement, as shown in Figure
5.5. The FEM analyses produced force-displacement curves from the output at the reference
point as well as the local stress and strain values for each element.
- 119 -
(a) Reference point coupling (b) Boundary conditions

Figure 5.4 – Typical restraints used during FEM analyses
60
40
Displacement (mm)
20
-20
-40
-60
0 2 4 6 8 10 12 14
Step
Figure 5.5 – Displacement history used for FEM analyses
Over thirty different FEM analyses were run to evaluate the performance of different flexural
plate connectors. A selection of the most relevant and influential connectors that were
modelled is included herein, which consisted of slotted flexural plates (SFP), flexural plates
with holes, J-shaped flexural plates and oval shaped flexural plates.
5.4.1 Slotted Flexural Plate (SFP)

The first connector modelled, SFP-1, consisted of a 127 mm by 177.8 mm, 6.35 mm thick plate
with two 25.4 mm wide horizontal slots (Figure 5.6a). The FEM was constructed as previously
described and the resulting deformed shape is shown in Figure 5.6b overlaid with the principal
strain field that developed when the connector was subjected to a relative displacement of
60 mm. Flexural yielding occurring at the ends of the horizontal webs is clearly visible. As
- 120 -
shown in Figure 5.6c, the predicted force-displacement response exhibited stable hysteretic
loops with large amounts of energy dissipation from the flexural yielding mechanism. The
predicted force-displacement response is plotted alongside the design envelope and it can be
seen that SFP-1 was about 10% below the required strength. The FEM did not include
prediction of the failure mechanism and thus showed an idealized response with no strength
degradation. Instead, prediction of the displacement capacity was achieved by monitoring the
peak strains that were generated at critical locations on the connector. Figure 5.6c also
includes predicted peak principal strains generated in the two most critical elements of the
connector as a function of displacement. It is observed that the strain demand reached 0.35
(35%) at the maximum displacement of 60 mm, which is well beyond the ultimate strain limit
of 0.10 chosen for the A50 steel. Using this strain limit, it is seen from Figure 5.6c that the
displacement capacity of SFP-1 was 11 mm, which was well short of the 60 mm target
displacement.
177.8
38.1 25.4
25.4
25.4 127
R12.7
60 mm
Plate thickness = 6.35 mm
(a) Sketch of connector (b) FEM at 60 mm disp.

40 0.35
30 0.3
20
0.25
10
Strain (m/m)
Force (kN)
0.2
0
0.15
-10
0.1
-20
Envelope
-30 FEM Force 0.05
FEM Strain
-40 0
-60 -40 -20 0 20 40 60
Displacement (mm)
(c) Force-displacement and strain prediction

Figure 5.6 – Dimensions and FEM analysis for SFP-1
- 121 -
In an attempt to reduce the strain demand on the slotted flexural plate connector, the aspect
ratio was increased by lengthening the plate from 177.8 mm to 254 mm. SFP-2 (see Figure
5.7a) was modelled with this increased length, which reduced the plastic rotations and
corresponding strain demand at the ends of the horizontal webs. Figure 5.7b shows the FEM
analysis output of SFP-2 at the peak 60 mm displacement. The force-displacement response of
this connector, shown in Figure 5.7c, produced good hysteretic loops, but the strength of the
connector was 54% below the required strength. SFP-2 was successful in reducing the strain
demand, with the predicted peak strain dropping to 0.2 and the displacement capacity at the
0.10 strain limit increased to 26 mm. While the connector strength could be increased by
increasing the number of connectors or the connector thickness, increasing the displacement
capacity was considered more critical. The strain demand could be reduced by further
increasing the aspect ratio of the connector. However, a length greater than 254 mm would
cause the connectors to be unsuitable given the dimensions of the prototype PreWEC system.
254
38.1 25.4
25.4
25.4 127
R12.7
60 mm

40 0.2
30
20 0.15
10
Strain (m/m)
Force (kN)
0 0.1
-10
-20 0.05
Envelope
-30 FEM Force
FEM Strain
-40 0
-60 -40 -20 0 20 40 60
Displacement (mm)

- 122 -
Another option trialled was to change the orientation of the slots from horizontal to vertical.
SFP-3 maintained the same plate dimensions as SFP-1, but the two horizontal slots were
substituted with three vertical slots of the same width, see Figure 5.8a. The FEM was
constructed in the same way and the resulting analysis output at 60 mm displacement is shown
in Figure 5.8b. Flexural yielding was again apparent at the ends of the now vertical webs. The
force-displacement prediction is plotted in Figure 5.8c, and it can be seen that the strength
increased substantially above that predicted by the FEM analysis of SFP-1 due to an increased
number of locations for flexural yielding. SFP-3 produced a backbone curve that exceeded the
required design envelope by 60%. However, due to the shortened length of the webs, the
plastic rotation demands increased, leading to predicted peak strains higher than those observed
for SFP-1. The vertical displacement capacity of the connector was reduced to only 8 mm.
178.8
25.4
25.4 25.4 127
38.1 R12.7
60 mm
Plate thickness =6.35 mm

60 0.45
0.4
40
0.35
20 0.3
Strain (m/m)
Force (kN)
0.25
0
0.2
-20 0.15
Envelope 0.1
-40
FEM Force
0.05
FEM Strain
-60 0
-60 -40 -20 0 20 40 60
Displacement (mm)

- 123 -
5.4.2 Flexural Plates with Holes

One of the problems observed with the SFP is that the locations of plastic deformation are
limited to small regions at the ends of the webs. To increase the potential locations for plastic
deformation the slots were replaced with six 25.4 mm diameter holes for connector H-1, as
shown in Figure 5.9a. The resulting FEM analysis is shown in Figure 5.9b with an increase in
the locations of severe plastic deformation. As a result of this, it can be seen from the force-
displacement prediction in Figure 5.9c that the strength of the connector increased to three
times the require design envelope strength. Additionally, the hysteresis loops showed large
amounts of energy dissipation. However, the strain demand was similar to SFP-1, peaking at
around 0.35 during the 60 mm cycle. This resulted in a predicted displacement capacity of just
13 mm when the 0.10 strain limit was reached. It was concluded that H-1 offered exceptional
strength, but that the displacement capacity was still well below the requirements for PreWEC.
177.8
R12.7
38.1 50.8 127
38.1
60 mm

150 0.4
0.35
100
0.3
50
0.25
Strain (m/m)
Force (kN)
0 0.2
0.15
-50
0.1
Envelope
-100
FEM Force 0.05
FEM Strain
-150 0
-60 -40 -20 0 20 40 60
Displacement (mm)

Figure 5.9 – Dimensions and FEM analysis for H-1
- 124 -
Again, to reduce the strain demand on the connector the aspect ratio of the plate was increased
by increasing its length. H-2, shown in Figure 5.10a, had a length of 254 mm and contained
eight 25.4 mm diameter holes with the same dimensions used for H-1. The FEM at peak
displacement is shown in Figure 5.10b with evenly distributed plastic action occurring over
several regions of the plate. The force-displacement history plotted in Figure 5.10c indicated
that H-2 performed well, with stable hysteresis loops and strength of about 2.5 times the
required design envelope. As expected, the strain demand reduced, with peak strains limited to
just 0.26 at the peak 60 mm displacement and an increased displacement capacity of 21 mm.
Although H-2 produced promising results with greater than required strength and energy
dissipation, the displacement capacity was well short of the 60 mm required. Increasing the
aspect ratio further, to reduce the strain demand, was again not considered a viable option for
the PreWEC system.
254
R12.7
50.8 50.8 127
38.1
60 mm

100 0.35
0.3
50
0.25
Strain (m/m)
Force (kN)
0.2
0
0.15
0.1
-50
Envelope
FEM Force 0.05
FEM Strain
-100 0
-60 -40 -20 0 20 40 60
Displacement (mm)

- 125 -
As previously mentioned, the connector for the PreWEC system experiences an unsymmetrical
displacement loading. To take advantage of this unsymmetrical loading, an innovative
approach using inclined elliptical holes was trailed. The inclined elliptical holes were chosen
because the strain demand would be much lower when the holes were stretched in the short
axis, providing an increased displacement capacity in the positive loading direction. Figure
5.11a shows the dimensions of H-3, which consisted of the longer 254 mm length plate and
included eight large elliptical holes orientated at 45 degrees. The FEM was constructed as
previously described, but the loading protocol was modified. Instead of loading the connector
symmetrically in both directions, it was decided to limit the negative displacement to 20 mm.
This limit prevented loading to a large negative displacement that the unsymmetrical connector
was not intended to be capable of withstanding. Figure 5.11b shows H-3 at the peak positive
displacement of 60 mm and the opening of the elliptical holes was apparent, with a more
circular profile observed. The force-displacement history, shown in Figure 5.11c, indicated
that the connector produced stable hysteresis loops and strength that exceeded the required
target envelope by over 40%. The strains were reduced when compared to the circular holes
used for H-2, with a predicted peak value of less than 0.3 at 60 mm, resulting in a displacement
capacity of 25 mm when using the 0.10 strain limit. Although H-3 was successful in proving
the viability of inclined elliptical holes, the displacement capacity was again well less than of
the 60 mm design requirement.
5.4.3 J-Shaped Flexural Plate (J-Connector)

After traditional flexural plate options failed to provide the required displacement capacity, a
new style of connector was trialled. The J-shaped flexural plate (or J-Connector) can be cut
from a steel plate, but the arrangement allows for a substantial reduction in strain demand as
there is no direct tension path between the two ends, which results in an almost pure flexural
deformation mechanism. The modelled J-Connector, shown in Figure 5.12a, consisted of a
25.4 mm wide J shape cut from a 12.7 mm steel plate with square ends which were welded on
three sides. The unsymmetrical design of the J shape was aimed at minimizing the strain
demand in the positive displacement direction. Again, an unsymmetrical loading history was
used, with the negative displacement capped at 20 mm.
- 126 -
254
50.8 50.8
127
60 30
38.1 45°
60 mm

40 0.25
30
0.2
20
10
Strain (m/m)
0.15
Force (kN)
0.1
-10
-20
Envelope 0.05
-30 FEM Force
FEM Strain
-40 0
-20 0 20 40 60
Displacement (mm)

The results from the FEM analysis that assumed no out-of-plane movement to the connector
are displayed in Figure 5.12b at the 60 mm positive displacement. The flexural yielding
mechanism is clearly visible in the two legs. Yielding was spread along almost the entire
length of the J-Connector legs, which significantly reduced the strain demand. The predicted
strain demand at the critical elements, plotted in Figure 5.12c, indicated that the connector can
be subjected to the full 60 mm vertical displacement while generating peak strains of only 0.08.
The drawback of the J-Connector concept was that the plastic deformation was less effective
and the resulting force-displacement response showed a relatively low strength capacity. The
force-displacement response indicated extensive energy dissipation with full hysteresis loops,
but the strength was below half the require design strength.
- 127 -
63.5 B 60 mm
57.2 A
25.4
127
76.2
R38.1
R63.5

40 0.08
30 0.07
20 0.06
10 0.05
Strain (m/m)
Force (kN)
0 0.04
-10 0.03
-20 0.02
Envelope
-30 FEM Force 0.01
FEM Strain
-40 0
-20 0 20 40 60
Displacement (mm)

Figure 5.12 – Dimensions and FEM analysis for J-Connector
5.4.4 Oval Shaped Flexural Plate (O-Connector)

To increase the strength of the J-Connector while still maintaining the same flexural
mechanism, an oval-shaped flexural plate O-Connector was considered. The first O-
Connector, shown in Figure 5.13a, was of similar dimensions to the J-Connector, with
25.4 mm wide and 76.2 mm long legs, and was cut from a 12.7 mm thick steel plate. The
connector was to be attached to a structural system using a 50.8 mm fillet weld on either side
of each leg, as shown in Figure 5.13a. The oval shape increased the potential location of
plastic deformation from two to four legs, resulting in twice the strength of the J-Connector.
The FEM was constructed in the same way as for the previous models. The welding conditions
were again simulated by restraining the degrees of freedom along the welded faces of the O-
Connector. The resulting strain field from the FEM analysis is shown in Figure 5.13b,
- 128 -
indicating that plastic yielding occurred in all four legs of the O-Connector. As shown in
Figure 5.13c, the force-displacement response predicted by the FEM analysis, confirmed that
the O-Connector achieved twice the strength of the J-Connector, which was 13% below the
require design strength. The strain demand was similar to the J-Connector with the flexural
yielding spread along a large length of each leg. A peak strain of just over 0.08 occurred
during the maximum 60 mm vertical displacement. The O-Connector successfully maintained
the displacement capacity of the J-Connector with strength close to the target design envelope.
76.2
50.8 60 mm
25.4
76.2
76.2
R38.1
R63.5

40 0.09
30 0.08
0.07
20
0.06
10
Strain (m/m)
Force (kN)
0.05
0
0.04
-10
0.03
-20
Envelope 0.02
-30 FEM Force

0.01
FEM Strain
-40 0
-60 -40 -20 0 20 40 60
Displacement (mm)

Figure 5.13 – Dimensions and FEM analysis for O-Connector
- 129 -
5.4.5 Summary of FEM Analyses

The FEM analysis results are summarised in Table 5.1. The table shows a comparison of the
eight connectors previously described with regard to their strengths and displacement
capacities. The results indicate that the flexural plates with slots or holes can provide useful
connectors when the displacement requirements are small, but are not suitable for the PreWEC
system designed for high seismic regions. From the FEM study it was concluded that the most
suitable connector was the oval shaped flexural plate (O-Connector), which provided adequate
strength while maintaining a large displacement capacity.
Table 5.1 – Summary of connector dimensions and capacities obtained from FEMs
Connector Length Openings Strength Displacement capacity

(mm) kN % of required mm % of required
capacity capacity
SFP-1 177.8 Horizontal slots 32 91 11 18
SFP-2 254 Horizontal slots 16 46 26 43
SFP-3 177.8 Vertical slots 56 160 8 13
H-1 177.8 Holes 108 309 13 22
H-2 254 Holes 86 246 21 35
H-3 254 Elliptical holes 50 143 25 42
J-Connector N/A N/A 15 43 60+ 100
O-Connector N/A N/A 31 87 60+ 100
5.4.6 Optimisation of O-Connector

To optimise the O-Connector and better understand its behaviour, a further series of FEM
analyses were conducted. In these analyses, the effect of changes to various dimensions on the
connector response was investigated, and the resulting strengths and displacement capacities
were compared. Ten different O-Connector options were modelled with changes including
connector length, width, loop radius, loop leg width, and plate thickness.
The FEMs that were generated to analyse and optimise the O-Connector were more
sophisticated than the previous models. The connector itself was modelled in the same way as
described previously, but the mesh was modified to increase the number of elements across the
section width and thus increase the accuracy of the strain predictions. Additionally, instead of
assuming that the weld provided a fully fixed restraint, the weld itself was introduced to the
model. The 9.5 mm fillet welds were modelled with 3D stress elements with a global mesh
- 130 -
size of approximately 4 mm. The weld material used a bilinear stress-strain definition with an
elastic modulus of 200 GPa, yield stress of 500 MPa, ultimate stress of 600 MPa, and ultimate
strain of 0.10. In addition to the introduction of the weld into the model, steel plates were
added to simulate the actual loading conditions. In a PreWEC wall system, the connectors
would be welded to steel plates cast into the wall and the end column. The plates were
modelled with 3D stress elements and used a 20 mm mesh. The weld bond between the
connector, weld and plates was modelled using a tie constraint between the adjacent surfaces.
The loading was applied as described in the previous section, although now through the steel
plates. The back surface of the left steel plate was restrained in all degrees of freedom and the
back surface of the right steel plate was constrained to a reference point through which the
displacement controlled loading history was applied. Loading was applied in 10 mm reverse
cyclic increments up to a 60 mm peak vertical displacement, as shown previously in Figure
5.5.
As explained earlier, several analyses were conducted to assess the influence of the connector
dimensions on both the strength capacity and strain demand. A summary of the findings from
these analyses is shown in Table 5.2. It was found that increasing the O-Connector length,
width, or loop radius resulted in a reduced strength and strain demand. Increasing the loop leg
width increased the strength and strain demand, and increasing the plate thickness increased the
strength but had no effect on the strain demand.
Table 5.2 – Effect of O-Connector dimensions on strength and strain demand
Connector Change Strength Strain demand

Dimension
Length Increased Reduced Reduced

Width Increased Slightly reduced Slightly reduced
Loop radius Increased Slightly reduced Slightly reduced
Loop leg width Increased Increased Increased
Plate thickness Increased Increased Negligible
A revised O-Connector design was obtained that optimised all the dimensional considerations.
The optimised O-Connector, shown in Figure 5.14a, consisted of an increased 31.75 mm loop
width, overall connector width of 152.4 mm and leg length of 88.9 mm. This resulted in a
higher strength than required so the plate thickness was reduced to 9.5 mm, which was a more
economical option. The FEM constructed for the optimised O-Connector can been seen in
- 131 -
Figure 5.14b with the strain field plotted at 60 mm displacement. The predicted force-
displacement history in Figure 5.14c again showed stable hysteresis loops with large amounts
of energy dissipation and strength just below the required design envelope. The low strain
demand was maintained with the peak strain of 0.08 at the maximum 60 mm vertical
displacement, which was 20% below the limiting strain of 0.10. The optimised O-Connector
appeared to be well suited for the PreWEC system with a satisfactory force-displacement
backbone curve, sufficient energy dissipation, and greater than required displacement capacity.
88.9
50.8
31.8
88.9
88.9
R44.5
R76.2
60 mm

40 0.09
30 0.08
0.07
20
0.06
10
Strain (m/m)
Force (kN)
0.05
0
0.04
-10
0.03
-20
Envelope 0.02
-30 FEM Force

0.01
FEM Strain
-40 0
-60 -40 -20 0 20 40 60
Displacement (mm)

Figure 5.14 – Dimensions and FEM analysis for optimised O-Connector
- 132 -
5.5 Experimental Validation
5.5 EXPERIMENTAL VALIDATION

To validate the FEM predictions and to confirm the expected connector performance when
subjected to cyclic loading, two tests were performed on the O-Connector by Aaleti [5-2],
namely A1 and A2. In each test, four individual connectors were tested to maintain symmetry
of the test setup. The dimensions of each connector were the same as those of the optimised O-
Connector, shown previously in Figure 5.14a, but with a reduced 44.45 mm weld length on
each side. The connectors were cut from 9.53 mm thick grade A50 steel plate using a laser
cutting technique in order to reduce the residual stresses induced during the fabrication process.
5.5.1 Test Setup

The test setup, shown in Figure 5.15a, was designed using steel tubes and steel plates to apply
the desired vertical loading to the O-Connectors. The test setup consisted of an outer U-frame
and a central H-section that represented the end columns and the wall in the PreWEC system,
respectively. In order to eliminate any eccentric loading, four O-Connectors were welded
between the U-frame and H-section and tested simultaneously. This additionally provided a
more accurate assessment of the mean connector response. The test rig was placed in a MTS
uniaxial testing machine with the U-frame attached to the bottom grip and the H-section
attached to the top grip. Loading was applied in a displacement control mode, subjecting the
O-Connectors to a relative vertical displacement history. In addition to the displacement and
force output from the MTS machine, external LVDTs and strain gauges were used for data
acquisition, as shown in Figure 5.15b. An LVDT was installed to measure the relative vertical
displacement between the U-frame and H-section, to ensure that the applied loading subjected
the connectors to the correct displacement. LVDTs were also installed to measure any slip that
might have occurred at the end grips of the MTS. No significant slip was observed or
measured at the end grips during testing and so the external LVDTs recorded connector
displacements similar to the recorded displacement output from the MTS. Up to four strain
gauges were mounted at the locations at which the FEM predicted the maximum strains to
occur in the O-Connector. These locations, shown in Figure 5.15b, were on the side face, or
extreme tension edge, of the connector and positioned approximately 20 mm along the leg
from the end of the weld.
- 133 -
H‐section
Strain gauges
Connectors
Strain gauges
LVDTs
U‐Frame
(a) Setup (b) Instrumentation

Figure 5.15 – Experimental test setup and instrumentation used for O-Connector tests
The loading protocol used for Test-A1 was developed to simulate the expected displacement
history to which the connectors would be subjected during a reverse cyclic load test of the
PreWEC-1 prototype specimen [5-1]. The displacement history, shown in Figure 5.16a,
consisted of an unsymmetrical reverse cyclic loading up to a maximum peak displacement of
50.8 mm in the positive direction. The displacements in the negative direction were capped at
12.7 mm. As explained previously, the loading of the connectors in the PreWEC system is
unsymmetrical and 12.7 mm represented a conservative negative displacement limit. At each
displacement level, the connectors were cycled three times to observe the stability of the force-
displacement response. During Test-A2, the loading protocol was modified and the connectors
were subjected to a true displacement history measured during the large-scale testing of the
PreWEC-1 specimen [5-4, 5-13], as shown in Figure 5.16b. The recorded displacement history
ended at a peak positive displacement of 53 mm, so the record was extrapolated to a peak of
71 mm until failure occurred to the connectors.
- 134 -
60
50
40
Displacement (mm)
30
20
10
-10
-20
0 10 20 30 40 50 60 70
Step
(a) Test-A1
80
60
Displacement (mm)
40
20
-20
0 1000 2000 3000 4000 5000 6000 7000 8000 9000
Step
(a) Test-A2
Figure 5.16 – Displacement history used for connector Tests-A1 & A2
To determine the true properties of the steel used to manufacture the tested connectors, three
tensile test coupons were machined from the same 9.53 mm thick A50 steel plate that was used
for the O-Connectors. The tensile tests were carried out according to ASTM standards for
tension testing of metallic materials [5-14] and are recorded in more detail by Aaleti [5-2].
Measured stress-strain results for two of the tensile coupons are plotted in Figure 5.17 with an
enlargement of the 0-0.04 strain region.
- 135 -
500
400
Stress (MPa) 500
300 400
Stress (MPa) 300
200 200
Coupon 1
100
Coupon 2
100
0
0 0.01 0.02 0.03 0.04
Strain (m/m)
0
0 0.05 0.1 0.15 0.2 0.25 0.3 0.35 0.4
Strain (m/m)
Figure 5.17 – Measured stress-strain results for tensile coupons
5.5.2 Detailed FEM of Test Configuration

To improve the accuracy of the analysis, improvements were made to the existing FEM by
incorporating a model for the entire test rig, as shown in Figure 5.18. The model of the actual
O-Connector remained the same but the grade A50 steel definition was modified from the
original assumed values to a true stress-strain curve measured during tensile coupon tests. A
combined kinematic/isotropic hardening model was used with a monotonic stress-strain input.
(a) FEM assembly (b) FEM at 60 mm displacement

Figure 5.18 – FEM of the test rig and O-Connectors
- 136 -
The FEM of the test rig was created using 3D linear stress elements based approximately on a
20 mm quadrilateral mesh. The individual parts of the rig used tie constraints at the adjoining
faces to simulate the effects of welds. The bottom loading plate used a boundary condition
restraint to simulate the grip, preventing movement to the surfaces of the plate in all directions.
The top loading plate was constrained in all degrees of freedom to a reference point. The
vertical displacement histories were applied via a series of displacement boundary conditions
to the reference point. The analysis of this test setup was run and is reported with the
experimental test results below.
5.5.3 Experimental and Analytical Results

The results from Test-A1 indicated that the O-Connector behaved as predominantly expected.
It can be seen from the force-displacement response in Figure 5.19 that the connectors
exhibited stable hysteresis loops with large energy dissipation. The O-Connectors began to
experience out-of-plane buckling during the 31.75 mm displacement cycle. As the out-of-
plane bucking became more pronounced during large displacement cycles, see Figure 5.20,
significant strength degradation occurred which was not desirable, especially because it
reduced the energy being dissipated. To prevent this out-of-plane movement occurring in
future tests, a retrofit to the connector was provided with a pair of steel restraining plates. A
photograph of one of the restraining plates used in Test-A2 is shown in Figure 5.21.
40
30
20
10
Force (kN)
-10
-20
-30
-40
-20 -10 0 10 20 30 40 50 60
Displacement (mm)
Figure 5.19 – Measured force-displacement response of Test-A1
- 137 -
Figure 5.20 – Out-of-plane buckling of the connectors in Test-A1 at a relative vertical

displacement of 50 mm
Connector
Restraining plate
Figure 5.21 – Restraining plate used during Test-A2
The results from Test-A2 showed improved performance with no out-of-plane buckling
observed. The force-displacement loops, shown in Figure 5.22a, were stable up to positive
displacements of 57 mm, with some strength degradation occurring during the cycle to 71 mm
when the connectors started to fracture. The restraining plates were successful in preventing
any out-of-plane buckling and allowed the full displacement capacity of the O-Connector to be
reached without any significant loss in strength. Ultimately, if the design predicted relative
displacements in excess of 25 mm, the restraining plates would be required.
- 138 -
40 40
30 30
20 20
10 10
Force (kN)
Force (kN)
0 0
-10 -10
-20 -20
-30 Test -30 Test

FEM FEM
-40 -40
-20 0 20 40 60 80 -20 0 20 40 60 80
Displacement (mm) Displacement (mm)
(a) Cyclic force-displacement (b) Force-displacement backbone

Figure 5.22 – Comparison of measured and calculated response of Test-A2
The FEM analysis was run for the test specimen as described earlier, using the measured steel
stress-stain properties from the material tests. A comparison of the predicted force-
displacement response of a single connector from Test-A2 is plotted alongside the test results
in Figure 5.22a. It is observed that the FEM provided accurate estimation of the connector’s
response throughout the hysteresis cycles, closely matching the loading and unloading stiffness
and only marginally underestimated the connector strength. The accuracy of the FEM is
highlighted in Figure 5.22b which shows the force-displacement backbone that was extracted
from the cycle peaks. At a more local level, the predicted strain from FEM elements at the
same location and direction as two mounted strain gauge (SG-1 and SG-2) is plotted for
comparison in Figure 5.23 for both the full cyclic displacement history and the backbone
extracted from displacement peaks. The experimental readings are terminated when the gauges
reached their measuring limit of approximately -0.04 and 0.03 for SG-1 and SG-2 respectively.
Overall, the FEM estimation of the strains was good, with the FEM being within 10% of the
measured strains. The FEM predicted a displacements capacity of 62.5 mm for the O-
Connector at the 0.10 strain limit, which correlated well with the actual connector failure which
occurred during the final cycle to 71 mm. Finally, the FEM predicted touching of the
connectors in the test setup to occur at a displacement of ~60 mm (as seen in Figure 5.18b).
Although the displacement exceeded 60 mm during the Test-A2, touching was not observed
because the connectors began to fracture during this load cycle, altering their displaced shape.
- 139 -
0.02 0.02
Test Test
0.01 0.01
FEM FEM
0 0
-0.01 -0.01
Strain (m/m)
Strain (m/m)
-0.02 -0.02
-0.03 -0.03
-0.04 -0.04
-0.05 -0.05
-0.06 -0.06
-20 0 20 40 60 -20 0 20 40 60
(a) Cyclic strain for SG-1 (b) Strain backbone for SG-1
0.05 0.05
0.04 0.04
0.03 0.03
Strain (m/m)
Strain (m/m)
0.02 0.02
0.01 0.01
0 0
Test Test
FEM FEM
-0.01 -0.01
-10 0 10 20 30 40 -10 0 10 20 30 40
(c) Cyclic strain for SG-2 (d) Strain backbone for SG-2
Figure 5.23 – Comparison of measured and calculated strains for connector in Test-A2
5.6 LARGE-SCALE PREWEC TEST

Experimental validation of the PreWEC concept was undertaken in collaboration between Iowa
State University, the National Centre for Research on Earthquake Engineering (NCREE) in
Taiwan, and the University of Auckland. The experimental validation is reported in detail by
Aaleti [5-2] and Sritharan et al. [5-13]. The PreWEC test specimen, shown in Figure 5.24, was
tested at the NCREE laboratory in Taiwan and consisted of a 6.100 m high and 1.830 m long
precast concrete panel with two concrete filled steel tube end columns. Following the
- 140 -
5.6 Large-scale PreWEC Test
successful design and experimental validation described above, the O-Connector was included
in the PreWEC test specimen. One leg of the O-Connector was welded to the steel end column
and the other leg was welded to a steel plate that was embedded into the concrete wall panel, as
shown in Figure 5.24b. The PreWEC test specimen was designed to have comparable
dimensions and material properties to a previously tested traditional monolithic reinforced
concrete wall, allowing for direct comparison between the two systems. A single hydraulic
actuator was used to apply a lateral load at the top of the wall to simulate the behaviour of the
system during an earthquake. An increasing pseudo-static cyclic displacement history was
applied and the wall behaviour was monitored with an extensive array of instrumentation.
(a) Test setup (b) O-Connector

Figure 5.24 – PreWEC specimen test setup
5.6.1 O-Connector Design and Testing

The connectors used during the large-scale PreWEC test had similar dimensions to the
previously tested O-Connector (Test-A), but the length was shortened by 25 mm and the weld
length was increased to 63.5 mm, as shown in Figure 5.25. Because of these revised
dimensions, as well as a change in material due to the steel grades available in Taiwan, further
tests were conducted to verify the behaviour of the O-Connector prior to the large-scale test of
- 141 -
the PreWEC specimen. Furthermore, these additional tests provided additional experimental
data to validate the developed FEM procedure.
69.9
63.5
31.8
69.9
88.9
R44.5
R76.2
Figure 5.25 – Sketch of O-Connector used for PreWEC test
The new O-Connectors were tested using the same setup as before (see Figure 5.15).
Additionally, the previous displacement history was used, which consisted of unsymmetrical
cyclic loading up to a peak positive displacement of 50.8 mm, as shown previously in Figure
5.16a. In total four tests were completed and are reported in detail by Aaleti [5-2]. The force-
displacement response of one of these tests, named Test-B, is reproduced in Figure 5.26. The
new O-Connector showed a similar response to the previously tested design (Test-A), with
stable and full hysteresis loops. Again, the restraining plates were successful in preventing any
out-of-pane buckling and ensured that the connector strength was maintained until failure
occurred due to fracture of the connector legs. The measured strength was 20% higher than the
previously tested design due to the reduced leg length and the Taiwan steel properties, which
were found to be of higher strength grade than the US equivalent. The shorter leg length also
led to higher strain demand, which along with less ductile steel, led to earlier failure of the O-
Connectors, which started to fracture during the final cycle to 44.45 mm. The fractures
continued to open up, causing some strength degradation during the cycles to 50.8 mm and
eventually led to the complete rupture of one of the O-Connector legs, as observed in Figure
5.27. However, even in their final fractured condition the connectors maintained
approximately 60% of their maximum strength and with significant energy dissipation still
observed in the hysteresis loops. Based on the connector response found from testing, the
PreWEC test specimen was designed using 20 O-Connectors to maximise the energy
dissipation while ensuring that self-centering would still occur [5-2].
- 142 -
5.6 Large-scale PreWEC Test
50 50
Force (kN)
Force (kN)
0 0
Test Test
FEM FEM
-50 -50
-20 0 20 40 60 -20 0 20 40 60
(a) Cyclic force-displacement (b) Force-displacement backbone

Figure 5.26 – Comparison of measured and calculated response of Test-B
Fracture
Figure 5.27 – Failure of O-Connectors in Test-B
As described before, an FEM was constructed to replicate the entire test setup. The new O-
Connector dimensions and weld length were used and the steel material definition was updated
based on the measured stress-strain response found from tensile coupon tests. A comparison
between the FEM calculated and experimental force-displacement response is shown in Figure
5.26. Again, the FEM provided good estimation of the connector behaviour, closely matching
the measured response and extracted backbone. The FEM underestimated the maximum
strength by 5% in the positive loading direction and underestimated the cyclic strain hardening,
especially in the negative loading direction where displacements were capped at -12.7 mm.
- 143 -
5.6.2 PreWEC Test Results and Connector Performance

As reported by Aaleti [5-2] and Sritharan et al. [5-13], the PreWEC specimen tested at NCREE
behaved as expected and showed satisfactory performance. Inelastic deformation was
concentrated at a single crack that opened up at the wall base and minimal damage resulted.
This is highlighted in Figure 5.28a, which shows the wall condition at 3% lateral drift. The O-
Connectors performed as expected and the yield lines and deformed shape are visible in Figure
5.28b. The measured force-displacement response of the PreWEC wall, shown in Figure
5.29a, indicated a stable response with energy dissipation resulting from the connectors and
with self-centering as a result of the unbonded post-tensioning. Additionally, Figure 5.29b
shows the vertical displacement measured across two connectors located at opposite ends of
the wall. The connector displacement showed an almost linear correlation with the wall lateral
displacement, and peaked at over 50 mm during the cycles to 3.5% drift. Some of the
connectors began to fracture during the 3% drift cycle, or a connector vertical displacement of
approximately 45 mm, which was similar to the failure displacement observed during the
connector testing. However, it should be noted that in their fractured state the connectors still
transferred forces, and only minor loss in strength and energy dissipation was observed during
the final cycles of the force-displacement response. At the conclusion of the test, over half the
O-Connectors had partially or fully fractured legs, as shown in Figure 5.30. This is considered
a desirable failure mechanism because the connectors can be easily replaced following a large
earthquake and the observed failure occurred in excess of the 2% design level drift.
(a) Wall condition at 3% drift (b) O-Connector at 3% drift

Figure 5.28 – Performance of PreWEC wall test specimen
- 144 -
5.7 Conclusions
Lateral Drift (%) Lateral Drift (%)

-3 -2 -1 0 1 2 3 -3 -2 -1 0 1 2 3
600 60
North
50
South

400
40
200
30
0 20
10
-200
0
-400
-10
-600 -20
-200 -100 0 100 200 -200 -100 0 100 200
(a) PreWEC lateral force-displacement (b) Connector vertical displacement

Figure 5.29 – Measured responses for PreWEC test
Fracture
Figure 5.30 – Failure of O-Connectors during PreWEC test
5.7 CONCLUSIONS
The PreWEC system overcomes several limitations that are associated with pervious self-
centering wall systems. The arrangement of elements in the PreWEC system maximises the
- 145 -
moment capacity of the wall to provide an economical construction alternative, and

additionally the connectors are located so that they can be easily inspected or replaced
following an earthquake, resulting in a seismic resilient self-centering wall system. This
chapter describes the design and analysis of energy dissipating steel connectors for use in the
PreWEC wall system. A review of previous studies indicated that the use of a flexural
deformation mechanism would be most suitable, providing stable hysteresis loops with large
amounts of energy dissipation and increased displacement capacity.
A series of finite element analyses were conducted to investigate the behaviour of the
following connector types: slotted flexural plates (SFP), flexural plates with holes, J-shaped
flexural plates (J-Connector), and oval shaped flexural plates (O-Connector). After comparing
the predicted force-displacement response and displacement capacity, the O-Connector was
selected as the most suitable connector type for PreWEC systems in high seismic regions.
However, other connectors may be used for PreWEC systems in low or moderate seismic
regions or for connecting other precast elements with lower displacement demands (e.g., panel-
to-panel connection in a precast flooring system and floor-to-wall-connections). Further FEM
analyses were conducted to determine the optimum dimensions of the O-Connector.
An experimental test program was conducted with two test units each containing four O-
Connectors. The experimental results validated the connector’s performance, demonstrating
excellent force-displacement characteristics with stable hysteresis loops and large energy
dissipation. The first test demonstrated the expected out-of-plane buckling of the connector
and its influence on the force-displacement response of the connector. This problem was
overcome by adding a simple restraining plate in the second test. The experimental results
provided successful validation to the FEM analysis procedure. The FEM was found to provide
accurate prediction for both the connector strength and the critical strains. The strain limit of
0.10 imposed to predict the maximum displacement capacity proved to satisfactorily estimate
the failure of the tested O-Connector.
Finally, large-scale testing of a PreWEC wall that incorporated the O-Connector design was
conducted. The connectors behaved as expected during the test, providing a stable response up
to and beyond the 2% design level drift. The O-Connector contributed to the successful
validation of the PreWEC concept, with the test specimen exhibiting sufficient energy
dissipation as well as a self-centering response and minimal structural damage. Due to a
- 146 -
5.8 References
change in the O-Connector dimensions and steel grade used during manufacturing, a lower
displacement capacity was observed, with failure initiating during cycles to 3% lateral drift.
However, no significant loss in wall strength was observed and the O-Connectors could be
easily removed and replaced if required.
5.8 REFERENCES
[5-1] Aaleti, S. and Sritharan, S. (2007) A precast wall with end columns (PreWEC) for
seismic applications. 8th Pacific Conference on Earthquake Engineering, Singapore,
Dec 5-7, Paper 157.
[5-2] Aaleti, S. (2009) Behavior of rectangular concrete walls subjected to simulated seismic
loading. PhD thesis. Dept. of Civil, Construction and Environmental Engineering, Iowa
State Univerisity, Ames, IA.
Journal, 44(6), 42-67.

1560-1570.
[5-5] Priestley, M. J. N., Seible, F., and Calvi, G. M. (1996) Seismic design and retrofit of
bridges. John Wiley and Sons, Inc., New York.
[5-6] Priestley, M. J. N. (1991) Overview of PRESSS research program. PCI Journal, 36(4),
50-57.
[5-7] Shultz, A. E. and Magana, R. A. (1996) Seismic behavior of connections in precast

concrete walls. Mete A. Sozen Symposium, ACI SP 162, American Concrete Institute,
Farmington Hills, MI.
[5-8] Kelly, J. M., Skinner, R. I., and Heine, A. J. (1972) Mechanisms of energy absorption
in special devices for use in earthquake resistant structures. Bulletin of the New Zealand
Society for Earthquake Engineering, 5(3).
[5-9] Sritharan, S., Aaleti, S., and Thomas, D. J. Seismic analysis and design of precast
concrete jointed wall systems. ISU-ERI-Ames Report ERI-07404, Department of Civil,
Construction and Environmental Engineering, Iowa State University, Ames, IA, 2007.
[5-11] Li, B., Reis, L., and de Freitas, M. (2006) Simulation of cyclic stress/strain evolutions
for multiaxial fatigue life prediction. International Journal of Fatigue, 28(5-6), 451-
458.
- 147 -
[5-12] Shen, C., Mamaghani, I. H. P., Mizuno, E., and Usami, T. (1995) Cyclic behavior of
structural steels. II: theory. Journal of Engineering Mechanics, 121(11), 1165-1172.
[5-14] ASTM Committee E-28. Standard test methods for tension testing of metallic materials.
American Society for Testing and Materials, Annual book of ASTM standards, 1991.
- 148 -
Chapter 6
Finite Element Modelling of the
PreWEC System
Following successful experimental validation of the PreWEC wall system as reported in

Chapter 5, a finite element model (FEM) of the test specimen was developed to assist in
understanding the observed behaviour. The FEM of the PreWEC test specimen extended upon
the previously developed modelling techniques, including the 3D FEM of the PT concrete wall
panels that was described in Chapter 4 and the O-Connector FEM that was described in
Chapter 5. The behaviour of the confined concrete in the wall toe was a critical component in
the PreWEC model, and the extensive investigation conducted into the procedures for
modelling confined concrete is described in this chapter. The FEM developed was used to
calculate the full cyclic response of the PreWEC test specimen and the influence of cyclic
damage parameters was examined.
Modifications to the PreWEC design were also investigated using the FEM. Several of the
main design attributes were varied to assess their effect on wall performance, including the
amount of post-tensioning and the number of energy dissipating connectors. Hysteretic energy
dissipation was examined in detail and the FEM was used to quantify the contribution from the
- 149 -
Chapter 6 Finite Element Modelling of the PreWEC System
connectors and develop a design equation to calculate the equivalent viscous damping of the
PreWEC system. Finally, a concept for a core wall system using PreWEC was analytically
investigated using the FEM.
6.1 MODEL FORMULATION

To accurately capture the response of the PreWEC wall, each of the components in the system
needed to be modelled correctly. As introduced in Chapter 5, the PreWEC system consists of a
precast concrete wall and two end columns, all of which are post-tensioned to the foundation
and connected together using special energy dissipating connectors, as shown in Figure 6.1.
Figure 6.1 – PreWEC wall system
The FEM of the individual PT walls that was described in Chapter 4 was extended in order to
develop the basis for the PreWEC FEM. The 3D FEM in Chapter 4 showed good correlation
with the experimental data, and was able to accurately capture the complexities of both the
wall-to-foundation interface and the compression wall toe. However, the PT wall panels tested
and analysed in Chapter 3 and 4 contained no specific confinement reinforcement in the wall
toe region. Because the PreWEC test specimen was designed for application in regions with
- 150 -
6.2 Confined Concrete
high seismicity, the wall toes must be designed to withstand the high compressive strains that
are generated when the wall is subjected to large lateral drifts. For this reason, confinement
reinforcement was required in the wall toes to prevent premature failure, as indicated in Figure
6.1. Modelling of this confined concrete region was complex, and an extensive investigation
was conducted to develop an accurate model of the confined concrete compressive response
that could be included in the PreWEC FEM.
The column elements were modelled using the same basic technique that was used for the
precast concrete wall, except that appropriate material definitions for the steel tube and
concrete core were required. Additionally, the O-Connectors that were used on the PreWEC
test specimen were modelled using the FEM that was developed and validated in Chapter 5.
6.2 CONFINED CONCRETE

6.2.1 Fundamental Equations
In seismic design, it is common to use stirrups or spirals as transverse reinforcement to provide
confinement to concrete sections where high strain demand is expected. When concrete is
subjected to a confining pressure, its behaviour is altered significantly from that of unconfined
concrete, with enhanced strength and a significant increase in strain capacity. Although several
models exist for predicting the stress-strain behaviour of confined concrete, the model
presented by Mander et al. [6-1] was chosen due to its widespread recognition within the
engineering community. Mander et al. proposed Eq. 6.1 and 6.2 to predict the confined
concrete compressive strength, fcc', and the corresponding strain, εcc.
 7.94 f l ' f l ' 

f cc'  f c'   1.254  2.254 1   2 (6.1)
 f c' f c' 

  f c' 
 cc   co 1  5  '  1 (6.2)
  f cc 
where fc' is the unconfined concrete compressive strength, fl', is the effective lateral confining
stress, and εco is the compressive strain at the unconfined concrete compressive strength (εco is
often assumed to be equal to 0.002).
- 151 -
Furthermore, Mander et al. proposed Eq. 6.3 to predict the backbone compressive stress-strain
response of confined concrete (fc, εc), where the definition of the equation terms is shown
visually on a typical stress-strain response in Figure 6.2.
 
r   c 
fc
   cc 
(6.3)
f cc'  
r
r  1   c 
  cc 
where
Ec
r
f cc' (6.4)
Ec 
 cc
Figure 6.2 – Confined concrete model proposed by Mander et al. [6-1]
The original model presented by Mander et al. [6-1] was limited to prediction of the backbone
compressive stress-strain response of confined concrete, but Chang and Mander [6-2] extended
the original procedure to provide a unified equation that could be used to approximate the
compressive and tensile stress-strain response of both confined and unconfined concrete.
Chang and Mander adopted a modified stress-strain backbone equation that is shown in
Eq. 6.5. It should be noted that for the compressive stress-strain response of confined concrete,
Eq. 6.5 results in a backbone response that is comparable to the original equation by
Mander et al. (Eq. 6.3).
- 152 -
 
n   c 
fc
   cc 
f cc'  c 
r
(6.5)
 
 r    c    cc 
1 n   
 r  1    cc  r 1
where
Ec  cc (6.6)
n
f cc'
n 1 (6.7)
r
n
E c  8200  f c'   3
8 (6.8)
Although Eqns. 6.5 and 6.6 represent the compressive stress-strain response of confined
concrete, the variables in Eq. 6.5 can easily be interchanged to allow for calculation of the
unconfined concrete response ( fcc' = fc' and εcc = εco) or tensile response ( fc = ft , εc = εt ,
fcc' = ft' and εcc = εto).
As well as equations to calculate the stress-strain backbone response, Chang and Mander
presented an extensive set of hysteresis rules to represent the cyclic behaviour of confined and
unconfined concrete. The primary rule that was appropriate to this investigation was the
unloading stiffness. Chang and Mander adopted Eq. 6.9 and 6.10 to predict the unloading
stiffness of the compressive and tensile response respectively (ESEC, c and ESEC, t).
  f un  
    0.57 
E
E SEC , c  E c   c cc   (6.9)
   
  un   0.57 
   cc  
  f un  
    0.67 
E
E SEC , t  E c   c to   (6.10)
     
  un o
  0.67 
   to  
where fun and εun are the stress and strain prior to unloading and εo is the strain at zero stress
prior to the tension load cycle.
- 153 -
6.2.2 ABAQUS Concrete Damaged Plasticity Model

The characteristic stress-strain response of the “concrete damaged plasticity” model available
is ABAQUS [6-3] is shown in Figure 6.3. The model allows the user to define the concrete
modulus of elasticity as well as the compression and tension inelastic stress-strain backbone
responses. By default the unloading stiffness is equal to the initial stiffness, Ec, but “damage
indices” (dc and dt) can be defined to reduce the slope of the unloading stiffness. The
flexibility of the concrete damaged plasticity model allowed the backbone response and
unloading stiffness to be defined using the Chang and Mander [6-2] formulas that were
described above.
(a) Tension (b) Compression

Figure 6.3 – ABAQUS concrete damaged plasticity model [6-3]
6.2.3 FEM Trials

A series of trials were conducted to assess the accuracy of different techniques for modelling
confined concrete behaviour using ABAQUS [6-3]. The original experimental tests that were
conducted by Mander et al. [6-4] were selected as the benchmark for the FEM trials. An FEM
of one of the test columns was constructed for the initial investigation, followed by analysis of
a wall test specimen to assess appropriate techniques to use for the PreWEC FEM.
6.2.3.1 Columns tests

Mander et al. [6-4] tested a total of twelve 1500 mm high and 500 mm diameter circular
concrete columns, with varying quantities of longitudinal and transverse reinforcing steel. The
columns were tested until failure by subjected them to an increasing monotonic compressive
load. The measured response was used to validate the proposed equations for calculating the
compressive stress-strain response of confined concrete.
- 154 -
Due to the availability of the data, the Column 7 specimen from Mander et al. [6-4] was
initially modelled. The details for the Column 7 test specimen are given in Table 6.1.
Column 7 was considered to have medium to high confinement, and the spiral transverse
reinforcing steel arrangement resulted in a high confinement effectiveness ratio, ke, of 0.987.
The measured strength of the concrete and reinforcing steel that was reported by Mander et al.
was used throughout the modelling and is also recorded in Table 6.1.
Table 6.1 – Details of Column 7 tested by Mander et al. [6-4]
Height 1500 mm
Diameter 500 mm
Core diameter 438 mm
fc'* 31 MPa
Longitudinal rebar / yield stress* 8 D28 / 296 MPa
Transverse rebar / yield stress* R12 spiral, 52 mm spacing / 340 MPa
Confinement effectiveness, ke 0.987
'#
Confining pressure, fl 3.35 MPa
* measured value # calculated based on yield strength of transverse reinforcement

The concrete column was modelled using 3D brick elements, with a mesh size of
approximately 60 mm. The concrete material defined for the column used the “concrete
damaged plasticity model”, and the concrete stress-strain behaviour was defined using the
backbone equation proposed by Chang and Mander (Eq. 6.5). The compressive stress-strain
response initially assumed unconfined concrete properties based on the concrete crushing
strength, fc', with the confinement effects included in the model using several different
techniques, which are described in detail below. The tensile stress-strain response was also
modelled using Eq. 6.5, with an assumed maximum tensile stress of 0.62 f c' (MPa), which is
consistent with recommendations by Paulay and Priestley [6-5]. Figure 6.4 shows the
unconfined compressive and tensile stress-strain responses that were input into the FEM. The
end face at the bottom of the column was restrained in all directions, and the end face at the top
of the column was restrained from rotation and subjected to an increasing vertical displacement
to apply the compressive load. The axial load was recorded, as well as the vertical strain
calculated over a central 450 mm gauge length, as was measured during the experimental tests.
- 155 -
-35
Unconfined
-30
Modified
-25
Stress (MPa)
-20
-15
-10
-5
5
0 -0.005 -0.01 -0.015 -0.02 -0.025
Strain (m/m)
Figure 6.4 – Concrete stress-strain input used for Column 7 FEM
For the first column trial, the detailed FEM was designed to replicate the physical conditions of
the experimental test. The concrete was modelled with an unconfined stress-strain response, as
described above, and the longitudinal and transverse reinforcing bars were modelled discretely
using wire truss elements. The wire element D28 longitudinal bars and R12 transverse spiral
were assigned material properties based on the measured properties reported by Mander et al.
[6-4], including a modulus of elasticity of 200 GPa and yield stresses of 296 MPa and 340 MPa
respectively. The reinforcing bars were embedded into the column elements with an idealised
bond that assumed no degradation or slip at the interface. The calculated response from the
column analysis using wire reinforcing steel is shown in Figure 6.5a. Although the FEM
accurately calculated the maximum confined strength of the column, the axial load response
descended more rapidly than the measured response from the test column. When the confining
pressure as the transverse reinforcing steel was engaged, the compressive strength of the
concrete elements was increased but the strain softening was not captured. The behaviour of
the model is further highlighted in Figure 6.5a, where the unconfined FEM response is also
plotted, representing the column without transverse reinforcing bars. The confining pressure
effectively increased the strength of the element’s stress-strain response by a constant factor,
but the general shape of the stress-strain response remained unchanged.
Inaccuracies in the FEM strain softening response of confined concrete elements has been
observed by several researchers [6-6, 6-7, 6-8], and is attributed to the influence of concrete
fracture mechanics. When concrete is loaded in axial compression, failure is attributed to
- 156 -
fracture of the specimen in a localised region. Due to strain localisation in the fracture region,
the response is more accurately represented by a fracture displacement rather than strains that
are dependent on the gauge length [6-9]. The fracture behaviour of concrete can be represented
by the crack band model [6-10], and implemented into finite element models using discrete or
smeared crack material models. However, because the objective was to model the cyclic
hysteresis behaviour of the PreWEC test specimen it was important to find a solution that could
be implemented using the concrete damaged plasticity model.
12 12
10 10
8 8
Axial load (MN)
Axial load (MN)
6 6
4 4
Test Test
2 2
FEM: Wire rebar FEM: Modified + Rebar
FEM: Unconfined FEM: Modified
0 0
0 0.01 0.02 0.03 0.04 0 0.01 0.02 0.03 0.04
Compressive Strain Compressive Strain
(a) Wire reinforcing bar (b) Modified strain softening
12
10
8
Axial load (MN)
2
Test
FEM: Modified + Pressure
0
0 0.01 0.02 0.03 0.04
Compressive Strain
(c) External pressure

Figure 6.5 – Results for FEM analysis of concrete confinement effects on Column 7
- 157 -
One method of modelled the softening behaviour of concrete in axial compression is to modify
the strain softening slope of the concrete material model to ensure that a constant fracture
energy is maintained [6-9]. After several trial analyses, the descending branch of the concrete
compressive stress-strain input was modified to a constant slope from the peak stress to a stress
of 10 MPa at a compressive strain of 0.025, as shown in Figure 6.4. The FEM calculated
response using this modified concrete strain softening slope, as well as the wire longitudinal
and transverse reinforcing steel, is shown in Figure 6.5b. The FEM response correlated well
with the experimental data, accurately capturing the descending path of the axial load response.
Additionally, the response of the column analysis that incorporated the modified concrete
strain softening slope and no transverse reinforcing steel is also shown in Figure 6.5b. The
analysis without the transverse reinforcing steel confirmed the increase in strength that
occurred when the confining pressure was provided by the transverse stirrups.
Finally, instead of discretely modelling the transverse reinforcing bars, a simplified technique
of applying the confining pressure directly to the outside face of the column was trialled. The
3.35 MPa confinement pressure that was calculated by Mander et al. [6-4] was applied to the
column during a separate analysis step prior to the application of the axial load. The concrete
input used during this analysis included the modified strain softening slope that was described
above and shown in Figure 6.4. The resulting FEM calculated axial load response, shown in
Figure 6.5c, indicated reasonable correlation with the experimental data, but the peak
compressive strength was over-predicted by 12%. This over-prediction in strength may have
been a result of the maximum confining pressure being applied from the start of the analysis,
whereas in reality the confining pressure increases from zero at the start of the test until the
transverse reinforcing bars reach their yield stress.
In the column FEM analyses described above, the tensile stress-strain response of the concrete
followed the realistic descending path defined by Eq. 6.5, as shown in Figure 6.4. Trials were
conducted with the concrete tensile strength capped at the peak stress with no strain softening.
The capped tensile stress minimised convergence problems in the model and was found to have
negligible effect on the global response of the column. For this reason, a capped tensile stress
was used during subsequent analysis.
Additional columns tested by Mander et al. [6-4] were also modelled, and the results from
these analyses confirmed the conclusions that were described above. Furthermore, it was
- 158 -
found that the strain softening slope of the modified concrete compressive response was
dependent on the confinement pressure as well as the mesh size due to the fracture energy of
each specimen. This observation meant that the concrete stress-strain definition used in the
FEM needed to be calibrated to match the experimental response for each column.
6.2.3.2 Wall tests

In addition to circular column tests, Mander et al. [6-4] carried out a total of 16 tests on
confined concrete wall sections. These wall tests provided suitable data to validate the
accuracy of FEM procedures for the confined toe region of the PreWEC wall. Of the 16
different test variations, specimen Wall 12 was selected from the investigation by Mander et al.
[6-4] because the configuration of confinement reinforcement was similar to that of the
PreWEC test specimen. The details for the Wall 12 test specimen are given in Table 6.2. The
reinforcing steel arrangement used for Wall 12 was considered to provide a high level of
confinement, but because transverse rectangular hoop provide a less efficient confinement
pressure than circular spirals, the confinement effectiveness ratio was only 0.634.
Table 6.2 – Details of Wall 12 tested by from Mander et al. [6-4]
Height 1200 mm
Length 700 mm
Thickness 150 mm
Cover to hoops 25 mm
fc’* 41 MPa
Longitudinal rebar / yield stress* 16 D16 / 290 MPa
Transverse rebar / yield stress* R10 spiral, 42 mm spacing / 360 MPa
Confinement effectiveness, ke 0.634
Confining pressure, f lx' / fly ' # 6.66 / 9.49 MPa
* measured values # calculated based on yield strength of transverse reinforcement

The FEM of Wall 12 was developed in the same manner as for the column analysis described
in section 6.2.3.1. The first wall FEM analysed the behaviour of an unconfined concrete wall,
with no longitudinal and transverse reinforcing steel. The unconfined concrete material stress-
strain definition was calculated using Eq. 6.5, and is shown in Figure 6.6. For the unconfined
wall analysis, the concrete tensile strength followed the descending strain softening curve
defined by Eq. 6.5. The monotonic axial load response calculated by the unconfined wall FEM
is compared with the measured experimental response in Figure 6.7a. Overall the FEM
calculated response showed good correlation with the experimental results, closely matching
- 159 -
the peak strength. There was some residual strength observed in the FEM response due to the
concrete compression and tension stresses not dropping completely to zero.
-80
-70
-60
-50
Stress (MPa)
Unconfined
-40
Unconfined: Modified
-30 Confined
Confined: Modified
-20
-10
10
0 -0.01 -0.02 -0.03 -0.04 -0.05
Strain (m/m)
Figure 6.6 – Concrete stress-strain input used for Wall 12 FEM
Following the unconfined analysis, the wall FEM was modified to include the longitudinal and
transverse reinforcing steel. Based on the findings from the column analysis, the reinforcing
steel was modelled as embedded wire truss elements and the unconfined concrete compressive
strain softening slope was modified to fit the shape of the descending path of the experimental
axial load response. For the wall analysis, the compressive stress dropped at a constant slope
from the maximum strength to a stress of 10 MPa at a strain of 0.035, as shown in Figure 6.6.
Additionally, the concrete tensile strength was capped at the peak stress with no strain
softening. It was found during the column analysis that capping the tensile strength at the peak
stress reduced convergence problems and had negligible effect on the model behaviour.
The axial load response calculated from the Wall 12 FEM analysis with wire reinforcing steel
is shown in Figure 6.7b, alongside the backbone of the measured experimental axial load
response. It can be seen that the FEM significantly overestimated the wall’s compressive
strength, by as much as 50%. This over-strength was attributed to the behaviour of the cover
concrete. Unlike the column, where the cover concrete constitutes only a small proportion of
the cross sectional area, the cover concrete in the Wall 12 specimen equated to approximately
45% of the cross-sectional area. The FEM did not accurately capture the crushing and spalling
of the cover concrete, which was represented by the dip in the experimental axial load
- 160 -
response. The transverse wire reinforcing steel was not only confining the core concrete, but
also restraining the cover concrete because the concrete model did not allow for deterioration
of the bond between the core and cover concrete that represented spalling. Additionally,
whereas the column had high confinement effectiveness, the effectiveness ratio of the wall
confinement was only 0.634. Because the FEM did not capture spalling of the cover concrete,
the area of confined core concrete was not accurately modelled. This deficiency led to more
effective confinement provided by the transverse reinforcing steel in the FEM, when compared
to the test specimen, which also contributed to the observed over-strength.
5 9
Test: Unconfined
8
FEM: Unconfined
4 7
6
Axial load (MN)
Axial load (MN)
3
5
4
2
3
Test
2 FEM: Wire rebar
1
FEM: Core only
1
FEM: Core + cover
0 0
0 0.005 0.01 0.015 0 0.01 0.02 0.03 0.04 0.05 0.06
(a) Unconfined (b) Rebar, core, and cover
8 8
Test: Cyclic
7 7
FEM
6 6 FEM: Damage indicies

Axial load (MN)
Axial load (MN)
5 5
4 4
3 3
2 2
1 Test 1
FEM: Modified input
0 0
0 0.01 0.02 0.03 0.04 0.05 0 0.01 0.02 0.03 0.04 0.05
(c) Modified input (d) Cyclic analysis

Figure 6.7 – Results for FEM analysis of concrete confinement effects on Wall 12
- 161 -
Two further analyses were conducted to investigate the influence of the cover concrete. The
first analysis involved modelling just the confined core region of Wall 12. The 30 mm cover to
the center of the transverse hoops was removed from the wall FEM, and the core was modelled
with embedded wire longitudinal and transverse reinforcing steel. For this analysis, the
concrete material definition used the same compressive stress-strain response, with the
modified strain softening slope. The calculated response of the “core only” FEM, shown in
Figure 6.7b, correlated well with the experimental data, but only after the cover concrete had
spalled during the test. The core FEM resulted in a significant under prediction of the initial
wall stiffness. Prediction of the initial stiffness is critical to the overall performance of the
PreWEC model and so the contribution of the cover concrete needed to be addressed.
Accurate modelling of spalling concrete can be achieved by modifying the elastic modulus of
the cover elements to zero after a tensile stress limit is exceeded at the boundary between the
cover and core elements [6-11, 6-12]. However, this technique for modelling cover concrete
could not be implemented into the ABAQUS FEM, and so instead an alternative analysis was
conducted that included the cover concrete as separate elements that were not connected to the
core at any stage during the analysis. In this analysis, the core was modelled as described for
the pervious analysis, and the cover concrete was modelled using the unconfined concrete
stress-strain response that was shown previously in Figure 6.6. The axial load response of this
“core + cover” FEM are also shown in Figure 6.7b. The core and cover FEM accurately
calculated the wall’s initial stiffness and captured the peak load and dip that occurred as the
cover crushed. The cover concrete still contributed some residual strength after it crushed,
unlike the complete loss of strength that occurs with real spalling, so the FEM slightly
overestimated the strength during the descending section of the axial load response.
Although modelling the core and cover separately resulted in reasonable correlation with the
experimental response, this technique was not considered suitable for use in the PreWEC FEM.
The complexity of the detailing required for the core plus cover FEM made it computationally
inefficient and caused instability during the analysis. Additionally, the core plus cover FEM
still required the compression strain softening response to be adjusted to match the
experimental results, which was considered not ideal for the PreWEC FEM. A simplified
procedure was developed to reduce the complexity, while still accounting for the confined core
and unconfined cover concrete. Instead of explicitly modelling the transverse reinforcement to
provide the confining pressure on the concrete core, the calculated confined concrete stress-
- 162 -
strain response can be directly applied to the concrete material definition in the FEM. Eq. 6.1
was used to calculate the confined concrete strength, fcc', which for the Wall 12 test specimen
was 79 MPa. Eq. 6.5 was then used to calculate the compressive stress-strain response of the
confined concrete, which is shown in Figure 6.6. The confined concrete stress-strain response
can be assumed for the entire confined core of the wall because the confinement effectiveness
was already included when calculating the effective confining pressure.
Applying the confined concrete stress-strain response to the core overcame the need to
explicitly model the transverse stirrups, but accurate representation of the cover concrete still
needed to be addressed. It was desirable to not model the core and cover separately in the
PreWEC FEM, so the behaviour of the core and cover elements were instead averaged across
the entire wall section. Before the unconfined compressive strength was reached, the confined
and unconfined stress-strain responses are approximately equal, see Figure 6.6. However, after
the cover crushes and spalling occurs, the cover was assumed to have negligible strength and
the confined core strength was averaged across the entire section. Excluding the 30 mm cover
to the centreline of the stirrups the core concrete for Wall 12 accounted for 55% of the cross-
section area. The confined stress-strain response was then modified, with all stresses in excess
of the unconfined strength multiplied by a factor of 0.55 to average the strength across the
entire section. This modified compressive stress-strain response is also shown in Figure 6.6.
The calculated axial load response of the Wall 12 FEM using this simplified averaging
technique is shown in Figure 6.7c. The calculated response showed good correlation with the
measured experimental response, closely matching the initial stiffness, peak strength and
descending curve. The FEM response did not reflect the immediate effects of the cover
spalling but instead produced a smooth averaged response though the middle of the peak and
dip associated with cover spalling during the experiment. Because of the accuracy and the
reduced complexity that minimised computational time and convergence issues, this simplified
technique was considered most suitable for inclusion in the PreWEC FEM.
The wall analysis above focused on the monotonic load response, whereas the PreWEC FEM
required a cyclic analysis to be completed, and so the cyclic hysteresis behaviour of the
confined concrete was also investigated. During the Wall 12 experimental test, the applied
axial force was unloaded and reloaded at two stages to measure the unloading stiffness. An
initial comparison between the cyclic axial load behaviour calculated by the simplified or
- 163 -
averaging FEM and the test results is shown in Figure 6.7d. As expected, because no damage
indices were defined in the concrete damaged plasticity model, the FEM unloading stiffness
was steeper than the unloading stiffness observed during the experimental test. Following this,
Eqns. 6.9 and 6.10 were used to calculate the unloading stiffness, which was introduced into
the FEM by defining damage indices for given tension and compression strain magnitudes.
The response of the FEM which included these calculated damage indices is also shown in
Figure 6.7d. With the damage indices included, the FEM accurately predicted the unloading
and reloading path of the experimental axial load response. Because the loading was similar to
the axial compression force on the wall toe, this analysis provided sufficient validation of the
averaging technique for modelling the confined concrete in the PreWEC FEM.
6.2.4 PreWEC Confined Toe Region

Using the knowledge gained from the FEM analyses of the Column 7 and Wall 12 test
specimen, the confined toe region of the PreWEC wall was modelled. As explained in section
6.2.3.2, the technique of directly defining the confined concrete stress-strain response to the
elements, and averaging the strength of the confined core and unconfined cover regions,
provided a simple and accurate modelling technique.
To define the behaviour of the confined region of the PreWEC test specimen, the effective
confinement pressure and section properties were calculated. A summary of calculated
confinement properties are presented in Table 6.3, and detailed calculations are included in
Appendix B. Drawings of the confined toe region of the PreWEC test specimen, as shown in
Figure 6.8, and additional detail of the test specimen has been published by Aaleti [6-13]. The
rectangular stirrups, placed at 50.8 mm centers in the corners of the wall, provided confinement
to the toe, where high strains are generated as the wall rocks. Using the procedure described
by Mander et al. [6-1] for rectangular concrete sections, the confinement effectiveness ratio
was calculated to be 0.587. Using Eq. 6.1, the confined concrete strength, fcc', was calculated
to be 1.46 fc' or 92 MPa.
- 164 -
Table 6.3 – Summary of PreWEC confined region properties
Property Calculated value

'
fc 63 MPa
Ec 38.78 GPa
εco 0.00244
ke 0.587
flx' 5.90 MPa
fly' 4.15 MPa
fcc'/ fc' 1.46
fcc' 92 MPa
εcc 0.00805
Cover factor* 0.87 [13% cover]
* the cover factor is the ratio of the core area to the total section area
1828.8
368.3
152.4
PreWEC wall cross-section

368.3
# 4 rebar (D13): Area = 127 mm^2
85.1 107.0 85.1
# 3 rebar (D10): Area = 71 mm^2
152.4 89.8
all dimensions in mm
Confined Toe Region
Figure 6.8 – Details of the confined toe in the PreWEC test specimen
The compressive stress-strain response of the confined core concrete in the PreWEC toe was
calculated using Eq. 6.5, and is plotted in Figure 6.9. As discussed during the modelling of
Wall 12, the strength of the confined core concrete was averaged over the entire cross-section
to account for the loss of strength when the cover concrete crushed and spalling occurred.
Because the PreWEC test specimen was constructed at 50% scale, the clear cover to the
transverse stirrups was only 10 mm. The confined core boundary was defined as the centreline
of the transverse stirrups, which resulted in a total cover depth of 15 mm. Because the
confined toe is at the ends of the wall cross-section, only three sides of the wall toe had cover
concrete that spalled during the test. Thus, the cover concrete was calculated to account for
- 165 -
13% of the cross-sectional area of the confined toe region. As per the technique validated for
Wall 12, the confined concrete strength of the toe region was modified by reducing all stresses
that were in excess of the unconfined concrete strength by 13%. This modified stress-strain
response was used as input for the material definition of the wall toe, and is also plotted in
Figure 6.9.
-100
-80
Stress (MPa)
-60
-40
-20 Unconfined
Confined
Modified
0
0 -0.005 -0.01 -0.015 -0.02
Strain (m/m)
Figure 6.9 – Concrete compressive stress-strain definitions for PreWEC
6.3 MODELLING OF PREWEC

This section describes the construction of the FEM of the PreWEC test specimen that was
tested at NCREE in Taiwan [6-14]. A summary of the specimen details is given, followed by
descriptions of the material properties, FEM formulation, and analysis solution control.
6.3.1 Details of PreWEC Test Specimen

A drawing of the PreWEC test unit and experimental setup is shown in Figure 6.10 and a
summary of the critical dimensions and properties of each element is included in Table 6.4. A
more extensive description of the PreWEC test specimen and experimental results has been
published by Aaleti [6-13].
The precast concrete wall was seated on a concrete foundation block, which was anchored to
the laboratory strong floor. The wall was inserted into a pocket that was constructed in the top
of the foundation block and seated on 31.8 mm high spacers. A high strength steel fiber grout
- 166 -
6.3 Modelling of PreWEC
was then placed into the 31.8 mm gap between the foundation block and the wall to provide a
robust contact surface for the wall to rock on. The columns were erected at each end of the
wall and also seated on the high strength grout bedding. The columns were constructed from a
rectangular steel tube section which was filled with a high strength concrete grout. A total of
20 O-Connectors (described in Chapter 5), 10 on each side, were welded to the steel column
and steel plates embedded into the precast concrete wall. Restraining plates were used to
prevent the connectors from experiencing any out-of-plane buckling. A T-shaped precast
concrete beam was seated on top of the wall to provide attachment for the actuator and a top
anchorage for the wall post-tensioning tendons. Post-tensioning tendons were inserted into
ducts that were cast into the center of the wall and column elements and anchored at the bottom
within the foundation block. Because all of the post-tensioning tendons were located within a
single duct at the center of the wall, the wall section was enlarged at the center to prevent the
hollow duct from weakening the precast concrete panel. Minimum required shear and
longitudinal reinforcing steel was placed within the wall, as well as confinement stirrups that
were placed in the wall toe. In addition, two steel channels were cast into the toe region at the
wall base to protect the corners from extensive damage when the wall rocked.
A single actuator was used to apply a reverse cyclic lateral displacement history to the loading
beam at the top of the wall. The loading height was calculated to be 6096 mm above the base
of the wall. Full details of the experimental performance of the PreWEC test specimen are
published by Aaleti [6-13] and Sritharan et al. [6-14].
Table 6.4 – Details of PreWEC test setup
Element Description
Wall 1828.8 mm long × 152.4 mm wide × 5875 mm high
Column 203.2 mm long × 152.4 mm wide × 5621 mm high (6.35 mm thick plate)
Connector Refer to O-Connector description in Chapter 5
Foundation block 3658 mm long × 1270 mm wide × 968 mm high
Embedded connector plate 159 mm long × 254 mm high × 10 mm thick
Wall PT 12 × 15.2 mm diameter strand (Aps = 1680 mm2 and lp = 7900 mm)
Column PT 3 × 15.2 mm diameter strand (Aps = 420 mm2 and lp = 6650 mm)
Wall shear rebar 2 × D10 @ 150 mm centers (Grade 60)
Wall longitudinal rebar 12 × D13 and 12 × D10, as per cross section (Grade 60)
Wall confinement rebar D10 stirrups @ 50.8 mm centers, as per cross section (Grade 60)
Channels at wall base 152.4 mm wide × 75 mm high × 610 mm long (6.35 mm thick plate)
Aps = total area of post-tensioning steel and lp = unbonded tendon length
- 167 -
1828.8 900.0
Actuator Loading beam
CFT Column
Connector
6096.0
Connector plate
5621.0
965.2
high-strength
558.8 grout
South North
967.5 end end
Base block
3657.6 1270.0
Elevation
203.2
152.4
1828.8
2286.0
Cross-section all dimensions in mm
Figure 6.10 – Drawing of PreWEC test setup
- 168 -
6.3.2 Material Properties

Samples taken from the materials used to construct the PreWEC specimen were tested to verify
their actual properties for use in analytical models. Details of these material tests were also
reported by Aaleti [6-13], and a summary of the measured properties of each material is shown
in Table 6.5.
Table 6.5 – Summary of measured PreWEC material properties
Material Location Property Average Measured Value

'
Concrete Wall fc 63 MPa
'
Grout Base of wall fc 92 MPa
Steel PT tendons Es 192 GPa
fy ~1600 MPa
Steel Connectors Es 214 GPa
fy 400 MPa
Steel Columns / plates Es 201 GPa
fy 359 MPa
Steel Rebar Es 200 GPa
fy 440 MPa
fu 650 MPa
6.3.2.1 Concrete and grout

The concrete materials in the PreWEC FEM used the “concrete damaged plasticity” model,
which was described in section 6.2.2. The stress-strain definition used for the unconfined
sections of the precast wall was derived from the measured compressive strength using Eq. 6.5.
However, during the analysis both the compression and tension strain softening paths were
modified to prevent convergence issues. In the compression region, the slope of the strain
softening section was reduced, ending at 0.1fc' at a compressive strain of 0.01. The tensile
strength was capped at the maximum stress of 0.62 f c' (MPa), and no strain softening was
permitted. The modified unconfined stress-strain definitions that were input into the FEM are
shown in Figure 6.11a, alongside the unmodified response calculated using Eq. 6.5. Since no
significant damage was observed in the unconfined regions of the wall panel, the reduced
compressive strain softening, and the capped tensile strength, had no significant effect on the
results of the analysis. The confined compressive stress-strain definition used the modified
response which was discussed in detail in Section 6.2.4 (also shown in Figure 6.11a). As
described previously, the modified confined stress-strain definition was applied to the whole
toe region, including the confined core and cover sections.
- 169 -
Although no test cylinders were cast from the concrete batch that was used to construct the
foundation block, the mix design was the same as for the wall concrete. For this reason, the
unconfined concrete definition was also applied to the foundation block, with an assumed
compressive strength of 63 MPa.
-90 2000
-80
-70
1500
-60
Stress (MPa)
Stress (MPa)
-50
Unconfined
-40 Unconfined (Modified) 1000
-30 Confined (Modified)
-20 Test coupon 1

500
-10 Test coupon 2
Test coupon 3
0
FEM input
10 0
0 -0.005 -0.01 -0.015 -0.02 -0.025 -0.03 0 0.005 0.01 0.015 0.02 0.025
(a) Concrete (b) PT tendons
600 500
500
400
400
Stress (MPa)
Stress (MPa)
300
300
200
200
Test coupon 1 Test coupon 1
Test coupon 2 100 Test coupon 2
100
Test coupon 3 Test coupon 3
FEM input FEM input
0 0
0 0.02 0.04 0.06 0.08 0 0.02 0.04 0.06 0.08
(c) O-Connector steel (d) CFT column / embedded plate steel

Figure 6.11 – Material stress-strain definitions used for the PreWEC FEM
High strength steel fiber reinforced grout was used as a bedding layer at the base of the wall.
Because the grout layer is highly confined, both from the steel fibers and its placement within a
pocket underneath the wall base, it was not expected to experience significant non-linear
- 170 -
behaviour. This assumption was confirmed by experimental observations, with all visible
damage isolated to the wall toe region and no crushing of the grout bedding layer observed.
For this reason, the bedding grout was modelled using an elastic stress-strain definition, with
an assumed modulus of elasticity of 40 GPa based on the manufacturer’s specification.
Additionally, the high strength grout that was used to fill the CFT columns was highly
confined by the surrounding steel shell and the columns were observed to behave elastically
during the experimental test. Therefore, the CFT concrete grout was modelled using the same
elastic material definition as the bedding layer grout.
6.3.2.2 Steel
The PreWEC FEM utilised five different steel material definitions for various elements in the
structure, including:
1. PT tendons,
2. O-Connector steel plate,
3. Steel plate used for CFT column and embedded plates,
4. Reinforcing steel,
5. Connector weld.
All of the steel materials were defined within the FEM using a standard plasticity model with
combined kinematic/isotropic hardening. With the exception of the reinforcing steel and the
connector weld, the stress-strain response for each material was defined based on the recorded
stress-strain response from the tensile coupon tests. The FEM stress-strain definitions used for
the PT tendons, O-Connector and CFT column/embedded plate are plotted in Figure 6.11
alongside the measured response from the three tensile coupons tested for each steel type.
Additionally, the elastic modulus defined for each steel type was the same as the experimental
values listed previously in Table 6.5.
The reinforcing steel tensile test coupons were not instrumented to obtain the full stress-strain
response, but instead only the yield strength and ultimate strength were determined from
testing. For this reason, the reinforcing steel material was defined using an assumed bi-linear
stress-strain response, with a modulus of elasticity of 200 GPa, yield strength of 340 MPa, and
ultimate strength of 650 MPa at a strain of 0.1. Additionally, the weld material used to attach
the O-Connectors used an assumed bi-linear stress-strain definition with a modulus of elasticity
of 200 GPa, yield strength of 450 MPa, and ultimate strength of 550 MPa at a strain of 0.1.
- 171 -
6.3.3 FEM Formulation

As explained earlier, the PreWEC FEM expanded on the previously validated 3D FEM of the
PT wall that was described in Chapter 4. The precast concrete wall was modelled using brick
elements. Due to meshing difficulties, the post-tensioning duct at the center of the wall panel,
as well as the enlarged wall thickness, was not included in the FEM. Instead, the wall was
modelled as a solid rectangular section, which was considered to have negligible effect on the
wall behaviour. The wall panel was partitioned to allow the modified confined concrete
material definition to be applied to the confined toe region, while the unconfined concrete
definition was applied to the remaining section of the wall, as seen from the FEM assembly
shown in Figure 6.12. The wall was meshed so that the element size was reduced towards the
corner toes of the wall, both along the wall base and the height. This discretisation allowed for
a more detailed mesh in the critical toe region, where most of the non-linear behaviour was
expected. The element size in the corner toe was approximately 45 mm high × 50 mm long,
and expanded to 150 mm high × 75 mm long at the top of the wall. Three elements were used
though the thickness of the wall panel.
The CFT columns were modelled in a similar way to the wall panel model. However, the
column section was partitioned to allow the shell to be assigned the steel CFT material
definition, while the elastic grout definition was used for the centre elements. This partitioning
meant that the surfaces between the steel shell and grout fill were restrained together with no
degradation, debonding, or slip accounted for in the FEM. The column mesh used six elements
along the length of the column to accurately capture the bending and neutral axis depth, and the
element height was approximately 120 mm.
The foundation block that was used during the experiment was also included in the model. The
foundation block was partitioned to separate the pocket below the wall that was assigned the
elastic grout bedding material definition, while the unconfined concrete definition was applied
to the remaining sections of the foundation block. The grout pocket in the foundation block is
visible in Figure 6.12c. The foundation block had a mesh size of approximately 150 mm, but
the mesh was reduced in the grout pocket to provide more accuracy at the wall-to-foundation
interface. As with the PT wall FEM that was described in Chapter 4, the horizontal joint
between the wall and foundation (and column and foundation) was modelled using a contact
interaction that provided “hard contact” under compressive stresses while allowing unrestricted
- 172 -
“gap opening”. Additionally, the friction mechanism for transferring shear forces was
sufficient to prevent any slip occurring during experimental testing, so a “no slip” condition
was used to prevent horizontal wall sliding.
Because the top of the wall was not critical to the PreWEC behaviour, the T-shaped loading
beam was modelled using a discreetly rigid member, which was located at the 6096 mm
loading height. The wall height was extended to meet the rigid loading beam, which was
constrained to the top of the wall with no slip or uplift allowed. The rigid member evenly
distributed forces to the top of the wall, including the horizontal shear force from the actuator
and the vertical compression force from the post-tensioning steel.
The unbonded tendons were modelled as truss elements and assigned non-linear stress-strain
properties, as described in the previous section. The tendon lengths were equal to the
unbonded length that was measured prior to testing, as listed in Table 6.4. The bottom tendon
anchorage was represented by restraining the horizontal and vertical degrees of freedom of the
bottom tendon node. The top tendon anchorage of the column post-tensioning was modelled
by coupling the horizontal and vertical degrees of freedom of the top tendon node to the top
surface of the column. The top tendon anchorage of the wall post-tensioning was modelled by
coupling the horizontal and vertical degrees of freedom of the top tendon node to the top
surface of a load cell element. The load cell element was tied to the rigid loading beam and
assigned an elastic stiffness to represent the stiffness of the load cell used during the test.
The O-Connectors, which are visible in Figure 6.12d, were modelled in the same way as
described previously in Chapter 5. The O-Connector and weld sections were assigned their
respective steel material definitions that were described in section 6.3.2.2. The weld bond was
modelled by restraining the weld surface to the column face on one side, and to the embedded
steel plate on the other side. The steel plate was embedded into the concrete wall with an ideal
bond assumed between the concrete and steel elements.
During the initial analysis the longitudinal and shear reinforcing steel were included in the
FEM using embedded truss elements with a 50 mm mesh size and a stress-strain definition that
was described in section 6.3.2.2. Additionally, the steel channels were modelled with solid
elements with a mesh size of approximately 40 mm. The channels were initially embedded
into the concrete wall elements with perfect bond to the concrete, but this detail was later
modified, as described in detail below.
- 173 -
(a) Unmeshed assembly (b) Meshed assembly
(c) Wall base (d) O-Connector

Figure 6.12 – Illustrations of the PreWEC FEM assembly
- 174 -
During the experimental test, pinned struts were placed between the columns and the wall to
ensure that the columns were subjected to the same lateral displacement as the wall. The struts
were found to be unnecessary during the FEM analysis, with the O-Connectors sufficient to
laterally tie the columns to the walls. Additionally, lateral restraint was provided to the wall
during the experimental test to prevent any out-of-plane displacement. This lateral resistant
was included in the FEM by simulating a roller restraint to the loading beam.
Loading was applied using a series of analysis steps. First, the tendons were prestressed by
specifying an initial stress in the truss elements and the top of the tendons were initially
restrained to prevent any displacement due to the initial stress condition. The tendon restraints
were removed during the first load step, which allowed the prestress force to be transferred into
the wall as a precompression. Following the prestressing step, the wall self-weight was applied
as a gravity load during the second load step. Lastly, the reference point at the end of the
loading beam at the top of the wall was subjected to a displacement controlled lateral load
history.
6.3.4 Solution Control and Monotonic Trials

6.3.4.1 Implicit vs. explicit solver
Initially, the lateral displacement was applied during a “static riks” analysis step, using the
implicit (standard) solver available within ABAQUS. The implicit solver uses Newton’s
method to solve the non-linear equilibrium equations, and is particularly useful for non-linear
static analysis, where the inertia effects can be ignored. When the PreWEC system was
subjected to an increasing monotonic lateral displacement history, the implicit solver was
efficient at solving the response. However, during the cyclic analysis, the implicit solver
suffered extensive convergence problems and was unable to progress beyond a single load
reversal, even after the convergence tolerances were increased. The convergence problems
were attributed to the wall-to-foundation interface, which became a complex contact problem
as the wall rocked back and forward and the wall base became deformed due to crushing in
both wall toes. To overcome the solution convergence issues, the FEM analysis was shifted to
the explicit solver in ABAQUS. The explicit dynamic procedure performs a large number of
small time increments using a central-difference time integration rule. Although it is typically
used for dynamic analysis problems, the explicit solver is suitable to perform quasi-static
analyses with complicated contact conditions.
- 175 -
To enable the PreWEC FEM to utilize the explicit solver, several changes needed to be made
to the model. Because the explicit solver used a dynamic analysis step, the inertia effects
needed to be considered, with each material assigned an appropriate density. Additionally,
loading was conducted dynamically, but the velocities were limited so that a quasi-static
analysis was conducted with negligible inertia forces. It was not desirable to impart
displacement controlled loading conditions to the model when using the explicit solver, so
instead a velocity history was applied. The velocity was slowly ramped up at a constant rate
from zero to a maximum velocity, where it was continued for the remainder of the analysis
step. The ramp time, maximum velocity, and total step time were calculated to ensure that the
target displacement was reached at the end of the analysis step. A series of trials that were
conducted to determine an appropriate loading speed are described in detail below. Also, the
use of an initial stress condition for the tendon prestress was not appropriate for the explicit
analysis because it caused the precompression force to be released instantaneously, which
induced severe dynamic behaviour. Instead, the prestress was applied by subjecting the bottom
tendon node to a velocity condition to elongate the tendon. This applied prestress velocity was
ramped up over the total step time of 20s, which was found to be slow enough to prevent any
dynamic effects. Because the prestress loading elongated the tendon, the initial tendon length
was reduced to allow the measured unbonded tendon length to be achieved following the
prestress being applied. Additionally, the gravity load that was applied during the second
analysis step was ramped up at a constant rate, over a total time step of 10s.
After the solution control parameters were appropriately defined, as discussed below, the
explicit solver produced a monotonic lateral load response that was comparable to that
obtained from the implicit solver, as seen in Figure 6.13a. This correlation provided sufficient
confidence in the explicit solving technique for the cyclic analysis of the PreWEC system to be
completed.
6.3.4.2 Explicit solver control parameters

As described above, the loading speed must be slow enough to ensure that the PreWEC
specimen maintained the quasi-static test response. However, the computational time is also
dependent on the step time, and so the optimum loading speed had to be determined. Several
monotonic load trials were conducted to assess the influence of different loading speeds. The
ramp and hold velocity condition was conducted with a total step time of 10, 20, 30, and 50
seconds, which corresponded to maximum lateral velocities of 37.5, 12.5, 7.5, and 4.2 mm/s
- 176 -
respectively. A comparison of the monotonic lateral load response of the PreWEC FEM when
subjected to each of these loading speeds is shown in Figure 6.13b. It was observed that the 10
and 20s step times produced some erratic behaviour in the later stages of the analysis, whereas
the analysis with 30 and 50s step times showed a more stable response. For this reason, a
maximum velocity of 7.5 mm/s was used for the PreWEC analysis, which represented the
optimum balance between quasi-static speed and computational time.
600 600

500 500
400 400
300 300
200 200
10s (37.5 mm/s)
20s (12.5 mm/s)
100 100
Implicit 30s (7.5 mm/s)
Explicit 50s (4.2 mm/s)
0 0
0 50 100 150 200 0 50 100 150 200
(a) Solver type (b) Loading speed
600 600
500 500
400 400
300 300
200 200
No mass scaling
100 100
10x mass scaling Rebar
100x mass scaling No Rebar
0 0
0 50 100 150 200 0 50 100 150 200
(c) Mass scaling (d) Reinforcing steel

Figure 6.13 – Lateral force-displacement response for monotonic trials of PreWEC FEM
- 177 -
The computational time of an explicit dynamic analysis can also be reduced by artificially
increasing the material density, often referred to as “mass scaling”. However, the increase in
mass must be conducted carefully, because it also increases the inertia forces. Further trials
were conducted to assess the effect of mass scaling on the monotonic lateral load response of
the PreWEC FEM. The density of all of the materials were increased by 10 and 100 times and
compared to the analysis conducted with the true densities (i.e. no mass scaling). As shown in
Figure 6.13c, when mass scaling of 10 or 100 times was applied, the analysis became erratic
because the inertia forces strongly affected the lateral force output. Although it reduced the
computational time significantly, mass scaling was found to be not appropriate for the PreWEC
FEM and was subsequently not used during the cyclic analysis discussed below.
The explicit solving method generates a large number of increments (> 1 million), which
means the solution results can be susceptible to round off errors. The explicit solver includes a
“double precision” option to reduce round off errors, at the expense of computational time.
Trials using a double precision analysis indicated that this had negligible effect on the PreWEC
response, and so single precision was used throughout all analyses.
6.3.4.3 Inclusion of wall reinforcing steel

As well as optimising the solution control parameters, monotonic load trails were conducted to
confirm the accuracy of certain modelling decisions. The longitudinal and shear reinforcing
steel in the precast wall were initially included in the PreWEC FEM. However, as shown in
Figure 6.13d, the inclusion of the reinforcing steel had negligible effect on the monotonic
response, with an almost identical lateral load behaviour observed when no reinforcing steel
was included. This similarity in response was occurred because the wall panel did not undergo
any non-linear flexural or shear deformation and so reinforcement was not required to control
cracking. Additionally, the influence of the longitudinal bars on the neutral axis depth was
found to be negligible. When the reinforcing steel was included, the large number of
embedded elements caused a significant increase in the computational time, and so the
reinforcing steel was excluded from the subsequent cyclic analysis this is discussed below.
6.3.4.4 Steel channels in the wall toe

During the monotonic trial analyses, the steel channels placed in the wall toe were also
investigated. Initially, the whole channel was constrained as an embedded element within the
wall, but this was found to provide far greater protection to the wall toe because the channel
- 178 -
6.4 PreWEC Cyclic Response
was restrained from delaminating from the wall as it bent. The stress-strain response of the
concrete elements that the channel was embedded in experienced unusually high strengths,
similar to that observed earlier when solid confinement reinforcing steel was used. Several
trials were conducted to assess different modelling techniques for including the steel channels
into the FEM, and the most reliable method developed involved embedding only the bottom
web of the channel into the concrete wall and creating a contact interaction at the surface
between the side of the wall and the channel side flanges. This contact interaction allowed the
channel to protect the wall toe from excessive lateral expansion and damage, while also
providing provision for the channel flange to detach from the wall face when it was subjected
to excessive bending.
6.4 PREWEC CYCLIC RESPONSE

Following the monotonic lateral load analysis trials, the PreWEC FEM was subjected to the
full reverse cyclic lateral load history that was applied during the PreWEC test and the
calculated response was compared with the measured experimental results. Excluding the
small force-based cycles at the start of the experiment, the FEM load history consisted of
displacement controlled load cycles to ±0.15, 0.3, 0.5, 0.75, 1.0, 1.5, 2.5, and 3.0% lateral drift.
During the experiment, the PreWEC system was subjected to three cycles at each displacement
level, while the FEM was subjected to a loading history that consisted of either one or three
cycles at each displacement level, as discussed in more detail below. In addition, the
experimental test also included a single cycle to 3.5%, but this load cycle was excluded from
the analysis because at this stage during the test the connectors had begun to fracture, and the
FEM was not intended to capture the connector failure mode.
As discussed previously, when the explicit solver was used the lateral displacements were
applied to the model using velocity commands that ensured that a quasi-static response was
maintained. Each load step included a ramp up time, a period at a constant (peak) velocity, and
a ramp down period. For all load steps the maximum velocity specified was less than 9 mm/s,
and the maximum acceleration was less than 1 mm/s2. This loading speed conformed to the
recommendations from the trials that were described in section 6.3.4.2, and ensured that a
stable response was maintained. Additionally, each complete load cycle was separated into
four load steps:
- 179 -
6. Loading from zero displacement to maximum positive lateral displacement.

7. Unloading to zero displacement.
8. Loading from zero displacement to maximum negative lateral displacement.
9. Unloading to zero displacement.
This four step loading ensured that the velocity was ramped down to zero before the loading
direction was reversed and when contact occurred at the wall-to-foundation interface. Trials
indicated that loading the specimen at full velocity through the contact phase resulted in an
unstable response, and so this was avoided during the analysis.
6.4.1 General Behaviour

In general, the PreWEC FEM was observed to behave in a similar manner to the PreWEC test
specimen. The displaced shape and stress contours of the PreWEC FEM at 3% lateral drift are
shown in Figure 6.14. The uplift at the base of the wall and column elements was clearly
visible, as well as the plastic deformation of the O-Connectors. High stresses and strains were
generated in the wall toe, but the confined concrete region was sufficient to prevent significant
crushing, which is comparable with the behaviour observed during the experimental test.
6.4.2 Concrete Cyclic Parameters

The observed lateral force-displacement response of the PreWEC test specimen did not follow
the idealised flag-shape hysteresis behaviour that is often assumed for self-centering systems.
The observed response was attributed to non-linear behaviour of the concrete in the wall toe,
which caused the wall to deviate from a bilinear elastic response. Except for the connectors
and the wall toe, all of the components in the PreWEC system remain predominantly within
their elastic state. Because the non-linear behaviour of the connectors was well captured by the
O-Connector FEM (see Chapter 5), the only uncertainly in the PreWEC FEM was the
behaviour of the wall toe.
The cyclic load behaviour of the confined concrete in the wall toe was critical to overall
response of the PreWEC FEM, and so three different cyclic analyses were conducted to
investigate the influence of the concrete cyclic parameters. The first FEM analysis was
conducted with a basic concrete model for the confined toe region that did not include the
cyclic damage indices, which meant that the unloading and reloading stiffness of the concrete
- 180 -
was equal to the initial elastic stiffness. Next, the analysis was rerun with the damage indices
included. The damage indices were calculated as described previously in Section 6.2.4, and
resulted in more realistic unloading and reloading paths for the concrete stress-strain response.
These first two analyses were conducted with only one cycle completed at each displacement
level in order to reduce computational time. The final analysis was conducted with the
concrete damage indices and a loading history that included the full three cycles at each
displacement level, as was completed during the experimental test.
(a) Side view (b) Wall base

Figure 6.14 – PreWEC FEM at 3% lateral drift (displacements magnified 2×)
6.4.2.1 Global response

The lateral force-displacement response for the three different cyclic FEM analyses are shown
in Figure 6.15 for both the full cyclic record and the backbone extracted from the cycle peaks.
Overall, the FEM showed satisfactory correlation with the experimental response, closely
matching the initial stiffness, strength, and shape of the hysteresis loops. Additionally, in all
- 181 -
cases, the backbone response was well captured by the FEM with minimal difference observed
between the backbone responses of the three different analyses.
The main difference observed between the three cyclic FEM analyses occurred in the
unloading path of the hysteresis loop, when the lateral force was close to zero. For the FEM
with no damage indices, the calculated hysteresis loops showed significant pinching as the
lateral force approached zero, as seen in Figure 6.15a. This pinching caused the FEM to
deviate from the experimental response and resulted in under prediction of the residual
displacements. Additionally, the degradation in the reloading stiffness was poorly captured by
the FEM with no damage indices. The FEM response showed improved performance when
cyclic damage indices were included to accurately capture the concrete cyclic unloading and
reloading behaviour. As seen in Figure 6.15c, the FEM with damage indices showed a
reasonable correlation with the experimental unloading stiffness, particularly for the larger
displacement cycles. However, pinching of the hysteresis loops was still observed, particularly
during the reloading section just after the lateral force crossed the zero axis. As seen in Figure
6.15e, the unloading response was further improved when the FEM was subjected to the full
displacement history with three cycles at each displacement level. The repeated cycles resulted
in increased damage, which was apparent in the FEM with increased degradation of both the
unloading and reloading stiffnesses. However, pinching behaviour was still observed in the
hysteresis loops at approximately zero lateral force, and the increased damage caused
significant crushing of the wall toe during the final cycles to 3% lateral drift, and this was
visible in the unusual bulging of the unloading hysteresis curves.
To further examine the global hysteresis behaviour, the energy dissipation and residual drift
calculated from the FEM analyses were compared with the experimental results. The energy
dissipation was calculated from the area enclosed by the lateral force-displacement hysteresis
loops, using two different methods. Traditionally, energy dissipation is represented in the form
of equivalent viscous damping, ξeq, using the expression shown in Eq. 6.11. Additionally,
guidelines for the design of unbonded post-tensioned concrete wall systems, recently published
by ACI [6-15], include an alternative expression, referred to as the relative energy dissipation
ratio, β. As shown in equation 6.12, the relative energy dissipation ratio is proportional to the
equivalent viscous damping ratio, and thus both can be plotted on the same graph. The
residual drift was calculated from the residual displacement at zero lateral force, after
unloading from the peak displacement of a cycle.
- 182 -

-3 -2 -1 0 1 2 3 -3 -2 -1 0 1 2 3
600 600

400
400
200 200
0 0
-200 -200
-400 -400
Test Test
FEM FEM
-600 -600
-200 -100 0 100 200 -200 -100 0 100 200
(a) No damage indices (b) No damage indices (backbone)
-3 -2 -1 0 1 2 3 -3 -2 -1 0 1 2 3
600 600
400 400
200 200
0 0
-200 -200
-400 -400
Test Test
FEM FEM
-600 -600
-200 -100 0 100 200 -200 -100 0 100 200
(c) With damage indices (d) With damage indices (backbone)
-3 -2 -1 0 1 2 3 -3 -2 -1 0 1 2 3
600 600
400 400
200 200
0 0
-200 -200
-400 -400
Test Test
FEM FEM
-600 -600
-200 -100 0 100 200 -200 -100 0 100 200
(e) Three cycles (f) Three cycles (backbone)
Figure 6.15 – Comparison of measured and calculated lateral force-displacement
responses for the PreWEC FEM
- 183 -
1 Aloop
 eq  (6.11)
4 Astrain
Aloop Aloop 
    eq (6.12)
Aideal 8  Astrain 2
where Aloop is the area enclosed by a single hysteresis loop, Astrain is the area equal to the strain
energy measured at the peak of that same cycle, and Aideal is the area of an ideal elasto-plastic
hysteresis loop representing maximum energy dissipation.
A comparison between the equivalent viscous damping and energy dissipation ratios calculated
for the experimental test and each of the three FEM cyclic analyses is shown Figure 6.16a. For
each analysis, the FEM produced almost identical energy dissipation, indicating that as more
cyclic damage occurred both the unloading and reloading paths were degraded, but the overall
hysteresis area remained unchanged. Overall, the FEM showed good calculation of the energy
dissipation when compared to the experimental test. The FEM slightly underestimated the
energy dissipation during the larger cycles to 2-3% lateral drift, which may be a result of the
pinching behaviour previously observed in the hysteresis loops.
20 1
0.3
Equivalent Viscous Damping,  eq (%)
Test
Relative Energy Dissipation Ratio, 
No damage indices
0.25 0.8
15 With damage indices
3 cycles
Residual Drift (%)
0.2
0.6
10 0.15
0.4
0.1
5 Test
No damage indices 0.05 0.2
With damage indices
0 3 cycles 0
0
0 0.5 1 1.5 2 2.5 3 3.5 -3 -2 -1 0 1 2 3
Lateral Drift Cycle (%) Lateral Drift Cycle (%)
(a) Energy dissipation (b) Residual drift

Figure 6.16 – Comparison of measured and calculated hysteresis loop parameters for the
PreWEC FEM
- 184 -
Additionally, in Figure 6.16b the residual displacements that were calculated for each of the
three FEM analyses are compared with the residual displacement measured during the
experimental test. As was observed previously from the hysteresis loops, the pinching
behaviour caused the FEM to under predict the residual displacement by up to 50%. Up until
the 2% lateral drift cycle, no significant difference was observed between the residual
displacements calculated for each of the three FEM analyses. At lateral drifts above 2%, the
cyclic damage accumulation caused the residual displacements to be larger when the damage
indices were included, and larger again when the FEM was subjected to three cycles at each
displacement level. The residual displacements calculated for the analysis with three cycles at
each displacement level showed reasonable correlation with the experimental results at 2.5%
drift, but overestimated the residual displacements at 3% drift. The pinching behaviour that
was observed from the hysteresis loops and the difficultly in accurately calculating the residual
displacements were attributed to the gap closure behaviour in the FEM, as discussed in more
detail below.
6.4.2.2 Local response

Local responses were also investigated to further assess the accuracy of the FEM for
calculating the cyclic response of the PreWEC specimen, including unbonded tendon stress in
the wall PT, neutral axis depth, base rotation, concrete compressive strains, and connector
displacements. The deficiency of the cyclic analysis that did not include cyclic damage indices
was obvious from the previous investigation of the global response, and so only the results
from the two cyclic analyses that included the damage indices are discussed in this section.
These two analyses are referred to as “1 cycle” and “3 cycles”, representing the number of load
cycles that the FEM was subjected to at each displacement level.
The FEM calculated tendon stress for the wall post-tensioning is compared with the
experimental results in Figure 6.17 for both the full cyclic and extracted backbone response. In
general, the FEM showed good correlation with the experimental tendon stress data for both
the 1 cycle and 3 cycle analyses. The backbone responses indicated that the increase in tendon
stress was well captured by the FEM, especially for low drift levels. Some deviation occurred
at high drift levels, with the FEM over predicting the measured peak tendon stresses. In
particular, the experiment showed unexplained large losses in prestress during the repeated
cycles to 3% drift, which was not accurately captured by either FEM. Overall, the 3 cycle
FEM captured the loss in initial prestress better than the 1 cycle FEM, with increased
- 185 -
degradation occurring during repeated cycles. It should be noted that the FEM indicated that
the prestress loss was predominantly caused by wall shortening, due to inelastic strains at the
wall base, and not yielding of the prestressing tendons, which remained close to their elastic
state. The overestimation of tendon stress during large drift cycles may have been contributed
to by seating losses at the post-tensioning anchor during the test, which was not captured in the
PreWEC FEM.

-3 -2 -1 0 1 2 3 -3 -2 -1 0 1 2 3
1600 1600
Test Test
FEM (1 cycle) FEM (3 cycles)
1500 1500
Stress (MPa)
Stress (MPa)
1400 1400
1300 1300
1200 1200
1100 1100
-200 -100 0 100 200 -200 -100 0 100 200
(a) 1 cycle at each peak (b) 3 cycles at each peak
Lateral Drift (%)

-3 -2 -1 0 1 2 3
1600
Test
FEM (1 cycle)
1500
FEM (3 cycles)
Stress (MPa)
1400
1300
1200
1100
-200 -100 0 100 200
(c) Backbone from peaks

Figure 6.17 – Comparison of measured and calculated wall unbonded tendon stresses for
the PreWEC FEM
- 186 -
To investigate the behaviour at the wall base, the neutral axis (NA), or contact depth, and the
base rotation were calculated. Both the NA and base rotation were calculated from the
displaced profile along the wall base, as described previously in Chapter 3. A comparison
between the FEM and experimentally calculated NA and base rotation responses is shown in
Figure 6.18. To allow for a visible comparison of the results, only the backbone responses that
were extracted from the cycle peaks are shown. The FEM showed excellent correlation with
the test results for both the NA and base rotation. Additionally, no significant difference was
observed between the results of the 1 cycle and 3 cycle FEM analyses. The accurately
calculated NA depth provided increased confidence in the technique used to model the
confined toe region, because the average strength of the toe region is directly related to the NA
depth.

-3 -2 -1 0 1 2 3 -3 -2 -1 0 1 2 3
 0.03
Test
//
1400 FEM (1 cycle) 0.02

FEM (3 cycles)
1200
Base Rotation (rad)
0.01
1000
NA (mm)
800 0
600
-0.01
400
Test
-0.02
200 FEM (1 cycle)
FEM (3 cycles)
0 -0.03
-200 -100 0 100 200 -200 -100 0 100 200
(a) Neutral axis (NA) depth (b) Base rotation

Figure 6.18 – Comparison of measured and calculated neutral axis depth and base
rotation backbones for the PreWEC FEM
To further investigate the behaviour of the confined wall toe, a comparison between the
calculated and experimentally measured concrete strains was also conducted. During the
experimental test, concrete strains were measured using strain gauges that were mounted
vertically in the wall toe, 100 mm in from the wall end face and 150 mm above the wall base,
at both corners of the wall. The extracted backbones of these experimental strain
measurements are shown in Figure 6.19, alongside the FEM calculated strains that were
extracted from elements at the corresponding location to the strain gauges. As the wall rocked,
- 187 -
the toe at the north end of the wall was in compression during positive displacement cycles and
the toe at the south end of the wall was in compression during negative displacement cycles.
The compressive strains that were measured and calculated when the toe was not in contact
with the foundation (negative displacements for the north toe and positive displacements for
the south toe), represent the residual strain after unloading from the maximum compressive
stress. The FEM calculated concrete strains at the north end of the wall correlated well with
the experimental measurements, which provided further confidence in the modelling of the
confined concrete in the wall toe. Additionally, the concrete strains calculated from the 3 cycle
FEM are similar to those calculated from the 1 cycle FEM, except that an increase in strain was
observed during repeated cycles. The FEM calculated strains at the south end of the wall were
slightly higher than the experimentally measured strains. Additionally, the measured strains at
the south end were lower than at the north end, which again highlighted the variability in
experimental strain measurements that was previously discussed in Chapter 3.

-3 -2 -1 0 1 2 3 -3 -2 -1 0 1 2 3
0.02 0.02
Test Test
FEM (1 cycle) FEM (1 cycle)
FEM (3 cycles) 0.015 FEM (3 cycles)
0.015
Compressive Strain
Compressive Strain
0.01 0.01
0.005 0.005
0 0
-200 -100 0 100 200 -200 -100 0 100 200
(a) North end strain gauge N1 (b) South end strain gauge S1
Figure 6.19 – Comparison of measured and calculated concrete strain backbones for the
PreWEC FEM
Lastly, the relative vertical displacement between the wall and columns at the connector
locations was investigated. Figure 6.20 shows a comparison between the backbones of the
measured and calculated vertical displacements at two connector locations, being at the second
connector up from the wall base on the south end and the fifth connector up from the wall base
on the north end. For both connector locations the FEM showed excellent correlation with the
- 188 -
experimental displacements, except for overestimation of the negative displacements during

large drift cycles. Again, no significant difference was observed between the 1 cycle and
3 cycle FEMs.

-3 -2 -1 0 1 2 3 -3 -2 -1 0 1 2 3
50 50
Test Test
40 FEM (1 cycle) 40 FEM (1 cycle)
Connector Displacement (mm)
Connector Displacement (mm)

FEM (3 cycles) FEM (3 cycles)
30 30
20 20
10 10
0 0
-10 -10
-20 -20
-200 -100 0 100 200 -200 -100 0 100 200
(a) South end connector 2 (b) North end connector 5

Figure 6.20 – Comparison of measured and calculated connector vertical displacement
backbone for PreWEC FEM
6.4.3 Wall-to-Foundation Contact Properties

As described above, the FEM captured the PreWEC experimental response with good
accuracy. The main discrepancy was that the FEM showed a pinching response in the lateral
force-displacement hysteresis loops, which caused the FEM response to deviate from the
experimental unloading path and substantially under predict the residual displacement. It was
suspected that this pinching may have been caused by the ideal conditions of the wall-to-
foundation contact. In the FEM, a hard contact property was used which meant that no force
was transferred until the two surfaces contacted. During the experiment, debris from the
spalling cover concrete was observed to fall into the open crack at the wall base. This problem
may have been exacerbated by the grout pocket detail, which prevented debris from escaping
once trapped between the wall and foundation.
In an attempt to recreate the behaviour of the test specimen, the contact definition at the wall-
to-foundation surface was modified. Instead of hard contact in compression, an exponential
over-closure contact definition was used, as shown in Figure 6.21. The over-closure was
- 189 -
defined such that as the gap closed, contact was initiated when the surfaces were 2 mm apart
(co value), and the contact pressure then increased exponentially until full closure was achieved
when the contact pressure reached 20 MPa (po value). Additionally, the FEM used for this
analysis included the cyclic concrete damage indices and was conducted with only one load
cycle at each displacement level. In Figure 6.22 the lateral force-displacement results of this
FEM analysis are compared against the experimental response. The unloading hysteresis
curves showed less abrupt pinching, but the residual displacements were actually reduced when
compared to the previous FEM with hard contact (Figure 6.15c). In addition to not resolving
the residual displacement issue, the FEM with the revised contact definition showed reduced
strength, with the lateral force-displacement backbone significantly below the experimental
response. Several trials were conducted with different parameters for the over-closure contact
definition, but no improvement of the unloading response or residual displacement calculation
was achieved.
Figure 6.21 – Over-closure contact definition in ABAQUS [6-3]
6.4.4 Recommendations
Based on the results of the PreWEC cyclic FEM analyses, the following recommendations
were made:
 Defining damage indices for the confined concrete in the wall toe was critical in order
to accurately capture the concrete unloading and reloading behaviour. Inclusion of the
damage indices led to improved calculation of the PreWEC cyclic response, with
degradation of the unloading and reloading stiffnesses and increased residual
displacements.
 Both the FEM analyses with 1 cycle and 3 cycles at each displacement level showed
reasonable correlation with the PreWEC cyclic response. The 3 cycle FEM showed
- 190 -
6.5 Modified PreWEC Designs
increased cyclic degradation, but this had no significant impact on many of the local
parameters investigated. When considering accuracy as well as computational time, the
1 cycle FEM provided a more useful analysis tool and was thus used for the subsequent
analytical investigations described below.
 Although the FEM accounted for the non-linear behaviour of the wall toe, the FEM did
not accurately capture the residual displacement of the PreWEC system. The prediction
of residual displacements is critical to the design of self-centering systems, as discussed
in Chapter 7, and this remains a topic for future investigation.

-3 -2 -1 0 1 2 3 -3 -2 -1 0 1 2 3
600 600

400 400
200 200
0 0
-200 -200
-400 -400
Test Test
FEM FEM
-600 -600
-200 -100 0 100 200 -200 -100 0 100 200
(a) Cyclic (b) Backbone

Figure 6.22 – Comparison of measured and calculated lateral force-displacement
response for PreWEC FEM with modified contact properties
6.5 MODIFIED PREWEC DESIGNS

Following the successful validation described above, the PreWEC FEM was used to study the
impact of the major design parameters. Although the connectors accounted for only 35% of
the total moment capacity of the PreWEC system, perfect self-centering behaviour was not
achieved because of the large in-elastic strains in the wall toe. A series of modified PreWEC
designs were created in ABAQUS by changing the PT configuration, number of connectors, or
the wall properties. The objective of these modified designs was to investigate the effect of
these changes on the PreWEC performance and identify where improvements to the design
- 191 -
could be achieved. The FEM of the PreWEC test specimen that included the cyclic damage
indices and one cycle at each displacement level was used as the control, referred to below as
PreWEC-t (where “t” represents test). A summary of the design changes for the five modified
PreWEC designs is given in Table 6.6.
Table 6.6 – Summary of changes to the PreWEC design
Label PT Connectors MPT : Mcon* Wall Thickness

PreWEC-t Aps = 1680 mm2 and fse = 1200 MPa 20 connectors 0.65 : 0.35 152.4 mm
PreWEC-A Aps = 1350 mm2 and fse = 1200 MPa 28 connectors 0.50 : 0.50 -
PreWEC-B Aps = 1900 mm2 and fse = 1200 MPa 16 connectors 0.72 : 0.28 -
PreWEC-C Aps = 1680 mm2 and fse = 890 MPa 28 connectors 0.50 : 0.50 -
PreWEC-D Elastic definition - - -
PreWEC-E - - - 200 mm
* MPT : Mcon is the ratio of the base moment provided by the PT to the base moment provided by the
connectors at 2% lateral drift
‘-’ indicates no change from the PreWEC-t specimen
6.5.1 Number of Connectors

The first three analyses were aimed at changing the ratio between the moment provided by the
PT and the moment provided by the connectors. The objective of this exercise was to modify
the PreWEC design, but not the target strength. This meant that as the connector moment
contribution was increased the PT moment contribution was reduced, and vice versa.
Monotonic lateral load analyses were conducted using the ABAQUS FEM to confirm the
lateral strength of the modified designs. The monotonic lateral force-displacement envelope of
the first three modified designs was comparable to the PreWEC test FEM (PreWEC-t), as
shown in Figure 6.23.
6.5.1.1 More connectors and less PT

PreWEC-A used an increased number of connectors, from 20 to 28 (14 on each side/end). To
maintain the original lateral strength, the PT area in PreWEC-A was reduced from 1680 mm2
to 1350 mm2. As expected, the resulting cyclic lateral force-displacement response, shown in
Figure 6.24a, indicated increased hysteretic energy dissipation, when compared to PreWEC-t.
Interestingly, even with the fatter hysteresis loops, the PreWEC-A response did not result in a
significant increase in the cyclic residual displacements. This was because the compression
force on the wall toe was less for PreWEC-A, which resulted in less damage to the wall when
compared to PreWEC-t. The PreWEC system consist of connectors at each end of the wall,
- 192 -
and as the wall rocks the connectors on one end of the wall are displaced in the positive vertical
direction, while the connectors at the other end of the wall are displaced in the negative vertical
direction. For large lateral drifts these vertical displacements cause the connectors at both ends
of the wall to yield, and in doing so the vertical connector forces imparted at each end the wall
are approximately equal and opposite. Therefore, the axial load on the wall is primarily a
result of the PT and gravity loads. As the number of connectors was increased and the PT was
reduced, the axial load on the wall, and resulting damage to the wall toe, was reduced. This
reduction in axial load is confirmed in Figure 6.25, which shows the calculated neutral axis
depth and critical concrete strain at the extreme compression fibre of the wall for PreWEC-t
and PreWEC-A. When compared to PreWEC-t, both the neutral axis depth and the concrete
strains were reduced for PreWEC-A, which indicated a reduced axial compression force in the
wall toe. This reduced axial compression force resulted in lower in-elastic compressive strains
in the wall toe of PreWEC-A, which reduced the component of the residual displacement that
was attributed to the wall hysteresis behaviour.
Lateral Drift (%)

0 0.5 1 1.5 2 2.5 3
600
500
400
300
200
PreWEC-t
PreWEC-A
100
PreWEC-B
PreWEC-C
0
0 50 100 150 200
Figure 6.23 – Monotonic lateral force-displacement response of modified PreWEC FEM
Because of the increased energy dissipation and reduced damage to the wall toe, PreWEC-A
would exhibit improved seismic performance when compared to PreWEC-t. Also, because of
the increased damage to the wall toe when a higher level of PT is used, increasing the number
of connectors will not necessarily result in increased cyclic residual displacements, as is
traditionally assumed. A detailed investigation of residual displacements and self-centering
behaviour is discussed in Chapter 7.
- 193 -

-3 -2 -1 0 1 2 3 -3 -2 -1 0 1 2 3
600 600

400 400
200 200
0 0
-200 -200
-400 -400
PreWEC-t PreWEC-t
PreWEC-A PreWEC-B
-600 -600
-200 -100 0 100 200 -200 -100 0 100 200
(a) PreWEC-A (b) PreWEC-B

Figure 6.24 – Lateral force-displacement response calculated for PreWEC-A & B

-3 -2 -1 0 1 2 3 -3 -2 -1 0 1 2 3
 0.035
PreWEC-t PreWEC-t
//
1400 PreWEC-E 0.03 PreWEC-A

PreWEC-B
1200
Compressive Strain
0.025
1000
NA (mm)
0.02
800
0.015
600
0.01
400
200 0.005
0 0
-200 -100 0 100 200 -200 -100 0 100 200
(a) Neutral axis (NA) depth (b) Critical concrete strain

Figure 6.25 – Local responses backbones calculated for PreWEC-A & B
6.5.1.2 Fewer connectors and more PT

PreWEC-B used a decreased number of connectors, from 20 to 16 (8 on each side/end). To
maintain the same lateral strength as PreWEC-t, the PT area in PreWEC-B was increased from
1680 mm2 to 1900 mm2. As shown in Figure 6.24b, the resulting cyclic lateral force-
displacement response indicated reduced hysteretic energy dissipation, but only during low
- 194 -
lateral drift cycles. During the larger drift cycles significant damage occurred to the wall toe as
a result of the increased axial load from the PT. This behaviour was opposite to that observed
for PreWEC-A, and the increased wall damage resulted in poor performance. The increased
axial compression force in the wall toe is highlighted in Figure 6.25, which shows an increase
in the neutral axis depth and concrete compressive strains in the wall toe for PreWEC-B, when
compared to PreWEC-t. Again, contrary to what would generally be expected, reducing the
number of connectors did not reduce the residual displacement because of the increased wall
hysteresis due to concrete crushing.
6.5.1.3 More connectors and lower PT stress

The initial tendon prestress level of 1200 MPa that was used for PreWEC-t and PreWEC-A
caused the tendon stress to increase beyond 1500 MPa during cyclic loading, and some
yielding of the tendons occurred. To improve on the performance of PreWEC-A, instead of
reducing the area of PT, for PreWEC-C the initial tendon stress was reduced to 890 MPa. As
shown in Figure 6.26a, investigation of the tendon stress response indicated that for PreWEC-C
the wall PT remained elastic, whereas some yielding of the PT and prestress loss was observed
for PreWEC-A. However, as observed in Figure 6.26b, the resulting lateral force-displacement
response for PreWEC-C showed no significant change when compared to the response of
PreWEC-A. This observation indicated that there was no significant advantage derived from
ensuring that the tendons remain completely elastic, when compared to allowing minor
yielding during large lateral drift cycles. Therefore, an economical design would minimise the
number of prestressing tendons that were used and allow limited yielding to occur in excess of
the design level lateral drift (2% for the PreWEC wall).
Currently, initial prestress limits are defined based on the idealised yield strength of the tendon.
For example, Appendix B of NZS 3101:2006 [6-16] recommends that the initial prestress be
designed to ensure that the tendon stress at the design level drift does not exceed 0.9fpy. Limits
defined from the idealised yield stress are difficult to relate to actual prestress loss, because in
reality the tendon is fully elastic up to a stress that is lower than the idealised yield stress. This
realistic elastic state can be observed in Figure 6.11b, where although the idealised yield stress
of the tendon was 1650 MPa, testing indicated that the tendon was fully elastic up until
1350 MPa, with significant yielding occurring after 1500 MPa. This means that some prestress
loss will occur even when the tendon stress does not exceed the idealised yield strength, as was
observed for PreWEC-t and PreWEC-A.
- 195 -

-3 -2 -1 0 1 2 3 -3 -2 -1 0 1 2 3
600
1600 PreWEC-t
PreWEC-A

1500 400
PreWEC-C
1400
200
Stress (MPa)
1300
0
1200
1100 -200
1000 PreWEC-t
-400
PreWEC-A
900 PreWEC-C
-600
-200 -100 0 100 200 -200 -100 0 100 200
(a) Wall tendon stress (b) Lateral force-displacement

Figure 6.26 – Lateral force-displacement and tendon stress response calculated for
PreWEC-C
6.5.2 Elastic Tendons

To investigate the influence of the tendon non-linear behaviour further, PreWEC-D was
analysed with an elastic tendon that was defined using a modulus of elasticity of 192 GPa and
with no plasticity defined. No significant difference was observed between the lateral force-
displacement responses of PreWEC-D and that of PreWEC-t, as shown in Figure 6.27a.
However, investigation of the tendon stress response provided more insight into the PreWEC
behaviour. As shown in Figure 6.27b, the tendon stress from PreWEC-D showed lower
prestress loss when compared to PreWEC-t. Because the tendon remained elastic, the observed
prestress loss for PreWEC-D was attributed to wall shortening due to non-linear behaviour at
the wall base, and it was not until the large drift cycles that yielding of the PT tendon caused
higher prestress loss for PreWEC-t.
- 196 -

-3 -2 -1 0 1 2 3 -3 -2 -1 0 1 2 3
600 1600
PreWEC-t
PreWEC-D
400
1500
200
Stress (MPa)
1400
1300
-200
1200
-400
PreWEC-t
PreWEC-D
-600 1100
-200 -100 0 100 200 -200 -100 0 100 200
(a) Lateral force-displacement (b) Wall tendon stress

Figure 6.27 – Lateral force-displacement and tendon stress response calculated for
PreWEC-D
6.5.3 Wall Thickness

Another method to reduce the axial compression force in the toe of a PT wall is to modify the
properties of the wall. Increasing the concrete strength was considered to not be a practical
option because the concrete strength of 63 MPa was already higher than standard precast
components. Instead, for PreWEC-E the wall thickness was increased from 152.4 mm to 200
mm and the resulting lateral force-displacement response is shown in Figure 6.28. The
increased thickness resulted in a small increase in strength due to the increased moment lever
arm. However, a more noticeable difference between the responses of PreWEC-E and
PreWEC-t was the change in the unloading path, which was a result of significantly less
damage to the wall toe. The PreWEC-E response shifted closer towards the ideal flag-shape
hysteresis behaviour and the residual displacements were almost zero, even during large drift
cycles. The behaviour of the PreWEC-E wall toe was confirmed by the neutral axis depth and
concrete strain at the extreme compression fibre, which are shown in Figure 6.29. A
significant reduction in the neutral axis depth was observed due to the increased contact width
and this resulted in a large reduction in concrete compressive strain. The lower strain demand
prevented any significant crushing in the wall toe, which ultimately led to a much more
desirable seismic response. Increasing the wall thickness would increase the cost of the
system, but the significant reduction in wall damage would probably outweigh the initial cost.
- 197 -
Lateral Drift (%)

-3 -2 -1 0 1 2 3
600

400
200
-200
-400
PreWEC-t
PreWEC-E
-600
-200 -100 0 100 200
Figure 6.28 – Lateral force-displacement response calculated for PreWEC-E

-3 -2 -1 0 1 2 3 -3 -2 -1 0 1 2 3
1600 0.03
PreWEC-t PreWEC-t
1400
PreWEC-E 0.025 PreWEC-E
1200
Compressive Strain
0.02
1000
NA (mm)
800 0.015
600
0.01
400
0.005
200
0 0
-200 -100 0 100 200 -200 -100 0 100 200

Figure 6.29 – Localised response backbones calculated for PreWEC-E
6.6 INFLUENCE OF CONNECTORS

In the previous section the number of connectors in the PreWEC system was both increased
and decreased, while at the same time the PT force was decreased or increased to maintain the
same lateral strength as the PreWEC test specimen. In this section, the influence of the number
- 198 -
6.6 Influence of Connectors
of connectors is investigated further. The original PreWEC test specimen contained a total of
20 O-Connectors (10 on each side/end), and two further FEM analyses were conducted with
the total number of O-Connectors both increased to 28 and decreased to 16 O-Connectors. For
these two analyses, the wall PT and the rest of the FEM were unchanged from the test
specimen in order to to isolate the influence of the connectors. As in the previous section, the
FEM of the PreWEC test specimen that included the cyclic damage indices and one cycle at
each displacement level was used as the benchmark.
The lateral force-displacement responses of the FEMs with 28 and 16 connectors are compared
with that of the 20 connector test specimen FEM in Figure 6.30. As expected, when more
connectors were added to the system there was a noticeable increase in lateral strength and
hysteretic energy dissipation. The opposite was observed when the number of connectors was
reduced, with decreased lateral strength and hysteresis area.

-3 -2 -1 0 1 2 3 -3 -2 -1 0 1 2 3
600 600
400 400
200 200
0 0
-200 -200
-400 -400
20 Con (test) 20 Con (test)
28 Con 16 Con
-600 -600
-200 -100 0 100 200 -200 -100 0 100 200
(a) More connectors (b) Less connectors

Figure 6.30 – Calculated lateral force-displacement response for PreWEC FEM with 28
and 16 connectors
As explained in the previous section, even though the displacement of the connectors at each
end of the wall was not equal, once the connectors have yielded the vertical connector forces at
each end of the wall are approximately equal and opposite. As a result, the number of
connectors had no significant influence on the cyclic behaviour of the concrete wall. Because
the magnitude of PT was unchanged throughout the analyses, the axial load on the wall was not
- 199 -
significantly affected by the number of connectors. The consistency in the wall behaviour is
highlighted in Figure 6.31, which shows a comparison of the calculated neutral axis depth and
critical concrete compressive strain in the wall toe for each analysis. Both the neutral axis
depth and the concrete compressive strain showed no significant change between the different
analyses, which confirmed that the wall behaviour was independent of the number of
connectors. The constant wall behaviour was also visually observed from the FEM analyses,
which exhibited similar levels of crushing at the wall toe.

-3 -2 -1 0 1 2 3 -3 -2 -1 0 1 2 3
1600 0.03
20 Connector (test) 20 Connector (test)
1400
16 Connector 0.025 16 Connector
1200 28 Connector 28 Connector

Compressive Strain
0.02
1000
NA (mm)
800 0.015
600
0.01
400
0.005
200
0 0
-200 -100 0 100 200 -200 -100 0 100 200

Figure 6.31 – Calculated local response backbones for PreWEC FEM with 28 and 16
connectors
Using Eqns. 6.11 and 6.12, the equivalent viscous damping and relative energy dissipation
ratio were calculated for each of the FEM analyses with a different number of connectors, as
plotted in Figure 6.32. As was observed from the force-displacement hysteresis loops, the
equivalent viscous damping increased as the number of connectors increased. Because the wall
behaviour remained unchanged between the three analyses, the increase in the equivalent
viscous damping was directly associated with the number of connectors that were used.
Additionally, as shown previously in Figure 6.16a, the FEM with only one cycle at each
displacement level was found to provide accurate calculation of the equivalent viscous
damping measured during the experimental test. For these reasons, the FEM analyses
described in this section could be used to develop an expression to calculate the equivalent
viscous damping for the PreWEC system.
- 200 -
6.6 Influence of Connectors

0.35

20
0.3
15 0.25
0.2
0.15
10
0.1
0.05
5 20 Connector (test)
0
16 Connector
28 Connector -0.05
0
0 0.5 1 1.5 2 2.5 3
Lateral Drift Cycle (%)
Figure 6.32 – Equivalent viscous damping for PreWEC FEMs with varied number of
connectors
The equivalent viscous damping is plotted against the number of connectors in Figure 6.33 for
each level of lateral drift from 1 to 3%. It was observed that at each drift level the equivalent
viscous damping was approximately proportional to the number of connectors, nc, with an
average slope of 0.4nc. Trend lines with a slope of 0.4nc are also plotted in Figure 6.33 and
show good correlation with the FEM calculated equivalent viscous damping. The intercept of
these trend lines, when the number of connectors is zero, is equal to the equivalent viscous
damping of the wall itself. As reported earlier, large in-elastic stains in the wall toe caused the
wall behaviour to deviate from the ideal bi-linear elastic response and added significant
hysteresis area to the PreWEC response. A parabolic curve was used to approximate the wall
equivalent viscous damping, resulting in the quadratic equation shown in Eq. 6.13. By adding
the wall and connector contributions, Eq. 6.14 was developed to calculate the equivalent
viscous damping for the PreWEC wall, depending on the number of O-Connectors and lateral
drift level. It should be noted that Eq. 6.13 and Eq. 6.14 are only valid for this particular
PreWEC design, and that if the wall properties or the PT configuration are modified, then the
equivalent viscous damping associated with the wall response will need to be recalculated.
 eq, Wall  1.03d 2  7.7d  3.3 (6.13)
 eq , Pr eWEC  0.4nc   1.03d 2  7.7d  3.3 (6.14)
where d is the lateral drift % and nc is the number of O-Connectors.
- 201 -
25

1% drift
1.5% drift
20 2% drift
2.5% drift
3% drift
15 0.4nc trend line
10
0 5 10 15 20 25 30
Number of Connectors, nc
Figure 6.33 – Equivalent viscous damping for each lateral drift level
The accuracy of Eq. 6.14 is demonstrated in Figure 6.34 which shown a plot of the calculated
equivalent viscous damping across a lateral drift range of 1-3% for PreWEC systems with 16,
20, and 28 O-Connectors. The equivalent viscous damping calculated using Eq. 6.14 showed
good correlation with the equivalent viscous damping calculated from the FEM hysteresis
responses. Additionally, the equivalent viscous damping attributed to the wall behaviour is
also plotted in Figure 6.34, using the definition provided in Eq. 6.13. It was observed that the
wall contributed significantly to the equivalent viscous damping of the PreWEC system.
25
FEM: 16 con 0.35

FEM: 20 con
20 0.3
FEM: 28 con
Eq. 6.14: 16 0.25
Eq. 6.14: 20
15 0.2
Eq. 6.14: 28
Wall only 0.15
10 0.1
0.05
5
0
-0.05
0
-0.5 0 0.5 1 1.5 2 2.5 3
Lateral Drift Cycle (%)
Figure 6.34 – Comparison of equivalent viscous damping calculated using Eq. 6.14 and
the PreWEC FEM
- 202 -
6.7 PreWEC Core Wall System
6.7 PREWEC CORE WALL SYSTEM

The PreWEC test specimen was designed as an in-plane structural wall. However, the
PreWEC concept can easily be extended for application in core walls, making the precast
construction option more appealing. Core walls typically have a high stiffness and are often
used as the primary lateral load resisting system in buildings. In high rise structures, the use of
a stiff reinforced concrete core wall is often necessary in order to control lateral deflections [6-
5]. Bi-directional seismic loading can lead to considerable demand on a reinforced concrete
core wall, causing extensive damage, especially to the corners. The use of a seismic resilient
self-centering wall system for the core wall structure would have considerable benefits, both
from the high initial stiffness and reduction in structural damage.
6.7.1 PreWEC Core Wall Concept

A square core wall design was developed that included four precast concrete wall elements and
four columns located at the corners, as shown in Figure 6.35. The core wall design was based
off the dimensions of the single PreWEC wall specimen and a summary of the details is
included in Table 6.7. The overall dimensions of the PreWEC core wall were 2388 mm ×
2388 mm, with a wall length of 1828.8 mm. The columns were modified to be an L-shape in
order to allow the connectors to be placed on each side of the wall and provide the connector
enough clearance for their expected displaced shape. The wall PT remained unchanged, but
because the columns were required to act in both directions the cross-sectional area of the
column PT was doubled to 6 tendons, placed in two separate ducts. This increased quantity of
column PT was designed to ensure that the column did not lift off from the foundation due to
the vertical connector forces. Additionally, the material properties were assumed to be the
same as for the PreWEC test specimen.
Table 6.7 – Details of PreWEC core wall
Element Description
Wall 1828.8 mm long × 152.4 mm wide × 6096 mm high
Column L-shaped, 203.2 mm long side × 152.4 mm short side × 5621 mm high
Connectors 10 × O-Connectors at each joint (5 on each side)
Wall PT 12 × 15.2 mm diameter strand (Aps = 1680 mm2 and lp = 7900 mm)
Column PT 6 × 15.2 mm diameter strand (Aps = 840 mm2 and lp = 6650 mm)
Wall reinforcement Same as PreWEC test specimen
- 203 -
254.0
2387.6 1828.8
254.0
152.4
254.0 1828.8 254.0

2387.6
Figure 6.35 – Cross-section of the PreWEC core wall design
6.7.2 FEM Analysis

A FEM was constructed to analyse the bi-directional loading behaviour of the PreWEC core
wall system. The FEM used the same modelling techniques that were described previously for
the single PreWEC wall specimen. Lateral loading was applied simultaneously to points at the
top of all four walls. The core wall FEM was subjected to lateral loading in two different
directions, being (i) the in-plane direction with the lateral displacement parallel to two of the
walls and (ii) diagonal displacement at an angle of 45 degrees to the in-plane direction of the
walls.
6.7.2.1 In-plane direction

The response of the PreWEC core wall FEM for the in-plane load direction is shown in Figure
6.36. The response of the in-plane loaded walls appeared to be similar to that of the single
PreWEC wall, with uplift at the wall base and plastic deformation of the O-Connectors. The
- 204 -
out-of-plane loaded walls at the ends of the core structure also rocked at the wall base, but did
not lift off completely from the foundation. Additionally, damage to the wall toes was of a
similar level to that observed for the single wall PreWEC test specimen. The lateral force-
displacement response of the core wall FEM for the in-plane load direction is shown in Figure
6.37, alongside the response of the single PreWEC wall FEM multiplied by a factor of two.
The core wall exhibited a lateral strength that was 20% higher than the strength of two single
PreWEC walls. This over-strength was attributed to the out-of-plane lateral strength of the end
walls. Furthermore, because the length of the L-shaped columns was larger than the wall
thickness, the uplift at the base of the column was greater than the uplift at the base of the out-
of-plane loaded wall. This displacement incompatibility caused a relative vertical
displacement of up to 2.3 mm across the connectors attached to the out-of-plane loaded wall.
As shown in the close up of the connectors in Figure 6.36, the out-of-plane loaded connectors
were subjected to significant deformation, and the 2.3 mm displacement resulted in a connector
force of 16.5 kN, which was approximately 40% of their maximum strength. This vertical
force in the out-of-plane loaded connectors contributed to the 20% increase in lateral strength.
(a) Side view (b) Angled view

Figure 6.36 – In-plane response of PreWEC core wall FEM at 3% lateral drift
- 205 -
Lateral Drift (%)

0 0.5 1 1.5 2 2.5 3
1600

1400
1200
1000
800
600
400 Core: In-plane
200 Core: 45 deg

Single PreWEC x2
0
0 50 100 150 200
Figure 6.37 – Lateral force-displacement response for PreWEC core wall FEM
6.7.2.2 Bidirectional loading

As shown in Figure 6.38, the PreWEC core wall also performed well when it was subjected to
diagonal loading at an angle of 45 degrees, with each wall subjected to in-plane and out-of-
plane loading. Each of the four concrete walls uplifted, with the toe of each wall remaining in
contact with the foundation. Additionally, all of the connectors were subjected to large in-
elastic deformation. As seen in Figure 6.38d, the uplift of the walls caused the connectors on
both sides of the trailing column to undergo the maximum positive displacement. However,
the increased column PT was sufficient to prevent the trailing column from lifting off from the
foundation. The lateral force-displacement response of the core wall FEM in the diagonal
direction is also plotted in Figure 6.37. The core wall exhibited a 30% strength increase in the
diagonal direction when compared to the response in the in-plane loading direction. This
strength gain was attributed to all four walls undergoing uplift and to the increased demand on
the connectors.
6.7.3 Summary
The FEM successfully demonstrated the viability of the PreWEC core wall concept, which
displayed excellent strength and ductility with only minimal structural damage resulting.
However, it was observed that the behaviour of the core wall became increasingly more
complex to understand and analyse when compared to a single PreWEC wall. Even in the in-
plane loading direction the relative displacement at every wall-to-column joint would need to
- 206 -
be considered in order to accurately calculate the connector behaviour and lateral strength of
the system. When considering the bi-directional response in a diagonal loading direction, the
interaction between the components became more complex, and in reality the behaviour of the
core wall system is highly dependent on the influence of the floor diaphragms, as stiff cast-in-
place floor diaphragms would not allow the walls to rock and uplift independently of each
other. The issue of wall-to-floor interaction is investigated in more detail in Chapter 8.
(b) Side view at base
(a) Side view
(c) End view (trailing column) (d) Connectors on trailing column

Figure 6.38 – Bidirectional response of PreWEC core wall FEM at 3% lateral drift
- 207 -
6.8 CONCLUSIONS
6.8.1 FEM Development
This chapter presented the development of a detailed finite element model (FEM) of the
PreWEC system that was an extension of the previously developed 3D FEM for single PT
walls and the 3D O-Connector FEM. Compared to the unconfined concrete PT walls that were
previously modelled, the PreWEC system required accurate representation of the confined
concrete in the wall toe.
Extensive trials were conducted in order to determine suitable and efficient techniques for
modelling confined concrete, by comparing FEM results with past experimental data found in
the literature. Instead of explicitly modelling the transverse reinforcement, a confined stress-
strain response was directly assigned to the concrete elements. Additionally, for concrete wall
sections the cover concrete constitutes a significant portion of the cross-sectional area and the
crushing and spalling behaviour of this cover was accurately accounted for.
Due to the complexity of the wall-to-foundation contact surface, the explicit solver available
within ABAQUS was relied upon to conduct the cyclic FEM analysis. Overall, the FEM
showed good correlation with the PreWEC experimental test for both the global and local
responses. The inclusion of cyclic damage parameters, to accurately model the concrete
unloading and reloading paths, was found to be critical to the calculated cyclic response and
degradation in the system’s strength and stiffness. Additionally, although increased cyclic
damage was observed when the FEM was subjected to three cycles at each displacement level,
the FEM analysis with a single cycle at each displacement level showed sufficient accuracy
when compared to the experimental response.
The PreWEC FEM was unable to accurately capture the experimental unloading and reloading
responses as the gap closed at the wall-to-foundation interface. The FEM displayed a pinching
shape in the lateral force-displacement hysteresis loops, which led to underestimation of the
residual drift that was observed during the test. Several trials that were conducted with
modified contact properties were unable to improve calculation of the residual displacements.
- 208 -
6.8 Conclusions
6.8.2 Investigation of PreWEC Behaviour

The FEM was used to investigate several modifications to the PreWEC design. From the
results of the analyses it was found that when the moment contribution from the connectors
was increased, by adding more connectors and reducing the magnitude of the PT force, the
PreWEC response was improved with increased energy dissipation and reduced wall damage.
Because of the PreWEC configuration, the vertical connector forces at each end of the wall are
approximately equal and opposite, which means that the wall axial load is primarily attributed
to the PT force and imposed gravity loads, and is independent of the number of connectors.
For this reason, increasing the moment contribution from the connectors reduced the demand
on the wall toe due to the reduced PT load, which reduced the residual drift caused by damage
to the wall toe. This finding means that contrary to what is traditionally assumed, increasing
the connector moment contribution may not always increase the residual displacement (and
vice versa).
Although the PT tendon stresses never exceeded the ideal yield stress, both the experimentally
measured and FEM calculated tendon stresses in the wall PT indicated some prestress loss
during cyclic loading. This prestress loss was attributed to the inelastic behaviour of the
tendon as well as wall shortening that resulted from in-elastic strains at the wall base. During
design, the idealised yield stress of the PT tendons should be treated with caution. In reality,
the tendon was fully elastic up until 80-90% of the idealised yield stress that was advertised by
the manufacturer. If the tendon stress exceeds the fully elastic state, some prestress loss due to
non-linear behaviour should be expected.
Increasing the wall thickness significantly reduced the damage observed in the wall toe, with
the calculated response being closer to the ideal flag-shape hysteresis rule. However, during
the design process the cost of increasing the wall thickness would need to be balanced against
the resulting performance benefits.
To isolate the influence of the connector, further analyses were conducted with a consistent
wall PT force and an increased or decreased number of connectors. The results from these
analyses were used to establish an equation to predict the energy dissipation in the PreWEC
system, which was separated into the contribution from wall and connector hysteresis. The
hysteresis behaviour of the wall cannot be assumed as bi-linear elastic and calculation of the
- 209 -
damage to the wall toe is essential for understanding the system’s energy dissipation and
residual drift.
Lastly, the viability of a PreWEC core wall concept was successfully investigated. The core
wall displayed excellent behaviour, but the interaction between the wall and column elements
would significantly increase the complexity of simplified analysis equations. The development
of simplified design procedures is essential to the viability of the core wall system, and remains
a topic for future investigation.
6.9 REFERENCES
[6-1] Mander, J. B., Priestley, M. J. N., and Park, R. (1988) Theoretical stress-strain model
for confined concrete. Journal of Structural Engineering, 114(8), 1804-1826.
[6-2] Chang, G. A. and Mander, J. B. (1994) Seismic energy based fatigue damage analysis
of bridge columns: Part 1 - Evaluation of seismic capacity. NCEER technical report no.
NCEER-94-0006, State University of New York, Buffalo, NY.
[6-4] Mander, J. B., Priestley, M. J. N., and Park, R. (1988) Observed stress-strain behavior
of confined concrete. Journal of Structural Engineering, 114(8), 1827-1849.
[6-6] Han, L. H., Wang, W. D., and Zhao, X. L. (2008) Behaviour of steel beam to concrete-
filled SHS column frames: Finite element model and verifications. Engineering
Structures, 30(6), 1647-1658.
[6-7] Hu, H. T., Huang, C. S., and Chen, Z. L. (2005) Finite element analysis of CFT
columns subjected to an axial compressive force and bending moment in combination.
Journal of Constructional Steel Research, 61(12), 1692-1712.
[6-8] Hu, H. T., Huang, C. S., Wu, M. H., and Wu, Y. M. (2003) Nonlinear analysis of
axially loaded concrete-filled tube columns with confinement effect. Journal of
Structural Engineering, 129(10), 1322-1329.
[6-9] Mier, J. G. M. (1997) Fracture processes of concrete: assesment of material

parameters for fracture models. CRC Press, Boca Raton, FL.
[6-10] Bažant, Z. P. and Oh, B. H. (1983) Crack band theory for fracture of concrete.
Matériaux et Constructions, 16(3), 155-177.
- 210 -
6.9 References
[6-11] Foster, S. J., Liu, J., and Sheikh, S. A. (1998) Cover spalling in HSC columns loaded in
concentric compression. Journal of Structural Engineering, 124(12), 1431.
[6-12] Liu, J. and Foster, S. J. (2000) A three-dimensional finite element model for confined
concrete structures. Computers & Structures, 77(5), 441-451.
New Zealand.
- 211 -
- 212 -
Chapter 7
Analysis of Residual Drifts
The main objective when designing seismic resilient building systems is to safely dissipate the
energy imparted to a structure during an earthquake, with minimal structural damage resulting.
Given the difficulty in straightening a building which is left with a significant residual drift
following an earthquake, self-centering behaviour is a critical aspect of seismic resilient design.
As described in previous chapters, the behaviour of self-centering structures is typically
characterised by a flag-shaped hysteresis response. However, when a flag-shaped cyclic
hysteresis response is used to define the self-centering behaviour of a PT concrete member,
two major factors are ignored. First, in reality the response of a concrete member with
unbonded PT does not follow a perfect bilinear-elastic hysteresis rule, and so an imperfect flag-
shape is often observed. Second, investigation of the cyclic hysteresis response does not
account for the dynamic response, and the resulting residual drift of the structure following an
earthquake.
This chapter demonstrates the inadequacies of the current methods that are used to ensure that
a self-centering response is achieved as part of the design of a seismic resilient structure.
Furthermore, an investigation into how self-centering should be accounted for during the
design process is included in this chapter. Extensive time-history analyses were performed to
- 213 -
Chapter 7 Analysis of Residual Drifts
investigate the self-centering behaviour of the PreWEC system during an earthquake, with the
results used to develop appropriate design recommendations.
7.1 BACKGROUND
7.1.1 Static Self-centering
As described previously, the response of a self-centering system is typically characterised by a
flag-shaped hysteresis loop. For a member with unbonded PT, the hysteresis response is
typically assumed to be bi-linear elastic, with zero residual displacement. Although shown to
be inaccurate in section 7.2, the combination of the bi-linear elastic response and the elasto-
plastic response of the energy dissipating components is often used to define the idealised flag-
shape response, as shown in Figure 7.1. The amount of PT and energy dissipating components
control the form of the flag-shape hysteresis, because as energy dissipating components are
added the hysteresis area of the flag gets larger.

F F F
∆ ∆ ∆
Bi-linear elastic + Elasto-plastic = Flag-shape
Figure 7.1 – Idealised flag-shape hysteresis response typically assumed for self-centering
systems
Based on the ideal flag-shape shown in Figure 7.1, the common perception during the PRESSS
program was that for hybrid systems, self-centering would be achieved if the moment provided
by the energy dissipating mild steel was less than the moment provided by the PT [7-1]. This
concept is referred to as “static self-centering”, were the hysteresis response returns to zero
displacement when the lateral load is slowly released to zero. Several recent studies into the
behaviour of concrete [7-2], steel [7-3] and timber [7-4] self-centering systems have continued
to rely on the concept of static self-centering, using an ideal flag-shaped hysteresis behaviour.
The bi-linear elastic hysteresis assumption may be valid for steel and timber members, but it is
seldom achieved for PT concrete members. Cyclic tests on both unbonded PT concrete frames
- 214 -
7.1 Background
[7-5] and walls [7-6] have indicated that the response of the PT concrete member included
significant stiffness degradation and cyclic hysteresis area. This imperfect bi-linear elastic
hysteresis response is caused by inelastic strains in the compression toe of the concrete
member, and significantly increases the residual drift observed in the cyclic hysteresis
response. This deviation from the idealised behaviour was further highlighted during tests of
hybrid and jointed self-centering frame and wall systems [7-1, 7-7, 7-8], as well as during the
PreWEC test and the analyses discussed in Chapters 5 and 6.
7.1.2 Current Design Procedures

Despite the experimentally observed behaviour of PT concrete members, current design
procedures that are used to ensure that self-centering is achieved are predominantly based upon
the ideal flag-shaped hysteresis response. As a result of the PRESSS research program, ACI
design guidelines for self-centering concrete frames included the expression shown in Eq. 7.1,
which states that the moment contribution from energy-dissipating reinforcement, Ms, must
equate to less than 50% of the probable flexural strength of the member, Mpr. In accordance
with capacity design procedures, the New Zealand concrete design standard adopted a similar
expression that also included the overstrength factor of the energy dissipating components.
The equation included in Appendix B of NZS 3101:2006 [7-9] is shown in Eq. 7.2, where the
moment contribution ratio, λ, must be greater than the overstrength factor of the energy
dissipating components, αo. In New Zealand, the overstrength factor for mild steel is typically
1.15 or greater [7-9], which implies that the moment contribution from the energy dissipating
components should be less than 46% of the total flexural strength. The design guidelines for
self-centering concrete systems published by the New Zealand Concrete Society [7-10] also
includes Eq. 7.2, but it is further recommended that the moment provided by the energy
dissipating components should not exceed 40% of the total moment resistance.
Ms
 0.5 (7.1)
M pr
M pt  M N
  o (7.2)
Ms
where Mpt is the moment provided by the unbonded PT and MN is the moment provided by
additional axial load (including self-weight).
- 215 -
Both Eqns. 7.1 and 7.2 are intended to be calculated at the design lateral drift, or ultimate
moment resistance, which is not sufficient to ensure static self-centering of the cyclic hysteresis
loops, as demonstrated later in section 7.2. Additionally, Eqns. 7.1 and 7.2 do not account for
the realistic cyclic hysteresis behaviour, or the dynamic response, when considering the self-
centering capability of a structure during an earthquake.
The ACI Innovation Task Group 5 (ITG-5), that developed code requirements for the
development and design of unbonded post-tensioned concrete walls did not include a specific
procedure to ensure that self-centering was achieved. However, ITG-5.1 [7-11] states that
studies by Kurama [7-12] indicated that the self-centering capability following a major
earthquake may be lost if more than 40% of the flexural capacity was provided by energy
dissipating mild steel. Additionally, ITG-5.2 [7-13], which provides design procedures for
self-centering wall systems, recommends that for hybrid walls, Eq. 7.3 be used to calculate the
minimum required prestress force to ensure that the base crack closes.
A ps f se  0.9 N  As f u (7.3)
where Aps is the area of post-tensioning tendons, fse is the effective stress in the post-tensioning
tendons, N is the self-weight of the wall plus any dead load, As is the area of energy dissipating
reinforcement, and fu is the ultimate tensile strength of the energy dissipating reinforcement.
Eq. 7.3 is a vertical force balance equation and therefore it does not necessarily ensure that
self-centering will be achieved, because the moments attributed to the PT and energy
dissipating elements are not specifically considered. Additionally, Eq. 7.3 is unique to the
hybrid wall system, where to ensure that the gap at the wall-to-foundation interface can close
when the wall is re-centered, the post-tensioning force (and dead load) must overcome the
maximum force that can develop in the energy dissipating components. Eq. 7.3 is not relevant
to the PreWEC system because the vertical connector forces at each end of the wall are
approximately equal and opposite, and do not prevent gap closure at the wall-to-foundation
interface when the wall is re-centered.
7.1.3 Dynamic Self-centering

A residual displacement observed from a cyclic hysteresis loop does not necessarily guarantee
poor building performance. Instead, the residual drift of a structure following excitation during
- 216 -
7.1 Background
an earthquake is critical when considering whether a self-centering response was achieved.

When subjected to an earthquake ground motion, a structure will typically display smaller load
reversals following the peak displacement as the earthquake intensity diminishes, as shown in
Figure 7.2. This post-peak behaviour is referred to as the “shake-down” phenomenon [7-14].
As a result, the residual drift at the end of excitation, dr, will typically be less than the
maximum possible residual drift that would be predicted from the hysteresis loops, dr, max.
F
dmax
dr
dr, max d
shake down
Figure 7.2 – Dynamic shake-down behaviour expected for seismic load resisting systems
After recognising that the residual drift of a structure was a function of the hysteresis behaviour
and the earthquake ground motion, MacRae and Kawashima [7-14] conducted a series of
dynamic analyses to investigate the behaviour of single-degree-of-freedom oscillators.
MacRae and Kawashima found that even for oscillators with elasto-plastic hysteresis rules,
significant reduction in the residual displacement was observed due to the shake-down effect.
The residual displacement at the end of the ground motion was normalised by the maximum
possible residual displacement to define the “residual displacement ratio”, or drr. Furthermore,
MacRae and Kawashima found that the magnitude of the residual displacement ratio was
primarily a function of the post-yield stiffness of the bilinear elasto-plastic hysteresis rule. A
large positive post-yield stiffness led to small residual displacements, while a negative post-
yield stiffness led to large residual displacements.
Residual drifts were also the focus of a performance based assessment of frame structures that
was conducted by Christopoulos et al. [7-15]. Christopoulos et al. carried out a series of
- 217 -
dynamic single-degree-of-freedom analyses to compare the seismic response of systems having

elasto-plastic, Takeda, and flag-shape hysteresis rules. Although noting that dynamic shake-
down reduced the residual drifts of the elasto-plastic and Tekada systems, Christopoulos et al.
argued that the full benefit of self-centering systems could not be assessed without considering
the residual drifts. However, because an ideal flag-shape hysteresis rule was used during the
analyses, a realistic assessment of the residual drifts observed in self-centering systems was not
established.
7.2 STATIC ANALYSIS OF FLAG-SHAPE RESPONSE

Although self-centering members exhibit a response that appears flag-shaped, the idealised
flag-shape representation that was shown in Figure 7.1 does not typically exist for two reasons.
First, the responses of the PT member and energy dissipating components cannot be simply
added together because energy dissipating elements are only engaged after decompression
occurs at the tension toe of the wall and uplift occurs at the wall-to-foundation interface.
Second, an imperfect flag-shape is achieved when realistic hysteresis responses are used for
both the PT member and the energy dissipating elements. This section investigates the flag-
shaped hysteresis response of a more realistic self-centering wall model. The inaccuracy of the
current procedures for ensuring self-centering is demonstrated, even when the residual drift
caused by wall hysteresis and the dynamic shake-down effects were ignored.
7.2.1 Lumped Plasticity Model

A realistic representation of the response of a self-centering concrete member can be achieved
using a lumped plasticity model, such as the two-spring model that was used by Palermo et al.
[7-2] to analyse the response of self-centering bridge piers. As shown in Figure 7.3, the
concrete member (wall, column, or beam) can be represented by an elastic beam member, with
the non-linear behaviour lumped into two rotational springs at the base of the member. The
two rotational springs, labelled A and B, represent the base moment-rotation behaviour of the
PT member and energy dissipating components, respectively. The moment-rotation response
at the base of a PT member is idealised as non-linear elastic, with a high initial stiffness and
non-linear behaviour initiating when the wall begins to uplift, as shown by Spring A. The
energy dissipating components are idealised using an elasto-plastic moment-rotation response,
as shown by Spring B.
- 218 -
7.2 Static Analysis of Flag-shape Response
7.2.2 Ruaumoko Analysis

To analyse the flag-shaped hysteresis behaviour, a two-spring lumped plasticity model was
constructed using the non-linear dynamic structural analysis program Ruaumoko [7-16]. The
model properties were approximately defined from the PreWEC test specimen, using an elastic
beam member with a length of 6.1 m, modulus of elasticity of 38 GPa, and moment of inertia
of 0.085 m4. The two rotational springs were defined with various properties, as described
below. The top of the beam member was then subjected to a single lateral displacement cycle
to 2.5% drift.
Spring A
θ
Elastic beam
member
M
Spring B
θ
A B
Figure 7.3 – Two-spring lumped plasticity representation of a self-centering wall system
The first analysis was conducted using idealised hysteresis models for the rotational springs,
including a bi-linear elastic definition for the PT spring and an elasto-plastic definition for the
energy dissipation (ED) spring. As shown in Figure 7.4a, both rotational springs had equal
yield moments of 1250 kN-m and ultimate moments of 1500 kN-m (corresponding to
approximately 2.5% lateral wall drift). The moment-drift response of the model, shown in
Figure 7.4b, indicated that the flag-shaped response consisted of five distinct sections. When
compared to the idealised flag-shape that was shown in Figure 7.1, the initial stiffness
- 219 -
exhibited two distinct sections. Initially, the stiffness of the system is controlled by the elastic
response of the wall, but once decompression occurs at the wall toe and the PT spring becomes
non-linear (point 1 in Figure 7.4), the stiffness is also influenced by the second slope of the PT
spring and the initial stiffness of the energy dissipating spring. Upon unloading, the hysteresis
response returned to the origin with zero residual drift. Using current self-centering procedures
described previously, a zero residual drift would be expected because the moment attributed to
the energy dissipating components did not exceed the moment attributed to the PT.
1500 1/5 2 4PT 3000 3

3
PT PT
2ED 2
1000 2500
Base Moment (kN-m)
Moment (kN-m)
500 2000
0 1500
1ED
1
-500 1000
-1000 500
4ED PT
5ED ED 5 4
-1500
0
0 0.005 0.01 0.015 0.02 0.025 0 0.5 1 1.5 2 2.5
Rotation,  (rad) Lateral Wall Drift (%)
(a) Rotational spring definitions (b) Total system hysteresis response

Figure 7.4 – Two spring analysis with idealised hysteresis and equal moment resistance
For the second analysis, the post-yield stiffness of the ED spring was increased, so that the
ultimate moment of the ED spring exceeded that of the PT spring, as shown in Figure 7.5a.
Interestingly, as observed from Figure 7.5b, the resulting moment-drift hysteresis response of
the system exhibited perfect self-centering with zero residual drift after unloading. This
demonstrated that the traditional criterion that was used to ensure self-centering occurred was
inaccurate and ambiguous, as perfect static self-centering was achieved even when the energy
dissipating components provided more than 50% of the total moment. Although there is some
correctness to the 50% ED moment criteria, the point at which this calculation is performed
during the response is critical. The proportion of the moment provided by the energy
dissipating components at the maximum lateral drift is not relevant, and instead the moment
balance calculation should be performed when the member is unloaded to zero displacement.
The negative moment resistance in the ED spring after unloading to zero displacement must
- 220 -
7.2 Static Analysis of Flag-shape Response
not exceed the yield (or decompression) moment of the PT spring. In the case of the analysis
shown in Figure 7.5, the 1250 kN-m yield moment PT spring exceeded the -1000 kN-m
unloaded moment in the ED spring, so a static self-centering response was achieved with zero
residual drift.
2000 3500
1500 3000
Base Moment (kN-m)

Moment (kN-m)
1000 2500
500 2000
0 1500
-500 1000
PT
500
-1000 ED
0
0 0.005 0.01 0.015 0.02 0.025 0 0.5 1 1.5 2 2.5

Figure 7.5 – Two spring analysis with idealised hysteresis and unbalanced moment
resistance
For real self-centering members, the hysteresis definitions of the PT and ED rotational springs
are not bi-linear elastic and elasto-plastic. The response of an undamaged PT member can be
more accurately approximated using a tri-linear elastic hysteresis (ignoring the wall hysteresis
energy dissipation) and the energy dissipating components typically display a smoother plastic
hysteresis response. For the third analysis, a tri-linear elastic definition was used for the PT
spring and a Ramberg-Osgood definition was used for the ED spring, as shown in Figure 7.6a,
and the yield moments and ultimate moments of the PT and ED springs were approximately
equal. As shown in Figure 7.6b, the resulting hysteresis response deviated from the ideal flag-
shape behaviour, with a more realistic smooth form. Although the moment provided by the
energy dissipating components did not exceed the moment provided by the PT, perfect self-
centering was not achieved and a small residual drift was observed. Again, this analysis
highlighted that the PT and ED moment contributions are only critical when the member is
unloaded to zero displacement. In this case, the unloaded moment in the ED spring of -1100
kN-m exceeded the 750 kN-m yield moment of the PT spring and thus perfect static self-
centering was not achieved.
- 221 -
1500 3000
1000
2500
Base Moment (kN-m)

Moment (kN-m)
500
2000
0
1500
-500
1000
-1000
PT 500
ED
-1500
0
0 0.005 0.01 0.015 0.02 0.025 0 0.5 1 1.5 2 2.5

Figure 7.6 – Two spring analysis with realistic hysteresis and equal moment resistance
For the final analysis, the realistic tri-linear and Ramberg–Osgood hysteresis definitions were
retained, but the PT moment contribution was increased, while the ED moment contribution
was reduced, as shown in Figure 7.7a. The resulting hysteresis response of the model
exhibited perfect static self-centering, with zero residual drift upon unloading. However, to
achieve perfect static self-centering, the moment provided by the energy dissipating
components contributed to only 41% of the total base moment at 2.5% lateral drift.
3000
1500
2500
Base Moment (kN-m)
1000
Moment (kN-m)
2000
500
1500
0
1000
-500
PT 500
-1000 ED
0
0 0.005 0.01 0.015 0.02 0.025 0 0.5 1 1.5 2 2.5

Figure 7.7 – Two spring analysis with realistic hysteresis and unbalanced moment
resistance
- 222 -
7.3 Residual Drift Limits
7.2.3 Summary
The inaccuracy of the current procedures that are used to ensure that self-centering is achieved
was demonstrated, and it was established that the moment contribution from the PT and energy
dissipating components is not critical at large drift levels. Instead, to achieve zero residual drift
in the static hysteresis response, the yield (or decompression) moment of the PT member must
exceed the moment provided by the energy dissipating components at zero base rotation after
prior displacement to the maximum design level drift. However, the design guidelines that
include a limit on the moment attributed to the energy dissipating components do not mention
at what point in the response this calculation must be preformed, and so it is assumed that the
limits would only be applied at the maximum design level drift/moment.
In reality, the response of a PT member does not have an easily defined yield point, with non-
linear behaviour initiating after decompression occurs at the tension wall toe. As a result, it is
difficult to ensure that zero residual drift is achieved. Small residual drifts should be expected
for real self-centering systems, even when the behaviour of the PT member remains non-linear
elastic. Additionally, large inelastic strains in the compression toe of a PT concrete wall will
cause the response to deviate from a non-linear elastic hysteresis behaviour and increase the
residual drifts that are observed from the cyclic hysteresis response.
7.3 RESIDUAL DRIFT LIMITS

As explained in previous sections, self-centering systems will typically exhibit some degree of
residual drift. However, because the response follows an approximately flag-shaped
behaviour, the residual drifts observed in a self-centering system are typically small when
compared to those of traditional construction systems. As observed by Kurama [7-12], due to
the shake-down effect, the small residual drift observed from the cyclic hysteresis loops of
hybrid walls led to almost zero residual drift during seismic excitation. If residual drifts still
occur in real self-centering systems, appropriate limits need to be set to define what is
classified as self-centering behaviour.
As discussed in section 7.1.3, several researchers have emphasised the importance of residual
drifts when conducting a performance based seismic assessment [7-14, 7-17], but no specific
guidelines for appropriate drift limits were discussed. Kawashima [7-18] reported that the
- 223 -
devastation during the 1995 Kobe earthquake highlighted the need to consider residual drift
performance criteria, and consequently the 1996 Japanese seismic design specification for
bridges [7-19] imposed a check on residual displacements, which were to be less than 1%
lateral drift.
To determine appropriate residual drift limits for a self-centering system, the target seismic
performance levels must first be determined. As outlined by Priestley [7-17] and the
performance based design appendix of the SEAOC blue book [7-20], the seismic performance
of structures can be categorised into four levels:
1. Immediate occupancy. Continued operation with negligible damage.

2. Occupancy. Continued operation with minor repairable damage and minor disruption
to non-essential services.
3. Life safety. Life safety is protected, with moderate to extensive damage.
4. Near collapse. Structural collapse is prevented, but with severe damage.
The design objective is then used to relate these four performance levels to four levels of
seismic hazard, as shown in Figure 7.8. The SEAOC blue book [7-20] defines the four levels
of seismic hazard as:
 EQ-I: 87% probability in 50 years (33% of EQ-III).

 EQ-II: 50% probability in 50 years (50% of EQ-III).
 EQ-III: 10% probability in 50 years [referred to as design base earthquake (DBE)].
 EQ-IV: 2% probability in 50 years (150% of EQ-III) [referred to as the maximum
credible earthquake (MCE)].
The basic, or minimum, design objective is typically used for the prevailing ductile seismic
design philosophy, where collapse is prevented by non-linear structural behaviour that results
in extensive damage. Resilient seismic design correlates to the enhanced objectives 1 and 2,
where the damage caused to the structure during the design level and maximum credible
earthquakes is reduced.
- 224 -
7.3 Residual Drift Limits
Figure 7.8 – Seismic performance objectives for buildings [7-20]
For PreWEC and other self-centering concrete systems, the design philosophy typically targets
the enhanced objective 1 performance levels. This target objective means that a building
designed using the PreWEC system would be expected to achieve full seismic resilience when
subjected to the design level earthquake, with only minimal and repairable damage resulting.
During the maximum credible earthquake, life safety is the main concern for the PreWEC
system, but damage should still be limited and in most cases repairable. To achieve enhanced
objective 1 for the entire building, the residual drift performance levels were defined as shown
in Figure 7.9. To limit the damage during the maximum credible earthquake to a repairable
state, the building should exhibit a self-centering response with permanent residual drifts that
allow for re-occupancy following the repair.
Using the performance targets shown in Figure 7.9, residual drift limits for the PreWEC system
were established for each hazard level. Residual drift limits that were developed by Rahman
and Sritharan [7-21, 7-22] were used as the performance targets for each earthquake
performance level. This procedure resulted in the residual drift limits described in Table 7.1,
which limits the permanent residual drift to 0.2% following the design level earthquake and
0.3% following the maximum credible earthquake. It was considered that these residual drift
levels provided realistic limits for a resilient building system, with a self-centering response
ultimately achieved.
- 225 -
Figure 7.9 – Residual drift performance objectives for PreWEC
Table 7.1 – PreWEC residual drift performance limits
Seismic hazard Performance level Residual drift limit (%)

EQ-I Immediate occupancy 0.1
EQ-II Immediate occupancy 0.1
EQ-III Immediate occupancy / occupancy 0.2
EQ-IV Occupancy 0.3
7.4 DYNAMIC ANALYSIS

To accurately assess the expected residual drifts of a structure following an earthquake, both
the realistic cyclic hysteresis response and the dynamic shake-down need to be included.
Using the PreWEC system as an example, a series of dynamic time-history analysis were
preformed to investigate the self-centering behaviour of realistic structures subjected to
earthquake ground motions. As well as determining whether the PreWEC system would meet
the residual drift performance limits that were set, the analysis allowed a parametric
investigation to be conducted to developed appropriate self-centering design procedures.
The structural analysis program Ruaumoko [7-16] was selected for this analytical
investigation. Ruaumoko is a non-linear dynamic structural analysis program capable of both
- 226 -
7.4 Dynamic Analysis
elastic and inelastic analysis of structures subjected to earthquake and other dynamic loadings.
Both two and three dimensional analyses can be conducted, with models created from various
line, plate and spring elements. The main advantage of Ruaumoko is the extensive library of
hysteresis rules that can be used to define the nonlinear characteristics of the model elements.
7.4.1 Model Development

The detailed finite element model, which was described in Chapter 6, provided a tool that was
used to closely examine the localised behaviour of the PreWEC specimen, but this localised
behaviour was not required for the dynamic analysis. For the purpose of the dynamic analysis,
a lumped plasticity model was used to represent the PreWEC system. The lumped plasticity
model was developed with a combination of non-linear springs, to closely match the recorded
hysteresis response of the PreWEC test specimen. The reduced complexity of the analytical
model used for the dynamic analyses allowed for a large parametric study to be conducted.
7.4.1.1 Two-spring model

Initially, a simple two-spring lumped plasticity model was trialled, similar to that shown
previously in Figure 7.3. The wall was modelled using a beam element with a 6.1 m length,
equal to the loading height of the PreWEC test specimen. The beam element was assigned an
elastic hysteresis rule based on the uncracked elastic properties of the concrete wall and
columns. Using the concrete stress-strain definition that was defined for the FEM in Chapter 6,
the modulus of elasticity of the concrete wall was calculated to be 38 GPa, and the calculated
in-plane moment of inertia of the wall plus the two end columns was equal to 0.0850 m4. The
elastic in-plane shear deformation of the wall panel was considered to be negligible and
therefore ignored during the analysis.
The two rotational springs designed to characterise the non-linear behaviour at the wall base
were calibrated using the experimental data. The test backbone moment-rotation relationship
at the wall base, calculated from the average of the experimental measurements in each loading
direction, is shown in Figure 7.10a. The test base moment-rotation backbone was idealised for
the model using five linear sections, which were then separated into the moment contribution
from the PT and from the energy dissipating connectors. The average vertical displacement
measured across the connectors during the PreWEC test showed a linear relationship with the
calculated base rotation. The moment provided by the connectors for each increment of the
base rotation was then calculated from the O-Connector tests that were described in Chapter 5.
- 227 -
The connector forces on each end of the wall were assumed to be equal and opposite, and thus
the connector moment was equal to the magnitude of the force in the 10 O-Connectors at one
end multiplied by a 1.828 m lever arm that corresponded to the wall length. Using this
approach the backbone base moment-rotation behaviour attributed to the connectors was
calculated and subtracted from the total base moment in order to calculate the backbone base
moment-rotation behaviour of the PT wall and column, as plotted in Figure 7.10a.
3000
2500
Base Moment (kN-m)
2000 Test
Model: Total
1500 Model: PT w all & col
Model: Connector
1000
500
0
0 0.005 0.01 0.015 0.02 0.025 0.03
Base Rotation (rad)
(a) Backbone moment-rotation response
Lateral Drift (%)

-3 -2 -1 0 1 2 3
3000
2000
Base Moment (kN-m)
1000
-1000
-2000
Test
-3000 Model
-200 -100 0 100 200

(b) Full cyclic response

Figure 7.10 – Two-spring lumped plasticity model of PreWEC
- 228 -
The wall PT rotational spring was modelled using a multi-linear elastic hysteresis rule, which
was defined to match the derived backbone response shown in Figure 7.10a. The connector
rotational spring behaviour was defined using a bounded Ramberg-Osgood hysteresis rule,
with a yield moment of 650 kNm, an initial stiffness of 250,000 kNm-rad, and a bi-linear
factor, r, of 10. The connector rotational spring properties were defined to match the backbone
base moment-rotation response as well as the cyclic behaviour observed from the O-Connector
tests. The bottom node of the rotational springs was restrained for all degrees of freedom, and
the node at the top of the springs was restrained for all degrees of freedom except the in-plane
rotation. To check the cyclic behaviour of the Ruaumoko model, the top node of the wall
element was subjected to the same cyclic lateral displacement history that was measured during
the PreWEC test. The velocity was limited to 0.5 mm/s to ensure that the dynamic inertia
effects were negligible.
The global moment-drift response of the two-spring lumped plasticity model is compared
against the experimental response in Figure 7.10b. It can be seen that the model exhibited an
almost ideal flag-shaped response, which significantly underestimated the hysteresis behaviour.
The multi-linear elastic hysteresis behaviour of the wall PT spring did not capture the true
hysteresis behaviour of the concrete wall, because the inelastic strains in the wall toe and the
resulting residual drift were ignored. The objective of the analysis was to investigate the
residual drift behaviour, and thus a more complex model was next developed in order to
improve the representation of the PreWEC hysteresis behaviour.
7.4.1.2 Seven-spring model

To accurately capture the PreWEC hysteresis behaviour, an arrangement of seven rotational
springs at the base of the wall was developed, as shown in Figure 7.11. It should be noted that
the layout shown in Figure 7.11 has been expanded in order to display the connections between
the springs, whereas in the actual model the nodes that the springs were connected to were all
located at the same position at the bottom of the elastic beam member. The connector spring,
C, was unchanged from the previous two-spring model, but the behaviour of the PT wall was
separated into six springs, two in parallel and a further four in series. Springs W1 and W2
controlled the general behaviour of the PT wall. W1 had a multi-linear elastic hysteresis
definition, while W2 used a degrading bi-linear elasto-plastic hysteresis definition to add the
necessary hysteresis area caused by inelastic strains in the wall toe. Although a trial model
with springs W1, W2 and C showed an improved hysteresis area, the residual displacement
- 229 -
could not be captured using springs that were placed in parallel. For this reason, four springs
were added to control the unloading path and the residual displacement. The springs, labelled
RD, acted as switches that turned on when critical moment limits were reached. The RD
springs had an extremely high initial stiffness, and after a specified yield moment was reached,
the springs began their non-linear rotation that caused a residual displacement when the lateral
force was unloaded to zero. The RD springs were designed to act in pairs that had yield
moments corresponding to points on the W1 spring multi-linear elastic hysteresis definition.
To accurately capture the unloading stiffness of the PT wall, each RD pair consisted of one
spring with a modified Tekada hysteresis definition and another spring with a revised origin
centered hysteresis definition.

me
W1 W2
he
RD1a
C
RD1b
RD2a
RD2b
Figure 7.11 – Ruaumoko seven spring lumped plasticity model for PreWEC
- 230 -
During the development of the seven spring lumped plasticity model, the combination of the
backbone response of each spring was calculated to ensure that the experimental base moment-
rotation envelope was closely matched. The properties of the rotational springs were revised
though a trial and error process until the best correlation was obtained with the experimental
PreWEC response. The final properties used in the seven spring lumped plasticity model are
given in Table 7.2, based on the hysteresis rules defined in Figure 7.12.
Table 7.2 – Element properties for PreWEC Ruaumoko model
Element Function Hysteresis rule Property* Value

Ec 38 GPa
Beam Wall panel response Elastic
Iz 0.0850 mm4
My 1335 kNm
K 11838544 kNm-rad
r 0.0142576
F1 0.000389
W1 Wall base PT response Multi-linear elastic D1 13.34
F2 0.000166
D2 17.44
F3 0.002423
D3 67.78
My 600 kN
Degrading Bi-linear Elasto- K 240000 kNm-rad
W2 Wall base hysteresis area
plastic r 0.048408
α 0.5
My 1931 kNm
K 200000000 kNm-rad
Residual drift offset r 0.001973
RD1a st Modified Tekada
(1 force limit) α 0.5
β 0
unloading Drain-2D
Residual drift unloading My 1931 kNm
RD1b stiffness correction Revised Origin Centered K 200000000 kNm-rad
(1st force limit) r 0.000489
My 2243 kNm
K 200000000 kNm-rad
Residual drift offset r 0.000234
RD2a nd Modified Tekada
(2 force limit) α 0.5
β 0
unloading Drain-2D
Residual drift unloading My 2243 kNm
RD2b stiffness correction Revised Origin Centered K 200000000 kNm-rad
(2nd force limit) r 0.000100
My 650 kNm
Connector hysteresis Bounded Ramberg-
C K 250000 kNm-rad
response Osgood
r 10
* where hysteresis rule properties are defined in Figure 7.12
- 231 -
(a) Modified Tekada
(b) Multi-linear elastic
(c) Degrading bi-linear elasto-plastic
(d) Revised origin centered (e) Bounded Ramberg-Osgood

Figure 7.12 – Hysteresis rules used for PreWEC Ruaumoko model [7-16]
As with the two spring model described earlier, the top node of the elastic beam, at a height of
6.1 m, was subjected to the cyclic lateral displacement history that was measured during the
PreWEC test. The results of the cyclic analysis of the Ruaumoko seven-spring lumped
plasticity model are compared against the PreWEC experimental response in Figure 7.13. It
can be seen that the combination of seven rotational springs accurately captured the
- 232 -
experimental response though all stages of the hysteresis loop. In particular the model
accurately captured the unloading curve, residual displacement, and reloading stiffness
degradation, which were not captured when previous researchers used idealised flag-shaped
hysteresis rules. A close up of the initial cycles of the response is shown in Figure 7.13b.
Accurate calculation of the PreWEC response was achieved for the small initial cycles, which
provided confidence that accuracy would be maintained for low earthquake excitation.

-3 -2 -1 0 1 2 3 -1 -0.5 0 0.5 1
3000 3000
2000 2000
Base Moment (kN-m)
Base Moment (kN-m)

1000 1000
0 0
-1000 -1000
-2000 -2000
Test Test
-3000 Model -3000 Model
-200 -100 0 100 200 -60 -40 -20 0 20 40 60

(a) Full response (b) Close up of initial cycles

Figure 7.13 – Calculated cyclic response of PreWEC Ruaumoko model
To further assess the suitability of the Ruaumoko model for conducting dynamic analysis of
the PreWEC system, the calculation of the hysteretic energy dissipation and residual drift of
the cyclic loops was also investigated. As observed in Figure 7.14a, the equivalent viscous
damping, calculated from the area of the Ruaumoko model hysteresis loops, compared well
with the experimental results. The Ruaumoko model did underestimate the equivalent viscous
damping during the first cycles to 2, 2.5 and 3% lateral drift, but closely matched the calculated
experimental equivalent viscous damping during the second and third cycles to the same drift
level. Additionally, the Ruaumoko model closely calculated the experimental residual drift, as
seen in Figure 7.14b. Given that the primary objective of the analytical investigation was to
assess the permanent residual drift of buildings constructed with PreWEC walls, it was
concluded that the critical cyclic residual drift was accurately captured using the lumped
plasticity model.
- 233 -
20 0.8
Test
0.7
Model
15 0.6
Residual Drift (%)

0.5
10 0.4
0.3
5 0.2
Test 0.1
Model
0 0
0 0.5 1 1.5 2 2.5 3 -4 -2 0 2 4
Lateral Drift Cycle (%) Lateral Drift Cycle (%)
(a) Equivalent viscous damping (b) Static residual drift

Figure 7.14 – Calculated hysteresis loop parameters for PreWEC Ruaumoko model
7.4.2 Displacement Based Design

Direct displacement based design (DDBD) was used to design the parametric data set of
PreWEC walls using the base moment-rotation behaviour of the PreWEC test specimen. As
described by Priestley et al. [7-23] DDBD typically involves the following steps:
1. Calculate the design lateral displacement of an idealised single degree of freedom

(SDOF) representation, ∆d, using Eq. 7.4 and the assumed displaced shape of the
structure.
   m  
n n
 d   mi 2i i i (7.4)
i 1 i 1
where n is the number of storeys, mi is the storey mass, and ∆i is the design
displacement at storey i.
2. Calculate the effective mass, me, and the effective height, he, of the SDOF
representation using Eqns. 7.5 & 7.6.
 
n
m e   mi  i d (7.5)
i 1
- 234 -
   m  
n n
he   mi  i hi i i (7.6)
i 1 i 1
where hi is the height to the ith storey.
3. Estimate the system damping, which should be checked and revised following member
design.
4. Use design displacement spectra to calculate the effective period of the structure, Te
based on the design displacement and damping calculated in steps 1 and 3.
5. Calculate the effective stiffness of the SDOF structure, Ke, using Eq. 7.7.
K e  4 2 me Te2 (7.7)
6. Calculate the design base shear of the structure, Vb, using Eq. 7.8.
Vb  K e  d (7.8)
7. Distribute the design base shear to the entire structure and calculate member actions.
The assumed displacement profile of the structure is required in order to estimate the design
displacement. For precast walls with unbonded post-tensioning, the deformation at the
maximum lateral displacement is dominated by the base rotation due to wall rocking, which
results in an almost linear displacement profile up the wall height. For this reason, Priestley
[7-24] recommends that for precast concrete wall buildings of up to 10 storeys tall the
displacement at each storey can be calculated using Eq. 7.9. The results from the PreWEC
FEM analysis in Chapter 6 confirmed the suitability of the linear displacement profile defined
using Eq. 7.9, with over 90% of the calculated lateral displacement attributed to base rotation.
 i  hi  Drift (7.9)
where ∆Drift is equal to the design level interstorey drift.
7.4.3 Parametric Matrix

In order to assess the seismic performance and permanent residual drift over a wide range of
PreWEC designs, an extensive parametric matrix was developed to vary the fundamental
period of the structure, quantity of hysteretic energy dissipation, viscous damping, seismic
- 235 -
hazard level, and earthquake ground motion. A total of 400 analyses were completed and
details of the parametric variations are discussed in this section.
The first set in the parametric matrix consisted of five single lumped mass models that were
designed to represent buildings with heights equivalent to 2, 4, 6, 8 and 10 storeys, as shown in
Figure 7.15a. For each of the five models, the lateral resistance was provided by a single
PreWEC wall with cross-sectional dimensions and base moment-rotation behaviour equal to
the prototype PreWEC test specimen that was described and analysed in Chapters 5 and 6.
Note that the five models in the set were not intended to represent an increasing number of
floors of the same floor-plan. Instead, each wall was designed with a tributary area (and mass)
sufficient to impose a seismic demand that the PreWEC test wall was designed to resist, as
shown in Figure 7.15b and discussed in more detail below. This first set of five wall models is
referred to as Ptest to represent the PreWEC test specimen.

Tributary area
PreWEC wall
me
he
10 storey
(b) PreWEC wall with tributary area

2 storey
4 storey
6 storey
8 storey
(a) Five lumped mass models

Figure 7.15 – Graphical representation of parametric model variations
- 236 -
In order to design a range of different structures using the same PreWEC wall design, the
DDBD procedure was back-calculated from the required base moment-rotation behaviour.
Because the PreWEC test was conducted with the specimen constructed at 50% scale, the
properties of the Ruaumoko spring model were first increased to full size by scaling the
dimensions, while retaining the same stress and strain levels. The scale factors used for each
variable of the model are summarised in Table 7.3, and the resulting full-scale base moment-
rotation behaviour that was used for all walls in set Ptest is plotted in Figure 7.16.
Table 7.3 – Scale factors for full size PreWEC model
Variable Unit Scale factor

Displacement, d m 2
Stress (σ) / strain (ε) / rotation (θ) MPa / m/m / rad 1
Force, F kN (MPa × m2) 4
Moment, M kN-m 8
Moment-rotation stiffness, K kN-m / rad 8
Moment of inertia, I m4 16
30
20
Base Moment (MN-m)
10
-10
-20
-30
-0.03 -0.02 -0.01 0 0.01 0.02 0.03
Base Rotation (rad)
Figure 7.16 – Base moment-rotation behaviour for all walls in the Ptest set
A further two parametric sets were developed to represent modified base moment-rotation
behaviours. The two sets were labelled Pmore and Pless, and represented base moment-rotation
behaviours with more or less hysteresis area. Lastly, wall set Ptest was analysed with three
different amounts of assumed viscous damping in the analysis.
- 237 -
7.4.3.1 Fundamental period / building height

The five wall models in set Ptest were designed to represent 2, 4, 6, 8 and 10 storey buildings,
with an assumed inter-storey height of 3.66 m. The walls were designed for a design level
displacement of 2% lateral drift, as recommended by the ACI ITG-5 documents [7-11, 7-13]
and used by previous researchers [7-7, 7-21, 7-25]. At 2% lateral drift the base moment
resistance of the full scale PreWEC test specimen was calculated to be 24800 kN-m. The
effective period and mass of the SDOF representation of each structure was then back-
calculated from this moment resistance using the following procedures:
1. Calculate the design lateral displacement of the idealised SDOF representation, ∆d,
using Eq. 7.4 and assuming a linear displaced shape as represented by Eq. 7.9. Note:
the mass was assumed to be constant at each storey level and so the mass terms in the
numerator and denominator of Eq. 7.4 cancelled each other out.
2. Calculate the effective height of the SDOF representation using Eq. 7.6 (again the mass
terms cancelled out).
3. Calculate the wall design base shear, Vb, by dividing the design moment resistance of
24800 kN-m by the effective SDOF height calculated from step 2.
4. Use the design base shear, Vb, to back calculate the effective tangent stiffness of the
SDOF system, Ke, using Eq. 7.8.
5. Calculate the effective building period, Te, from the design displacement spectra (as
discussed below).
6. Back calculate the effective SDOF mass, me, using Eq. 7.7 and the calculated value of
Ke and Te. This was the mass required to excite the building to 2% lateral drift during
the design base earthquake (DBE).
7. Calculate the initial stiffness of the building, Ti, [using the Ruaumoko model].
To calculate the effective period, Te, (step 5 above), an estimation of the equivalent viscous
damping was required to interpret the displacement response spectra. As described by
Priestley et al. [7-23], the total equivalent viscous damping, ζeq, is equal to the sum of the
elastic and hysteretic damping, ζel and ζhyst respectively. However, this calculation must be
performed at the design level lateral displacement which results in Eq. 7.10. The factor κ is
used to determine the proportion of the building’s inherent elastic damping that acts at the
design level displacement, which is calculated from the displacement ductility factor, μ, and
the secant stiffness correction factor, λ, as shown in Eq. 7.11.
- 238 -
 eq     el   hyst (7.10)
   (7.11)
The elastic damping component of the equivalent viscous damping is typically included in
time-history analyses to account for damping that is not captured by the hysteretic model. For
a concrete structural system, elastic damping is attributed to several sources, including
nonlinear behaviour of materials prior to yield (such as micro-cracking of concrete), friction,
and behaviour of non-structural elements. Additionally, for a rocking wall system, such as
PreWEC, the elastic damping is also relied on to account for the radiation or contact damping
as the wall impacts with the foundation. Typically 5% elastic damping is used when modelling
reinforced concrete buildings, but the elastic damping in a self-centering system will be less
because of the reduced cracking to the wall panel. After calculating the equivalent viscous
damping of PT concrete masonry walls from shake-table tests conducted by Wight et al. [7-26],
Ma [7-27] recommended that 3% elastic damping was appropriate for self-centering wall
systems. For this reason, 3% initial stiffness proportional damping was used for this analytical
investigation, as was also used by Kurama et al. [7-12, 7-28] during investigation of a hybrid
self-centering wall system. Additionally, the sensitivity of the chosen elastic damping was
investigated during the parametric study, as discussed in more detail in section 7.4.3.3.
Recommendations provided by Priestley et al. [7-23] were used to calculate the secant stiffness
correction factor, λ. As described earlier, the PreWEC hysteresis behaviour deviated from an
ideal flag-shape hysteresis rule, and instead could be represented by a pinched Takeda
hysteresis rule. Therefore, a λ value of 0.310 was selected, which represented an intermediate
value to that suggested by Priestley et al. [7-23] for ideal flag-shape and Tekada hysteresis
rules. It was difficult to define a displacement ductility factor for the PreWEC system, because
there is no well defined yield point, and so an estimate of μ = 6 at the design level 2% lateral
drift was used, based on the hysteresis response shown previously in Figure 7.13a with an
estimated yield drift of 0.33%. Using Eq. 7.11, these assumptions resulted in a κ factor of 1.70,
which equated to an elastic damping at the design level displacement of 5.2%. Additionally,
the hysteretic damping at the design level 2% lateral drift was calculated to equal 15.3% for the
Ruaumoko model, as seen previously in Figure 7.14a. Using Eq. 7.10, the total equivalent
viscous damping for the PreWEC specimen at the design level displacement was calculated to
be 20.5%.
- 239 -
The exact site condition was not critical to the outcome of this investigation, so for the purpose
of this study the buildings were located in California seismic Zone 4 on intermediate soil, Sc, as
defined by SEAOC [7-20]. Because the calculated equivalent viscous damping was equal to
20.5% at the design level drift, a 20% damped displacement response spectrum was used
during the design process. Following the SEAOC performance based design appendix [7-20],
the 20% damped displacement response spectrum for Zone 4, soil type Sc, and seismic hazard
level EQ-III was used to design the buildings for the parametric study, as shown in Figure 7.17.

Sd (m)
Sdd = 0.366 m
S dd T
Sd 
Td
Ts = 0.67s Td = 4.45s T (s)
Figure 7.17 – 20% damped displacement response spectrum for Zone 4, soil type Sc, and
seismic hazard level EQ-III
Following the design procedure described above, the five walls in set Ptest were designed, and a
summary of the calculated design parameters is given in Table 7.4. Because the wall moment
resistance was kept constant, the effective mass, me, was equal for all five walls, and the
fundamental period of the walls was varied by increasing the wall height, or equivalent
building height. As shown in Table 7.4, this variation resulted in a wide range of fundamental
periods (Ti) between 0.38 and 2.93s, which corresponds to a realistic range for concrete wall
buildings designed using DDBD.
Table 7.4 – DDBD parameters for the five buildings designed with PreWEC
Wall No. of ∆d (m) he (m) Vb (kN) Ke (kN-m) Te (s) me Ti (s)

storeys (tonne)
1 2 0.122 6.1 4066 33324 1.48 1860.3 0.38
2 4 0.220 10.98 2259 10285 2.67 1860.3 0.86
3 6 0.317 15.86 1564 4930 3.86 1860.3 1.46
4 8 0.415 20.74 1196 2883 5.05 1860.3 2.15
5 10 0.512 25.62 968 1889 6.24 1860.3 2.93
- 240 -
7.4.3.2 Hysteretic energy dissipation

A further two sets of five walls that had a base moment-rotation behaviour with either more or
less hysteresis area were included in the parametric study, referred to as Pmore and Pless. The
modified hysteresis area was achieved by increasing or reducing the number of energy
dissipating connectors, in a similar manner to the FEM analyses reported in Chapter 6. The
hysteresis behaviour was modified by multiplying the connector spring moments by a factor of
1.4 (14 connectors instead of 10) or 0.8 (8 connectors instead of 10), while the moment
resistance of the six remaining springs that represented the PT wall behaviour were reduced or
increased by the same factors respectively. The modified base moment-rotation behaviours of
the modified Ruaumoko models are shown in Figure 7.18. It was observed that the area of the
hysteresis loops increased and decreased accordingly and that the cyclic residual displacements
were also affected. The modified base moment-rotation behaviours were then applied to all
five walls in the set shown previously in Table 7.4, providing an increased data set to assess the
residual drift and self-centering behaviour of the PreWEC system.
30 30
20 20
Base Moment (MN-m)
Base Moment (MN-m)
10 10
0 0
-10 -10
-20 Ptest -20 Ptest

Pmore Pless
-30 -30
-0.03 -0.02 -0.01 0 0.01 0.02 0.03 -0.03 -0.02 -0.01 0 0.01 0.02 0.03
Base Rotation (rad) Base Rotation (rad)
(a) More connectors (14 each side) (b) Less connectors (8 each side)
Figure 7.18 – Ruaumoko base moment-rotation behaviour for wall sets Pmore and Pless
7.4.3.3 Viscous damping

The influence of the assumed elastic viscous damping was also investigated. As described
above, 3% damping was assumed during the design and analytical investigation. Analyses for
wall set Ptest were repeated with 1% and 5% viscous damping to assess the sensitivity of the
assumed elastic damping, as well as investigate any potential relationship between the elastic
damping and the residual drift.
- 241 -
7.4.4 Earthquake Records

To analyse the behaviour of the PreWEC walls, the Ruaumoko spring model was subjected to a
series of earthquake ground motions representing the design level (EQ-III) and maximum
credible (EQ-IV) seismic hazard. All the ground motions used during this investigation were
soured from the PEER strong motion database (http://peer.berkeley.edu/smcat/).
7.4.4.1 Ground motion selection

A total of eight earthquake ground motions (GM) were selected for this investigation, with the
details given in Table 7.5. The ground motion records were primarily sourced from previous
studies [7-3, 7-22, 7-25] and selected to represent a wide range of excitation that was expected
for a high seismic zone and intermediate soil (Sc) condition around the world. Except for El
Centro and Imperial Valley, all of the recording stations were located in USGS soil class B
which corresponds to an intermediate soil class Sc as defined by SEAOC [7-20]. The El Centro
and Imperial Valley recording stations were located on softer class C soil, but the records were
still found to be compatible with the Sc response spectrum.
Table 7.5 – Summary of earthquake ground motions (GM)
GM Earthquake Year MW Station Direction Soil Distance PGA PGV

(USGS) (km) (g) (m/s)
1 Chi-Chi 1999 7.6 CHY080 NS B 6.95 0.902 1.024
2 El Centro 1940 7 El Centro Array EW C 8.3 0.215 0.302
o
3 Imperial Valley 1979 6.5 Cucapah 85 N C 23.6 0.309 0.363
4 Kobe 1995 6.9 KJMA NS B 0.6 0.821 0.813
5 Loma Prieta 1989 6.9 Gillroy Array #6 NS B 19.9 0.126 0.128
6 Northridge 1994 6.7 Coldwater Can EW B 14.6 0.271 0.222
7 Northridge 1994 6.7 Mt Gleason Ave NS B 17.7 0.127 0.138
8 Tabas 1978 7.4 Tabas LN B 1.2 0.836 0.978
MW is the moment magnitude, PGA is the peak ground acceleration, and PGV is the peak ground velocity
The unscaled acceleration records for each of the eight earthquake ground motions are plotted
in Figure 7.19. Although, no provision for near fault effects was included during the wall
design, the Chi-Chi, Kobe, and Tabas ground motion records were identified during a study by
Cox and Ashford [7-29] to contain a strong velocity pulse. A strong velocity pulse was
considered to have a greater chance of causing a significant residual drift, by reducing the self-
centering ability during dynamic shake-down.
- 242 -
1
0.2
0.5
Acceleration (g)
Acceleration (g)
0.1
0 0
-0.1
-0.5
-0.2
GM-1: Chi-Chi GM-2: El Centro
-1
0 10 20 30 40 50 60 70 0 10 20 30 40
Time (s) Time (s)
0.8
0.3
0.6
0.2
0.4
Acceleration (g)
Acceleration (g)
0.1 0.2
0 0
-0.2
-0.1
-0.4
-0.2
-0.6
-0.3 GM-3: Imperial GM-4: Kobe
-0.8
0 10 20 30 40 0 10 20 30 40
Time (s) Time (s)
0.15 0.3
0.1 0.2
Acceleration (g)
Acceleration (g)
0.05 0.1
0 0
-0.05 -0.1
-0.1 -0.2
GM-5: Loma P GM-6: N-Ridge 1
0 10 20 30 40 0 5 10 15 20 25
Time (s) Time (s)
0.15 1
0.1
0.5
Acceleration (g)
Acceleration (g)
0.05
0 0
-0.05
-0.5
-0.1
GM-7: N-Ridge 2 GM-8: Tabas
-1
0 5 10 15 20 25 30 0 5 10 15 20 25 30 35
Time (s) Time (s)
Figure 7.19 – Unscaled selected earthquake acceleration records
- 243 -
7.4.4.2 Ground motion scaling

The earthquake acceleration records were scaled to match the 5% damped response spectrum
for California seismic zone 4, soil type Sc, and seismic hazard levels EQ-III and EQ-IV, which
are shown in Figure 7.20. The scaling was performed as outlined by the SEAOC guidelines [7-
20], which included minimising the sum of the squared difference between spectral ordinates,
while also ensuring that all spectral ordinates were at least 70% of the design response
spectrum. This scaling procedure was conducted for each of the five walls in the design set,
with the response spectrum matched between the fundamental period, Ti, and the effective
period, Te. The calculated scale factors used during the analysis are summarised in Table 7.6,
and a graphical representation of the scaled response spectra is included in Appendix C.
2
EQ-III
EQ-IV
Spectral Acceleration, Sa (g)
1.5
0.5
0
0 1 2 3 4 5
Peroid (s)
Figure 7.20 – SEAOC 5% damped acceleration response spectrum for Zone 4 soil type Sc.
Except for Chi-Chi (GM-1) and Kobe (GM-4), the acceleration response spectra from all
ground motion records were found to be compatible, with the average matching closely with
the SEAOC design response spectrum, as shown in Appendix C. The Chi-Chi and Kobe
ground motions included high peaks in the spectral acceleration between 0.6 to 1.2s, which
caused some discrepancy with the design spectrum for the period range of walls 2 and 3.
Without Chi-Chi and Kobe included, the average of the scaled ground motion response spectra
closely matched the design spectrum for walls 2 and 3.
- 244 -
Table 7.6 – Scale factors for earthquake ground motions
Hazard Wall Ground motion

level
GM-1 GM-2 GM-3 GM-4 GM-5 GM-6 GM-7 GM-8
Chi-Chi El Centro Imperial Kobe Loma P N-Ridge 1 N-Ridge 2 Tabas
1 0.47 1.85 1.71 0.43 3.84 1.87 3.01 0.64
2 0.7 1.75 1.6 0.77 3.77 1.92 3.53 0.6
EQ- III 3 0.97 1.73 1.81 1.13 5.01 1.92 3.63 0.62
4 0.99 1.78 2.2 1.4 5.35 2.11 4.54 0.62
5 1.08 2.01 2.64 1.53 5.59 2.32 4.41 0.66
1 0.7 2.78 2.57 0.64 5.75 2.8 4.51 0.96
2 1.05 2.62 2.4 1.15 5.65 2.88 5.29 0.89
EQ-IV 3 1.45 2.59 2.71 1.69 7.52 2.88 5.44 0.93
4 1.49 2.68 3.3 2.1 8.02 3.17 6.81 0.94
5 1.62 3.01 3.95 2.29 8.38 3.48 6.61 1
7.4.5 Analysis Details

The dynamic time-history analyses were conducted using the lumped plasticity model of the
PreWEC system, which was described previously in section 7.4.1.2. The analyses were
conducted using the lumped mass, or SDOF representation, that was used during the design
process. The entire effective mass, me, was located at the top node of the wall element, which
had a length equal to the effective wall height, he, as was shown previously in Figure 7.11 and
Figure 7.15a. In addition to the lumped mass, me, a negligible 0.1 kg mass was added to the
rotational degree of freedom of each of the base springs to prevent instability of the analysis
solution.
The analyses used an integration time step of 0.001s with the Newmark constant average
acceleration solution method. This time step was reduced to 0.0005s for walls 1 and 2, after
some instability in the solution was observed. No gravity or static loads were applied to the
model because the moment-rotation behaviour of the base springs was defined to include the
effect of the wall self-weight. As described above, 3% initial stiffness proportional Rayleigh
viscous damping was used in the model. The horizontal earthquake ground motions were
applied separately to the structure and all nodes were restrained from movement in the out-of-
place direction. Additionally, the nodes between the rotational springs, and at the bottom of
the wall beam member, were constrained to allow only in-plane rotation to occur. The analysis
time was equal to the full length of the ground motion record plus a free vibration time that
- 245 -
was equal to at least 20Ti. The free vibration time allowed the structure sufficient time for the
response to decay completely, allowing accurate calculation of the final residual drift.
7.4.6 Analysis Results and Discussion

A summary of the results from the 400 analyses conducted during the parametric investigation
are discussed in this section. The typical observed response is discussed, followed by a
discussion of the maximum drift demand and residual drift calculated for wall sets Ptest, Pmore,
and Pless. A detailed investigation of the residual drift ratio and the correlation with design
parameters is also included. Lastly, the results from the trials with different assumed viscous
damping are discussed.
7.4.6.1 Typical responses

The calculated response from the Ruaumoko time-history analyses followed the moment-
rotation hysteresis defined using the seven rotational springs. The calculated lateral drift time-
history and the base moment-lateral drift response for two example analyses are shown in
Figure 7.21, where the lateral drift was calculated by dividing the lateral displacement
measured at the top node by the effective wall height, he. Additionally, the calculated lateral
drift time-history and the base moment-lateral drift response for all analyses are included in
Appendix D.
Figure 7.21a shows a typical symmetrical response, where the lateral drift demand was
approximately equal in both the positive and negative directions. The base moment-drift
response closely followed the expected hysteresis behaviour observed from the pseudo-static
cyclic analysis, reaching the 2% design target drift on subsequent positive and negative load
cycles. The decay observed during the shake-down period (between 22 and 40s) returned the
wall to negligible residual drift, as was expected because of the symmetry of the response.
In contrast to the symmetrical response, Figure 7.21b shows an example of an asymmetrical

response, which exhibited a single large pulse in one direction. This asymmetrical response
typically occurred when the frequency of a velocity pulse in the ground motion matched the
fundamental period of the wall. In this case, the wall was pushed beyond the design level drift
during the early stages of the ground motion and the response only returned to about a quarter
of this peak displacement during the load reversal in the positive direction. It was expected
that this pulse type response would led to a large residual drift. However, the subsequent
- 246 -
smaller load cycles were sufficient to re-center the wall, with minimal residual drift observed at
the end of the analysis. This finding highlighted the significance of the shake-down effect, and
the ability of the PreWEC hysteresis behaviour to self-center during a small load reversal that
followed the peak lateral displacement.
30
2
Base Moment (MN-m)

20
Lateral Drift (%)
1 10
0 0
-1 -10
-20
-2
-30
0 10 20 30 40 50 -2 -1 0 1 2
Time (s) Lateral Drift (%)
(a) Ptest wall-3 GM-6 EQ-III
30
2
Base Moment (MN-m)
20
Lateral Drift (%)
1 10
0 0
-1 -10
-20
-2
-30
0 10 20 30 40 50 60 70 80 -2 -1 0 1 2
(b) Ptest wall-4 GM-2 EQ-III

Figure 7.21 – Example responses from the dynamic analyses of PreWEC walls
7.4.6.2 Wall set Ptest

The calculated maximum lateral drift results for wall set Ptest are shown in Figure 7.22 for
hazard levels EQ-III (DBE) and EQ-IV (MCE). The plotted results were grouped into both
wall and ground motion number to assess if any trends existed. As shown in Figure 7.22a, the
average maximum drift calculated for each of the five walls correlated well with the 2% design
target drift, which is represented by the dot-dashed line. This correlation highlighted the
accuracy of the displacement based design process used to construct the wall matrix.
Additionally, as shown in Figure 7.22b, no ground motion was significantly more demanding
- 247 -
than any other, with the average maximum drift calculated for each ground motion also
correlating well with the 2% design target drift. Although some of the earthquake ground
motions could be considered to include near-fault effects, the dominant velocity pulse in each
ground motion was typically only noticeable for one of the five walls. When the frequency of
the earthquakes velocity pulse matched the fundamental period of the wall a large pulse was
observed in one direction. However, no single earthquake caused a large pulse response for all
five walls.
The maximum drift demand also matched the target drift for hazard level EQ-IV. As shown in
Figure 7.22c, the average maximum drift for each wall correlated well with the target drift of
3% for the maximum credible earthquake. As with the hazard level EQ-III analyses, strong
velocity pulses in the ground motions caused some outlier responses, in particular GM-3 for
wall 1 and GM-6 for wall 2. As shown in Figure 7.22, GM-6 was the most violent, causing
lateral drifts in excess of the 3% target for all but one wall, while GM-7 resulted in lateral drifts
that were mostly well below the 3% target.
The main objective of the analytical investigation was to investigate the residual drift that the
building exhibited following the earthquake. The calculated residual drifts for wall set Ptest are
shown in Figure 7.23 for hazard levels EQ-III and EQ-IV, and are also shown grouped into
both wall and ground motion number. For all analyses, the residual drift was below the
corresponding performance limits that were defined in section 7.3. For the design hazard level
EQ-III, the residual drift limit of 0.2% was not exceeded and the average residual drift was
approximately 0.05%. Additionally, the 0.3% residual drift limit set for the maximum credible
earthquake, corresponding to hazard level EQ-IV, was not exceeded during any of the analyses.
These results confirmed that the PreWEC system met the performance based assessment
criteria to be classified a seismic resilient or self-centering structure.
The calculated residual drift did not display any significant trend with regard to the natural
period (wall number), or the earthquake ground motion. As seen in Figure 7.23, the average
residual drift was relatively constant for both the wall number and the ground motion.
Interestingly, some of the analyses that exhibited an outlying maximum drift, that may have
been considered a consequence of a strong velocity pulse, did not result in large residual drifts.
For example, for hazard level EQ-IV, GM-3 for wall 1 and GM-6 for wall 2 both exhibited
maximum drifts in excess of 4% drift, but neither had a residual drift larger than 0.1%.
- 248 -
3 (a) Hazard level EQ-III (grouped by wall) GM-1

GM-2
Max Wall Drift (%)
2.5
GM-3
2 GM-4
GM-5
1.5
GM-6
1 GM-7
GM-8
0.5
Ave
0
1 2 3 4 5
Wall
3 (b) Hazard level EQ-III (grouped by ground motion) w all 1

w all 2
Max Wall Drift (%)
2.5
w all 3
2 w all 4
w all 5
1.5
Ave
1
0.5
0
1 2 3 4 5 6 7 8
Ground Motion
5 (c) Hazard level EQ-IV (grouped by wall) GM-1

GM-2
Max Wall Drift (%)
4
GM-3
GM-4
3
GM-5
2 GM-6
GM-7
1 GM-8
Ave
0
1 2 3 4 5
Wall
5 (d) Hazard level EQ-IV (grouped by ground motion) w all 1

w all 2
Max Wall Drift (%)
4
w all 3
w all 4
3
w all 5
2 Ave
0
1 2 3 4 5 6 7 8
Ground Motion
Figure 7.22 – Maximum drift results from the dynamic analyses of wall set Ptest
- 249 -
(a) Hazard level EQ-III (grouped by wall) GM-1

0.25
GM-2
Residual Drift (%)
0.2 GM-3
GM-4
0.15 GM-5
GM-6
0.1
GM-7
0.05 GM-8
Ave
0
1 2 3 4 5
Wall
(a) Hazard level EQ-III (grouped by wall) GM-1

0.25
GM-2
Residual Drift (%)
0.2 GM-3
GM-4
0.15 GM-5
GM-6
0.1
GM-7
0.05 GM-8
Ave
0
1 2 3 4 5
Wall
0.4
(c) Hazard level EQ-IV (grouped by wall) GM-1
GM-2
Residual Drift (%)
0.3
GM-3
GM-4
0.2 GM-5
GM-6
GM-7
0.1
GM-8
Ave
0
1 2 3 4 5
Wall
0.4
(d) Hazard level EQ-IV (grouped by ground motion) w all 1
w all 2
Residual Drift (%)
0.3
w all 3
w all 4
0.2 w all 5
Ave
0.1
0
1 2 3 4 5 6 7 8
Ground Motion
Figure 7.23 – Residual drift results from the dynamic analyses of wall set Ptest
- 250 -
7.4.6.3 Wall sets Pmore and Pless

The calculated maximum wall drift results for the wall sets with more and less connectors,
Pmore and Pless, are shown in Figure 7.24 for the design hazard level EQ-III. As expected, the
increased hysteretic energy dissipation resulted in lower lateral drift demand for walls in set
Pmore, with the average maximum drift being below the 2% design target drift for all five walls.
The opposite was observed for wall set Pless, with less energy dissipation resulting in higher
maximum lateral drifts. However, when compared to the design wall set Ptest, the difference in
the lateral displacement demand for wall sets Pmore and Pless was small, possibly because the
peak lateral displacement typically occurred during a large pulse near the beginning of the
ground motion, before the hysteretic energy dissipation was activated. This observation was
confirmed by the analysis described in section 7.4.6.6, which indicated that the elastic viscous
damping had a greater influence on the lateral displacement demand than the hysteresis area.
3
(a) Wall set Pmore GM-1
2.5
GM-2
Max Wall Drift (%)
2 GM-3
GM-4
1.5 GM-5
GM-6
1
GM-7
0.5 GM-8
Ave
0
1 2 3 4 5
Wall
3
(b) Wall set Pless GM-1
2.5
GM-2
Max Wall Drift (%)
2 GM-3
GM-4
1.5 GM-5
GM-6
1
GM-7
0.5 GM-8
Ave
0
1 2 3 4 5
Wall
Figure 7.24 – Maximum drift results for wall set Pmore and Pless, hazard level EQ-III
The calculated residual drifts for wall sets Pmore and Pless are shown in Figure 7.25, for the
design hazard level EQ-III. When compared to the design wall set Ptest, the average residual
drifts were higher for wall set Pmore and lower for wall set Pless. The increase and decrease in
- 251 -
the average residual drift was expected after observing the same trend in the cyclic hysteresis
loops that were shown previously in Figure 7.18. Two analyses from the Pmore set resulted in a
calculated residual drift that slightly exceeded the residual drift performance limit of 0.2%, and
both analyses also exhibited a high outlying lateral drift demand, as seen from Figure 7.24.
The same occurred in wall set Pless, with the analysis with the highest lateral drift demand also
slightly exceeding the 0.2% residual drift limit. An asymmetrical response was observed for
all three of the analyses reported in Figure 7.25 that exceeded the 0.2% residual drift limit.
This result provided some evidence to support the hypothesis that earthquake ground motions
with a strong velocity pulse would cause higher residual drifts.
(a) Wall set Pmore GM-1

0.25
GM-2
Residual Drift (%)
0.2 GM-3
GM-4
0.15 GM-5
GM-6
0.1
GM-7
0.05 GM-8
Ave
0
1 2 3 4 5
Wall
(b) Wall set Pless GM-1

0.25
GM-2
Residual Drift (%)
0.2 GM-3
GM-4
0.15 GM-5
GM-6
0.1
GM-7
0.05 GM-8
Ave
0
1 2 3 4 5
Wall
Figure 7.25 – Residual drift results for wall set Pmore and Pless, hazard level EQ-III
7.4.6.4 Pulse type ground motions

The influence of the asymmetrical loading response on the residual drift of PreWEC walls was
investigated further. It was difficult to identify the pulse type ground motions because as
explained earlier, no single ground motion inflicted this type of response on all five walls.
Instead, pulse type analyses were identified by the asymmetry observed in the lateral
displacement history. The difference between the magnitude of the lateral drift peaks in the
- 252 -
positive and negative load directions was used to characterise the extent of the asymmetric
response. In Figure 7.26 the residual drift is plotted against the calculated asymmetry in the
positive and negative load peaks for all analyses from wall sets Ptest, Pmore, and Pless. The data
labelled “Velocity pulse” corresponds to analyses conducted using the Chi-chi, Kobe, or Tabas
ground motions that were previously identified as containing a predominant velocity pulse.
Although there is significant scatter observed, the plotted trend lines show a positive
correlation between the residual drift and the asymmetry of the peak drifts. Symmetric
responses with the magnitude of peak drifts in each loading direction being within 0.6%
typically resulted in low residual drifts of less than 0.06%, which confirmed that an
asymmetric seismic response was more likely to result in a larger residual drift because the
shake-down effect was unable to completely re-center the wall following a large pulse.
Interestingly, the ground motions that were identified as containing a strong velocity pulse did
not result in the most asymmetric responses. However, as seen from the two trend lines plotted
in Figure 7.26, the velocity pulse ground motions typically resulted in larger residual drifts.
0.35
0.3
0.25
Residual Drift (%)
No velocity pulse
0.2
Velocity pulse
No velocity pulse (trend)
0.15
Velocity pulse (trend)
0.1
0.05
0
0 0.5 1 1.5 2 2.5 3 3.5 4
Asymmetry of Peak Drifts (%)
Figure 7.26 – Effect of asymmetrical response on residual drift for all wall sets
7.4.6.5 Residual drift ratio

To quantify the effectiveness of the shake-down effect and to develop a non-dimensional
design parameter, the residual drift ratio that was utilised by MacRae and Kawashima [7-14]
was examined. The residual drift ratio, drr, is calculated by dividing the residual drift at the
end of the analysis, dr, by the maximum possible residual drift, dr, max, which is found by
unloading slowly from the maximum lateral displacement (also referred to as maximum static
or cyclic residual drift).
- 253 -
The calculated residual drift ratios for all analyses in sets Ptest, Pmore, and Pless are plotted
against both the residual drift and maximum possible residual drift in Figure 7.27. The highest
residual drift ratio was calculated to be 0.44, which means that the shake-down effect reduced
the residual drift to less than 44% of the maximum possible residual drift. This reduction
confirmed that self-centering should not be exclusively defined from the cyclic hysteresis
loops, but that the residual drift following an earthquake is more important during design. As
expected, the lower the residual drift, the smaller the residual drift ratio, as observed in Figure
7.27a. However, no significant trend was observed between the residual drift ratio and the
maximum lateral drift, as shown by the plot in Figure 7.27b.
0.45
0.4
0.35
Residual Drift Ratio, drr
0.3 Ptest (III)
0.25 Ptest (IV)

Pmore (III)
0.2
Pmore (IV)
0.15
Pless (III)
0.1
Pless (IV)
0.05 95% confidence limit
0
0 0.05 0.1 0.15 0.2 0.25 0.3 0.35
Residual Drift, dr (%)
(a) vs. residual drift
0.45
0.4
0.35
0.3 Ptest (III)

Ptest (IV)
0.25
Pmore (III)
0.2
Pmore (IV)
0.15 Pless (III)
0.1 Pless (IV)
0
0 0.5 1 1.5
Maximum Possible Residual Drift, dr, max (%)
(b) vs. maximum possible (static) residual drift

Figure 7.27 – Residual drift ratio for wall sets Ptest, Pmore and Pless
- 254 -
The residual drift ratio can be used as an important design tool for assessing the maximum
predicted residual drift. A 95% upper bound confidence limit equal to 0.298 was established
from the residual drift ratio data, as shown in Figure 7.27. This confidence limit was rounded
to 0.3 to provide an appropriate upper limit for the residual drift ratio that could be used during
design of buildings with PreWEC walls, as discussed below in section 7.5. However, there
was significant scatter observed in the results and more in-depth investigation was conducted
in search of variables that influenced the residual drift ratio.
To investigate the relationship between fundamental period and the residual drift ratio, the
average residual drift ratio was calculated for each wall in sets Ptest, Pmore, and Pless for both
hazard level EQ-III and EQ-IV. As seen in Figure 7.28, the residual drift ratio showed no
significant trend as the fundamental period increased, with only minor inconsistent variations
observed. The residual drift ratio showed a peak for wall 4 during hazard level EQ-IV, but this
was considered a random event with no obvious global trend. Additionally, no significant
difference was observed between the analysis sets with different hysteretic energy dissipation.
Wall Wall
1 2 3 4 5 1 2 3 4 5
0.35 0.35
Ptest (ave) Ptest (ave)
0.3 0.3
Pmore (ave) Pmore (ave)
Pless (ave) Pless (ave)

0.25 0.25
0.2 0.2
0.15 0.15
0.1 0.1
0.05 0.05
0 0
0 0.5 1 1.5 2 2.5 3 3.5 0 0.5 1 1.5 2 2.5 3 3.5
Fundamental Period, Ti (s) Fundamental Period, Ti (s)
(a) Hazard level EQ-III (b) Hazard level EQ-IV

Figure 7.28 – Influence of fundamental period on the residual drift ratio
The influence of the ground motion parameters on the residual drift ratio was also investigated.
The residual drift ratio for all analyses in sets Ptest, Pmore, and Pless are plotted in Figure 7.29
against the ground motion peak ground acceleration, peak velocity, and distance from the
source. Both the scaled peak ground acceleration and peak ground velocity showed large
- 255 -
scatter, with no significant correlation observed with the residual drift ratio. The average
residual drift ratio for each ground motion also indicated no significant correlation with the
distance from the earthquake source. Additionally, in Figure 7.29d the analyses that were
conducted using ground motions with a strong velocity pulse (Chi-chi, Kobe and Tabas) are
separated from those with no significant velocity pulse. It can be seen that there was also no
significant difference in the calculated residual drift ratio between ground motions with or
without a predominant velocity pulse.
0.45 0.45
Ptest Ptest
0.4 0.4
Pmore Pmore
0.35 0.35
Pless Residual Drift Ratio, drr Pless

0.3 0.3
0.25 0.25
0.2 0.2
0.15 0.15
0.1 0.1
0.05 0.05
0 0
0 0.5 1 1.5 2 0 0.5 1 1.5 2
Scaled Peak Ground Acceleration (g) Scaled Peak Ground Velocity (m/s)
(a) Peak ground acceleration (b) Peak ground velocity

0.35 0.45
Ptest (ave) No velocity pulse
0.4
0.3 Velocity pulse
Pmore (ave)
0.35
0.25 Pless (ave)

0.3
0.2 0.25
0.15 0.2
0.15
0.1
0.1
0.05
0.05
0 0
0 5 10 15 20 25 30 0 0.5 1 1.5
Distance from Source (km) Maximum Possible Residual Drift, dr, max (%)
(c) Distance from source (d) Velocity pulse ground motions

Figure 7.29 – Influence of ground motion parameters on the residual drift ratio
- 256 -
7.4.6.6 Assumed viscous damping

As explained earlier, the analyses for wall set Ptest were repeated with 1% and 5% viscous
damping to assess the effect of the assumed elastic viscous damping. The calculated maximum
drift results for the analyses with 1% and 5% damping are shown in Figure 7.30 for the design
hazard level EQ-III. As expected, when 1% damping was used for the analyses, the average
maximum drifts for each wall were higher than the 2% design target, while the average
maximum drifts for the 5% damping analyses were below the 2% design target. Additionally,
the reduced 1% elastic damping resulted in some high outlier maximum drift demands during
pulse type excitation. These outliers can be seen in Figure 7.30a, where many analyses
approach 3% drift and one analysis for wall 2 reached over 3.5% lateral drift, which was 75%
higher than the 2% design target. The outliers highlighted the importance of the viscous
damping in reducing lateral displacements, especially during ground motions that have a strong
velocity pulse.
(a) Maximum wall drift 1% damping GM-1

3 GM-2
Max Wall Drift (%)
GM-3
GM-4
2
GM-5
GM-6
1 GM-7
GM-8
Ave
0
1 2 3 4 5
Wall
(b) Maximum wall drift 5% damping GM-1

3 GM-2
Max Wall Drift (%)
GM-3
GM-4
2
GM-5
GM-6
1 GM-7
GM-8
Ave
0
1 2 3 4 5
Wall
Figure 7.30 – Maximum drift results for wall set Ptest with 1% and 5% viscous damping,
EQ-III
- 257 -
A comparison of the average residual drift calculated for each wall with 1%, 3% and 5%
viscous damping is shown in Figure 7.31 for the design hazard level EQ-III. The analyses with
1% damping resulted in higher residual drifts than the 3% design analyses, while the residual
drifts for the 5% damping analyses were below that of the 3% analyses. This trend in the
residual drift behaviour was primarily attributed to the difference observed in the maximum
drift demand.
0.08
0.07
Average Residual Drift (%)
0.06
0.05
0.04
0.03
0.02
1% damping
0.01 3% damping
5% damping
0
1 2 3 4 5
Wall
Figure 7.31 – Residual drift results for wall set Ptest with 1% and 5% viscous damping,
EQ-III
Most importantly, the effect of the elastic damping selection on the residual drift ratio was
investigated. The calculated residual drift ratio for all analyses conducted with 1% and 5%
viscous damping are shown in Figure 7.32. A similar trend was observed to the design
analyses conducted with 3% viscous damping, with all calculated residual drift ratios being
below 0.40. The 95% confidence limit for the 1% and 5% damping analyses was calculated to
be 0.281, as shown in Figure 7.32. This calculated confidence limit was just below the 0.298
limit that was calculated for the 3% damping analyses. It was concluded that the assumed
elastic damping did not appear to have any significant impact on the residual drift ratio, which
provided confidence in the use of an average residual drift ratio limit for the design of self-
centering systems regardless of the elastic or hysteretic damping in the building.
- 258 -
7.5 Design Procedure
0.4
0.35
0.3
1% damping (III)
0.25
1% damping (IV)
0.2 5% damping (III)
5% damping (IV)
0.1
0.05
0
0 0.5 1 1.5
Maximum Possible Residual Drift, dr, max (%)
Figure 7.32 – Residual drift ratio for wall set Ptest with 1% and 5% viscous damping
7.5 DESIGN PROCEDURE

After analysing the results from the dynamic analysis described above, a procedure for a
simple design check was established to include a residual drift limit in performance based
design of self-centering systems. The design check included the following steps:
 Step 1: Establish suitable residual drift performance limits, [for PreWEC and other
self-centering systems the suggested limits are 0.2% for the design hazard level (EQ-
III) and 0.3% for the maximum credible hazard level (EQ-IV)].
 Step 2: Complete preliminary design of the structure using force-based or
displacement-based design.
 Step 3: Estimate the cyclic hysteresis behaviour of the system using either
experimental, finite element, or simplified analysis responses.
 Step 4: From the assumed cyclic hysteresis behaviour, estimate the maximum possible
residual drift, dr, max, corresponding to the design target drift at each performance level.
 Step 5: Estimate the upper bound residual drift ratio, drr, which is equal to 0.3 based on
the dynamic analyses of the PreWEC system.
 Step 6: Multiply dr, max by drr to calculate the upper limit of the design residual drift,
dr, design.
 Step 7: Check if dr, design is below the residual drift performance limit for each hazard
level. If not, revise design accordingly.
- 259 -
7.6 IMPLICATIONS TO OTHER SYSTEMS

Self-centering systems can be divided into two categories based on the behaviour of the
compression toe and the level of damage expected:
 Systems where the compression toe of the PT member remains essentially elastic, with
no visible damage.
 Systems where significant inelastic strains are expected in the compression toe of the
PT member.
If the compression toe remains essentially elastic, a multi-linear elastic hysteresis rule would be
sufficient to model the PT member, and self-centering can be easily ensured by checking that
that the static residual drift observed from the cyclic hysteresis loops was below the target
performance limits.
When significant inelastic strain are expected in the compression toe, the true hysteresis
behaviour and resulting static residual drift need to be considered when modelling the cyclic
response of the PT member. If the resulting static residual drift observed from the cyclic
hysteresis loops is above the target performance limits, either a known residual drift ratio, drr,
or dynamic analysis would be required to ensure that the structure was adequately designed to
self-center during a possible earthquake at the geographic location of the structure.
For self-centering concrete members, it is challenging to determining at what level the strains
in the compression toe become significant enough that the member requires more detailed
hysteresis modelling. Resptrepo and Rahman [7-8] suggested that the initial axial load on PT
walls should be kept below 0.15 fc' to prevent excessive damage and possible bucking of the
wall toe. However, the initial axial load on the PreWEC test wall was only 0.12 fc' and yet
inelastic strains in the wall toe still caused significant deviation from a multi-linear elastic
hysteresis response. The inclusion of a residual drift check in the design process relied heavily
on the estimation of the cyclic hysteresis response. For this investigation, the experimental
response was used to predict the static residual drift behaviour. Further investigation into the
calculation of the static residual drift would greatly simplify the self-centering design process.
Although the dynamic analysis concentrated on the PreWEC system, it is considered that the
hysteresis behaviour is typical of realistic self-centering wall and frame members that
- 260 -
7.7 Conclusions
experience limited damage in the compression toe region. For instance, similarly shaped
hysteresis behaviour was observed for the jointed wall system during the 5 storey PRESSS
building test [7-7]. Further investigation would determine whether the upper bound residual
drift ratio limit of 0.3 was also applicable to other types of self-centering systems.
7.7 CONCLUSIONS
Although the term self-centering is typically applied to any structure designed with unbonded
PT, limited research has been conducted to investigate how self-centering is defined and the
magnitude of residual drifts that occur in realistic concrete members with unbonded PT. This
chapter examined the validity of current design procedures, and a parametric study was
preformed to investigate dynamic self-centering behaviour using the PreWEC system as an
example.
The current procedures that are used to ensure that self-centering occurs in structures with a
combination of unbonded PT and energy dissipating elements are inaccurate and do not
account for the dynamic self-centering behaviour that exists when the structure is subjected to
an earthquake. Using a two-spring lumped plasticity model, a series of simple analyses
showed that prefect self-centering could still be observed from the cyclic load response even
when the energy dissipating elements contribute to over 50% of the total moment resistance.
Additionally, the analysis showed that even when the PT wall response followed a multi-linear
elastic response, a small residual drift can occur when the moment contributions from the PT
and energy dissipating elements are balanced.
Ensuring that self-centering will occur in real self-centering systems, such as PreWEC, is
difficult because inelastic strains in the compression toe lead to a response that includes
additional hysteretic energy dissipation, stiffness degradation and residual drift. Simplified
techniques cannot be used to ensure perfect self-centering behaviour for realistic self-centering
concrete structures.
A lumped plasticity model was developed using a total of seven rotational springs to accurately
capture the PreWEC behaviour observed during the experimental test. Unlike previous
modelling, the developed model accurately accounted for the wall hysteresis, stiffness
degradation and the residual drift that was observed from the cyclic hysteresis loops.
- 261 -
The developed model was used to conduct an extensive parametric study to investigate the
design and residual drift of the PreWEC system when subjected to earthquake ground motions.
The results of the dynamic analyses showed that the PreWEC system can be accurately
designed using direct displacement based design with an assumed 3% viscous damping. The
average maximum drift for each of the walls in the design set closely matched the target 2%
design level drift and 3% maximum allowable drift levels.
The residual drifts at the conclusion of the dynamic analyses were much lower than the
maximum possible residual drift observed from the cyclic hysteresis loops. As a result, the
PreWEC system was also found to meet performance based residual drift limits that could be
used to define realistic self-centering behaviour.
A residual drift ratio was established to define the dynamic shake-down behaviour. No
significant correlation was observed between the residual drift ratio and the fundamental
period, the ground motion parameters, or the assumed viscous damping. The lack of obvious
trends implied that the upper bound confidence limit of 0.3 could be used during the design
process for all types of PreWEC structures subjected to any earthquake ground motion.
A simplified design process was developed to incorporate a check on the estimated residual
drift during the design process. Although the analysis was performed using the PreWEC
system as an example, the results are also applicable to other self-centering systems that have
similar hysteresis behaviour.
7.8 REFERENCES
[7-1] Stanton, J., Stone, W. C., and Cheok, G. S. (1997) A hybrid reinforced precast frame
for seismic regions. PCI Journal, 42(2), 20-32.
[7-2] Palermo, A., Pampanin, S., and Marriott, D. (2007) Design, modeling, and
experimental response of seismic resistant bridge piers with posttensioned dissipating
connections. Journal of Structural Engineering, 133(11), 1648-1661.
[7-3] Christopoulos, C., Filiatrault, A., and Folz, B. (2002) Seismic response of self-centring
hysteretic SDOF systems. Earthquake Engineering and Structural Dynamics, 31(5),
1131-1150.
- 262 -
7.8 References
[7-4] Newcombe, M. P., Pampanin, S., Buchanan, A., and Palermo, A. (2008) Section
analysis and cyclic behavior of post-tensioned jointed ductile connections for multi-
story timber buildings. Journal of Earthquake Engineering, 12(SUPPL. 1), 83-110.
[7-5] Palmieri, L., Sagan, E., French, C., and Kreger, M. (1997) Ductile connections for
precast concrete frame systems. Mete A. Sozen Symposium, ACI SP 162, American
Concrete Institute, Farmington Hills, MI, 313-355.
[7-6] Perez, F. J., Sause, R., and Lu, L. W. (2003) Lateral load tests of unbonded post-
tensioned precast concrete walls. ACI Special Publication, 211, 161-182.
Journal, 44(6), 42-67.

1560-1570.
New Zealand.
[7-10] Pampanin, S., Marriott, D., and Palermo, A. (2010) PRESSS design handbook. New
Zealand Concrete Society, Auckland, N.Z.
[7-12] Kurama, Y. C. (2002) Hybrid post-tensioned precast concrete walls for use in seismic
regions. PCI Journal, 47(5), 36-59.
[7-14] Macrae, G. A. and Kawashima, K. (1997) Post-earthquake residual displacements of

bilinear oscillators. Earthquake Engineering and Structural Dynamics, 26(7), 701-716.
[7-15] Christopoulos, C., Pampanin, S., and Priestley, M. J. N. (2003) Performance-based

seismic response of frame structures including residual deformations. Part I: Single-
[7-16] Carr, A. RUAUMOKO - Inelastic Dynamic Analysis Program. University of

Canterbury, Christchurch, New Zealand, 2003.
[7-17] Priestley, M. J. N. (2000) Performance based seismic design. Bulletin of the New
Zealand Society for Earthquake Engineering, 33(3), 325-346.
- 263 -
[7-18] Kawashima, K. (1997) The 1996 Japanese seismic design specifications of highway
bridges and the performance based design. Seismic design methodologies for the next
generation of codes, ed. P. Fajfar and Krawinkler, H. Balkema, Rotterdam/Brookfield.
[7-19] Japan Road Association (JRA). Design specification of highway bridges, Part V
Seismic design. 1996.
[7-20] Seismology Committee. Recommended lateral force requirements and commentary

(Blue book). Structural Engineers Association of California (SEAOC), 1999.
[7-21] Rahman, M. A. and Sritharan, S. (2006) An evaluation of force-based design vs. direct
displacement-based design of jointed precast post-tensioned wall systems. Earthquake
Engineering and Engineering Vibration, 5(2), 285-296.
[7-22] Rahman, M. A. and Sritharan, S. (2007) Performance-based seismic evaluation of two

five-story precast concrete hybrid frame buildings. Journal of Structural Engineering,
133(11), 1489-1500.
[7-23] Priestley, M. J. N., Calvi, G. M., and Kowalsky, M. J. (2007) Displacement-based

seismic design of structures. IUSS Press.
[7-24] Priestley, M. J. N. (2002) Direct displacement-based design of precast/prestressed

concrete buildings. PCI Journal, 47(6), 66-79.
[7-25] Pennucci, D., Calvi, G. M., and Sullivan, T. J. (2009) Displacement-based design of
precast walls with additional dampers. Journal of Earthquake Engineering, 13(1
SUPPL. 1), 40-65.
[7-26] Wight, G. D., Kowalsky, M. J., and Ingham, J. M. (2004) Shake table testing of post-
tensioned concrete masonry walls. RD-04-04, North Carolina State University, Raleigh,
NC, USA.
[7-27] Ma, Q. T. (2010) The mechanics of rocking structures subjected to ground motions.
PhD thesis. Dept. of Civil and Environmental Engineering, University of Auckland,
New Zealand.
[7-28] Kurama, Y. C., Sause, R., Pessiki, S., and Lu, L.-W. (2002) Seismic response
evaluation of unbonded post-tensioned precast walls. ACI Structural Journal, 99(5),
641-651.
[7-29] Cox, K. E. and Ashford, S. A. (2002) Characterization of large velocity pulses for
laboratory testing. PEER 2002/22, Pacific Earthquake Engineering Research Center,
Berkeley, CA.

- 264 -
Chapter 8
Analysis of a Self-centering Building
In the previous chapters, the behaviour of the self-centering wall system was analysed in
isolation. To fully appreciate the expected behaviour during an earthquake and ensure that
self-centering is achieved, the response of the entire structure needs to be considered.
Although concrete shear walls are often designed as the primary lateral load resisting system,
the wall still interacts with the surrounding structure, including other gravity load elements,
which can potentially influence the wall response. For example, Waugh and Sritharan [8-1]
found that the inclusion of the floor diaphragms and the framing action of the gravity columns
was critical in order to accurately predict the dynamic response of a seven storey reinforced
concrete building. Additionally, there is no advantage in designing a seismic resilient wall
system if extensive damage would be expected in other parts of the structure, or if the wall was
unable to self-center the entire building following an earthquake.
This chapter investigates how the interaction between self-centering wall systems and the
surrounding structure affects the building’s seismic performance, focusing in particular on the
wall-to-floor interaction. Typical types of wall-to-floor connections were identified and their
suitability for use with self-centering wall systems is discussed. The PreWEC finite element
model that was developed in Chapter 6 was extended to investigate the behaviour of the floor
- 265 -
Chapter 8 Analysis of a Self-centering Building
diaphragms in a prototype building for two extreme connection details, being a rigid cast-in-
place wall-to-floor connection and a fully isolated wall-to-floor connection. Using the wall-to-
floor connection response calculated from the finite element model, a series of dynamic time-
history analyses were conducted to compare the seismic response of the prototype building that
utilised a PreWEC wall system and different wall-to-floor connection types.
8.1 INTRODUCTION
A critical consideration when examining the seismic response of a self-centering building is
how the wall system interacts with the floor diaphragms. When subjected to a lateral load, the
flexural deformation of a self-centering wall is concentrated at a single crack that opens up at
the base of the wall, as shown in Figure 8.1. The uplift at the base of the wall causes a vertical
displacement and rotation at the wall-to-floor connection. During conventional design
practice, the floors are assumed to act as rigid diaphragm elements and the effect of this
displacement incompatibility is ignored. To ensure that a seismically resilient building is
achieved, the vertical displacement of the wall needs to be accounted for when designing both
the floor diaphragms and the wall system. Note that the displacement incompatibility between
the wall and floor is not limited to self-centering precast wall systems, and also affects
traditional systems including reinforced concrete (RC) walls. It is likely that the vertical
displacement at the wall-to-floor connection would be greater for a RC wall, where the flexural
deformation and cracking of the wall is spread over a plastic hinge region that experiences
progressive damage and incomplete crack closure.
PT tendon
Wall
uplift
Foundation
Figure 8.1 – Lateral load behaviour of a PT wall
- 266 -
8.2 Wall-to-Floor Connection Details
The robustness of the wall-to-floor connection is critical to ensure that both seismic and gravity
load paths are maintained. During an earthquake, the inertia forces that are generated by the
buildings mass must be transferred from the floor diaphragms to the lateral load resisting shear
walls. Additionally, if the shear wall is also designed as a bearing wall it is required to transfer
gravity loads from the floors to the foundation. However, as detailed in the literature review in
Chapter 2, little research has been conducted to investigate the seismic behaviour of wall-to-
floor connections, especially for self-centering precast concrete systems. Primarily researchers
have focused on the development, testing, design, and analysis of self-centering wall systems,
without considering how these walls interact with the surrounding structure (as was the case
for the PreWEC system that was discussed and analysed in Chapters 5 and 6). In the few
projects that involved experimental testing of a building system that utilised self-centering
precast concrete members, such as the PRESSS five storey building [8-2] and the precast
building tested on the UCSD shake table [8-3], the wall-to-floor connections were isolated in
order to minimise the uncertainty in the behaviour due to wall uplift.
Standards that outline the seismic design of self-centering precast concrete wall systems in the
United States [8-4, 8-5] and New Zealand [8-6, 8-7] primarily focus on the wall component
behaviour, and merely state that the interaction between structural elements, including wall and
floor diaphragms, needs to be considered during design.
8.2 WALL-TO-FLOOR CONNECTION DETAILS

The behaviour of the wall-to-floor interface is highly dependent on the connection detail. The
connection type determines how the lateral inertia forces from the diaphragm are transferred to
the wall, as well as how the floor diaphragms respond to the vertical uplift of the wall.
Typically the wall-to-floor connection is critical to the lateral load resistance path and thus
capacity design principles should be used to ensure that the connection does not fail
prematurely. Several types of wall-to-floor connection were identified based on the type and
orientation of the floor system.
8.2.1 Cast-in-Place Floor

A commonly used detail is a reinforced concrete cast-in-place floor slab that is fully tied into
the wall system, as shown in Figure 8.2. The cast-in-place wall-to-floor connection would be
- 267 -
constructed using reinforcing bars that extend from the precast wall into the floor, providing a
moment resisting connection. The cast-in-place connection detail is easy to construct, but the
floor diaphragm would be constrained to deform with the wall system. The uplift occurring at
one end of the wall would push the floor slab up, with the rigid wall-to-floor connection
providing resistance. A concern with the cast-in-place detail is that it can cause damage to the
floor slab around the wall and possibly reduce the ability of the connection to transfer lateral
and gravity loads from the floor to the wall. Note that the behaviour and possible damage to
the cast-in-place connection would be highly dependent on the floor layout and the boundary
conditions provided by the frame elements at the edge of the floor diaphragm.
floor
wall
Figure 8.2 − Cast-in-place floor
8.2.2 Precast Floor Units Parallel to the Wall

When precast floor units are used, the connection type depends on the orientation of the precast
units and the load paths that are required. If the precast floor unit spans parallel to the length of
the wall, as shown in Figure 8.3, then the concrete shear wall may be designed to not carry any
of the gravity load from the floor area. However, the seismic inertia forces generated in the
floor diaphragm must be transferred to the lateral load resisting shear wall.
If the precast floor unit was connected to the wall using a welded detail, as shown in Figure
8.4, the relatively stiff precast floor unit would resist the uplift and rotation of the precast shear
wall, which may have a detrimental effect on the wall behaviour. If the wall is restrained from
- 268 -
uplift, the increased lateral resistance could result in shear sliding failure at the wall base.
Additionally, the welded connections that are shown in Figure 8.4 would not have adequate
ductility to allow for relative movement between the wall and floor. It is likely that a welded
wall-to-floor connection would fail during a large earthquake, resulting in loss in the lateral
resistance load path and potential catastrophic failure.
Figure 8.3 − Precast floor unit parallel to the wall
Figure 8.4 – Floor to shear wall connections detailed in PCI design handbook [8-8]
- 269 -
An alternative to a rigid wall-to-floor connection is to isolate the floor system from vertical
uplift of the wall, as was done during the precast parking structure test at USCD [8-3]. If the
wall is not required to carry vertical gravity loads from the floor diaphragms, the wall-to-floor
connection can be designed to transfer horizontal inertia loads with unrestrained displacement
in the vertical direction. This isolated wall-to-floor connection could be achieved by using a
slotted connector, such as the PSA slotted insert designed by JVI Inc. [http://www.jvi-
inc.com], which is shown in Figure 8.5. The slotted connection would allow the wall and floor
to be constrained in the horizontal direction while leaving the wall free to uplift in the vertical
direction. This solution could reduce the damage caused to the floor system by allowing the
floor to remain in an undeformed flat position. In Figure 8.5 the slotted connector was placed
on the top of the precast unit for clarity, but in reality the connectors could be hidden by
placing them on the bottom side of the precast floor unit.
(b) Close-up of slotted connector
(a) Slotted connectors placed between wall and floor (c) PSA slotted insert by JVI Inc.
[http://www.jvi-inc.com]
Figure 8.5 – Possible slotted wall-to-floor connection to isolate the floor from the wall
uplift
- 270 -
8.2.3 Precast Floor Units Bearing on Wall

When the precast floor units are orientated perpendicular to the length of the wall, as shown in
Figure 8.6, the wall is required to carry both vertical gravity loads and horizontal seismic
inertia forces. Because the wall is relied on for the gravity load path, the slotted connection
that was described above cannot be used. Typical detailing of a bearing connection would
include the precast flooring seated on a corbel attached to the wall, as shown in Figure 8.7
which shows different wall-to-floor connection details for bearing walls. When an insitu
concrete topping is used, the topping reinforcement would be tied into the wall panel, and for
pretopped floor units a welded steel plate would connect the top of the precast floor unit to the
precast wall, as seen in the center images in Figure 8.7.
Figure 8.6 − Precast floor with insitu topping bearing on the wall
When considering the seismic behaviour of the wall-to-floor connection with precast floor
units bearing on the self-centering wall, the expected behaviour would be somewhere between
the rigid cast-in-place connection and the isolated slotted connection. The typical connection
details which are shown in Figure 8.7 would provide partial moment resistance and thus
behave differently to a full cast-in-place connection. It is critical to consider the connection
between the precast units and the end seating detail in order to ensure that a desirable seismic
response is expected when uplift and rotation of the wall panel occurs.
- 271 -
Figure 8.7 – Floor to bearing wall connections detailed in PCI design handbook [8-8]
8.2.4 Possible Connections for PreWEC Systems

When compared to a single precast concrete panel, the arrangement of the PreWEC system can
offer unique solutions for wall-to-floor connections. In the PreWEC system, the column uplift
is negligible when compared to the wall uplift, and so the floor diaphragms can be attached to
the columns to isolate the floor from the vertical displacement of the wall. Two possible
isolating wall-to-floor connections that could used in conjunction with the PreWEC wall
system are shown in Figure 8.8.
Precast floor units spanning parallel to the wall length could be attached to the columns using a
hinged connection that allows rotation, as shown in Figure 8.8a. Because the uplift of the
PreWEC end columns is negligible, the floor will not be subjected to any significant vertical
- 272 -
displacement. However, inertia forces must be transferred from the floor diaphragm into the
lateral force resisting shear wall, which could be achieved by attaching pinned struts between
the column and wall, similar to that used during the large-scale PreWEC test [8-9, 8-10].
(a) Parallel precast flooring
(b) Precast flooring bearing on wall

Figure 8.8 – Possible isolating wall-to-floor connections suitable for PreWEC wall system
- 273 -
Alternatively, perpendicular precast floor units could be seated on a steel angle that was only
attached at each end to the PreWEC end columns, as shown in Figure 8.8b. The gravity load
from the floor area would be transferred to the column elements, and the seismic inertia forces
could again be transferred to the wall using pinned struts attaching the column to the wall.
8.3 PROTOTYPE BUILDING

A prototype structure was designed to assess the seismic response of a building that included a
self-centering wall system, with particular focus on the interaction with the floor diaphragms.
The prototype structure was based on the four storey building design that was used during the
dynamic analysis in Chapter 7, and the layout of the prototype building is shown in Figure 8.9.
The inter-storey height used previously was 3.66 m, which resulted in a total building height of
14.64 m. The lateral resistance in the N-S direction relied exclusively on two shear walls, with
the lateral resistance of the concrete perimeter frame ignored. In the E-W the lateral resistance
was provided by two moment resisting frames, which are highlighted in black. As with the
building design used in Chapter 7, the shear walls were designed using PreWEC walls with the
same cross-sectional dimensions and moment resistance as the full scale equivalent of the
PreWEC test specimen reported in Chapters 5 and 6. From Chapter 7, the equivalent weight,
We, of the four-storey building, which was designed using direct displacement based design,
was 18250 kN, which equated to a weight of 5475 kN per floor for each wall, or a factored
dead and live load of 11.4 kPa for the building layout shown in Figure 8.9.
48m
12m
Cast-in-place slab Double T Hollow core
20m
Wall Wall
W E
S Moment resisting frame
Figure 8.9 – Four-storey prototype building layout
- 274 -
8.3 Prototype Building
Based on preliminary design calculations, the concrete columns in the prototype building were
assumed to be 500 mm square, and the beams were assumed to be 400 mm wide and 600 mm
deep. Also, except for the PreWEC wall, the concrete compressive strength was assumed to be
30 MPa. As shown in Figure 8.9, the prototype building was designed with various possible
options for the floor diaphragms, including a fully cast-in-place floor slab, precast double-T
units spanning N-S and precast hollowcore units spanning E-W. The details of each of these
floor types are discussed below.

The cast-in-place slab was designed to represent a rigid wall-to-floor connection, where the
floor slab would be subjected to the full vertical displacement and rotation of the PreWEC
wall. The cast-in-place floor slab was attached directly to the PreWEC wall panel, and block-
outs were provided around the end columns of the PreWEC wall to constrain the floor
movement to that of the wall, while still allowing the relative vertical displacement between
the wall and end column to occur.
Based on preliminary design calculations, a slab thickness of 250 mm was selected using
concrete with a compressive strength of 30 MPa. The slab reinforcement consisted of
Grade 500 mild steel reinforcing bars, with a yield strength of 500 MPa. In the longitudinal N-
S direction, 20 mm diameter reinforcing bars (DH-20) were placed at 300 mm centers at both
the top and bottom of the slab, and 10 mm diameter reinforcing bars (DH-10) were placed at
300 mm centers at both the top and bottom of the slab in the transverse E-W direction.
8.3.2 Precast Floor Units

The precast double-T units were designed to span between the perimeter beams with no
intermediate support, and so the PreWEC wall would not be required as a gravity load path.
The double-T units would most likely be pre-topped and the wall-to-floor connection could be
detailed using a slotted connector to isolate the floor diaphragm from the vertical displacement
of the wall, as shown earlier in Figure 8.5.
The precast hollow-core units were designed to span 12 m to an intermediate beam in line with
the PreWEC wall. The wall-to-floor connection detail could be either a semi-rigid connection,
with the hollow-core seated on a corbel and the toping concrete tied into the precast concrete
- 275 -
wall, or an isolated connection with the hollowcore seated on an angle attached to the PreWEC
end columns, as shown in Figure 8.8.
When considering a fully isolated wall-to-floor connection, the difference in behaviour

between different precast floor and connection types would be negligible. For the purpose of
the analysis, the behaviour of an ideal isolated wall-to-floor connection was examined, which
could represent either the double-T or hollow-core isolated connections. The behaviour of the
partially-rigid hollow-core connection was not examined during this study.

Initial investigation of the lateral load response of the prototype building and the wall-to-floor
interaction was carried out analytically using a detailed finite element model (FEM). An FEM
was constructed for one half of the 4 storey prototype building described above, with both a
rigid and an isolated wall-to-floor connection modelled. Following the building model, a
single storey floor and wall section was modelled to quantify the moment rotation behaviour of
the rigid wall-to-floor connection.
8.4.1 Description of Model

The model of the prototype building was an extension of the PreWEC FEM that was developed
in Chapter 6, with the floors and frame elements added around the wall. Unless stated
otherwise, all elements and material properties of the PreWEC wall were modelled as
described previously in Chapter 6.
8.4.1.1 Material properties

The concrete that was used for the floor and frame elements of the building was defined in a
similar manner to the “concrete damaged plasticity” model used for the PreWEC FEM. The
stress-strain definition was derived for an assumed 30 MPa compressive strength (fc') using
Eq. 6.5, and a maximum tensile stress of 0.62 f c' (MPa). As before, the slope of the strain
softening section was reduced, ending at 0.1 fc' at a compressive strain of 0.01. Because the
cast-in-place floor was expected to crack, a realistic tensile strain softening slope was defined,
decreasing to 0.1 MPa at a strain of 0.0005. The concrete stress-strain definition that was input
into the FEM for the floor and frame elements is shown in Figure 8.10a.
- 276 -
700
-30
600
-25
500
-20
Stress (MPa)
Stress (MPa)
400
-15
300
-10
200
-5
0 100
5 0
0 -0.003 -0.006 -0.009 0 0.02 0.04 0.06 0.08 0.1
(a) Concrete (b) Reinforcing bar

Figure 8.10 – Material stress-strain definitions used in floor FEM
The reinforcing steel in the cast-in-place floor slab was defined using a bi-linear stress-strain
response, with an assumed modulus of elasticity of 200 GPa, yield strength of 500 MPa, and
ultimate strength of 650 MPa at a strain of 0.1, as shown in Figure 8.10b.
8.4.1.2 FEM construction

As stated previously, the PreWEC walls that were designed for the prototype building had the
same cross-sectional properties and moment-rotation behaviour as the full scale version of the
PreWEC test specimen. Therefore, the FEM of the PreWEC wall was constructed as was
described in detail in Chapter 6, except that the dimensions were scaled to full size. To achieve
the correct height for the four-storey building, the height of the PreWEC wall had to be
extended to 14.8 m. However, to achieve the same moment-rotation behaviour at the wall
base, the unbonded tendon length remained the same as the test specimen (15.8 m at full scale).
Due to symmetry in the building layout, only one PreWEC wall and one half of the floor area
was modelled. The four 20 m long by 24 m wide floor diaphragms were modelled using plane-
stress shell elements, with the concrete damaged plasticity model that was described above.
Cut-outs in the floor were provided around the end columns of the PreWEC wall because the
two elements would not be connected in the structure. The floor used an approximately
180 mm square mesh with seven integration points through the 250 mm slab depth. Both the
mesh size and the number of integration points were selected following a mesh sensitivity
- 277 -
study, which found that further reducing the mesh size or increasing the number of integration
points had no significant affect on the model accuracy.
The meshed assembly of the prototype building FEM is shown in Figure 8.11. The beams
were modelled as wire beam elements and the columns modelled as truss elements, all with
gross cross-sectional properties. All of the beam-column joints were idealised as pinned
connections, and the edges of the floor diaphragm were tied to the adjacent beam edge. The
base of the columns were also idealised as pinned connections, with the displacement
restrained in all directions. Although the pinned joints ignored the moment contribution
provided by the gravity frame members, framing action between the columns at each end of the
floor slab was considered to be critical in assessing the moment contribution provided by the
entire gravity frame system, as found during previous analysis by Waugh and Sritharan [8-1].
Figure 8.11 – FEM assembly of 4-storey prototype building
The application of an inverse triangular load pattern was difficult to implement into the FEM
without significant convergence problems, and so the lateral load behaviour of the prototype
building FEM was instead determined by applying a monotonic lateral displacement to the top
of the wall. Because the wall flexural deformation is primarily concentrated at the base, it was
considered that the single lateral load applied at the top of the wall was adequate to accurately
capture the displaced profile of the building. As with the PreWEC FEM described in
- 278 -
Chapter 6, the explicit solver in ABAQUS was used with a dynamic analysis step, and a slow
loading speed of 7.4 mm/s was used to ensure that a pseudo-static response was achieved.

The first analysis was conducted to simulate the behaviour of a cast-in-place floor slab with a
rigid wall-to-floor connection.
8.4.2.1 Connection modelling

The floor diaphragm shell element extended all the way through the concrete wall, and the
cast-in-place connection was modelled using an embedded constraint to couple the floor nodes
within the connection region to adjacent nodes of the wall. This constraint created a rigid
connection, where the anchorage and strain penetration of the connection reinforcement was
ignored for simplicity.
8.4.2.2 Results
The deformed shape of the building FEM at 3% lateral wall drift is shown in Figure 8.12, with
the deformations magnified three times. The PreWEC wall behaved in the same manner as the
previous FEM of the individual PreWEC specimen, with deformation primarily concentrated at
the single crack at the wall base. The rigid behaviour of the cast-in-place wall-to-floor
connections is visible, with the floors constrained to uplift and rotate with the wall. Because
the edges of the floor were constrained to the gravity frame, which did not experience
significant vertical displacement, the uplift and rotation at the wall-to-floor connection caused
bending of the floor in the out-of-plane direction. Because the deformation of the floor is
controlled by the magnitude of wall uplift and rotation and the distance to the perimeter gravity
frame, the behaviour is highly dependent on the wall length and the building layout.
Several parameters were examined to quantify the extent of damage that would result in a cast-
in-place floor slab. Figure 8.13 plots the maximum principal strains that were calculated for
the concrete in the building FEM at 3% wall drift, where tensile strains that exceed the
concrete cracking strain of 0.000138 are represented by the grey region. It can be seen that the
area of the floor that was predicted to undergo cracking is extensive, and because the wall
deformation was primarily concentrated at the wall base, the subsequent uplift and rotation at
each floor level is similar, which causes comparable damage for each floor. However, the top
- 279 -
floor experienced a slightly larger uplift due to elastic deformation of the wall panel, and so the
maximum strains occurred near the wall-to-floor connection on the top floor.
Figure 8.12 – Elevation view showing the calculated displaced shape of the prototype
building FEM with cast-in-place floor (displacements magnified 3 times)
Figure 8.13 – Calculated principal strain contours of the prototype building FEM with
cast-in-place floor at 3% lateral drift
- 280 -
An elevation view of the concrete strains in the top surface of the top floor is shown in Figure
8.14, which plots the principal strain contours alongside vectors representing the magnitude
and direction of the maximum principal strain. At 2% lateral wall drift, which corresponds to
the design level displacement, the area of the cracked concrete is already significant, and at the
maximum considered displacement of 3% lateral drift, the cracked concrete accounts for over
45% of the total floor area. Also, it should be noted that these strains were developed during
monotonic loading, and that almost the entire floor area would be expected to be cracked
during reverse cyclic loading. Additionally, the strain vectors indicate the cracking pattern on
the floor slab, with larger cracks concentrated along yield planes that spread out from the end
of the wall-to-floor connection. However, strains of a large magnitude are confined to a small
region directly adjacent to the end of the wall and around the cut-out provided for the PreWEC
end column.
(a) 2% lateral drift (b) 3% lateral drift

Figure 8.14 – Calculated principal strain contours of the top cast-in-place floor slab
To determine how wide the cracks in the floor slabs would open, strain magnitudes in the steel
reinforcement were investigated. The FEM calculated maximum strain in the top layer of
reinforcing steel in the top floor slab is plotted in Figure 8.15, with strains exceeding the
0.0025 yield strain of the grade 500 reinforcing steel represented by the grey region. At both
2% and 3% lateral drift only a small portion of the reinforcing steel was expected to yield,
which indicated that most of the concrete cracks should close up when the load is removed.
Additionally, the maximum tensile strain at 2% drift was only 0.0098, which just reaches the
- 281 -
seismic serviceability limit strain of 0.010, as suggested by Priestley et al. [8-11] for members
without axial load. Consequently, during design level earthquake loading the damage to the
floor slab would result in residual crack widths being less than 1.0 mm, which are easily
repaired. However, yielding of the reinforcement results in hysteretic energy dissipation,
which combined with the residual crack width, may result in larger residual displacements and
a loss of the wall system’s self-centering advantage. The cyclic floor hysteresis is investigated
in more detail in section 8.4.4.
(a) 2% lateral drift (b) 3% lateral drift

Figure 8.15 – Calculated reinforcing steel strains of the top cast-in-place floor slab
The monotonic moment-drift response calculated from the FEM of the prototype building with
cast-in-place (CIP) floor is plotted in Figure 8.16, alongside the FEM response of just the
PreWEC wall (ignoring the floors and gravity frame). It can be seen that the inclusion of the
floors significantly altered the response, with the deformation and bending of the floor
diaphragms increasing the moment resistance of the building by approximately 44% at 2%
lateral wall drift and 50% at 3% lateral wall drift. This over-strength would have a significant
effect on the building design and seismic response, particularly when considering capacity
design principles. The increased moment resistance resulted in an increased shear demand on
the PreWEC wall, which could lead to undesirable failure mechanisms such as shear failure or
base sliding of the wall panel. Clearly the influence of the floor diaphragm should be
considered and quantified during the design process and that analysing wall by itself is not
sufficient, even when the wall is the primary lateral resisting structural element.
- 282 -
40
2500
35
30
Base Moment (MN-m)

2000
Base Shear (kN)

25
1500
20
15 1000
10
500
5 Wall only
Building: CIP floor
0 0
0 0.5 1 1.5 2 2.5 3
Lateral Wall Drift (%)
Figure 8.16 – Calculated response of the prototype building FEM with cast-in-place floor
8.4.3 Isolated Floor

In order to simulate the opposite extreme from a rigid cast-in-place connection, a fully isolated
wall-to-floor connection was also modelled. This isolated connection represented the use of a
slotted connector, or other techniques that could isolate the floors from the vertical uplift of the
wall while still providing a load path for the horizontal inertia forces. As stated earlier, an
isolated connection could be achieved using any type of floor, including a flat cast-in-place
slab or precast floor units, and so for the purpose of this investigation the floor shell element
remained unchanged from the cast-in-place analysis described above. Because the floor was
isolated from the deformation of the wall, the relative strength and stiffness of the floor
diaphragm was not expected to have a significant influence on the results of the FEM analysis.
8.4.3.1 Connection modelling

Instead of the floor diaphragm extending through the wall, as was the case with the rigid cast-
in-place connection, a cut-out gap was provided in the floor diaphragm around the entire
PreWEC wall. The slotted connectors were then modelled between the floor and wall by
coupling the horizontal displacements of a floor node to a corresponding node of the wall while
leaving the vertical displacements unconstrained. The slotted connector couplings were
modelled at four locations for each wall-to-floor connection, with two on each side of the wall.
- 283 -
8.4.3.2 Results
The deformed shape of the building FEM with isolated floors at 3% lateral wall drift is shown
in Figure 8.17, with the deformations magnified three times. Again, the PreWEC wall behaved
in the same manner to the previous FEM of the individual PreWEC specimen, with
deformation primarily concentrated at the single crack at the wall base. However, the uplift
and rotation of the wall was not transferred to the floors, and as a result the floor diaphragms
remained relatively undeformed as they displaced horizontally with the wall. Upon closer
inspection, no significant damage was calculated for the floor diagrams with strains found to
not exceed the concrete cracking strain at any location.
Figure 8.17 – Elevation view showing the calculated displaced shape of the prototype
building FEM with isolated floor diaphragms (displacements magnified 3 times)
As well as preventing damage to the floor diaphragms, the isolated wall-to-floor connection
resulted in a more predictable and dependable lateral strength. Figure 8.18 shows a
comparison of the moment-drift response of the building with an isolated wall-to-floor
connection, alongside the response of only the PreWEC wall and the building with a cast-in-
place (CIP) wall-to-floor connection. Unlike the cast-in-place connection, which showed a
significant increase in strength, the response of the building with isolated floor connections is
almost identical to the analysis of the individual PreWEC wall. This means that when the
floors are isolated, the entire lateral resistance is provided by the PreWEC system and that the
seismic response of the building can be easily determined.
- 284 -
40
2500
35
30
Base Moment (MN-m)

2000
Base Shear (kN)

25
1500
20
15 1000
10
Wall only
500
5 Building: CIP floor
Building: Isolated floor
0 0
0 0.5 1 1.5 2 2.5 3
Lateral Wall Drift (%)
Figure 8.18 – Calculated response of the prototype building FEM with isolated floors
8.4.4 Section Component Model

Because of the complexity of the FEM of one half of the prototype building, a full cyclic
analysis could not be conducted. In order to assess the cyclic behaviour and isolate the
response of the cast-in-place wall-to-floor connection, a component model was developed with
a single floor diaphragm and a section of the wall.
8.4.4.1 Description of model

As shown in Figure 8.19, the section component model included a single floor diaphragm,
which was modelled in the same manner as described above. The floor was embedded into a
section of the precast concrete wall that had a height of 7.32 m, equivalent to two stories. The
restraint provided to the floor diaphragm by the gravity frame was idealised using a roller
restraint on the exterior edges of the floor slab, which allowed only in-plane horizontal
displacement. The wall section was then displaced in order to approximate the uplift and
rotation of the wall at the floor level. Because the wall deformation is primarily concentrated
at the wall base, the relative displacements at each floor level are similar, and the average point
of rotation from all four floor levels was calculated from the building FEM to be located at
700 mm in from the edge of the wall. The point of rotation was simulated in the component
model using two compression only springs at the base of the wall section, which allowed for
migration of the point of rotation during reverse cyclic loading. As with previous FEM
analysis, a velocity command was used to displace a reference point at the top of the wall
- 285 -
section in order to approximate the wall movement during loading of the entire building. Both
monotonic and reverse cyclic loading was applied up to a maximum 3% lateral wall drift.
Figure 8.19 – FEM representation of the wall and floor section
8.4.4.2 Results
As seen from Figure 8.20, the displaced shape of the floor diaphragm in the section component
FEM appeared similar to that of the floor diaphragm in the full building FEM. Additionally,
the relative uplift and rotation calculated for the wall section closely matched the average
deformation of the wall at the four floor levels in the building FEM.
Figure 8.20 – Calculated displaced shape of the floor section FEM at 3% wall drift
(displacements magnified 10 times)
The monotonic lateral load response of the section component FEM could be extrapolated in
order to adequately predict the response of the entire building. The moment-drift response of
the wall-to-floor section FEM was calculated, multiplied by four floors, and then added to the
response of the FEM of just the PreWEC wall. As seen in Figure 8.21a, the calculated wall
plus section response closely matched the response of the building FEM. This correlation
highlighted that although the interaction between components in the building was complex,
good estimation of the building response could be achieved by combining the wall behaviour
with the behaviour of the floor diaphragm subjected to a rotational deformation.
- 286 -
40 4
2500
35 3
30 2
Base Moment (MN-m)
Base Moment (MN-m)

2000
Base Shear (kN)

25 1
1500
20 0
15 1000 -1
10 -2
Wall only
500
5 Building: CIP floor -3 Monotonic
Wall only + section Cyclic
0 0 -4
0 0.5 1 1.5 2 2.5 3 -3 -2 -1 0 1 2 3
Lateral Wall Drift (%) Lateral Wall Drift (%)
(a) Monotonic response of the full building and (b) Monotonic and cyclic response of the
wall + floor section FEM cast-in-place floor section
Figure 8.21 – Calculated response of the floor section FEM
After confirming the validity of the component section FEM, reverse cyclic loading was
applied in order to characterise the hysteresis behaviour of the cast-in-place wall-to-floor
connection. The calculated cyclic moment-drift response of the section component FEM is
shown alongside the monotonic response in Figure 8.21b. The cyclic behaviour follows typical
hysteresis behaviour of reinforced concrete sections, with a degrading stiffness unloading and
reloading behaviour. Because only a small portion of the reinforcing steel in the floor slab
yielded, the hysteresis loops are relatively narrow, which should minimise the negative effects
of the floor hysteresis on the ability of the building to self-center following an earthquake.
Additionally, the amount of hysteretic energy dissipation was already accounted for during the
self-centering analysis conducted in Chapter 7, and if the hysteresis behaviour is correctly
quantified a self-centering response can be ensured during the design process.
8.5 DYNAMIC ANALYSIS

Time history analysis was used to further investigate the seismic response of the prototype
building and the effect of the interaction between the wall and the surrounding structure. The
four storey prototype building was modelled with two different wall-to-floor connections,
being the rigid cast-in-place connection and an isolated connection. The isolated wall-to-floor
connection represented a wall-to-floor connection where the floor diaphragms were isolated
- 287 -
from the vertical uplift of the wall, or alternatively the isolated connection could represent the
result of an analysis conducted where the influence of the wall-to-floor interaction was ignored.
8.5.1 Description of Model

The 2D model that was used to conduct the dynamic analyses was an extension of the
Ruaumoko lumped plasticity model that was developed previously in Chapter 7. The seven
spring configuration that was calibrated to match the base moment-rotation behaviour of the
PreWEC wall was used in the lumped plasticity model of the prototype building, as seen in
Figure 8.22. The PreWEC wall was again represented by elastic frame elements, except that
instead of a single lumped mass, the building mass was lumped at each of the four floor levels.
The four storey prototype building that was designed in Chapter 7 had an effective mass, me, of
1860.3 tonne, which equated to a floor mass, mi, of 558.1 tonne. The gravity frame columns
were included with elastic properties and the floor diaphragms were represented using rigid
elements. As with the building FEM described above, the floor-to-column connections and the
column-to-foundation connection were idealised using pinned restraints.
mi
3.66 m
mi
Wall element 3.66 m
mi
3.66 m
Floor element Pin
mi
Wall-to-floor spring
Column element 3.66 m
PreWEC 7-spring configuration
10.0 m 10.0 m
Figure 8.22 – Lumped plasticity model of four storey prototype building
- 288 -
Because the lumped plasticity model did not explicitly model the wall uplift, the behaviour of
the wall-to-floor connection was represented by a rotational spring that was placed between the
wall and the floor elements. For the rigid cast-in-place wall-to-floor connection, the floor
behaviour that was determined from the section component FEM of a single floor diaphragm
was used to calibrate the wall-to-floor rotational springs. The moment-rotation behaviour
calculated by the FEM was split into the two spring locations on either end of the wall, and as
seen in Figure 8.23a, a Modified Tekada hysteresis rule with the properties given in Table 8.1
showed good correlation with the FEM response. For the isolated wall-to-floor connection the
wall-to-floor rotational spring was changed to a pinned connection with no moment resistance.
Table 8.1 – Properties of cast-in-place the wall-to-floor rotational spring
Yield Elastic Bilinear Unloading Reloading

Hysteresis model
moment, My stiffness, K factor, r factor, α factor, β
Modified Tekada 750 kN 150000 kNm-rad 0.25 0.5 1
2 40
1.5 30
1 20
Base Moment (MN-m)
Base Moment (MN-m)
0.5 10
0 0
-0.5 -10
-1 -20
-1.5 Section FEM (M/2) -30 Isolated

Tekada spring model Cast-in-place
-2 -40
-3 -2 -1 0 1 2 3 -3 -2 -1 0 1 2 3
Lateral Wall Drift (%) Lateral Wall Drift (%)
(a) Wall-to-floor rotational spring (b) Building model response

Figure 8.23 – Cyclic behaviour of the prototype building using the lumped plasticity
Ruaumoko model
To verify the lateral load response of the prototype building lumped plasticity model, a reverse
cyclic pseudo-static displacement history was applied to the top node of the wall, and the
calculated base moment-lateral drift response is shown in Figure 8.23b for both the cast-in-
place and isolated wall-to-floor connection options. The over-strength attributed to the floor
- 289 -
deformation is clearly visible for the building response with the cast-in-place connection, and
as expected the building response with the isolated connection was similar to the PreWEC
response from Chapter 7. Interestingly, the additional hysteresis area provided by the cast-in-
place wall-to-floor connection behaviour did not result in any significant increase in the static
residual displacements when compared to the isolated wall-to-floor connection. The negligible
increase in static residual displacements indicated that self-centering should still be achieved,
as was found in Chapter 7.
A modal analysis was conducted prior to the dynamic excitation in order to determine the
corresponding natural frequencies for the four storey prototype building. The fundamental
period of the building with the isolated wall-to-floor connection was 0.88s which corresponded
well with the 0.86s fundamental period from the SDOF representation modelled in Chapter 7.
However, the increased stiffness provided by the rigid cast-in-place floor diaphragms resulted
in a reduced fundamental period of 0.81s. This reduction in natural period may lead to higher
seismic demand for the building with rigid cast-in-place wall-to-floor connections. A similar
trend was also observed for higher mode periods as well.
Note that only the lateral resistance of the PreWEC wall was used in Chapter 7 to design the
four storey prototype building that is used throughout this chapter. This means that the
building model with isolated wall-to-floor connections should be considered to be the design
structure, and the modified behaviour that resulted from the rigid cast-in-place wall-to-floor
connection should be considered as an unexpected phenomenon that was not accounted for
during the seismic design of the building.
8.5.2 Earthquake Ground Motions

The same eight earthquake ground motions that were used during the dynamic analysis in
Chapter 7 were applied to the four storey prototype building. A summary of the details of the
eight ground motion records is given in Table 7.5. As in Chapter 7, the earthquake
acceleration records were scaled to match the 5% damped response spectrum for Zone 4, soil
type Sc, and seismic hazard levels EQ-III and EQ-IV, as outlined by the SEAOC guidelines [8-
12]. The scale factors calculated for the four storey prototype building with isolated wall-to-
floor connections are reported in Table 7.5. Because the wall-to-floor interaction was assumed
to not be accounted for during the building design, the same scale factors were used for the
building with rigid cast-in-place wall-to-floor connections. The use of the same scale factors,
- 290 -
or seismic intensity, also allowed for direct comparison between the results from the building
analyses with each wall-to-floor connection type.
Table 8.2 – Summary of ground motions used for the dynamic analyses
Scale factor
GM Earthquake Year MW Station Direction
Level III Level IV
1 Chi-Chi 1999 7.6 CHY080 NS 0.70 1.05
2 El Centro 1940 7 El Centro Array EW 1.75 2.62
o
3 Imperial Valley 1979 6.5 Cucapah 85 N 1.60 2.40
4 Kobe 1995 6.9 KJMA NS 0.77 1.15
5 Loma Prieta 1989 6.9 Gillroy Array #6 NS 3.77 5.65
6 Northridge 1994 6.7 Coldwater Can EW 1.92 2.88
7 Northridge 1994 6.7 Mt Gleason Ave NS 3.53 5.29
8 Tabas 1978 7.4 Tabas LN 0.60 0.89
8.5.3 Analysis Details

The non-linear time-history analysis used a Newmark constant average acceleration solution
method with an integration time step of 0.0005s. As with the analysis in Chapter 7, 3% initial
stiffness proportional Rayleigh viscous damping was used in the model to account for the
elastic damping of the structure, including the radiation damping as the wall impacts with the
foundation. The single direction horizontal ground motion records were applied separately to
the structure and all nodes were restrained from movement in the out-of-place direction. The
analysis time was equal to the full length of the earthquake record plus a free vibration time
that was equal to at least 20Ti. The free vibration time allowed the structure sufficient time for
the response to decay completely, allowing accurate calculation of the final residual drift.
8.5.4 Results
An example of the lateral drift response at the top of the building calculated from the
Ruaumoko time-history analysis is shown in Figure 8.24 for GM-8, hazard level III. In
general, the response of the buildings with the cast-in-place and isolated wall-to-floor
connections was similar. However, the building with the cast-in-place wall-to-floor connection
typically resulted in lower lateral displacements due to the increased moment resistance and
hysteretic energy dissipation.
- 291 -
2
Top Lateral Drift (%)
-1
Isolated
Cast-in-place
-2
5 10 15 20 25 30 35 40
Time (s)
Figure 8.24 – Top lateral drift time history responses for GM-8, hazard level III
Several key performance criteria were used in order to compare the performance of the four
storey prototype building with cast-in-place and isolated wall-to-floor connections, including
maximum interstorey drift, residual interstorey drift, peak floor acceleration, and peak base
shear.
8.5.4.1 Interstorey drift

Due to the stiff elastic concrete wall panel, the building response was dominated by the
fundamental mode shape, and so for all of the building analyses, the maximum interstorey drift
occurred at the top storey. The calculated maximum interstorey drift results for all of the
building analyses are shown in Figure 8.25 for hazard level EQ-III (DBE) and EQ-IV (MCE).
Both the buildings with cast-in-place and isolated wall-to-floor connections performed well,
with maximum interstorey drifts typically below the 2% design level and 3% maximum
considered drift targets. Except for a few earthquake ground motions, the building with cast-
in-place wall-to-floor connections resulted in lower interstorey drift demand than the building
with isolated wall-to-floor connections. The average interstorey drift for the cast-in-place
connection was 8% below the isolated connection for the design hazard level EQ-III, and 14%
below for the maximum credible hazard level EQ-IV. Although the reduction in fundamental
period due to the stiffness of the cast-in-place wall-to-floor connection would increase the
seismic demand, the lateral resistance over-strength and increased hysteretic energy dissipation
ultimately reduced the interstorey drift demand of the building with cast-in-place wall-to-floor
connections. Therefore, when considering the interstorey drift demand, failure to account for
the wall-to-floor interaction would not necessarily lead to poor seismic performance..
- 292 -
3
(a) Hazard level EQ-III Isolated
Max Interstorey Drift (%)
2.5
Cast-in-place
2
1.5
0.5
0
GM-1 GM-2 GM-3 GM-4 GM-5 GM-6 GM-7 GM-8 Ave
Ground Motion
4 (b) Hazard level EQ-IV Isolated

Max Interstorey Drift (%)
Cast-in-place
3
0
Ground Motion
Figure 8.25 – Maximum interstorey drift results for the prototype building
8.5.4.2 Residual interstorey drift

When considering residual displacement, the residual interstorey drift was almost identical for
each storey level, with permanent displacements caused by the rotational springs at the wall
base. The calculated maximum residual interstorey drift results for all of the building analyses
are shown in Figure 8.26 for hazard level EQ-III (DBE) and EQ-IV (MCE). The results are
somewhat scattered, with both the cast-in-place and isolated wall-to-floor connections resulting
in larger residual interstorey drifts for different earthquake ground motions. However, on
average there appeared to be a reduction in the expected residual interstorey drift for the
building with cast-in-place wall-to-floor connections of up to 75% for hazard level EQ-IV.
Additionally, from observation of the lateral displacement time-history responses, the building
with cast-in-place wall-to-floor connections appeared to self-center faster than the building
with isolated wall-to-floor connections. The ability to self-center quickly is most likely a result
of the increased hysteretic energy dissipation provided by the cast-in-place wall-to-floor
connections. Therefore, when considering the residual interstorey drift, failure to account for
the wall-to-floor interaction would also not necessarily lead to poor seismic performance or
- 293 -
failure of the building to self-center. However, this finding is obviously dependent on the
building configuration. If the lateral resistance over-strength resulting from the wall-to-floor
interaction was increased to the stage where there was a significant increase in the static
residual displacement observed from the hysteresis loops, the ability of the building to self-
centering following an earthquake would be compromised.
0.05
Residual Intersorey Drift (%)

0.04 Cast-in-place
0.03
0.02
0.01
0
Ground Motion
0.12
Residual Intersorey Drift (%)
(b) Hazard level EQ-IV Isolated

0.1
Cast-in-place
0.08
0.06
0.04
0.02
0
Ground Motion
Figure 8.26 – Residual displacement results for the prototype building
8.5.4.3 Floor acceleration

The calculated peak floor acceleration results for all of the building analyses are shown in
Figure 8.27 for hazard level EQ-III (DBE) and EQ-IV (MCE). The peak floor accelerations
were not isolated to any particular floor level, with the peak acceleration at different floor
heights dependent on the characteristics of the earthquake ground motion. Except for a few
earthquake ground motions, the building with cast-in-place wall-to-floor connections resulted
in higher peak floor accelerations than the building with isolated wall-to-floor connections, but
only by up to 6.5% on average. The higher accelerations are likely caused by the reduced
fundamental period and the increased inelastic stiffness caused by the over-strength.
- 294 -
Therefore, when considering peak accelerations, failure to account for the wall-to-floor
interaction may result in less desirable seismic performance, but not by any significant amount.
10
Peak Floor Accel (m/s 2)
8 Cast-in-place
0
Ground Motion
15
Peak Floor Accel (m/s 2)
Cast-in-place
10
0
Ground Motion
Figure 8.27 – Peak floor acceleration results for the prototype building
8.5.4.4 Base shear

The calculated peak base shear results for all of the building analyses are shown in Figure 8.28
for hazard level EQ-III (DBE) and EQ-IV (MCE). For all except one earthquake record, the
building with cast-in-place wall-to-floor connections resulted in higher peak base shear
demand than the building with isolated wall-to-floor connections, due to the increased lateral
resistance provided by the cast-in-place floors. On average the increase in base shear was 10%
for the design hazard level EQ-III and 16% for the maximum credible hazard level EQ-IV.
Therefore, when considering peak base shear, failure to account for the wall-to-floor
interaction may result in less desirable seismic performance, with potential shear or sliding
failure of the concrete wall.
- 295 -
10000
Peak Base Shear (kN)
8000 Cast-in-place
6000
4000
2000
0
Ground Motion
12000
Peak Base Shear (kN)
10000
Cast-in-place
8000
6000
4000
2000
0
Ground Motion
Figure 8.28 – Peak base shear results for the prototype building
8.6 CONCLUSIONS
Although several successful self-centering precast concrete wall concepts have been
developed, the interaction between the wall and the surrounding structure has previously
received little attention. As a self-centering wall is displaced horizontally, uplift occurs at the
base, causing a relative vertical displacement and rotation at the wall-to-floor connections
while the edges of the floor are restrained vertically by the gravity columns. This chapter
investigated the influence of different types of wall-to-floor connections on the building
response.
Typical wall-to-floor connections can be grouped into three categories: rigid cast-in-place
connection, fully isolated connection, and semi-rigid connection. Isolated wall-to-floor
connections, such as that achieved with a slotted connector, allow the wall to uplift without
deforming the floor diaphragm, and provide a load path for the transfer of seismic inertia forces
from the floor to the lateral load resisting shear wall.
- 296 -
8.6 Conclusions
The PreWEC system can offer unique solutions to isolate floor diaphragms from wall uplift
and subsequent damage. Because the end columns do not experience significant uplift, the
floor diaphragms can be supported on the end columns, which could additionally be used to
carry gravity loads. The seismic inertia forces can then be transferred to the precast wall panel
using either a slotted connector, or a pinned strut between the wall and end column.
Finite element modelling of a four storey prototype building indicated that when a rigid cast-
in-place wall-to-floor connection is used, the deformation of the floor diaphragms was
significant. Although extensive cracking of the concrete floor and some yielding of the
reinforcing steel occurred at the 2% design level lateral drift, the damage would typically be
within serviceability limits and be easily repairable. However, the magnitude of damage
expected at the wall-to-floor connection is dependent on the building layout and the wall
length. The response of the building was dramatically altered when the wall-to-floor
interaction was accounted for, with up to 50% increase in the lateral strength at the design level
lateral drift. This over-strength has potential consequences for the seismic design of the
building, especially when considering possible shear or sliding failure of the wall.
Additionally, if the cyclic hysteresis behaviour of the wall-to-floor connection can be
quantified, self-centering can be ensured using the procedures developed in Chapter 7.
In contrast to a rigid cast-in-place connection, finite element modelling indicated that the
isolated wall-to-floor connection can effectively eliminate any damage to the floor diaphragm
and also provide a more dependable and predictable lateral strength.
A series of dynamic time-history analyses were conducted to assess the seismic behaviour of
the four storey prototype building with PreWEC walls and either rigid cast-in-place or isolated
wall-to-floor connections. Both building configurations performed well, with maximum and
residual interstorey drifts typically being below the design target limits. From the dynamic
analyses it was found that the building with rigid cast-in-place wall-to-floor connections had
lower maximum and residual interstorey drifts, and higher peak floor accelerations and peak
base shear, than the building with isolated wall-to-floor connections. The differences in the
dynamic behaviour highlighted that although life-safety is not at stake, the wall-to-floor
interaction should be accounted for during seismic design in order to ensure that the building
conforms to performance based design objectives.
- 297 -
8.7 REFERENCES
[8-1] Waugh, J. D. and Sritharan, S. (2010) Lessons learned from seismic analysis of a
seven-story concrete test building. Journal of Earthquake Engineering, 14(3), 448-469.
Journal, 44(6), 42-67.
[8-3] Schoettler, M. J., Belleri, A., Dichuan, Z., Restrepo, J. I., and Fleischman, R. B. (2009)
Preliminary results of the shake-table testing for the development of a diaphragm
seismic design methodology. PCI Journal, 54(1), 100-124.
[8-6] NZS 3101:2006. Appendix B: Special provisions for the seismic design of ductile
jointed precast concrete structural systems. Concrete structures standard, Standards
New Zealand, Wellington, New Zealand.
[8-7] Pampanin, S., Marriott, D., and Palermo, A. (2010) PRESSS design handbook. New
Zealand Concrete Society, Auckland, N.Z.
[8-8] PCI design handbook: Precast prestressed concrete. 5th ed. Precast/Prestressed
Concrete Institute, Chicago, IL, 1999.
[8-11] Priestley, M. J. N., Calvi, G. M., and Kowalsky, M. J. (2007) Displacement-based

seismic design of structures. IUSS Press.
[8-12] Seismology Committee. Recommended lateral force requirements and commentary

(Blue book). Structural Engineers Association of California (SEAOC), 1999.

- 298 -
Chapter 9
Conclusions
In Chapter 1 it was stated that the main objective of this study was to overcome several
identified limitations, to allow for widespread implementation of self-centering precast
concrete wall systems. The literature review confirmed these current limitations, which
included:
 complex design procedures that prevent the use of self-centering systems for simple
building designs,
 a reduction in moment capacity when compared to traditional reinforced concrete walls,
which leads to larger and less economical walls,
 failure of the wall to provide a fully seismic resilient system that requires only minor
repair following a large earthquake,
 uncertainty in the behaviour of the complete building and the interaction between self-
centering components and other structural and non-structural elements.
To overcome the above mentioned limitations, this study investigated seismic resilient self-
centering wall systems that are suitable for application in regions of both low and high
seismicity, and focused on modelling the realistic seismic behaviour of the entire building.
- 299 -
Chapter 9 Conclusions
The experimental testing and analysis of self-centering individual PT concrete walls that had
no specific confinement reinforcement in the wall toe was reported in Chapters 3 and 4. These
individual PT walls displayed stable lateral resistance with no visible damage for low lateral
drift demands that are appropriate for seismic design in regions of low seismicity. Simplified
design equations were developed to accurately predict the wall nominal flexural strength,
which greatly reduced the design complexity while retaining the benefits of an easy to
construct self-centering precast concrete wall system.
The PreWEC system was introduced in Chapter 5 as an alternative self-centering wall system
that is suitable for application in regions of moderate to high seismicity. Due to the
arrangement of its components, the PreWEC system has an increased moment capacity when
compared to previous jointed or coupled wall systems. To maximise the seismic resilience of
the PreWEC wall system, the design of easily replaceable energy dissipating O-Connectors
was described. Additionally, Chapter 6 reported extensive finite element modelling that was
performed to further understand the response of the PreWEC system and to investigate the
influence of several important design parameters.
Previous seismic analysis of self-centering structures has relied on idealised behaviour of

systems that are analysed in isolation from the rest of the building. Chapter 7 investigated the
realistic behaviour of self-centering systems, and developed appropriate design procedures that
quantify the expected residual drift of the system following an earthquake. To conclude,
Chapter 8 described finite element and time-history analyses of a prototype building, with the
results highlighting the importance of accounting for the interaction between the self-centering
wall system and the surrounding gravity load elements of the building in order to ensure that a
seismic resilient building design is achieved.
A summary of the main conclusions from each section of work are given below, followed by
several recommendations for future research into the topic of self-centering concrete walls and
seismic resilient building design.
9.1 INDIVIDUAL PT WALLS – CHAPTERS 3 & 4

The behaviour of individual precast concrete wall panels with unbonded post-tensioning was
examined by both experimental testing and analytical FEM analysis. In general, the wall
- 300 -
9.1 Individual PT Walls – Chapters 3 & 4
panels behaved as was expected for a member with unbonded PT, with deformation
concentrated at a single crack that opened up at the base of the wall and only minor localised
damage occurring in the wall toe. Additionally, the developed 3D FEM was able to capture
both the global and local response of the walls with good correlation when compared to the
experimental results.
Experimental measurement of concrete compression strains at the extreme compressive fibre of

the PT walls was difficult, and a large variation in the observed readings was obtained because
localised behaviour strongly influenced the gauge reading. However, the 3D FEM allowed for
accurate calculation of the average strain at a position close to the extreme compression fibre,
which could be used to determine the ultimate limit state of the individual PT walls with no
specific confinement reinforcement in the wall toe.
The lateral load resistance of the wall was maintained well beyond the code defined ultimate
concrete compressive strain of 0.003, and minimal concrete damage was observed at this limit
state. Experimental strain measurements indicated that a higher compressive strain limit of
0.005 or greater may be more suitable for describing the nominal flexural strength of PT
concrete walls.
In Chapter 4 a simple set of equations was developed to predict the wall response at the
nominal flexural strength limit state. Prediction of the nominal flexural strength of a PT wall
requires accurate estimation of the unbonded tendon stresses. A proposed equation for
predicting the unbonded tendon stresses in post-tensioned concrete walls was developed which
provided good estimation of unbonded tendon stress across a wide range of wall parameters
when compared with both finite element analyses and experimental test results. Additionally,
the proposed equation was found to be accurate for an ultimate concrete compressive strain of
0.003 and 0.005. When the new equation for predicting unbonded tendon stresses was used,
accurate prediction of the nominal flexural strength was also achieved for all walls analysed.
Additionally, an equation for predicting the wall lateral displacement was also developed,
which included components from gap opening at the wall base and flexural deformation of the
wall panel. The equation provided good estimation of the wall displacement, when compared
to the finite element analyses, for an ultimate compressive concrete strain of between 0.003 and
0.005.
- 301 -
9.2 PREWEC SYSTEM – CHAPTERS 5 & 6

The PreWEC system overcomes several limitations that are associated with pervious self-
centering wall systems. The arrangement of elements in the PreWEC system maximises the
moment capacity of the wall to provide an economical construction alternative, and
additionally the connectors are located so that they can be easily inspected or replaced
following an earthquake, resulting in a seismic resilient self-centering wall system.
9.2.1 Connector Design

A series of finite element analyses were conducted to investigate the behaviour of energy
dissipating connectors. After analysing several connector types, the O-Connector was selected
as the most suitable connector type for PreWEC systems. An experimental test program
validated both the connector’s performance and the FEM analysis procedure. The FEM was
found to provide accurate prediction of both the connector strength and the critical strains. The
strain limit of 0.10 imposed to predict the maximum displacement capacity proved to
satisfactorily estimate the failure of the tested O-Connector. Lastly, the O-Connectors behaved
as was expected during large-scale testing of a PreWEC wall specimen, providing a stable
response up to and beyond the 2% design level lateral drift.
9.2.2 PreWEC FEM

This previously developed finite element models for both the individual PT walls and the O-
Connector were extended to model the cyclic behaviour of the PreWEC test specimen.
Compared to the unconfined individual concrete PT walls, the PreWEC system required
accurate representation of the confined concrete in the wall toe. Overall, the FEM showed
good correlation with the PreWEC experimental test for both the global and local responses.
The inclusion of cyclic damage parameters, to accurately model the concrete unloading and
reloading paths, was found to be critical to the calculated cyclic response and degradation in
the system’s strength and stiffness.
The FEM was used to investigate several modifications to the PreWEC design. One of the
major findings from these analyses was that when the moment contribution from the
connectors was increased, by adding more connectors and reducing the magnitude of the PT
force, the PreWEC response was improved with increased energy dissipation and reduced wall
- 302 -
9.3 System Behaviour – Chapters 7 & 8
damage. Because of the PreWEC configuration, the wall axial load is primarily attributed to
the PT force and imposed gravity loads, and is independent of the number of connectors. For
this reason, increasing the moment contribution from the connectors reduced the demand on
the wall toe due to the reduced PT load, which reduced the residual drift caused by damage to
the wall toe. This finding means that contrary to what is traditionally assumed, increasing the
connector moment contribution may not always increase the residual displacement (and vice
versa).
To isolate the influence of the connector, further analyses were conducted with a consistent
wall PT force and an increased or decreased number of connectors. The results from these
analyses were used to establish an equation to predict the energy dissipation in the PreWEC
system, which was separated into the contribution from wall and connector hysteresis. It was
shown that the hysteresis behaviour of the wall cannot be assumed as bi-linear elastic and
calculation of the damage to the wall toe is essential for understanding the system’s energy
dissipation and residual drift.
9.3 SYSTEM BEHAVIOUR – CHAPTERS 7 & 8

The experimental response of the PreWEC test specimen and the finite element modelling from
previous sections were used to investigate design aspects of self-centering systems that are
subjected to earthquake actions. The analytical investigation focused on the expected residual
drifts as well as the response of the entire building.
9.3.1 Residual Drift Analysis

The current procedures that are used to ensure that self-centering occurs in structures with a
combination of unbonded PT and energy dissipating elements are inaccurate and do not
account for the dynamic self-centering behaviour that exists when the structure is subjected to
an earthquake. A series of simple analyses showed that prefect self-centering could still be
observed from the cyclic load response even when the energy dissipating elements contribute
to over 50% of the total moment resistance. Additionally, the analysis showed that even when
the PT wall response followed a multi-linear elastic response, a small residual drift can occur
when the moment contributions from the PT and energy dissipating elements are balanced.
- 303 -
A lumped plasticity model was developed to accurately capture the PreWEC behaviour
observed during the experimental test, including the stiffness degradation and the residual drift
that was observed from the cyclic hysteresis loops. The developed model was used to conduct
an extensive parametric study to investigate the residual drift of the PreWEC system when
subjected to earthquake ground motions. The analysis results showed that the residual drifts at
the conclusion of the ground motion record were much lower than the maximum possible
residual drift observed from the cyclic hysteresis loops. As a result, the PreWEC system was
also found to meet performance based residual drift limits that could be used to define realistic
self-centering behaviour.
A residual drift ratio was established to define the dynamic shake-down behaviour, and an
upper bound confidence limit of 0.3 could be used during the design process for all types of
PreWEC structures subjected to any earthquake ground motion. A simplified design process
was developed to incorporate a check on the estimated residual drift using the residual drift
ratio and the estimated hysteresis response of the system.
9.3.2 Building Analysis

Although several successful self-centering precast concrete wall concepts have been
developed, the interaction between the wall and the surrounding structure has previously
received little attention. As a self-centering wall is displaced horizontally, uplift occurs at the
base, causing a relative vertical displacement and rotation at the wall-to-floor connections
while the edges of the floor are restrained vertically by the gravity columns.
The response of a building with self-centering walls would be dependent on the type of wall-
to-floor connection. Finite element modelling of a four storey prototype building indicated that
when a rigid cast-in-place wall-to-floor connection is used, the deformation of the floor
diaphragms was significant. Extensive, but perhaps reparable, cracking of the concrete floor
and yielding of the reinforcing steel occurred, and a 50% increase in the lateral strength was
observed when the wall-to-floor interaction was accounted for. This over-strength has
potential consequences for the seismic design of the building, especially when considering
possible shear or sliding failure of the wall. Additionally, if the cyclic hysteresis behaviour of
the wall-to-floor connection can be quantified, self-centering can be ensured using the
procedures developed in Chapter 7.
- 304 -
9.4 Recommendations for Future Research
In contract to a rigid cast-in-place connection, finite element modelling indicated that the
isolated wall-to-floor connection can effectively eliminate any damage to the floor diaphragm
and also provide a more dependable and predictable lateral strength. Additionally, the
arrangement of the component in the PreWEC system offers some unique isolated solutions,
such as the floors being connected to the end columns instead of directly to the wall.
Lastly, a series of dynamic time-history analyses were conducted to assess the seismic
behaviour of the four storey prototype building with PreWEC walls and either rigid cast-in-
place or isolated wall-to-floor connections. Both building configurations performed well, with
maximum and residual interstorey drifts typically being below the design target limits. From
the dynamic analyses it was found that the building with rigid cast-in-place wall-to-floor
connections had lower maximum and residual interstorey drifts, and higher peak floor
accelerations and peak base shear, than the building with isolated wall-to-floor connections.
The differences in the dynamic behaviour highlighted that although life-safety is not at stake,
the wall-to-floor interaction should be accounted for during seismic design in order to ensure
that the building conforms to performance based design objectives.
9.4 RECOMMENDATIONS FOR FUTURE RESEARCH

As reported in Chapter 3, the lateral load resistance of individual PT walls without specific
confinement reinforcement in the wall toe was maintained beyond concrete compressive strains
of 0.003 at the extreme compression fiber. Although the simplified equations developed in
Chapter 4 were validated for ultimate compressive strain in excess of 0.003, only eight wall
tests achieved reliable strain measurements in excess of 0.003. It is suggested that further
experimental or analytical investigation into the concrete compressive strains in PT walls and
the confinement provided by the foundation would supply evidence to support the use of an
increased ultimate concrete compressive strain during design of PT walls.
The equation to predict the unbonded tendons stresses at nominal flexural strength that was
developed in Chapter 4 was exclusively validated using the matrix of PT concrete walls. The
equation may also be appropriate for use in unbonded PT concrete beams, where a similar
deformation mechanism exists. It is recommended that the proposed equation be compared
against experimental or analytical data of typical PT beam details to assess whether it remains
accurate for members with a higher length/depth ratio.
- 305 -
The concept of a core wall design using the PreWEC system was presented in Chapter 6. The
finite element model indicated that similar behaviour to the in-plane PreWEC response could
be achieved in all loading directions, with gap opening at the wall base and relative vertical
displacements at the connector locations. However, the core wall concept require experimental
validation and simplified design equations that are used to predict the in-plane response of the
PreWEC system need to be expanded to allow for the design of a 3D wall system.
The dynamic self-centering analysis that was described in Chapter 7 used the PreWEC system
as an example. However, it is considered that the PreWEC hysteresis behaviour is typical of
realistic self-centering wall and frame members that experience limited damage in the toe
region. It is suggested that further investigation is required to determine whether the upper
bound residual drift ratio limit of 0.3 was also appropriate for designing other types of self-
centering systems. Also, the inclusion of a residual drift check in the design process relied
heavily on estimation of the cyclic hysteresis response. In Chapter 7 the experimental response
was used to predict the static residual drift behaviour. Further investigation into the calculation
of the static residual drift would greatly simplify the self-centering design process, possibly by
linking the wall damage state to the compressive strain in the wall toe.
Currently dynamic experimental data of self-centering concrete walls is scarce. Shake-table or

mobile shaker testing of self-centering wall systems would provide further validation of the
dynamic self-centering of real systems, as well as assisting in quantifying the radiation
damping as the wall impacts with the foundation.
Lastly, the analysis in Chapter 8 highlighted the importance of modelling the entire building to
account for the interaction between the self-centering wall system and the surrounding gravity
load elements. It is suggested that experimental tests should be conducted to validate the finite
element model that was developed and an extensive parametric study should be conducted to
investigate the response of different building configurations and types of wall-to-floor
connection.
- 306 -
APPENDIX A:
Individual PT Wall Test Results
Complete documentation of results for each of the 32 PT wall tests reported in Chapter 3 is
included within this appendix. The measured results plotted include:
 Lateral force-displacement,
 Unbonded tendon stresses,
 Uplift at the base of the wall,
 Concrete strains from all types of instrumentation.
Additionally, derived results are plotted for:

 Neutral axis depth,
 Wall base rotation for walls A and B,
 Wall lateral deformation components for walls C1, C2, D1, D2, E and F.
Lastly photos taken at the maximum lateral displacement for each test are shown to indicate the
extent of damage to the wall toe.

- 307 -
Appendix A: Individual PT Wall Test Results
A1: Test A-i
250 950
Ten #1
900
Ten #2
200 850 Ten #3

Ten #4
800
Ten #5
Stress (MPa)
150
750
700
100
650
50 600
550
0 500
0 5 10 15 0 5 10 15
(a) Lateral force-displacement (b) Tendon stress
8 350
6 300
250
4
Uplift (mm)
200
2
150
0
100
-2 SGR: 25 mm high
50
SGR: 50 mm high
-4 0
0 5 10 15 0 5 10 15
(c) Wall base uplift (11 gauges) (d) Concrete strain
Figure A1 − Measured results for test A-i
- 308 -
A1: Test A-i
-3
x 10
1200 5
1000
4
Base Rotation (rad)

800
3
NA (mm)
600
2
400
1
200
0 0
0 5 10 15 0 5 10 15
(a) Neutral axis depth (a) Wall base rotation

Figure A2 − Calculated results for test A-i
Figure A3 – Photos of final damage for test A-i
- 309 -
A2: Test A-ii
250 1000
Ten #1
Ten #2
900
200 Ten #3
Ten #4
800
Ten #5
Stress (MPa)
150
700
100
600
50
500
0 400
0 5 10 15 20 25 30 0 5 10 15 20 25 30
15 800
700
600
10
500
Uplift (mm)
400
5
300
200
0
100
25 mm high
0
50 mm high
-5 -100
0 5 10 15 20 25 30 0 5 10 15 20 25 30
Figure A4 − Measured results for test A-ii
- 310 -
A2: Test A-ii
-3
x 10
1200 10
1000 8
Base Rotation (rad)

800 6
NA (mm)
600 4
400 2
200 0
0 -2
0 5 10 15 20 25 30 0 5 10 15 20 25 30

Figure A5 − Calculated results for test A-ii
Figure A6 – Photos of final damage for test A-ii
- 311 -
A3: Test A-iii
120 1100
Ten #1
1000
Ten #2
100
Ten #3
900
Ten #4
80
Ten #5
Stress (MPa)
800
60 700
600
40
500
20
400
0 300
0 10 20 30 40 0 10 20 30 40
25 600
20 500
15 400
Uplift (mm)
10 300
5 200
0 100 SGR: 25 mm high

SGR: 50 mm high
-5 0
0 10 20 30 40 0 10 20 30 40
Figure A7 − Measured results for test A-iii
- 312 -
A3: Test A-iii
-3
x 10
1200 12
1000 10
Base Rotation (rad)

800
NA (mm)
6
600
4
400
2
200
0
0 -2
0 10 20 30 40 0 10 20 30 40

Figure A8 − Calculated results for test A-iii
Figure A9 – Photos of final damage for test A-iii
- 313 -
A4: Test A-iv
140 1200
Ten #1
120 Ten #2
1100
Ten #3
100 Ten #4
1000
Ten #5
Stress (MPa)
80
900
60
800
40
700
20
0 600
0 10 20 30 40 0 10 20 30 40
25 1000
20
800
15
Uplift (mm)
600
10
400
5
200
0 SGR: 25 mm high
SGR: 50 mm high
-5 0
0 10 20 30 40 0 10 20 30 40
Figure A10 − Measured results for test A-iv
- 314 -
A4: Test A-iv
1200 0.012
1000 0.01
Base Rotation (rad)

800 0.008
NA (mm)
600 0.006
400 0.004
200 0.002
0 0
0 10 20 30 40 0 10 20 30 40

Figure A11 − Calculated results for test A-iv
Figure A12 – Photos of final damage for test A-iv
- 315 -
A5: Test B-i
30 1100
Ten #1
1000 Ten #2
25
Ten #3
900
20
Stress (MPa)
800
15
700
10
600
5
500
0 400
0 20 40 60 80 100 120 0 20 40 60 80 100 120
35 3000
SGS #1
30
2500 SGS #2
25 SGS #3
2000 SGR #1
20
Uplift (mm)
SGR #2
15 1500
10
1000
5
500
0
-5 0
0 20 40 60 80 100 120 0 20 40 60 80 100 120
Figure A13 − Measured results for test B-i
- 316 -
A5: Test B-i
500 0.035
0.03
400
0.025
Base Rotation (rad)

300
NA (mm)
0.02
0.015
200
0.01
100
0.005
0 0
0 20 40 60 80 100 120 0 20 40 60 80 100 120

Figure A14 − Calculated results for test B-i
Figure A15 – Photos of final damage for test B-i
- 317 -
A6: Test B-ii
60 1100
Ten #1
1000 Ten #2
50
Ten #3
900
40
Stress (MPa)
800
30
700
20
600
10
500
0 400
0 10 20 30 40 50 60 0 10 20 30 40 50 60
20 3000
SGS #1
2500 SGS #2
15
SGS #3
2000 SGR #1
Uplift (mm)
10
SGR #2
1500
5
1000
0
500
-5 0
0 10 20 30 40 50 60 0 10 20 30 40 50 60
Figure A16 − Measured results for test B-ii
- 318 -
A6: Test B-ii
500 0.018
0.016
400 0.014
Base Rotation (rad)

0.012
300
NA (mm)
0.01
0.008
200
0.006
100 0.004
0.002
0 0
0 10 20 30 40 50 60 0 10 20 30 40 50 60

Figure A17 − Calculated results for test B-ii
Figure A18 – Photos of final damage for test B-ii
- 319 -
A7: Test B-iii
80 1100
Ten #1
70 1000 Ten #2
60 Ten #3
900
Stress (MPa)
50
800
40
700
30
600
20
10 500
0 400
0 10 20 30 40 50 60 0 10 20 30 40 50 60
20 6000
SGS #1
5000 SGS #2
15
SGS #3
4000 SGR #1
Uplift (mm)
10
SGR #2
3000
5
2000
0
1000
-5 0
0 10 20 30 40 50 60 0 10 20 30 40 50 60
Figure A19 − Measured results for test B-iii
- 320 -
A7: Test B-iii
500 0.018
0.016
400 0.014
Base Rotation (rad)

0.012
300
NA (mm)
0.01
0.008
200
0.006
100 0.004
0.002
0 0
0 10 20 30 40 50 60 0 10 20 30 40 50 60

Figure A20 − Calculated results for test B-iii
Figure A21 – Photos of final damage for test B-iii
- 321 -
A8: Test B-iv
80 1100
Ten #1
70
Ten #2
1000
60 Ten #3
900
Stress (MPa)
50
40 800
30
700
20
600
10
0 500
0 10 20 30 40 50 60 0 10 20 30 40 50 60
20 5000
SGS #1
SGS #2
15 4000
SGS #3
SGR #1
Uplift (mm)
10 3000
SGR #2
5 2000
0 1000
-5 0
0 10 20 30 40 50 60 0 10 20 30 40 50 60
Figure A22 − Measured results for test B-iv
- 322 -
A8: Test B-iv
500 0.018
0.016
400 0.014
Base Rotation (rad)

0.012
300
NA (mm)
0.01
0.008
200
0.006
100 0.004
0.002
0 0
0 10 20 30 40 50 60 0 10 20 30 40 50 60

Figure A23 − Calculated results for test B-iv
Figure A24 – Photos of final damage for test B-iv
- 323 -
A9: Test C1-i
70 1700
Ten #1
1600
60 Ten #2
1500 Ten #3
50
1400
Stress (MPa)
40 1300
30 1200
1100
20
1000
10
900
0 800
0 20 40 60 80 0 20 40 60 80
4
x 10
25 2.5
SGS #1
20 2 SGS #2
Ext #1
15 1.5 Ext #2
Uplift (mm)
10 1
5 0.5
0 0
-5 -0.5
0 20 40 60 80 0 20 40 60 80
Figure A25 − Measured results for test C1-i
- 324 -
A9: Test C1-i
500 80
 Sum
70
 Gap
400
60  Flxure
Deformation (mm)
50  Shear
300
NA (mm)
40
200
30
20
100
10
0 0
0 20 40 60 80 0 20 40 60 80
(a) Neutral axis depth (a) Wall deformation components

Figure A26 − Calculated results for test C1-i
Figure A27 – Photos of final damage for test C1-i
- 325 -
A10: Test C1-ii
70 1600
Ten #1
1500
60 Ten #2
1400 Ten #3
50
1300
Stress (MPa)
40 1200
30 1100
1000
20
900
10
800
0 700
0 20 40 60 80 0 20 40 60 80
20 2500
2000
15
1500
Uplift (mm)
10
1000
5
500 SGS #1
SGS #2
0
0
Ext #1
Ext #2
-5 -500
0 20 40 60 80 0 20 40 60 80
Figure A28 − Measured results for test C1-ii
- 326 -
A10: Test C1-ii
500 80
 Sum
70
 Gap
400
60  Flxure
Deformation (mm)
50  Shear
300
NA (mm)
40
200
30
20
100
10
0 0
0 20 40 60 80 0 20 40 60 80

Figure A29 − Calculated results for test C1-ii
Figure A30 – Photos of final damage for test C1-ii
- 327 -
A11: Test C1-iii
100 1600
Ten #1
Ten #2
1500
80
Ten #3
1400
Stress (MPa)
60
1300
40
1200
20
1100
0 1000
0 10 20 30 40 50 60 0 10 20 30 40 50 60
20 5000
15 4000
3000
Uplift (mm)
10
2000
5
SGS #1
1000
SGS #2
0
Ext #1
0
Ext #2
-5
0 10 20 30 40 50 60 0 10 20 30 40 50 60
Figure A31 − Measured results for test C1-iii
- 328 -
A11: Test C1-iii
500 60
 Sum
50  Gap
400
 Flxure
Deformation (mm)
40  Shear
300
NA (mm)
30
200
20
100
10
0 0
0 10 20 30 40 50 60 0 10 20 30 40 50 60

Figure A32 − Calculated results for test C1-iii
Figure A33 – Photos of final damage for test C1-iii
- 329 -
A12: Test C1-iv
90 1600
Ten #1
80
Ten #2
1500
70 Ten #3
60 1400
Stress (MPa)
50
1300
40
30 1200
20
1100
10
0 1000
0 10 20 30 40 0 10 20 30 40
10 4500
4000
8
3500
6 3000
Uplift (mm)
2500
4
2000
2 1500 SGS #1
1000 SGS #2
0
Ext #1
500
Ext #2
-2 0
0 10 20 30 40 0 10 20 30 40
Figure A34 − Measured results for test C1-iv
- 330 -
A12: Test C1-iv
700 40
 Sum
600 35
 Gap
30  Flxure
500
Deformation (mm)
25  Shear
NA (mm)
400
20
300
15
200
10
100 5
0 0
0 10 20 30 40 0 10 20 30 40

Figure A35 − Calculated results for test C1-iv
Figure A36 – Photos of final damage for test C1-iv
- 331 -
A13: Test C2-i
90 1500
Ten #1
80 1400
Ten #2
70 Ten #3
1300
60
Stress (MPa)
1200
50
1100
40
1000
30
900
20
10 800
0 700
0 10 20 30 40 0 10 20 30 40
14 2500
SGS #1
12
SGS #2
2000
10 SGS #3
SGS #4
8
Uplift (mm)
1500
1000
4
2
500
0
-2 0
0 10 20 30 40 0 10 20 30 40
Figure A37 − Measured results for test C2-i
- 332 -
A13: Test C2-i
600 40
 Sum
35
500  Gap
30  Flxure
Deformation (mm)
400  Shear
25
NA (mm)
300 20
15
200
10
100
5
0 0
0 10 20 30 40 0 10 20 30 40

Figure A38 − Calculated results for test C2-i
Figure A39 – Photos of final damage for test C2-i
- 333 -
A14: Test C2-ii
90 1500
Ten #1
80
1400 Ten #2
70 Ten #3
1300
60
Stress (MPa)
50 1200
40 1100
30
1000
20
900
10
0 800
0 10 20 30 40 0 10 20 30 40
14 3500
SGS #1
12 3000
SGS #2
10 SGS #3
2500
SGS #4
8
Uplift (mm)
2000
6
1500
4
1000
2
0 500
-2 0
0 10 20 30 40 0 10 20 30 40
Figure A40 − Measured results for test C2-ii
- 334 -
A14: Test C2-ii
600 40
 Sum
35
500  Gap
30  Flxure
Deformation (mm)
400  Shear
25
NA (mm)
300 20
15
200
10
100
5
0 0
0 10 20 30 40 0 10 20 30 40

Figure A41 − Calculated results for test C2-ii
Figure A42 – Photos of final damage for test C2-ii
- 335 -
A15: Test C2-iii
140 1600
Ten #1
120 Ten #2
1500
Ten #3
100
1400
Stress (MPa)
80
1300
60
1200
40
1100
20
0 1000
0 5 10 15 20 25 30 0 5 10 15 20 25 30
10 3000
SGS #1
8 2500 SGS #2
SGS #3
6 2000 SGS #4
Uplift (mm)
4 1500
2 1000
0 500
-2 0
0 5 10 15 20 25 30 0 5 10 15 20 25 30
Figure A43 − Measured results for test C2-iii
- 336 -
A15: Test C2-iii
600 30
 Sum
500 25  Gap
 Flxure
Deformation (mm)
400 20  Shear
NA (mm)
300 15
200 10
100 5
0 0
0 5 10 15 20 25 30 0 5 10 15 20 25 30

Figure A44 − Calculated results for test C2-iii
Figure A45 – Photos of final damage for test C2-iii
- 337 -
A16: Test C2-iv
140 1600
Ten #1
120 Ten #2
1500
Ten #3
100
1400
Stress (MPa)
80
1300
60
1200
40
1100
20
0 1000
0 5 10 15 20 25 30 0 5 10 15 20 25 30
10 3500
SGS #1
8 3000
SGS #2
SGS #3
2500
6 SGS #4
Uplift (mm)
2000
4
1500
2
1000
0
500
-2 0
0 5 10 15 20 25 30 0 5 10 15 20 25 30
Figure A46 − Measured results for test C2-iv
- 338 -
A16: Test C2-iv
600 30
 Sum
500 25  Gap
 Flxure
Deformation (mm)
400 20  Shear
NA (mm)
300 15
200 10
100 5
0 0
0 5 10 15 20 25 30 0 5 10 15 20 25 30

Figure A47 − Calculated results for test C2-iv
Figure A48 – Photos of final damage for test C2-iv
- 339 -
A17: Test D1-i
200 1500
Ten #1
1400 Ten #2
Ten #3
150
1300 Ten #4
Ten #5
Stress (MPa)
1200
100
1100
1000
50
900
0 800
0 10 20 30 40 50 0 10 20 30 40 50
25 3500
SGS #1
20 3000
SGS #2
SGS #3
2500
15
Uplift (mm)
2000
10
1500
5
1000
0
500
-5 0
0 10 20 30 40 50 0 10 20 30 40 50
Figure A49 − Measured results for test D1-i
- 340 -
A17: Test D1-i
1200 50
 Sum
1000  Gap
40
 Flxure
Deformation (mm)
800  Shear
30
NA (mm)
600
20
400
10
200
0 0
0 10 20 30 40 50 0 10 20 30 40 50

Figure A50 − Calculated results for test D1-i
Figure A51 – Photos of final damage for test D1-i
- 341 -
A18: Test D1-ii
250 1600
Ten #1
1500
Ten #2
200 Ten #3
1400
Ten #4
Ten #5
Stress (MPa)
1300
150
1200
100
1100
1000
50
900
0 800
0 10 20 30 40 50 0 10 20 30 40 50
25 3500
SGS #1
20 3000
SGS #2
SGS #3
2500
15
Uplift (mm)
2000
10
1500
5
1000
0
500
-5 0
0 10 20 30 40 50 0 10 20 30 40 50
Figure A52 − Measured results for test D1-ii
- 342 -
A18: Test D1-ii
1200 50
 Sum
1000  Gap
40
 Flxure
Deformation (mm)
800  Shear
30
NA (mm)
600
20
400
10
200
0 0
0 10 20 30 40 50 0 10 20 30 40 50

Figure A53 − Calculated results for test D1-ii
Figure A54 – Photos of final damage for test D1-ii
- 343 -
A19: Test D1-iii
300 1600
Ten #1
1500 Ten #2
250
Ten #3
1400 Ten #4
200
Ten #5
Stress (MPa)
1300
150
1200
100
1100
50
1000
0 900
0 5 10 15 20 25 30 0 5 10 15 20 25 30
14 3500
SGS #1
12
3000
SGS #2
10
SGS #3
2500
8
Uplift (mm)
6 2000
4 1500
2
1000
0
500
-2
-4 0
0 5 10 15 20 25 30 0 5 10 15 20 25 30
Figure A55 − Measured results for test D1-iii
- 344 -
A19: Test D1-iii
1200 30
 Sum
1000 25  Gap
 Flxure
Deformation (mm)
800 20  Shear
NA (mm)
600 15
400 10
200 5
0 0
0 5 10 15 20 25 30 0 5 10 15 20 25 30

Figure A56 − Calculated results for test D1-iii
Figure A57 – Photos of final damage for test D1-iii
- 345 -
A20: Test D1-iv
350 1600
Ten #1
300 Ten #2
1500
Ten #3
250 Ten #4
1400
Ten #5
Stress (MPa)
200
1300
150
1200
100
1100
50
0 1000
0 5 10 15 20 25 30 0 5 10 15 20 25 30
16 3500
SGS #1
14
3000
SGS #2
12
SGS #3
2500
10
Uplift (mm)
8 2000
6 1500
4
1000
2
500
0
-2 0
0 5 10 15 20 25 30 0 5 10 15 20 25 30
Figure A58 − Measured results for test D1-iv
- 346 -
A20: Test D1-iv
1200 30
 Sum
1000 25  Gap
 Flxure
Deformation (mm)
800 20  Shear
NA (mm)
600 15
400 10
200 5
0 0
0 5 10 15 20 25 30 0 5 10 15 20 25 30

Figure A59 − Calculated results for test D1-iv
Figure A60 – Photos of final damage for test D1-iv
- 347 -
A21: Test D2-i
250 1600
Ten #1
Ten #2
1400
200 Ten #3
Ten #4
1200
Ten #5
Stress (MPa)
150
1000
100
800
50
600
0 400
0 5 10 15 20 25 30 0 5 10 15 20 25 30
25 3500
SGS #1
20 3000
SGS #2
SGS #3
2500
15
Uplift (mm)
2000
10
1500
5
1000
0
500
-5 0
0 5 10 15 20 25 30 0 5 10 15 20 25 30
Figure A61 − Measured results for test D2-i
- 348 -
A21: Test D2-i
1200 30
 Sum
1000 25  Gap
 Flxure
Deformation (mm)
800 20  Shear
NA (mm)
600 15
400 10
200 5
0 0
0 5 10 15 20 25 30 0 5 10 15 20 25 30

Figure A62 − Calculated results for test D2-i
Figure A63 – Photos of final damage for test D2-i
- 349 -
A22: Test D2-ii
250 1500
Ten #1
1400
Ten #2
200 Ten #3
1300
Ten #4
Ten #5
Stress (MPa)
1200
150
1100
100
1000
900
50
800
0 700
0 5 10 15 20 25 30 0 5 10 15 20 25 30
20 3500
SGS #1
3000
SGS #2
15
SGS #3
2500
Uplift (mm)
10
2000
1500
5
1000
0
500
-5 0
0 5 10 15 20 25 30 0 5 10 15 20 25 30
Figure A64 − Measured results for test D2-ii
- 350 -
A22: Test D2-ii
1200 30
 Sum
1000 25  Gap
 Flxure
Deformation (mm)
800 20  Shear
NA (mm)
600 15
400 10
200 5
0 0
0 5 10 15 20 25 30 0 5 10 15 20 25 30

Figure A65 − Calculated results for test D2-ii
Figure A66 – Photos of final damage for test D2-ii
- 351 -
A23: Test D2-iii
450 1500
Ten #1
400 1400
Ten #2
350 Ten #3
1300
Ten #4
300
Ten #5
Stress (MPa)
1200
250
1100
200
1000
150
900
100
50 800
0 700
0 5 10 15 20 25 30 0 5 10 15 20 25 30
20 3500
SGS #1
3000
SGS #2
15
SGS #3
2500
Uplift (mm)
10
2000
1500
5
1000
0
500
-5 0
0 5 10 15 20 25 30 0 5 10 15 20 25 30
Figure A67 − Measured results for test D2-iii
- 352 -
A23: Test D2-iii
1200 30
 Sum
1000 25  Gap
 Flxure
Deformation (mm)
800 20  Shear
NA (mm)
600 15
400 10
200 5
0 0
0 5 10 15 20 25 30 0 5 10 15 20 25 30

Figure A68 − Calculated results for test D2-iii
Figure A69 – Photos of final damage for test D2-iii
- 353 -
A24: Test D2-iv
400 1500
Ten #1
350
Ten #2
1400 Ten #3
300
Ten #4
Ten #5
Stress (MPa)
250
1300
200
1200
150
100
1100
50
0 1000
0 5 10 15 0 5 10 15
10 3500
SGS #1
3000
SGS #2
SGS #3
2500
5
Uplift (mm)
2000
1500
0
1000
500
-5 0
0 5 10 15 0 5 10 15
Figure A70 − Measured results for test D2-iv
- 354 -
A24: Test D2-iv
1200 15
 Sum
1000  Gap
 Flxure
Deformation (mm)
800 10  Shear
NA (mm)
600
400 5
200
0 0
0 5 10 15 0 5 10 15

Figure A71 − Calculated results for test D2-iv
Figure A72 – Photos of final damage for test D2-iv
- 355 -
A25: Test E-i
40 1600
Ten #1
35
Ten #2
1500
30 Ten #3
Stress (MPa)
25
1400
20
1300
15
10
1200
5
0 1100
0 20 40 60 80 100 0 20 40 60 80 100
25 16000
SGS #1
14000
20 SGS #2
12000
SGS #3
15 10000 SGS #4
Uplift (mm)
8000 SGE #1
10
SGE #2
6000
5 4000
2000
0
0
-5 -2000
0 20 40 60 80 100 0 20 40 60 80 100
Figure A73 − Measured results for test E-i
- 356 -
A25: Test E-i
600 100
 Sum
500  Gap
80
 Flxure
Deformation (mm)
400  Shear
60
NA (mm)
300
40
200
20
100
0 0
0 20 40 60 80 100 0 20 40 60 80 100

Figure A74 − Calculated results for test E-i
Figure A75 – Photos of final damage for test E-i
- 357 -
A26: Test E-ii
70 1600
Ten #1
60 1500
Ten #2
1400 Ten #3
50
Stress (MPa)
1300
40
1200
30
1100
20
1000
10 900
0 800
0 20 40 60 80 0 20 40 60 80
20 10000
8000
15
6000
Uplift (mm)
10
SGS #1
4000
SGS #2
5
2000 SGS #3
SGS #4
0
0 SGE #1
SGE #2
-5 -2000
0 20 40 60 80 0 20 40 60 80
Figure A76 − Measured results for test E-ii
- 358 -
A26: Test E-ii
600 80
 Sum
70
500  Gap
60  Flxure
Deformation (mm)
400  Shear
50
NA (mm)
300 40
30
200
20
100
10
0 0
0 20 40 60 80 0 20 40 60 80

Figure A77 − Calculated results for test E-ii
Figure A78 – Photos of final damage for test E-ii
- 359 -
A27: Test E-iii
90 1500
Ten #1
80 1400
Ten #2
70 1300 Ten #3
60 1200
Stress (MPa)
50 1100
40 1000
30 900
20 800
10 700
0 600
0 20 40 60 80 0 20 40 60 80
20 6000
5000
15
4000
Uplift (mm)
10
3000
SGS #1
2000 SGS #2
5
SGS #3
1000 SGS #4
0
SGE #1
0
SGE #2
-5 -1000
0 20 40 60 80 0 20 40 60 80
Figure A79 − Measured results for test E-iii
- 360 -
A27: Test E-iii
600 80
 Sum
70
500  Gap
60  Flxure
Deformation (mm)
400  Shear
50
NA (mm)
300 40
30
200
20
100
10
0 0
0 20 40 60 80 0 20 40 60 80

Figure A80 − Calculated results for test E-iii
Figure A81 – Photos of final damage for test E-iii
- 361 -
A28: Test E-iv
100 1550
Ten #1
1500
Ten #2
80 1450 Ten #3
1400
Stress (MPa)
60
1350
1300
40
1250
20 1200
1150
0 1100
0 10 20 30 40 0 10 20 30 40
10 7000
6000
8
5000
6
4000
Uplift (mm)
SGS #1
4 3000
SGS #2
2000
2 SGS #3
1000 SGS #4
0 SGE #1
0
SGE #2
-2 -1000
0 10 20 30 40 0 10 20 30 40
Figure A82 − Measured results for test E-iv
- 362 -
A28: Test E-iv
600 40
 Sum
35
500  Gap
30  Flxure
Deformation (mm)
400  Shear
25
NA (mm)
300 20
15
200
10
100
5
0 0
0 10 20 30 40 0 10 20 30 40

Figure A83 − Calculated results for test E-iv
Figure A84 – Photos of final damage for test E-iv
- 363 -
A29: Test F-i
200 1500
Ten #1
1400 Ten #2
Ten #3
150
1300 Ten #4
Ten #5
Stress (MPa)
1200
100
1100
1000
50
900
0 800
0 10 20 30 40 0 10 20 30 40
25 16000
SGS #1
14000
20 SGS #2
12000 SGS #3
15 SGS #4
10000
Uplift (mm)
SGE #1
10 8000
SGE #2
6000
5
4000
0
2000
-5 0
0 10 20 30 40 0 10 20 30 40
Figure A85 − Measured results for test F-i
- 364 -
A29: Test F-i
1200 40
 Sum
35
1000  Gap
30  Flxure
Deformation (mm)
800  Shear
25
NA (mm)
600 20
15
400
10
200
5
0 0
0 10 20 30 40 0 10 20 30 40

Figure A86 − Calculated results for test F-i
Figure A87 – Photos of final damage for test F-i
- 365 -
A30: Test F-ii
250 1600
Ten #1
Ten #2
200 1500 Ten #3

Ten #4
Ten #5
Stress (MPa)
150 1400
100 1300
50 1200
0 1100
0 10 20 30 40 0 10 20 30 40
20 16000
SGS #1
14000
SGS #2
15
12000 SGS #3
SGS #4
10000
Uplift (mm)
10
SGE #1
8000
SGE #2
5
6000
4000
0
2000
-5 0
0 10 20 30 40 0 10 20 30 40
Figure A88 − Measured results for test F-ii
- 366 -
A30: Test F-ii
1200 40
 Sum
35
1000  Gap
30  Flxure
Deformation (mm)
800  Shear
25
NA (mm)
600 20
15
400
10
200
5
0 0
0 10 20 30 40 0 10 20 30 40

Figure A89 − Calculated results for test F-ii
Figure A90 – Photos of final damage for test F-ii
- 367 -
A31: Test F-iii
300 1500
Ten #1
1400
Ten #2
250
Ten #3
1300
Ten #4
200
Ten #5
Stress (MPa)
1200
150 1100
1000
100
900
50
800
0 700
0 10 20 30 40 0 10 20 30 40
20 14000
12000
15
10000
Uplift (mm)
10
8000
SGS #1
6000 SGS #2
5
SGS #3
4000 SGS #4
0
SGE #1
2000
SGE #2
-5 0
0 10 20 30 40 0 10 20 30 40
Figure A91 − Measured results for test F-iii
- 368 -
A31: Test F-iii
1200 40
 Sum
35
1000  Gap
30  Flxure
Deformation (mm)
800  Shear
25
NA (mm)
600 20
15
400
10
200
5
0 0
0 10 20 30 40 0 10 20 30 40

Figure A92 − Calculated results for test F-iii
Figure A93 – Photos of final damage for test F-iii
- 369 -
A32: Test F-iv
350 1600
Ten #1
300 Ten #2
1500
Ten #3
250 Ten #4
1400
Ten #5
Stress (MPa)
200
1300
150
1200
100
1100
50
0 1000
0 5 10 15 20 25 30 0 5 10 15 20 25 30
15 8000
7000
10 6000
5000
Uplift (mm)
SGS #1
5 4000
SGS #2
3000
SGS #3
0 2000 SGS #4
SGE #1
1000
SGE #2
-5 0
0 5 10 15 20 25 30 0 5 10 15 20 25 30
Figure A94 − Measured results for test F-iv
- 370 -
A32: Test F-iv
1200 30
 Sum
1000 25  Gap
 Flxure
Deformation (mm)
800 20  Shear
NA (mm)
600 15
400 10
200 5
0 0
0 5 10 15 20 25 30 0 5 10 15 20 25 30

Figure A95 − Calculated results for test F-iv
Figure A96 – Photos of final damage for test F-iv
- 371 -
- 372 -
APPENDIX B:
PreWEC Confinement Calculations
B1: Confinment Effectivness

The effectiveness of the confinement depends on the detailing of the transverse and
longitudinal reinforcing bars. Mander et al. [B-1] developed expressions to calculate the
effective core of confined rectangular sections by assuming parabolic surfaces between the
reinforcing bars, as shown in Figure B1. Using the dimensions of the PreWEC confined toe,
shown in Figure B2, the effectiveness coefficient, ke, was calculated, as shown below.
- 373 -
Appendix B: PreWEC Confinement Calculations
Figure B1 – Effective confined core for rectangular sections (by Mander et al. [B-1])
1828.8
368.3
152.4
PreWEC wall cross-section

368.3
# 4 rebar (D13): Area = 127 mm^2
85.1 107.0 85.1
# 3 rebar (D10): Area = 71 mm^2
152.4 89.8
Confined Toe Region
Figure B2 – Details of the confined toe in the PreWEC test specimen
For the PreWEC confined region:
bc  348.3 mm
d c  122.4 mm
s  50.8 mm
s '  40.8 mm
Ac  bc d c  42632 mm 2
- 374 -
B2: Confined Concrete Properties
As 4  71  4  127
 cc    0.01858
Ac 42632
Ai  
n
w 
' 2
i
 2
89.8 2
 4
85.12
 2
107 2
 11332 mm 2
i 1 6 6 6 6
 s'  s' 
Ae   Ac  Ai 1   1    24556 mm 2
 2bc   2d c 
Acc  Ac 1   cc   41840 mm 2
Ae
ke   0.587
Acc
B2: Confined Concrete Properties

Using the ke value calculated above, details of the PreWEC confined toe (shown in Figure B2),
the measured unconfined concrete strength, fc', of 63 MPa, and the measured yield strength of
the transverse stirrups, fyh, of 440 MPa, the confined concrete strength can be calculated. For
rectangular sections, the confining pressure in each direction must be calculated, as shown
below. Mander et al. [B-1] provided a chart to calculate the confined strength from two lateral
confining pressures, as shown in Figure B3.
Asx 2  71
x   0.02284
sd c 50.8  122.4
Asy 4  71
y   0.01605
sbc 50.8  348.3
f lx'  k e  x f yh  5.90 MPa
f ly'  k e  y f yh  4.15 MPa
- 375 -
Appendix B: PreWEC Confinement Calculations
f c'
 1.46 (from Figure B3)
f cc'
f cc'  92 MPa
  f c' 
 cc   co 1  5  '
 1  0.00805

  f cc 
Figure B3 – Confined strength chart for rectangular sections (by Mander et al. [B-1])
B3: References
[B-1] Mander, J. B., Priestley, M. J. N., and Park, R. (1988) Theoretical stress-strain model
for confined concrete. Journal of Structural Engineering, 114(8), 1804-1826.

- 376 -
APPENDIX C:
Ground Motion Scaling Graphs
The scaled response spectra for all five walls that were designed for the parametric study that
was reported in Chapter 7 are plotted in this appendix. As reported in Chapter 7, the ground
motions were scaled to minimise the sum of the squared difference between spectral ordinates
for a time range between the fundamental period, Ti, and the effective period, Te, while also
ensuring that all spectral ordinates were at least 70% of the design response spectrum. The
scaled spectral accelerations for all eight ground motions records as well as the average are
plotted alongside the SEAOC Zone 4 soil type Sc design spectrum for both the design hazard
level EQ-III and the maximum credible hazard level EQ-IV.
- 377 -
Appendix C: Ground Motion Scaling Graphs
Design spectrum
0.7 x Design spectrum
Average
Average (excl. GM-1 & GM-4)
Scaled Sa for each record
1.5
2

1 1.5
0.5
0.5
Ti Te Ti Te
0 0
0.5 1 1.5 0.5 1 1.5
Peroid (s) Peroid (s)

Figure C1 – Scaled spectral accelerations for Wall-1
Design spectrum
Average
1.5
2
1 1.5
0.5
0.5
Ti Te Ti Te
0 0
1 1.5 2 2.5 1 1.5 2 2.5

- 378 -
Design spectrum
Average
1 1.5

0.8
1
0.6
0.4
0.5
0.2
Ti Te Ti Te
0 0
1.5 2 2.5 3 3.5 4 1.5 2 2.5 3 3.5 4

Design spectrum
Average
0.5 0.8
0.7
0.4
0.6
0.5
0.3
0.4
0.2
0.3
0.2
0.1
0.1
Ti Te Ti Te
0 0
2 2.5 3 3.5 4 4.5 5 2 2.5 3 3.5 4 4.5 5

- 379 -
Design spectrum
Average
0.3 0.45
0.4
0.25

0.35
0.2 0.3
0.25
0.15
0.2
0.1 0.15
0.1
0.05
0.05
Ti Te Ti Te
0 0
3 3.5 4 4.5 5 5.5 6 3 3.5 4 4.5 5 5.5 6

- 380 -
APPENDIX D:
Residual Drift Analysis Graphs
The dynamic analysis results that were generated for the parametric study detailed in Chapter 7
and included within this Appendix. These results include the lateral drift time history and the
base moment-lateral drift hysteresis for the following data sets:
 Set Ptest, hazard level EQ-III [walls 1-5 for all eight ground motions].
 Set Ptest, hazard level EQ-IV [walls 1-5 for all eight ground motions].
 Set Pmore, hazard level EQ-III [walls 1-5 for all eight ground motions].
 Set Pmore, hazard level EQ-IV [walls 1-5 for all eight ground motions].
 Set Pless, hazard level EQ-III [walls 1-5 for all eight ground motions].
 Set Pless, hazard level EQ-IV [walls 1-5 for all eight ground motions].
- 381 -
Appendix D: Residual Drift Analysis Graphs
30
2
Base Moment (MN-m)

20
Lateral Drift (%)
1 10
GM-1 0 0
-1 -10
-20
-2
-30
0 20 40 60 80 -2 -1 0 1 2
30
2
Base Moment (MN-m)

20
Lateral Drift (%)
1 10
GM-2 0 0
-1 -10
-20
-2
-30
0 10 20 30 40 50 -2 -1 0 1 2
30
2
Base Moment (MN-m)
20
Lateral Drift (%)
1 10
GM-3 0 0
-1 -10
-20
-2
-30
0 10 20 30 40 50 -2 -1 0 1 2
30
2
Base Moment (MN-m)
20
Lateral Drift (%)
1 10
GM-4 0 0
-1 -10
-20
-2
-30
0 10 20 30 40 50 -2 -1 0 1 2
Figure D1 – Dynamic analysis results for Set Ptest, wall-1, hazard level EQ-III, GM 1-4
- 382 -
Ptest Hazard Level EQ-III
30
2
Base Moment (MN-m)

20
Lateral Drift (%)
1 10
GM-5 0 0
-1 -10
-20
-2
-30
0 10 20 30 40 50 -2 -1 0 1 2
30
2
Base Moment (MN-m)

20
Lateral Drift (%)
1 10
GM-6 0 0
-1 -10
-20
-2
-30
0 5 10 15 20 25 30 -2 -1 0 1 2
30
2
Base Moment (MN-m)
20
Lateral Drift (%)
1 10
GM-7 0 0
-1 -10
-20
-2
-30
0 5 10 15 20 25 30 35 40 -2 -1 0 1 2
30
2
Base Moment (MN-m)
20
Lateral Drift (%)
1 10
GM-8 0 0
-1 -10
-20
-2
-30
0 5 10 15 20 25 30 35 40 -2 -1 0 1 2
- 383 -
30
2
Base Moment (MN-m)

20
Lateral Drift (%)
1 10
GM-1 0 0
-1 -10
-20
-2
-30
0 20 40 60 80 100 -2 -1 0 1 2
30
2
Base Moment (MN-m)

20
Lateral Drift (%)
1 10
GM-2 0 0
-1 -10
-20
-2
-30
0 10 20 30 40 50 60 -2 -1 0 1 2
30
2
Base Moment (MN-m)
20
Lateral Drift (%)
1 10
GM-3 0 0
-1 -10
-20
-2
-30
0 10 20 30 40 50 60 -2 -1 0 1 2
30
2
Base Moment (MN-m)
20
Lateral Drift (%)
1 10
GM-4 0 0
-1 -10
-20
-2
-30
0 10 20 30 40 50 60 -2 -1 0 1 2
- 384 -
30
2
Base Moment (MN-m)

20
Lateral Drift (%)
1 10
GM-5 0 0
-1 -10
-20
-2
-30
0 10 20 30 40 50 60 -2 -1 0 1 2
30
2
Base Moment (MN-m)

20
Lateral Drift (%)
1 10
GM-6 0 0
-1 -10
-20
-2
-30
0 5 10 15 20 25 30 35 40 -2 -1 0 1 2
30
2
Base Moment (MN-m)
20
Lateral Drift (%)
1 10
GM-7 0 0
-1 -10
-20
-2
-30
0 10 20 30 40 50 -2 -1 0 1 2
30
2
Base Moment (MN-m)
20
Lateral Drift (%)
1 10
GM-8 0 0
-1 -10
-20
-2
-30
0 10 20 30 40 50 -2 -1 0 1 2
- 385 -
30
2
Base Moment (MN-m)

20
Lateral Drift (%)
1 10
GM-1 0 0
-1 -10
-20
-2
-30
0 20 40 60 80 100 -2 -1 0 1 2
30
2
Base Moment (MN-m)

20
Lateral Drift (%)
1 10
GM-2 0 0
-1 -10
-20
-2
-30
0 10 20 30 40 50 60 70 -2 -1 0 1 2
30
2
Base Moment (MN-m)
20
Lateral Drift (%)
1 10
GM-3 0 0
-1 -10
-20
-2
-30
0 10 20 30 40 50 60 70 -2 -1 0 1 2
30
2
Base Moment (MN-m)
20
Lateral Drift (%)
1 10
GM-4 0 0
-1 -10
-20
-2
-30
0 10 20 30 40 50 60 70 -2 -1 0 1 2
- 386 -
30
2
Base Moment (MN-m)

20
Lateral Drift (%)
1 10
GM-5 0 0
-1 -10
-20
-2
-30
0 10 20 30 40 50 60 70 -2 -1 0 1 2
30
2
Base Moment (MN-m)

20
Lateral Drift (%)
1 10
GM-6 0 0
-1 -10
-20
-2
-30
0 10 20 30 40 50 -2 -1 0 1 2
30
2
Base Moment (MN-m)
20
Lateral Drift (%)
1 10
GM-7 0 0
-1 -10
-20
-2
-30
0 10 20 30 40 50 60 -2 -1 0 1 2
30
2
Base Moment (MN-m)
20
Lateral Drift (%)
1 10
GM-8 0 0
-1 -10
-20
-2
-30
0 10 20 30 40 50 60 -2 -1 0 1 2
- 387 -
30
2
Base Moment (MN-m)

20
Lateral Drift (%)
1 10
GM-1 0 0
-1 -10
-20
-2
-30
0 20 40 60 80 100 120 -2 -1 0 1 2
30
2
Base Moment (MN-m)

20
Lateral Drift (%)
1 10
GM-2 0 0
-1 -10
-20
-2
-30
0 10 20 30 40 50 60 70 80 -2 -1 0 1 2
30
2
Base Moment (MN-m)
20
Lateral Drift (%)
1 10
GM-3 0 0
-1 -10
-20
-2
-30
0 10 20 30 40 50 60 70 80 -2 -1 0 1 2
30
2
Base Moment (MN-m)
20
Lateral Drift (%)
1 10
GM-4 0 0
-1 -10
-20
-2
-30
0 20 40 60 80 -2 -1 0 1 2
- 388 -
30
2
Base Moment (MN-m)

20
Lateral Drift (%)
1 10
GM-5 0 0
-1 -10
-20
-2
-30
0 10 20 30 40 50 60 70 80 -2 -1 0 1 2
30
2
Base Moment (MN-m)

20
Lateral Drift (%)
1 10
GM-6 0 0
-1 -10
-20
-2
-30
0 10 20 30 40 50 60 -2 -1 0 1 2
30
2
Base Moment (MN-m)
20
Lateral Drift (%)
1 10
GM-7 0 0
-1 -10
-20
-2
-30
0 10 20 30 40 50 60 70 -2 -1 0 1 2
30
2
Base Moment (MN-m)
20
Lateral Drift (%)
1 10
GM-8 0 0
-1 -10
-20
-2
-30
0 10 20 30 40 50 60 70 -2 -1 0 1 2
- 389 -
30
2
Base Moment (MN-m)

20
Lateral Drift (%)
1 10
GM-1 0 0
-1 -10
-20
-2
-30
0 50 100 150 -2 -1 0 1 2
30
2
Base Moment (MN-m)

20
Lateral Drift (%)
1 10
GM-2 0 0
-1 -10
-20
-2
-30
0 20 40 60 80 -2 -1 0 1 2
30
2
Base Moment (MN-m)
20
Lateral Drift (%)
1 10
GM-3 0 0
-1 -10
-20
-2
-30
0 20 40 60 80 -2 -1 0 1 2
30
2
Base Moment (MN-m)
20
Lateral Drift (%)
1 10
GM-4 0 0
-1 -10
-20
-2
-30
0 20 40 60 80 100 -2 -1 0 1 2
- 390 -
30
2
Base Moment (MN-m)

20
Lateral Drift (%)
1 10
GM-5 0 0
-1 -10
-20
-2
-30
0 20 40 60 80 -2 -1 0 1 2
30
2
Base Moment (MN-m)

20
Lateral Drift (%)
1 10
GM-6 0 0
-1 -10
-20
-2
-30
0 10 20 30 40 50 60 70 80 -2 -1 0 1 2
30
2
Base Moment (MN-m)
20
Lateral Drift (%)
1 10
GM-7 0 0
-1 -10
-20
-2
-30
0 10 20 30 40 50 60 70 80 -2 -1 0 1 2
30
2
Base Moment (MN-m)
20
Lateral Drift (%)
1 10
GM-8 0 0
-1 -10
-20
-2
-30
0 20 40 60 80 -2 -1 0 1 2
- 391 -
4 30
Base Moment (MN-m)

20
2
Lateral Drift (%)
10
GM-1 0 0
-10
-2
-20
-4 -30
0 20 40 60 80 -4 -2 0 2 4
4 30
Base Moment (MN-m)

20
2
Lateral Drift (%)
10
GM-2 0 0
-10
-2
-20
-4 -30
0 10 20 30 40 50 -4 -2 0 2 4
4 30
Base Moment (MN-m)
20
2
Lateral Drift (%)
10
GM-3 0 0
-10
-2
-20
-4 -30
0 10 20 30 40 50 -4 -2 0 2 4
4 30
Base Moment (MN-m)
20
2
Lateral Drift (%)
10
GM-4 0 0
-10
-2
-20
-4 -30
0 10 20 30 40 50 -4 -2 0 2 4
Figure D11 – Dynamic analysis results for Set Ptest, wall-1, hazard level EQ-IV, GM 1-4
- 392 -
Ptest Hazard Level EQ-IV
4 30
Base Moment (MN-m)

20
Lateral Drift (%) 2
10
GM-5 0 0
-10
-2
-20
-4 -30
0 10 20 30 40 -4 -2 0 2 4
4 30
Base Moment (MN-m)

20
2
Lateral Drift (%)
10
GM-6 0 0
-10
-2
-20
-4 -30
0 5 10 15 20 25 30 -4 -2 0 2 4
4 30
Base Moment (MN-m)
20
2
Lateral Drift (%)
10
GM-7 0 0
-10
-2
-20
-4 -30
0 5 10 15 20 25 30 35 40 -4 -2 0 2 4
4 30
Base Moment (MN-m)
20
2
Lateral Drift (%)
10
GM-8 0 0
-10
-2
-20
-4 -30
0 5 10 15 20 25 30 35 40 -4 -2 0 2 4
- 393 -
4 30
Base Moment (MN-m)

20
2
Lateral Drift (%)
10
GM-1 0 0
-10
-2
-20
-4 -30
0 20 40 60 80 100 -4 -2 0 2 4
4 30
Base Moment (MN-m)

20
2
Lateral Drift (%)
10
GM-2 0 0
-10
-2
-20
-4 -30
0 10 20 30 40 50 60 -4 -2 0 2 4
4 30
Base Moment (MN-m)
20
2
Lateral Drift (%)
10
GM-3 0 0
-10
-2
-20
-4 -30
0 10 20 30 40 50 60 -4 -2 0 2 4
4 30
Base Moment (MN-m)
20
2
Lateral Drift (%)
10
GM-4 0 0
-10
-2
-20
-4 -30
0 10 20 30 40 50 60 -4 -2 0 2 4
- 394 -
4 30
Base Moment (MN-m)

20
Lateral Drift (%) 2
10
GM-5 0 0
-10
-2
-20
-4 -30
0 10 20 30 40 50 60 -4 -2 0 2 4
4 30
Base Moment (MN-m)

20
2
Lateral Drift (%)
10
GM-6 0 0
-10
-2
-20
-4 -30
0 5 10 15 20 25 30 35 40 -4 -2 0 2 4
4 30
Base Moment (MN-m)
20
2
Lateral Drift (%)
10
GM-7 0 0
-10
-2
-20
-4 -30
0 10 20 30 40 50 -4 -2 0 2 4
4 30
Base Moment (MN-m)
20
2
Lateral Drift (%)
10
GM-8 0 0
-10
-2
-20
-4 -30
0 10 20 30 40 50 -4 -2 0 2 4
- 395 -
4 30
Base Moment (MN-m)

20
2
Lateral Drift (%)
10
GM-1 0 0
-10
-2
-20
-4 -30
0 20 40 60 80 100 -4 -2 0 2 4
4 30
Base Moment (MN-m)

20
2
Lateral Drift (%)
10
GM-2 0 0
-10
-2
-20
-4 -30
0 10 20 30 40 50 60 70 -4 -2 0 2 4
4 30
Base Moment (MN-m)
20
2
Lateral Drift (%)
10
GM-3 0 0
-10
-2
-20
-4 -30
0 10 20 30 40 50 60 70 -4 -2 0 2 4
4 30
Base Moment (MN-m)
20
2
Lateral Drift (%)
10
GM-4 0 0
-10
-2
-20
-4 -30
0 10 20 30 40 50 60 70 -4 -2 0 2 4
- 396 -
4 30
Base Moment (MN-m)

20
Lateral Drift (%) 2
10
GM-5 0 0
-10
-2
-20
-4 -30
0 10 20 30 40 50 60 70 -4 -2 0 2 4
4 30
Base Moment (MN-m)

20
2
Lateral Drift (%)
10
GM-6 0 0
-10
-2
-20
-4 -30
0 10 20 30 40 50 -4 -2 0 2 4
4 30
Base Moment (MN-m)
20
2
Lateral Drift (%)
10
GM-7 0 0
-10
-2
-20
-4 -30
0 10 20 30 40 50 60 -4 -2 0 2 4
4 30
Base Moment (MN-m)
20
2
Lateral Drift (%)
10
GM-8 0 0
-10
-2
-20
-4 -30
0 10 20 30 40 50 60 -4 -2 0 2 4
- 397 -
4 30
Base Moment (MN-m)

20
2
Lateral Drift (%)
10
GM-1 0 0
-10
-2
-20
-4 -30
0 20 40 60 80 100 120 -4 -2 0 2 4
4 30
Base Moment (MN-m)

20
2
Lateral Drift (%)
10
GM-2 0 0
-10
-2
-20
-4 -30
0 10 20 30 40 50 60 70 80 -4 -2 0 2 4
4 30
Base Moment (MN-m)
20
2
Lateral Drift (%)
10
GM-3 0 0
-10
-2
-20
-4 -30
0 10 20 30 40 50 60 70 80 -4 -2 0 2 4
4 30
Base Moment (MN-m)
20
2
Lateral Drift (%)
10
GM-4 0 0
-10
-2
-20
-4 -30
0 20 40 60 80 -4 -2 0 2 4
- 398 -
4 30
Base Moment (MN-m)

20
Lateral Drift (%) 2
10
GM-5 0 0
-10
-2
-20
-4 -30
0 10 20 30 40 50 60 70 80 -4 -2 0 2 4
4 30
Base Moment (MN-m)

20
2
Lateral Drift (%)
10
GM-6 0 0
-10
-2
-20
-4 -30
0 10 20 30 40 50 60 -4 -2 0 2 4
4 30
Base Moment (MN-m)
20
2
Lateral Drift (%)
10
GM-7 0 0
-10
-2
-20
-4 -30
0 10 20 30 40 50 60 70 -4 -2 0 2 4
4 30
Base Moment (MN-m)
20
2
Lateral Drift (%)
10
GM-8 0 0
-10
-2
-20
-4 -30
0 10 20 30 40 50 60 70 -4 -2 0 2 4
- 399 -
4 30
Base Moment (MN-m)

20
2
Lateral Drift (%)
10
GM-1 0 0
-10
-2
-20
-4 -30
0 50 100 150 -4 -2 0 2 4
4 30
Base Moment (MN-m)

20
2
Lateral Drift (%)
10
GM-2 0 0
-10
-2
-20
-4 -30
0 20 40 60 80 -4 -2 0 2 4
4 30
Base Moment (MN-m)
20
2
Lateral Drift (%)
10
GM-3 0 0
-10
-2
-20
-4 -30
0 20 40 60 80 -4 -2 0 2 4
4 30
Base Moment (MN-m)
20
2
Lateral Drift (%)
10
GM-4 0 0
-10
-2
-20
-4 -30
0 20 40 60 80 100 -4 -2 0 2 4
- 400 -
4 30
Base Moment (MN-m)

20
Lateral Drift (%) 2
10
GM-5 0 0
-10
-2
-20
-4 -30
0 20 40 60 80 -4 -2 0 2 4
4 30
Base Moment (MN-m)

20
2
Lateral Drift (%)
10
GM-6 0 0
-10
-2
-20
-4 -30
0 10 20 30 40 50 60 70 80 -4 -2 0 2 4
4 30
Base Moment (MN-m)
20
2
Lateral Drift (%)
10
GM-7 0 0
-10
-2
-20
-4 -30
0 10 20 30 40 50 60 70 80 -4 -2 0 2 4
4 30
Base Moment (MN-m)
20
2
Lateral Drift (%)
10
GM-8 0 0
-10
-2
-20
-4 -30
0 20 40 60 80 -4 -2 0 2 4
- 401 -
30
2
Base Moment (MN-m)

20
Lateral Drift (%)
1 10
GM-1 0 0
-1 -10
-20
-2
-30
0 20 40 60 80 -2 -1 0 1 2
30
2
Base Moment (MN-m)

20
Lateral Drift (%)
1 10
GM-2 0 0
-1 -10
-20
-2
-30
0 10 20 30 40 50 -2 -1 0 1 2
30
2
Base Moment (MN-m)
20
Lateral Drift (%)
1 10
GM-3 0 0
-1 -10
-20
-2
-30
0 10 20 30 40 50 -2 -1 0 1 2
30
2
Base Moment (MN-m)
20
Lateral Drift (%)
1 10
GM-4 0 0
-1 -10
-20
-2
-30
0 10 20 30 40 50 -2 -1 0 1 2
Figure D21 – Dynamic analysis results for Set Pmore, wall-1, hazard level EQ-III, GM 1-4
- 402 -
Pmore Hazard Level EQ-III
30
2
Base Moment (MN-m)

20
Lateral Drift (%)
1 10
GM-5 0 0
-1 -10
-20
-2
-30
0 10 20 30 40 50 -2 -1 0 1 2
30
2
Base Moment (MN-m)

20
Lateral Drift (%)
1 10
GM-6 0 0
-1 -10
-20
-2
-30
0 5 10 15 20 25 30 -2 -1 0 1 2
30
2
Base Moment (MN-m)
20
Lateral Drift (%)
1 10
GM-7 0 0
-1 -10
-20
-2
-30
0 5 10 15 20 25 30 35 40 -2 -1 0 1 2
30
2
Base Moment (MN-m)
20
Lateral Drift (%)
1 10
GM-8 0 0
-1 -10
-20
-2
-30
0 5 10 15 20 25 30 35 40 -2 -1 0 1 2
- 403 -
30
2
Base Moment (MN-m)

20
Lateral Drift (%)
1 10
GM-1 0 0
-1 -10
-20
-2
-30
0 20 40 60 80 100 -2 -1 0 1 2
30
2
Base Moment (MN-m)

20
Lateral Drift (%)
1 10
GM-2 0 0
-1 -10
-20
-2
-30
0 10 20 30 40 50 60 -2 -1 0 1 2
30
2
Base Moment (MN-m)
20
Lateral Drift (%)
1 10
GM-3 0 0
-1 -10
-20
-2
-30
0 10 20 30 40 50 60 -2 -1 0 1 2
30
2
Base Moment (MN-m)
20
Lateral Drift (%)
1 10
GM-4 0 0
-1 -10
-20
-2
-30
0 10 20 30 40 50 60 -2 -1 0 1 2
- 404 -
30
2
Base Moment (MN-m)

20
Lateral Drift (%)
1 10
GM-5 0 0
-1 -10
-20
-2
-30
0 10 20 30 40 50 60 -2 -1 0 1 2
30
2
Base Moment (MN-m)

20
Lateral Drift (%)
1 10
GM-6 0 0
-1 -10
-20
-2
-30
0 5 10 15 20 25 30 35 40 -2 -1 0 1 2
30
2
Base Moment (MN-m)
20
Lateral Drift (%)
1 10
GM-7 0 0
-1 -10
-20
-2
-30
0 10 20 30 40 50 -2 -1 0 1 2
30
2
Base Moment (MN-m)
20
Lateral Drift (%)
1 10
GM-8 0 0
-1 -10
-20
-2
-30
0 10 20 30 40 50 -2 -1 0 1 2
- 405 -
30
2
Base Moment (MN-m)

20
Lateral Drift (%)
1 10
GM-1 0 0
-1 -10
-20
-2
-30
0 20 40 60 80 100 -2 -1 0 1 2
30
2
Base Moment (MN-m)

20
Lateral Drift (%)
1 10
GM-2 0 0
-1 -10
-20
-2
-30
0 10 20 30 40 50 60 70 -2 -1 0 1 2
30
2
Base Moment (MN-m)
20
Lateral Drift (%)
1 10
GM-3 0 0
-1 -10
-20
-2
-30
0 10 20 30 40 50 60 70 -2 -1 0 1 2
30
2
Base Moment (MN-m)
20
Lateral Drift (%)
1 10
GM-4 0 0
-1 -10
-20
-2
-30
0 10 20 30 40 50 60 70 -2 -1 0 1 2
- 406 -
30
2
Base Moment (MN-m)

20
Lateral Drift (%)
1 10
GM-5 0 0
-1 -10
-20
-2
-30
0 10 20 30 40 50 60 70 -2 -1 0 1 2
30
2
Base Moment (MN-m)

20
Lateral Drift (%)
1 10
GM-6 0 0
-1 -10
-20
-2
-30
0 10 20 30 40 50 -2 -1 0 1 2
30
2
Base Moment (MN-m)
20
Lateral Drift (%)
1 10
GM-7 0 0
-1 -10
-20
-2
-30
0 10 20 30 40 50 60 -2 -1 0 1 2
30
2
Base Moment (MN-m)
20
Lateral Drift (%)
1 10
GM-8 0 0
-1 -10
-20
-2
-30
0 10 20 30 40 50 60 -2 -1 0 1 2
- 407 -
30
2
Base Moment (MN-m)

20
Lateral Drift (%)
1 10
GM-1 0 0
-1 -10
-20
-2
-30
0 20 40 60 80 100 120 -2 -1 0 1 2
30
2
Base Moment (MN-m)

20
Lateral Drift (%)
1 10
GM-2 0 0
-1 -10
-20
-2
-30
0 10 20 30 40 50 60 70 80 -2 -1 0 1 2
30
2
Base Moment (MN-m)
20
Lateral Drift (%)
1 10
GM-3 0 0
-1 -10
-20
-2
-30
0 10 20 30 40 50 60 70 80 -2 -1 0 1 2
30
2
Base Moment (MN-m)
20
Lateral Drift (%)
1 10
GM-4 0 0
-1 -10
-20
-2
-30
0 20 40 60 80 -2 -1 0 1 2
- 408 -
30
2
Base Moment (MN-m)

20
Lateral Drift (%)
1 10
GM-5 0 0
-1 -10
-20
-2
-30
0 10 20 30 40 50 60 70 80 -2 -1 0 1 2
30
2
Base Moment (MN-m)

20
Lateral Drift (%)
1 10
GM-6 0 0
-1 -10
-20
-2
-30
0 10 20 30 40 50 60 -2 -1 0 1 2
30
2
Base Moment (MN-m)
20
Lateral Drift (%)
1 10
GM-7 0 0
-1 -10
-20
-2
-30
0 10 20 30 40 50 60 70 -2 -1 0 1 2
30
2
Base Moment (MN-m)
20
Lateral Drift (%)
1 10
GM-8 0 0
-1 -10
-20
-2
-30
0 10 20 30 40 50 60 70 -2 -1 0 1 2
- 409 -
30
2
Base Moment (MN-m)

20
Lateral Drift (%)
1 10
GM-1 0 0
-1 -10
-20
-2
-30
0 50 100 150 -2 -1 0 1 2
30
2
Base Moment (MN-m)

20
Lateral Drift (%)
1 10
GM-2 0 0
-1 -10
-20
-2
-30
0 20 40 60 80 -2 -1 0 1 2
30
2
Base Moment (MN-m)
20
Lateral Drift (%)
1 10
GM-3 0 0
-1 -10
-20
-2
-30
0 20 40 60 80 -2 -1 0 1 2
30
2
Base Moment (MN-m)
20
Lateral Drift (%)
1 10
GM-4 0 0
-1 -10
-20
-2
-30
0 20 40 60 80 100 -2 -1 0 1 2
- 410 -
30
2
Base Moment (MN-m)

20
Lateral Drift (%)
1 10
GM-5 0 0
-1 -10
-20
-2
-30
0 20 40 60 80 -2 -1 0 1 2
30
2
Base Moment (MN-m)

20
Lateral Drift (%)
1 10
GM-6 0 0
-1 -10
-20
-2
-30
0 10 20 30 40 50 60 70 80 -2 -1 0 1 2
30
2
Base Moment (MN-m)
20
Lateral Drift (%)
1 10
GM-7 0 0
-1 -10
-20
-2
-30
0 10 20 30 40 50 60 70 80 -2 -1 0 1 2
30
2
Base Moment (MN-m)
20
Lateral Drift (%)
1 10
GM-8 0 0
-1 -10
-20
-2
-30
0 20 40 60 80 -2 -1 0 1 2
- 411 -
4 30
Base Moment (MN-m)

20
2
Lateral Drift (%)
10
GM-1 0 0
-10
-2
-20
-4 -30
0 20 40 60 80 -4 -2 0 2 4
4 30
Base Moment (MN-m)

20
2
Lateral Drift (%)
10
GM-2 0 0
-10
-2
-20
-4 -30
0 10 20 30 40 50 -4 -2 0 2 4
4 30
Base Moment (MN-m)
20
2
Lateral Drift (%)
10
GM-3 0 0
-10
-2
-20
-4 -30
0 10 20 30 40 50 -4 -2 0 2 4
4 30
Base Moment (MN-m)
20
2
Lateral Drift (%)
10
GM-4 0 0
-10
-2
-20
-4 -30
0 10 20 30 40 50 -4 -2 0 2 4
Figure D31 – Dynamic analysis results for Set Pmore, wall-1, hazard level EQ-IV, GM 1-4
- 412 -
Pmore Hazard Level EQ-IV
4 30
Base Moment (MN-m)

20
Lateral Drift (%) 2
10
GM-5 0 0
-10
-2
-20
-4 -30
0 10 20 30 40 50 -4 -2 0 2 4
4 30
Base Moment (MN-m)

20
2
Lateral Drift (%)
10
GM-6 0 0
-10
-2
-20
-4 -30
0 5 10 15 20 25 30 -4 -2 0 2 4
4 30
Base Moment (MN-m)
20
2
Lateral Drift (%)
10
GM-7 0 0
-10
-2
-20
-4 -30
0 5 10 15 20 25 30 35 -4 -2 0 2 4
4 30
Base Moment (MN-m)
20
2
Lateral Drift (%)
10
GM-8 0 0
-10
-2
-20
-4 -30
0 5 10 15 20 25 30 35 40 -4 -2 0 2 4
- 413 -
4 30
Base Moment (MN-m)

20
2
Lateral Drift (%)
10
GM-1 0 0
-10
-2
-20
-4 -30
0 20 40 60 80 100 -4 -2 0 2 4
4 30
Base Moment (MN-m)

20
2
Lateral Drift (%)
10
GM-2 0 0
-10
-2
-20
-4 -30
0 10 20 30 40 50 60 -4 -2 0 2 4
4 30
Base Moment (MN-m)
20
2
Lateral Drift (%)
10
GM-3 0 0
-10
-2
-20
-4 -30
0 10 20 30 40 50 60 -4 -2 0 2 4
4 30
Base Moment (MN-m)
20
2
Lateral Drift (%)
10
GM-4 0 0
-10
-2
-20
-4 -30
0 10 20 30 40 50 60 -4 -2 0 2 4
- 414 -
4 30
Base Moment (MN-m)

20
Lateral Drift (%) 2
10
GM-5 0 0
-10
-2
-20
-4 -30
0 10 20 30 40 50 60 -4 -2 0 2 4
4 30
Base Moment (MN-m)

20
2
Lateral Drift (%)
10
GM-6 0 0
-10
-2
-20
-4 -30
0 5 10 15 20 25 30 35 40 -4 -2 0 2 4
4 30
Base Moment (MN-m)
20
2
Lateral Drift (%)
10
GM-7 0 0
-10
-2
-20
-4 -30
0 10 20 30 40 50 -4 -2 0 2 4
4 30
Base Moment (MN-m)
20
2
Lateral Drift (%)
10
GM-8 0 0
-10
-2
-20
-4 -30
0 10 20 30 40 50 -4 -2 0 2 4
- 415 -
4 30
Base Moment (MN-m)

20
2
Lateral Drift (%)
10
GM-1 0 0
-10
-2
-20
-4 -30
0 20 40 60 80 100 -4 -2 0 2 4
4 30
Base Moment (MN-m)

20
2
Lateral Drift (%)
10
GM-2 0 0
-10
-2
-20
-4 -30
0 10 20 30 40 50 60 70 -4 -2 0 2 4
4 30
Base Moment (MN-m)
20
2
Lateral Drift (%)
10
GM-3 0 0
-10
-2
-20
-4 -30
0 10 20 30 40 50 60 70 -4 -2 0 2 4
4 30
Base Moment (MN-m)
20
2
Lateral Drift (%)
10
GM-4 0 0
-10
-2
-20
-4 -30
0 10 20 30 40 50 60 70 -4 -2 0 2 4
- 416 -
4 30
Base Moment (MN-m)

20
Lateral Drift (%) 2
10
GM-5 0 0
-10
-2
-20
-4 -30
0 10 20 30 40 50 60 70 -4 -2 0 2 4
4 30
Base Moment (MN-m)

20
2
Lateral Drift (%)
10
GM-6 0 0
-10
-2
-20
-4 -30
0 10 20 30 40 50 -4 -2 0 2 4
4 30
Base Moment (MN-m)
20
2
Lateral Drift (%)
10
GM-7 0 0
-10
-2
-20
-4 -30
0 10 20 30 40 50 60 -4 -2 0 2 4
4 30
Base Moment (MN-m)
20
2
Lateral Drift (%)
10
GM-8 0 0
-10
-2
-20
-4 -30
0 10 20 30 40 50 60 -4 -2 0 2 4
- 417 -
4 30
Base Moment (MN-m)

20
2
Lateral Drift (%)
10
GM-1 0 0
-10
-2
-20
-4 -30
0 20 40 60 80 100 120 -4 -2 0 2 4
4 30
Base Moment (MN-m)

20
2
Lateral Drift (%)
10
GM-2 0 0
-10
-2
-20
-4 -30
0 10 20 30 40 50 60 70 80 -4 -2 0 2 4
4 30
Base Moment (MN-m)
20
2
Lateral Drift (%)
10
GM-3 0 0
-10
-2
-20
-4 -30
0 10 20 30 40 50 60 70 80 -4 -2 0 2 4
4 30
Base Moment (MN-m)
20
2
Lateral Drift (%)
10
GM-4 0 0
-10
-2
-20
-4 -30
0 20 40 60 80 -4 -2 0 2 4
- 418 -
4 30
Base Moment (MN-m)

20
Lateral Drift (%) 2
10
GM-5 0 0
-10
-2
-20
-4 -30
0 10 20 30 40 50 60 70 80 -4 -2 0 2 4
4 30
Base Moment (MN-m)

20
2
Lateral Drift (%)
10
GM-6 0 0
-10
-2
-20
-4 -30
0 10 20 30 40 50 60 -4 -2 0 2 4
4 30
Base Moment (MN-m)
20
2
Lateral Drift (%)
10
GM-7 0 0
-10
-2
-20
-4 -30
0 10 20 30 40 50 60 70 -4 -2 0 2 4
4 30
Base Moment (MN-m)
20
2
Lateral Drift (%)
10
GM-8 0 0
-10
-2
-20
-4 -30
0 10 20 30 40 50 60 70 -4 -2 0 2 4
- 419 -
4 30
Base Moment (MN-m)

20
2
Lateral Drift (%)
10
GM-1 0 0
-10
-2
-20
-4 -30
0 50 100 150 -4 -2 0 2 4
4 30
Base Moment (MN-m)

20
2
Lateral Drift (%)
10
GM-2 0 0
-10
-2
-20
-4 -30
0 20 40 60 80 -4 -2 0 2 4
4 30
Base Moment (MN-m)
20
2
Lateral Drift (%)
10
GM-3 0 0
-10
-2
-20
-4 -30
0 20 40 60 80 -4 -2 0 2 4
4 30
Base Moment (MN-m)
20
2
Lateral Drift (%)
10
GM-4 0 0
-10
-2
-20
-4 -30
0 20 40 60 80 100 -4 -2 0 2 4
- 420 -
4 30
Base Moment (MN-m)

20
Lateral Drift (%) 2
10
GM-5 0 0
-10
-2
-20
-4 -30
0 20 40 60 80 -4 -2 0 2 4
4 30
Base Moment (MN-m)

20
2
Lateral Drift (%)
10
GM-6 0 0
-10
-2
-20
-4 -30
0 10 20 30 40 50 60 70 80 -4 -2 0 2 4
4 30
Base Moment (MN-m)
20
2
Lateral Drift (%)
10
GM-7 0 0
-10
-2
-20
-4 -30
0 10 20 30 40 50 60 70 80 -4 -2 0 2 4
4 30
Base Moment (MN-m)
20
2
Lateral Drift (%)
10
GM-8 0 0
-10
-2
-20
-4 -30
0 20 40 60 80 -4 -2 0 2 4
- 421 -
30
2
Base Moment (MN-m)

20
Lateral Drift (%)
1 10
GM-1 0 0
-1 -10
-20
-2
-30
0 20 40 60 80 -2 -1 0 1 2
30
2
Base Moment (MN-m)

20
Lateral Drift (%)
1 10
GM-2 0 0
-1 -10
-20
-2
-30
0 10 20 30 40 50 -2 -1 0 1 2
30
2
Base Moment (MN-m)
20
Lateral Drift (%)
1 10
GM-3 0 0
-1 -10
-20
-2
-30
0 10 20 30 40 50 -2 -1 0 1 2
30
2
Base Moment (MN-m)
20
Lateral Drift (%)
1 10
GM-4 0 0
-1 -10
-20
-2
-30
0 10 20 30 40 50 -2 -1 0 1 2
Figure D41 – Dynamic analysis results for Set Pless, wall-1, hazard level EQ-III, GM 1-4
- 422 -
Pless Hazard Level EQ-III
30
2
Base Moment (MN-m)

20
Lateral Drift (%)
1 10
GM-5 0 0
-1 -10
-20
-2
-30
0 10 20 30 40 50 -2 -1 0 1 2
30
2
Base Moment (MN-m)

20
Lateral Drift (%)
1 10
GM-6 0 0
-1 -10
-20
-2
-30
0 5 10 15 20 25 30 -2 -1 0 1 2
30
2
Base Moment (MN-m)
20
Lateral Drift (%)
1 10
GM-7 0 0
-1 -10
-20
-2
-30
0 5 10 15 20 25 30 35 40 -2 -1 0 1 2
30
2
Base Moment (MN-m)
20
Lateral Drift (%)
1 10
GM-8 0 0
-1 -10
-20
-2
-30
0 5 10 15 20 25 30 35 40 -2 -1 0 1 2
- 423 -
30
2
Base Moment (MN-m)

20
Lateral Drift (%)
1 10
GM-1 0 0
-1 -10
-20
-2
-30
0 20 40 60 80 100 -2 -1 0 1 2
30
2
Base Moment (MN-m)

20
Lateral Drift (%)
1 10
GM-2 0 0
-1 -10
-20
-2
-30
0 10 20 30 40 50 60 -2 -1 0 1 2
30
2
Base Moment (MN-m)
20
Lateral Drift (%)
1 10
GM-3 0 0
-1 -10
-20
-2
-30
0 10 20 30 40 50 60 -2 -1 0 1 2
30
2
Base Moment (MN-m)
20
Lateral Drift (%)
1 10
GM-4 0 0
-1 -10
-20
-2
-30
0 10 20 30 40 50 60 -2 -1 0 1 2
- 424 -
30
2
Base Moment (MN-m)

20
Lateral Drift (%)
1 10
GM-5 0 0
-1 -10
-20
-2
-30
0 10 20 30 40 50 60 -2 -1 0 1 2
30
2
Base Moment (MN-m)

20
Lateral Drift (%)
1 10
GM-6 0 0
-1 -10
-20
-2
-30
0 5 10 15 20 25 30 35 40 -2 -1 0 1 2
30
2
Base Moment (MN-m)
20
Lateral Drift (%)
1 10
GM-7 0 0
-1 -10
-20
-2
-30
0 10 20 30 40 50 -2 -1 0 1 2
30
2
Base Moment (MN-m)
20
Lateral Drift (%)
1 10
GM-8 0 0
-1 -10
-20
-2
-30
0 10 20 30 40 50 -2 -1 0 1 2
- 425 -
30
2
Base Moment (MN-m)

20
Lateral Drift (%)
1 10
GM-1 0 0
-1 -10
-20
-2
-30
0 20 40 60 80 100 -2 -1 0 1 2
30
2
Base Moment (MN-m)

20
Lateral Drift (%)
1 10
GM-2 0 0
-1 -10
-20
-2
-30
0 10 20 30 40 50 60 70 -2 -1 0 1 2
30
2
Base Moment (MN-m)
20
Lateral Drift (%)
1 10
GM-3 0 0
-1 -10
-20
-2
-30
0 10 20 30 40 50 60 70 -2 -1 0 1 2
30
2
Base Moment (MN-m)
20
Lateral Drift (%)
1 10
GM-4 0 0
-1 -10
-20
-2
-30
0 10 20 30 40 50 60 70 -2 -1 0 1 2
- 426 -
30
2
Base Moment (MN-m)

20
Lateral Drift (%)
1 10
GM-5 0 0
-1 -10
-20
-2
-30
0 10 20 30 40 50 60 70 -2 -1 0 1 2
30
2
Base Moment (MN-m)

20
Lateral Drift (%)
1 10
GM-6 0 0
-1 -10
-20
-2
-30
0 10 20 30 40 50 -2 -1 0 1 2
30
2
Base Moment (MN-m)
20
Lateral Drift (%)
1 10
GM-7 0 0
-1 -10
-20
-2
-30
0 10 20 30 40 50 60 -2 -1 0 1 2
30
2
Base Moment (MN-m)
20
Lateral Drift (%)
1 10
GM-8 0 0
-1 -10
-20
-2
-30
0 10 20 30 40 50 60 -2 -1 0 1 2
- 427 -
30
2
Base Moment (MN-m)

20
Lateral Drift (%)
1 10
GM-1 0 0
-1 -10
-20
-2
-30
0 20 40 60 80 100 120 -2 -1 0 1 2
30
2
Base Moment (MN-m)

20
Lateral Drift (%)
1 10
GM-2 0 0
-1 -10
-20
-2
-30
0 10 20 30 40 50 60 70 80 -2 -1 0 1 2
30
2
Base Moment (MN-m)
20
Lateral Drift (%)
1 10
GM-3 0 0
-1 -10
-20
-2
-30
0 10 20 30 40 50 60 70 80 -2 -1 0 1 2
30
2
Base Moment (MN-m)
20
Lateral Drift (%)
1 10
GM-4 0 0
-1 -10
-20
-2
-30
0 20 40 60 80 -2 -1 0 1 2
- 428 -
30
2
Base Moment (MN-m)

20
Lateral Drift (%)
1 10
GM-5 0 0
-1 -10
-20
-2
-30
0 10 20 30 40 50 60 70 80 -2 -1 0 1 2
30
2
Base Moment (MN-m)

20
Lateral Drift (%)
1 10
GM-6 0 0
-1 -10
-20
-2
-30
0 10 20 30 40 50 60 -2 -1 0 1 2
30
2
Base Moment (MN-m)
20
Lateral Drift (%)
1 10
GM-7 0 0
-1 -10
-20
-2
-30
0 10 20 30 40 50 60 70 -2 -1 0 1 2
30
2
Base Moment (MN-m)
20
Lateral Drift (%)
1 10
GM-8 0 0
-1 -10
-20
-2
-30
0 10 20 30 40 50 60 70 -2 -1 0 1 2
- 429 -
30
2
Base Moment (MN-m)

20
Lateral Drift (%)
1 10
GM-1 0 0
-1 -10
-20
-2
-30
0 50 100 150 -2 -1 0 1 2
30
2
Base Moment (MN-m)

20
Lateral Drift (%)
1 10
GM-2 0 0
-1 -10
-20
-2
-30
0 20 40 60 80 -2 -1 0 1 2
30
2
Base Moment (MN-m)
20
Lateral Drift (%)
1 10
GM-3 0 0
-1 -10
-20
-2
-30
0 20 40 60 80 -2 -1 0 1 2
30
2
Base Moment (MN-m)
20
Lateral Drift (%)
1 10
GM-4 0 0
-1 -10
-20
-2
-30
0 20 40 60 80 100 -2 -1 0 1 2
- 430 -
30
2
Base Moment (MN-m)

20
Lateral Drift (%)
1 10
GM-5 0 0
-1 -10
-20
-2
-30
0 20 40 60 80 -2 -1 0 1 2
30
2
Base Moment (MN-m)

20
Lateral Drift (%)
1 10
GM-6 0 0
-1 -10
-20
-2
-30
0 10 20 30 40 50 60 70 80 -2 -1 0 1 2
30
2
Base Moment (MN-m)
20
Lateral Drift (%)
1 10
GM-7 0 0
-1 -10
-20
-2
-30
0 10 20 30 40 50 60 70 80 -2 -1 0 1 2
30
2
Base Moment (MN-m)
20
Lateral Drift (%)
1 10
GM-8 0 0
-1 -10
-20
-2
-30
0 20 40 60 80 -2 -1 0 1 2
- 431 -
4 30
Base Moment (MN-m)

20
2
Lateral Drift (%)
10
GM-1 0 0
-10
-2
-20
-4 -30
0 20 40 60 80 -4 -2 0 2 4
4 30
Base Moment (MN-m)

20
2
Lateral Drift (%)
10
GM-2 0 0
-10
-2
-20
-4 -30
0 10 20 30 40 50 -4 -2 0 2 4
4 30
Base Moment (MN-m)
20
2
Lateral Drift (%)
10
GM-3 0 0
-10
-2
-20
-4 -30
0 10 20 30 40 50 -4 -2 0 2 4
4 30
Base Moment (MN-m)
20
2
Lateral Drift (%)
10
GM-4 0 0
-10
-2
-20
-4 -30
0 10 20 30 40 50 -4 -2 0 2 4
Figure D51 – Dynamic analysis results for Set Pless, wall-1, hazard level EQ-IV, GM 1-4
- 432 -
Pless Hazard Level EQ-IV
4 30
Base Moment (MN-m)

20
Lateral Drift (%) 2
10
GM-5 0 0
-10
-2
-20
-4 -30
0 10 20 30 40 50 -4 -2 0 2 4
4 30
Base Moment (MN-m)

20
2
Lateral Drift (%)
10
GM-6 0 0
-10
-2
-20
-4 -30
0 5 10 15 20 25 30 -4 -2 0 2 4
4 30
Base Moment (MN-m)
20
2
Lateral Drift (%)
10
GM-7 0 0
-10
-2
-20
-4 -30
0 5 10 15 20 25 30 35 40 -4 -2 0 2 4
4 30
Base Moment (MN-m)
20
2
Lateral Drift (%)
10
GM-8 0 0
-10
-2
-20
-4 -30
0 5 10 15 20 25 30 35 40 -4 -2 0 2 4
- 433 -
4 30
Base Moment (MN-m)

20
2
Lateral Drift (%)
10
GM-1 0 0
-10
-2
-20
-4 -30
0 20 40 60 80 100 -4 -2 0 2 4
4 30
Base Moment (MN-m)

20
2
Lateral Drift (%)
10
GM-2 0 0
-10
-2
-20
-4 -30
0 10 20 30 40 50 60 -4 -2 0 2 4
4 30
Base Moment (MN-m)
20
2
Lateral Drift (%)
10
GM-3 0 0
-10
-2
-20
-4 -30
0 10 20 30 40 50 60 -4 -2 0 2 4
4 30
Base Moment (MN-m)
20
2
Lateral Drift (%)
10
GM-4 0 0
-10
-2
-20
-4 -30
0 10 20 30 40 50 60 -4 -2 0 2 4
- 434 -
4 30
Base Moment (MN-m)

20
Lateral Drift (%) 2
10
GM-5 0 0
-10
-2
-20
-4 -30
0 10 20 30 40 50 60 -4 -2 0 2 4
4 30
Base Moment (MN-m)

20
2
Lateral Drift (%)
10
GM-6 0 0
-10
-2
-20
-4 -30
0 5 10 15 20 25 30 35 40 -4 -2 0 2 4
4 30
Base Moment (MN-m)
20
2
Lateral Drift (%)
10
GM-7 0 0
-10
-2
-20
-4 -30
0 10 20 30 40 50 -4 -2 0 2 4
4 30
Base Moment (MN-m)
20
2
Lateral Drift (%)
10
GM-8 0 0
-10
-2
-20
-4 -30
0 10 20 30 40 50 -4 -2 0 2 4
- 435 -
4 30
Base Moment (MN-m)

20
2
Lateral Drift (%)
10
GM-1 0 0
-10
-2
-20
-4 -30
0 20 40 60 80 100 -4 -2 0 2 4
4 30
Base Moment (MN-m)

20
2
Lateral Drift (%)
10
GM-2 0 0
-10
-2
-20
-4 -30
0 10 20 30 40 50 60 70 -4 -2 0 2 4
4 30
Base Moment (MN-m)
20
2
Lateral Drift (%)
10
GM-3 0 0
-10
-2
-20
-4 -30
0 10 20 30 40 50 60 70 -4 -2 0 2 4
4 30
Base Moment (MN-m)
20
2
Lateral Drift (%)
10
GM-4 0 0
-10
-2
-20
-4 -30
0 10 20 30 40 50 60 70 -4 -2 0 2 4
- 436 -
4 30
Base Moment (MN-m)

20
Lateral Drift (%) 2
10
GM-5 0 0
-10
-2
-20
-4 -30
0 10 20 30 40 50 60 70 -4 -2 0 2 4
4 30
Base Moment (MN-m)

20
2
Lateral Drift (%)
10
GM-6 0 0
-10
-2
-20
-4 -30
0 10 20 30 40 50 -4 -2 0 2 4
4 30
Base Moment (MN-m)
20
2
Lateral Drift (%)
10
GM-7 0 0
-10
-2
-20
-4 -30
0 10 20 30 40 50 60 -4 -2 0 2 4
4 30
Base Moment (MN-m)
20
2
Lateral Drift (%)
10
GM-8 0 0
-10
-2
-20
-4 -30
0 10 20 30 40 50 60 -4 -2 0 2 4
- 437 -
4 30
Base Moment (MN-m)

20
2
Lateral Drift (%)
10
GM-1 0 0
-10
-2
-20
-4 -30
0 20 40 60 80 100 120 -4 -2 0 2 4
4 30
Base Moment (MN-m)

20
2
Lateral Drift (%)
10
GM-2 0 0
-10
-2
-20
-4 -30
0 10 20 30 40 50 60 70 80 -4 -2 0 2 4
4 30
Base Moment (MN-m)
20
2
Lateral Drift (%)
10
GM-3 0 0
-10
-2
-20
-4 -30
0 10 20 30 40 50 60 70 80 -4 -2 0 2 4
4 30
Base Moment (MN-m)
20
2
Lateral Drift (%)
10
GM-4 0 0
-10
-2
-20
-4 -30
0 20 40 60 80 -4 -2 0 2 4
- 438 -
4 30
Base Moment (MN-m)

20
Lateral Drift (%) 2
10
GM-5 0 0
-10
-2
-20
-4 -30
0 10 20 30 40 50 60 70 80 -4 -2 0 2 4
4 30
Base Moment (MN-m)

20
2
Lateral Drift (%)
10
GM-6 0 0
-10
-2
-20
-4 -30
0 10 20 30 40 50 60 -4 -2 0 2 4
4 30
Base Moment (MN-m)
20
2
Lateral Drift (%)
10
GM-7 0 0
-10
-2
-20
-4 -30
0 10 20 30 40 50 60 70 -4 -2 0 2 4
4 30
Base Moment (MN-m)
20
2
Lateral Drift (%)
10
GM-8 0 0
-10
-2
-20
-4 -30
0 10 20 30 40 50 60 70 -4 -2 0 2 4
- 439 -
4 30
Base Moment (MN-m)

20
2
Lateral Drift (%)
10
GM-1 0 0
-10
-2
-20
-4 -30
0 50 100 150 -4 -2 0 2 4
4 30
Base Moment (MN-m)

20
2
Lateral Drift (%)
10
GM-2 0 0
-10
-2
-20
-4 -30
0 20 40 60 80 -4 -2 0 2 4
4 30
Base Moment (MN-m)
20
2
Lateral Drift (%)
10
GM-3 0 0
-10
-2
-20
-4 -30
0 20 40 60 80 -4 -2 0 2 4
4 30
Base Moment (MN-m)
20
2
Lateral Drift (%)
10
GM-4 0 0
-10
-2
-20
-4 -30
0 20 40 60 80 100 -4 -2 0 2 4
- 440 -
4 30
Base Moment (MN-m)

20
Lateral Drift (%) 2
10
GM-5 0 0
-10
-2
-20
-4 -30
0 20 40 60 80 -4 -2 0 2 4
4 30
Base Moment (MN-m)

20
2
Lateral Drift (%)
10
GM-6 0 0
-10
-2
-20
-4 -30
0 10 20 30 40 50 60 70 80 -4 -2 0 2 4
4 30
Base Moment (MN-m)
20
2
Lateral Drift (%)
10
GM-7 0 0
-10
-2
-20
-4 -30
0 10 20 30 40 50 60 70 80 -4 -2 0 2 4
4 30
Base Moment (MN-m)
20
2
Lateral Drift (%)
10
GM-8 0 0
-10
-2
-20
-4 -30
0 20 40 60 80 -4 -2 0 2 4
- 441 -

SC Pre Conc Wall in Building For Low To Medium Seismicity PDF

Încărcat de

Informații document

Titlu original

Drepturi de autor

Formate disponibile

Partajați acest document

Partajați sau inserați document

Opțiuni de partajare

Vi se pare util acest document?

Este necorespunzător acest conținut?

Drepturi de autor:

Formate disponibile

SC Pre Conc Wall in Building For Low To Medium Seismicity PDF

Încărcat de

Drepturi de autor:

Formate disponibile

http://researchspace.auckland.ac.

x Any use you make of these documents or images must be for

To request permissions please use the Feedback form on our webpage.

General copyright and disclaimer

Note : Masters Theses

The digital copy of a masters thesis is as submitted for examination and

A thesis submitted in partial fulfilment of the requirements

for the degree of Doctor of Philosophy

Supervised by Assoc. Prof. Jason Ingham and Prof. Sri Sritharan

Department of Civil and Environmental Engineering

The University of Auckland

A series of experimental tests confirmed the behaviour of individual unbonded PT concrete

Chapter 1 Introduction ............................................................................... 1

Chapter 2 Literature Review ..................................................................... 9

2.7 References ................................................................................................................... 30

Chapter 3 Testing of Post-tensioned Wall Panels.................................... 37

3.6.4 Initial Stiffness...................................................................................................... 73

Chapter 4 Analysis and Design of Post-tensioned Wall Panels .............. 79

Chapter 5 PreWEC Connector Design .................................................. 113

5.6.1 O-Connector Design and Testing....................................................................... 141

Chapter 6 Finite Element Modelling of the PreWEC System ................. 149

6.7.1 PreWEC Core Wall Concept .............................................................................. 203

Chapter 7 Analysis of Residual Drifts .................................................... 213

7.6 Implications to Other Systems .................................................................................. 260

Chapter 8 Analysis of a Self-centering Building ..................................... 265

Chapter 9 Conclusions ........................................................................... 299

9.2 PreWEC System – Chapters 5 & 6 ............................................................................ 302

APPENDIX A: Individual PT Wall Test Results .................................. 307

APPENDIX B: PreWEC Confinement Calculations ............................ 373

APPENDIX C: Ground Motion Scaling Graphs .................................. 377

APPENDIX D: Residual Drift Analysis Graphs ................................... 381

Figure 6.21 – Over-closure contact definition in ABAQUS [6-3] ........................................... 190

Aideal = area of an ideal elasto-plastic hysteresis loop

Aloop = area enclosed by a single hysteresis loop

A ps = prestressing steel area

As = mild reinforcing steel area

c = neutral axis depth at nominal flexural strength

d n , fl = flexural displacement at he corresponding to nominal flexural strength

d n , gap = gap opening displacement at he corresponding to nominal flexural strength

dr = residual displacement at the end of excitation

d r ,max = maximum possible residual displacement

d rr = residual drift ratio (dr / dr, max)

e = eccentricity of load parallel to axis measured from centroid of section

E ps = prestressing steel modulus of elasticity

E SEC ,c = concrete compression unloading secant stiffness (cyclic hysteresis parameter)

E SEC ,t = concrete tension unloading secant stiffness (cyclic hysteresis parameter)

fc = concrete compressive stress

f c' = concrete compressive strength

f cc' = confined concrete compressive strength

fl' = effective lateral confining stress

fm = masonry axial stress

f m' = masonry compressive strength

f ps = prestressing steel stress at nominal flexural strength

f pu = ultimate rupture stress of prestressing steel

f py = yield stress of prestressing steel

f se = initial stress in prestressing steel after losses

ft = concrete tensile stress

ft' = concrete tensile strength

f un = stress prior to unloading (cyclic hysteresis parameter)

fu = ultimate stress of mild reinforcing steel

fy = yield stress of mild reinforcing steel

GC = shear modulus of concrete

he = equivalent or effective wall height