Investigation of Effects of Soil-Structure Interaction On The Seismic Response of RC Bridges Using A Performance-Based Design Approach

INVESTIGATION OF EFFECTS OF SOIL-STRUCTURE INTERACTION ON THE SEISMIC
RESPONSE OF RC BRIDGES USING A PERFORMANCE-BASED DESIGN APPROACH
by
Kianosh Ashkani Zadeh
B.Sc., Civil Engineering, University of Isfahan, 1993
MASc, Structural and Earthquake Engineering, The University of British Columbia, 2013
A THESIS SUBMITTED IN PARTIAL FULFILLMENT OF
THE REQUIREMENTS FOR THE DEGREE OF
Doctor of Philosophy
in
THE FACULTY OF GRADUATE AND POSTDOCTORAL STUDIES
(Civil Engineering)
THE UNIVERSITY OF BRITISH COLUMBIA
(Vancouver)
October 2020
© Kianosh Ashkani Zadeh, 2020

The following individuals certify that they have read, and recommend to the Faculty of Graduate
and Postdoctoral Studies for acceptance, the dissertation entitled:
Investigation of Effects of Soil-Structure Interaction on Seismic Response of RC Bridges
Using a Performance-Based Design Approach
submitted by Kianosh Ashkani Zadeh in partial fulfillment of the requirements for
the degree of Doctor of Philosophy
in Civil Engineering
Examining Committee:
Professor Ronald Clowes, Department of Department of Earth, Ocean and Atmospheric Sciences
Examination Chair
Professor Carlos E. Ventura, Department of Civil Engineering
Supervisor
Professor Liam Finn, Department of Civil Engineering
Co-supervisor
Professor Reza Vaziri, Department of Civil Engineering
University Examiner
Professor Anasavarapu Srikantha Phani, Department of Mechanical Engineering
University Examiner
Additional Supervisory Committee Members:
Professor Farzad Naeim, Department of Civil and Environmental Engineering, UCI
Supervisory Committee Member
Dr. Upul Atukorala, Golder Associates Ltd.
Professor Ahmed Elgamal, Department of Structural Engineering, UCSD
Dr. Anoosh Shamsabadi, California High-Speed Rail
ii
Abstract
Soil-structure interaction (SSI) can have a significant impact on the response of structures
subjected to strong earthquakes. Despite its major effects, SSI is not sufficiently addressed in
current design code and practice. In this study, the effect of SSI on the seismic response of
Reinforced Concrete (RC) bridges is studied within the framework of performance-based
earthquake engineering. This thesis is organized in three phases that investigate three different
aspects of SSI in relation to analysis, assessment and design. The Meloland Road Overcrossing
(MRO) in California is chosen as the case study in this research.
Phase one is focused on investigating the kinematic effect and variation of foundation
motion from the free-field motion. To achieve this, detailed 3D continuum models are developed.
Response time history analyses are performed on the models using ten unscaled ground motions
to investigate variation of bridge foundation motions from free-field motions. The finite element
simulation results show that an amplification of the free-field motions takes place in the low
frequency regime that covers the first few natural frequencies of the system. The tau-averaging
method and Elsabee and Morray Transfer function are unable to predict the amplification regime
observed in the simulations.
In phase two, a discrete simulation approach is adopted to carry out performance
assessment of RC bridges considering soil-structure interaction. Four archetype models with
various levels of SSI representation are developed. Incremental Dynamic Analysis (IDA) is
performed using a set of 22 ground motions to derive collapse fragility curves for each archetype
model. The role of SSI in the calculated collapse fragility curves and corresponding failure modes
is investigated. A FEMA-based collapse assessment procedure is proposed to quantify the
performance of RC bridges.
In phase three, a comprehensive nonlinear continuum model of the MRO is developed.
Seismic response of the continuum model in terms of drift, base shears, and spectral acceleration
is compared to the discrete model developed in previous phase. It is shown that the responses
predicted using the discrete and continuum approaches are significantly different mainly due to
their differences in material constitutive models or representation of SSI effects, specifically the
kinematic effect.
iii
Lay Summary
Despite of soil-structure interaction (SSI) potentially important role on seismic response of

bridges, there is a lack of performance evaluation guidelines in the seismic design provisions.
This research aims to investigate SSI effects on the seismic response of bridges in three
phases. Phase one focuses on the variation of bridge foundation motion from the free-field motion.
Investigation shows an amplification of the bridge foundation motion in the low frequency regime.
In phase two, the role of SSI in the probability of collapse of RC bridges and change of
failure mode is investigated using the discrete method, and a collapse assessment procedure is
proposed for evaluation of seismic performance of RC bridges.
In the third phase, seismic response of a nonlinear continuum model of the MRO bridge in
California is compared with the response obtained from the discrete model. This comparison
showed a significant difference in predicted responses from two models.
iv
Preface
This thesis titled “Investigation of effects of soil-structure interaction on seismic response of RC

bridges using performance-based design approach” presents the research performed by Kianosh
Ashkani Zadeh. The research was supervised by Professor Carlos Ventura and co-supervised by
Professor Liam Finn of Department of Civil Engineering at the University of British Columbia,
Vancouver, Canada.
Some parts of this thesis, listed below, have been published in conference proceedings by the same
authors.
– A summary of the research presented in Chapter 3 has been published as a conference
proceeding in the 12th. Canadian Conference on Earthquake Engineering (12CCEE):
Ashkani Zadeh, K., Ventura, C., and Finn, L. “Effect of Kinematic Soil-Structure
Interaction on the Foundation Motion of Integral Abutment Bridges” proceedings of the
12th. Canadian Conference on Earthquake Engineering (12CCEE), Québec City, Québec,
Canada, in June 2019. I was the principal author of the paper and conducted all the
numerical analyses and writing of the paper. Professors Ventura and Finn supervised this
research and aided in revising this manuscript.
– A version of Chapters 4 and 5 has been summarized into a conference proceeding published
by the 17WCEE in September 2020: Ashkani Zadeh, K., Ventura, C., and Finn, L.
“Collapse Assessment of RC Bridges Considering Soil Structure Interaction (SSI)”
proceedings of the 17th. World Conference on Earthquake Engineering (17WCEE), Sendai,
Japan. I was the principal author of the paper and conducted all the numerical analyses and
writing of the paper. Professors Ventura and Finn supervised this research and aided in
revising this manuscript.
– A version of Chapter 6 has been summarized into a paper published by the 17WCEE in
September 2020: Ashkani Zadeh, K., Ventura, C., and Finn, L. “Evaluation of Seismic
Response of Soil+Bridge System Using 3D Discrete and Continuum Simulation
Approaches” proceedings of the 17th. World Conference on Earthquake Engineering
(17WCEE), Sendai, Japan. I was the principal author of the paper and conducted all the
numerical analyses and writing of the paper. Professors Ventura and Finn supervised this
research and aided in revising this manuscript.
v
Table of Contents
Abstract ......................................................................................................................................... iii
Lay Summary ............................................................................................................................... iv
Preface .............................................................................................................................................v
Table of Contents ......................................................................................................................... vi
List of Tables .............................................................................................................................. xiii
List of Appendix Tables............................................................................................................ xvii
List of Figures ........................................................................................................................... xviii
List of Appendix Figures ....................................................................................................... xxviii
List of Symbols ........................................................................................................................ xxxii
List of Abbreviations .................................................................................................................. xli
Acknowledgements ................................................................................................................... xliv
Dedication ................................................................................................................................... xlv
Chapter 1: Introduction ................................................................................................................1
1.1 Overview, and Motivations ............................................................................................. 1
1.2 Design Codes and SSI Effect .......................................................................................... 3
1.3 Implementation of SSI Effects in Analysis and Design.................................................. 4
1.4 Goals, Objectives, and Tasks .......................................................................................... 7
1.5 Thesis organization .................................................................................................... 8
Chapter 2: Literature Survey .....................................................................................................12
2.1 Scientific Understanding of SSI: Early History to Current State ................................. 12
2.2 Kinematic SSI Effect .................................................................................................... 16

vi
2.3 Simulating SSI Effects .................................................................................................. 16
2.4 Summary of Key SSI Research Milestones .................................................................. 19
2.5 SSI in Design Codes ..................................................................................................... 21
2.6 Gaps in Knowledge ....................................................................................................... 25
Chapter 3: Investigation of Variation of Bridge Foundation Motion from the Free-Field
Motion ...........................................................................................................................................27
3.1 Introduction and Methodology ..................................................................................... 27
3.2 Meloland Road Overcrossing (MRO) ........................................................................... 29
3.3 Continuum Model to Investigate Kinematic Effects .................................................... 31
3.3.1 Elastic Model Description......................................................................................... 32
3.3.2 Eigenvalue Analysis on both Soil Free-Field and Bridge+Soil Models ................... 35
3.3.3 Choice of Ground Motions ....................................................................................... 37
3.3.4 Damping and Elastic Analysis .................................................................................. 39
3.4 Results ........................................................................................................................... 41
3.4.1 Effect of Kinematic Soil-Structure Interaction on Foundation Motion: Amplification
or Reduction? ........................................................................................................................ 41
3.4.2 Applicability of Tau-averaging Method to Estimate Foundation Motion ................ 46
3.5 Chapter Summary ......................................................................................................... 55
Chapter 4: Collapse Assessment of the Meloland Road Overcrossing (MRO) ......................57
4.1 Introduction and Methodology ..................................................................................... 57
4.2 Development of nonlinear Archetype Models for Collapse Assessment ..................... 61
4.2.1 Choice of Software for Discrete Approach ............................................................... 63
4.2.2 Element Type ............................................................................................................ 64

vii
4.2.3 Nonlinear Geometry Model ...................................................................................... 67
4.2.4 Material Properties .................................................................................................... 67
4.2.5 Sectional Response of Pier Column and Abutment Backwalls ................................ 69
4.2.6 Structural Damping ................................................................................................... 71
4.2.7 SSI Features in D1, D2 and D3 Index Archetype Models .......................................... 75
4.2.8 SSI Features in the D4 Index Archetype Model ........................................................ 75
4.2.8.1 Idealized Soil Layers......................................................................................... 76
4.2.8.2 Simulation of the Abutment-backfill Soil Interaction ...................................... 77
4.2.8.3 Soil-pile interaction ........................................................................................... 83
4.2.8.3.1 Lateral soil-pile interaction ......................................................................... 85
4.2.8.4 Vertical Soil-Pile Interaction ............................................................................ 91
4.2.8.4.1 Pile skin Resistance ..................................................................................... 91
4.2.8.4.2 Pile Tip Resistance ...................................................................................... 94
4.3 Eigen Value Analysis .................................................................................................... 95
4.4 Performance Criteria ..................................................................................................... 96
4.4.1 Bridge Components Limit State................................................................................ 97
4.4.2 Shear and chord rotation capacity of bridge members............................................ 101
4.5 Pushover Analysis ....................................................................................................... 101
4.6 Hazard Analysis .......................................................................................................... 103
4.7 IDA results .................................................................................................................. 109
4.8 Identification and comparison of Structural Failure Modes in the archetype models 116
4.9 Fragility Curves .......................................................................................................... 121
4.10 Effect of Archetype Model on Fragility Curves ......................................................... 123

viii
4.11 Chapter Summary ....................................................................................................... 126
Chapter 5: Performance Evaluation of the Bridge Systems ..................................................129
5.1 Introduction and Methodology ................................................................................... 129
5.2 Collapse Margin Ratio ................................................................................................ 129
5.3 Adjusted Collapse Margin Ratio (ACMR) ................................................................. 134
5.4 Proposed Simplified Bridge Seismic Performance Evaluation Procedure (SBSPEP) 136
5.5 Example Application: Evaluation Seismic Performance of the Meloland Road
Overcrossing (MRO) .............................................................................................................. 149
5.5.1 Performance-Based Design Requirement Required Performance Level (PL) and
Seismic Retrofit Categories (SRC) For the MRO Bridge ................................................... 149
5.5.2 Developing Maximum Considerable Earthquake (MCE) For MRO ...................... 151
5.5.3 Nonlinear Analysis and Predicting Parameters ( ˆ and ˆ ) (Step 2.) ..................... 152
5.5.4 Calculation of CMRacceptable (Step 3.2) .................................................................... 152
5.5.5 Calculating Spectral shape factor (SSF) (Step 3.3) ................................................ 153
5.5.5.1 Mean Expected Epsilon,  0 ( T ) .................................................................... 153
5.5.5.2 Calculating  ( T1 )records , β coefficients, and SSF ............................................ 156
5.5.6 Check Validity of ACMRs (Step 3.4) ..................................................................... 156
5.5.7 Comparison of Predicted ACMR and ACMRacceptable (Step 3.5) ............................ 158
5.5.8 Calculation of Collapse Capacity (Sam(T1)collapse ) (Step 3.6) ............................. 159
5.5.9 Documentation and Peer Review (Step 4) .............................................................. 160
5.6 Chapter Summary ....................................................................................................... 160
Chapter 6: Nonlinear Response of the MRO Using A Continuum Model ...........................162
ix
6.1 Introduction and Methodology ................................................................................... 162
6.2 Continuum Model (C Model ) of the Meloland Road Overcrossing .......................... 166
6.2.1 Nonlinear Material Models for Structural Members .............................................. 169
6.2.1.1 Unconfined and Confined Concrete Model .................................................... 169
6.2.1.2 Concrete Damage Plasticity (CDP) Model ..................................................... 174
6.2.1.3 Concrete Damage Plasticity Model Verification ............................................ 177
6.2.1.4 Mesh Dependency and Localization Issue ...................................................... 177
6.2.1.5 Nonlinear model for Steel Rebars ................................................................... 181
6.2.2 Nonlinear Material Models for Soil ........................................................................ 184
6.3 Collapse-level Ground Motions .................................................................................. 189
6.4 Nonlinear Dynamic Analysis on the Continuum Model............................................. 191
6.4.1 Pushover Analysis of Pier Column and Abutment Backwall ................................. 191
6.4.2 Time History Analysis (THA) ................................................................................ 193
6.5 Results and Discussion ............................................................................................... 193
6.5.1 Investigation of Damage in the Main Structural Members ..................................... 193
6.5.2 Comparison of Nonlinear Dynamic Analysis response from the continuum and
Discrete Models .................................................................................................................. 199
6.6 Chapter Summary ....................................................................................................... 205
Chapter 7: Summary, Conclusions and Future Work ...........................................................207
7.1 Summary ..................................................................................................................... 207
7.2 Concluding Remarks ................................................................................................... 208
7.3 Statement of Contributions ......................................................................................... 212
7.4 Limitations, Recommendations and Future Work ...................................................... 213

x
Bibliography ...............................................................................................................................215
Appendices ..................................................................................................................................233
Appendix A Meloland Road Overcrossing (MRO) Drawing Details ..................................... 233
A.1 MRO layout and Elevation View ............................................................................ 233
A.2 MRO Deck and Girder Outline ............................................................................... 234
A.3 Deck and Pier RC Detail ......................................................................................... 235
A.4 Abutment and Wingwalls Outline and RC detail ................................................... 236
Appendix B Time Histories of Ground Motions used for Analyses ....................................... 237
B.1 Time History of Ground Motions are Used for Elastic Analyses of the Continuum
Model E to Investigate Variation of the Bridge Foundation Motion from the Free-Field
Motion (Phase 1 Study) ...................................................................................................... 237
B.2 Time History of Ground Motions used for IDA Analyses of the Discrete Models
(Phase 2 Study) ................................................................................................................... 238
B.3 Time History of Ground Motions Used for Nonlinear Analyses of the Continuum
Model C at Collapse Level (Phase 3 Study) ....................................................................... 241
Appendix C Variation of Bridge Foundation Motion from The Free-Field Motion .............. 242
C.1 MRO Abutment Footing Foundation ...................................................................... 242
C.2 MRO Pier Pilecap Foundation ................................................................................ 243
Appendix D Calculation of Reinforced Concrete Member Capacitates ................................. 244
D.1 Performance Criteria for Determining Bridge Shear Strength Capacity ................ 244
D.2 Performance Criteria for Determining Bridge Displacement Capacity .................. 247
Appendix E AASHTO (2017) Site Class Definitions............................................................. 253
Appendix F California Mean Predicted ε0 Values for Periods of 0.2s and 1.0 s ................... 254
xi
Appendix G Relationship between ln (Sc(T1)) and ε(T1) ....................................................... 255
G.1 D1 model ................................................................................................................. 255
G.2 D2 model.................................................................................................................. 255
G.3 D3 model ................................................................................................................. 256
Appendix H Collapse Failure Mode of Discrete Archetype Models D1, D2, D3, and D4 ....... 257
H.1 Effect of the Numerical Models on Failure Modes and Their Sequence of Occurrence
257
H.2 Effect of Earthquake Ground Motions on Failure Modes and Their Sequence of
Occurrence .......................................................................................................................... 260
H.3 Sequence of Failure for All the Collapsed Level Ground Motions and Index Archetype
Models................................................................................................................................. 264
Appendix I Concrete Tensile and Compressive Damage Results for the Continuum Model C
................................................................................................................................................. 266
I.1 Tensile Damage Results .......................................................................................... 266
I.2 Compressive Damage Results................................................................................. 270
Appendix J Results of Flagged Yielded Soil Layers at Collapse Level ................................. 274
Appendix K Execution Time of the Continuum Models (C, E, and F Model) ....................... 279
Appendix L Validation of ABAQUS One-way Infinite Element CN3D8 for the Quite
Boundaries (SIMULIA, 2018) ................................................................................................ 280
xii
List of Tables
Table 1-1 Capabilities of the different SSI modeling methods for a large 3D bridge model ......... 6
Table 2-1 A summary of research on SSI at UBC since 2000...................................................... 21
Table 3-1 Material properties of the soil layers (Kwon and Elnashai, 2008) ............................... 35
Table 3-2 Period and mode shape of the soil and Soil+Bridge models ........................................ 36
Table 3-3 The selected ground motions and their magnitudes and mechanisms .......................... 38
Table 3-4 The selected ground motions and their rupture distances and characteristics .............. 38
Table 3-5 Defined Rayleigh damping coefficients in the Soil and Soil+Bridge models .............. 40
Table 3-6 Error in estimated motion using Tau-averaging method compare to foundation motion
calculated using the FE model ...................................................................................................... 55
Table 4-1 Assessment framework for performance-based earthquake engineering ..................... 59
Table 4-2 Proposed seismic design performance criteria by the Seismic Subcommittee of the
CHBDC (CSA S6-06, 2006) ......................................................................................................... 60
Table 4-3 Nonlinear concrete material properties for the MRO pier column and abutment
backwalls (Werner et al., 1993) .................................................................................................... 68
Table 4-4 Parameters of assigned elastic material model to the deck slab, foundations and piles68
Table 4-5 Material properties of reinforcing steel bars for the MRO archetype models .............. 69
Table 4-6 Modal frequencies (ωj) and damping ratios percentage (ξj) of Meloland Road
Overcrossing (Zhang and Makris, 2002) ...................................................................................... 72
Table 4-7 Spring and dashpot values that approximate the presence of the approach embankments
and pile foundation for the Meloland Road Overcrossing (Zhang and Makris, 2002) ................. 75
Table 4-8 EHFD Coefficients Back-Calculated Using Log-Spiral Hyperbolic (LSH) simulation
modified by a wall height-adjustment factor (Shamsabadi et al., 2010) ...................................... 78
Table 4-9 Calculated parameters to define fully characterize the tri-linear symmetric response
curve for pier pilecap and footing foundation of Meloland Road Overcrossing .......................... 83
Table 4-10 Equivalent pile diameter considered for each soil layer in calculating pile load-
displacement relationship in the index archetype model D4 ......................................................... 84
Table 4-11 Lateral soil resistance-deflection relationships for piles in soft clay (API, 2007) ..... 88
Table 4-12 Calculated parameters to define p-y curve for clay layers surrounding the abutment and
pier piles in the index archetype model D4 ................................................................................... 90
xiii
Table 4-13 Calculated parameters to define p-y curve for soil layer B and D surrounding the
abutment and pier piles in the index archetype model D4 ............................................................ 91
Table 4-14 Piles’ skin rsistance in clay and sand layers (API, 2007) .......................................... 91
Table 4-15 Calculated parameters to define t-z curve for the soil layers A, C and E surrounding
the abutment and pier piles in the index archetype model D4....................................................... 93
Table 4-16 Defined parameters to define t-z curve for soil layer B and D surrounding the abutment
and pier piles in the index archetype model D4 ............................................................................ 93
Table 4-17 Tip-load-Displacement relationship for sand and clay soils (API, 2007) .................. 94
Table 4-18 Model parameters for Q-z curve for abutment and pier piles in the index archetype
model D4 ....................................................................................................................................... 95
Table 4-19 Modal period of the models D1, D2, D3, and D4 along with modal period obtained from
Ambient Vibration Test (AVT) for the MRO ............................................................................... 96
Table 4-20 Performance criteria for the Pier Column of the MRO .............................................. 99
Table 4-21 Performance criteria for the Abutment Backwall of the MRO ................................ 100
Table 4-22 Performance criteria for deck slab and pier diaphragm of the MRO ....................... 100
Table 4-23. Calculated plastic hinge length and hinge to total length ratio for the pier column.102
Table 4-24 Summary of the selected ground motions using NGA-West 2 ground motion database
(PEER, 2019) .............................................................................................................................. 106
Table 4-25 Total and significant duration values of the input time histories used in IDA analyses
of the MRO discrete models ....................................................................................................... 107
Table 4-26 Comparison of longitudinal top pier drift and its corresponding base shear of the index
archetype model D4 using the entire and the significant duration of Imperial Valley-06 ground
motion ......................................................................................................................................... 109
Table 4-27 Summary of IDA log for each index archetype model............................................. 110
Table 4-28 Calculated IDA parameters for the models .............................................................. 122
Table 4-29 Sequence of failure modes of the index archetype models D1 and D4 at onset of collapse
..................................................................................................................................................... 124
Table 5-1 5% damped 6th generation seismic hazard values (in g unit) for the major Canadian
cities, for 2% in 50 year probability on Site Class C (Adams et al., 2019) ................................ 142
xiv
Table 5-2 Proposed total system collapse uncertainty (𝛽TOT) based on quality of model and design
for the period-based ductility, T  3 (FEMA P695, 2009) ..................................................... 145
Table 5-3 Required procedures of the proposed seismic performance methodology for RC bridges.
..................................................................................................................................................... 147
Table 5-4 Proposed seismic design performance criteria by the Seismic Subcommittee of the
CHBDC (CSA S6-06, 2006) ....................................................................................................... 149
Table 5-5 Seismic Performance Categories ................................................................................ 150
Table 5-6 Performance-based Seismic Hazard Categories based on Table 1-6 of the Seismic
retrofitting manual for highway structures-bridges (FHWA, 2006) ........................................... 151
Table 5-7 AASHTO coefficients and parameter for Meloland response spectrum .................... 151
Table 5-8 Calculated value of  ( T1 )records using different attenuation equations. ...................... 157
Table 5-9 Calculated value of β1 using different attenuation equations ..................................... 157
Table 5-10 Calculated Spectral Shape Factors, SSF, using different attenuation equations ...... 158
Table 5-11 ACMR, ACMRaceptable and their ratio (ACMR/ ACMRaceptable) corresponding to MCE
level (2% in 50 years), Sa(T1)MCE=2.65g .................................................................................... 159
Table 5-12 Summarized modified collapse capacity of the archetype models using different
attenuation relationships ............................................................................................................. 160
Table 6-1 Section and material properties are used in the MRO continuum model (Werner et al.,
1993). .......................................................................................................................................... 167
Table 6-2 Comparison of the modal period of different simulated FE models and vibration test of
the MRO...................................................................................................................................... 169
Table 6-3 Comparison of different representation of rebars in a FE continuum model ............. 183
Table 6-4 Five M-C plasticity parameters of the soil layers defined in the nonlinear continuum
model (C model) ......................................................................................................................... 188
Table 6-5 Collapse level ground motions used in nonlinear analyses of the MRO continuum model
..................................................................................................................................................... 190
Table 6-6 Total and significant duration values of the input time histories used in nonlinear
analyses of the MRO continuum model...................................................................................... 190
Table 6-7 Comparison of the MRO pier column ultimate moment and shear capacity using
different software ........................................................................................................................ 192
xv
Table 6-8 Comparison of the maximum drift of the discrete and continuum model at the collapse
level ............................................................................................................................................. 201
xvi
List of Appendix Tables
Table E- 1Site Class Definitions specified in Table 3.10.3.1-1 AASHTO (AASHTO, 2017) ... 253
Table F- 1 Mean Predicted ε0 values for periods of 0.2s and 1.0 s, sorted by Seismic Design
Category (SDC), for California sites and selected California cities for site class B (Haselton et al.,
2011) ........................................................................................................................................... 254
Table H- 1 Sequence of failure modes of the index archetype models D1, D2, D3, and D4 at onset
of collapse ................................................................................................................................... 264
Table K- 1 Ground motions used in the continuum analyses ..................................................... 279
Table L- 1 Calculated moduli and dashpot parameters .............................................................. 282

Table L- 2 Calculated moduli and dashpot parameters .............................................................. 282
xvii
List of Figures
Figure 1-1 Failure of Mission Gothic bridge column due to the Northridge shallow crustal
earthquake in California (Aschheim, 1994) .................................................................................... 1
Figure 1-2 (a) Collapse of the Cypress Structure, the freeway approach to the Bay Bridge from
Oakland due to Loma Prieta Earthquake (Steinbrugge, 1989), (b) Collapse of a 630m long segment
of the elevated Hanshin Expressway during Kobe Earthquake Japan (Tsikada and Shimbun, 1995).
......................................................................................................................................................... 2
Figure 1-3 Diagram of investigating of effect soil-structure interaction ........................................ 7
Figure 1-4 Flow chart of the study phases and their corresponding chapters ............................... 11
Figure 2-1 Timeline of some SSI key studies have been carried out on response of structures ... 20
Figure 3-1 Kinematic SSI main causes (a) base slab averaging, (b) wave scattering, (c) embedment
effect, and (d) vertically propagated shear waves. Image reproduced from (Kramer, 1996) ....... 28
Figure 3-2 Various aspects of kinematic effect in bridges ........................................................... 28
Figure 3-3 (a) Meloland Road Overcrossing (MRO) in the background and a kink in Highway 8
due to continuous crippling of the exposed Imperial fault, (b) Panoramic view of MRO (Ventura
et al., 2011) ................................................................................................................................... 30
Figure 3-4 California quaternary fault and location of the bridge and alignment of the Imperial
Fault (USGS Interactive Fault Map., 2019) .................................................................................. 30
Figure 3-5 MRO plan, elevation and section along with sensor locations (Ventura et al., 2011) 31
Figure 3-6. a) Discretized layout plan and 3D view of the soil and Soil+Bridge structure (SSI)
model showing finite element mesh, b) massless structure modeled in the SSI model and
measurement points, c) discretized layout plan and infinite element boundary walls of the SSI
model, and d) discretized layout plan and infinite element boundary walls of the soil free-field
model and measurement points..................................................................................................... 33
Figure 3-7 3D Continuum model simulated in ABAQUS as Meloland Road Overcrossing (MRO)
bridge reference model and surrounding infinite boundary element walls................................... 34
Figure 3-8 Mode shapes of the first three modes of the continuum Soil+bridge combined model
(SSI model) ................................................................................................................................... 36
Figure 3-9 Response spectra of the selected input motion with 5% damping and periods of interest.
....................................................................................................................................................... 39
xviii
Figure 3-10 Identification of damping properties of the MRO: (a) Damping ratio (ξ) vs. frequency
for the soil layers, (b) Modal damping vs bridge frequencies obtained from the Ambient Vibration
Test (Ventura et al., 2011), (c) and (d) An impact type earthquake loading used to identify damping
ratio (ξ) of the superstructure based on the calculated decaying (Δu) by Kwon and Elnashai (2008),
respectively. .................................................................................................................................. 40
Figure 3-11 Elevation view and boundary conditions (a) SSI model, (b) free-field soil model... 41
Figure 3-12 Transmissibility factors calculated from RTH analyses for a) abutment footing
foundation and b) pier foundation................................................................................................. 42
Figure 3-13 Average amplitude of transmisibility function plotted as a dimensionless unique
function of α=ɷD/Va (a) Abutment foundation with D/Va=0.004s (b) Pier foundation with
D/Va=0.02s .................................................................................................................................... 44
Figure 3-14 Comparison of the calculated transmissibility factor with the transfer function
amplitude model by Elsabee and Morray (1977) for the (a) abutment foundation, (b) pier
foundation ..................................................................................................................................... 45
Figure 3-15 Comparison of averaging model with observed Spectral motions at the foundation
base of the Hollywood storage building during the 1994 Northridge Earthquake in the North-South
direction (Pandey, 2013). .............................................................................................................. 47
Figure 3-16 Rigid rectangular foundation of a large structure ..................................................... 47
Figure 3-17 τ-factor as a function of frequency and apparent wave velocity (Clough and Penzien,
1995). ............................................................................................................................................ 50
Figure 3-18 Schematic Tau-average method procedures to estimate foundation motion from free
field motion ................................................................................................................................... 51
Figure 3-19 Schematic concept of base averaging in small and large foundations ...................... 51
Figure 3-20 a) calculated Tau parameter for abutment and pier foundations, Comparison of Hector
Mine spectrum with 5% damping of a foundation motion with the free-field and estimated
foundation motion using Tau-averaging method, b) Abutment footing foundation, and c) pier
pilecap foundation ......................................................................................................................... 52
Figure 3-21 Calculated accelerations for the various depths illustrating the embedment effect due
to Imperial Valley-06 ground motion ........................................................................................... 53
xix
Figure 3-22 One dimensional shear beam site amplification. Abutment embankment is not shown
for clarity ....................................................................................................................................... 53
Figure 3-23 Spectra of the Abutment footing foundation, free-field, and estimated foundation
based on the Tau-averaging method with 5% damping ................................................................ 54
Figure 3-24 Spectra of the Pier pilecap foundation, free-field, and estimated foundation based on
the Tau-averaging method with 5% damping ............................................................................... 54
Figure 4-1 A typical structural performance and associated behavior and damage states (after
Ghobarah, 2001) ........................................................................................................................... 60
Figure 4-2 An elevation view of idealized model: (a) Viscoelastic embankments and center bent,
(b) and (c) Elastic support at embankments and center bent ........................................................ 62
Figure 4-3 General framework of the study research presented in Chapter 4 .............................. 62
Figure 4-4 Different modeling methods of the structural components (NEHRP, 2013). ............. 64
Figure 4-5 Discretization of a typical reinforced concrete cross-section (SeismoSoft, 2020) ..... 65
Figure 4-6 Section discretization for the MRO's index archetype models (a) Pier pilecap, (b) Pier
column, and (c) Abutment backwalls ........................................................................................... 67
Figure 4-7 Sectional response of pier column of the Meloland Road Overcrossing (a) AASHTO-
99 LRFD M-V interaction (Response-2000, 2001) , (b) Moment-Curvature diagram calculated by
Werner et al (1993) and Response-2000 software (2019) ............................................................ 70
Figure 4-8 Sectional response of abutment backwalls of the Meloland Road Overcrossing (a)
AASHTO-99 LEFD M-V interaction (Response-2000, 2001) , (b) Moment-Curvature diagram
calculated by using Response-2000 software (2019).................................................................... 70
Figure 4-9 First six modal frequencies, damping ratios, and modes computed for stick model (left)
and 3D FEM model (right) of the MRO model by Zhang and Makris (2002) ............................. 74
Figure 4-10 3D view of the index archetype model D4 constructed using SeismoStruct software.
....................................................................................................................................................... 76
Figure 4-11 Reproduced description of soil layers on the geotechnical borehole log B2 of the MRO
(Caltrans, 1969)............................................................................................................................. 77
Figure 4-12 Idealized soil layers for the MRO and elevational view of the index archetype model
D4 .................................................................................................................................................. 77
xx
Figure 4-13 Caltrans nonlinear abutment model (a) seat-type abutment (b) diaphragm-type
abutment (Caltrans, 2019a) ........................................................................................................... 79
Figure 4-14 Effective Abutment height and width (Caltrans, 2019a) .......................................... 81
Figure 4-15 Nonlinear abutment response in longitudinal direction of the Meloland Road
Overcrossing (MRO) .................................................................................................................... 81
Figure 4-16 Force-displacement relationship for pier pilecap and footing foundation of the MRO
....................................................................................................................................................... 82
Figure 4-17 Tri-linear symmetric curve employed to model idealized trilinear behavior in
SeismoStruct adopting an isotropic hardening rule (SeismoSoft, 2020) ...................................... 82
Figure 4-18 Abutment backwalls’ springs and API springs arrangement are considered in the
archetype model D4 to simulate Abutment-backfill Soil Interaction and soil-pile interaction,
respectively ................................................................................................................................... 85
Figure 4-19 Coefficients used to calculate the ultimate resistance, pu (API, 2007) ..................... 86
Figure 4-20 Relationship between initial modulus of subgrade reaction and internal friction angle
(API, 2007) ................................................................................................................................... 87
Figure 4-21 Calculated p-y curves for soil layers for the index archetype model D4 surrounded the
(a) abutment piles, (b) pier piles ................................................................................................... 90
Figure 4-22 A symmetric bi-linear curve with its three required parameters to fully characterize
this response curve in SeismoStruct software (SeismoSoft, 2020)............................................... 91
Figure 4-23 The relationship between mobilized soil-pile shear transfer and local pile deflection
for the abutment and pier piles...................................................................................................... 94
Figure 4-24 Calculated pile tip load-displacement for abutment and pier of the Meloland Road
Overcrossing ................................................................................................................................. 95
Figure 4-25 Calculated initial and final shear strength of pier column of the MRO .................... 98
Figure 4-26 Force-displacement curve for the MRO simulated in SeismoStruct commercial
software (a) Pier column with axial load=5382.4kN (b) Abutment backwall with axial
load=2691.2kN ........................................................................................................................... 103
Figure 4-27 Significant historic Earthquakes greater than magnitude 5.5 between 1800 and 2000
in California (Toppozada and Branum, 2008) ............................................................................ 104
Figure 4-28 Spectra (5% damping) of the selected ground motions for IDA ............................. 105
xxi
Figure 4-29 Comparison of pier column drifts of the archetype model D4 in the longitudinal and
transverse directions, simulated using entire ground motion duration and significant duration. 108
Figure 4-30 Hysteresis curves of the pier column calculated performing response time history
analysis on the index archetype model D4 using Imperial Valley-06 ground motion (a) transverse
direction (b) longitudinal direction ............................................................................................. 108
Figure 4-31 Comparison of Collapse Scale Factors (CSFs) ratio relative to model D4 ............. 111
Figure 4-32 IDA curves for pier column of the MRO in x-direction (across the bridge deck) for
the index archetype models: (a) D1, (b) D2, (c) D3, and (d) D4. .................................................. 112
Figure 4-33 IDA curves for pier column of the MRO in y-direction (along the bridge deck) for the
index archetype models: (a) D1, (b) D2, (c) D3, and (d) D4......................................................... 113
Figure 4-34 IDA curves for the MRO abutment backwall of the MRO in y-direction (across the
bridge deck) for the index archetype models: (a) D1, (b) D2, (c) D3, and (d) D4. ....................... 114
Figure 4-35 Calculated hysteresis curves of the pier column using D1 and D4 model at their
corresponding collapse levels using (a) Imperial Valley-06 ground motion in the transverse
direction (b) Imperial Valley-06 ground motion in the longitudinal direction, (c) El Mayor ground
motion in the longitudinal direction (d) El Mayor ground motion in the transverse direction. .. 115
Figure 4-36 Failure mode and their sequence of occurrence for the models when subjected to the
Imperial Valley-06, Superstition Hills-02, Landers, and El Mayor earthquake ground motions at
the collapse levels corresponding to each model ........................................................................ 117
Figure 4-37 Failure mode and their sequence of occurrence for the models when subjected to the
San Fernando, Northridge-01, Loma Prieta, and Parkfield earthquake ground motions at the
collapse levels corresponding to each model .............................................................................. 118
Figure 4-38 Failure mode sequence of the models D1, D2, D3, and D4 due to the collapse level
ground motions (a) Imperial Valley-06 (SF= 1.4, 1.8, 1.6 and 1.6), (b) El Mayor-5630 (SF= 3.6,
3.6, 4.2 and 3.6), (c) Superstition Hills (SF= 1.8, 1.6, 1.6 and 1.4) and (d) Landers (SF= 2, 2, 2.6
and 1.4) ....................................................................................................................................... 119
Figure 4-39 Failure mode sequence of the models D1, D2, D3, and D4 due to the collapse level
ground motions (a) San Fernando (SF=1, 1, 1, and 1), (b) Northridge (SF= 1.4, 1.2, 1.2 and 1.4),
(c) Loma Prieta (SF= 1.4, 1.6, 1.6 and 1.4) and (d) Parkfield (SF= 2, 1.8, 2 and 1.6) .............. 120
Figure 4-40 Fragility curves for the index archetype models D1 to D4 for the MRO Bridge. .... 123
xxii
Figure 4-41 Fragility curves for the index archetype models D1 and D4 for the MRO Bridge. . 125
Figure 5-1 Procedure for calculating Collapse Margin Ratio (CMR) ........................................ 130
Figure 5-2 Reduction of CMR and increase of Probability of Collapse in SSI archetype model D 1
and D4.......................................................................................................................................... 132
Figure 5-3 Reproduced IDA collapse fragility curve (curve a) with modifications to account for
modeling uncertainty (curve b) and spectral shape effects (curve c) (Deierlein et al., 2008) .... 133
Figure 5-4 Imperial Valley-06 spectrum with 5% damping at the collapse level (SF=1.6) ,
computed mean spectrum, and predicted epsilon ....................................................................... 135
Figure 5-5 Proposed Simplified Bridge Seismic Performance Evaluation Procedure (SBSPEP)
..................................................................................................................................................... 140
Figure 5-6 AASHTO response spectrum for probability of exceedance 2% in 50 years (reproduced
based on AASHTO, 2017). ......................................................................................................... 141
Figure 5-7 Design spectrum with 5% damping and probability of exceedance 2% in 50 years for
Victoria, Vancouver, and Queen Charlotte Island (Haida Gwaii) in the British Columbia ...... 143
Figure 5-8 Detailed procedures for Step 2 to 4 of the proposed Simplified Bridge Seismic
Performance Evaluation Procedure (SBSPEP) ........................................................................... 148
Figure 5-9 Developed MCE using AASHTO LRFD (2017) spectrum with 5% damping and design
spectrum (2/3 MCE) for the MRO site ....................................................................................... 152
Figure 5-10 Relationship between total uncertainty of the system (βTOT) and acceptable collapse
margin ration (CMRacceptable) for the index archetype models in logarithmic scale ................... 153
Figure 5-11 Predicted  0 for the Western United States and for period of 1.0s and a 2% in 50 years
(Harmsen et al., 2002). ................................................................................................................ 154
Figure 5-12 Seismic deaggregation for MRO hazard site using Dynamic for 2% in 50 years hazard
level based on Conterminous U.S. 2008 (v3.3.1) data (USGS, 2019). ....................................... 155
Figure 5-13 Expected epsilon value versus period ..................................................................... 155
Figure 5-14 Estimated β0 and β1 coefficients (intercept and slope of the linear regression line) for
the D4 archetype model ............................................................................................................... 156
Figure 6-1 3D simulation of the MRO developed by Kwon and Elnashai (2008) using
multiplatform soil+bridge system technique .............................................................................. 163
xxiii
Figure 6-23D Soil+bridge continuum model of the PSO developed and used by Jeremic and co-
workers (2009) ............................................................................................................................ 163
Figure 6-3 Numerical simulation of the Humboldt Bay Middle Channel Bridge using (a) 2D
Continuum model developed by Zhang et al. (2008) and (b) 3D Continuum model developed by
Elgamal et al. (2008). .................................................................................................................. 164
Figure 6-4 3D continuum SSI model of Meloland Road Overcrossing visualized by the GiD
interactive graphical user interface (Rahmani, 2014) ................................................................. 165
Figure 6-5 Developed 3D nonlinear continuum model (C) using ABAQUS software as Meloland
Road Overcrossing (MRO) bridge reference model. .................................................................. 168
Figure 6-6 Unconfined concrete compression stress-strain response model (a) Hognestad (1951),
(b) Modified Hognestad, and (c) Todeschini et al. (1964) ......................................................... 170
Figure 6-7 Calculated unconfined concrete compression response using different analytical
approximation methods based on considering concrete strength of 35.62 MPa......................... 171
Figure 6-8 Stress-strain relationship of plain and reinforced concrete in tension and tensile
stiffening with a cut-off (Adebar, 2015). .................................................................................... 172
Figure 6-9 Simplified tensile stress-strain model (a) Nayal and Rasheed (2006) tension stiffening
model, (b) Wahalathantri et. al (2011) tension stiffening model ................................................ 173
Figure 6-10 Simplified tensile stress-strain relationship proposed by Allam et al. (2012) ........ 173
Figure 6-11 Stress-strain curve for the unconfined concrete main structural members of MRO
based on concrete compression strength ( f c ) and ultimate strain ( ε cu ) value 35.6MPa and
0.003mm/mm, respectively. Compression strength is considered a negative value. .................. 174

Figure 6-12 Behavior of concrete under axial strength (a) compressive and (b) tension (SIMULIA,
2008) ........................................................................................................................................... 175
Figure 6-13 Comparison of stress-strain curves of ABAQUS input with (a) single element
subjected uniaxial compression, (b) the element subjeted to uniaxial tension. .......................... 177
Figure 6-14Maximum principal plastic strain at capacity of the abutment backwall with (a) fine
mesh size (0.3m along the height of wall) (b) coarse mesh size (0.9m along the height of wall)
..................................................................................................................................................... 178
xxiv
Figure 6-15 Maximum principal plastic strain at capacity of the pier column with (a) fine mesh
size (0.3m along the height of the column) (b) coarse mesh size (0.9m along the height of the
column) ....................................................................................................................................... 179
Figure 6-16 Effect of mesh size in force-displacement response of the structural components (a)
pier column (Vertical load=5382.3kN), and (b) abutment backwall (Vertical load=2691.2kN) 179
Figure 6-17 Comparison of the simulation of the pier steel cage in the MRO 3D continuum model
(a) 3D solid representation approach (b) Wire (1D) embedded reinforcing steel bars’ approach
..................................................................................................................................................... 182
Figure 6-18 Reinforcing steel bars’ stress-strain response curve considered in the 3D continuum
model........................................................................................................................................... 184
Figure 6-19 Real and idealized stress-strain behavior of Soils (a) real response, (b) linear elastic,
(c) nonlinear elastic, (d) Perfectly plastic response, and (e) hardening response (Desai and Zaman,
2014) ........................................................................................................................................... 185
Figure 6-20 Mohr-Coulomb model (SIMULIA, 2018) .............................................................. 186
Figure 6-21 Mohr-Coulomb deviatoric plane. Image reproduced from ABAQUS user’s manual
(SIMULIA, 2018) ....................................................................................................................... 187
Figure 6-22 Embankment and supporting soil layers in the simulated nonlinear continuum model
of the MRO ................................................................................................................................. 188
Figure 6-23 Investigation stress-strain response of soils using direct shear test on a unit cube model
..................................................................................................................................................... 189
Figure 6-24 Force-displacement curve for the pier and abutment backwall component of the MRO
simulated in ABAQUS and SeismoStruct commercial software (a) Pier column with axial
load=5382.4kN (b) Abutment backwall with axial load=2691.2kN .......................................... 191
Figure 6-25 Monitoring performance of the simulated 1D wire reinforcing steel bars using truss
element (T3D2) embedded in concrete after performing pushover analysis on the components
using ABAQUS software (a) pier column (f= 2.95 MPa) and (b) abutment backwall (f= 0.
0.57MPa)..................................................................................................................................... 192
Figure 6-26 Predicted concrete (a) tensile damage and (b) compressive damage parameters of the
MRO continuum model at the collapsed level using El Mayor Earthquake (SF=3.6). .............. 194
xxv
Figure 6-27 Predicted concrete (a) tensile damage and (b) compressive damage parameters of the
MRO continuum model at the collapsed level using Imperial Valley-06 Earthquake (SF=1.6). 195
Figure 6-28 Maximum principal stress in the reinforcing steel bars (a) abutment wall steel cage (b)
Pier steel cage due to the Imperial Valley-06 Earthquake at collapse level (SF =1.6) ............... 196
Figure 6-29 Maximum principal strain in the reinforcing steel bars (a) abutment wall steel cage (b)
Pier steel cage due to the Imperial Valley-06 Earthquake at collapse level (SF =1.6) ............... 197
Figure 6-30 Yielding state of various soil layers around the piles subjected to collapse level El
Mayor ground motion (SF=3.6). Red color is an indicator of yielding state around the piles and the
observed time in the calculated acceleration time histories. ....................................................... 198
Figure 6-31 Yielding state of various soil layers around the piles subjected to collapse level
Imperial Valley-06 ground motion (SF=1.6). Red color is an indicator of yielding state around the
piles and the observed time in the calculated acceleration time histories................................... 198
Figure 6-32 Spectra with 5% damping of the bottom pier pilecap using the collapsed level ground
motions (a) Imperial Valley-06 with 1.6 scale factor in the transverse direction (b) Imperial
Valley-06 with 1.6 scale factor in the longitudinal direction (c) El Mayor with 3.6 scale factor in
the transverse direction (d) El Mayor with 3.6 scale factor in the longitudinal direction ......... 199
Figure 6-33 Comparison of calculated drift using C and D4 models and collapse level ground
motion (a) pier column drift in the transverse direction (b) pier column drift in the longitudinal
direction (c) abutment backwall drift in the transverse direction and (d) abutment backwall drift
in the longitudinal direction. ....................................................................................................... 200
Figure 6-34 Comparison maximum base shear results at the collapse level (a) pier column shear in
the transverse direction (b) pier column shear in the longitudinal direction (c) abutment backwall
shear in the transverse direction (d) abutment backwall shear in the longitudinal direction ...... 202
Figure 6-35 Calculated hysteresis curves of the pier column using the C and D4 model (a) Imperial
Valley-06 with SF=1.6 in the transverse direction (b) Imperial Valley-06 with SF=1.6 in the
longitudinal direction (c) El Mayor with SF=3.6 in the transverse direction (d) El Mayor with
SF=3.6 in the longitudinal direction ........................................................................................... 203
Figure 6-36 Calculated hysteresis curves of the abutment backwall using the C and D4 model (a)
Imperial Valley-06 with SF=1.6 in the transverse direction (b) Imperial Valley-06 with SF=1.6 in
xxvi
the longitudinal direction (c) El Mayor with SF=3.6 in the transverse direction (d) El Mayor with
SF=3.6 in the longitudinal direction ........................................................................................... 204
xxvii
List of Appendix Figures
Figure A- 1 Meloland Road Crossing (MRO) dimension detail (a) Lauout plan view, (b) Elvation
view (Caltrans, 1969) .................................................................................................................. 233
Figure A- 2 MRO Concrete outline and RC detail (a) Typical deck slab's section, (b) Girder layout
(c) Deck longitudinal section (Caltrans, 1969) ........................................................................... 234
Figure A- 3 MRO concrete outline and RC detail (a) Sectional elevation at Pier, (b) Pier footing
foundation plan, and (c) Pier column section (Caltrans, 1969) .................................................. 235
Figure A- 4 MRO Abutment and Wing walls RC detail (a) Abutment plan, (b) Abutment section,
(c) Wingwall elevation, (d) Wingwall section, and (e) Wingwall corner detail (Caltrans, 1969)
..................................................................................................................................................... 236
Figure B- 1 Ground motions used as input motions in response time history analyses of the elastic
3D continuum model (E model) of the MRO ............................................................................. 237
Figure B- 2 Imperial Valley-06, Superstition Hills-02, Loma Prieta, Northridhe, Bam-Iran, Cape
Mendocino, Chi Chi-Taiwan, and Christchurch New Zealand time histories (PEER, 2018) .... 238
Figure B- 3 Duzce-Turkey, El Mayor-Cucapah Mexico, Gazli-USSR, Hector Mine, Imperial
Valley-02, Irpinia Italy-01, Kobe Japan, and Kocaeli turkey time histories (PEER, 2018) ...... 239
Figure B- 4 Landers, Nahanni Canada, Parkfield, San Fernando, Spitak-Armenia and Tabas-Iran
time histories (PEER, 2018) ....................................................................................................... 240
Figure B- 5 Collapse level ground motions used as input motion in response time history analyses
of the nonlinear 3D continuum model (C Model) of the MRO .................................................. 241
Figure C- 1 Spectra of the Abutment footing foundation, free-field, and estimated foundation based
on the Tau-averaging method with 5% damping ........................................................................ 242
Figure C- 2 Spectra of the Pier pilecap foundation, free-field, and estimated foundation based on
the Tau-averaging method with 5% damping ............................................................................. 243
Figure G- 1 Linear regression analysis for determining β0 and β1 coefficients (intercept and slope
of the linear regression line) for the D1 archetype model .......................................................... 255
Figure G- 2 Regression analysis for determining β0 and β1 coefficients (intercept and slope of the
linear regression line) for the D2 archetype model .................................................................... 255
Figure G- 3 Regression analysis for determining β0 and β1 coefficients (intercept and slope of the
linear regression line) for the D3 archetype model ..................................................................... 256
xxviii
Figure H- 1 Failure modes sequence of the index archetype models D1, D2, D3, and D4 due to the
collapse level ground motions (a) Hector Mine (SF=2, 1.6, 1.8 and 1.6), (b) Cape Mendocino (SF=
2.4, 2, 2 and 1.6), (c) Christchurch (SF= 2.2, 2, 1.8and 2.2) and (d) El Centro (SF= 2.8, 2.4, 2.6
and 2.2) ....................................................................................................................................... 257
collapse level ground motions (a) Kobe (SF=0.8, 1.4, 1 and 1.2), (b) Duzce (SF=1.4, 1.4, 1.2 and
0.8), (c) Kocaeli (SF=1.8, 1.8, 2 and 1.6) and (d) Bam (SF=1.2, 1.8, 1.4 and 1.2) ................... 258
collapse level ground motions (a) Irpinia (SF= 4.2, 4, 4 and 3.4), (b) Gazli (SF= 1.4, 1.6, 1.8 and
1.4), (c) Tabas (SF= 2, 1.8, 1.8 and 2) and (d) Nahanni (SF= 1.8, 1.6, 1.4 and 1.2) .................. 259
collapse level ground motions (a) Spitak (SF= 4.4, 3.8, 3 and 3.2), and (b) Chi Chi (SF= 1, 0.8,
1and 1) ........................................................................................................................................ 259
Figure H- 5 Failure modes and their sequence of occurrence for the index archetype models when
subjected to the Hector Min, Cape Mendocino, and El Centro earthquake ground motions at their
collapse levels corresponding to each archetype model ............................................................. 260
subjected to the Kobe, Duzce, Kocaeli, and Bam earthquake ground motions at their collapse levels
corresponding to each archetype model ...................................................................................... 261
subjected to the Irpinia, Gazli, Tabas, and Nahanni earthquake ground motions at their collapse
levels corresponding to each archetype model ........................................................................... 262
subjected to the Spitak, Chi Chi, and Christchurch earthquake ground motions at their collapse
levels corresponding to each archetype model ........................................................................... 263
xxix
Figure I- 1 MRO continuum model’s concrete tensile damage at the collapsed level (a) Hector
Mine (Scale Factor=1.6) (b) Landers (Scale Factor=1.4) ........................................................... 266
Figure I- 2 MRO continuum model’s concrete tensile damage at the collapsed level (a) Parkfield
(Scale Factor=1.6) (b) Superstition Hills (Scale Factor=1.4) ..................................................... 267
Figure I- 3 MRO continuum model’s concrete tensile damage at the collapsed level (a) Bam (Scale
Factor=1.2) (b) Kobe (Scale Factor=1.2).................................................................................... 268
Figure I- 4 MRO continuum model’s concrete tensile damage at the collapsed level (a) Duzce
(Scale Factor=0.8) (b) Kocaeli (Scale Factor=1.6) ..................................................................... 269
Figure I- 5 MRO continuum model’s concrete compressive damage at the collapsed level (a)
Hector Mine (Scale Factor=1.6) (b) Landers (Scale Factor=1.4) ............................................... 270
Figure I- 6 MRO continuum model’s concrete compressive damage at the collapsed level (a)
Parkfield (Scale Factor=1.6) (b) Superstition Hills (Scale Factor=1.4) ..................................... 271
Figure I- 7 MRO continuum model’s concrete compressive damage at the collapsed level (a) Bam
(Scale Factor=1.2) (b) Kobe (Scale Factor=1.2)......................................................................... 272
Figure I- 8 MRO continuum model’s concrete compressive damage at the collapsed level (a) Duzce
(Scale Factor=0.8) (b) Kocaeli (Scale Factor=1.6) ..................................................................... 273
Figure J- 1 Yielding state of various soil layers around the piles subjected to collapse level Hector
Mines ground motion (SF=1.6). Red color is an indicator of yielding state around the piles and the
Figure J- 2 Yielding state of various soil layers around the piles subjected to collapse level Landers
ground motion (SF=1.4). Red color is an indicator of yielding state around the piles and the
Figure J- 3 Yielding state of various soil layers around the piles subjected to collapse level
Parkfield ground motion (SF=1.6). Red color is an indicator of yielding state around the piles and
the observed time in the calculated acceleration time histories. ................................................. 275
xxx
Figure J- 4 Yielding state of various soil layers around the piles subjected to collapse level
Superstition Hills Ground Motion (SF=1.4). Red color is an indicator of yielding state around the
piles and the observed time in the calculated acceleration time histories................................... 276
Figure J- 5 Yielding state of various soil layers around the piles subjected to collapse level Bam
Figure J- 6 Yielding state of various soil layers around the piles subjected to collapse level Kobe
Figure J- 7 Yielding state of various soil layers around the piles subjected to collapse level Duzce
Figure J- 8 Yielding state of various soil layers around the piles subjected to collapse level Kocaeli
Figure K- 1 Wallclock time for the continuum models using a workstation with 24 CPUs @ 3.0
GHz and 512 GB RAM............................................................................................................... 279
Figure L- 1 (a) Continuum single element model with a CIN3D8 one-way infinite element
extension, (b) Continuum single element model with dashpots connected to the ground at the base
of the model (SIMULIA, 2018). ................................................................................................. 280
Figure L- 2 Comparison of the calculated harmonic amplitude at x-direction at the node A for T1
and T2 Models ............................................................................................................................. 283
Figure L- 3 Comparison of the calculated harmonic amplitude at y-direction at the node A for T1
and T2 Models 283
xxxi
List of Symbols
A Factor to account for cyclic or static loading condition

Abh Area of one spiral or hoop bar
Ae Effective shear area
Ag Gross area for rectangular and circular sections
Ap Gross end area of pile
A( ) Amplitude of Transmissibility function
a x ( t ) Acceleration of the rigid base in x-direction
a r & br Back-calculated coefficients for the abutment backwall reference height

c Undrained shear strength, depth from the extreme compression fiber to the neutral axis, or
cohesion
C Damping matrix
Cx Dashpot damping value in x-direction
Cy Dashpot damping value in y-direction
Cz Dashpot damping value in z-direction
Ci=1 to 3 Coefficients determined as function of internal friction of sand
D Pile diameter, foundation dimension, or overall depth of section
D´ Distance between the outer layers of longitudinal steel in a rectangular section
D″ Diameter of transverse hoop or spiral
Dequivalent Equivalent diameter of pile
Di Discrete index archetype model i ( i=1to 4)
db Diameter of the longitudinal tension reinforcing steel bars
ds Cover of the column rebars
dc Compressive damage
dt Tensile damage
d' Depth from the extreme compression fiber to the center of the compression reinforcing
steel bars
d1 First branch displacement limit in a tri-linear response curve
d2 Second branch displacement limit in a tri-linear response curve
xxxii
E Modulus of elasticity
Ec Undamaged modulus of elasticity
E0 Initial modulus of elasticity
Ect Tangent modulus
EI Flexural Strength
e Embedment or deviatoric eccentricity
Fy Soil yield strength
F(y) Abutment embankment induced lateral force
Fabut Idealized ultimate passive capacity of the backfill behind abutment backwall or diaphragm
Fa Coefficient stands for acceleration responses of the soil to short periods
Fv Coefficients stands for acceleration responses of the soil to long periods
Fpga Coefficients stands for acceleration responses of the soil to zero periods
fc Compressive strength
f´c Expected concrete strength
f´cc Cconfined concrete strength
f´ce Unconfined concrete compressive strength
fct Concrete tensile stress
fct,max Maximum concrete tensile stress
fcr crack stress
ft Tensile strength
fy Yield strength
fyh Yield stress of the transverse hoops or spirals
fi-Soil Soil’s ith. mode frequency
G Shear Modulus or modulus of rigidity
Gmax Maximum Shear Modulus
Gf Fracture energy
gf Fracture energy density
H Depth of soil layer
Hw(ω) Transfer function of the embedded shallow foundations
Hw(ω) Amplitude of the Transfer function

xxxiii
hd Height of the damaged zone
he Height of element
hw Height of wall
h abut Abutment backwall height
hdia* Effective height of abutment backwall if diaphragm is not designed for full soil pressure
hdia** Effective height of backwall if diaphragm is designed for full soil pressure
i Unit imaginary number ( −1 )
Ig Gross moment of inertia
Icrack Crack moment of inertia
IA Arias Intensity
J Dimensionless empirical constant
jd Internal lever arm
k Initial modulus of subgrade
ke Confinement effectiveness coefficient for spirals and hoop steel
K Stiffness matrix
Kx Spring stiffness value in x-direction
Ky Spring stiffness value in y-direction
Kz Spring stiffness value in z-direction
Kr Spring rotational stiffness value
Kxr Spring stiffness value about x-direction
Kyr Spring stiffness value about y-direction
Kabut Stiffness of abutment due to embankment's passive pressure force resisting movement
behind the abutment backwall
Keff Abutment effective stiffness
K0 Initial stiffness
K1 Second branch stiffness in a tri-linear response curve
K2 Third branch stiffness in a tri-linear response curve
L Length of a structural element (pile, abutment, column, etc.) and shear span or effective
height
xxxiv
Lp Plastic hinge length
M Earthquake Magnitude Scale (Richter magnitude scale)
M Mass matrix
Mp Plastic moment capacity
N standard penetration resistance
Nf Effective number of equal-amplitude cycles of loading that lead to fracture
Nq Pile end bearing factor
p Actual pile lateral resistance or equivalent pressure stress or axial compression load on
member
pe Axial load on the section
PLi Performance Level i (i=0,1, 2, and 3)
p´0 Effective overburden pressure at the pile tip depth
pus Ultimate resistance in kN/m for shallow depth
pud Ultimate resistance in kN/m for deep depth
P Pier column axial load
Pacceptable Acceptable probability of collapse
P-Δ Geometrical nonlinearity

Pu Soil ultimate lateral bearing capacity
PI Plasticity index
Q Mobilized end bearing capacity
Qp Total pile end bearing capacity
q Mises equivalent stress
R sk Skew reduction factor and for abutments passive capacity with skew angle (   66 )
r Deviatoric stress
Rrup Distance to the rupture plane
S Stress deviator matrix
s Center-to-center longitudinal spacing of the transverse hoop steels or spirals
S (T) The design spectral response acceleration, expressed as a ratio to gravitational
acceleration, for a period of T
xxxv
Sa (T) The 5% damped spectral response acceleration, expressed as a ratio to gravitational
acceleration, for a period of T
SD1 Parameter to define Seismic design categories based on AASHTO (2014)
SDS Parameter to define Seismic hazard level based on FHWA (2006) (AASHTO, 2014)
Su Un-drained shear strength
SCT Spectral acceleration correspond to median collapse to the 5%-damped spectral
acceleration of the MCE ground motions
SCT10 % Spectral acceleration corresponds to10% probability of collapse
SCT20 % Spectral acceleration corresponds to 20% probability of collapse
c
Sa 50% Median value of spectral acceleration at the fundamental period
SaMT Intensity at MCE-level ground motions correspond to fundamental period of a structureT

Period of a structure
SaTau Predicted spectral acceleration- Tau-averaging method
Sam (T1) Modified collapse capacity
Sa(T1)MCE Spectral acceleration at the fundamental period
SaFound. Predicted spectral acceleration- Bridge foundation
S1 Horizontal response spectral acceleration coefficient at 1.0 sec period on rock (Site Class
B)
Ss Horizontal response spectral acceleration coefficient at 0.2-sec period on rock (Site Class
B),
t Mobilized soil pile adhesion
tmax Maximum soil pile adhesion or unit skin friction capacity
Tm Period of vibration of mth. mode
Tn Natural period of vibration of the bridge
Tp Predominant period of a ground motion record
T1 Fundamental natural period
TF( ) Transmissibility function
u bf Bridge foundation acceleration
u gx Free-field ground acceleration
xxxvi
u ff Free-field acceleration
V Base shear of the structure

Va Apparent velocity
Vf Final shear capacity
Vi Initial shear
Vm Shear demand
Vp Shear ccontribution provided by arch (strut) action
Vcf Final shear strength carried by the concrete
Vci Shear resistance provided by the diagonal tension field in the concrete
Vs Rebar truss action contribution to the shear strength or shear wave velocity
VS30 Average shear wave velocity of top 30 meters of the site
Wabut Width of the abutment wall
Wbw Width of the backwall of the seat-type abutments
Wdia. Width of the backwall of the diaphragm abutments
X Depth below soil surface
y Abutment longitudinal displacements
y50 Deflection at one-half the ultimate resistance
Z Local pile sliding

α Corner-to-corner strut angle and ratio of average concrete stress in compression zone to
confined concrete strength
α & β Raleigh damping coefficients
β Shaft friction factor or standard deviation of ln Sa (dispersion of Sa ) and depth of stress
block
ˆ Standard deviation of the normal distribution representing the ln IM values
β RTR Record-to-record collapse uncertainty
β DR Design requirements-related collapse uncertainty
β TD Test data-related collapse uncertainty
β MDL Modeling-related collapse uncertainty
β TOT Total uncertainty of system
xxxvii
β1 Sensitivity of collapse-level spectral acceleration to variation of epsilon of that ground
motion
δ Logarithmic decrement
σ Normal stress or standard deviation
σ Normal stress matrix
γ Effective unit weight or specific weight or Ramberg-Osgood parameter
γi Mode shape of the ith. mode
Δ eff Effective abutment longitudinal displacement
ε Strain in structural analysis and Number of standard deviations by which an observed

logarithmic spectral acceleration differs from the mean logarithmic spectral acceleration in
hazard analysis.
ε Meridional eccentricity
εb Buckling strain in the longitudinal reinforcing steel bars
εy Yield strain of the longitudinal reinforcing steel bars,
0 Predicted mean epsilon
 ( T1 )records Mean epsilon value of the Far-Field ground motion evaluated at the period, T1
εc Strain at peak stress
εcr Crack strain
εcu Concrete ultimate strain
εep Plastic strain amplitude
εsu Strain at the maximum stress of the transverse reinforcing steel bars
εs,max Steel fracture strain
εu Ultimate strain
ε50 Strain at 50% stress level
ε cpl Compressive plastic strain
ε in
c
Inelastic plastic strain
ε pl
t
Tensile plastic strain
ε ck
t
Cracking strain
xxxviii
 Characteristic vector
Φ () Standard normal cumulative density function (CDF)
Φ-1 Inverse cumulative normal distribution function
φ Latitude coordinate, internal angle of friction, or angle of failure envelope
φp Plastic curvature
φy Yielding curvature
φ´ Angle of internal friction of sand
 Complex characteristic value
µ Strain Hardening Parameter or mean value
µT Period-based ductility
 Mean annual frequency or longitudinal coordinate
t Ratio of the total area of longitudinal reinforcing steel bars to the gross section area
cc Ratio of area of longitudinal reinforcing steel bars to area of core of section
s Volumetric ratio of spirals or circular hoops to the core concrete
v Volumetric ratio of transverse steel
ln (  ) Standard deviation of the normal distribution representing the ln IM values

θ Deviatoric polar angle, angle of the principal crack plane, or temperature
 ( ) Phase angle
θp Plastic hinge rotation
θy Yield rotation
τ Tau-averaging factor or critical shear stress
ν Poisson's ratio
 Fixity factor
 Ccoefficient for spirals and hoops
ω Cyclic frequency
 Frequency
ψ Dilation angle
ξ Damping ratio
xxxix
ξn Modal damping ratio
~ Mean “about” when placed before a number and indicate a range between two numbers
when placed between them
Subscript
Accept Acceptable values of (adjusted) collapse margin ratio

MT Maximum Considered Earthquake (MCE) ground motion intensity
CT Median collapse intensity
SSI Soil structure interaction
TOT Total
α% Value relates to α% probability of collapse
i Denotes index archetype model i
Superscript
ˆ Estimated value
_
Average Value
°
Degree
xl
List of Abbreviations
AASHTO: American Association of State Highway and Transportation Officials

ABCs: Absorbing Boundary Conditions
ACI : American Concrete Institute
ACMR: Adjusted Collapse Margin Ratio
ACMRacceptable Acceptable Adjusted Collapse Margin Ratio
ACMRmedian: Adjusted Collapse Margin Ratio correspond to median (50%)
ACMRi: Adjusted Collapse Margin Ratio correspond to i % probability of collapse
API: American Petroleum Institute
ASCE: American Society of Civil Engineers
ATC: Applied Technology Council
AVT : Ambient Vibration Test
Caltrans: California Department of Transportation
CDF: Cumulative Distribution Function
CFC: Collapse Fragility Curve
CHBDC: Canadian Highway Bridge Design Code
CMR: Collapse Margin Ratio
CMRacceptable: Acceptable Collapse Margin Ratio
CMRmedian: Collapse Margin Ratio correspond to median
CMRα%: Collapse Margin Ratio correspond to α% probability of collapse
CDP: Concrete Damage Plasticity
CFA Collapse Fragility Curve
CP: Collapse Prevention
CPT: Cone Penetration Test
CSA: Canadian Standards Association
CSD: Caltrans Seismic Design Criteria
CSF: Collapse Scale Factor
CU: Consolidated Undrained
DB: Displacement Base
DM: Damage Measures
xli
DV: Decision Variable
EC: Euro Code
EDPs: Engineering Demand Parameters
EFDD Enhanced Frequency Domain Decomposition
EHFD: Extended Hyperbolic Forced Displacement
FB: Force-Based
FE: Finite Element
FEA: Finite Element Analysis
FEMA: Federal Emergency Management Agency
FFT: Fast Fourier Transform
FHWA: Federal Highway Administration
FT: Fourier Transform
GM: Ground Motion
IAB: Integral Abutment Bridge
IBEs: Infinite Boundary Elements
IDA: Incremental Dynamic Analysis
IFT: Inverse Fourier Transform
IM: Intensity Measure
infrmDB: Inelastic Displacement-Based Element
infrmFB: Inelastic Force-Based Element
infrmFBPH: Inelastic Force-Based Plastic Hinge Element
IO: Immediate Occupancy
JSCE: Japan Society of Civil Engineers
LRFD: Load and Resistance Factor Design
LS: Life Safety
LSH: Log-Spiral Hyperbolic
M-C: Mohr-Coulomb
MCE: Maximum Considered Earthquake or Event
MDOF: Multiple Degrees of Freedom
MoTI: Ministry of Transportation and Infrastructure
xlii
MRO: Meloland Road Overcrossing
NC: Near Collapse
NEHRP: National Earthquake Hazards Reduction Program
NGA: Next Generation Attenuation
NRC: Natural Resources Canada
NSC: Non-Simulated Collapse Modes
NZS: New Zealand Standard
OP: Operational
PBD: Performance Based Design
PBEE: Performance Based Earthquake Engineering
PEER: The Pacific Earthquake Engineering Research Center
PGA: Peak Ground Acceleration expressed as a ratio to gravitational acceleration
PL: Performance Level
PSHA: Probabilistic Seismic Hazard Analysis
RC: Reinforced Concrete
RTH: Response Time History Analysis
SBT: Soil Behaviour/classification Type
SBSPEP: Simplified Bridge Seismic Performance Evaluation Procedure
SC: Simulated Collapse Modes
SDC: Seismic Design Categories
SDOF: Single Degree of Freedom
SF: Scale Factor (Ground Motion)
SFRS: Seismic Force Resisting System(s)
SRC: Performance Retrofit Category
SSF: Spectral Shape Factor
SSI: Soil Structure Interaction
TF: Transmissibility Function
UHS: Uniform Hazard Spectra
USGS: U.S. Geological Survey
xliii
Acknowledgements
I would like to express my deepest gratitude to my academic supervisor, Professor Carlos Ventura,
for his outstanding guidance in research and generous advice in teaching and many other aspects
offered by him. I feel truly honoured to have the opportunity of being his student. The completion
of this thesis would have not been possible without his insightful comments and valuable feedback.
My many special thanks to Professor Liam Finn, my co-supervisor, for his amazing
support/advising/guidance through all the years of my studies. His penetrating questions taught
me to question more deeply. I thank him for enlarging my vision of science and providing coherent
answers to my endless questions.
I offer my enduring gratitude to my supervisory committee, Dr. Farzad Naeim, Dr. Upul
Atukorala, Dr. Anoosh Shamsabadi, and Professor Ahmed Elgamal. Their feedback was essential
to the success of this research.
I offer my enduring gratitude to the faculty of the department of Civil Engineering,
especially, Professors Reza Vaziri and Ricardo Foschi, who have inspired me to continue my work
in this field.
I owe particular thanks to Dr. Alireza Forghani, who was a true mentor to me. He guided
me to move forward in my research while keeping my ideas and efforts focused on the main theme
of the thesis. His comments and feedback inspired me to excel my work every step of the way.
I am deeply thankful to my wife, Maryam, and my son, Mohammad, for their continual
support, inspiration, and being there for me at hard times and sharing the happy moments
throughout the years of studying and living in Vancouver. Words cannot express my appreciation
to them. I am eternally grateful to my parents who always listened to me and shared a piece of
advice on the different challenges I faced. Their love and comfort were present every moment and
they taught me the path of success under God’s will, being patient and working hard to succeed
while serving others. This degree and all my achievements would not be possible without their
encouragements and continuous support.
xliv
Dedication
Dedicated to my parents, Yaghoub and Zahra, to my wife, Maryam and to my son, Mohammad
for their unconditional love and support
xlv
Chapter 1: Introduction
1.1 Overview, and Motivations
Strong earthquake shaking has resulted in collapse of several pile supported bridges worldwide.
Lack of understanding and consideration of the effect of Soil-Structure Interaction (SSI) is among
the major reasons behind these devastating collapses (Mylonakis et al., 1997). One of the famous
examples where SSI had a major contribution to bridge failure is the collapse of the Cypress
Structure in Oakland California during the Loma Prieta earthquake in 1989. The loose sand that
the structure was built on, contributed to a more severe response of the structure which ultimately
resulted in structural collapse of many sections of this bridge (Ohuchi et al., 1992) (Barbosa et al.,
2014).
During the Northridge earthquake in 1994, several bridge piers were damaged due to soil-
pile-bridge seismic interaction (Mitchell et al.,1995 and Mylonakis et al., 1997).
Figure 1-1 shows a bridge column failure due to the Northridge Earthquake in 1994.
Figure 1-1 Failure of Mission Gothic bridge column due to the Northridge shallow crustal earthquake in
California (Aschheim, 1994)
Another example where SSI had a major contribution to bridge collapse is the Hanshin
Expressway Bridge during the Kobe Earthquake in Japan in 1995. An analytical study of the
Hanshin Expressway Bridge showed that soil characteristics had a significant effect on the
behavior of this structure. The elongation of the fundamental period of system (soil+structure) and
modification of earthquake frequency content due to soft soil, soil categorized as having a standard
1
penetration of equal or less than 10 (N≤10), resulted in intensifying the structural response
(Mylonakis et al., 2000). Figure 1-2 shows the catastrophic failure of the two bridge structures due
to the Loma Prieta and Kobe earthquakes.
(a) (b)
Figure 1-2 (a) Collapse of the Cypress Structure, the freeway approach to the Bay Bridge from Oakland due
to Loma Prieta Earthquake (Steinbrugge, 1989), (b) Collapse of a 630m long segment of the elevated Hanshin
Expressway during Kobe Earthquake Japan (Tsikada and Shimbun, 1995).
Since then many studies have been conducted that showed the potentially detrimental role
of soft soil in intensifying the structural response. Soft soil can result in elongation of the system’s
period which may result in severe effects such as large displacements (worsening the P-Delta
effect) and deformations in critical regions of the bridge. Furthermore, soft soil can result in
amplification of the ground motion causing a more severe response of a structure (Rayhani and El
Naggar, 2008). These examples show the importance of consideration of SSI effects in the analysis
and design of bridges. This would include consideration of SSI in calculation of deflections,
ductility demand, and forces induced in the structure. As Ashkani Zadeh and Ventura (2017)
showed, SSI effects can lead to redistribution of loading in structural members, thus changing the
failure mode and an increase in probability of collapse in RC Integral Abutment Bridges (IABs).
Seismic response of a ‘soil+bridge’ system is highly influenced not only by the
characteristics of the superstructure, but also by the characteristics of the foundation and the soil
(Mylonakis et al., 2000). In such systems, response of structure and soil are inter-dependent
(Tuladhar et al., 2008).
2
Bridge structures play important roles in the economy, politics, culture, as well as national
defense. The possibility of severe damage to bridges that are subjected to rare earthquakes leads
to the necessity of collapse potential evaluation of bridges, particularly those which have been or
will be designed as lifeline or major route bridges in the high seismic zones. When it comes to
performance assessment, there is a lack of guidelines and pre-defined workflows to properly
address the SSI effects and site hazards on performance of bridges. There is a crucial need for
developing a practical methodology for quantitatively determining global seismic performance for
use in seismic design and retrofit.
1.2 Design Codes and SSI Effect
Currently, design codes in North America require consideration of SSI effects in specific cases.
The Canadian Highway Bridge Design Code (CHBDC) version S6-14 states that “the interaction
of soil-structure foundation systems to earthquake loadings shall be evaluated when deemed
appropriate by the Regulatory Authority.” Regulatory Authorities require consideration of the
nonlinear behavior due to soil-structure interaction in performing seismic analyses for a probability
of exceedance of 2% in 50 years and 5% in 50 years for the following bridge types, which are
classified as seismic performance category 2 and 3 (CSA-S6-14, 2014):
− Regular or irregular lifeline bridges
− Irregular major route bridges
However, seismic design codes do not provide comprehensive provisions to address SSI
effects. Some codes provide simplified approaches such as application of factors to account for
SSI effects. For example, in CSA S6-14, in the analysis of concrete buried structures, the effect of
soil-structure interaction on the horizontal and vertical soil pressures are considered using λh and
λv factors, respectively. The issue is that such approach is only applicable to the case that was
tested and cannot offer a general solution to variety of soil types and layers.
In addition, the CHBDC code provisions in clause 4.6.4, recommend designers to employ
established analysis methods to incorporate SSI effects in their design. The code provides
guidelines to include SSI effects in simplified discrete models in an approximate manner such as
springs and spring-dashpot systems. These guidelines, however, may not be able to capture all
3
aspects of SSI such as slab averaging and wave scattering effects. In addition, the code does not
provide specific guidelines on collapse assessment (CSA-S6-19, 2019).
To minimize seismic risk and avoid catastrophic failure of bridges, changes in the structural
response due to soil-structure interaction need to be effectively addressed in the next generation of
bridge design codes. This is especially important for bridges in regions with soft soil and high
earthquake risks.
1.3 Implementation of SSI Effects in Analysis and Design
To acknowledge the difference between free-field motion from the embedded foundation motion
and superstructure response, SSI effects on structures such as bridges were categorized into two
phenomena, namely inertial and kinematic interactions.
a) The interaction corresponding to mass related effect is called inertial interaction (Clough
and Penzien, 1995). Presence of soil alters the dynamic characteristics of the system by
elongating its natural periods and introducing additional damping. Inertia developed in the
structure due to its vibrations creates base shear and moment, which in turn leads to
displacements of the foundation relative to the free-field (Stewart et. al., 1998 and 1999).
b) The interaction effect associated with the stiffness of the structure is termed kinematic
interaction (Clough and Penzien, 1995). Kinematic interaction is about the effect of the
structure’s foundation on the soil motion. The extent of this effect is dependent on the
stiffness and geometry of the foundation and soil. Large and stiff foundations can
significantly alter the effective motion of the foundation compared to free-field motion. The
kinematic SSI effect can be captured through base-slab averaging, embedment effect and
wave scattering (Stewart et. al., 1998 and 1999).
There are two possible approaches to consider the effect of Soil Structure Interaction (SSI);
namely empirical and analytical. In the empirical approach, results of several test measurements
are distilled into models that can be used in analysis and design to estimate the seismic response
of a structure. One of the best examples of the experimental measurements of bridge abutment
behavior is full-scale testing of monolithic abutments. This class of tests was carried out on
cohesive backfill by Romstad et al. (1995) and on granular backfill by Stewart et al. (2007) and
Lemnitzer et al. (2009) at the University of California, Davis (UCD) and the University of
4
California, Los Angeles (UCLA), respectively. Although empirical models offer a valuable insight
into the effect of soil on structural response, their application is limited to the cases studied in tests.
Constraints in scale, material, and geometric configuration limits the applicability of test
observations to real situations experienced by the structure. In the long-term, accurate evaluation
of the effects of soil-structure interaction (SSI) on the seismic response of structures using the
empirical method will be possible when a significant body of strong motion data becomes available
in future from sites with instrumented structures and free-field accelerographs. Prior to availability
of data from sites with instrumented structures, study of the Inertial and Kinematic interaction
phenomena in a general sense can be only done by analytical methods (Stewart et al., 1998).
In analytical SSI methods, soil-structure models are constructed, and the seismic response
of this combined system is predicted using computational methods such as nonlinear dynamic
analysis. To appropriately estimate geotechnical and structural demands, detailed numerical
models that consider geotechnical and structural components are required in analyses. Accurate
representation of behavior of soil as well as structural components going through a nonlinear
regime is a key factor in analytical SSI models.
In practice most bridge engineers use simplified discrete models to analyze SSI effects.
These models include a limited set of SSI aspects. As discussed earlier, there is a lack of clear and
unified guidelines for engineers to follow in developing the SSI models.
Within the analytical framework, there are two approaches to simulate soil-structure
interaction effect; namely, discrete and continuum approaches. In the discrete approach, the effect
of soil is represented by sets of springs and dashpots attached to piles, abutment backwalls and
abutment buried foundations. Properties of these discrete elements are determined based on soil
characteristics. API (2007), AASHTO (2017), and CALTRANS (2019a) offer methods to define
properties of springs and dashpots for piles and abutments. Although the discrete approach offers
a computationally effective solution, as discussed by Finn (2005), this approach is inherently
incapable to accurately capture the inertial and kinematic SSI effects. Simple discrete models
which consist of springs and dashpots to represent soil (also known as pseudo-static methods) are
inherently incapable of accurately capturing inertial and kinematic effects. More complex models
that include lumped masses, e.g. models proposed by (Wolf, 1985), offer an improved
representation of inertial effects but the kinematic effects are still missing.
5
Discrete models are also incapable of simulating the dynamic effects of pore-water pressure
in soil. It should be noted that the discrete method is only capable of capturing some limited aspects
of SSI.
In the continuum approach, the soil is simulated by continuum elements allowing for a
more physical representation of its behavior. Soil layers are explicitly simulated in this approach
using continuum elements. In this approach effect of pore-pressure (which is the cause of
liquefaction) can be also simulated. Although the continuum approach offers a more realistic
representation of the soil, its complexity and computational cost compared to discrete approach
makes it a less favorable choice in engineering practice. The ability to capture different soil-
structure interaction effects of the discrete and continuum approaches for large-scale models such
as 3D bridge models are compared in Table 1-1.
Table 1-1 Capabilities of the different SSI modeling methods for a large 3D bridge model
Method
Effect to Evaluate
Discrete Continuum
Inertial Effect Yes Yes
Kinematic Effect No Yes
Pore pressure No Yes
Liquefaction No Yes
Computational Efficiency Yes No
Model Complexity Low to Medium High
A detailed 3D continuum modeling method where soil and structure are simulated together,
would capture both kinematic and inertial soil-structure interaction phenomena. Thus, the 3D
continuum modeling method with appropriate modeling of soil-structure interfaces provides us
with a better insight into the SSI effects when compared with the discreet method. The following
diagram (Figure 1-3) illustrates different models, analysis methods, and simulation approaches
which have been used in investigation of SSI effects.
6
Investigating SSI effects
Empirical Analytical
Model Model
Direct Approach Substructure

Analysis Approach Analysis
Kinematic Interaction Foundation

3D Continuum 3D Discrete Input Motion (FIM)
Simulation Simulation
+
Impedance Function (stiffness and

damping characteristics of foundation-
soil interaction)
+
Analysis of structure on
compliant base subjected to FIM
Figure 1-3 Diagram of investigating of effect soil-structure interaction
1.4 Goals, Objectives, and Tasks
The main goal of this research is to provide a better understanding of soil-structure interaction
effects on the seismic response of RC bridges and their collapse potential using a performance-
based framework. The following are the major objectives of study to achieve the goal.
a) Investigating variation of bridge foundation motion from the free-field motion
b) Identifying and comparing structural failure modes due to SSI
c) Proposing a procedure to assess seismic performance of the RC bridges
d) Comparison of ability of discrete and continuum approaches to capture different soil-
structure interaction effects for the 3D bridge models
The following major tasks are carried out in this study to meet these objectives:
i. Constructing 3D finite element models of soil and soil+ bridge systems using ABAQUS
and SeismoStruct software
7
ii. Comparing the predicted bridge foundation motion using FE model with the estimated
foundation motion from the free-field motion using the Tau-averaging method (Clough and
Penzien, 1995) and the analytical transfer function (Elsabee and Morray, 1977).
iii. Introducing seismic performance criteria for the main structural components in the
developed bridge SSI-Discrete archetype models and performing Incremental Dynamic
Analyses (IDA) on these models
iv. Predicting probability of collapse
v. Calculating and comparing Collapse Margin Ratio (CMR) and acceptable Adjusted
Collapse Margin Ratio (ACMRacceptable) of the SSI-Discrete archetype models
vi. Comparing seismic response of the SSI-Discrete model with its corresponding 3D
continuum bridge model
1.5 Thesis organization
This thesis is organized in three phases and seven chapters. Phase 1 of this study focuses on
variation of the bridge foundation motions from the free field motions which is known as kinematic
effect of soil-structure interaction. The focus of Phase 2 is on seismic evaluation and collapse
assessment of the RC bridges and proposal of a practical procedure for performing this assessment.
Phase 3 includes a comparison of predicted seismic response of main structural components of
MRO performing nonlinear analysis using collapse level ground motions on validated discrete and
continuum model of the bridge. These three phase studies are presented in seven chapters as
follows:
− Chapter 1: Provides an overview of past soil-structure interaction role in collapse of
bridges. Discusses the different analysis and modeling approaches in studying SSI.
− Chapter 2: Provides a literature review on SSI effects in response of RC bridge structures.
This chapter reviews current implementation of SSI effects in standard design codes. In
addition, it emphasizes the necessity of providing detailed guidelines currently missing in
the seismic design codes that help designers to incorporate SSI effects and reveal the
motivation for pursuing this research.
− Chapter 3: Studies kinematic aspect of SSI in RC bridges founded on pile foundations
investigating variation of bridge foundation motion from the free-field motion.
8
− Chapter 4: Provides collapse assessment of the Meloland Road Overcrossing (MRO) using
four different index archetype models considering soil-structure interaction and code-base
performance criteria. It compares the derived fragility curves of the models and considers
the significance role of SSI in shifting these curves. In addition, it reveals the role of SSI
effects in changing failure modes at the onset of collapse and collapse level.
− Chapter 5: Proposes a FEMA-base collapse assessment guideline for RC bridge structures.
It evaluates the performance of Meloland Road Overcrossing (MRO) bridge using the
proposed procedure.
− Chapter 6: Provides a comparison of seismic responses calculated using both discrete and
continuum methods. It shows significant differences in potential design parameters such as
drift ratio and base slab shear for the different methods. It also investigates compressive
and tensile damages.
− Chapter 7: This chapter summarizes all chapter conclusions of this research and provides
some recommendations for future studies.
Appendices are organized as follow:
− Appendix A provides the Meloland Road Overcrossing (MRO) Drawing Details used in
constructing bridge discrete models in Chapter 4 and Chapter 5 and bridge continuum
model used in Chapter 3, and Chapter 6.
− Appendix B presents the time histories of ground motions used in IDA analyses in Chapter
4 and ground motions used in analyses of the continuum models in Chapter 3, and Chapter
6.
− Appendix C presents spectra that shows variation of bridge foundation motion from the
free-field motion discussed in Chapter 3.
− Appendix D presents equations used in calculation of performance criteria for the MRO
discrete models presented in Chapter 4 based on Seismic Retrofitting Manual of Highway
Structures-Part 1.
− Appendix E provides AASHTO (2017) site class definitions.
− Appendix F provides a table summarized California mean predicted ε0 values for periods
of 0.2s and 1.0 s for California cities for site class B (Haselton et al., 2011).
9
(
− Appendix G provides relationships between ln Sc ( T1 ) ) and  ( T1 ) used in Chapter 5.
− Appendix H presents graph of failure modes of the archetype models obtained due to the
collapse level ground motions.
− Appendix I shows result of concrete tensile and compressive damage for the nonlinear
continuum model C.
− Appendix J shows results of yielded soil layers due to the collapse level ground motions
discussed in Chapter 6.
− Appendix K shows the running time required for performing different analyses employing
the bridge continuum models (C, E, and F models).
− Appendix L shows validation of Abaqus one-way infinite element, CN3D8, for the Quite
Boundaries
The roadmap of all studies that are carried out in this research is presented in the flowchart shown
in Figure 1-4.
10
Introduction and Literature Review:
Chapter 1: Introduction
Chapter 2: Literature review
Study Phases:
Phase 1: Chapter 3
Chapter 3: Investigating variation of pile supported bridge foundation from the free-
field motion (SSI Kinematic effect)
Phase 2 : Chapter 4 and 5 Phase 3: Chapter 6
Chapter 4: Determining probability Chapter 6: Comparing response of

of collapse of MRO using four discrete and continuum models of
archetype models and investigating MRO at the collapse level.
their collapse failure modes investigation of soil damage and
failure in the main structural
members in the nonlinear continuum
model
Chapter 5: Comparing ACMRs as a
collapse safety measure, determining
collapse capacity of the models, and
proposing a simplified performance
evaluation process for RC bridges
Chapter 7: Concluding remarks for the

Appendices
all three phase studies
Figure 1-4 Flow chart of the study phases and their corresponding chapters
11
Chapter 2: Literature Survey
2.1 Scientific Understanding of SSI: Early History to Current State
SSI effect was first noticed by Lord Kelvin in the 19th century when analyzing the problem of
determining displacements caused by concentrated static forces acting at some arbitrary point in
an elastic infinite solid (Kausel, 2010). Up to the 20th century, the soil effect was only considered
in a static manner. In early 20th century Sir Horace Lamb studied the dynamics of wave propagation
in soil (Kausel, 2010). However, the influence of soil-structure interaction effects on structural
response was first considered by Sezawa and Kanai in 1935 where they investigated decaying
vibration of a structure due to dissipation of its energy into the ground (Sezawa and Kanai, 1935).
Martel studied SSI for the first time in the US in 1940 (Martel, 1940). He reported observations of
the performance of the Hollywood Storage Building during the Long Beach earthquake in 1933.
Bycroft (1956) studied response of a circular rigid plate on an elastic bed subjected to harmonic
excitation.
By performing an analytical study of elastoplastic systems on rigid and flexible bases,
Newmark and Hall (1973) concluded that SSI has a beneficial effect on the response of structure
by decreasing ductility demand due to its contribution to elongating the period of the system.
Kausel in 1974 used a numerical modeling approach to investigate the effects of Soil-Structure
Interaction. He developed a finite element-based- method using the Fourier expansion technique
to determine response of axisymmetric foundations and soil-structure systems. This study showed
a good agreement between the predicted response of a three-dimensional model and an equivalent
plane strain model of a nuclear power plant constructed on a stratum. He was one of the first to
point out the importance of considering the effect of foundation rigidity on the earthquake waves
(later titled as kinematic SSI effect) in addition to the effect of inertial forces from the structure on
soil deformation (later named as inertial effect). Kausel also called for more investigation on the
effect of foundation flexibility on structural response as he found a significant mat distortion in
the 3-D model (Kausel, 1974). Veletsos (1977) studied the effect of soil-structure interaction on
the effective damping and seismic response of structures. Later, his seminal work was incorporated
in the ATC-3 (1978) provisions.
12
The terms inertial and kinematic interaction were introduced by Whitman (1970) to
recognize two types of interaction (Roesset, 2013). The importance of kinematic interaction effects
particularly for embedded foundations was confirmed by subsequent studies (Elsabee et al., 1977).
In the 70’s, several studies were conducted to evaluate the relative importance of inertial and
kinematic interaction, developing SSI method of analysis, and determining the dynamic stiffness
of different types of foundations.
There are two general categories in analysis of Soil-Structure Interaction, namely the direct
method and sub-structuring (also known as multi-step) method (Kramer, 1996). The effect of soil-
structure interaction on structural response has been investigated by many scholars using these
general methods of analysis. The Substructure approach divides the system into three sub-parts to
calculate the response of the SSI system: determination of the motions of the foundation without
any structure due to the seismic waves; computation of a dynamic stiffness matrix for the
foundation; and dynamic analysis of the structure with the foundation matrix added and subjected
to the motion computed in the first step (Roesset, 2013). The substructuring approach assumes a
simplified linear or bi-linear elastic behavior for both soil and the structure. The direct analysis
approach is known as the ‘complete solution’ by Seed et al. (1977). The direct approach is the best
approach to conduct nonlinear analysis if a detailed 3D model of soil and structure with appropriate
nonlinear constitutive relations for the soil is employed (Seed et al., 1977). Both discrete and
continuum methods are classified under direct approach. Discrete modeling simulation method is
one of the most popular methods to calculate response of the soil+bridge system. In this approach,
soil and its interaction with the pile foundation and the abutment systems are represented by a set
of springs and dashpots. In continuum modeling the response of soil and structure system are
concurrently simulated.
Earlier studies on SSI effect concluded that SSI has a beneficial effect on seismic response
of structures due to its contribution to increases in the period of structure and the damping of the
system. The pioneering work of Newmark and Hall (1973) and their conclusion on beneficial effect
of SSI was later shown not to be generally applicable to all cases. Priestley and Park (1987)
revealed the disadvantageous aspects of SSI showing that the additional flexibility of an
elastoplastic bridge pier can reduce the ductility capacity of the system (Gazetas and Mylonakis,
2001). In addition, Miranda and Bertero claimed that the proposed Newmark and Hall relationships
13
do not apply to ground motions recorded on soft soils (Miranda and Bertero, 1994). Reviewing
the past investigations of strength reduction factors, Miranda and Bertero showed that magnitude
of the strength reductions is mainly a function of period of the system, maximum tolerable
displacement ductility demand, and the site soil conditions. Strength reduction factor known as R
factor is conventionally defined as the ratio of elastic strength (maximum strength if the structure
remained elastic) to yield strength. Miranda and Bertero concluded that using R factor derived for
sites with very dense soils for the short period structures (structures with the fundamental period
less than two thirds of predominant period of ground motion) founded on soft soils results in a
significant increase in displacement ductility demands compared to the ductility demands in the
original design (Miranda and Bertero, 1994).
Considering SSI in inelastic response of bridge piers, Ciampoli and Pinto (1995) pointed
out the importance of the stiffness ratio of structure relative to soil and height-to-width ratio of
structure in determining seismic response of inelastic structures. Based on this, they identified two
parameters, namely wave parameter and slenderness parameter. The wave parameter represents
the relative soil-to-structure stiffness whereas slenderness is a geometrical parameter (Anand and
Kumar, 2018).
Ciampoli and Pinto (1995) noticed that although considering SSI in inelastic response of
bridge piers results in a favorably reduced ductility demand, but also results in an increase in top
displacement demand.
Mylonakis and Gazetas (2000) reviewed SSI effects on seismic response of bridge
structures using recorded motions and theoretical considerations. In contrast with the previously
agreed-upon perception of favorable effect of SSI on decreasing ductility demand, they showed
that the increase in the fundamental natural period of a moderately flexible structure due to SSI
may have a detrimental effect on the imposed seismic demand for specific seismic and soil
environments. Mylonakis and Gazetas also showed that the detrimental effects of SSI are
substantial for heavy and rigid structures constructed on soft soils. They argued that limiting the
analysis ductility concepts and geometric aspects (as introduced by Ciampoli and Pinto (1995))
can lead to wrong assessment of structures’ seismic performance (Mylonakis and Gazetas, 2000).
The same authors in their later work investigated the collapse of Fukae section of Hanshin
Expressway during 1995 Kobe Earthquake. They identified the significant role of SSI in the
14
collapse, in addition to issues in the structural design observed in earlier investigations (Gazetas
and Mylonakis, 2001).
Finn (2005) investigated SSI effects on bridge response using a three-span continuous box
girder bridge model. He concluded that fixed-base models are only applicable when the ratio of
the superstructure stiffness to the foundation stiffness is small. In this case, the effect of inertial
interaction on system frequency is negligible compared to the effect of kinematic interaction (Finn,
2005).
Ghalibafian and co-workers (2008) investigated the effect of SSI on ductility demand, and
in particular its relation to the ratio of the flexible-based to fixed-based period. They observed that
role of SSI in response of piers with the pile foundations becomes more significant when the ratio
of piers elongated period to the fixed-base natural period of pier increases (Ghalibafian et al.,
2008).
Aviram et al. (2008) and Carvajal Uribe (2011) claimed that abutments can significantly
affect the response of an entire bridge system undergoing strong earthquake motion particularly
for bridges with short spans and relatively high superstructure stiffness. They found that using a
simplified abutment model can be acceptable in practice in the longitudinal direction comparing
to more sophisticated spring models. However, it can substantially underestimate the
displacements in the transverse direction of the bridge (Aviram et al., 2008).
Jarernprasert et al. (2013) and Aydemir (2013) studied the effect of stiffness degradation
on strength reduction factors and inelastic displacement ratios considering soil-structure
interaction. They suggested that SSI effects can be incorporated in the seismic codes in terms of
modified strength reduction factors and inelastic displacement ratios.
Investigating damaged asymmetrical buildings which were supported on piled raft during
the 2015 Nepal Earthquake, Badry and Satyam (2017) discovered that detrimental effects of SSI
can be significantly intensified by asymmetry in the geometry of the superstructure.
Stefanidou et al. (2017) studied effects of the Soil-Structure Interaction in the seismic
fragility analysis of RC bridges, both at the component and the global level. They found that
although SSI effects on the probability of system failure is minimal, it may have a significant effect
on component’s fragility (Stefanidou et al., 2017).
15
2.2 Kinematic SSI Effect
The Kinematic aspect of SSI has been the subject of research since the mid 70’s initiated by
Kausel’s work in 1974 as discussed earlier (Kausel, 1974). Elsabee and Morray (1977), Veletsos
and Prasad (1989) and Veletsos et al. (1997) showed the importance of the kinematic SSI effect in
structures supported on large and stiff foundations. They developed several transfer functions to
express translational and rotational foundation motions in terms of free field ground motion. These
transfer functions were calibrated later by Kim and Stewart (2003) against observed foundation
and free field motions.
Following the proposed averaging method by Newmark et al. (1977) and Morgan et al.
(1983) that estimates the foundation motion from free-field motion, Clough and Penzien (1995)
proposed Tau-averaging method. Clough and Penzien (1995) employed the idea of averaging
harmonic waves when encountered by a rigid massless foundation to estimate the spectral
acceleration felt by the foundation as a function free-field spectrum (Clough and Penzien, 1995).
2.3 Simulating SSI Effects
Numerical simulation has been an effective tool to investigate SSI effects. Models span from
simple discrete spring-dashpot models to very complex 3D continuum models.
Discrete models have been popular solutions due to their simple implementation and low
computational cost. In these models the effect of soil is represented using a series of springs and
dashpots. Discrete models are not able to directly capture kinematic effects, but if properly
calibrated, they can be an effective tool to capture the effect of soil on dynamic characterises of
the structural response, specially in elastic regime.
Kausel et al. (1978) presented a series of practical tables and charts using combined
analytical numerical methods to estimate translational, rotational, and cross-coupling spring and
dashpot coefficients for mat foundations and single piles embedded in elastic soil deposits. Gazetas
(1991) developed analytical equations and dimensionless graphs to calculate dynamic stiffness of
a foundation system as a function of its shape, material properties of the underlying soil, the
amount of embedment, and the frequency of the input dynamic load. Later, Makris and co-workers
(1994) proposed a procedure using these analytical equations to calculate dynamic stiffness of
bridges supported by pile-foundations. These works were mainly limited to linear small
16
deformations regime and were not applicable when deformation was significant during high-
intensity excitations. To address this shortcoming, Zafir (Zafir, 2002) and Lam and co-workers
(Lam et al., 2007) used secant stiffness values at the expected displacement level during the
earthquake to characterize the nonlinear inelastic response of the foundation soil and its interaction
with the superstructure. Although the secant approach represents the reduction of stiffness due to
nonlinear soil behaviour, it is unable to capture the hysteretic behaviour and energy dissipation
processes in soil.
Elgamal and co-workers (2002) and Yang and co-workers (2003) developed fully coupled
solid-fluid 3D elements and corresponding soil material models to simulate water pore pressure
generation/redistribution/dissipation and drained/undrained cyclic behavior for both sand and clay
materials. They developed multi-yield surface plasticity models implemented as Pressure
Dependent Multi-Yield (PDMY) and Pressure Independent Multi-Yield (PIMY) model in
OpenSees software framework (Elgamal et al., 2002) and (Yang et al. 2003). PDMY and PIMY
model are capable of simulating liquefaction effects and soil dilatancy of the pressure-dependent
(sand and silts) and independent (clay and silts) materials, respectively (see (Elgamal et al., 2003),
(Elgamal et al., 2008) and (Elgamal, 2010)).
Jeremic and co-workers (Jeremic et al., 2009), by analyzing a 3D continuum bridge model
found that there is a significant difference between the free field motions and observed motions at
the bridge columns. They concluded that the use of free field motions as input for structural models
would not be realistic.
Linear spring models are substituted with a more sophisticated approach known as the
Beam on a Nonlinear Winkler Foundation (BNWF). In this approach, the foundation is simulated
as a beam supported on a series of discrete nonlinear springs to simulate the supporting soils. The
BNWF approach is largely used in practice to simulate nonlinear response of soil-structure
systems.
The American Petroleum Institute (API, 2007) has provided nonlinear backbone curves for
the analyses of soil-pile interaction using discrete springs. The API model includes both lateral
and vertical (frictional and end bearing) spring properties.
Allotey and El Negar (Allotey and El Negar, 2008), developed nonlinear p-y links to be
used in discrete models to represent effect of soil. These p-y links were based on the Winkler
17
Model and could capture the hysteretic behaviour due to cyclic loading. The p-y models are
implemented in commercial software such as SeismoStruct and have become an integrated tool in
SSI simulation in practice. This model was employed by (Ashkani et al., 2017) to model SSI effects
in integral abutment bridges. The shortcoming of the p-y link model is its complexity and
introduction of nineteen model parameters, which require extensive model characterization and
calibration.
Correia and co-workers proposed an advanced pile-head macro-element model for
simulation of inertial soil-structure effects in the seismic analysis of deep bridge foundations. Their
proposed model was validated through comparison of numerical and experimental outcomes. The
model demonstrated to be an efficient tool for performance-based seismic design (Correia et al.,
2012).
The California Department of Transportation (Caltrans, 2019a) has recommended
nonlinear backbone curves for the embankment-abutment interaction. Caltrans provisions are
based on the work of Shamsabadi and co-workers (Shamsabadi et al., 2010), who provided
numerical models to count for Abutment-backfill soil interaction and estimate its influence on
seismic response of bridges. These models were validated against experimental measurements
performed by UCLA and UCD (Shamsabadi et al., 2010).
Continuum 3D models have been rarely used to simulate the SSI effects in Soil+Bridge
systems. This is due the significant complexity of such models and their high computational costs.
However, the introduction of powerful computational resources has made such analyses feasible
these days.
Mc-Callen and Romstad (1994) are among the pioneers in 3D continuum modeling of SSI
systems. They modeled the Painter Street Overpass in Northern California using a 3D finite
element model representing the nonlinear hysteric behavior of the soil-structure (Mc-Callen and
Romstad,1994). Jeremic et al. (2009) assessed the influence of SSI on response of a prototype
soil–bridge system considering non–uniform soil conditions on a four-span prototype concrete
bridge. Due to computational limitations, they did not simulate the abutments in their 3D
continuum model of the bridge as they considered bridge deck is apart from the abutments.
Shamsabadi et al. (2007), Kwon and Elnashai (2008), Elgamal et al. (2008), Lu et al.
(2011), and Rahmani et al. (2016) are among the more recent 3D continuum modeling efforts.
18
Kwon and Elnashai (2008) modeled Meloland Road Overcrossing (MRO) using a multi-
platform analysis approach to combine the soil and structural models. In this model, the bridge is
simulated using the Zeus-NL FE software (Elnashai et al., 2002) which employs fiber-based frame
analysis. Embankments, abutments, and pile groups are modeled in OpenSees (McKenna and
Fenves 2001). The two models were combined using UI-SimCor to capture the SSI effects (Kwon
and Elnashai, 2008).
Elgamal and co-workers (2008) developed one of the first 3D continuum models of bridge
structure and supporting soil of the Humboldt Bay Middle Channel Bridge in California using the
OpenSees software. The bridge model was used by Lu et al. (2011) and Rahmani (2014) to predict
the seismic response of the bridge-soil system using high-performance computing techniques to
reduce computational time of required nonlinear dynamic analyses.
Stefanidou and co-workers (2017) studied inertial and kinematic SSI effects for the seismic
fragility analysis of a typical three-span overpass using the direct analysing method and discrete
modeling approach. They proposed a methodology to quantify the fragility of a typical highway
overpass at both component and system level (Stefanidou et al., 2017).
2.4 Summary of Key SSI Research Milestones
Figure 2-1 summarizes some key studies conducted on the influence of soil-structure interaction
effects on structural response in a chronological order.
19
2020
Dynamic Soil-Structure Interaction Analyses of Two Important Structures Affected by Liquefaction
during the Canterbury Earthquake Sequence (Luque and Bray, 2020)
2015
SSI Bridge: Soil-Bridge Interaction During Long-
Predict Seismic Response the Bridge- Ground System Using duration Earthquake Motions (Barbosa et al. , 2014)
2010 High-performance Computing Techniques (Lu et al. , 2011)
Three-Dimensional Seismic Response of Humboldt Bay
Bridge-Foundation-Ground System (Elgamal et al. , 2008) A Study of Piles during Earthquakes: Issues of Design and
2005 Analysis (Finn, 2005)
Modeling of Cyclic Mobility in Saturated Cohesionless Kinematic Soil-Structure Interaction from Strong Motion
Soils (Elgamal, et al., 2003) Recordings (Kim and Stewart, 2003)
2000
Seismic Soil-Structure Interaction: Beneficial or Transfer Functions for Rigid Rectangular Foundations
Detrimental (Mylonakis and Gazetas, 2000) (Veletsos et al., 1997)
Dynamics of Structures (Clough and Panzin, 1995) Effects of Soil-Structure Interaction on Inelastic Seismic
1995
Evaluation of Strength Reduction Factors of Response of Bridge Piers (Ciampoli and Pinto, 1995)
Earthquake-Resistant (Miranda and Bertero, 1994)
1990
Seismic Interaction of Structures and Soils:Stochastic
Approach (Veletsos and Prasad, 1989) Strength and Ductility of Concrete Bridges Columns under
Seismic Loading (Priestley and Park, 1987)
1985
Seismic Soil-Structure Interaction (Wolf, 1985)
1980
Sub-Structure and Direct Approach (Seed et al.,1977) Dynamic behavior of Embedded Foundation
1975 (Elsabee and Morray, 1977)
Seismic Design Criteria for Nuclear Reactor Facilities Inertial and Kinematic Interaction Effect (Kausel, 1974)
(Newmark and Hall,1973)
1970
Forced Vibrations of a Rigid Circular Plate on Semi-Infinite
1950 Elastic Space and on An Elastic Stratum (Bycroft, 1956)
Observations of Performance of The Hollywood Storage

Building during The Long Beach Earthquake (Martel, 1940) Decaying Vibration of A Structure Due to Dissipation of Its
Energy into The Ground (Sezawa and Kanai, 1935)
1930
Figure 2-1 Timeline of some SSI key studies have been carried out on response of structures
The University of British Columbia (UBC) has been at the forefront of research on SSI in
the past couple of decades. Table 2-1 summarizes some of the key research projects which have
been conducted on soil-structure interaction at UBC.
20
Table 2-1 A summary of research on SSI at UBC since 2000
Author Title Date
Sepideh Ashtari Evaluating the performance-based seismic design of RC bridges according to the 2018
2014 Canadian highway bridge design code
Elnaz Amirzehni Seismic Assessment of Basement Walls in British Columbia 2016
Mohammad Numerical study on the response of pile groups under lateral loading 2015
Sajjad Fayyazi
Ramin Latifi Effect of pulse-like near-fault ground motions on inelastic response of structures 2014
including foundation flexibility
Amin Rahmani Three-dimensional nonlinear analysis of dynamic soil-pile-structure interaction 2014
for bridge systems under earthquake shakings
Seku Samory Displacement demands for performance-based design of skewed bridges with seat 2014
Catacoli type abutments
Kianosh Seismic analysis of the RC integral abutment bridges using performance-based 2013
Ashkani Zadeh design approach including soil structure interaction
Bishnu Hari Investigation of variation of motions between free field and foundation in seismic 2013
Pandey soil-structure interaction of structures with rigid shallow foundation
Alireza Response of underground power transmission cables subjected to earthquake 2011
Ahmadnia induced permanent ground deformations
Juan Carlos Seismic embankment-abutment-structure interaction of integral abutment bridges 2011
Carvajal
Human Evaluation of effect of nonlinear soil structure interaction on the inelastic Seismic 2006
Ghalibafian response of pile supported bridge piers
Thuraisamy Seismic analysis of pile foundation for bridges 2000
Thavaraj
2.5 SSI in Design Codes
The first generation of SSI provisions for design in the US was drafted by the Applied Technology
Council (ATC) in 1978 as ATC 3-06 (ATC, 1978). These provisions suggested a reduction in
design base shear due to elongation of natural period and higher damping. These provisions which
were based on elastic analysis, led to a reduction in design base shear.
ASCE/SEI 7-10 (2010) introduced a lower limit on base shear reduction by restricting
modified design base shear to 70% of the original value. FEMA P-1050 (2015) introduced an
equivalent lateral force procedure for base shear reduction. ASCE/SEI 7-16 (2016) allows using a
21
linear dynamic analysis which employs either an SSI modified site-specific response spectrum
(developed by the design engineer) or the SSI modified general design response spectrum (as
specified in the code) (Anand and Kumar, 2018).
AASHTO LRFD Bridge Design (2017) requires explicit consideration of soil-structure in
pile groups and deep foundations (AASHTO LRFD, 2017).
The backfill passive pressure resisting movement on the abutment varies nonlinearly with
abutment displacement and is dependent on the material properties of the backfill. To estimate the
backfill passive pressure resisting movement at the abutment, Caltrans SDC-2.0 (2019a) suggested
an empirical bilinear model based on experimental and calibrated analytical models performed by
Shamsabadi et al. (2007) for straight abutments and Rollins and Jessee (2013) and Shamsabadi
and co-workers (2020) for the skew abutments, respectively. In these studies, they used engineered
structural backfill to a relative compaction of at least 95% (Caltrans, 2019a).
Caltrans SDC-2.0 (2013) classifies a bridge as non-standard when its supporting and
surrounding soil is poor. In this case, it requires specific design criteria that address soil-structure
interaction (SSI) effects. The mechanisms that should be addressed within SSI framework are earth
pressure generated by lateral ground displacement, dynamic settlement, and the effect of
foundation flexibility on the response of the entire bridge. Depending on soil properties, Caltrans
SDC-2.0 (2019) categorized soils into two classes, namely, Class S1 and Class S2. Class S1
represents competent soils with superior mechanical properties which are not susceptible to
liquefaction, lateral spreading, or scour. Poor soil has conventionally been categorized as having
a standard penetration (N) of less than 10, and marginal soil is defined as the range of soil that
cannot be classified as either poor or competent soil. Class S2 represents non-competent soils such
as marginal soil, poor soil, soft soil, potentially liquefiable soil and soil susceptible to lateral
spreading.
The previous version of seismic design criteria such as Caltrans SDC-1.7 (2013) lacked a
precise specification and designers were allowed to use their engineering judgement and
experience when assessing the impact of marginal soils on the overall dynamic response of a bridge
(Caltrans, 2013). Caltrans SDC-2.0 (2019a) on the other hand provides a model for lateral analysis
of pile foundations surrounded by Class S2 soil (Caltrans, 2019a).
22
Currently, Canadian design codes require consideration of SSI effects in specific cases. As
an example, based on clause 4.11.14 of the Canadian Highway Bridge Design Code (CHBDC),
the interaction of soil-structure foundation systems to earthquake loadings needs to be evaluated
when required by the Regulatory Authorities (CSA-S6-14, 2014). Bridge Standards and
Procedures Manual (BSM) which is a supplement to the CSA requires performing soil-structure
interaction analysis for lifeline and emergency-route bridges with Soil Profiles III and IV in
Seismic Performance Zone 2 and for all bridges in Seismic Performance Zones 3 and 4 (BC MoTI,
2016).
Although the code recommends designers to employ established analysis methods to
incorporate the SSI effect in their design in clause 4.6.4 of CSA, the code does not provide a
specific guideline for designers on how to establish a method to incorporate SSI effect in their
analysis and design. In addition, the code allows that the soil-foundation-structure system is
represented using simplified models of soil and foundation, and it permits the response assessed
using the pseudo-static method (CSA S6-14, 2014). As discussed earlier, the simplified models do
not offer a comprehensive SSI framework that is cable of capturing the key SSI effects such as
kinematic effects and liquefaction.
Japanese standard specifications for concrete structures (JSCE 15, 2007) recommend
consideration of dynamic interaction between the structure and the ground in design of bridge
abutments, retaining walls, and foundation structure and underground structures. For other types
of structures, JSCE guideline recommends consideration of SSI effects depending on type and
characteristics of structure and ground. This code allows designers to implement SSI effect using
either direct or the substructure analyzing methods. (Anand and Kumar, 2018).
Eurocode 8 (EN 1998-5, 2004) suggests SSI to be considered for slender structures or
structures with significant p-delta effect. Considering SSI effects in design is recommended for
structures supported by piles such as bridge piers or massive buried foundations such as silos and
nuclear plants. The code also recommends considering SSI effects when dealing with deposits
soils with high plasticity index and low shear strength. Despite requiring consideration of SSI
effects for these specific cases, the code does not provide specific guidelines required for
quantification of SSI effects (Anand and Kumar, 2018).
23
Likewise, New Zealand standard (NZS 1170.5, 2004) does not provide specific guidelines
for incorporating SSI in design practice and relies on engineering judgement. However, it
recommends use of a measuring parameter called structural performance factor which is a function
of period of the seismic resisting system, damping, material and shape of the structure, and
interaction of the structure with the supporting soil. New Zealand seismic code also requires
consideration of foundation flexibility in calculation of structure deformations. However,
foundation flexibility is often not considered in design by structural designers due to a lack of
specific guidelines on how to account for such effect (NZS 1170.5, 2004) and (Anand and Kumar,
2018)).
To minimize the seismic risk and avoid catastrophic failure of bridges, change in the
structural response due to soil-structure interaction needs to be effectively addressed in the next
generation of bridge design codes. This is especially important for bridges built on soft soil and in
regions with high earthquake risk. Following points need to be considered to mitigate seismic
vulnerability in the design process of structures whose responses are sensitive to the SSI effects:
a) For structures founded on landfills and soft soils, establishing procedures in the
seismic guidelines to account for SSI effects is necessary to ensure implementation
and quantification of SSI.
b) Seismic standards mainly suggest conditions for performing an SSI analysis in
design practice but do not establish procedure for it, and there is no consensus in
codes about the specifics of how to consider SSI effect in structural analysis and
design. For example, as discussed, Canadian codes require consideration of SSI
effect in design of critical structures such as lifeline and emergency route bridges
located in critical zones (SPZ3 and SPZ4). However, code only provides general
guidelines as how to incorporate the SSI effects for soil types III and IV, but details
are left unspecified.
c) As incorporating SSI effects in structural design is not sufficiently addressed in
most of standards, incorporating SSI effects mainly depends on capability and
judgment of designers. ASCE 7-16 (2016) appears to be the most developed code
in term of inclusion of SSI in structural design comparing to other discussed
standards.
24
d) Ductility demand, modal characteristics, seismic vulnerability, fragility, inelastic
displacement ratio, and strength reduction factor are the key parameters in
considering SSI effects (Anand and Kumar, 2018).
e) Close collaboration between geotechnical and structural engineers is essential to
address the SSI-related risks and uncertainties. Ability to provide a platform for
knowledge sharing among various stakeholders of a project might help in reducing
its seismic vulnerability (Anand and Kumar, 2018).
2.6 Gaps in Knowledge
Many lifeline and emergency-route bridges have been constructed or will be constructed in seismic
hazard zones in North America where regulatory authorities require soil-structure interaction
analysis. Unfortunately, performing this type of analysis is a cumbersome process for practicing
engineers. Current bridge seismic design codes are lacking adequate guidelines on how to deal
with SSI effects and providing guidance on how to assess failure/collapse probability. Even SSI
experts have rarely used detailed continuum modeling to study the effect of soil-structure
interaction in structural response due to the complexity of the analysis and the computational
demands. Instead, a discrete approach has been widely used both in research and practice, despite
of the fact that its accuracy has not been appropriately assessed for the nonlinear dynamic analyses
of large-scale SSI problems, such as multi span RC bridges (Tyapin, 2007).
Assessment of nonlinear seismic response and probability of the collapse of RC bridges
located in seismic prone zones have not been sufficiently investigated within a performance-based
framework using the 3D continuum modeling approach. As discussed by Tyapin (Tyapin, 2007),
still a large gap exists between SSI simulation tools used by researchers and practitioners. One of
the major gaps in SSI simulation tools for bridge-soil systems is lack of an objective measure
which can be used to characterize the collapse safety and to compare collapse of bridges. This
thesis focuses on the following aspects which are lacking in the current approach to SSI effects
when it comes to performance-based analysis and design of bridges:
a) Lack of a comprehensive soil-structure model capable of predicting the collapse of a
structure:
a.1) Detailed continuum soil models do not typically have a comprehensive structural
model
25
a.2) Detailed structural models do not typically pay much attention to soil (discrete
elements are often used for soil simulation).
b) Lack of comprehensive IDA analysis using enough number of ground motions: to obtain a
reasonable collapse fragility curve, adequate number of simulations using a variety of
ground motions are required.
c) Lack of a comprehensive study on the SSI effects on the fragility curves.
In this research the following topics were investigated:
i. Variation of free-field and bridge foundation motions
ii. Variation of bridge response depending on SSI representation
iii. Variation of collapse mode depending on SSI representation
iv. Collapse assessment of RC bridges
v. Comparison between continuum and discrete models
26
Chapter 3: Investigation of Variation of Bridge Foundation Motion from the
Free-Field Motion
3.1 Introduction and Methodology
The presence of embedded foundations in soil affects the characteristics of the earthquake motion
at the site of the structure. As a result, excitation of the structure foundation is different from the
free-field motion. This phenomenon is known as Kinematic Soil-Structure Interaction effect.
Kinematic effect is about altering of ground motion due to its interaction with the foundation.
Multiple mechanisms contribute to Kinematic SSI effect. Kramer (1996) has listed the following
mechanisms as the main contributors to Kinematic SSI due to presence of a stiff foundation
(Kramer, 1996):
a) Base Slab Averaging
b) Wave Scattering
c) Embedment Effect
Base slab averaging results from the stiffness of foundation altering the free-field motion
even for a massless foundation. Wave Scattering (reflecting, refracting, etc.) happens due to lack
of homogeneity along the wave’s travel path, which can result in seismic wave incoherency
(Kramer, 1996). Embedment effect is the variation of ground motion with depth in the soil
medium. The amplification of motion is generally reduced with depth (Chang et al., 1989).
In addition to the aspects discussed above, vertically propagated shear waves and framing
action can be important contributors to the kinematic SSI effect. Vertically propagated shear waves
can cause rocking as well as translation of the foundation (Gicev et al., 2016). The framing action
occurs due to relatively stiff deck. It can be especially important for monolithic bridges with
diaphragm abutments where individual foundations are connected through a stiff upper structure
(Ashkani et al., 2018).
27
(a)
(b)
(c) (d)
Figure 3-1 Kinematic SSI main causes (a) base slab averaging, (b) wave scattering, (c) embedment effect, and
(d) vertically propagated shear waves. Image reproduced from (Kramer, 1996)
Figure 3-2 provides a summary of structural components that contribute to kinematic SSI
effect.
✓ Exist
Piles  ✓ ✓ ✓ ✓
 Not Exist
Foundations
✓ ✓ ✓ ✓ ✓
Deck Slab
    ✓
Figure 3-2 Various aspects of kinematic effect in bridges
28
To study the variation of bridge foundation motion from the free-field motion, the
following objectives are considered:
a) Investigate amplification or reduction of the foundation motions compared to free-field
motion. The concept of transmissibility function is employed to study amplification or
reduction of foundation motion from the free-field motion.
b) Investigate the applicability of the Tau-average method proposed by Clough and Penzien
(1995) for estimating foundation motion in integral abutment bridges.
c) Examine the applicability of the transfer function proposed by Elsabee and Morray (1977)
for estimating embedded foundation motion.
In this study, a detailed continuum finite element approach is employed to investigate the
kinematic SSI effects.
This chapter includes five sections. In the second section, a brief description of the model
and FE analyses are presented. In the third section, the differences between free field and
foundation motions are investigated using the finite element method. The transmissibility concept
is applied to determine amplification and de-amplification of the motion. The fourth section is on
the Elsabee and Morray transfer function and Tau-averaging method, where the capability of these
approaches to accurately estimate the foundations’ motion is examined. Finally, concluding
remarks are presented in the last section.
3.2 Meloland Road Overcrossing (MRO)
The Meloland Road Overcrossing (MRO) in California is chosen as the case study. The bridge is
located near the Imperial Fault Zone in southern California and has experienced a number of
earthquakes in the past decades, including the Imperial Valley earthquake in 1979, Westmorland
earthquake in 1981, Superstition Hills earthquake in 1987, and El Mayor earthquake in 2010 (see
Figure 3-4).
29
(a) (b)
Figure 3-3 (a) Meloland Road Overcrossing (MRO) in the background and a kink in Highway 8 due to
continuous crippling of the exposed Imperial fault, (b) Panoramic view of MRO (Ventura et al., 2011)
California Quaternary Faults:
Figure 3-4 California quaternary fault and location of the bridge and alignment of the Imperial Fault (USGS
Interactive Fault Map., 2019)
The MRO is a two-span Integral Abutment Bridge (IAB) with a single column pier.
According to earlier studies by Douglas and co-workers (1984), Maragakis and co-workers (1991),
Werner and co-workers (1993), and Kwon and Elnashai (2008), the pier and abutments are
supported on timber piles. The bridge abutments rest on a single row pile foundation. The pier
column rest on a footing and pilecap foundation supported by 5×5 group pile. Meloland Road
Overcrossing plan, elevation and section along with its seismic instrumentation are shown in
Figure 3-5. More detailed bridge drawings are presented in Appendix A.
30
Legend:
Positive direction of the
measured acceleration
Out of the plane positive
direction of the measured
acceleration
Figure 3-5 MRO plan, elevation and section along with sensor locations (Ventura et al., 2011)
The MRO has shown good performance during past earthquakes (Werner et al., 1993). This
research attributes such performance to the following factors:
a) The simple configuration of the MRO (e.g., unskewed, monolithic abutments, etc.)
b) The use of relatively tightly-pitched spiral steel as the transverse reinforcing steel bars for the
MRO's central pier, which was well beyond Caltrans' more typical practice at the time for the
transverse reinforcement of bridge columns.
The bridge is heavily instrumented and has been the subject of various studies. Zhang and
Makris (2002), Werner et al. (1987), Wilson and Tan (1990), Gates (1993) and Werner (1994)
determined natural frequencies and modal damping ratios using recorded motions. Kwon and
Elnashai (2008) used impact-type earthquake loading and analysis of decay in response to
determine the damping ratio of the super structure. Ventura and co-workers (2011) performed
ambient vibration tests to identify the dynamic characteristics of this bridge. The identification
was performed using the Enhanced Frequency Domain Decomposition (EFDD) technique,
implemented in ARTeMIS Modal (2010) software.
3.3 Continuum Model to Investigate Kinematic Effects
Continuum models of soil deposit and Soil+Bridge are constructed and used to simulate the bridge
response using the ABAQUS FEA software (SIMULIA, 2018).
A detailed 3D continuum model is developed for this study. Time history analyses were
performed using ten unscaled strong motions chosen from the PEER NGA West-2 Ground Motion
31
Database (PEER, 2018) were used as input motions. In this chapter, unscaled ground motions are
used as an input applied to the base of the 3D model.
A continuum representation of soil, foundation, and the bridge was constructed and
analyzed. To investigate the difference between bridge foundation motion and the free-field
motion, two models were developed for the MRO:
a) Soil Model only
b) Soil + Bridge Model (SSI model)
3.3.1 Elastic Model Description
The soil model is developed as an elastic continuum with soil material parameters summarized in
Table 3-1. The model includes a simplified representation of the soil surrounding bridge underpass
as shown in Figure 3-6(d). Infinite Boundary Elements (IBEs) are used as Absorbing Boundary
Conditions (ABCs) in the model’s boundary walls to simulate unbounded domain. IBEs ensure
that energy is dissipated at the model edges rather than reflecting into the system. It serves to
represent the far-field regions and provide “quiet” boundaries to the FE model in dynamic analyses
(SIMULIA, 2018). The soil+structure model includes a simplified representation of the 63.4m
long two-span concrete box girder bridge with a single pier column and the integral abutments.
The key structural components of the bridge, including the abutments, pier column, deck slab,
foundations and piles, and wing walls are considered in the model. The model also includes an
extended representation of the soil embankment behind the abutment backwalls due to its
significance in the system response (see Figure 3-6(a) and (c)). To investigate the Kinematic SSI
effect, soil and structural components are considered massless in the soil and soil+bridge models
to minimize inertial SSI effect. The FE model of the structure used in the soil+bridge model is
shown in Figure 3-5(b). To focus the study in this section on kinematic effects, elastic models for
soil and soil+bridge system are used in the dynamic analyses. This figure shows the location of
the points on the foundations (points 1a and 1b) and on the free field (points 2a and 2b) at which
the motions have been compared (see Figures 3-5(b) and 3-5(d)).
32
Embankments
7.5m
20m A
BC Point 1a Point 1b
D
E
150m long x 70m width
(a) (b)
Infinite Element Walls Point 2a Point 2b
(IBEs)
(c) (d)
Figure 3-6. a) Discretized layout plan and 3D view of the soil and Soil+Bridge structure (SSI) model showing
finite element mesh, b) massless structure modeled in the SSI model and measurement points, c) discretized
layout plan and infinite element boundary walls of the SSI model, and d) discretized layout plan and infinite
element boundary walls of the soil free-field model and measurement points
In the FE models of the structure+soil system, the soil representation is limited to a finite
region surrounding the structure. To incorporate the far-field effect of soil, non-reflecting
boundary conditions are employed. In Abaqus software, Infinite Boundary Elements (IBEs) can
be employed to simulate the absorbing boundary condition due to the far-field soil surroundings.
These elements will result in energy dissipation at the model edges rather than reflected back into
the system.
To solve boundary value problems defined in unbounded domains, built-in infinite
elements are provided in ABAQUS (SIMULIA, 2018). The infinite elements in ABAQUS are
developed based on the work of Lysmer and Kuhlemeyer (1969) on quiet boundaries to the FE
models performing dynamic analyses and Zienkiewicz et al. (1983) on stiffness of the infinite
element performing static solid continuum analyses (Edip et al., 2018).
In this research, standard finite elements and infinite elements are utilized to model the
region of interest and the far-field region, respectively. The infinite element CIN3D8 shown in
boundary walls in Figure 3-6 is used as Absorbing Boundary Conditions (ABCs) to simulate
33
unbounded domain and to provide far-field condition. CIN3D8 is a 3D solid continuum infinite
element with active displacement degrees of freedom in x, y, and z directions. This linear infinite
element consists of 8 nodes as shown in Figure 3-7 (SIMULIA, 2018). CIN3D8 is one-way infinite
element, i.e., defined in ABAQUS only to allow one-way propagation of wave (SIMULIA, 2018).
As a result, orientation of the element in the model to simulate quite boundaries is very important.
8
7
5
6
4 CIN3D8 Element
3
1
2
Figure 3-7 3D Continuum model simulated in ABAQUS as Meloland Road Overcrossing (MRO) bridge
reference model and surrounding infinite boundary element walls
Performance of the ABAQUS CIN3D8 infinite elements is validated in Appendix L by

comparing the predictions of a single element model supported by this element to another case
where the infinite element is replaced by a set of dashpots.
34
Table 3-1 Material properties of the soil layers (Kwon and Elnashai, 2008)
Thick.† Density† Bulk † Shear† Modulus of* Poisson’s* Shear Wave*

Soil Layer (ρ) Modulus Modulus (G) Elasticity (E) Ratio (ν) Velocity (Vs)
(m) (Kg/m3) (K) (MPa) (MPa) (MPa) (-) (m/s)
Embankments 7.43 1600 90 19 53.3 0.40 109.0

A 2.17 1500 300 60 169 0.41 200.0
B 3.36 1900 200 75 200 0.33 198.7
C 4.57 1800 750 150 422 0.41 288.7
D 4.57 1900 200 75 200 0.33 198.7
E 5.33 1800 750 150 422 0.41 288.7
†
These property values of the of the soil layers and embankments are adopted from Kwon and Elnashai
(2008).
* These value of the of the of the soil layers and embankments are calculated based on the above adopted
values using Equation (3.1).
Density, bulk and shear modulus of the soil layers are adopted from previous studies
conducted by Kwon and Elnashai (2008). The modulus of elasticity, shear wave velocity and
Poisson’s ratio of the soil layers are calculated using the Elastic moduli relationships for
homogeneous isotropic materials given Equation (3.1).
3K - 2G G max
ν= , Vs = and E = 2G max (1 + ν )
2 ( 3K + G )
(3.1)
ρ
where,
G and Gmax are the shear modulus and maximum shear modulus, respectively,
E is the modulus of elasticity,
K is the bulk modulus,
ρ is the density,
ν is the Poisson’s ratio, and
Vs is the shear wave velocity.
3.3.2 Eigenvalue Analysis on both Soil Free-Field and Bridge+Soil Models
Eigenvalue analyses are carried out using the same FE model in ABAQUS to obtain the natural
periods and mode shapes of the system. The eigenvalue analysis is also employed to compute the
Rayleigh damping coefficients of the soil model to be used in the time history analysis. Periods
35
and mode shape description of the first six modes of the Soil model only and SSI models are
presented in Table 3-2. The frequency values in parenthesis obtained from the ambient vibration
test carried out at the MRO structure by Ventura and co-workers (2011).
Table 3-2 Period and mode shape of the soil and Soil+Bridge models
Soil Model SSI model

Mode Frequency Mode Description Frequency Mode Description
[fi (Hz)] [fi (Hz)]
1 2.89 transverse mode 3.60 (3.37) vertical anti-symmetric mode
2 2.92 1st. torsional mode 3.95 (3.63) transverse mode
3 2.99 1st. longitudinal mode 4.34 (4.47) vertical symmetric mode
4 3.05 2nd. torsional mode 6.67 (6.74) 1st. torsional mode
5 3.13 2nd. longitudinal mode 10.00 (9.72) 2nd. torsional mode
6 3.23 3rd. torsional mode 11.11 (11.36) 2nd. vertical anti-symmetric mode
The mode shapes of the first three modes of the SSI model are shown in Figure 3-8.
Mode 1 (vertical anti-symmetric mode): T1=0.28s
Mode 2 (transverse mode): T2=0.25s
Mode 3 (vertical symmetric mode): T3=0.23s
Figure 3-8 Mode shapes of the first three modes of the continuum Soil+bridge combined model (SSI model)
36
3.3.3 Choice of Ground Motions
As mentioned earlier, a set of ten unscaled ground motions from the NGA-West2 ground motion
database of the Pacific Earthquake Engineering Research Center (PEER) (2018) are chosen to
perform response history analyses of the soil only and SSI models, respectively. Strong ground
motions with magnitudes between 6.5 and 7.5 were chosen as input motions for response time
history analysis in this study. The distance to the rupture plane (Rrup) of the selected ground
motions ranges from 1km to 30km. The average shear wave velocity of top 30 meters of the sites
(VS30) from 200m/s to 560m/s and with an average VS30 of 460m/s. In addition, ground motion
records are selected with different predominant periods (Tp), the period at which the maximum
spectral acceleration occurs in an acceleration response spectrum calculated at 5% damping.
Similarly, the records are chosen with different Peak Ground Acceleration (PGA). The
predominant period of ground motions are calculated in the horizontal direction. The maximum
and minimum considered Tp of the selected motions are 0.64s and 0.08s for the motion RSN#723
and RSN# 879, respectively. The maximum PGA of the chosen motions are 0.81g for the motion
RSN#4040 and RSN#1602 and the minimum PGA is 0.22g belonging to earthquake record RSN#
288 (see Table 3-3). In this study, the length of the records used for input was determined as
significant duration between 5% and 95% of the Arias Intensity (AI) of each motion. Spectra of
the selected input motions for comparison are shown in Figure 3-9. The mechanism and
characteristics of the selected input motions are shown in Table 3-3 and Table 3-4, respectively.
37
Table 3-3 The selected ground motions and their magnitudes and mechanisms
Earthquake Name PEER Year Station Name M Mechanism

RSN No
Imperial Valley-06 184 1979 El Centro Differential Array 6.53 Strike-slip
Superstition Hills-02 723 1987 Parachute Test Site 6.54 Strike-slip
Kobe-Japan 1119 1995 Takarazuka 6.9 Strike-slip
Hector Mine 1787 1999 Hector 7.13 Strike-slip
Duzce-Turkey 1602 1999 Bolu 7.14 Strike-slip
Kocaeli-Turkey 1158 1999 Duzce 7.51 Strike-slip
Landers 879 1992 Lucerne 7.28 Strike-slip
Bam-Iran 4040 2003 Bam 6.6 Strike-slip
El Mayor-Cucapah-Mexico 5836 2010 El Centro - Meloland Geot. Array 7.2 Strike-slip
Irpinia Italy-01 288 1980 Brienza 6.9 Normal
Table 3-4 The selected ground motions and their rupture distances and characteristics
Earthquake Name Rrup (km) Vs30 (m/sec) Tp (s) / PGA (g)
Imperial Valley-06 5.09 202.26 0.40 0.48

Superstition Hills-02 0.95 348.69 0.64 0.43
Kobe-Japan 0.27 312 0.46 0.70
Hector Mine 11.66 726 0.50 0.33
Duzce-Turkey 12.04 293.57 0.4 0.81
Kocaeli-Turkey 15.37 281.86 0.38 0.36
Landers 2.19 1369 0.08 0.79
Bam-Iran 1.7 487.4 0.20 0.81
El Mayor-Cucapah-Mexico 29 264.57 0.18 0.23
Irpinia Italy-01 22.56 213.44 0.14 0.22
The spectra of the selected ground motions along with the fundamental period of the SSI
and soil model are shown in Figure 3-9 and time history of the selected ground motions can be
found in Appendix B.1.
38
3
Bam Duzce
2.5 El Mayor Hector Mine
Imperial Valley Irpinia
2 Kobe Kocaeli
Landers Superstition Hills
Sa (g)
1.5 T1-Structure=0.31s T1-Soil=0.35s

T1: Fundamental Natural Period
1
0.5
0
0 0.5 1 1.5 2 2.5 3 3.5 4
T1-Str=0.31s T1-Soil=0.35s Period (s)
Figure 3-9 Response spectra of the selected input motion with 5% damping and periods of interest.
3.3.4 Damping and Elastic Analysis
Due to various dissipative mechanisms involved, damping of a soil-structure system becomes a

complex topic. Various researchers and code provisions suggest a range of damping ratios for the
entire soil-structure system. In 1969, The U.S. Nuclear Regulatory Commission (NRC) defined an
upper limit of 15% for the soil-structure interaction damping (Stevenson, 1980). Caltrans’s seismic
design criteria suggests a 10% damping ratio for bridges that are heavily influenced by energy
dissipation at the abutments (Caltrans, 2013). Feng and Lee (2009) suggest a damping ratio
between 3 to 12% for the entire bridge system. In this study, Rayleigh damping with two separate
damping ratios are incorporated for the soil and the concrete components. As suggested by Kwon
and Elnashai (2008) a damping ratio of 4% is applied to structural components on modes 1 and 10
(see Figure 3-10 (b) and (c)). The frequencies used to define damping ratio are chosen according
to Ventura et al. (2011). A damping ratio of 12.5% is applied to soil layers including the abutment
embankments on modes 1 and 12 (see Figure 3-10 (a)).
Rayleigh damping coefficients (α and β) used in both bridge and soil models are estimated
and summarized in Table 3-5 using damping ratio 4% and 12.5% for structural components and
soil layers, respectively.
39
Table 3-5 Defined Rayleigh damping coefficients in the Soil and Soil+Bridge models
Rayleigh Damping Coefficients, C=αM+βK

Component 2ξ1 12 2ξ
α= β=
( 1 + 12 )
 (1 + 12 )
Structural elements 1.485 0.0005
Soil layers and embankments 2.58 0.006
In addition to Rayleigh damping, the models include non-reflecting boundary condition

that incorporates the effect of radiation damping. 3
2.5
0.25 3
Damping Ratio ModalDamping-
Modal Damping-
2
AmbientVibration
Ambient Vibration Test
Test
2.5
ξn (%)
0.2 1.5
ξ1=1.4% Modal Damping-
2 1
0.15 ξn (%) Ambient Vibration Test
ξ=0.125 1.5 0.5
ξ1=1.4%
ξ10=0.1%
ξ
0.1
1 0
Deck relative displacement (cm) Free-field acceleration (g)
0.05 0 10 20 30
0.05 0.5 f1=3.37Hz f10=23.94Hz
f (Hz)
f12-Soil=3.80 Hz ξ10=0.1%
0 0
0 2 4 6 8 10 12 0.00 0 10 20 30
f1-Soil=2.89 Hz f (Hz) f1=3.37Hz f10=23.94Hz
f Time
(Hz) (s)
(a) -0.05 (b)
Impact-type earthquake loading
Maximum response
Subsequent peaks from near-free vibration
Deck relative displacement (cm) Free-field acceleration (g)
0.05 0.6
Δu ξ <0.1 ξ =δ/2π ; δ=ln(Δu)
0.00 0.0
Time (s) Time (s)

-0.05 -0.6
Impact-type earthquake loading
Maximum response
(c) from near-free vibration
Subsequent peaks (d)
0.6
Δu ξ <0.1of damping
Figure 3-10 Identification ξ =δ/2π ; δ=ln(Δu)
properties of the MRO: (a) Damping ratio (ξ) vs. frequency for the soil
layers, (b) Modal damping vs bridge frequencies obtained from the Ambient Vibration Test (Ventura et al.,
0.0 (c) and (d) An impact type earthquake loading used to identify damping ratio (ξ) of the superstructure
2011),
Time (s) (Δu) by Kwon and Elnashai (2008), respectively.
based on the calculated decaying
-0.6 (c)
To extract the bridge foundation motions and free-field motions, Response Time History
(RTH) analysis is performed on the SSI and free-field model. In these models the input excitations
40
is applied at the rock base (bottom of the supporting deposit soil) in y-direction. Excitations and
boundary conditions in the SSI model and free-field soil model are shown in Figure 3-11.
z
y
Soil+Bridge Model
(SSI Model)
Infinite Infinite
Boundary Walls (a)
Boundary Walls
Soil Model
(Free-field Model)
(b)
Figure 3-11 Elevation view and boundary conditions (a) SSI model, (b) free-field soil model
To avoid losing ground motion peaks, finite element time steps are chosen to match the
time intervals of the ground motion records which ranged between 0.0029s and 0.01s. These input
motions are applied to the base of the bottom soil layer along the bridge deck direction.
As the peak accelerations does not offer an objective measure of the system’s response,
spectral acceleration is considered as an output of the finite element simulation. Spectral
acceleration not only offers a more objective assessment of the system behavior, but also it is
consistent with code provisions.
3.4 Results
3.4.1 Effect of Kinematic Soil-Structure Interaction on Foundation Motion: Amplification
or Reduction?
To study the variation of the bridge foundation motion from the free-field motion, the concept of
transmissibility function is employed here. The Transmissibility Function (TF) is defined by the
ratio between the Fourier Transform (FT) of the bridge foundation acceleration ( u bf ) to that of
41
the free-field acceleration ( u ff ). The Transmissibility Function given in Equation (3.2) can be
used to determine the spectral amplifications and reduction as a function of frequency.
FT(u bf ) i ( )
TF( ) = = A( ) e (3.2)
FT(u ff )
where, FT is the Fourier Transform function,  is frequency, A( ) = TF( ) is the amplitude of the
transmissibility function, and  () is the phase angle which is an indicator of the delay between
the input and output signals is defined as per Equation (3.3).
ln ( TF( ) A( ) )
 ( ) = (3.3)
i
where,
ln is the natural logarithm, and
i is the unit imaginary number
The numerical evaluation of Equation (3.2) is performed using the Fast Fourier Transform
(FFT) algorithm. Figure 3-12 shows the transmissibility amplitude (A) as a function of frequency.
This is obtained using the response of points 1 and 2 as shown in Figure 3-6.
2 2
1.8 Amplification Reduction 1.8 Bam
1.6 1.6 Amplification Reduction Duzce
1.4 1.4 El Mayor
1.2 1.2 Hector Mine
1 1 Imperial Valley
A
0.8 0.8 Irpinia

0.6 0.6 Kobe
0.4 0.4 Kocaeli
0.2 0.2 Landers
0 0 Superstition Hills
0 2 4 5.5 6 8 10 12 0 2 4 6 8 10 12
f1-Soil=2.89Hz Frequency (Hz) f3-Soil=3Hz Frequency (Hz)
(a) (b)
Figure 3-12 Transmissibility factors calculated from RTH analyses for a) abutment footing foundation and b)
pier foundation.
As shown in Figure 3-12, the pier foundation shows an initial slight amplification up to
3Hz, which corresponds to the soil’s third mode frequency (f3-Soil ≈3Hz). Past this point, a
reduction is observed. The abutment foundation motion shows an amplification in the frequency
42
range below 5.5Hz with the peak occurring at soil’s natural frequency of 2.89Hz (f1-Soil). The
occurrence of the peak amplification at the natural frequency of the soil system was also observed
by Pandey (2013). A de-amplification is also observed in the abutment motion past the 5.5Hz
range. It is worth noting that the natural frequency of the bridge system resides within the
amplification regime. As a result, the analysis based on the free-field as the input excitation would
underestimate the bridge response.
To compare amplification of foundation motion, the amplitude of the transmissibility
function for abutment and pier foundation are presented in Figure 3-13 as a function of a
dimensionless parameter, α, shown in Equation (3.4). The α parameter is a measure of ratio of the
foundation dimension to wavelength of the harmonic motion moving at the wave velocity, which
allows to express the transfer function amplitude model in a unique way and more convenient
form.
D 2πD
α= = (3.4)
Va λ( )
where, ɷ is the cyclic frequency, D is the foundation dimension along the wave direction, Va is
apparent velocity of the waves with respect to the ground surface, and λ is the wavelength.
The graphs shown in Figure 3-13 are calculated for D/Va ratios equal to 0.004s
(corresponding to the abutment foundation as shown in Figure 3-13(a)) and 0.02s (corresponding
to the pier foundation as shown in Figure 3-13(b)) for abutment and pier cases, respectively. These
values are calculated given the abutment and pier foundation dimensions encountering the wave.
43
1.8 1.8
1.6 1.6
mp ification
1.4 1.4
1.2 eduction 1.2
1 1
mp ification
0.8 0.8
0.6 0.6
eduction
0.4 0.4
0.2 0.2
0 0
0 π 32 π 16 3π 32 π 0 π8 π4 3π π2
α= a α= a
(a) (b)
Figure 3-13 Average amplitude of transmisibility function plotted as a dimensionless unique function of
α=ɷD/Va (a) Abutment foundation with D/Va=0.004s (b) Pier foundation with D/Va=0.02s
As it can be seen from Figure 3-13, an amplification is occurred in the short frequency
region in the abutment and pier foundation motions compared to the free-field motions.
Particularly, abutment foundation motion demonstrates a higher amplification compared to the
pier foundation. This amplification increase is due to the topographic amplification effect or
kinematic embedment effect as the abutment foundations were built at a higher site elevation.
Input motions should be selected in a way that outcrop of these motions leading to their
corresponding free-field motions. As predicting such input motions practically is not feasible, in
this study selected ground motions from the Peer NGA West2 database were used as input motions
and outcrop of such input motions are considered as their corresponding free-field motions. The
measured free-field motions demonstrate an amplification compared to the real free-field motions
due to the kinematic embedment effect. As a result, predicted variation of bridge foundation
motions from their corresponding free field motions could be different if the employed input
motions had an outcrop which perfectly matching with their corresponding free-field motions.
Previous studies by Elsabee and Morray (1977), Veletsos and Prasad (1989), and Veletsos
and co-workers (1997), and Kim and Stewart (2003) have developed analytical equations for
transfer functions considering the slab averaging, embedment phenomena, incoherence effects and
inclination angle. The transfer functions proposed by these works are all calculated for harmonic
shear wave propagation. The common feature of all these models is that the amplitude of the
transfer function is equal or smaller than 1.0. In other words, these models always predict reduction
and fail to predict amplification effects.
44
The work by Elsabee and Morray (1977) has been presented in FEMA 440 (2005). They
developed an analytical equation for the transfer function describing the motion on an embedded
foundation shown by Equation (3.5) (FEMA 440, 2005). They performed a series of parametric
studies on a 3D finite element cylindrical embedded foundation using the substructuring approach.
The model was subjected to vertically propagated coherent shear wave motion (Elsabee and
Morray,1977).
  eω 
 Hw(ω) = cos  
  Vs  (3.5)
Hw(ω)  0.454

where e, ω, and Vs are foundation embedment, frequency, and shear wave velocity, respectively.
The Elsabee and Morray transfer function amplitude model is plotted with the calculated
average amplitude of transmissibility function for the pier and abutment foundation and compared
with the reference line in Figure 3-14(a) and Figure 3-14(b).
1.6 Abutment Foundation Pier Foundation

1.6
Elsabee and Morray (1977) Elsabee and Morray (1977)
1.2 Reference Line (TF=1) Reference Line (TF=1)
1.2
TF or H
TF or H
0.8 0.8
0.4 0.4
0 0
0 0.3 0.6 0.9 1.2 1.5 0 0.2 0.4 0.6 0.8 1
eω Vs eω Vs
(a) (b)
Figure 3-14 Comparison of the calculated transmissibility factor with the transfer function amplitude model
by Elsabee and Morray (1977) for the (a) abutment foundation, (b) pier foundation
As it can be seen, the transfer function amplitude model by Elsabee and Morray (1977)
predicts a reduction over the whole (eɷ/Vs) range. Whereas the results of the finite element model
show both amplification and reduction in the amplitude of the transmissibility function. It should
be noted that, as discussed in FEMA 440, the Elsabee and Morray’s approach does not apply to
pile foundations. As a result, this model is not capable of predicting variation of the bridge pile
supported foundation motion compared to free field accurately.
45
3.4.2 Applicability of Tau-averaging Method to Estimate Foundation Motion
Slab averaging caused by interaction of the rigid foundation and the earthquake wave in general
leads to a reduction in the magnitude of motion experienced by the foundation. This effect is more
significant when the foundation dimension is large relative to excitation wavelength. When
foundation dimensions along the direction of the earthquake wave propagation are larger than
wavelength, an averaging effect would take place which will result in a reduction in the foundation
motion’s magnitude when compared to free field motion. Reduction in the magnitude of the motion
at the foundation increases when the ratio of the dimension of the foundation to the wavelength of
the motion increases (Clough and Penzien, 1995).
Newmark et al. (1977) introduced a numerical averaging scheme to estimate the foundation
motion from the free field based on the concept that the averaging of motion happens over a time
lag in the excitation at both sides of the foundation. Clough and Penzien (1995) later proposed a
method based on averaging of harmonic waves, known as the “Tau-averaging” scheme, using the
Fourier amplitude of the original motion and then taking the inverse Fourier transform to estimate
the foundation motion. This approach assumes a massless rigid slab foundation and calculates the
averaging of the incoming wave due to its kinematic interaction with the foundation. These
averaging techniques proposed by Newmark et al. (1977) and Clough and Penzien (1995) only can
accurately estimate the foundation motion in a specific period range. For example, Newmark’s
method can only estimate the foundation motion accurately in the short period regime, and it
overestimates the motion in the medium period range. On the contrary, the Tau-averaging method
developed by Clough and Penzien overestimate the foundation motion in the short period and
provides a good estimate in the medium range period. Both techniques provide a good estimate of
foundation motion in the long period range, where the estimated foundation motions are very close
to free-field motion (Pandey, 2013). Pandey (2013) estimated the foundation motion by applying
both techniques on the data for the Hollywood Storage Building in the Northridge Earthquake in
the north-south direction. Comparison of these estimated foundation motions. This is shown in
Figure 3-15.
46
1500
Free-field (N-S)
1200 Foundation-base (N-S)
Wave Passage Model (Newmark et al.)
Tau-factor Model (Clough and Penzin)
Sa (cm/sec2)
900
600
300
0
0 0.3 0.6 0.9 1.2 1.5
Period, T (sec)
Figure 3-15 Comparison of averaging model with observed Spectral motions at the foundation base of the
Hollywood storage building during the 1994 Northridge Earthquake in the North-South direction (Pandey,
2013).
Closed-form solutions exist for interaction of a rigid massless mat foundation with the
incoming harmonic wave and this is known as the Tau effect. To investigate the Tau effect, Clough
and Penzien considered a rigid massless rectangular mat shown in Figure 3-16. They showed that
resulting slab motions will be some average of the free-field motions in that area if the mat
dimensions (D and L) are comparable to the wave lengths (Clough and Penzien,1995).
0 x
L
Figure 3-16 Rigid rectangular foundation of a large structure
To derive a closed-form expression for Tau-averaging effect, Clough and Penzien assumed
that free-field motions only act in one direction and they are independent of that direction. A shear
wave that includes excitation in one direction and propagates along another direction is the best
47
example for imagining such motions. The ground acceleration function u gx ( y, t ) can be expanded
into an orthogonal series shown in Equation (3.6), if these motions are caused by a single wave
train moving in the y-direction with an apparent wave velocity, Va:
u gx ( y, t ) =  a i x ( t ) γ i ( y ) (3.6)
i
where, γ i ( y ) are dimensionless displacement functions that satisfy the orthogonality condition as
per Equation (3.7).

D
 γ ( y ) γ ( y ) dy = 0 (i  k )
0
i k
(3.7)
Assuming rigid foundation, the rigid body motion can be expressed in terms of translation
and rotation, depicted by functions γ1 and γ 2 in Equation (3.8).
 γ1 ( y ) = 1

 2y (0 < y < D) (3.8)
 γ 2 ( y ) = 1-
 D
The first two dimensionless displacement functions in Equation (3.8) express the
rigid-body (translation and rotation) modes of a uniform free-free beam. Where, the first term
represents uniform rigid-body translation in the x-direction over the entire base area and the second
term represents rigid-body rotation about the vertical z-axis. The acceleration coefficient a1x given
in Equation (3.9) are obtained by inserting γ1 ( y ) in to Equation (3.6) and considering the
orthogonality condition specified in Equation (3.7), and integrating with respect to y from zero to
D.
1 D
a1x ( t ) = u gx ( y, t ) dy
D 0
(3.9)
The first shape function is effective as a translational input to the structure, and the
remaining terms represent motions would be filtered out for a rigid foundation. Thus, the resulting
acceleration of the rigid base, a x ( t ) , can be written as:
1 D
ax (t) = u gx ( y, t ) d y
D 0
(3.10)
48
Using the Fourier integral, free-field ground acceleration can be expressed at an arbitrary
value of y as the combination of a sequence of harmonic terms as shown in Equation (3.11).
1 
u gx ( t ) = A ( i ) exp ( i t ) d
2π -
(3.11)
where the A ( i ) function is the Fourier transform of u gx ( t ) , given by Equation (3.12).


A ( i ) =  u gx ( t ) exp ( -it ) d (3.12)
-
The accelerations at any other location of y and u gx ( y, t ) can be written as per Equation (3.13).
1    y 
u gx ( y,t ) =  A ( i ) exp i  t -   d (3.13)
2π -   Va  
Substituting Equation (3.13) into Equation (3.10) and integrating over y the modified rigid-base
translational input acceleration, a x ( t ) , is given as:
  D  
 exp  -i  -1 
ax (t) =
1 
A ( i )   Va  
exp ( it ) d
2π -
(3.14)
  D  
  -i  
  Va  
In the Tau-average method, the  factor is defined as the ratio of amplitudes of the
harmonics in the rigid-base translational motion to the corresponding free-field amplitudes given
in Equation (3.14) and Equation (3.11), respectively.
τ ( ) =
1
2 (1- cosα ) (3.15)
α
where, α is defined in Equation (3.4), and λ is the wavelength of the ground motion and can be
calculated using Equation (3.16).
Va
λ = 2π (3.16)

The Tau factor decays from unity at α=0 (for an infinite wavelength, λ=∞) to zero value at
α=2π when the base dimension is equal to the wavelength of motion (λ=D). The variation of τ
factor as a function of frequency and apparent wave velocity is shown over the range 0< α ≤ 3π in
Figure 3-17.
49
1
0.8
0.6
τ (ω)
0.4
0.2
0
0 π/2 π 3π/2 2π 5π/2 3π
α=ωD/Va=2πD/λ
Figure 3-17 τ-factor as a function of frequency and apparent wave velocity (Clough and Penzien, 1995).
As discussed earlier, the Fourier Transform of the foundation motion is estimated by

applying a reduction multiplier, τ , to the Fourier Transform of the free-field motion as shown in
Equation (3.17).
( )
FT u est
bf
= τ( ) FT ( u ff ) (3.17)
The foundation acceleration in the time domain ( uest

bf ) is then calculated using the inverse Fourier
Transformation as shown in Equation (3.18).
u estbf = IFT ( τ( ) FT ( u ff ) ) (3.18)
The factor τ is a function of the frequency (  ), the foundation dimension along the wave direction
(D) and apparent wave velocity (Va). The Tau-average method procedure to estimate bridge
foundation motion from a free field motion is summarized in Figure 3-18.
50
Time Domain Frequency Domain
Free Field Motion ( u ff ) FT Free Field Motion ( FT ( u ff ) ) :
va   D  
× τ ( ) = 2 1- cos   
D   Va  
FT ( u est
bf ) = τ (  ) × FT ( u ff )
Estimated Foundation
( )
Motion u est
bf
IFT
Figure 3-18 Schematic Tau-average method procedures to estimate foundation motion from free field motion
When the base dimension of the foundation (D) is large compared with the wavelength of
the ground motion (λ= 2π Va/ɷ ), the Tau-effect is significant, and the resulting averaged
magnitude of foundation motion will be smaller than the free-field ground motion. Imagining
small boat and large ferry in the sea is a good example to explain slab averaging. As the
wavelength of a sea wave is relatively small compared to the dimension of a ferry, the sea wave is
felt less due to the averaging effect. While the same sea wave will be felt significantly in the small
boat as the difference of wavelength and the boat dimension is minimal. Figure 3-19 Shows the
base slab averaging effect for the relatively small and large foundations, respectively.
Dj
Di Foundations
Tj Tj
Figure 3-19 Schematic concept of base averaging in small and large foundations
The abutment foundation dimension along the encountering seismic wave is about 0.9m
whereas the pier pile cap foundation dimension is about 4.6m. Therefore, as shown in Figure
3-20(a) the Tau factor is about 1.0 for the abutment footing due to its relatively small foundation
dimension compared to the wave length of the ground motions. As a result, the abutment
foundation has little kinematic effect on the ground motions. The tau-averaging method predicts a
foundation motion that is almost identical to the free-field motion, but the model analyses show
51
that there is a significant difference between these motions. This is confirmed by Figure 3-20(a)
and (b), which shows that the free field motions are unaffected by application of the τ factor.
1
3 3
0.9 2.5 2.5

Free-Field Free-Field
2 2 Pier Foundation
Abutment Foundation
0.8
Sa (g)
Sa (g)
τ
Pier Foundation 1.5 Tau Averaging 1.5 Tau Averaging
Abutment Foundation 1 1
0.7
0.5 0.5
0.6 0 0
0 5 10 15 0 1 2 3 0 1 2 3
Frequency (Hz) Period (s) Period (s)
(a) (b) (c)
Figure 3-20 a) calculated Tau parameter for abutment and pier foundations, Comparison of Hector Mine
spectrum with 5% damping of a foundation motion with the free-field and estimated foundation motion using
Tau-averaging method, b) Abutment footing foundation, and c) pier pilecap foundation
The Tau-averaging method introduces a simplified approach to incorporate base slab

averaging due to kinematic constraint caused by the stiff surface foundation. This method was
originally designed for mat slab foundations and does not capture other kinematic SSI aspects such
as embedment and wave scattering effects. It is considered that the stiffness of massless
foundations results in base slab averaging and preventing them from matching free-field
deformations. However, as it was shown earlier, the Tau-averaging method did not accurately
estimate the foundation motions for the abutment and pier foundations in this case study. Tau
parameter is always less or equal to one as it is only relying on base slab averaging effect, therefore
it cannot predict amplification of the foundation motion due to the embedment and wave scattering
effects. The amplification observed for abutment foundation is believed to be mainly due to the
embedment effect due to the increase in the height of the soil at the abutment footing foundation
compared to the ground grade level, at which the free field motion is observed. Chang et al. have
investigated the effect of depth on intensity of the motion in soil. According to this work, both
peak acceleration and response spectra of ground motions significantly varied with depth (Chang
et al., 1989). To investigate this matter, the variation of acceleration with depth calculated from
free-field model using the Imperial Valley-06 record with a of 6.6 seconds. The calculated time
history accelerations of the first four seconds for various depths are shown in Figure 3-21.
52
1
Depth=2.17m
Acceleration (g)
0.6 (Bottom of Pier
0.2 Pilecap)
Depth=17.17m (Pier
-0.2 Pile Toe level)
-0.6
Depth=20m (Bottom
-1 of the Model)
0 0.5 1 1.5 2 2.5 3 3.5 4
Time (s)
Figure 3-21 Calculated accelerations for the various depths illustrating the embedment effect due to Imperial
Valley-06 ground motion
As it can be seen from Figure 3-21, the amplitude of the calculated motion is reduced with
depth. The elevations of the Abutment footing and Pier pilecap foundation are shown and
compared with the ground level in Figure 3-22. The slight amplification of the Pier foundation can
be due to other aspects of kinematic SSI such as vertically propagated shear waves and the wave
scattering effect of footing and pilecap foundations. As discussed by Gicev et al. (2016), wave
propagation along the base of the structure must be considered for structures with multiple
supports. This effect can lead to rocking and foundation translation.
Abutment Abutment
Approaching Approaching
Abutment Footing Abutment Footing
Embankment GL Embankment
Pier Footing
Abutment Pier Pilecap Abutment
Loose soil 7x1 Pile Group
Pier 5x5 7x1 Pile Group
Pile Group
Wave Refraction
Stiff soil
Earthquake Source
Figure 3-22 One dimensional shear beam site amplification. Abutment embankment is not shown for clarity
To assess the performance of the Tau method in averaging the foundation motion, the
spectra of the estimated motions are compared with the spectra of the foundation motion calculated
from FE analysis. The SeismoSignal (2018) software was employed to calculate the spectra of
free-field motion, estimated foundation motion as well as foundation motion extracted from FE
simulation. Figure 3-23 and Figure 3-24 show a comparison of the calculated spectral acceleration
53
using the FE analysis and Tau method for the Abutment and Pier foundations for the Imperial
Valley, Landers, and El Mayor ground motions, respectively.
Imperial Valley-06 Landers El Mayor
5 4 3
4
3
2 Free-Field
3
Sa (g)
Sa (g)
Sa (g)
2 Abutment Foundation
2 Tau Averaging
1
1
1
0 0 0
0 1 2 3 0 1 2 3 0 1 2 3
Period (s) Period (s) Period (s)
Figure 3-23 Spectra of the Abutment footing foundation, free-field, and estimated foundation based on the Tau-
averaging method with 5% damping
Imperial Valley-06 Landers El Mayor

5 3 3
4
2 2
3 Free-Field
Sa (g)
Sa (g)
Sa (g)
Pier Foundation
2
1 1 Tau Averaging
1
0 0 0
0 1 2 3 0 1 2 3 0 1 2 3
Figure 3-24 Spectra of the Pier pilecap foundation, free-field, and estimated foundation based on the Tau-
To obtain a measure of accuracy of the Tau-averaging method in estimating the spectral

acceleration of the foundation motion, an error estimator is defined as:
 (Sa
2
1 Tau
- Sa found. )
Error (%) = × 100
2 (3.19)
N Sa found.
where, N is number of pair comparison, and Safound. and SaTau are calculated spectral acceleration
of foundation using FE model and Tau-averaging method, respectively. In the above equation,
Error (%) is defined as an average of the ratio of the pairwise Euclidian distance between the
spectral accelerations and foundation spectral acceleration. The computed error for each analysis
is listed in Table 3-6.
54
Table 3-6 Error in estimated motion using Tau-averaging method compare to foundation motion calculated
using the FE model
Error of the Estimated Motions by Tau-averaging Method (%)
Event/Foundation Abutment Footing Pier Pilecap
Imperial Valley-06 12.2 11.6
Superstition Hills-02 10.8 11.1
Kobe-Japan 12.4 12.5
Hector Mine 12.8 12.4
Duzce-Turkey 12.3 12.3
Kocaeli-Turkey 13.2 11.0
Landers 13.9 13.4
Bam-Iran 10.5 14.4
El Mayor-Cucapah-Mexico 11.3 13.5
Irpinia-Italy-01 14.0 12.8
As it can be seen from the Table 3-6, the estimated foundation motion using the Tau-
averaging method shows up to14.5% difference between results for the foundation motion
calculated from FE analysis for the abutment and pier foundations. The error percentages
summarized in Table 3-6 are calculated using the error estimator provided in Equation (3.19) for
the entire period regime. The predicted errors had been substantially increased if the error estimator
was used for a shorter period regime, where the difference between the estimated acceleration
using Tau-averaging method and bridge foundation motion calculated from FE model is
significant. In this case study where natural periods of system reside in the shorter period regime,
the difference can potentially lead to an unsafe design of bridge foundation if the Tau-averaging
method is used to estimate the bridge foundation motion from the free-field motion.
3.5 Chapter Summary
In this research, numerical simulation was used to investigate the kinematic aspect of soil-
structure interaction effect on an integral abutment bridge. The comparison of the bridge’s
abutment footing and pier pile cap foundation motions with the free-field motion showed zones
with amplification of the excitation in the lower frequency regime and zones that exhibit reduction
in the higher frequency regime. The Tau-averaging method and Elsabee and Morray’s transfer
function amplitude models were applied to estimate bridge foundation motion, and it was shown
that these methods are not capable of capturing the amplification regime in the short frequency
55
ranges. It is important to note that the natural frequency of the bridge system resides within the
amplification regime. The observed peak amplification occurred at the natural frequency of the
soil model. Therefore, use of the spectra from the free-field in design may lead to an
underestimated design.
In this study, the record ground motions from the PEER NGA West2 database were
employed as the input motion to the base of the finite element models. This is an inherent limitation
of the numerical modeling due to lack of record motion data at the bedrock. This might lead to
alteration of the both free-field and foundation responses.
In addition to the slab averaging mechanism that leads to reduction in excitation magnitude,
there are other kinematic SSI effects (e.g. embedment effect, wave scattering and vertically
propagated shear waves) that can lead to both amplification and reduction of the foundation motion
compared to the free-field.
56
Chapter 4: Collapse Assessment of the Meloland Road Overcrossing (MRO)
Performance-based seismic design approach offers an objective framework for design of

new structures and assessment of performance of existing structures. The main feature of
Performance-Based Design (PBD) is that it allows engineers to design a structure based on
functional objectives of service and damage states. It also provides flexibility of choosing different
materials and designs if performance objectives are met (Mitchell, 2019). The employed
performance-based methodology includes the following benefits:
1) The performance assessment procedure follows a logical sequence of steps including
seismic hazard characterization, simulation of structural response, and damage and loss
assessment.
2) The outcomes of a performance-based analysis can be summarized in terms of four well
established output variables: Earthquake Intensity Measure (IM), Engineering Demand
Parameters (EDP), Damage Measures (DM), and Decision Variables (DV).
3) The performance-based approach enhances seismic risk decision-making through
objective design and assessment methods with a sound scientific base.
4) The performance-based methodology provides a probabilistic framework that can be
employed by the next generation of seismic design codes.
Canada has been among the first countries to adopt performance-based design for its bridge code
and seismic retrofit guidelines for low and medium rise buildings. The Canadian Highway Bridge
Design Code (CHBDC) has incorporated performance-based approach in its provisions (CSA S6-
14, 2014). In addition, the US Federal Highway Administration (FHWA) Seismic Retrofitting
Manual for Highway Structures has employed PBD approach to assess performance of bridges
(FHWA, 2006). However, the effect of Soil-Structure Interaction (SSI) has not been considered in
detail in these code provisions.
In this study, the PBD approach is employed to assess the performance of integral abutment
bridges while considering the effect of soil-structure interaction (SSI). The main objective of this
study is to provide a better understanding of how soil-structure interaction could affect the collapse
57
potential of RC bridges subjected to earthquakes. To achieve this goal, a performance-based
framework is employed to assess seismic response of RC bridges.
The relationship between PBEE parameters is given by a triple-integral formula known as
the “PEER equation”, originally introduced by Cornell and Krawinkler (2000). The equation
became a framing equation to coordinate the research work by the Pacific Earthquake Engineering
Research Center. However, it is not particularly suitable for obtaining probability distributions for
cost using loss curves (Cornell and Krawinkler , 2000)( Moehle and Deierlein, 2004). The PEER
equation is given as:
   d G( dm | edp ) dG ( edp | im ) dG ( im )
G (dv) = 
0   G ( dv | dm ) .
0 0 d dm
.
dedp
.
dim
d dm.d edp.dim (4.1)
where,
G () is complementary CDF of a random variable,
dv is decision variable such as monetary loss due to repairs, downtime, and
potential injuries and deaths,
dm is damage measure,
edp is engineering demand parameter, and
im is ground motion intensity measure.
The methodology of PBEE is shown in the above PEER triple integral equation is based
on the total probability theory. The multiple conditional form of this equation allows for
discerption of the outcome of interest (e.g. cost of failure) in terms of intermediate parameters such
as hazard intensity measure, engineering demand parameters, and damage measures. This
approach provides a framework for collaboration of various disciplines including seismologists,
geotechnical and structural engineers. The outcome of this approach can be used by decision
makers (governments, city, DoTs, etc.), and insurance providers to name a few.
An assessment framework using performance-based earthquake engineering along with
parameters for collapse assessment within this research are presented in Table 4-1. This study
focuses on the damage measures and risk of failure as an outcome of the performance-based
framework, and it does not include the cost aspects.
58
Table 4-1 Assessment framework for performance-based earthquake engineering
Process Output Variable Discipline Key Parameters
Seismic Hazard IM: Intensity Measure Seismology, a) Fault location & type
Analysis • Sa(T1) Geotechnical b) Location & length of rupture (M-R)
Site → IM Engineering c) Site soil condition
Structural EDP: Engineering Demand Structural a) Model parameters
Analysis Parameter and b) SFRS response
IM → E P • Abutment drift Earthquake c) Incremental Dynamic Analysis (IDA)
• Pier drift Engineering
Damage DM: Damage Measure Structural a) Limit State
Assessment • Component Damage and b) Simulated and non-simulated collapse modes
E P→ M • Collapse Earthquake c) Probability of collapse
Engineering d) Adjusted collapse Margin Ratio (ACMR)
In this chapter, spectral acceleration is considered as the earthquake Intensity Measure

(IM). Drift ratio and base shear of pier columns and abutment backwalls are the Engineering
Demand Parameters (EDPs). Probability of collapse is used as the outcome of this performance-
based approach.
This chapter includes a description of the performance assessment calculated using the
approach detailed below:
a) Developing discrete finite element model(s) to simulate the bridge and soil
b) Performing Incremental Dynamic Analyses (IDA)
c) Predicting probability of collapse
d) Calculating and comparing the Adjusted Collapse Margin Ratio (ACMR) of the SSI-
Discrete and SSI-Continuum bridge models
The discrete approach provides a practical way of incorporation of effect of soil in
structural analysis and has gained significant interest in engineering practice.
In this Chapter, simulation includes a detailed representation of the Meloland bridge and
its behavior through failure. Depending on the functionality of the bridge, code provisions mandate
certain service and damage-level requirements. As an example, Table 4-2 summarizes the CHBDC
code requirements for the lifeline, emergency and other bridges. Performance-based analysis
framework can be employed to assess the probability of bridge failure where failure event is
defined according to performance requirements established by design codes. In this study, collapse
is defined as failure of the Seismic Force Resisting System (SFRS), which is leading to the
structural instability or catastrophic failure. In this study, collapse does not include local failure of
59
components not contributing to the SFRS, e.g., potential life-threatening failure of the non-
structural systems.
Table 4-2 Proposed seismic design performance criteria by the Seismic Subcommittee of the CHBDC (CSA S6-
06, 2006)
Seismic Ground motion Probability Service Level Damage Level

of Exceedance (return period)
Lifeline Bridges
2% in 50 years (2475 years) Possible loss of service Significant (No collapse)
5% in 50 years (975 years) Limited Repairable
10% in 50 years (475 years) Immediate Minimal
Emergency-Route Bridges
Other Bridges
2% in 50 years (2475 years) Possible loss of service No collapse
In this research, different significant failure modes of primary structural components that
contribute to collapse are simulated in all developed bridge models. Different damage states for a
typical reinforced concrete member are indicated in Figure 4-1.
Behavior Elastic Inelastic Collapse (structural instabilities)

Damage Repairable Irreparable Severe Extreme Key:
Performance Level IO Op LS CP NC IO: Immediate Occupancy
OP: Operational
LS: Life Safety
Ultimate Capacity
CP: Collapse Prevention
Lateral Load
NC: Near Collapse
Yield of steel
Reinforcement
Concrete cracking
Drift
Figure 4-1 A typical structural performance and associated behavior and damage states (after Ghobarah, 2001)
60
4.2 Development of nonlinear Archetype Models for Collapse Assessment
As described in FEMA P695 (2009), an index archetype configuration is a nonlinear prototypical

representation of a Seismic-Force-Resisting System (SFRS) configuration. This model represents
the key structural components and the behaviors associated with collapse performance when the
system is subjected to a strong earthquake motion. The index archetype model can be basic or
complex idealization of a system; however, should be able to capture significant behavioral modes
and main design features of that system. For the collapse assessment purpose, a sufficient number
of the archetype models, which is practically feasible for the collapse evaluation, needs to be
developed to capture all expected failure modes of a system.
In this study, four different discrete archetype models representing the Meloland Road
Overpassing (MRO) bridge are developed. The difference between these archetype models (D1-
D4), is in the way that SSI features are represented in the FE model.
D1, D2, and D3 are models of MRO based on previous studies considering viscoelastic (D1
model) or elastic (D2 and D3 models) simulation of embankments and center bent (Zhang and
Makris, 2002; Douglas and co-workers, 1991; and Caltrans Method A, 1989). In these models,
which are called simple-SSI models in this thesis, piles are not explicitly simulated, and instead,
the effect of soil is represented using lumped springs acting on pier pile cap and abutment footing
foundations (see Figure 4-2). Archetype model D1, which is based on work of Zhang and Makris
(2001) includes springs and dampers representing soil stiffness in both horizontal and vertical
directions. Archetype model D2, based on Douglas and co-workers (1991) includes only springs
in the horizontal and vertical directions. D3 archetype model is based on Caltrans Method A (1989)
and includes only horizontal springs (Zhang and Makris, 2001).
Archetype model D4 includes representation of individual piles supporting pier and
abutment walls. D4 model setup is discussed in detail in Section 4.2.8. To account for additional
bridge deck concrete top slab and barriers, a distributed mass equal to 1.64 tons/m is assigned in
all the models.
61
D1 Model D2 Model
Free-field motions Free-field motions

(a) (b)
D3 Model
Free-field motions
(c)
Figure 4-2 An elevation view of idealized model: (a) Viscoelastic embankments and center bent, (b) and (c)
Elastic support at embankments and center bent
Incremental Dynamic Analysis (IDA) is performed on these archetype models using 22

selected ground motions with the aim of constructing fragility curves for each archetype model.
Failure mode at collapse level is studied from the simulations of each bridge.
Step 1 Step 2
Discrete index Determining probability of collapse

archetype models of the index archetype models and
IDA investigating their failure modes
of MRO (D1, D2,
D3 and D4 model) leading to collapse
Figure 4-3 General framework of the study research presented in Chapter 4
62
4.2.1 Choice of Software for Discrete Approach
In this study, the SeismoStruct software was employed to develop and simulate the discrete
models. Some of the key features of SeismoStruct software are summarized in the following list
(SesimoSoft, 2019):
a. The nonlinear material behavior can be simulated using the fiber-based modeling
approach.
b. The SeismoStruct finite element package has a large library of hysteretic models
and 3D elements that can be used with a wide variety of pre-defined steel, concrete,
and composite section configurations.
c. The nonlinear analysis solver of the software accounts both for geometric and
material nonlinearities.
d. The software provides both auto-defined and user-defined performance criteria
with the ability of code-based capacity checks for different structural parameters,
such as frame element chord rotation, section curvature, strain for steel, concrete
and composite sections, and frame element shear capacity.
e. It allows performing different types of analysis including nonlinear dynamic time-
history analysis, pushover analysis, incremental dynamic analysis, eigenvalue, and
response spectrum analysis.
As discussed earlier, four 3-D finite element models of Meloland Overcrossing Road
(MRO) are developed in SeismoStruct software to study soil-structure effect in response of the
structure. These numerical models are developed considering different discrete type soil-structure
interaction features and indexed as D1, D2, D3, and D4 archetype model. Pier piles are not modeled
in D1, D2 and D3 index archetype models. Instead, the presence of the approach embankments and
pile foundation of the MRO are approximated using spring and dashpot values adopted from
previous studies. Detail of D1, D2 and D3 index archetype models is presented in Section 4.2.7. In
the index archetype model D4, abutment and pier piles are represented in detail. In this archetype
model, the presence of the abutment embankment (or lateral response of abutment systems) is
considered using a numerical simulation model proposed by Shamsabadi et al. (2010). In this
archetype model, lateral soil resistance around the abutment and pier piles are considered using
lateral pile-soil support curves (p-y curves) included in the API (2007) provisions and the manual
63
of computer program Ensoft LPile. More details about the SSI features of the archetype model D4
are provided in Section 4.2.8.
4.2.2 Element Type
There are various approaches to model the nonlinear behaviour of beam-columns including
concentrated plasticity (plastic hinge), distributed plasticity fiber elements, or continuum finite
element method. In the concentrated plasticity approach, all the nonlinear effects are lumped into
inelastic springs known as plastic hinges. These plastic hinges provide a macro-scale
representation of effective moment-rotation of the structural element in the plastic zone.
Distributed plasticity fiber elements use longitudinal fibers with uniaxial behaviour representing
effective response of the reinforced concrete. In terms of level of detail in representation, the fiber
approach offers a meso-scale model that sits in between the localized hinge models and the
continuum approach. The continuum approach offers a micro-scale representation of the
reinforced concrete member which typically includes three-dimensional representation of concrete
and reinforcing steel bar. All the above modeling methods rely on mathematical models that are
calibrated to simulate nonlinearity observed in tests. As presented in NEHRP (2013), different
ways of modeling structural elements are shown in Figure 4-4. The fiber method is used for the
discrete modeling of the MRO bridge.
Plastic Nonlinear Finite Length Fiber Finite

Hinge Spring Hinge Hinge Zone Section Element
Macro-Models Meso-Models Micro-Models
Figure 4-4 Different modeling methods of the structural components (NEHRP, 2013).
In the discrete simulations presented in Chapter 4 and Chapter 5 key structural components
in the models are defined using distributed inelasticity elements to simulate structural nonlinearity.
64
In SeismoStruct, different types of distributed inelasticity elements are available to simulate
nonlinear behavior of the structural components based on the fiber approach, which can be used
to determine the cross-section behavior of the beam-column members. In the fiber approach, a
uniaxial stress-strain constitutive relation is associated with each fiber. As a result, the sectional
stress-strain state of the beam-column elements can be obtained through the integration of the
nonlinear uniaxial stress-strain response of the individual fibers. The inelastic hysteresis behavior
of the cross-sections is captured by the nonlinear constitutive law assigned to the fibers and
therefore there is no need to explicitly assign a hysteresis loop to the sections. The implementation
of inelasticity distribution along beam-column elements using the fiber approach in SeismoStruct
software is shown in Figure 4-5 (SeismoSoft, 2020).
Gauss Node B
Section b
Gauss
Section a
σ Node A
A
L/2 3 L/2
σ
σ = + +
ε
RC Section Unconfined Confined Steel Fibers
Concrete Concrete
ε Fibers Fibers
Figure 4-5 Discretization of a typical reinforced concrete cross-section (SeismoSoft, 2020)
Inelasticity distribution can be implemented in force-based and displacement formulations.

In force-based formulation, forces and moments are employed as the degrees of freedom;
therefore, the equilibrium condition is automatically satisfied. The force-based approach tends to
improve the convergence of the numerical solution over displacement-based formulation in highly
nonlinear conditions.
65
In force-based formulation, accuracy of the solution can be improved by increasing either
number of the elements or integration points; especially, increasing number of integration points
results in faster convergence. While in displacement-base formulation, the accuracy of solution
only improves by increasing number of elements.
In SeismoStruct, both force-based and displacement-based elements are developed based
on the fiber approach. The inelastic force-based element is recommended to use when material
nonlinearities are presented in the modeling members and an efficient solution is required. In this
study, inelastic force-based 3D beam-column element is used to simulate deck slab, webs and
diaphragms, abutment walls and footings, and pier footing and pilecap foundations in the discrete
models. A total number of 1000 of section fibers are considered for abutment walls, deck slab
segments and diaphragms to capture inelasticity. Similarly, 500 section fibers are considered for
each web and foundation. Gauss-Lobatto quadrature scheme is used to analyze the FB section
behavior in SeismoStruct. To capture the inelastic effects in these elements 10 integration sections
are considered at the following locations: [-1 -0.920 -0.739 -0.478 -0.165 0.165 0.478 0.739 0.920
1] × L/2, where, L represents length of an element.
In SeismoStruct software, ‘inelastic force-based plastic hinge element’ is developed based
on plastic hinge integration method, which introduced by Scott and Fenves (2006). This element
type limits the development of inelastic response in the member to plastic hinges formed with a
given length at both ends of the beam-column member (SeismoSoft, 2020). The inelastic force-
based plastic hinge frame element type with a total of 1000 section fibers and 15% ratio of plastic
hinge length to the length of column (Lp/L) is considered for the pier column in all the archetype
models. The discretization of sections of the Pier pilecap, Pier column and abutment walls are
shown in Figure 4-6.
66
(a) (b)
(c)
Figure 4-6 Section discretization for the MRO's index archetype models (a) Pier pilecap, (b) Pier column, and
(c) Abutment backwalls
4.2.3 Nonlinear Geometry Model
Large displacements/rotations due to cumulative gravity forces is known as P-Δ effect. It is

important to consider the P-Δ effect since it reduces the effective stiffness of the structure and is a
major cause of failure. P-Δ effects are especially significant for bridge structures with long and
relatively flexible pier columns. P-Δ effect is considered by selecting nonlinear geometry option
in Response History Analysis (RHA).
4.2.4 Material Properties
A linear nonlinear concrete model proposed by Mander and co-workers (Mander et al., 1988) is
used as the material model for the pier columns and abutment backwalls in the archetype models.
The confinement effects provided by the lateral transverse reinforcement are incorporated through
the rules proposed by Mander et al. (1988) in this constitutive model. In addition, confining
pressure is assumed constant throughout the entire stress-strain range. Material properties for the
pier column and abutment backwalls are identical to those considered by Werner et al. (1993).
These values are summarized in Table 4-3 and Table 4-5.
67
Table 4-3 Nonlinear concrete material properties for the MRO pier column and abutment backwalls (Werner
et al., 1993)
Parameter Value
Compressive strength (f’c) 35.6 (MPa)
Tensile strength (f’t) 3.6 (MPa)
Modulus of elasticity (E) 28248 (MPa)
Strain at peak stress (εc) 0.0022 (mm/mm)
U timate Strain (εu) 0.003 (mm/mm)
Specific weight (γ) 23.54 (kN/m3)
Linear elastic material models are considered for the deck slab, foundations and piles.
These models are simplified uniaxial elastic material models with symmetric behavior in tension
and compression. The assigned modulus of elasticity and specific weight values considered for the
deck slab, foundations, and piles (only D4 model) are summarized in Table 4-4 .
Table 4-4 Parameters of assigned elastic material model to the deck slab, foundations and piles
Parameter Deck Slab and Foundations Timber Piles (only D4 Model)
Modulus of elasticity (E) 28248† (MPa) 12400* (MPa)
Specific weight (γ) 23.54† (kN/m3) 6.86* (kN/m3)
Note:
†
Values adopted from Werner and co-worker (1993)
* Values adopted from Kwon and Elnashai (2008)
Menegotto-Pinto steel model is considered as a material model for the reinforcing steel
bars. Ten model calibrating parameters are defined in order to fully describe the mechanical
characteristics of the material. These parameters are summarized in Table 4-5.
68
Table 4-5 Material properties of reinforcing steel bars for the MRO archetype models
Material Properties Value
Modulus of elasticity (E) 2.0E5† (MPa)
Yield strength (fy) 312.3† (MPa)
Strain Hardening Parameter (μ) 0.005* (-)
Transition curve initial shape parameter (R0) 20* (-)
Transition curve shape calibrating coefficients (a1 & a2) 18.5 and 0.15* (-)
Isotropic hardening calibrating coefficients (a3 & a4) 0 and 1* (-)
Fracture buck ing strain (εult) 0.1* (-)
Specific weight (γ) 78* (kN/m3)
Note:
†
Values adopted from Werner and co-workers (1993)
* SeismoSoft (2020)
4.2.5 Sectional Response of Pier Column and Abutment Backwalls
To visualize the sectional response, the Response-2000 software was employed. Response-2000
is a sectional analysis program developed at the University of Toronto, which can accurately
calculate the strength and ductility of a reinforced concrete cross-section subjected to axial load,
shear, and moment simultaneously (Bentz and Collins, 2001). The predicted section response
considering a 5382.35 kN axial load (Axial load/Axial Capacity =15.2%) is compared with the
column response given by Werner et al. (1993). The results are shown in Figure 4-7.
69
12,000
10,000
Bending Momemnt (kN m)

5,000
Shear Failure 8,000 Werner et al. (1993)
4,000
Flexure-Shear Response 2000
Failure
Shear (kN)
3,000 6,000
2,000 4,000
1,000 Flexure 2,000

Failure
0 0
0 3000 6000 9000 12000 0 0.002 0.004 0.006 0.008
Moment (kNm) Curvature (rad/m)
(a) (b)
Figure 4-7 Sectional response of pier column of the Meloland Road Overcrossing (a) AASHTO-99 LRFD M-
V interaction (Response-2000, 2001) , (b) Moment-Curvature diagram calculated by Werner et al (1993) and
Response-2000 software (2019)
Figure 4-8 shows the calculated sectional response of the MRO abutment backwall using
Response-2000 considering an axial force of 2691kN (Axial load/Axial Capacity=16%).
3000 Shear Failure 3000
2500 2500
Moment (kNm)
2000 2000
Shear (kN)
Flexure-Shear
Failure 1500
1500
1000
1000
500
500 Flexure
Failure 0
0 0 0.02 0.04 0.06 0.08
0 500 1000 1500 2000 Curvature (rad/m)
Moment (kNm)
(a) (b)
Figure 4-8 Sectional response of abutment backwalls of the Meloland Road Overcrossing (a) AASHTO-99
LEFD M-V interaction (Response-2000, 2001) , (b) Moment-Curvature diagram calculated by using Response-
2000 software (2019)
The M-V interaction and M-ϕ diagrams shown in Figure 4-7 and Figure 4-8 are used to
visualize interactive behavior between the bending ‘Moment and Shear’ and ‘Moment and
Curvature’ responses induced in the section of the pier column and abutment backwall of the MRO.
To accurately calculate these diagrams, actual member’s axial force should be considered. As
indicated in Figure 4-7(a) and Figure 4-8(a), the M-V interaction diagrams can be classified into
70
three regions: shear failure, flexural-shear failure, and flexural failure. Indicated shear-flexure
mode in these figures is a combined mode that is typically initiated by flexural cracking and
progresses into diagonal shear cracks.
4.2.6 Structural Damping
Damping of a soil-structure system becomes a complex topic due to various dissipative

mechanisms involved. Various researchers and code provisions suggest a range of damping ratios
for the entire soil-structure system. Modal frequencies and damping ratios of the Meloland Road
Overcrossing have been studied by many researchers. A summary of studies carried out by Zhang
and Makris (2002), Werner et al. (1987), Wilson and Tan (1990), Gates (1993) and Werner (1994)
is presented in Table 4-6. The modal frequencies and damping ratio percentages shown in this
table for the first six modes of the MRO have been analytically computed in these studies. As it
can be seen, the computed modal damping ratios,  j , vary significantly between the studies and
are noticeably larger than the 5% modal damping prescribed by Caltrans. The values of the first
modal damping ratio reported by Werner (1994) are the highest among the presented
corresponding values computed by other researchers. It is worthy to note that these values are
much greater than the values reported by the Werner and co-workers (1987) and other scholars
during earlier studies on the same bridge (Zhang and Makris, 2002). In addition, the value of the
computed second modal damping by Zhang and Makris (2002) is the highest ( 2 = 56% ) among
all the damping ratio values calculated by other researchers. According to Zhang and Makris, the
main reason for the such high value of modal damping is mobilization of a large volume of the
high damping soil due to the framing action that takes place as a result of the bridge configuration
(integral abutments and straight configuration of the deck).
71
Table 4-6 Modal frequencies (ωj) and damping ratios percentage (ξj) of Meloland Road Overcrossing (Zhang
and Makris, 2002)
Modes 1 2 3 4 5
ωj ξj ωj ξj ωj ξj ωj ωj ξj
1st Transverse 13.8 18.7 15.5 7.2 14.3 ~ 15.7 6.6 ~ 12.7 15.6 16.3 19 ~ 26
Longitudinal 19.4 56.8 16.7
1st Vertical (antisymmetric) 21.1 8.3 17.5
Torsion about vertical axis 17.9 100 Critically Damped Mode
2nd Transverse/torsion 26.9 28.2
about longitudinal axis
2nd Vertical (symmetric) 28.3 10.2 28.7 5.8 27.4 ~ 29.9 3.1 ~ 7.4 27.6
1: Zhang and Makris, 2002.
Note 2: Werner, Beck, and Levine ,1987
3: Wilson and Tan, 1990
4: Gates, 1993
5: Werner 1994
~ Indicate a range between two modal frequencies or damping ratio
percentages
Modal damping ratios of the bridge foundation system depicted in Figure 4-9 are calculated
by Zhang and Makris (2002) using the complex eigenvalue approach presented by Veletsos and
Ventura (1986) for the homogeneous equation of motion based on Equation (4.2).
Mu + Cu + Ku = 0 (4.2)
where M,C, and K are the mass, damping, and stiffness matrices of the bridge-foundation
idealization, respectively, and u is the free vibration response vector given as per Equation (4.3).
u = eiΩt (4.3)
where,  ,  , and i are the associated characteristic vector (mode shape), the complex
characteristic value, and unit imaginary number, respectively.
The damping matrix, C, was constructed by Zhang and Makris (2002) adopting the concept
of Rayleigh damping for the bridge superstructure and appending pre-identified lumped dashpots
at the locations where the superstructure interacts with its foundation. They assigned a 5% modal
damping ratio for the first and second modes of the undamped idealized model (bridge deck with
springs) and added a damping constant, c , that represents the presence of the embankments and
pile foundations. Considering the above, a nonclassical damping matrix of the bridge-foundation
system, C , is given in Equation (4.4).
72
C = M +  K +c (4.4)
Substituting Equation (4.3) in to Equation (4.2) results in a standard eigenvalue problem shown in
Equation (4.5).
( − M + iC + K )  = 0
2
(4.5)
Using the associated equation of a single-degree-of-freedom oscillator and considering  j
as the frequency domain parameter, the complex characteristic values Ωj can be evaluated as per
Equation (4.6) (Veletsos and Ventura, 1986).
 j =  j 1 −  2j + i j j ; j =1,2,....N (4.6)
where  j and  j are modal frequencies and damping ratios, respectively.
Using Equation (4.6) and solving for ɷj and ξj , these values can be calculated as per Equation
(4.7).
 = 2 + 2
 j jR jI
  jI (4.7)
 j =
 j
where  jI and  jR are the imaginary and real parts of the characteristic value  j , respectively.
73
f1=2.1 Hz ξ1=18.7% f1=1.96 Hz f4=4.00 Hz ξ4=100% f4=3.90 Hz
f2=2.85 Hz ξ2=56.8% f2=2.55 Hz f5=4.03 Hz ξ5=28.2% f5=4.30 Hz
f3=3.58 Hz ξ3=8.3% f3=3.41 Hz f6=4.20 Hz ξ6=10.2% f6=4.89 Hz
Figure 4-9 First six modal frequencies, damping ratios, and modes computed for stick model (left) and 3D FEM
model (right) of the MRO model by Zhang and Makris (2002)
As discussed earlier, the modal damping ratios of the bridge foundation system,  j ,
presented in Figure 4-9 were computed using Equation (4.7), which are obtained from solving for
the complex eigenvalues of the homogeneous equation presented in Equation (4.2). It should be
noted that Zhang and Makris used only imaginary part of the As it can be seen from Table 4-6,
modal damping ratio for the first transvers mode is identified as 18.7% by Zhang and Makris
(2002) and between 19%-26% by Werner (1994). For the discrete models in this study, a Rayleigh
damping with a damping ratio of 4% as suggested by Kwon and Elnashai (2008) is applied to
structural components on modes 1 and 10 as global damping (see Figures 3-4 (c)).
As discussed by Werner and co-workers, the effective damping of the soil+bridge system
is significantly higher compared to typical values used in structural analysis. They report effective
damping ratios between 19% to 26% for the first transverse mode of vibration (Werner et al.,
1993). Zhang and Makris (2002) have estimated a damping ratio of 18.7%. A summary of stiffness
and damping ratios presented in previous studies is presented in Table 4-6. In this study, a damping
74
ratio of 25% is applied to abutment backwalls, foundations and piles to capture the damping effect
of embankment and surrounding soil.
4.2.7 SSI Features in D1, D2 and D3 Index Archetype Models
The approximate value of springs and dashpots of studies previously carried out by Zhang and
Makris (2002), Douglas et al. (1991), and Caltrans (1989) are summarized in Table 4-7 (Zhang
and Makris, 2002).
Table 4-7 Spring and dashpot values that approximate the presence of the approach embankments and pile
foundation for the Meloland Road Overcrossing (Zhang and Makris, 2002)
Parameters D11 D22 D33
Kx (MN/m) 21+56* 91(365) 607+549

Pile Foundations
Embankment +
Ky (MN/m) 21+56 91(365) 596+49

Kz (MN/m) 78+356 263(1051) -
Cx (MN.s/m) 1.5+4.5 - -
Cy (MN.s/m) 1.5+4.5 - -
Cz (MN.s/m) 3+28 - -
Kx , Ky (MN/m) 260 254 (876) 175
Pile Foundation of
Kr (MN.m/rad) 7611 1888 (6509) -

Center Bent
Kxr ,Kyr (MN/rad) -409 - -

Kz (MN/m) 887 550 (1898) -
Cx , Cy (MN.s/m) 6 - -
Cz (MN.s/m) 25 - -
1. Zhang and Makris, 2002 (Shear modulus (G)=2.0 MPa and hysteretic damping coefficient (η)=0.52
for embankment soil). Values from the Zhang and Makris study are associated with the intensity of the
1979 Imperial Valley earthquake.
Note 2. Douglas, Maragakis and Vrontinos, 1991. Values in parenthesis are the optimal values identified
from dynamic tests.
3. Caltrans Method A, 1989.
(*) n1+n2 the first and second term values are corresponding to the embankment and pile foundation,
respectively.
4.2.8 SSI Features in the D4 Index Archetype Model
The archetype model D4 developed in this study includes a more detailed representation of SSI
features in the discrete model. The D4 archetype model includes explicit representation of
75
abutment and pier piles. This model also includes abutment wall-backfill soil interaction, pier
foundations-surrounded backfill soil interactions and pile lateral and vertical resistance are
calculated and considered in the model. A 3D view of this index archetype model is shown in
Figure 4-10. Detail of the calculated SSI features and their implementation of them in model D4
are presented in the upcoming sections.
Abutment Backwall Box Girder Deck Slab Pier Diaphragm Box Girder Deck Slab
Abutment
Footing
Intermediate Pier Column
Diaphragm Intermediate
Pier Footing Diaphragm
Pier Pilecap
1× 7 Abutment Pile Group

1× 7Abutment
5×5Pier Pile Group Pile Group
Figure 4-10 3D view of the index archetype model D4 constructed using SeismoStruct software.
4.2.8.1 Idealized Soil Layers
To incorporate the effect of soil surrounding the piles and determine the corresponding vertical
and lateral stiffnesses, a detailed study of soil layers is carried out. Bore hole logs of the MRO
were obtained prior to construction of the bridge and the results are shown in Figure 4-11. Based
on the description of soil layers obtained from the soil investigation report, Kwon and Elnashai
(2008) used Idealized soil layers for the MRO as shown in Figure 4-12 with material properties
summarized in Table 3-1. In this study, idealized soil layers for the bridge and their material
properties are identical to the soil layers adopted by Kwon and Elnashai (2008) in their works.
76
Site Elev.
GL B-2
481’
Soft red brown moist sandy clay 50% sand 50% clay 3.35m
470’
Compact red brown saturated fine silty sand interbedded with very stiff silty clay 3.05m
460’
No information was provided 3.05m
450’
Slightly compact brown silty to fine silty sand 3.05m
440’ 24.7m
Compact red brown fine silty sand and stiff red brown clay interbedded with slightly compact fine sand 3.05m
430’
Compact red brown fine silty sand interbedded with (sand 75% clay 25%) 3.05m
420’
Compact red brown silty sand interbedded with clay 3.05m
410’
Dense red brown silty sand 3.05m
400’
Figure 4-11 Reproduced description of soil layers on the geotechnical borehole log B2 of the MRO (Caltrans,
1969)
Embankment Embankment
7.5m Gravel Clay Gravel Clay
3.30m GL B-2 3.30m
2.17m Layer A
1.18m
3.36m Layer B
4.57m Layer C
4.57m Layer D
2.5m
5.33m Layer E
Figure 4-12 Idealized soil layers for the MRO and elevational view of the index archetype model D 4
4.2.8.2 Simulation of the Abutment-backfill Soil Interaction
Abutment-backfill soil interaction can be simulated using the Extended Hyperbolic Forced
Displacement (EHFD) backbone curves developed by Shamsabadi et al. (2010) to represent force
deflection relationships for walls of varying height. These EHFD backbone curves can be obtained
using Equation (4.8). The equation describes a hyperbolic relationship between lateral load per
unit width of the abutment wall and the wall deflection and are amendable to practical application
in seismic response simulations of bridge systems. It provides abutment longitudinal
displacements at residual capacity ymax of 0.05H for the granular backfill and 0.1H for the cohesive
backfill (Shamsabadi et al., 2010).
n
ar y  H
F( y) =   (4.8)
 H  Hr 
 +
 r b y
 Hr 
77
where,
F(y) is the induced force in kN/m
y is the abutment longitudinal displacements in cm
H is the backwall height in meter and Hr is the reference backwall height and Hr =1m
n is a dimensionless exponent
a r in kN/cm/m and br in 1/cm are back-calculated coefficients for the reference height.
These EHFD equations are empirically developed based on tests performed at the
University of California, Los Angeles (UCLA) and University of California, Davis (UCD) on
abutment backwalls subjected to compacted embankment soils. The UCLA tests were performed
on granular embankment material, whereas UCD tests were performed on embankment filled using
a cohesive soil. Table 4-8 lists the model parameters fitted for the granular and cohesive soils
(Shamsabadi et al., 2010).
Table 4-8 EHFD Coefficients Back-Calculated Using Log-Spiral Hyperbolic (LSH) simulation modified by a
wall height-adjustment factor (Shamsabadi et al., 2010)
Site backfill Type UCLA/granular UCD/cohesive

H(m) a b n a b n
(kN/cm/m) (1/cm) (-) (kN/cm/m) (1/cm) (-)
1.00 410.6 1.867 1.56 249.1 0.8405 1.05
1.25 316.6 1.468 1.56 199.4 0.6755 1.05
1.5 258.4 1.206 1.56 166.1 0.5637 1.05
1.67 230.8 1.073 1.56 149.6 0.5084 1.05
1.75 218.5 1.020 1.56 142.9 0.4856 1.05
2.00 190.2 0.8836 1.56 125.6 0.4270 1.05
2.25 168.7 0.7784 1.56 112.2 0.3811 1.05
2.5 152.8 0.6954 1.56 101.6 0.3446 1.05
Using the parameters listed in Table 4-8, the embankment passive force can be described
using Equation (4.9) for the UCLA and UCD tests.
78
 410.6y  H
1.56
   , y  0.05H (UCLA's silty sand backfill)

 ( H 1m ) + 1.867y  1m 
F( y) =  1.05
(4.9)
 249.1y  H
 ( H 1m ) + 0.8405y  1m  , y  0.1H (UCD's clayey silt backfill)

Based on the most recent Caltrans guideline, abutment effective stiffness (Keff) for seat-
type and diaphragm-type abutments can be obtained from an idealization of the initial stiffness of
the nonlinear abutment-backfill force-deformation curve as shown in Figure 4-13, and as per
Equation (4.10)(Caltrans, 2019a).
Force Force
Fabut Fabut
K eff K abut = K eff

K abut
 gap Deflection Deflection

 eff
 eff
(a) (b)
Figure 4-13 Caltrans nonlinear abutment model (a) seat-type abutment (b) diaphragm-type abutment
(Caltrans, 2019a)
Fabut
K eff = (4.10)
Δ eff
where, Fabut is the idealized ultimate passive capacity of the backfill behind abutment backwall or
diaphragm in kips unit, and it can be calculated using Equation (4.11).
  h abut
2.5

Fabut = Wabut   R sk (4.11)
 1+  h abut 
where, the β coefficient is equal to 1,567 (kN/m3.5) and 5.5 (kip/ft3.5) in metric and US customary
units, respectively,
The  coefficient is equal to 6.86 (1/m) and 2.09 (1/ft) in metric and US customary units,
respectively,
79
Rsk is the skew reduction factor and for abutments passive capacity with skew angle (θ) less than
66° is given as per Equation (4.12) (Shamsabadi and Rollins , 2014) and (Shamsabadi et al., 2020).
−
R sk = e 45
(4.12)
Wabut and habut are the width and height of the abutment wall in m (ft) as shown in Figure 4-14,
respectively.
Δeff is effective abutment longitudinal displacement in m (inch) when the passive force reaches,
and given as per Equation (4.13).
Δ + Δ abut Seat - type
Δ eff =  gap (4.13)
 Δ abut Diaphrag − type
where,
Δgap is the width of the expansion gap at the seat abutment in m (inch) for seat-type abutment, and
Δabut is the abutment displacement at idealized yield in m (inch) and given as Equation per (4.14)
.
Fabut
Δ abut = (4.14)
K abut
where, Fabut is the abutment bilinear-equivalent ultimate passive force shown in Figure 4-13 and
given in Equation (4.11) in kN (kip), and Kabut in kN/m (kip/in) unit is calculated as Equation per
(4.15) for the US customary units.
K abut = Wabut ( h abut +  ) R sk (4.15)
where,
The  coefficient is equal to 10,372 (kN/m/m2) and 5.5 (kip/in/ft2) for metric and US customary
units, respectively,
The  coefficient is equal to 11,496 (kN/m/m) and 20 (kip/in/ft) for the metric and US customary
units, respectively.
80
Wbw Wdia
Wabut
hbw h*dia
h**dia
Figure 4-14 Effective Abutment height and width (Caltrans, 2019a)
The model parameters fitted for granular backfill using EHFD approach were employed
here to develop the passive pressure response curve. A tri-linear curve was fitted to the data as
shown in Figure 4-15. In this figure, the passive pressure response is also compared with the
seismic design criteria revision 2.0 of the Caltrans (2019a) approach. The initial stiffness obtained
from the EHFD approach matches well with the proposed Caltrans model. The asymptotic passive
force shows good agreement between the two models.
16,000
14,000
12,000
Passive Force (kN)
10,000 EHFD (Granular Backfill)-

Shamsabadi et al. (2010)
8,000 SDC Rev 2.0 (Caltrans, 2019a)
6,000 Tri-linear Fit

4,000
2,000
0
0 0.05 0.1 0.15 0.2 0.25
Displacement (m)
Figure 4-15 Nonlinear abutment response in longitudinal direction of the Meloland Road Overcrossing (MRO)
The initial stiffness of the backfill is applied in the discrete model using five springs
distributed along the height of the abutment backwall. The gap-hook link model in the
81
SeismoStruct software was used. An initial gap of 25mm was used to represent a typical gap
between the backfill soil and the wall. A similar approach was also applied to calculate the passive
soil resisting the wing walls in the transverse direction.
The effective stiffness of soil supporting pier pilecap and footing foundation is calculated
using a similar method based on Equation (4.9). The calculated force-displacement curves for the
pier pilecap and footing foundation are shown in Figure 4-16. This behavior is simulated in
SeismoStruct using a tri-linear material model presented in Figure 4-17. The tri-linear behavior is
used to define the spring properties representing supporting soil for pier pilecap and abutment
footing in both longitudinal and transverse directions. A summary of assigned spring properties is
given in Table 4-9.
1,200
1,000
800
Force (kN)
Pier Pilecap
600 Pier Footing
400
200
0
0 0.02 0.04 0.06 0.08
Displacement (m)
Figure 4-16 Force-displacement relationship for pier pilecap and footing foundation of the MRO
Force or K2 Force or
Moment Moment
K1 r.K0
K0 Fy K0
K0 K0
-d2 -d1 d1 d2 Displacement Displacement
or Rotation K0 Fy or Rotation
K0 K1 r.K0
K2
(b)
Figure 4-17 Tri-linear symmetric curve employed(a)to model idealized trilinear behavior in SeismoStruct
adopting an isotropic hardening rule (SeismoSoft, 2020)
82
Table 4-9 Calculated parameters to define fully characterize the tri-linear symmetric response curve for pier
pilecap and footing foundation of Meloland Road Overcrossing
Pier Foundation d1 (m) d2 (m) K0 (kN/m) K1 (kN/m) K2 (kN/m)

Pilecap 0.01 0.02 70780 15730 4512
Footing 0.01 0.02 23350 3277 1917
4.2.8.3 Soil-pile interaction
The American Petroleum Institute (API) guidelines to simulate soil- pile interactions are based on
the results of static or slow cyclic loading tests. Despite that many scholars, including Murchison
and O’Neill (1984), Gazioglu and O’Neill (1984), Finn (2005), and Choi et al. (2013), have argued
against the validity of API soil-pile interactions curves for dynamic analysis, they have been
widely adopted in practice for solving large-scale seismic problems such as bridge system. This is
due to their simplicity, less computationally effort requirement, and lack of an alternative efficient
technique (Rahmani, 2014).
API guideline and the L-pile software manual are used to simulate soil- pile interactions in
the index archetype model D4. The API p-y springs are employed to simulate lateral soil-pile
interaction of the A, B and D layers (refer to Figure 4-12 for soil layer labels) (API, 2007). L-Pile
software’s technical manual is used to calculate p-y curves corresponding to C and E layers
(ENSOFT Inc., 2013). In addition, the API guideline is used to calculate vertical component of
soil-pile interaction (t-z and Q-z springs) (API, 2007). Abutment and pier have timber piles with
diameter of 320 mm at the top and 200 mm at the bottom. The modulus of elasticity of the timber
piles is estimated to be 12.4GPa based on the work of Maragakis et al. (1994) and Kwon and
Elnashai (2008). In the index archetype model D4, the piles are assumed to be prismatic with
circular cross section. Thus, an equivalent circular diameter is calculated as per Equation (4.16) to
ensure equivalency of flexural rigidities along the length of the piles.
32I
Dequivqlent = 4 (4.16)

where, Dequivalent is equivalent diameter of pile and I is the moment of Inertia. Given that the pile
dimensions taper off along the length, equivalent diameter of piles in the middle of each soil layer
83
is calculated and used in computing the API load-displacement curves. These equivalent diameters
are summarized in Table 4-10 for the abutment and pier piles and for different soil layers.
Table 4-10 Equivalent pile diameter considered for each soil layer in calculating pile load-displacement
relationship in the index archetype model D4
Abutment Piles Pier Piles

Soil Layer
T d a b Dequivalent t d a b Dequivalent
(m) (m) (m) (m) (m) (m) (m) (m) (m) (m)
Embankment 3.30 1.65 0.32 0.30 0.3 - - - - -
A 2.17 4.385 0.30 0.28 0.28 - - - - -
B 3.36 7.15 0.28 0.26 0.26 1.80 3.85 0.32 0.29 0.29
C 4.57 11.12 0.26 0.23 0.23 2.17 7.82 0.29 0.26 0.26
D 4.6 15.7 0.23 0.20 0.21 3.36 12.39 0.26 0.22 0.23
E - - - - - 2.50 15.92 0.22 0.20 0.20
where,
Dequivalent as defined previously is the equivalent diameter of piles for each soil layer,
d is the depth of center of soil layers from ground level,
a and b are diameter of piles at the top and bottom of the soil layers, respectively.
Figure 4-18 shows a schematic view of the springs’ arrangement is considered to simulate
soil-pile interaction in the index archetype model D4.
84
Caltrans Caltrans
Embankment Springs Springs Embankment
(Gravel Clay) (Gravel Clay)
7.5m
p-y 3.30m 3.30m

p-y
Soft Sandy Clay Pier Pilecap Pier Footing

2.17m Soil Layer A
(Sand 50%, Clay50%) Springs Springs
t-z p-y Soil Layer B

3.36m Compact Fine Silty Sand
Compact Fine Silty

4.57m Sand and Stiff Clay t-z t-z Soil Layer C
4.57m Compact Fine Silty Sand Soil Layer D
Compact Fine Silty Sand 2.5m

5.33m (Sand75%, clay25%)
Soil Layer E
interbedded with Stiff Clay Q-Z Q-Z
Figure 4-18 butment backwa s’ springs and API springs arrangement are considered in the archetype model
D4 to simulate Abutment-backfill Soil Interaction and soil-pile interaction, respectively
4.2.8.3.1 Lateral soil-pile interaction
The ultimate lateral bearing capacity (pu) for sand varies with depth. It can be calculated using
Equation (4.17) and Equation (4.18) for shallow and deep depth, respectively. The equation giving
the smallest value of pu should be used to calculate the ultimate lateral bearing capacity for sand
at a given depth (API, 2007).
p us = ( C1  H + C2  D )    H (4.17)
pud = C3  D    H (4.18)
where,
pus and pud are ultimate resistance in kN/m for shallow and deep depth, respectively,
γ is effective soil weight in kN m3,
H is depth in m,
85
C1, C2, C3 are Coefficients determined as function of φ´ (angle of internal friction of sand
in deg.) calculated based on graph shown in Figure 4-19,
D is the average pile diameter from surface to depth in m.
Figure 4-19 Coefficients used to calculate the ultimate resistance, pu (API, 2007)
The recommended p-y curve by API (2007) for laterally loaded piles, at any specific depth
H, in sands is given by Equation (4.19):
 kH  (4.19)
P = Ap u tanh   y
 Ap 
where,
A is a factor to account for cyclic or static loading condition is given by Equation (4.20):
 0.9 for cyclic loading

A =  0.8H  (4.20)
 3.0 − D   0.9 for static loading
 
H and y are depth and lateral deflection, respectively in m,
pu is the ultimate bearing capacity at depth H in kN/m,
86
k is initial modulus of subgrade reaction in kN/m3 which can be determined as function of
angle of internal friction, φ´, from Figure 4-20.
Figure 4-20 Relationship between initial modulus of subgrade reaction and internal friction angle (API, 2007)
The ultimate unit lateral bearing capacity of soft clay, pu, varies from 3c to 9c as depth
increases from 0 to XR as per Equation (4.21) (API, 2007).
 cX
3c +  X + J for X  X R
pu =  D (4.21)
 9c for X  X R
where,
pu is the ultimate resistance in kPa,
c is the undrained shear strength for undisturbed clay soil samples, kPa,
D is the pile diameter, mm,
γ is the effective unit weight of soil, MN/m3,
J is the dimensionless empirical constant with values ranging from 0.25 to 0.5 determined
from field testing,
87
X is the depth below soil surface, mm,
XR is the depth below soil surface to bottom of reduced resistance zone in mm. For a
condition of constant strength with depth, it can be calculated using Equation (4.22).
6D
XR = (4.22)
D
+J
c
When the strength varies with depth, XR should be determined by plotting the pu vs. depth using
Equation (4.21). In general, minimum values of XR should be about 2.5 pile diameters (API, 2007).
Based on the API code, lateral soil resistance-deflection relationships (p-y curves) for piles
in soft clay for the case, where the equilibrium has been reached under cyclic loading, are generated
from Table 4-11 (API, 2007).
Table 4-11 Lateral soil resistance-deflection relationships for piles in soft clay (API, 2007)
X > XR X < XR
p/pu y/yc p/pu y/yc
0.00 0.0 0.00 0.0
0.23 0.1 0.23 0.1
0.33 0.3 0.33 0.3
0.50 1.0 0.50 1.0
0.72 3.0 0.72 3.0
0.72  0.72 X/XR 15.0
0.72 X/XR 
where,
p is the actual lateral resistance in kPa,
y is the actual lateral deflection in m,
yc is calculated using Equation (4.23).
yc = 2.5 c D (4.23)
where,  c is the strain corresponding to one-half the maximum stress determined from
unconsolidated undrained compression tests on undisturbed soil samples.
88
Stiff clays are generally more brittle than soft clays and typically have nonlinear stress-
strain relationships. A good judgement is required to consider rapid deterioration of load capacity
at large deflections for stiff clays in developing stress-strain curves and subsequent p-y curves for
cyclic loads (API, 2017).
p-y curves for stiff clay and for static loading can be calculated using Equation (4.24)
(ENSOFT Inc., 2013).
 1 for y>1
p 
= y 1/4 (4.24)
p u 0.5( ) for y  1
 y50
where,
y50 is the deflection at one-half the ultimate resistance given by Equation (4.25).
y50 = 2.5 50 D (4.25)
where,  50 is strain at 50% stress level of clay with typical values of 0.01, 0.005-0.007 for medium
clay and stiff clay, respectively (ENSOFT Inc., 2013).

Using Equations (4.19),(4.21), and (4.24), force-deflection curves for abutment and pier piles were
calculated and presented in Figure 4-21.
89
2,500 200
2,000 160 Soil Layer A

Soil Layer C
p (kN/m)
p (kN/m)
1,500 120
Embankment
1,000 80
Soil Layer D
500 Soil Layer B 40
0 0
0 0.02 0.04 0.06 0.08 0.1 0 0.02 0.04 0.06 0.08 0.1
y (m) y (m)
(a)
2,500 200
2,000 160
p (kN/m)
p (kN/m)
1,500 Soil Layer D 120

Soil Layer C
Soil Layer B Soil Layer E
1,000 80
500 40
0 0
0 0.05 0.1 0 0.05 0.1
y (m) (b) y (m)
Figure 4-21 Calculated p-y curves for soil layers for the index archetype model D4 surrounded the (a) abutment
piles, (b) pier piles
A symmetric tri-linear curve as shown in Figure 4-17 with a zero third branch stiffness
(K2=0) fitted to the calculated p-y curves for the clay-dominant layers (layers A, C and E) around
the abutment and pier piles and implemented in the archetype model D4 using a corresponding
zero-link element in SeismoStruct. Model parameters for these curves for the corresponding layers
are summarized in Table 4-12.
Table 4-12 Calculated parameters to define p-y curve for clay layers surrounding the abutment and pier piles
in the index archetype model D4
Layer d1 (m) d2 (m) K0 (kN/m) K1 (kN/m) d1 (m) d2 (m) K0 (kN/m) K1 (kN/m)
Soil Layer A 0.0014 0.042 32200 2370 - - - -
Soil Layer C 0.0003 0.048 708700 11400 0.0004 0.0528 708500 11400
Soil Layer E - - - - 0.0003 0.0403 439600 7016
In addition, a symmetric bilinear curve (see Figure 4-22) is employed to model idealised
elastic-plastic behaviour of soil layers B and D (refer to Figure 4-12 for soil layer labels)
surrounding the abutment and pier piles.
90
ce or K2 Force or
ment Moment
K1 r.K0
K0 Fy K0
K0 K0
-d1 d1 d2 Displacement Displacement
or Rotation K0 Fy or Rotation
r.K0
(a) Figure 4-22 A symmetric bi-linear curve with its three (b)
required parameters to fully characterize this response
curve in SeismoStruct software (SeismoSoft, 2020)
Model parameters defining response curve for the sand layers surrounding the abutment
and pier piles are presented in Table 4-13.
Table 4-13 Calculated parameters to define p-y curve for soil layer B and D surrounding the abutment and
pier piles in the index archetype model D4

Layer K0 (kN/m) Fy (kN) r (-) K0 (kN/m) Fy (kN) r (-)
Soil Layer B 287600 4314 0.0003 191000 2292 0
Soil Layer D 1056000 10560 0 756400 9077 0
4.2.8.4 Vertical Soil-Pile Interaction
4.2.8.4.1 Pile skin Resistance
The relationship between mobilized soil-pile shear transfer and local pile deflection (t-z curve) at
any depth for axially loaded piles and for non-carbonate soils is given by API (2007) as per Table
4-14.
91
Table 4-14 Pi es’ skin rsistance in c ay and sand ayers ( PI, 2007)
Clays Sands
Z/D t/tmax Z (in.) t/tmax
0.0016 0.30 0.000 0.00
0.0031 0.50 0.100 1.00
0.0057 0.75  1.00
0.0080 0.90
0.0100 1.00
0.0200 0.70 to 0.90
 0.70 to 0.90
where
Z is local pile sliding in mm (in),
D is pile diameter in mm (in) ,
t is mobilized soil pile adhesion in kPa (lb/ft2),
tmax is the maximum soil pile adhesion or unit skin friction capacity in kPa (lb/ft2) computed
as per Equation (4.26).
  c for clay
t max =  (4.26)
  p0 for sand
where,
β is the shaft friction factor, and is equal to 0.29 and 0.37 for medium dense and medium
dense-dense soils, respectively,
p0 is the effective overburden pressure at the pile tip depth,
c is undrained shear strength
 is a coefficient given by Equation (4.27),
 0.5 −0.5   1
 = (4.27)
0.5  1
−0.25
c
where,  = .
p0
92
A Tri-linear response curve shown in Figure 4-17 with a zero third branch stiffness (K2=0)
is used to simplify and implement the above calculated t-z curves for the soil layers A, C and E.
The required parameters to define this response curve summarized in Table 4-15.
Table 4-15 Calculated parameters to define t-z curve for the soil layers A, C and E surrounding the abutment
and pier piles in the index archetype model D4

Layer d1 (m) d2 (m) K0 (kN/m) K1 (kN/m) d1 (m) d2 (m) K0 (kN/m) K1 (kN/m)
Soil Layer A 0.001 0.003 23860 7014 - - - -
Soil Layer C 0.0004 0.002 206200 91650 0.0004 0.0026 206200 91640
Soil Layer E - - - - 0.0003 0.002 126900 56420
Model parameters defining this response curve for the sand-dominant layers surrounding
the abutment and pier piles are calculated and summarized in Table 4-16.
Table 4-16 Defined parameters to define t-z curve for soil layer B and D surrounding the abutment and pier
piles in the index archetype model D4

Layer K0 (kN/m) Fy (kN) r (-) K0 (kN/m) Fy (kN) r (-)
Soil Layer B 19410 49 0 21634 55 0
Soil Layer D 14339 36 0 3496 9 0
t-z curves for the abutment and pier piles are calculated using Table 4-14 and shown in Figure
4-23.
93
Layer A-Abutment Layer B-Abutment
Layer C-Abutment Layer C-Pier
Layer B-Pier Layer D-Abutment
Layer E-Pier
350 60 Layer D-pier
300
50
250
40
t(kN/m)
t (kN/m)
200
30
150
100 20
50 10
0 0
0 0.005 0.01 0 0.005 0.01
y (m) y (m)
Figure 4-23 The relationship between mobilized soil-pile shear transfer and local pile deflection for the
abutment and pier piles
4.2.8.4.2 Pile Tip Resistance
According to API (2007), a pile tip displacement up to 10 percent of the pile diameter is required
for full mobilization in both sand and clay soils. Pile tip resistance or end bearing capacity
recommended for both sands and clays can be determined using Table 4-17 (API, 2007).
Table 4-17 Tip-load-Displacement relationship for sand and clay soils (API, 2007)
Z/D 0.002 0.013 0.042 0.073 0.100

Q/Qp 0.25 0.50 0.75 0.90 1.00
where
Z is axial tip deflection, in mm,
D is pile diameter in mm,
Q is mobilized end bearing capacity in kN, and
Qp is total pile end bearing in kN is given by Equation (4.28).
 9cA p for cohessive soils (4.28)
Qp = 
 N q p0 A p for cohessionless soils
where,
c is undrained shear strength,
Ap is gross end area of pile in m2,
94
p0 is effective overburden pressure at the pile tip depth,
Nq is end bearing factor is 12 or 20 for medium dense and dense soils, respectively.
Pile tip load-displacement is calculated and plotted for abutment and pier using Table 4-17
and Equation (4.28).
250
200
150
Q(kN)
Abutment
Pier
100
50
0
0 0.01 0.02 0.03
z (m)
Figure 4-24 Calculated pile tip load-displacement for abutment and pier of the Meloland Road Overcrossing
The tri-linear response curve shown in Figure 4-17 is used to simplify and implement the
above calculated pile tip load-displacement relationship. The required parameters to define this
response curve are summarized in Table 4-18.
Table 4-18 Model parameters for Q-z curve for abutment and pier piles in the index archetype model D 4
Component d1 (m) d2 (m) K0 (kN/m) K1 (kN/m) K2 (kN/m)

Abutment and Pier Piles 0.0005 0.0034 112500 20450 5172
4.3 Eigen Value Analysis
An eigenvalue analysis using the Lanczos algorithm (SeismoStruct, 2020) was performed to
determine the modal periods of models D1, D2, D3, and D4. The obtained modal period and mode
shapes are compared with their corresponding modal properties obtained from Ambient Vibration
Test (AVT) by Ventura et al. (2011) for the first three modes. These modal characteristics are
summarized in Table 4-19. As it can be seen, the periods of the first three modes of vibration
95
calculated for all archetype models are in good agreement with corresponding measured periods
from the ambient test. The mode shapes calculated from the models for the first six modes are also
identical to the corresponding measured mode shapes using the ambient vibration tests.
Table 4-19 Modal period of the models D1, D2, D3, and D4 along with modal period obtained from Ambient
Vibration Test (AVT) for the MRO
Period Ti (s)
Mode D1, D2, and D3 D4 AVT Mode Description
1 0.280 0.298 0.297 Vertical anti-symmetric mode

2 0.264 0.280 0.275 Transverse mode
3 0.210 0.236 0.224 Vertical symmetric mode
4 0.10 0.128 0.148 First torsional mode
5 0.054 0.109 0.103 Second torsional mode
6 0.044 0.072 0.088 Second vertical anti-symmetric mode
As presented Table 4-19, all six modal periods of the D4 model are matched with a
reasonably good accuracy with their corresponding periods calculated using the ambient vibration
tests. Whereas the simplified models of MRO (D1, D2 and D3) demonstrate a lower accuracy,
especially, in the higher modes. The simplified models are based on the previous studies and were
not calibrated in this research to match the measured periods.
4.4 Performance Criteria
Using the guidelines provided by the Seismic Retrofitting manual for the highway structures part
1 (FHWA, 2006), strength and displacement capacity of bridge members for various limit states
are provided in Section 7.4D.1 and Section 7.4D.2 of Appendix D, respectively. As discussed in
Section 7.4D.1, the shear strength of the main structural members is calculated based on the
contribution of concrete and reinforcing steel bars. The flexural plastic curvature capacity is
calculated considering initial and final shear capacities of the cross-section. The initial shear
capacity (Vi) refers to the shear strength of uncracked cross section whereas the final shear capacity
(Vf) considers cracking in the member due to cyclic loading. Detailed equations describing this
semi-ductile behaviour are presented in Section 7.4D.1.
96
Plastic curvature corresponding to various limit states in reinforced concrete columns and
beams are based on Seismic Retrofitting Manual for Highway Structures and presented in Section
7.4D.2 of Appendix D. This includes calculating the plastic curvatures for the following
phenomena:
- Compression failure of unconfined concrete (applied to cover sections to track spalling
failure)
- Compression failure of confined concrete (corresponding to failure of core concrete in
main structural members).
- Low-cycle fatigue of longitudinal reinforcing steel bars
- Buckling of longitudinal bars.
- Fracture of longitudinal rebars under tension.
Section 7.4D.2 also includes calculation of plastic hinge length (Lp) for the main structural
members. For each member, the ratio of plastic hinge length to the length of the structural member
(Lp/L) is calculated and defined in the finite element model. Performance criteria of the bridge
members defined in the bridge archetype model are calculated based on the above guidelines and
presented in Section 4.4.1. In addition, shear and chord rotation capacity of the bridge members
calculated by SeismoStruct software based on ASCE41-17 presented in Section 4.4.2.
4.4.1 Bridge Components Limit State
To calculate curvature corresponding to semi-ductile shear failure of pier column of the MRO (see
Equation (D1.2)), the initial and final shear strength of the pier column are calculated based on
Equation (D1.4) and Equation (D1.12), respectively. The calculated shear strengths (V) and
curvature of the pier column (φ) are shown in Figure 4-25.
97
7,000
6,000
Vi=5465.5kN
5,000
4,000
V(kN)
Vf=3288.1kN
3,000
2,000
1,000
0
0 0.006 0.01 0.016 0.02 0.03
ф (rad/m)
Figure 4-25 Calculated initial and final shear strength of pier column of the MRO
In addition, plastic curvatures corresponding to the limit states illustrated in Appendix D.1 and
Appendix D.2 for the bridge components are calculated and summarized in Table 4-20, Table
4-21, and Table 4-22.
98
Table 4-20 Performance criteria for the Pier Column of the MRO
Criterion Type Value Equation
Spalling Cover Failure Section Curvature 0.0072 (D2.1)

(rad/m)
Confined Concrete Section Curvature 0.0307 (D2.9)

Compression Failure (rad/m)
Ultimate Compression Concrete Strain -0.0085 (D2.10)

Strain (m/m)
Low-cycle Fatigue of Section Curvature 0.0374 (D2.11)

Longitudinal (rad/m)
Reinforcing Steel Bars
Buckling of Longitudinal Section Curvature 0.0035 (D2.14)

Reinforcing Steel Bars (rad/m)
Fracture of the Section Curvature 0.148 (D2.16)

Longitudinal (rad/m)
Semi-ductile Shear Section Curvature 0.0134 (D1.2)

Failure (rad/m)
Shear Capacity Frame Element Shear Auto-defined ASCE 41-17, Eq. (10-3)
Capacity and ACI 318-11 Eq. (11-
5), Eq. 11-27 and Eq. (11-
28)
Chord Rotation Frame Element Auto-defined ASCE 41-17, Tables10-

Capacity Chord Rotation 7,10-8 and 10-19
Capacity
99
Table 4-21 Performance criteria for the Abutment Backwall of the MRO
Ultimate Compression Concrete Strain (m/m) -0.0064 (D2.10)

Strain
Fracture of the Reinforcement Strain (m/m) 0.06 -

Longitudinal
Crack Curvature Section Curvature (rad/m) 0.0013 M-φ
Yield Curvature Section Curvature (rad/m) 0.0052 M-φ
Semi-ductile Shear Section Curvature (rad/m) 0.039 (D1.2)

Failure
Chord Rotation Yielding Frame Element Chord Rotation Auto-defined ASCE4 1-17, Table10-5,
Yielding (rad/m) Eq. (10-5)
Table 4-22 Performance criteria for deck slab and pier diaphragm of the MRO
Pier Diaphragm Concrete Strain -0.006 -

Ultimate Compression (m/m)
Strain
Deck Ultimate Concrete Strain -0.006 -

Compression Strain (m/m)
Deck Diaphragms Steel Reinforcement 0.06 -

Rupture Strain (m/m)
Deck Steel Rupture Reinforcement 0.06 -

Strain (m/m)
Pier Diaphragm Shear Frame Element Auto-defined ASCE 41-17, Eq. 10-3
Capacity Shear Capacity and ACI 318-11 Eq.
11-5, Eq. 11-27 and
Eq. 11-28
100
As summarized in Table 4-20 and Table 4-21, semi-ductile shear failure of the pier column
and abutment backwalls are considered in all models based on their associated section curvatures,
which were calculated using Equation (D1.2).
Details of calculation of the performance criteria of the main structural components are
presented in Appendix D. It should be noted that strain and curvature performance checks are
carried out at the integration sections, which specified in Section 4.2.2, of the selected elements.
Structural failure modes in the index archetype models are identified and compared in Section 4.8.
4.4.2 Shear and chord rotation capacity of bridge members
Chord rotation yielding, chord rotation capacity and element shear capacity checks can be
conducted using SeismoStruct. The software calculates the capacity of the elements during the
analysis, according to the selected equations, and checks it against the corresponding demand.
It is noted that expected values are calculated considering mean material strengths and no safety
factors are incorporated.
In addition to the user-defined strength and displacement capacity criteria illustrated in
Section D.1 and Section D.2, auto-defined shear strength and curvature capacities implemented in
SeismoStruct based on selected equations from ASCE 41-17 and ACI 318-11 provisions are used
to identify the instants at which different performance limit states such as structural damage and
collapse are reached. In this case, the expected material values without safety or confidence factors
are used for chord rotation yielding, chord rotation capacity and shear capacity (SeismoSoft, 2020).
4.5 Pushover Analysis
To determine ultimate capacity of the abutment backwall and pier column, a pushover analysis is
carried out using SeismoStruct software. The program requires plastic hinge length to be defined
for inelastic force-based plastic hinge (infrmFBPH) frame element type, as it considers inelasticity
within a fixed length of the element. The plastic hinge length is calculated as 0.9m based on the
semi-empirical formula recommended in the Seismic Retrofitting Manual for Highway Structures-
Part1 and presented in Equation (D2.18) considering 5382.4kN pier column axial load and 312.3
MPa steel yield stress as suggested by Werner et al. (1993).
101
To cross-check, the plastic hinge length is also calculated using the European seismic
design code for bridges as shown in Equation (4.29) (EC8-2, 2005). Using this equation, the plastic
hinge length is calculated as 0.97m for the pier column.
Lp = 0.1L + 0.015f yds (4.29)
where,
Lp is the length of plastic hinge in m,
L is the length of the pier column in m,
fy is the yield stress of the pier column rebar in MPa,
ds is the cover of the column rebars in m.
The Ratios of the pier column plastic length to its height (L/Lp) are calculated as 14.4% and 15.5%
using the Retrofitting Manual of Highway Bridges-part1and EC8, respectively. As a result, an
average ratio of the plastic hinge length to its height of 15% is considered and used in the
SeismoStruct model for the pier column.
Table 4-23. Calculated plastic hinge length and hinge to total length ratio for the pier column.
Calculation method Lp L/Lp

FHWA (2006) 0.9 (m) 14.4 (%)
EC8-2 (2005) 0.97 (m) 15.5 (%)
Average 0.94 15 (%)
For abutment backwall, plastic hinge length (Lp) is calculated as per Equation (4.30) based
on a minimum height of plastic hinge regions for the walls specified in Clause 21.5.2.1.2 of the
design of concrete structures of the Canadian Standards Association (CSA) code, (A23.3‐14,
2014).
Lp = 0.5w + 0.1h w (4.30)
where, w is the width and hw is the height of wall. Using Equation (4.30), the ratio of the plastic
hinge length to the wall length (Lp/hw) for abutment wall is calculated as 16.4%.
A nonlinear static pushover analysis is carried out in SeismoStruct software for the pier
column and abutment backwall in the longitudinal direction considering the above calculated ratio
102
of the plastic hinge length to the height of the components in the model. These pushover curves
are shown in Figure 4-26.
2,500 700
2,000 600
500
Force (kN)
Force (kN)
1,500
400
1,000 300
200
500
100
0 0
0.00 0.03 0.05 0.08 0.10 0.13 0.15 0.00 0.05 0.10 0.15
Displacement (m) Displacement (m)
Figure 4-26 Force-displacement curve for the MRO simulated in SeismoStruct commercial software (a) Pier
column with axial load=5382.4kN (b) Abutment backwall with axial load=2691.2kN
4.6 Hazard Analysis
The MRO is located in Southern California (φ=32.774°N and λ=115.448°W). The bridge is very
close to the Imperial fault (about 200m from the surface projection of the fault).
The probability of earthquake shaking with a certain magnitude in a region can be
determined knowing the average rate of historical events. Figure 4-27 shows significant historic
Earthquakes with magnitude greater than 5.5 during the 19th and 20th centuries.
103
Figure 4-27 Significant historic Earthquakes greater than magnitude 5.5 between 1800 and 2000 in California
(Toppozada and Branum, 2008)
Earthquakes in California are caused by the movement of the Pacific and North American
plates. The Pacific plate is moving northwest pushing the North American plate along the San
Andreas fault with an average annual rate of movement of 30mm to 50mm over the last 10 million
years (The Geological Society, 2018). Many of the major and minor earthquakes of the region are
produced by the San Andreas fault and some other parallel faults such as the Elsinore, San Jacinto,
and Imperial faults. As shown in Figure 4-27, many earthquakes of M5 and above have occurred
in this region within the past two centuries. The NGA-West 2 ground motion database provided
by the Pacific Earthquake Engineering Research Center (PEER) is employed here to select the
ground motions. This database includes a comprehensive set of globally recorded shallow crustal
earthquakes in active tectonic regimes. Each earthquake in the database contains a set of metadata
104
including various site characterizations, different distance measures, and earthquake source data
(PEER, 2019). In this research, a set of crustal ground motions are chosen as the MRO bridge is
located in a region which is mainly affected by shallow crustal earthquakes. In addition to hypo-
central depth, soil conditions and ground motions’ dynamics characteristics such as predominant
period and peak ground acceleration are considered in selecting of the input ground motions.
To perform Incremental Dynamic Analysis (IDA), a set of 22 ground motions is selected
using the PEER NGA-West 2 ground motion database considering different magnitude, peak
ground acceleration, predominant period and mechanism to account for hazard uncertainty. The
case study bridge, MRO, was built in an active crustal region close to the Imperial Valley Fault
zone (USGS, 2019). The region has been affected by strong and major crustal earthquakes both
historically and recently. As a result, the earthquake magnitude scale (M) of the selected
earthquake records ranges between M6.2 and M7.5. In addition, ground motion records are
selected with different Predominant Periods (Tp), the period at which the maximum spectral
acceleration occurs in an acceleration response spectrum calculated at 5% damping, to account for
hazard uncertainty. Similarly, the records are chosen with different Peak Ground Acceleration
(PGA). The maximum and minimum considered Tp of the selected motions are 0.64s and 0.08s for
the motion RSN#723 and RSN# 879, respectively. The maximum unscaled PGA of the chosen
motions are 1.22g for the motion RSN#77 and the minimum unscaled PGA is 0.2g belonging to
earthquake record RSN# 730 (see Table 4-24). The spectra of the selected ground motions are
shown in Figure 4-28. Information of these ground motions is summarized in Table 4-24 and time
stories of them are presented in Appendix B.2.
3.5
3
2.5
2
Sa (g)
1.5
1
0.5
0
0 1 2 3 4
T(s)
Figure 4-28 Spectra (5% damping) of the selected ground motions for IDA
105
Table 4-24 Summary of the selected ground motions using NGA-West 2 ground motion database (PEER, 2019)
Event PEER Earthquake Name Year Station Name Magnitude Mechanism Tp (s) / PGA(g)
No. RSN
No
1 184 Imperial Valley-06 1979 El Centro Differential 6.53 Strike Slip 0.4 / 0.48
Array
2 723 Superstition Hills- 1987 Parachute Test Site 6.54 Strike Slip 0.64 / 0.43
02
3 1119 Kobe_ Japan 1995 Takarazuka 6.9 Strike Slip 0.46 / 0.7
4 1787 Hector Mine 1999 Hector 7.13 Strike Slip 0.5 / 0.33
5 1602 Duzce_ Turkey 1999 Bolu 7.14 Strike Slip 0.36 / 0.81
6 1158 Kocaeli_ Turkey 1999 Duzce 7.51 Strike Slip 0.28 / 0.36
7 879 Landers 1992 Lucerne 7.28 Strike Slip 0.08 / 0.79
8 4040 Bam_ Iran 2003 Bam 6.6 Strike Slip 0.2 / 0.81
9 5836 El Mayor-Cucapah_ 2010 El Centro - Meloland 7.2 Strike Slip 0.14 / 0.23
Mexico Geot. Array
10 6 Imperial Valley-02 1940 El Centro Array #9 6.95 Strike Slip 0.46 / 0.28
11 30 Parkfield 1966 Cholame - Shandon 6.19 Strike Slip 0.36 / 0.44
Array #5
12 288 Irpinia_ Italy-01 1980 Brienza 6.9 Normal 0.14 / 0.22
13 77 San Fernando 1971 Pacoima Dam (upper 6.61 Reverse 0.38 / 1.22
left abut)
14 126 Gazli_ USSR 1976 Karakyr 6.8 Reverse 0.14 / 0.39
15 953 Northridge-01 1994 Beverly Hills - 14145 6.69 Reverse 0.54 / 0.49
Mulhol
16 3746 Cape Mendocino 1992 Centerville Beach_ 7.01 Reverse 0.24 / 0.48
Naval Fac
17 139 Tabas_ Iran 1978 Dayhook 7.35 Reverse 0.38 / 0.41
18 495 Nahanni_ Canada 1985 Site 1 6.76 Reverse 0.1 / 1.2
19 741 Loma Prieta 1989 Bran 6.93 Reverse 0.14 / 0.5
Oblique
20 730 Spitak_ Armenia 1988 Gukasian 6.77 Reverse 0.34 / 0.2
Oblique
21 1197 Chi-Chi_ Taiwan 1999 CHY028 7.62 Reverse 0.28 / 0.76
Oblique
22 8067 Christchurch_ New 2011 Christchurch Cashmere 6.2 Reverse 0.22 / 0.4
Zealand High School Oblique
The selected ground motions are applied to the models without any modification or scaling
at the period of vibration of the bridges in the analysis as IDA requires performing multiple
nonlinear dynamic analyses at different scale factors.
To optimize the analysis runtime, the significant duration of the selected unscaled ground
motions using the Arias Intensity (IA) between 5% and 95% was used to define the duration of
106
shaking when performing incremental dynamic analyses. The total duration and defined duration
of shaking for each ground motion are summarized in Table 4-25.
Table 4-25 Total and significant duration values of the input time histories used in IDA analyses of the MRO
discrete models
Event Number Total Duration Significant Duration

(s) (s)
1 39.1 6.6
2 22.3 10.6
3 40.9 4.4
4 45.3 9.7
5 55.9 9.0
6 27.2 10.9
7 48.1 13.8
8 66.5 8.0
9 87.5 17.8
10 53.7 24.3
11 44.0 15.1
12 35.0 10.3
13 41.7 7.0
14 13.1 7.0
15 30.0 8.2
16 28.6 10.4
17 21.0 11.0
18 10.3 7.3
19 25.0 9.8
20 20.0 10.5
21 90.0 5.9
22 40.0 4.9
To investigate the effect of using significant duration of motions a nonlinear response

analysis is carried out on archetype index model D4 using the entire and the significant duration of
107
the ground motions. These ground motions were applied in the longitudinal and transversal
directions (parallel and perpendicular to the deck alignment). No significant difference between
responses predicted using the significant versus entire ground motion was observed. As an
example, Figure 4-29 and Figure 4-30 show the calculated drift and hysteresis curves for Figure 4
28 and Figure 4 29 the Imperial Valley-06 ground motion.
0.6
Pier Transversal drift (%)
0.4 Entire Duration (39.1)

Significant Duration (6.6s)
0.2 0.6
0.5
0.0 0.4
0.3
Acceleration (g)
-0.2 0.2
0.1
-0.4 0
-0.1
0.4
Pier Longitudinal drift (%)
-0.2
-0.3
0.2
-0.4
-0.5
0.0 0 10 20 30 40 50
Time (s)
-0.2
-0.4
5.25 6.25 7.25 8.25 9.25 10.25 11.25
Time (s)
Figure 4-29 Comparison of pier column drifts of the archetype model D4 in the longitudinal and transverse
directions, simulated using entire ground motion duration and significant duration.
8,000 8,000
Significant Duration (6.6s) Significant Duration (6.6s)
6,000 Entire Duration (39.1s) 6,000 Entire Duration (39.1s)
4,000 4,000
Base Shear (kN)
Base Shear (kN)
2,000 2,000
0 0
-2,000 -2,000
-4,000 -4,000
-6,000 -6,000
-8,000 -8,000
-0.03 -0.02 -0.01 0.00 0.01 0.02 0.03 -0.03 -0.01 0.01 0.03
Relative Displacement (m) Relative Displacement (m)
Figure 4-30 Hysteresis curves of the pier column calculated performing response time history analysis on the
index archetype model D4 using Imperial Valley-06 ground motion (a) transverse direction (b) longitudinal
direction
108
Top drift and base shear value in the transverse direction using the entire and the significant
durations are compared in Table 4-26.
Table 4-26 Comparison of longitudinal top pier drift and its corresponding base shear of the index archetype
model D4 using the entire and the significant duration of Imperial Valley-06 ground motion
Maximum Drift (%) Maximum Base Shear (kN)

Pier Column /Duration Entire Significant Entire Significant
(39.1s) (6.6s) (39.1s) (6.6s)
Transverse Direction 0.37 0.36 5728.2 5688.1
Longitudinal Direction 0.36 0.36 6674.5 6683.4
4.7 IDA results
Incremental Dynamic Analysis (IDA) is an analysis method developed by Vamvatsikos and

Cornell (2002) to study seismic performance predictions of structures subjected to different
seismic excitation levels. Nonlinear dynamic analysis is performed using an earthquake record
multiplied by a scale factor. The scale factor is gradually increased until structural collapse occurs.
This procedure is repeated for all the ground motions.
IDA is a useful analytical tool that can bring a valuable insight into the response of structure
subjected to a variety of ground motion intensities all the way to collapse levels. It also provides
information on change in structural response, including sequence of damage/failure events, when
subjected to different levels of ground motion intensity. Given IDA is typically performed using a
number of different ground motions, it provides a means to study the effect of ground motion on
the structural response and its collapse and variability due to inherent characteristics of the
structure. If IDA is performed on a rich set of ground motions, it can also be used to perform
statistical analysis and calculate fragility curves. In this research, the IDA method is used to predict
collapse conditions and to calculate the collapse fragility curves for each archetype model. The
IDA analyses are performed applying a significant duration of the selected ground motions
corresponding to the 5%-95% of the Arias intensity in the direction parallel and across to the deck
alignment (x and y axis) with a 0.02s timestep size for each archetype model. IDA scale factors
started at a value of 0.2 with a 0.2 step increase until collapse of the simulated structure. A
summary of the analyses’ log listing the collapse scale factors, the scale factor that associated with
109
the failure of the main structural components, along with their mean and standard deviation for
each archetype model subjected to the 22 ground motions is presented in Table 4-27.
Table 4-27 Summary of IDA log for each index archetype model
Event No Event Collapse Scale Factor (CSF)

D1 D2 D3 D4
1 "Imperial Valley-06" 1.4 1.8 1.6 1.6
2 "El Mayor-Cucapah_ Mexico_5836" 3.6 3.6 4.2 3.6
3 "Hector Mine" 2 1.6 1.8 1.6
4 "Landers" 2 2 2.6 1.4
5 "Parkfield" 2 1.8 2 1.6
6 "Superstition Hills-02" 1.8 1.6 1.6 1.4
7 "Kobe-Japan" 0.8 1.4 1 1.2
8 "Duzce-Turkey" 1.4 1.4 1.2 0.8
9 "Kocaeli-Turkey" 1.8 1.8 2 1.6
10 "Bam-Iran" 1.2 1.8 1.4 1.2
11 "Irpinia-Italy-01" 4.2 4 4 3.4
12 "San Fernando" 1 1 1 1
13 "Gazli- USSR" 1.4 1.6 1.8 1.4
14 "Northridge-01" 1.4 1.2 1.2 1.4
15 "Cape Mendocino" 2.4 2 2 1.6
16 "Tabas-Iran" 2 1.8 1.8 2
17 "Nahanni-Canada" 1.8 1.6 1.4 1.2
18 "Loma Prieta" 1.4 1.6 1.6 1.4
19 "Spitak-Armenia" 4.4 3.8 3 3.2
20 "Chi-Chi-Taiwan" 1 0.8 1 1
21 "Christchurch-New Zealand" 2.2 2 1.8 2.2
22 "El Centro" 2.8 2.4 2.6 2.2
Mean of CSF: 2.0 1.9 1.9 1.7
Standard Deviation of CSF: 1.0 0.8 0.9 0.8
110
This table provides a comparison between collapse scale factors predicted for different
archetype models. Choosing model D4 as the benchmark, the ratio of the collapse-level scale factor
for D1, D2, and D3 relative to the D4 model are plotted in Figure 4-31. As shown in this figure,
these three models, on average show a higher collapse scale factor compared to the model D4.
These ratios are 1.06, 1.13, and 1.15 for D1/D4, D2/D4 and D3/D4, respectively.
D1/D4
D2/D4 D3/D4
1.13 1.15
1.06
Figure 4-31 Comparison of Collapse Scale Factors (CSFs) ratio relative to model D 4
The results of IDA analysis are typically presented in terms of an Intensity Measure (IM)
versus a Damage Measure (DM) (Vamvatsikos and Cornell, 2002). In this research, spectral
acceleration (Sa) at system’s fundamental period (T1) is employed as the intensity measure (IM)
and abutment and pier column drifts are considered as the Engineering Demand Measures (EDM).
Results of the IDA for the MRO are presented in Figure 4-32 to Figure 4-34.
111
3 3
2.5 2.5
Sa (T1=0.28s) [g]
Sa (T1=0.28s) [g]
2 2
1.5 1.5
1 1
0.5 0.5
0 0
0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6
Drift (%) Drift (%)
3
(a) 3 (b)
2.5 2.5
Sa (T1=0.28s) [g]
Sa (T1=0.3s) [g]
2 2
1.5 1.5
1 1
0.5 0.5
0 0
0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6
Drift (%) Drift (%)
(c) (d)
Figure 4-32 IDA curves for pier column of the MRO in x-direction (across the bridge deck) for the index
archetype models: (a) D1, (b) D2, (c) D3, and (d) D4.
112
3
3
2.5 2.5
Sa (T1=0.28s) [g]
Sa (T1=0.28s) [g]
2 2
1.5 1.5
1 1
0.5 0.5
0 0
0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6
Drift (%) Drift (%)
(a) 3 (b)
3
2.5 2.5
Sa (T1=0.3s) [g]
Sa (T1=0.28s) [g]
2 2
1.5 1.5
1 1
0.5
0.5
0
0
0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6
0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6
Drift (%)
Drift (%)
(c) (d)
Figure 4-33 IDA curves for pier column of the MRO in y-direction (along the bridge deck) for the index
archetype models: (a) D1, (b) D2, (c) D3, and (d) D4.
113
3 3
2.5 2.5
Sa (T1=0.28s) [g]
Sa (T1=0.28s) [g]
2 2
1.5 1.5
1 1
0.5 0.5
0 0
0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6
Drift (%) Drift (%)
(a) 3 (b)
3
2.5
2.5
Sa (T1=0.3s) [g]
Sa (T1=0.28s) [g]
2
2
1.5 1.5
1 1
0.5 0.5
0 0
0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6
Drift (%) Drift (%)
(c) (d)
Figure 4-34 IDA curves for the MRO abutment backwall of the MRO in y-direction (across the bridge deck)
for the index archetype models: (a) D1, (b) D2, (c) D3, and (d) D4.
In these figures, each curve corresponds to a particular ground motion and each point on
the curves represents a given scale factor applied to that ground motion. The points demonstrate
the seismic demand (drift in this case), corresponding to a specific level of intensity (represented
by Sa here). It can be seen that drift at collapse level in the archetype model D4 is associated with
the smaller spectral value compared to the corresponding values in the other models. This is
consistent with the observation of the collapsed scale factor ratios presented on Figure 4-31.
The weaving effect observed in some of the IDA cases, can be explained as the change in
the monotonic pattern of increasing the DM (drift) as IM (Sa) increases where under some
conditions, drift decreases with a higher level of excitation intensity. This effect was also reported
and discussed by Vamvatsikos and Cornell (2002). The weaving effect usually takes place in the
nonlinear regime past a stage of softening behaviour. The apparent hardening behaviour seen in
114
weaving effect is dependent on the pattern and timing of the events in the ground motion rather
than necessarily the excitation intensity (Vamvatsikos and Cornell, 2002). This effect can be due
to a change in the failure mode, due to a sudden change in stiffness characteristics of the structure.
This, in turn, can lead to subsequent changes of drift amplitudes. An earlier yield due to a high
intensity excitation can lead to changes in the failure sequence, and thus reduction in drifts.
The hysteresis curves of the pier column for model D1 and D4 in the transverse and
longitudinal direction at their corresponding collapse levels due to the Imperial Valley-06, and El
Mayor ground motions are compared and presented in Figure 4-35.
8000 8000
D1 Model D1 Model
D4 Model D4 Model
Base Shear (kN)
Base Shear (kN)

4000 4000
0 0
-4000 -4000
-8000 -8000
-0.04 -0.02 0.00 0.02 0.04 -0.04 -0.02 0.00 0.02 0.04
(a) (b)
8000 8000
D1 Model D1 Model
D4 Model D4 Model
Base Shear (kN)
Base Shear (kN)
4000 4000
0 0
-4000 -4000
-8000 -8000
-0.04 -0.02 0.00 0.02 0.04 -0.04 -0.02 0.00 0.02 0.04
(c) (d)
Figure 4-35 Calculated hysteresis curves of the pier column using D1 and D4 model at their corresponding
collapse levels using (a) Imperial Valley-06 ground motion in the transverse direction (b) Imperial Valley-06
ground motion in the longitudinal direction, (c) El Mayor ground motion in the longitudinal direction (d) El
Mayor ground motion in the transverse direction.
115
As shown in Figure 4-35, the hysteresis loops for the D1 and D4 models due to El Mayor
ground motion show similar response envelopes. For this ground motion, both models collapsed
at the same scale factor (SF=3.6). In the case of Imperial Valley ground motion, the two hysteresis
curves show significantly different response envelopes. As it can be seen from Figure 4-35 (a) and
(b), D4 model dissipated more energy at the collapse level. It should be noted that in this case, the
scale factors are different for the D1 and D4 models being, 1.4 and 1.6, respectively.
4.8 Identification and comparison of Structural Failure Modes in the archetype models
The sequence of failure modes predicted in the IDA analyses depends on many factors, including
ground motion characteristics and model details. The sequence of failure modes at the collapse
level for all the archetype models subjected to all ground motions were extracted from the analyses
results. Figure 4-36 and Figure 4-37 collectively show the sequence of failure for models D1 to D4
for eight of the ground motions. The rest of the results are presented in Appendix H . As shown
in these figures, an overall similarity in the sequence of failure modes is observed, generally
starting with failure of the pier and propagating into the abutment. Structural integrity cannot be
maintained when one or more main structural components belonging to the shear force resisting
system of a bridge structure fail. Thus, in this study, collapse of the models is defined as sequence
of failure of the main structural members, which lead to the model instability. In many of these
cases, collapse has taken place as a result of abutment steel rupture leading to model instability.
116
Failure Mode D1 Model D2 Model
Abutment steel rupture X Imperial Valley-
X X X X X X
Abutment curvature capacity 06
Abutment ultimate compression strain Superstition
Hills-02
Pier shear capacity
Landers
Pier core compression failure
Pier plastic curvature
El Mayor
Pier ultimate compression failure
Pier longitudinal bars buckling X Collapse
Pier spalling cover failure
th.
1st. 2nd. 3rd. 4th. 5th. 6th. 1st. 2nd. 3rd. 4th. 5th. 6 7th.
Sequence of Failure Event Sequence of Failure Event
(a) (b)
Abutment steel rupture X X X X X
Abutment curvature capacity X X
Abutment ultimate compression strain X
Pier shear capacity
Pier longitudinal bars buckling
1st. 2nd. 3rd. 4th. 5th. 1st. 2nd. 3rd. 4th. 5th. 6th.
(c) (d)
Figure 4-36 Failure mode and their sequence of occurrence for the models when subjected to the Imperial
Valley-06, Superstition Hills-02, Landers, and El Mayor earthquake ground motions at the collapse levels
corresponding to each model
117
Abutment steel rupture X X X X X San Fernando
Abutment curvature capacity X
Abutment ultimate compression strain Northridge-01
Pier shear capacity X
Pier core compression failure Loma Prieta
Parkfield
1st. 2nd. 3rd. 4th. 5th. 6th. 7th. st. 2nd. 3rd. 4th. 5th.
1 6th. 7th. 8th. 9th.
(a) (b)
Abutment steel rupture X X X X
Abutment curvature capacity
Abutment ultimate compression strain
Pier ultimate compression failure X
1st. 2nd. 3rd. 4 5
th. th.
6th. 7th. 1st. 2nd. 3rd. 4th. 5th. 6th.
(c) (d)
Figure 4-37 Failure mode and their sequence of occurrence for the models when subjected to the San Fernando,
Northridge-01, Loma Prieta, and Parkfield earthquake ground motions at the collapse levels corresponding to
each model
To investigate the SSI impact on the failure mode of the models, the relationship between
failure modes of each model and the sequence of their occurrence are plotted and presented in
Figure 4-38 and Figure 4-39. The ground motions considered are the Imperial Valley-06, El
Mayor, Superstition Hills, Landers, San Fernando, Northridge, Loma Prieta, and the Parkfield
earthquake ground motion. In some cases, for example the Superstition Hills earthquake shown in
Figure 4-38 (c), a very similar sequence of failure is observed in all the models. In this case, in all
models failure, starts with pier longitudinal bars buckling, followed by other pier failure modes
and then finally collapsing due to abutment curvature capacity failure and steel rupture. In other
cases, the models D1 to D4 show a very different response. For instance, in the case of the Imperial
Valley-V6 event shown in Figure 4-38 (a), sequence of failure in D1 to D4 models are quite
118
different. Unlike many other ground motions analyzed here, in the Imperial Valley-V6 event, much
fewer failures occur before collapse.
Failure Mode:
Pier shear capacity
D1 D1
Pier plastic curvature D2 D2
Pier ultimate compression failure D3 D3
Pier longitudinal bars buckling D4 D4
th. th.
1st. 2nd. 3rd. 4th. 5 6 7
th.
1st. 2nd. 3rd. 4th. 5th.
(a) (b)
Failure Mode:
Pier shear capacity
D1 D1
1st. 2nd. 3rd. 4th. 5th. 6
th. 1st. 2nd. 3rd. 4th. 5th. 6th.
(c) (d)
Figure 4-38 Failure mode sequence of the models D1, D2, D3, and D4 due to the collapse level ground motions
(a) Imperial Valley-06 (SF= 1.4, 1.8, 1.6 and 1.6), (b) El Mayor-5630 (SF= 3.6, 3.6, 4.2 and 3.6), (c) Superstition
Hills (SF= 1.8, 1.6, 1.6 and 1.4) and (d) Landers (SF= 2, 2, 2.6 and 1.4)
119
Failure Mode:
Pier shear capacity
Pier core compression failure D1 D1
Pier ultimate compression failure D3 X D3
1st. 2nd. 3rd. 4th. 5
th. 6th. th. th.
1st. 2nd. 3rd. 4th. 5 6 7th.
(a) (b)
Failure Mode:
Pier shear capacity X X
D1 D1
D2 D2
Pier spalling cover failure th.
th. th. th. th.
1st. 2nd. 3rd. 4th. 5 6 7 8 9th. 1st. 2nd. 3rd. 4th. 5 6th.
(c) (d)
Figure 4-39 Failure mode sequence of the models D1, D2, D3, and D4 due to the collapse level ground motions
(a) San Fernando (SF=1, 1, 1, and 1), (b) Northridge (SF= 1.4, 1.2, 1.2 and 1.4), (c) Loma Prieta (SF= 1.4, 1.6,
1.6 and 1.4) and (d) Parkfield (SF= 2, 1.8, 2 and 1.6)
The following list summarizes the observations based on the failure sequences presented
in Figure 4-36 to Figure 4-39:
– Model details and ground motion characteristics contribute to the sequence of failures
and final collapse mode in the models analyzed in this study.
– In some ground motion cases, the ground motion characteristics such as predominant
period and peak ground acceleration appear to be the dominant factor in response of
the archetype models. As a result, similar failure sequence and collapse modes were
observed in all models (e.g. Superstition Hills shown in Figure 4-38 (c).
– In some other cases, details of the model, including the SSI representation, appear to
dominate the response and failure modes. As an example, this effect was clearly
observed in the case of the Imperial Valley-06 shown in Figure 4-38 (a).
120
– In some earthquake events, models experience multiple modes of failure before
collapse. As a result, the structure has been able to utilize its ductility and energy
absorption (for instance see the Loma Prieta event shown in Figure 4-39(c)).
– Based on the variation of the sequence of the failure shown in the presented graphs,
which including the effect of the models and earthquakes dynamic characteristics such
as ground motion’s predominant period and peak ground acceleration, a statistical
approach needs to be employed to analyze the data. Special attention needs to be paid
in:
- Defining the details of structural model, including SSI representation
- Performing the analysis on a wide number of ground motions which will allow
for meaningful statistical analysis of results.
More figures showing the dependency of failure modes and sequence of their occurrence
of the understudied soil+bridge system to the numerical models and the selected ground motions
are presented in Appendix H.
4.9 Fragility Curves
Fragility curves offer a statistical approach to collapse assessment of structural systems and
provide an objective tool for comparison. Fragility curves are powerful statistical tools to study
the probability of reaching or exceeding a given failure state. In this study, fragility curves
represent probability of collapse calculated as a function of spectral acceleration (Sa). A common
approach employed by many researchers is to fit a log-normal function to the Cumulative
Distribution Function (CDF) that describes the probability of collapse calculated from IDA
analysis of a given structure (Baker, 2015).
The log-normal distribution function (Equation (4.31)) provides a continuous estimate of
probability of collapse as a function of Sa.
P ( C | Sa = x ) =  
( )
 ln x 
  (4.31)
  
 
where,
P (C | Sa=x) is the probability that a ground motion with Sa=x will cause the structure
to collapse,
121
Φ () is the standard normal cumulative function (CDF),
θ is the median of the fragility function (the Sa level with 50% probability of collapse) and
β is the standard deviation of ln Sa or dispersion of Sa
In this study, fragility curves are calculated based on the results of the Incremental dynamic
analysis (IDA). As Ibarra and Krawinkler (Ibarra and Krawinkler, 2005) suggested, fragility
function parameters can be estimated from this data by taking logarithms of each ground motion’s
IM value associated with onset of collapse, and computing their mean and standard deviation as
per Equation (4.32) and Equation (4.33) (Baker, 2015).
1 n
ln ˆ =  ln IM i (4.34)
n i =1
1 n
( ( ))
2
ˆ =  ln IMi ˆ
n − 1 i =1
(4.35)
where n is the number of ground motions considered, and IMi is the IM value (Sa in this case)
associated with the onset of collapse for the ith. ground motion. This is a method of moments
estimator, as ln ˆ and ˆ are the mean and standard deviation of the normal distribution
representing the ln IM values, respectively. As the mean of (ln IM) is equal to the median of IM
in the case that IM is lognormally distributed, sample mean in this manner produces an estimate
of θ. The mean and standard deviation, or moments, of the distribution are estimated using the
sample moments from a set of data as per Equations (4.34) and (4.35). Values of ˆ and ˆ are
calculated for each model and shown in Table 4-28.
Table 4-28 Calculated IDA parameters for the models
Index Archetype Model ˆ ln ˆ ˆ
D1 1.713 0.379 0.275
D2 1.698 0.538 0.243
D3 1.673 0.529 0.295
D4 1.461 0.515 0.226
122
Using this fragility fitting approach and a MATLAB code developed by Baker (2015),
fragility curves for all archetype models are calculated and shown in Figure 4-40.
1
0.9
Probability of Collapse, P(C|Sa i)observed
0.8
0.7 D1 Model
0.6 D2 Model
D3 Model
0.5
D4 Model
0.4
0.3
0.2
0.1
0
0 0.5 1 1.5 2 2.5 3 3.5 4
Spectral Acceleration at the Fundamental Period, Sa (T 1) [g]
Figure 4-40 Fragility curves for the index archetype models D1 to D4 for the MRO Bridge.
4.10 Effect of Archetype Model on Fragility Curves
Figure 4-40 shows that the models D1, D2 and D3 resulted in very similar fragility curves. But
model D4, which includes a more comprehensive SSI representation including piles (see Section
4.2.8), shows a significantly different global collapse fragility curve.
D1 model, which was originally developed by Zhang and Makris (2001), is the most
feature-rich model which includes representation of the soil effect on embankment and foundations
using springs and dashpots comparing to the D2 and D3 model as discussed earlier in Section 4.2.
D1 and D4 models are compared below to demonstrate the effect of archetype model on the fragility
curves and probability of collapse (see Figure 4-2(a) and Figure 4-10).
Figure 4-41 shows the comparison of the fragility curves calculated for models D1 and D4.
The figure shows a significant difference between these two collapse prediction curves. Model D4
shows a higher probability of collapse for a given spectral acceleration. For instance, the spectral
123
acceleration corresponding to a 50% probability of collapse for model D1 is 1.72g and for model
D4 is 1.47g, which corresponds to a 14.5% difference in the levels of spectral accelerations.
In this case, model D4 has an increase in probability of collapse and, consequently, a poor
performance when soil supporting layers are included in the analysis. This example demonstrates
the sensitivity of model response on the of choice of SSI features in the model. This is not only in
terms of component failure, but also on the global collapse predictions. It also highlights the
uncertainty around the model features and the essential need to include various models when
studying collapse assessment of bridge structures considering SSI effects.
The sequence of failure modes leading to collapse and spectral acceleration at the
fundamental period of the D1 and D4 models at collapse level corresponding to 11 earthquakes are
summarized in Table 4-29.
Table 4-29 Sequence of failure modes of the index archetype models D 1 and D4 at onset of collapse
Event No. Event D1 Model D4 Model

Sa(T1) SFMs Sa(T1) SFMs
1 Imperial Valley-06 1.23 b→d→f→g→i 1.57 d→e→g→i
2 Superstition Hills-02 1.85 b→d→c→f→g→i 1.16 b→d→c→e→h→g→i
3 Hector Mine 1.71 b→f→g→i 1.38 d→b→g→i
4 Landers 1.90 b→d→f→g→i 1.28 f→b→g→i→h
5 El Mayor -5836 1.96 b→a→f→g→i 2.01 f→b→d→g→i
6 San Fernando 2.11 b →g→i 1.88 b→d→f→g→i
7 Northridge-01 1.62 b→d→c→f→g→i 1.76 f→b→c→d→g→i
8 Cape Mendocino 2.19 d→a→f→g→i 1.31 d→c→e→g→i
9 Loma Prieta 1.88 b→d→c→g→i→h→f 1.59 f→a→d→b→g→i
10 El Centro 2.01 b→d→f 1.43 a→g→i
11 Parkfield 1.47 b→f→c→d→i 1.54 a→b→c→d→i
Key:
a: Pier spalling cover failure f: Pier shear capacity
SFMs: Sequence of Failure Modes
b: Pier longitudinal bars buckling g: Abutment ultimate
CSF: Collapse Scale Factor
c: Pier ultimate compression strain compression strain
T1: Fundamental period in second
d: Pier plastic curvature h: Abutment curvature capacity
Sa(T1): Spectral acceleration at CSF in g
e: Pier core compression failure i: Abutment steel rupture
To investigate how components affect the calculated collapse fragility curve of the models
as well as global collapse predictions, these 11 earthquake events are incorporate in the calculated
fragility curves for D1 and D4 models shown in Figure 4-41. The sequence of failure modes leading
to collapse of these motions are summarized in Table 4-29.
124
1
Probability of Collapse, P(C|Sai) observed (5)
(6) (8) D1 Model
0.8 (6) D4 Model
(7)
(9) (10) (#): Event
(1) (5) Number
0.6 (11) (4) & (9)
(10) (2)
(3) (3)
0.4 (2) (7)
(8)
(11)
(4)
0.2 (1)
0
0 0.5 1 1.5 2 2.5 3 3.5 4
Spectral Acceleration at the Fundamental Period, Sa (T1) [g]
Figure 4-41 Fragility curves for the index archetype models D1 and D4 for the MRO Bridge.
Based on Table 4-29 and Figure 4-41, the following remarks can be concluded:
a) Models experience multiple modes of failure before collapse, which are not identical in
terms of sequence and numbers of failure modes prior to collapse and dissipation of
energy.
b) Models collapse in a different measuring intensities and probabilities of collapse associated
with these intensities for a specific ground motion. For example, collapse spectral
acceleration and its corresponding probability of collapse for D1 and D4 model are (1.23g,
18%) and (1.57g and 65%) for Imperial Valley event, respectively.
c) For a specific model, failure sequence and collapse modes are dependent on the excitation.
For example, collapse progress of the D4 model starts with pier shear capacity and ends
with abutment curvature capacity failure mode when the model was subjected to Landers
earthquake. While, collapse progress of the model starts with pier plastic curvature and
125
ends with abutment steel rupture failure mode when the model was subjected to Imperial
Valley earthquake.
d) Both model features and ground motion characteristics are important factors in response
and failure modes; as a result, special attention needs to be made to the development of
models including SSI representation and ground motion selection in performing seismic
collapse assessment
4.11 Chapter Summary
In this chapter, the Discrete Approach was employed to study the Soil-Structure effects on the
response of the MRO Bridge. Four different discrete archetype models were constructed and used
to perform IDA analyses. Each of these archetype models included a different level of SSI
representation.
Provisions from Caltrans (2019a), API (2007) and reported literature values were used to
define the stiffnesses and damping parameters for springs and dashpots representing effect of soil
on abutment, backwalls and soil supporting piles.
Behavior of reinforced concrete for structural components were defined based on the work
by Werner et al. (1993). The performance criteria for main structural components were calculated
based on recommendations of the Seismic Retrofitting Manual for Highway Structures: Part1-
Bridges (FHWA, 2006) and the Canadian Highway Bridge Design Code (CSA, 2014).
IDA was performed on all the archetype models using 22 selected ground motions adopted
from the PEER NGA West 2 database. The results of the IDA simulations were studied to
investigate the effects of choice of discrete modeling details representing soil and structure.
Sequence of failure modes in different models was also studied. Finally, fragility curves for all the
models were calculated using methods in Baker (2015).
The following major tasks were carried out to investigate the collapse failure modes and to
calculate the collapse fragility curves:
a) Simulating main structural components such as abutment foundations, back walls
and wing walls, pier column and foundations (footing and pilecap), deck slab and
the abutment and pier piles. Including the material and geometry nonlinearities of
the simulated components in the models
126
b) Simulating SSI effects of the bridge supporting soil layers and abutment
embankments using series of springs and dashpots defined based on the API codes
and Caltrans provisions
c) Validating the mode shapes and frequencies of the developed models with the
results of Ambient Vibration Testing
d) Calculating the performance criteria for the simulated structural components using
the Seismic Retrofitting Manual for Highway Structures: Part1-Bridges and the
Canadian Highway Bridge Design Code
e) Selecting a sufficient number of the earthquake ground motions from the PEER
strong ground motion database, NGA West 2, considering their magnitude,
mechanism, hypocentral and rupture distance and dynamic characteristics such as
predominant period and peak ground acceleration
f) Performing IDA analyses on the models using the selected ground motions
considering suitable increasing scale factor
g) Investigating collapse failure modes and sequence of the models considering effects
of the modeling details and ground motions on the mode and sequence of failure of
the structural components
h) Calculating the collapse fragility curves using the method developed by Baker
(2015)
i) Comparing the calculated collapse fragility curves
The following conclusions were made as a result of investigation of the collapse failure
modes and calculation of the collapse fragility curves of the models:
1) It was found that both model details and ground motion characteristics contribute to
sequence of failure and final collapse mode.
2) In the cases where ground motion characteristics appear to be more impactful in response
of the archetype models, similar failure sequence and collapse modes were observed in all
the models.
3) In some other cases, it was observed that details of the model including the SSI
representation dominate the response and failure modes. It highlights the uncertainty
127
around the SSI features representation and their potential impact on structural collapse
predictions and the need to consider various models in collapse assessment of bridges.
4) In some earthquake events, structure was able to utilize its ductility and energy absorption
as the models experienced a sequence of multiple modes of failure before collapse.
5) Based on the variability observed in this study and effect of both model details and
earthquakes’ dynamic characteristics, special attention needs to be paid to defining the
details of structural model, including SSI representation and performing the analysis on a
sufficient number of ground motions which would allow for meaningful statistical analysis
of results.
6) Selecting suitable ground motions and combining the different hazards could be a
challenging task due to the uncertainties associated with them, especially for the regions
which different earthquake types are prevalent.
128
Chapter 5: Performance Evaluation of the Bridge Systems
In this chapter a methodology is introduced for performance evaluation of bridge structures

considering SSI. Although a simplified approach for performance evaluation of building structures
was commissioned by the Federal Emergency Management Agency (FEMA) under the ATC-63
Project, there is a lack of similar guidelines for evaluation of seismic performance of bridges.
Selecting ground motion with the appropriate magnitude, distance and shape for each site,
structural period of interest, and hazard level is the most rational way to account for seismic
performance of bridges. There is a lack of a generalized method for evaluating the collapse
performance of a specific type of bridge structure with different configurations and site class
categories. To address this issue a simplified process, similar to FEMA P695 methodology, for
performance evaluation of bridge structures is needed. The purpose of the methodology proposed
here is to provide a rational basis for determining bridge system performance. When properly
implemented in the seismic design or considered in retrofit of seismic-force-resisting systems in
existing bridges, this guideline can help ensure safety against collapse. The proposed procedure
can be used for both the design of new bridge structures or performance evaluation and retrofit of
existing bridge structures.
The concepts of Collapse Margin Ratio (CMR) and Adjusted Collapse Margin Ratio
(ACMR) are employed here for development of a performance assessment procedure. A workflow
is developed to assist engineers in implementing the proposed procedure. Finally, the MRO bridge
simulations presented in the previous chapter are used as a case study to demonstrate the proposed
procedure.
5.2 Collapse Margin Ratio
In this research, the Collapse Margin Ratio (CMR) is employed as the primary parameter used in
characterization of collapse safety of a bridge structure. The CMR introduced in the ATC-63
project report, FEMA P695 (FEMA P695, 2009) is an index widely used to evaluate global seismic
collapse resistance or capacity of buildings. CMR is defined as the ratio of the median spectral
acceleration of the collapse level ground motions, SˆCT , to the 5%-damped spectral acceleration of
129
the Maximum Considered Earthquake (MCE) ground motions, SMT, at the fundamental period of
a seismic force resisting system, T1 (FEMA P695, 2009). The collapse Margin Ratio alternatively
can be defined as the ratio of corresponding spectral displacements of the above defined spectral
accelerations as follows:
ŜCT SDCT
CMR = = (5.1)
SMT SD MT
The following tasks are needed to calculate the CMR:

– Task 1: Selecting a sufficient number of ground motions
– Task 2: Performing IDA analyses and calculating the IDA curves
– Task 3: Calculating the Collapse Fragility Curve (CFC)
– Task 4: Determining the Maximum Considered Earthquake (MCE) based on the
relevant soil site class and Calculating Collapse Margin Ratio (CMR)
Figure 5-1 demonstrates the sequence of the tasks required to calculate a CMR-Median for
a typical archetype model in this study:
3.5 Task 1 Task 2

3
3
2.5
Ground Motion (g)
2.5
IDA 2
Sa (T1) (g)
2
1.5
1.5
1
1
0.5 0.5
0 0
0 1 2 3 4 0 0.25 0.5 0.75 1 1.25
Drift (%)
Period (s)
1
3 Task 4 Task 3
Probability of Collapse, P(C|Sai)observed
SMT=2.65
2.5 MCE CFC
0.75
2
P(C | Sa = SC (T1 )) = 0.5
Sa (g)
1.5 0.5  Ln x −  
CMRMedian=SCT (T1) / SMT (T1) P(C | Sa = x) =   
  
1
0.25
0.5
SCT (T1)
0 0
0 T1=0.3s 1 2 3 4 0 0.5 1 1.5 2 2.5 3 3.5 4
Spectral Acceleration at the Fundamental Period, Sa (T 1) [g]
T (s)
Figure 5-1 Procedure for calculating Collapse Margin Ratio (CMR)

130
The CMR offers an objective measure of assessment of structural collapse. This parameter
combines the fragility curve with a site-specific response spectrum. FEMA P695 recommends a
multiplier of 1.2 for predicted CMR values when in a 3D nonlinear dynamic time history analysis,
the ground motions are applied in both directions simultaneously (paired) (FEMA P695). The
rationale behind this coefficient as presented by FEMA P695 is that 3D simulations with paired
ground motion on average show about 20% reduction in predicted CMR values when compared
to 2D counterparts. This is due to the coupling of forces induced in members due to the use of
ground motion in pairs. It should be noted that this multiplier is different from the SSF multiplier
employed to incorporate spectral shape effects discussed in Section 5.3 of this thesis.
CMR is tightly related to the probability of collapse. Figure 5-2 shows two examples of
fragility curves. The curve on the left, exhibits higher probability of collapse and a lower CMR
value comparing to the curve on the right. The comparison of the fragility curves can be interpreted
in two ways:
a) For a given spectral acceleration, the probability of collapse is higher for the model D4.
b) For a given probability of collapse, the spectral acceleration associated with the model
D4 is smaller compared to D1, resulting in a lower CMR value for D4.
As CMR is the ratio of collapse intensity to MCE intensity, a larger CMR indicates that
structure has more resistance against collapse. Thus, the higher the CMR parameter, the safer the
structure.
131
1.0
Probability of Collapse, P(C|Sai)Observed

0.9
0.8
CMR Decreases
0.7
0.6
D1 Model
0.5
D4 Model
0.4
0.3
Probability of Collapse
0.2 Increases
0.1
0
0 1 2 3 4
Spectral Acceleration at the Fundamental Period, Sa (T1)[g]
Figure 5-2 Reduction of CMR and increase of Probability of Collapse in SSI archetype model D1 and D4
CMR relates global collapse resistance of the structural system to its site class and the
hazard level of interest. Thus, CMR offers an objective safety measure in the period of interest
(fundamental period of the model). In this research, collapse margin ratio for 10% , 20% and 50%
(median) are calculated for each model based on the FEMA P695 recommendation. These values
are also required for purpose of the bridge seismic assessment using the proposed seismic
performance methodology in Section 5.4.
The calculated IDA collapse fragility curves in Section 4.9 cannot be directly used in the
collapse assessment of the structural system due to the modeling uncertainties and spectral shape
effects of the selected ground motions (Deierlein et al., 2008). As discussed by (Deierlein et al.,
2008), the effect of modeling uncertainty (curve b) and spectral shape effects (curve c) are
incorporated by modifying collapse fragility curve (curve a) as shown schematically in Figure
5-3.
132
1
(a)
(b)
PCollapse SSF
0.5
(c)
0 CMR ACMR
1 2 3 4 5
Sa/SMT
Figure 5-3 Reproduced IDA collapse fragility curve (curve a) with modifications to account for modeling
uncertainty (curve b) and spectral shape effects (curve c) (Deierlein et al., 2008)
Fragility curves (a) and (b) are corresponding to two different levels of uncertainty. The
curve (a) reflects record-to-record collapse uncertainty (βRTR) or ˆ defined by Equation (4.35).
While, the curve (b) reflects total system collapse uncertainty (βTOT) defined by Equation (5.13)
and summarized in Table 5-2. Curve (c) is calculated by incorporating the ground motions’ spectral
shape effects discussed in Section 5.3.
In this study, modeling uncertainties are addressed by defining the acceptable collapse
margin ratio (CMRacceptable) which incorporates various aspects of data and modeling uncertainties
for each archetype model. This topic is discussed in detail in Section 5.3. The concept of spectral
shape effects is introduced to address the difference between the spectrum of the selected ground
motions and the Maximum Considered Earthquake (MCE) spectrum (Baker and Cornell, 2006).
In addition to this, the structure’s ductility leads to a decrease in the effective stiffness which; as a
result, increases the period of the structure. FEMA P695 has introduced a Spectral Shape Factor
(SSF) to account for these phenomena, known as spectral shape effects (discussed in detail in
Section 5.3).
133
5.3 Adjusted Collapse Margin Ratio (ACMR)
The Spectral Shape Factor (SSF) is a parameter to account for spectral shape effects. The Adjusted
Collapse Margin Ratio (ACMR) is defined as the product of SSF parameter and the CMR as shown
in Equation (5.2).
ACMR = SSF × CMR (5.2)
The epsilon parameter is used to calculate the SSF. The epsilon parameter (ε) is representative of
difference between ground motion’s spectral acceleration and mean spectral acceleration of a
predicted ground motion (attenuation). Epsilon is a measure of the spectral shape of the records. It
is defined as the number of standard deviations by which a given ln (Sa) value differs from the
mean predicted ln (Sa) value for a given magnitude and distance. This difference is expressed in
terms of number of standard deviations in logarithmic space as shown in Equation (5.3) (Baker,
2011).
ln ( Sa ) − lnSa ( M, R, T )
 (T) = (5.3)
 lnSa
where, ln Sa ( M, R, T ) and  lnSa are mean and standard deviation of ln ( Sa ) and are calculated
using one or more ground motion attenuation equations. The epsilon parameter of each record
varies with the period of the linear system. As an example, the predicted mean epsilon (  0 ) using
Abrahamson and Silva (1997) attenuation relationship is calculated for the Imperial Valley ground
motion and shown in Figure 5-4. In this figure, the predicted epsilon value corresponding to the
fundamental period of the MRO soil+bridge system (T1=0.3s ) is indicated.
134
10
Imperial Valley (SF=1.6)
ε(T)
ε (T1)=1.1 Mean Spectrum-Abrahamson
and Silva (1997)
Sa (T) [g] or ε(T)
0.1
0 T1=0.3s 0.5 1 1.5 2
T(s)
Figure 5-4 Imperial Valley-06 spectrum with 5% damping at the collapse level (SF=1.6) , computed mean
spectrum, and predicted epsilon
In Figure 5-4, mean spectrum is the predicted spectral acceleration using the Abrahamson
and Silva (1997) attenuation. The Epsilon is the number of standard deviations by which
logarithmic spectral acceleration of the Imperial Valley ground motion (SF=1.6) differs from the
mean logarithmic spectral acceleration of the Abrahamson and Silva (1997) ground-motion
prediction equation.
In an IDA analysis, the selected ground motions leading to failure are inherently different
from the Maximum Considerable Earthquake (MCE). Thus, the response spectrum of the motions
has a different epsilon parameter compared to MCE. To take this difference into account, the SSF
parameter shown in Equation (5.4) is calculated as suggested in FEMA P695 (2009).
SSF = exp 1 ( 0 ( T1 ) −  ( T1 )records )  (5.4)
where,
 0 ( T1 ) is the expected or target epsilon value for the site and hazard-level of interest
obtained from the deaggregation of the seismic hazard of the site. For ductile structures,
the probability of exceedance of 0.5% in 50 years is recommended to calculate the mean
epsilon values at collapse level (FEMA P695, 2009) (Tehrani, 2012).
135
 ( T1 )records is the mean epsilon value of the Far-Field ground motion set, evaluated at
period, T1 . The β1 parameter is representative of the sensitivity of collapse-level spectral
acceleration to the variation of epsilon of that ground motion as described in Equation (5.5)
. A large value indicates that predictions are more sensitive to the spectral shape at the
fundamental period of the structure. The value of 1 can be calculated by performing a
regression analysis to derive a relationship between the natural logarithm of the collapse
capacities and the value of  ( T1 ) for each record.
ln ( Sc ( T1 ) ) = 1 ( T1 ) + 0 (5.5)
where,  0 is the average collapse capacity when  ( T1 ) = 0 .
5.4 Proposed Simplified Bridge Seismic Performance Evaluation Procedure (SBSPEP)
A methodology is proposed here for seismic assessment of reinforced concrete bridges. The
methodology of the proposed Simplified Bridge Seismic Performance Evaluation Procedure
(SBSPEP) is based on a comparison of values of the calculated ACMR (discussed in the previous
section) with the acceptable ACMR as recommended in the provisions of FEMA P695 (FEMA
P695, 2009).
The proposed methodology is shown to be sensitive to level of detail and fidelity of the
archetype models employed. To achieve a better understanding of the structure’s performance, it’s
best to choose a set of archetype models to cover the modelling uncertainty. The approach is
flexible though and can be incorporated when developing only a single archetype model is feasible,
given that the model provides a good representation of the key features of the structure/soil system
and the possible failure modes. This methodology allows using either discrete or continuum
approaches in defining the model.
This methodology can be employed with and without incorporating SSI effects. The SSI
effects need to be included when the situation demands; for instance, bridges built on the soft soil.
Although in this study an acceptable collapse probability of 10% or less ( Pacceptable  10 ) is
considered, it is not prescriptive about the choice of acceptable probability of collapse. The
proposed methodology can be incorporated with more restrictive probability of collapse as
applicable.
136
The predictions of the proposed methodology are dependent on the fidelity of its various
components including representation of key structural components, possible damage and failure
modes, choice of ground motions and their applicability to the studied site. Details such as damping
employed can also have a major effect on seismic performance predictions. Therefore, special
attention needs to be paid to calibration and validation of various components of the model.
In the case study of the MRO bridge presented here, a great attention was paid to
development of model using the available data provided in the literature and reports as well as
incorporating models developed by researchers in previous studies of the bridge. When applicable,
some aspects of the model were validated against field measurements (e.g. natural frequencies and
mode shapes and material strength). Nevertheless, inherent aleatoric as well as epistemic
uncertainties exist about the state of the studied soil+structure system. These uncertainties are
addressed in various aspects of the assessment framework. For instance, the TOT parameter
described later in this section incorporates an assessment of total uncertainty present due to various
sources including record data, design requirements, test data and modelling uncertainties.
The level of effort required to incorporate the proposed methodology for design and
assessment of structures is highly dependent on the size complexity of the structure. The available
computational resources are also a key factor in feasibility of the proposed methodology. Latest
advancements in cloud computing and parallel processing are making assessment of even most
complex structures feasible using the proposed methodology.
In order to use the proposed methodology, the following technical skills and competencies
are required by design engineers:
a) Model development skills to build FE representations of the studied structure system,
including knowledge of model verification and validation. Knowledge of developing SSI
models if applicable.
b) Knowledge of performance-based engineering and ability to develop performance
criteria based on available regional codes and best practices.
c) Knowledge of hazard analysis and deaggregation including selecting a set of suitable
ground motions, developing Maximum Considered Earthquake (MCE) and calculating
epsilon for the studied site
137
d) Knowledge of using advanced computational resources including cloud and parallel
computing when applicable to be used for the large and complex models.
To determine an acceptable value of ACMR, the acceptable risk (Pacceptable) and the total
uncertainty of the system (represented by TOT ) need to be identified. The proposed methodology
consists of four main steps as follows:

1. Nonlinear model development for collapse assessment
2. Nonlinear time history analyses
3. Seismic performance evaluation
4. Documentation and peer review
A schematic of the proposed SBSPEP workflow is illustrated in Figure 5-5. A detailed
description of each step is presented below.
In step (1), the finite element model of the system (including SFRSs and the soil if
applicable) is developed. Best practices should be employed in developing the numerical
representation of the archetype models. This includes selection of the main structural components
such as seismic resisting force systems (SFRSs). Both material and geometric nonlinearities play
key roles in response of structures subjected to extreme loading. Therefore, special attention needs
to be paid in incorporating material models representing structural and soil components (when
applicable). Various types of tests, including non-destructive and destructive tests can be used to
characterize the nonlinear behavior of the material. In order to build confidence in performance of
the structural model, a validation procedure is recommended. Performance of structural
components such as piers, abutment backwalls, girders, etc. can be validated against component-
level test data. Ambient vibration data is typically used to validate the global elastic response of
the structure and damping.
The next step (step (2)) is performing nonlinear time history analysis on the developed
model. A key aspect in this step is selection of ground motions. Best practices in selecting ground
motions prescribed by applicable codes and guidelines such as NEHRP (2011) need to be
employed. This includes consideration of an adequate number of suitable ground motions
considering their mechanism, hypocentral and rupture distance, predominant period, peak ground
acceleration, and spectral shape effects and soil conditions of the site. Available online databases
such as the PEER ground motion database and GMPEs offer useful tools for selection of ground
138
motions. An incremental nonlinear dynamic time history analysis is performed on the system to
determine the fragility curves of each archetype model. This step includes calculation of fragility
curves corresponding to each archetype model and estimation of IDA parameters ( ˆ and ˆ )
using Equation (4.34) and Equation (4.35), respectively.
Step (3) includes evaluation of simulation results. To evaluate performance of the
archetype models, adjusted collapse margin ratio (ACMR) is compared with an acceptable
threshold of adjusted collapse margin ratio (ACMRacceptable). The ACMRacceptable is calculated
considering a given probability of collapse when the models are subjected to MCE-level ground
motions. Ultimately, the capacity of the system is calculated considering spectral shape effects.
139
Step 1: Non-linear
Obtain Required
Model Development
Information
for Collapse
Assessment
Develop Archetype
Model(s)
Validate Model(s)
Obtain Required Information
Perform non-linear Step 2: Non-linear

Analyse: Analyses
• Static
• Dynamic and IDA
For Design or Retrofit
Modify the model(s)
and Repeat the Calculate Fragility Curves
Procedures for the Archetype Model(s)
Step 3: Performance
• Calculate CMR(s) , SSF(s)
Evaluation
and ACMR(s)
• Consider Acceptable
Pcollapse
• Calculate Acceptable
ACMR(s)
• Check Validity of ACMR(s)
• Calculate Collapse
Capacity
No ACMR
1
ACMR acceptable
Yes
Step 4: Documentation
Document Results and Peer Review
Figure 5-5 Proposed Simplified Bridge Seismic Performance Evaluation Procedure (SBSPEP)
140
The methodology requires archetypical systems being evaluated for collapse using the
Maximum Considered Event (MCE) with the hazard level at probability of exceedance of 2% in
50 years (return period of 2475 years) for a specific site (FEMA P695, 2009).
The MCE spectrum can be determined from appropriate regulatory bodies including
AASHTO for instance (Figure 5-6).
S DS = FaSS
Csm = A S + ( S DS -A S )( Tm − T0 )
Elastic Seismic Coefficient, Csm
S D1
CSM =
Tm
S D1 = Fv S1 S D1
CSM =
A S = Fpga PGA T2m
0.2 S D1 1.0
TS =
T0 = 0.2Ts S DS
Period Tm (s)
Figure 5-6 AASHTO response spectrum for probability of exceedance 2% in 50 years (reproduced based on
AASHTO, 2017).
where in Figure 5-6,

S1 is horizontal response spectral acceleration coefficient at 1.0 sec period on rock (Site
Class B),
Ss is horizontal response spectral acceleration coefficient at 0.2-sec period on rock (Site
Class B),
PGA is peak ground acceleration coefficient on rock (Site Class B),
Tm is period of vibration of mth mode (s), and
Coefficients Fpgs, Fa, FV are obtained considering seismic design categories for different
site classes using Table 3.10.3.2-1 to 3 Section 3: Loads and Load factors of the AASHTO
provision (AASHTO, 2017).
141
The Canadian provisions for the PBD design require a 5% damped design target response
spectrum with the 2% in 50-year hazard level for the structural damage assessment of RC bridges
(EGBC, 2018). The design target response spectrum or a site-specific response spectrum can be
developed based on the Canadian code provisions if the 5% damped seismic hazard values for that
location corresponding to the 2% in 50 years hazard level are available. Such spectrum can be
considered as an equivalent to MCE spectrum with 5% damping and similar hazard level specified
by the US code provisions.
Currently, seismic design values for the Canadian code provisions for various soil classes
are calculated based on the Canada’s 6th generation seismic hazard model (Adams et al., 2019).
These seismic hazard values for the major Canadian cities and for the probability of exceedance
2% in 50 years and the soil site class C are listed in Table 5-1.
Table 5-1 5% damped 6th generation seismic hazard values (in g unit) for the major Canadian cities, for 2% in
50 year probability on Site Class C (Adams et al., 2019)
Lat.N Long.W Sa(0.2) Sa(0.5) Sa(1.0) Sa(2.0) Sa(5.0) Sa(10.0) PGA
St. John's 47.57 52.72 0.18 0.13 0.079 0.041 0.012 0.0044 0.072
Halifax 44.65 63.6 0.21 0.14 0.082 0.042 0.012 0.0043 0.085
Moncton 46.1 64.78 0.29 0.17 0.1 0.047 0.013 0.0047 0.13
Fredericton 45.95 66.65 0.38 0.21 0.12 0.056 0.016 0.0055 0.17
Québec 46.8 71.23 0.77 0.41 0.21 0.1 0.026 0.0087 0.38
Montréal 45.51 73.55 0.84 0.43 0.22 0.1 0.025 0.0084 0.43
Ottawa 45.42 75.69 0.66 0.34 0.18 0.082 0.022 0.0073 0.33
Niagara Falls 43.1 79.07 0.44 0.22 0.1 0.045 0.011 0.0038 0.24
Toronto 43.65 79.38 0.36 0.19 0.093 0.042 0.011 0.0037 0.19
Windsor 42.3 83.02 0.17 0.1 0.049 0.022 0.0052 0.0018 0.071
Winnipeg 49.89 97.15 0.08 0.044 0.02 0.008 0.0016 0.0005 0.036
Edmonton 53.55 113.47 0.15 0.08 0.04 0.021 0.011 0.0075 0.073
Calgary 51.05 114.08 0.23 0.14 0.068 0.033 0.015 0.01 0.1
Kelowna 49.88 119.48 0.18 0.13 0.09 0.067 0.029 0.018 0.083
Kamloops 50.67 120.32 0.18 0.13 0.09 0.068 0.03 0.018 0.083
Vancouver 49.25 123.12 1.1 0.82 0.47 0.28 0.075 0.032 0.49
Victoria 48.43 123.37 2 1.5 0.87 0.51 0.12 0.047 0.83
Tofino 49.12 125.88 1.8 1.7 1 0.66 0.15 0.06 0.76
Q. C Island 53.26 132.08 2.3 1.8 1.1 0.6 0.16 0.068 0.96
The West Coast region of Canada has been affected by three distinctive active sources of
earthquakes including shallow crustal, deep subcrustal sources, and the Cascadia subduction zone
(NRCan, 2020). As a result, the Canadian cities which are in the West Coast region, have the
highest seismic hazard values compared to other regions in the country.
142
As an example, the design target response spectrum with 5% damping and probability of
exceedance 2% in 50 years for Vancouver, Victoria, and Queen Charlotte Island (Haida Gwaii) in
the British Columbia (BC) is calculated for the Soil Class C. These spectra are calculated using
the seismic hazard values given in the Table 5-1 and recommendations given by the Clauses 4.4.3.3
and 4.4.3.4 and Table 4.1 of the S6-14 Canadian Highway Design code (2014). The calculated
design spectra for these cities are shown in Figure 5-7.
2.5
Vancouver
2
Victoria
Sa(T) [g]
1.5
Queen Charlotte Island
(Haida Gwaii)
1
0.5
0
0 1 2 3 4 5
T (s)
Figure 5-7 Design spectrum with 5% damping and probability of exceedance 2% in 50 years for Victoria,
Vancouver, and Queen Charlotte Island (Haida Gwaii) in the British Columbia
The most accurate and reliable approach to calculate a site-specific hazard spectrum, which
can be considered as an equivalent to desired level of hazard MCE, is to perform a Probabilistic
Seismic Hazard Analysis (PSHA) and deaggregation. The PSHA and deaggregation should be
performed considering the desired hazard level and damping, period of interest, applicable
attenuations, and the average shear wave velocity of top 30m of the site. In this approach, the mean
value of the ground motion magnitude and epsilon for the location can be obtained, which can
itself be used in selecting the suitable ground motions required for the IDA analysis. The PSHA
can be performed using commercial software to develop site-specific earthquake hazard analysis.
Site class is typically categorized by soils stiffness. AASHTO (2017) provides the site
classification of the soil based on the soil characteristics such as average shear wave velocity of
top 30 meters of the site ( VS30 ), standard penetration resistance (N), undrain shear strength (Su),
143
and Plasticity Index (PI). AASHTO’s site classification is summarized in Table E- 1 in Appendix
E. For the Canadian sites, soil classifications are provided for the seismic site response based on
recommendations given in Table 4.1 of the S6-14 Canadian Highway Design code (2014).
Based on the proposed Simplified Bridge Seismic Performance Evaluation Procedure
(SBSPEP), a value of acceptable collapse margin ratio (CMRacceptable) is required to evaluate a
bridge performance. The CMRacceptable can be concluded as:
  Sa 50%
c

 − 
 ln ( Sa MCE ) − ln ( Sa 50% )  =    Sa MCE
c ln
P ( failure | Sa = Sa MCE ) =    (5.6)
  TOT    TOT 
   
 
c
where, Sa 50% = SCT (T1 ) is the median value of spectral acceleration at the fundamental period, T1
, of a calculated probability of collapse curve for an index archetype model. The Collapse Margin
Ratio (CMR) is defined in Equation (5.7):
c
Sa 50%
CMR = (5.7)
Sa MCE
Substituting Equation (5.8) in Equation(5.9), results in Equation (5.10).
 − ln ( CMR ) 
P ( failure | Sa = Sa MCE ) =    (5.10)
  TOT 
By defining P C = P ( failure | Sa = Sa MCE ) and solving Equation (5.10) for CMR, Equation (5.11) is
obtained (Tehrani and Mitchell, 2012).

1
CMR acceptable =
( )
(5.11)
exp TOT   −1 ( Pacceptable
C
)
ACMR acceptable = SSF  CMR acceptable (5.12)
where,
Φ-1 is the inverse cumulative normal distribution function,
c
Pacceptable is the acceptable probability of collapse,
SSF is given by Equation (5.4) , and
144
TOT represents the total system uncertainty in predicting the collapse capacity of the structure. It
ranges from 0.275 to 0.95. Based on FEMA P695, the total system collapse uncertainty can be
calculated as per (5.13).
TOT =  RTR 2 +  DR 2 + TD 2 +  MDL 2 (5.13)
where,
 RTR is the record-to-record collapse uncertainty (0.20 – 0.40)
 DR is the design requirements-related collapse uncertainty (0.10 – 0.50)
TD is the test data-related collapse uncertainty (0.10 – 0.50)
 MDL is the modeling-related collapse uncertainty (0.10 – 0.50).
FEMA P695 provides a simplified assessment method to estimate the total
uncertainty in the prediction of the collapse capacity. Values of total system collapse uncertainty,
𝛽TOT, for superior model quality and index archetype models with a period-based ductility, T  3
are provided in Table 5-2.
Table 5-2 Proposed total system collapse uncertainty (𝛽TOT) based on quality of model and design for the
period-based ductility, T 3 (FEMA P695, 2009)
Quality of Test Quality of Design Requirements

Data (A) Superior (B) Good (C) Fair (D) Poor
(A) Superior 0.425 0.475 0.550 0.650
(B) Good 0.475 0.500 0.575 0.675
(C) Fair 0.550 0.575 0.650 0.725
(D) Poor 0.650 0.675 0.725 0.825
Selection of the total system collapse uncertainty (𝛽TOT) itself is a source of uncertainty
since it includes a judgemental decision. Performance evaluation using the proposed method
highly depend on the assumed value of 𝛽TOT and special attention needs to be made in selecting
this parameter in the evaluation process. This can be achieved by performing an investigation on
record to record, test data, and modeling requirement uncertainties and performing a sensitivity
analysis prior to performance evaluation.
An acceptable collapse probability of 10% or less is consistent with the collapse
performance objectives of this methodology (FEMA P695, 2009).
145
Pacceptable = 
( )   10%
 ln IM
 ˆ (5.14)
  TOT 
 
However, to calculate the collapse capacities, the spectral acceleration corresponding to
50% probability of collapse (SCT), should be considered. The modified collapse capacity (Sam (T1))
can be calculated using Equation (5.15) (Tehrani, 2012).
Sa m ( T1 )collapse =ACMR  Sa ( T1 )MCE =SSF  SCT ( T1 ) (5.15)
where, Sa(T1)MCE is the spectral acceleration at the fundamental period of the structure from the
Maximum Considered Event (MCE) spectrum. Table 5-3 summarises the required steps for the
seismic assessment of RC bridges based on the proposed methodology as discussed earlier is
shown in Figure 5-5. A more detailed procedure for Step 2 to 4 of the proposed Simplified Bridge
Seismic Performance Evaluation Procedure (SBSPEP) is summarized in Figure 5-8. The procedure
shown in this figure uses an acceptable probability of collapse of 10%. This workflow can be
adopted for smaller collapse probabilities as well.
146
Table 5-3 Required procedures of the proposed seismic performance methodology for RC bridges.
Step Description
2) Perform Nonlinear Analysis This step includes calculation of fragility curves corresponding to each
And Calculate Parameters archetype model and estimation of IDA parameters (θ an 𝛽) using Equation
(4.34) and Equation (4.35), respectively.
3.1) Calculate CMR acceptable Total uncertainty of system (𝛽TOT) can be estimated using Table 5-2. Based
on FEMA P695, an acceptable ceiling of collapse probability (Pacceptable) of
10% and 20% are consistent with the collapse performance objectives of
the methodology for group and individual performance comparison,
respectively (FEMA P695, 2009). Calculate CMRacceptable using Equation
(5.11).
3.2) Calculate Spectral Shape The expected  value for the site and hazard-level of interest ,  0 ( T1 ) ,
Factor (SSF): can be determined by seismic hazard deaggregation or using predicted
 0 ( T1 ) map or relevant table for the required seismic design category and
site and for the period of the structure. Use  0 ( T1 ) corresponding to
probability of exceedance 0.5% in 50 year (FEMA P695, 2009) based on
FEMA P695 recommendation, (Haselton et al., 2011). The mean epsilon
value of the Far-Field ground motion set at period T1,  ( T1 )records , can be
predicted using Equation (5.3). Perform regression analysis for ln(Sa(T1))
obtained from fragility curves and calculated  ( T1 ) to calculate coefficient
𝛽1 using Equation (5.5). Finally, SSF for each archetype model can be
calculated knowing  0 ( T1 ) ,  ( T1 ) , and 𝛽1 using Equation (5.4).
3.3) Calculate ACMR and Calculate ACMR and ACMRacceptable using Equation (5.2) and Equation
ACMRacceptable (5.12), respectively.
3.4) Check Validity of ACMRs Given that there are a variety of attenuation equations available, it is
required to investigate the variability introduced to the results using
different attenuation models. To assess the variability, different attenuation
equations should be used to predict  ( T1 ) and recalculate SSFand ACMR.
The calculated SSF and ACMR values for each attenuation model should
be compared with the corresponding calculations in Steps 3.2 and 3.3.
3.5) Compare ACMR and Base on FEMA P695, in order to achieve acceptable performance,
ACMRacceptable following two criteria needs to be satisfied:
a) The average value of adjusted collapse margin ratio for each
performance group( ACMR i ) exceeds ACMR 10% ( ACMR i  ACMR10% ).
b) Individual values of adjusted collapse margin ratio for each index
archetype model (ACMRi ) within a performance group exceeds
ACMR20% (ACMRi ≥ ACMR20%)
3.6) Calculate Collapse Capacity If both conditions in Step 3.5 are fulfilled, the performance is deemed
(Sam(T1)collapse ) acceptable and modified collapse capacity (Sam(T1)collapse ) for each
archetype model can be calculated using Equation (5.15). If either of the
criteria is not met, another iteration of design or retrofit needs to be
performed as shown in Figure 5-5.
4) Documentation and peer Documentation and review process.
review
147
Step 2. : Estimate IDA parameters (ˆ and ˆ )
IDA: 1 n 
ˆ = exp   ln IM j 
n j= 1 
ˆ =
1 n
(
 ln IM j ˆ
n − 1 j= 1
)
Step 3. : Performance Evaluation
Step 3.1 : Calculate CMRacceptable

Step 3.3 : Calculate ACMR and ACMRacceptable
 TOT , Pacceptable =  
(
 ln IM ˆ 
  10% ) ACMR i = SSFi  CMR i
ACMR acceptable,i = SSFi  CMR acceptable,i
  TOT 
 
1 Step 3.4 : Check Validity of ACMRs

CMR acceptable =
(
exp  TOT   −1
(P
acceptable ))
Different Attenuation →  new ( T1 )records
 new ( T1 )records → SSFnew → ACMR new
Check  ACMR  
Step 3.2 : Calculate Spectral Shape Factor
(SSF)
 0 ( T1 )
Step 3.5 : Compare Average ACMR Ratio (R) and
Calculate Modified Collapse Capacity of the Archetype
ln ( Sa ) − lnSa ( M, R, T )
 ( T1 ) = Model(s), Sam(T1)collapse
 lnSa
No Modify the models

ln ( Sc ( T1 ) ) =  1  ( T1 ) + 0
ACMR
R= 1 and Repeat the
ACMR acceptable
Procedures
( )
Yes
SSF = exp   1  0 ( T1 ) −  ( T1 ) records 
  Sam (T1 )collapse =ACMR  Sa ( T1 ) MCE =SSF  S CT
Step 4. : Probability of collapse (partial or

global instability of SFRs of the archetype
models) is less than 10%. Proceed with
documentation and peer review.
Figure 5-8 Detailed procedures for Step 2 to 4 of the proposed Simplified Bridge Seismic Performance
Evaluation Procedure (SBSPEP)
148
5.5 Example Application: Evaluation Seismic Performance of the Meloland Road
Overcrossing (MRO)
5.5.1 Performance-Based Design Requirement Required Performance Level (PL) and
Seismic Retrofit Categories (SRC) For the MRO Bridge
In this section, the performance requirements of both Canadian code, CHBDC (CSA S6-06, 2006),
and Seismic Retrofitting Manual for Highway Structures-Bridges (FHWA, 2006) are investigated.
The Performance Based Design (PBD) approach requires meeting certain performance criteria for
a specific bridge category, described as tolerated levels of structural damage, and serviceability
objectives at three hazard levels with 2%, 5% and 10% probabilities of exceedance in 50 years.
Based on the Canadian Highway Bridge Design Code provisions, bridges are classified into three
importance categories including lifeline bridges, emergency-route bridges and other bridges (CSA
S-6, 2006). Service and damage level of each category bridge for different hazard level are
summarized in Table 5-4.
Table 5-4 Proposed seismic design performance criteria by the Seismic Subcommittee of the CHBDC (CSA S6-
06, 2006)
Seismic Ground Motion Probability Service Level Damage Level

of Exceedance (Return Period)
Lifeline Bridges
10% in 50 years (475 years) Immediate Minimal
Emergency-Route Bridges
Other Bridges
2% in 50 years (2475 years) Possible loss of service No collapse
Based on the Canadian Highway Bridge Design Code provisions, The MRO bridge can be
considered as an Emergency-route bridge. As it can be seen form Table 5-4, Emergency-route
bridge must be repairable after the occurrence of a design earthquake with 10% probability of
149
exceedance in 50 years. In the case of occurrence of MCE level ground motions with 2%
probability of exceedance in 50 years hazard, significant damage is expected but collapse should
be prevented. Based on the required performance level, the damage in the structural components
should not exceed acceptable limits, ensuring prevention of collapse. Also, code provisions require
no collapse in all hazard levels.
Based on the Seismic Retrofitting Manual for Highway Structures-Bridges (FHWA, 2006),
seismic Performance Level (PL) of bridges is categorized into four main groups as summarized in
Table 5-5.
Table 5-5 Seismic Performance Categories
Performance Level Description

Performance Level 0 (PL0) No minimum level of performance is recommended.
Performance Level 1 (PL1) Life safety. Significant damage is sustained during an
earthquake and service is significantly disrupted, but
life safety is assured. The bridge may need to be
replaced after a large earthquake.
Performance Level 2 (PL2) Operational. Damage sustained is minimal and full
service for emergency vehicles should be available after
inspection and clearance of debris. Bridge should be
reparable with or without restrictions on traffic flow.
Performance Level 3 (PL3) Fully Operational. Damage sustained is negligible and
full service is available for all vehicles after inspection
and clearance of debris. Any damage is repairable
without interruption to traffic.
According to Table 1-5 of the Seismic Retrofitting Manual for Highway Structures-
Bridges (FHWA, 2006), seismic hazard level of the MRO site is categorized zone IV as calculated
SD1 and SDs parameters listed in Table 5-7 are greater than 0.4 and 0.6, respectively.
150
Table 5-6 Performance-based Seismic Hazard Categories based on Table 1-6 of the Seismic retrofitting manual
for highway structures-bridges (FHWA, 2006)
Seismic Performance Level

Hazard Level
During Upper Level Earthquake During Lower Level Earthquake
PL0 PL1 PL2 PL0 PL3
I A A B A C
II A B B A C
III A B C A C
IV A C D A D
Based on Table 5-6 and considering that the MRO bridge should remain operational
(Performance Level PL2), performance-based Seismic Retrofit Category (SRC) of the bridge can
be classified as “D” during the upper level earthquake. In addition, Table 5-1 of the manual
recommends nonlinear dynamic method and inelastic time history analysis as evaluation method
and analysis for existing bridges with Seismic Retrofit Category (SRC) of D (FHWA, 2006).
5.5.2 Developing Maximum Considerable Earthquake (MCE) For MRO
Using USGS Vs30 map viewer, average shear wave velocity in top 30m is listed as 227.8m/s for
the Meloland site (USGS, 2019). According to the table provided in Appendix E, site Class D
should be considered for calculating response spectrum. In addition, the following coefficients
and parameters listed in Table 5-7 are obtained using the USGS provided output for the Meloland
site considering risk category IV (essential facilities) (USGS, 2019) based on AASHTO Guide
Specifications for LRFD Seismic Bridge Design (AASHTO, 2017).
Table 5-7 AASHTO coefficients and parameter for Meloland response spectrum
Coefficient or PGA (g) S1(g) Ss (g) SD1 (g) SDS (g) As (g)
Parameter
1.092 1.156 2.645 1.734 2.645 1.092
The response spectrum with 5% damping and probability of exceedances 2% in 50 years

along with ⅔rd. of it are calculated using the AASHTO (2017) general procedure shown in Figure
5-6. These developed spectra are shown in Figure 5-9 for comparison.
151
3
2.5 MCE
Design Response
2
Spectrum (2/3 MCE)
Sa(g)
1.5
0.5
0
0 T1=0.3s 1 2 3 4 5 6
T (s)
Figure 5-9 Developed MCE using AASHTO LRFD (2017) spectrum with 5% damping and design spectrum
(2/3 MCE) for the MRO site
5.5.3 Nonlinear Analysis and Predicting Parameters ( ˆ and ˆ ) (Step 2.)
To evaluate performance of the models using the proposed methodology, calculated fragility
curves for the index archetype model D1, D2, D3 and D4 are used. IDA parameters ˆ and ˆ are
calculated using Equation (4.34) and Equation (4.35), respectively. The predicted parameters are
listed in Table 4-28 for each archetype.
5.5.4 Calculation of CMRacceptable (Step 3.2)
The total system collapse uncertainty (𝛽TOT) equal to 0.475 is deemed appropriate for the bridges
considered in this study based on Table 5-2 as modeling quality and quality of design requirement
are superior, and quality of test data used for the nonlinear hysteretic models were considered as
“good”. Acceptable CMR is calculated using Equation (5.1) and considered total system collapse
uncertainty. Figure 5-10 shows the relationship between the acceptable collapse margin ratio
(CMRacceptable) and the total system collapse uncertainty (βTOT) for the index archetype models for
a probability of collapse 10% and 20%.
152
2.0 D1 Model 2.0
D1 Model
D2 Model
D2 Model
D3 Model D3 Model
D4 Model D4 Model
CMR20% (log scale)

CMR10% (log scale)
0.2 0.2
0.2 0.4 0.475 0.6 0.8 1 0.2 0.4 0.475 0.6 0.8 1
βTOT βTOT
Figure 5-10 Relationship between total uncertainty of the system (βTOT) and acceptable collapse margin ration
(CMRacceptable) for the index archetype models in logarithmic scale
As shown in this figure, βTOT parameter influences the predicted acceptable collapse
margin ratio (CMRacceptable) and, consequently, the predicted collapse capacity of each archetype
model. Thus, it is highly important to accurately estimate uncertainty of the system.
5.5.5 Calculating Spectral shape factor (SSF) (Step 3.3)
The spectral shape factor (SSF) is required to calculate the collapse capacity of the archetype
models. Based on the methodology,  0 ( T1 ) ,  ( T1 ) records and β1 required to be calculated.
5.5.5.1 Mean Expected Epsilon,  0 (T )
Three different methods are investigated to calculate the expected  0 (T ) value. This parameter
depends on the site and hazard level of interest.

a) Mean Predicted epsilon  0 values for California Sites Cities Suggested by Haselton and co -
workers (2011) for Site Class B

The first approach uses the methodology suggested by Haselton and co-workers (2011). In this
paper they have listed the mean predicted  0 for cities in California for different seismic design
category. Table F- 1 (see Appendix F ) lists the  0 values for different Seismic Design Category
153
(SDC) for periods of 0.2s and 1.0s. As per this table, the value of  0 is interpolated to about 2.0 for
a period of 0.3s (fundamental period of the MRO).

As discussed by Haselton et al. (2011), the values of  0 correspond to site class B (i.e. rock
sites). For sites with soft soil (e.g. the case of MRO bridge) subjected to strong shaking will require
further investigations.
b) USGS map for Western United States
One of the alternative methods to obtain is  0 ( T ) using the Figure 5-11 prepared by Harmsen and
co-workers (2002) and presented by the USGS for the Western United States for a period of 1.0s.
The value of epsilon for period of 0.3s and for Meloland bridge site is about 2.0 for 2% in 50 years
hazard level. Similar to Haselton’s paper, the values presented on this map also correspond to site
class B and further investigation is required to apply it to soft soil.
Figure 5-11 Predicted 0 for the Western United States and for period of 1.0s and a 2% in 50 years (Harmsen
et al., 2002).
c) USGS Unified Hazard Tool (USGS, 2019)

Given the shortcoming of the above two approaches in providing values for  0 for the site class D
(soft soils), seismic hazard deaggregation needs to be considered. Using USGS unified hazard tool,
considering Boore Atkinson (2008), Campbell and Bozorgnia (2008), and Chiou and Youngs
(2008) attenuation equations for the Meloland bridge location and period of interest (USGS, 2019).
Result of this deaggregation is shown in Figure 5-12.
154
Figure 5-12 Seismic deaggregation for MRO hazard site using Dynamic for 2% in 50 years hazard level based
on Conterminous U.S. 2008 (v3.3.1) data (USGS, 2019).
Figure 5-13 shows the mean expected epsilon versus period. This graph is calculated for
0.2s, 1s, and PGA by combining results of Seismic hazard deaggregation. Based on this graph a
value of 1.88 is estimated for  0 (T ) .
2.5
1.5
ε(T)
0.5
0
0 0.5 1 1.5
T (S)
Figure 5-13 Expected epsilon value versus period
Among the calculated values of  0 calculated considering the above mentioned three
approaches, the value of  0 (T1 = 0.3s) = 2.0 is employed for probability of failure 0.5% in 50 year
as per FEMA P695 (FEMA P695, 2009).
155
5.5.5.2 Calculating  ( T1 )records , β coefficients, and SSF
The mean and standard deviation of the logarithmic spectral acceleration (μlnSa(T1) and σlnSa(T1)) are
required to calculate ε(T1)records. These parameters are calculated using MATLAB codes developed
by Baker and co-workers (2019) based on Boore and Atkinson (2008), Abrahamson and Silva
(1997), Boore et al (1997) , and Campbell (1997) attenuation relationships. Then mean epsilon of
the records (  ( T1 )records ) is calculated at the fundamental period of the archetype models using
Equation (5.3). Calculated  ( T1 )records for each archetype models are summarized in Table 5-8.
To estimate β0 and β1 coefficients, a regression analysis is carried out for each archetype
model. The relationship between ε(T1) and ln(Sc(T1)) for archetype model D4 is shown in Figure
5-14.
1.0
lnSc(T1) = 0.122 ε(T1) + 0.314
0.8
0.6
ln Sc (T1)
0.4
0.2
0.0
-1.0 -0.5 0.0 0.5 1.0 1.5
-0.2
-0.4
ε(T1)
Figure 5-14 Estimated β0 and β1 coefficients (intercept and slope of the linear regression line) for the D4
archetype model
This calculation is performed for all the other archetype models and the results are
presented in Appendix G.
5.5.6 Check Validity of ACMRs (Step 3.4)
The accuracy of the performance evaluation method is influenced by the selection of ground
motions and ground motion prediction equations (attenuations). To validate the ACMR, different
156
attenuation equations are considered and the calculated SSF and ACMR values are compared prior
to calculation of the modified capacity of the system. As a result, Abrahamson and Silva (1997),
Campbell (1997), and Boore et al (1997) and Boore and Atkinson (2008) attentions equations are
considered and calculated for comparison. The calculated average epsilon of the records,
 ( T1 )records , and regression coefficient, β1, for each archetype and for each attenuation equation
are shown in Table 5-8 and Table 5-9, respectively.
Table 5-8 Calculated value of  ( T1 ) records using different attenuation equations.
Predicted  ( T1 )records
Index
Archetype Abrahamson and Boore et al. Campbell Boore and Atkinson Average of All
Model Silva (1997) (1997) (1997) (2008) Attenuation
Equations
D1 0.43 0.26 0.34 0.85 0.47
D2 0.43 0.26 0.34 0.85 0.47
D3 0.43 0.26 0.34 0.85 0.47
D4 0.40 0.16 0.26 0.78 0.40
Table 5-9 Calculated value of β1 using different attenuation equations
Regression Coefficient β1
Index Abrahamson and Boore et al. Campbell Boore and Atkinson
Archetype
Silva (1997) (1997) (1997) (2008)
Model
D1 0.168 0.119 0.128 0.199
D2 0.155 0.130 0.152 0.187
D3 0.163 0.138 0.141 0.228
D4 0.121 0.067 0.084 0.155
Ultimately, the spectral shape factor of each archetype is calculated and summarized in
Table 5-10. The calculated epsilon using the Boore and Atkinson attenuation equation is
significantly larger than the values using other attenuation relationships shown in Table 5-8.
However, the calculated spectral shape factors using Boore and Atkinson attenuation equation are
similar to the values calculated using other attenuations. See Table 5-10.
157
Table 5-10 Calculated Spectral Shape Factors, SSF, using different attenuation equations
Spectral Shape Factors (SSF)

Index Archetype Abrahamson and Boore et al. Campbell Boore and Atkinson
Model Silva (1997) (1997) (1997) (2008)
D1 1.31 1.24 1.24 1.27
D2 1.29 1.26 1.30 1.25
D3 1.30 1.28 1.27 1.32
D4 1.22 1.14 1.16 1.22
The calculated SSF value using Equation (5.4) does not show significant variability when
different attenuation equations are used in calculation of  ( T1 )records . As a result, the adjusted
collapse margin ratios and collapse capacity of the archetype models can be calculated using
Equation (5.2) and Equation (5.4), respectively.
5.5.7 Comparison of Predicted ACMR and ACMRacceptable (Step 3.5)
Based on FEMA P695, an acceptable performance is achieved when the Adjusted Collapse Margin
Ratio (ACMR) of each performance group for each model meet the following conditions:
a) The average value of adjusted collapse margin ratio for each performance group ( ACMR i
) exceeding ACMR10% ( ACMR i  ACMR10% ).
b) Individual values of adjusted collapse margin ratio for each index archetype model (
ACMR i ) within a performance group exceeding ACMR20% ( ACMR i  ACMR 20% )
ACMR and ACMRacceptable are calculated and a total system uncertainty of 0.475 ( TOT ) using
Equation (5.2) and Equation (5.12), respectively. The calculated ACMRacceptable and ACMR values
are listed in Table 5-11.
158
Table 5-11 ACMR, ACMRaceptable and their ratio (ACMR/ ACMRaceptable) corresponding to MCE level (2% in
50 years), Sa(T1)MCE=2.65g
Acceptable Probability of Collapse

C
Pacceptable = 10% C
Pacceptable = 20%
Index Archetype
Model SSF ACMR ACMRacceptable Ratio ACMRacceptable Ratio
D1 1.33 1.04 0.95 1.10(Y) 0.86 1.21(Y)
D2 1.35 1.04 0.96 1.08(Y) 0.87 1.19(Y)
D3 1.38 1.04 1.00 1.04(Y) 0.90 1.16(Y)
D4 1.22 0.81 0.98 0.83(N) 0.89 0.92(N)
Average 1.32 0.98 0.97 1.01(Y) 0.88 1.12(Y)
(Y) : Methodology requirement is fulfilled
Note
(N) : Methodology requirement is NOT fulfilled
As shown in Table 5-11, the average value of adjusted collapse margin ratio for the
performance group ( ACMR ) exceeds ACMR 10% by 1%. In addition, individual values of adjusted
collapse margin ratio for each model within the performance group ( ACMR i ) except i model D4
exceeds ACMR 20% . As a result, model D4 does not have enough collapse resistance and the results
indicate that retrofit for the bridge needs to be considered.
5.5.8 Calculation of Collapse Capacity (Sam(T1)collapse ) (Step 3.6)
The methodology requires evaluation of collapse under Maximum Considered Earthquake (MCE)
for various geographic regions of seismicity ground motions. The Meloland bridge soil Site Class
is D (stiff soil) with risk category IV and the value of the spectral acceleration of its corresponding
MCE at its fundamental period, Sa(T1)MCE is about 2.65g. AASHTO Code requires the
superstructure remain elastic if an inelastic time-history method is used in evaluation performance
of the bridge structures (AASHTO, 2017). Given that a nonlinear time history analysis is employed
here, the response modification factor, R, equal to 1.0 must be employed.
Considering a total system uncertainty ( TOT ) 0.475, the probability of collapse of models,
D1, D2, D3, and D4 facing ground motions with a probability of exceedance 2% in 50 years (return
period 2475) is calculated 5%, 6%, 7%, and 33%, respectively.
159
The modified collapse capacity of the archetype models (Sam(T1)collapse) is calculated and
listed in Table 5-12 using Equation (5.15) and the value of the calculated Spectral Shape Factors
(SSF) for each archetype model.
Table 5-12 Summarized modified collapse capacity of the archetype models using different attenuation
relationships
Modified Collapse Capacity, Sam(T1)collapse (g)

Index Archetype
Abrahamson and Boore et al. Campbell Boore and Atkinson Total Attenuation
Model
Silva (1997) (1997) (1997) (2008) Equations
D1 2.25 2.13 2.13 2.18 2.29

D2 2.19 2.14 2.20 2.12 2.29
D3 2.18 2.14 2.13 2.20 2.30
D4 1.79 1.67 1.70 1.79 1.79
5.5.9 Documentation and Peer Review (Step 4)
The proposed methodology requires that average adjusted collapse margin ratio ( ACMR ) of the
archetype models equal or greater than average acceptable adjusted collapse margin ratio (
ACMR acceptable ) for the probability of collapse equal or less than 10% for the performance
evaluation within 2% in 50 years hazard level. In addition, it implies that the adjusted collapse
margin ratio of each archetype model ( ACMR i ) equal or greater than its corresponding acceptable
adjusted margin ratio (ACMRi, acceptable) for the probability of collapse 20% with MCE level ground
motions. One of the archetype models did not meet the requirement of the proposed methodology.
As retrofitting of the MRO bridge is beyond the scope of this study, re-modification of the
archetype models and repeating the procedures as well as documentation peer review are not
carried out.
5.6 Chapter Summary
In this chapter included assessment of performance of the archetype model bridges of Meloland
Road Overcrossing (MRO). A Simplified Bridge Seismic Performance Evaluation Procedure
(SBSPEP) was introduced for RC bridges. This evaluation process which is developed based on
FEMA P695 is applicable for design, assessment and retrofitting of RC bridges and can include
160
SSI effects. This methodology can potentially consider soil-structure interaction effects, but
including these effects is not essential in this approach. However, when SSI effects are a key
contributor to response of the system, e.g. bridges built on soft soils, it is highly recommended to
consider the SSI effects in the archetype models. The methodology includes calculation of collapse
margin ratio (CMR) and adjusted collapse margin ratio (ACMR) by incorporating spectral shape
effects. This methodology considers modeling uncertainty in calculation of acceptable collapse
marginal ratio. Comparison of the calculated ACMR with acceptable ACMR values provides a
means of assessment for seismic performance evaluation of RC bridges. As a case study, this
evaluation procedure was employed to assess the performance of model of the MRO bridge. It
was concluded that the bridge model does not satisfy the performance requirements of the
proposed methodology.
The ground motions spectral shape effects play an important role in assessing collapse
performance of a bridge system. A simplified method needs to be developed to account for the
ground motions’ spectral shape effects in Canada, especially for the West Coast.
161
Chapter 6: Nonlinear Response of the MRO Using A Continuum Model
In Chapter 3 using the continuum representation of soil in SSI mode, it was shown that Kinematic
SSI has a significant effect on bridge foundation motion. However, in IDA analysis using a discrete
representation presented in Chapters 4 and 5, the approach that is most commonly used in practice,
the applied motions are considered uniform through the depth of foundation. Rahmani and co-
workers (2016) used a sub-structuring method to incorporate the variation of ground motion along
the depth of the piles. Given the inherent assumptions in the sub-structuring method, its application
is limited to elastic systems. In the case of collapse analysis considering SSI, stiffness of the
structure can significantly change during the event leading to changes in the interaction between
structure and soil.
The continuum modeling approach offers the most comprehensive representation of soil
and structural components allowing for simulation of nonlinear response of soil and structure in a
single simulation. Despite that 3D continuum modeling offers the most comprehensive
representation of soil+bridge system, this approach has been rarely employed in studying the
effects of soil structure interaction on structural response due to the complexity of the nonlinear
analysis and its required significant computational time. Comparisons of the ability of discrete and
continuum approaches to capture different soil-structure interaction effects for large-scale models
such as 3D bridge models are presented in Table 1-1.
Accurate and efficient modeling of soil+bridge systems is very cumbersome. To perform
analyses of such systems, a multiplatform simulation is used as no software alone has the capability
to include all SSI features (Kwon and Elnashai, 2008). One of the 3D multiplatform simulation
of bridges is the MRO model developed by Kwon and Elnashai (2008). In this model,
embankments, abutments, and pile groups are modeled using OpenSees software considering
nonlinear soil material model and the bridge is modeled in Zeus-NL software. Then both models
were integrated using the multiplatform simulation platform, UI-SimCor. Then model was verified
through comparison with previous studies and measured data (Kwon and Elnashai, 2008). The
configuration of the combined model is shown in Figure 6-1. It should be noted that in
multiplatform simulations, analysis is performed in a sequential order. This results in neglecting
162
the interaction between two different platforms, as such simulations require applying output from
one software as an input to another software.
Figure 6-1 3D simulation of the MRO developed by Kwon and Elnashai (2008) using multiplatform soil+bridge
system technique
Another 3D multiplatform simulation of highway overcrossing bridges is a 3D continuum

bridge model developed by Jeremic and co-workers (2009) simulating the Painter Street Overpass
(PSO) in Northern California. The model was developed to investigate of the influence of the non–
uniform soil conditions on the bridge. Due to computation limitations, the abutments were not
modeled, and it was assumed that the bridge deck and abutments were not connected. The infinite
element model of the bridge had 151, 264 elements and 484,104 degrees of freedom. Figure 6-23
shows the 3D model used by (Jeremic et al., 2009).
Figure 6-23D Soil+bridge continuum model of the PSO developed and used by Jeremic and co-workers (2009)
163
Zhang et al. (2008) and Elgamal et al. (2008) simulated 2D and 3D models of the Humboldt
Bay Middle Channel Bridge in Northern California using the open-source software platform
OpenSees. Later, this model was used by Lu et al. (2011) and Rahmani (2014) to predict seismic
response of the soil-bridge system using high-performance computing techniques to reduce the
computational time for the required nonlinear dynamic analyses. The soil domain represented in
this computational model is 650 m long, 151 m wide, and 74.5 m deep. The Humboldt Bay Middle
Channel Bridge model is shown in Figure 6-3.
(a)
(b)
Figure 6-3 Numerical simulation of the Humboldt Bay Middle Channel Bridge using (a) 2D Continuum model
developed by Zhang et al. (2008) and (b) 3D Continuum model developed by Elgamal et al. (2008).
The Meloland Road Overcrossing (MRO) also was simulated by Rahmani (2014) in
OpenSees using the 3D continuum modeling method to demonstrate the potential of the large-
scale continuum modeling for practical applications and providing a baseline data for evaluating
existing simplified approaches used in practice for dynamic analysis of bridges. In Ramani’s
model, nonlinear hysteresis responses of foundation soils and the critical structural components of
164
the bridge such as piers, were taken into account (Rahmani et al., 2014). Rahmani used solid
elements to model the soil domain and the pile cap. In addition, shell elements were employed to
model the back walls, wing walls, and bridge deck. The structural and geotechnical components
of the developed FE model of the MRO is shown in Figure 6-4.
Figure 6-4 3D continuum SSI model of Meloland Road Overcrossing visualized by the GiD interactive graphical
user interface (Rahmani, 2014)
To offer a comprehensive representation of a soil and structure system capable of capturing

key aspects of the bridge response, a continuum modeling approach is presented in this chapter.
This continuum model includes the following:
Soil:
- Nonlinear elasto-plastic response of soil
- Radiation damping
165
Structure:
- Continuum representation of main structural components including shear force
resisting systems
- Nonlinear material representation for both concrete and rebars including damage and
failure
- Nonlinear geometry to capture large deformations and P-Delta effect
Advancements in commercial finite element tools in terms of their offering of a
comprehensive library of constitutive models, as well as availability of powerful computational
tools and clusters have made use of continuum models in SSI analysis more feasible now.
In this chapter a detailed 3D representation of the MRO bridge is developed and presented
as a case study to showcase the capabilities of continuum models in predicting SSI features and
their effect on structural damage and collapse.
6.2 Continuum Model (C Model ) of the Meloland Road Overcrossing
The 3D model developed in this study includes detailed representation of the MRO bridge features
including:
a) Box girder deck slab (10.363m×62.943m) with one interior and two exterior webs and
including pier and intermediate diaphragms,
b) Abutment walls (0.457m×3.581m×10.363m)
c) Abutment wing walls 0.305m thickness seated on the Abutment footing and connected to
the Abut. backwall,
d) Abutment footing foundations (0.914m×0.533m×10.668m),
e) Pier Column with 6.25m height and 1.524m diameter,
f) Pier footing (3.048m×3.048m×0.61m) and pilecap (1.524m×1.524m×1.067m)
foundations,
g) 1×7 18m long abutment and 5×5 15m long pier pile group timber piles with modulus f
elasticity, E=1240 MPa, and Poisson ratio, ʋ=0.2. Piles are 0.32m at Top and 0.20m at
bottom and 0.152m imbedded in the foundations
Soil representation of the model includes representation of:
a) Supporting underground soil layers down to 20m,
b) Embankment behind abutment backwall up to 7.5m with 2H:1V side slopes, and
166
c) Soil embankment in front of abutment with 1.5H:1V slope.
3D view of the constructed continuum model in ABAQUS considered as a reference
analysis model is shown in Figure 6-5. Embankment and idealized soil layers are shown in Figure
3-6(a) and their elastic material properties are summarized in Table 3-1.
The following section and material properties summarized in Table 6-1 are considered in
calculation of stress-strain relationship and force-displacement of the MRO’s RC members based
on Table 2.2 of the bridge model identification report (Werner et al.,1993).
Table 6-1 Section and material properties are used in the MRO continuum model (Werner et al., 1993).
Parameter Value (unit)

2
(EI)crack 3.37E+09 (Nm )
2
(EI)Uncrack 7.48E+09 (Nm )
Crack (α) 0.45(%)
f'c 35.6 (MPa)
εu 0.003 (m/m)
4
Ig 2.65E-01 (m )
4
Icrack 1.19E-01(m )
Steel yield Stress (fy) 312.3 (MPa)
Poisson atio (νc) 0.17(-)
Pier Column axial load (P) 5,382.4 (kN)
Soil elastic material properties are identical to the developed elastic model of the MRO
described in Table 3-1 in Chapter 3. In addition, the Rayleigh damping coefficients used in the 3D
continuum model for the structural and soil components are similar to the developed elastic model
of the MRO described in Chapter 3. The calculated Rayleigh damping coefficients using a damping
ratio 4% and 12.5% for structural components and soil layers are summarized in Table 3-5.
The model consists of a total of 66 dependent part instances. It has also a total 158,442,
and 427,332 number of nodes and elements, respectively. The elements include: 361,544 linear
tetrahedral elements of type C3D4, 23,546 quadratic tetrahedral elements of type C3D10, 240
quadratic tetrahedral elements of type C3D10M, 41,122 linear line elements of type T3D2, and
880 linear hexahedral elements of type CIN3D8. A four-node tetrahedral element (C3D4) is used
167
to model structural and soil components. Like in the development of the elastic model, a CIN3D8
element is used in the boundary wall as Absorbing Boundary Conditions (ABCs) to simulate
unbounded domain as shown in Figure 3-6(c), Figure 3-6(d) and Figure 3-7.
3D nonlinear continuum model (C) is developed in ABAQUS and used in nonlinear
analyses of the MRO bridge reference model. The Discretized model is shown in Figure 6-5.
20m
z
y
x
Wingwalls
Abutment
Wall
7×1 Pile group 18m 15m

7×1 Pile group
5×5 Pile group
Figure 6-5 Developed 3D nonlinear continuum model (C) using ABAQUS software as Meloland Road
Overcrossing (MRO) bridge reference model.
Mode shapes of the first three modes of the developed nonlinear continuum model of the
MRO is shown in Figure 3-8 and compared with the result from carried out ambient vibration test
by Ventura et al. (2011) and previous studies in Table 6-2.
168
Table 6-2 Comparison of the modal period of different simulated FE models and vibration test of the MRO
Study 1 2 3 4
Mode Period (s)
1 0.28 0.32 0.35 0.30
2 0.25 0.31 - 0.28
3 0.23 - - 0.22
(1) This study
Note: (2) FE model developed by Kwon and Elnashai (2008) and shown in Figure 6-1.
(3) FE model developed by Rahmani (2014) and shown in Figure 6-4.
(4) Result obtained from ambient vibration test carried out by Ventura et al. (2011).
6.2.1 Nonlinear Material Models for Structural Members
The constitutive model is normally expressed in terms of stress-strain relations in continuum

mechanics. Stress-strain relations members are investigated for concrete and steel structural
members in the following sub-section and defined in the MRO continuum model. In modeling
reinforced structural members, concrete and steel are both represented in the finite element model.
Concrete is represented using continuum solid elements and reinforcing steel bars are modelled
using embedded wires in the continuum elements.
6.2.1.1 Unconfined and Confined Concrete Model
Uniaxial compression stress-strain response results from the gradual formation of microcracks
within the structure of the concrete. The compression stress-strain response for unconfined
concrete consists of a parabola from zero stress to the compressive strength of the concrete, for
normal strength concrete. Then stress–strain curve rises as a parabola curve formed to a maximum
stress at the strain that corresponds to the peak compressive stress, is often considered to be 0.002
and reaches a strain 0.003 by a descending branch. The equation for this parabola was originally
introduced by Hognestad (1951). After, many researchers including Todeschini et al. (1964) and
Thorenfeldt et al. (1987) have modified the Hognestad parabola considering different analytical
approximations (Wight and MacGregor, 2012).
169
Some of the commonly used concrete compression response curves are shown in Figure
6-6. In this study, the concrete compression stress-strain relation is calculated based on the widely
used parabola described by Equation (6.1), Equation (6.2), and Equation (6.3).
f c f c = 0.9f c
Linear f c = 0.9f c
f c f c
0.15f c
Stress, fc
Stress, fc
Stress, fc
  ε   ε  2

f c = f c 2   -      ε   ε 2  2f c( ε ε 0 )
f c = f c 2   -    fc =
  0   0     ε 0   ε 0   1+ ( ε ε 0 )
2
E ct = tan α E ct = tan α E ct = tan α
2f  ε 0 = 1.8f c E ct ε 0 = 1.7f c E ct

ε0 = c
Strain, ε E ct Strain, ε Strain, ε
(a) (b) (c)
Figure 6-6 Unconfined concrete compression stress-strain response model (a) Hognestad (1951), (b) Modified
Hognestad, and (c) Todeschini et al. (1964)
  ε   ε 2 

fc = fc 2   -    (6.1)
  ε 0   ε 0  
where,
f c is the maximum compression strength,
0 is the strain corresponding to the maximum compression strength, f c , and calculated as

per Equation (6.2)
2f c
ε0 = (6.2)
E ct
E ct is tangent modulus of concrete and can be calculated as per Equation (6.3) in MPa.
E ct = 5500 f c (6.3)
In this study, the Hognestad model is used to develop stress–strain response curve for the
abutment and pier column of the MRO considering an ultimate strain of ε cu = 0.003mm / mm and
concrete compression strength ( fc = 35.6MPa ) specified by Werner et al. (1993) for the main
structural component of the bridge. The stress-strain response curve for this study along with other
response curves considering different analytical approximations are presented in Figure 6-7.
170
40
35
30
Stress, fc (MPa)
25 Hognestad (1951)
20 Modified Hognestad Parabola
15 Todeschini et al. (1964)
10 Thorenfeldt et al. (1984) for HSC

(15-125MPa)
5
0
0 0.0005 0.001 0.0015 0.002 0.0025 0.003 0.0035 0.004
Strain, ε (m/m)
Figure 6-7 Calculated unconfined concrete compression response using different analytical approximation
methods based on considering concrete strength of 35.62 MPa
Tension stiffening is a key attribute in response of reinforced concrete members that needs
to be considered. When reinforced concrete is subjected to tension, concrete assists the reinforcing
steel bars in carrying the tensile force even after severe cracking has developed. This concrete
contribution is called tension stiffening (Carreira and Chu, 1986).
Various researchers have investigated the tension stiffening effects using both
experimental and analytical approaches. This includes works of Nilson (1972), (Carreira and Chu,
1986), Vecchio and Collins (1986), Massicotte and co-workers (1990), Nayal and Rasheed’s
(2006), Wahalathantri and co-workers (2011), Alih and Khelil, (2012), and Allam and co-workers
(2012).
The stress–strain response of concrete loaded in axial tension in the modified Vecchio and
Collins (1986) model can be divided into three phases: A linear phase prior to the maximum
concrete tensile stress (fct,max), a curve portion is calculated using Equation when concrete is
cracked (6.4), and a linearly varying cut-off portion when reinforcing steel bars are yielded. The
stress–strain response of concrete loaded in axial tension is shown in Figure 6-8 (Adebar, 2015).
f cr
fc t = (6.4)
1+ 500 ( ε - ε cr )
171
where,
f ct is the concrete tensile stress,
 cr is the crack strain, and is given by Equation (6.5),
f cr
ε cr = (6.5)
E ct
f cr is the crack stress (modulus of rupture), and
E ct is the tangent modulus and can be calculated as per Equation (6.3).
f cr
f ct =
f ct,max 1+ 500 ( ε - ε cr )
Equilibrium Check
Phase 2: Cracked Concrete in at Crack
f ct  ( f y - f s ) ρ
Reinforced Concrete
Stress, fct
Phase 1:
Elastic Regime Phase 3:
Reinforcement Yielding
Plain Concrete
Strain, ε
Figure 6-8 Stress-strain relationship of plain and reinforced concrete in tension and tensile stiffening with a
cut-off (Adebar, 2015).
Wahalathantri and co-workers (2011) modified the proposed Nayal and Rasheed’s (2006)
tension stiffening model to calculate the stress–strain response of concrete loaded in axial tension
in order to avoid ABAQUS run time errors (See Figure 6-9). They changed sudden drop at critical
tensile strain  cr in Nayal and Rasheed (2006) model from (  cr , f t 0 ) to ( 1.25 cr , 0.77f t 0 ) as
shown in Figure 6-9(b) to avoid convergence issues in the ABAQUS material model. By applying
this modification, the stress-strain curve follows the Nayal and Rasheed (2006) tension stiffening
model but terminated at ( 8.7 cr , 0.1f t 0 ) to avoid ABAQUS simulation convergent issues
(Wahalathantri et al., 2011).
172
ft
f t0
f t0
0.8f t0
Tensile Stress , ft
Tensile Stress , ft
0.8f t0 Primary
Cracking Stage
Secondary 0.45f t0
0.45f t0
Cracking Stage
0.1f t0
ε ε r 1.25ε r 4ε r 8.7ε r ε
εr 4ε r 10ε r
Tensile Strain, ε t Tensile Strain, ε t
(a) (b)
Figure 6-9 Simplified tensile stress-strain model (a) Nayal and Rasheed (2006) tension stiffening model, (b)
Wahalathantri et. al (2011) tension stiffening model
Allam et al. (2012) proposed a modified tension stiffening curve performing nonlinear
analysis and using the finite element method to investigate the concrete contribution in tension
after flexural cracking of reinforced concrete members. Their proposed tension stiffening curve is
shown in Figure 6-10.
ft
f ct,max
Tensile Stress , ft
2 3f ct,max
3 8f ct,max
1 6f ct,max
ε cr 3ε cr 5.875ε cr 10.5ε cr ε max,t = 16ε cr εt

Tensile Strain, ε t
Figure 6-10 Simplified tensile stress-strain relationship proposed by Allam et al. (2012)
Compressive and tensile behavior of the unconfined concrete of the pier column and
abutment walls and wing walls in the continuum reference model of the MRO are calculated based
173
on Hognestad’s parabola shown in Figure 6-7 and Allam and co-workers (2012) simplified tension
stiffening model presented in Figure 6-10. The concrete compressive strength ( f c ) and crack stress
( f cr ) values used in calculating compressive and tensile behavior of the RC members are based on
the values reported by Werner et al. (1993) and considering one tenth of the concrete strength (
0.1f c  0.6 f c ), respectively. The calculated stress-strain response for the unconfined concrete of
the pier and abutment walls and wing walls is shown in Figure 6-11.
10
0
-0.004 -0.003 -0.002 -0.001 0 0.001 0.002 0.003
-5
Stress (MPa)
-10
-15
-20
-25
-30
-35
-40
Strain (m/m)
Figure 6-11 Stress-strain curve for the unconfined concrete main structural members of MRO based on
concrete compression strength ( f c ) and ultimate strain ( ε cu ) value 35.6MPa and 0.003mm/mm, respectively.
Compression strength is considered a negative value.
6.2.1.2 Concrete Damage Plasticity (CDP) Model
The Concrete Damage Plasticity (CDP) model is a constitutive model used to predict the
constitutive behavior of concrete. CDP describes the constitutive behavior of concrete by
introducing scalar damage variables for compressive and tensile response of concrete. The
concrete Damage Plasticity (CDP) model for concrete in ABAQUS has the following attributes
(SIMULIA, 2018):
a) Isotropic linear representation of elastic behavior of the material
174
b) Isotropic tensile and compressive plasticity along with notions of isotropic elastic damage
represent the inelastic behavior of concrete.
c) Capable of modeling monotonic, cyclic, and/or dynamic loading under low confining
pressures,
d) Permits user control of stiffness recovery effects during cyclic load reversals,
The concrete damage plasticity model characterizing compressive and tensile response of
concrete in ABAQUS are shown Figure 6-12.
ft 0
fc 0
t
t
t
c0
t0
(a) (b)
Figure 6-12 Behavior of concrete under axial strength (a) compressive and (b) tension (SIMULIA, 2008)
The reduction of the elastic modulus is given in terms of a scalar degradation variable (d)
in the concrete damaged plasticity model as shown in Equation (6.6).
E = (1- d)E 0 (6.6)
where, E 0 is the initial (undamaged) modulus of the material.
The CDP model for concrete compressive and tensile behavior associated with
compression hardening and tension stiffening are given in Equation (6.7) and Equation (6.8),
respectively (SIMULIA, 2018).
dcfc
ε cpl = ε inc - (6.7)
(1- d c ) E c
dtft
ε plt = ε ck
t - (6.8)
(1- d t ) E c
where,
f c and f t are the concrete compressive and tensile stress (strength),
175
Ec is the initial (undamaged) elastic stiffness of the concrete is given by (6.3),
ε cpl and ε pl
t
are the compressive and tensile plastic strain, respectively,
ε in
c
and εck
t
are the compressive inelastic and cracking tensile strain, respectively, and
d c and d t are compressive and tensile damage variables, respectively. These variables
characterize the degradation of the elastic stiffness on the strain softening branch of the stress-
strain curve and ranging from 0 to 1.
In ABAQUS, compressive stress data are provided as a tabular function of inelastic strain
( ε in
c
) instead of plastic strain ( ε cpl ). Inelastic strain is calculated using Equation (6.9).
ε inc = ε c - ε cel0 (6.9)
where, εel0c , is given by Equation (6.10).
f c0
ε elc0 = (6.10)
Ec
ABAQUS automatically converts the inelastic strain ( ε in

c
) values to compressive plastic strain (
ε cpl ) values using the Equation (6.7). Unloading data are provided to ABAQUS for tension
stiffening in terms of the cracking strain ( εck

t
). Cracking strain is calculated using Equation (6.11)
ε ck
t = ε t - ε t0
el
(6.11)
where, ε elt , is given by Equation (6.12).

0
f t0
ε elt0 = (6.12)
Et
Using the Equation (6.8), ABAQUS automatically converts the cracking strain ( εck
t
) values to
tensile plastic strain ( ε pl

t
) values.
In addition, the plastic damage model in ABAQUS requires the values of Poisson’s ratio
(ν) and five plastic damage parameters to describe compressive and tensile behavior. The five
plastic damage parameters are the dilation angle (ψ), the flow potential eccentricity (e), the ratio
of initial equiaxial compressive yield stress (fb0) to initial uniaxial compressive yield stress (fc0),
176
the ratio of the second stress invariant on the tensile meridian to that on the compressive meridian
(k), and viscosity parameter (Ve) that defines viscoplastic regularization. ABAQUS
documentation recommends values of e=0.1, fb0/fco =1.16, k=0.66, and Ve=0.0 for concrete
material (SIMULIA, 2018). The dilation angle (ψ) and Poisson’s ratio (ν) are considered as 37°
and 0.17, respectively.
6.2.1.3 Concrete Damage Plasticity Model Verification
To verify the Concrete Damage Plasticity (CDP) model in ABAQUS, input and output stress-strain
relationship for a single element model subjected to pure compressive and tensile force are
compared and presented in Figure 6-13.
40 4
35 3.5
30 3
Stress (MPa)
Stress (MPa)
25 2.5
Input
20 Input 2
Output
15 Output 1.5
10 1
5 0.5
0 0
0.0 0.0 0.0 0.0 0 0.0005 0.001 0.0015 0.002 0.0025
Strain (m/m) Strain (m/m)
(a) (b)
Figure 6-13 Comparison of stress-strain curves of ABAQUS input with (a) single element subjected uniaxial
compression, (b) the element subjeted to uniaxial tension.
As it can be seen from Figure 6-13, ABAQUS predicted stress-strain response (ABAQUS
model output) in both uniaxial compressive and tensile loading scenarios are in good agreement
with the model input.
6.2.1.4 Mesh Dependency and Localization Issue
As discussed in the previous section, a plasticity-damage material model that exhibits softening
stress-strain response, in both tension and compression, is being employed. Softening models lead
to the well-known phenomena of strain localization where inelastic strain gets concentrated in a
finite zone in the structural components. The predicted size of the damage zone in a finite element
177
simulation is typically proportional to the element size. As a result, the finite element predictions
show a dependency on element size, known as mesh size dependency.
In the reinforced concrete beam elements, softening of the hinge is caused by strain
softening of concrete and yielding of steel (Bazant and Cedolin, 1991). In the discrete model,
presented in Chapter 4, the size of the plastic hinge zone, which exhibits strain localization, was
specified to be 15% of the height of the abutment backwall and pier column ( lp l  15% ). In the
continuum model presented in this chapter, the size of the plastic hinge is dependent on the mesh
size.
A mesh size dependency study was performed on both abutment backwall and pier columns
to investigate the effect of mesh size on push-over response and developed size of the plastic zone.
A series of the pushover analyses are carried out on both pier column and abutment backwall
models in ABAQUS software considering different mesh sizes along the height of these members.
Figure 6-14 and Figure 6-15 show the plastic strain in abutment backwall and pier columns,
respectively.
(a) (b)
Figure 6-14Maximum(a) (b)backwall with (a) fine mesh size (0.3m
principal plastic strain at capacity of the abutment
along the height of wall) (b) coarse mesh size (0.9m along the height of wall)
178
(a) (b)
(a) (b)
Figure 6-15 Maximum principal plastic strain at capacity of the pier column with (a) fine mesh size (0.3m along
the height of the column) (b) coarse mesh size (0.9m along the height of the column)
The calculated pushover curves are shown in Figure 6-16.
2000
900
Snap Through
Horizontal Applied Force (kN)
1800
Horizontal Applied Force (kN)
800
1600
700
1400
600
1200
Element Height # 0.3m 500
1000 Element Height # 0.3m
800 Element Height # 0.5m 400
Element Height # 0.5m
Element Height # 0.7m
Element Height # 1.0m Element Height # 0.9m
200 100
0 0
0 0.05 0.1 0.15 0.2 0 0.05 0.1 0.15 0.2
Top Pier Column Horizontal Displacment (m) Top Abutment Wall Horizontal Displacment (m)
(a) (b)
Figure 6-16 Effect of mesh size in force-displacement response of the structural components (a) pier column
(Vertical load=5382.3kN), and (b) abutment backwall (Vertical load=2691.2kN)
As shown in Figure 6-14 and Figure 6-15, the predicted size of the plastic hinge where
plastic strain localization takes place, is equal to the height of an element. As it can be seen in
Figure 6-16, the push over response of both structural members show a strong dependency on the
179
mesh size. While the elastic response of the members is independent of the mesh size, the response
starts to deviate as soon as damage initiates. As a result, the peak force, and the post-peak behavior
can be significantly different when the mesh size is changed. The dissipated energy in the damage
process, which is the area under the load-displacement curves, is also strongly mesh-size
dependent. When larger element sizes were employed, both structural members show a more
ductile behavior and exhibit higher amount of energy dissipation in the damage process.
The response curve of the pier column (see Figure 6-16 (a)) has a smooth softening
behavior as the column’s dissipative capacity is more than its stored strain energy. While the
abutment wall’s response curve shows a sudden drop in the load which is a sign of snap-through
(see Figure 6-16(b)) In this case, the dissipative capacity is less compared to its sored strain energy,
leading to an unstable response in the post-peak regime reaching the peak force (Bazant and Oh
1983).
Various remedies have been proposed to address the localization issue. One of the simplest
remedies is the crack-band method proposed by Bazant and Oh (Bazant and Oh, 1983). Albeit the
approach cannot resolve the mathematical issues resulting the localization problem, the crack-band
method can be used to prevent the mesh-size dependency in the finite element simulations (Sluys,
1992). The fracture energy, G f , can be calculated using Equation (6.13).
G f = g f× h d (6.13)
where, gf and hd are the area under the stress-strain curve known as the fracture energy density and
the height of the damaged zone, respectively.
To keep the energy dissipated during the damage progression consistent and independent
of element height, the softening part of the stress-strain relation is scaled according to the height
of the element in the crack-band approach. In a local finite element analysis, damage localizes in
a single row of elements. Therefore, height of the damaged zone (hd) would be equal to the height
of that element (he). The crack-band method makes the fracture energy fairly consistent and
independent of the mesh size by scaling the fracture energy density, gf , according to the height of
element as shown in Equation (6.14) .
Gf
g fFE = (6.14)
he
180
Other solutions to localization issue include using non-local approaches, which are not
currently available in ABAQUS software. Using rate-dependent damage-plasticity models can
offer another remedy to the underlying mathematical issue that leads to the localization problem.
Given the lack of data on rate-dependency in concrete damage, this option was not employed.
Given the limitations of ABAQUS software, the local approach had to be employed. In this
case, mesh size had to be chosen to 1) ensure consistency between the predicted size of the plastic
hinge and experimental observations in terms of damage height and 2) achieve appropriate load-
displacement and energy dissipation response that agrees with experimental data (if available).
The mesh size in pier column and abutment backwalls in the finite element model for the
MRO were chosen accordingly to make sure the predicted height of damage is around the expected
size of the plastic hinge.
6.2.1.5 Nonlinear model for Steel Rebars
In continuum modeling, three approaches can be used to simulate RC components in FE software:

namely, homogenized approach, solid (3D) rebar representation, wire (1D) embedded rebar
representation.
In this research, the homogenized approach is used to simulate the bridge deck slab,
foundations, and abutment wing walls. Rebars in abutment backwalls and pier column are
simulated using embedded 1D wires.
In the homogenized approach, the rebars are not explicitly represented. Instead, the effect
of the reinforcing steel bars, and confinement can be incorporated in the reinforced concrete
response in a homogenized and smeared manner. In this approach, the tensile and compressive
behavior of the RC concrete members can be considered using the tension stiffening concept and
theoretical stress-strain model for confined concrete introduced by Mander (Mander et al., 1988).
The unified stress-strain approach for confined concrete suggested by Mander et al. (1988) is based
on monotonic loading at slow strain rates. It allows for cyclic loading and includes the effect of
strain rate. In this approach, the monotonic loading stress-strain curve is assumed to form an
envelope to the cyclic loading stress-strain response (Mander et al., 1988). As shown in Table 6-3,
although performing an analysis is feasible in the homogenized representation in large and
complex simulations, there is a significant convergence issue due to localization of plastic strain.
181
In solid (3D) rebar representation, rebars are represented in the FE simulation using solid
elements. Since rebars are explicitly present in the model, the model for concrete needs to reflect
unconfined condition, i.e., to consider unconfined concrete stress-strain response curve in the
model. The rebars will be responsible to bring the effect of confinement, and rebars and concrete
are connected using tie constraint. This approach is suitable for component-level simulation (piers
or walls, etc.). The 3D approach requires a very fine mesh for the rebar and in concrete surrounding
the rebars. The very fine mesh makes the model computationally costly and impractical for large-
scale simulations. Solid (3D) representation of the pier column rebar is shown in Figure 6-17(a).
Total number of
Total number of
elements: 385,254
elements: 1,854
Total number of
Total number of
nodes: 119,515
nodes: 1,872
(a) (b)
Figure 6-17 Comparison of the simulation of the pier steel cage in the MRO 3D continuum model (a) 3D solid
representation approach (b) Wire (1D) embedded reinforcing steel bars’ approach
In embedded wire (1D) approach, rebars are represented using 1D members with truss
cross-section assignment. The embedded region interaction feature in ABAQUS allows for
coexistence of solid elements (representing concrete) and the 1D elements (representing the rebars)
enforcing continuity in displacement field. This approach does not require similar meshes in
concrete and rebar and as a result, a relatively coarse mesh can be employed. This feature makes
the embedded approach feasible for simulation of large-scale reinforced concrete structures. In this
approach, behavior of unconfined concrete is assigned to the concrete elements, and steel’s
nonlinear behavior is assigned to the wire rebars. Markou and Papadrakakis (2011) have shown
182
that the result obtained using wire (1D) rebar representation is in a good agreement with the solid
(3D) representation of rebars. This makes the 1D wire embedded approach a feasible solution
without sacrificing simulation accuracy.
The homogenized approach where reinforced concrete behavior is modelled in a smeared
sense and is applied to the whole cross section. Whereas in discrete modelling of the rebars (in
both solid and 1D wires), the location of rebars are in accordance with their actual placement in
the concrete member. The advantages and disadvantages of the different approaches in simulations
of RC components in a FE model are summarized in the Table 6-3.
Table 6-3 Comparison of different representation of rebars in a FE continuum model
Criterion RC Simulation Approach

Homogenized Solid (3D) Rebar Wire (1D) Rebar
Representation Representation Representation
Rebar positions Effect of rebar modelled in Rebars are Rebars are
representations a smeared manner simulation in their simulation in their
exact positions exact positions
Confinement and tension Implicitly simulated Explicitly Explicitly
stiffening effects thought the homogenized simulated simulated
stress-strain response
Computational cost Moderate Costly Moderate
Analysis feasibility for large Feasible Infeasible Feasible
and complex models
Given the advantages of the wire embedded 1D method, this approach has been employed
for modelling abutment backwalls, wing walls and pier column in MRO 3D model.
An elastic-perfectly plastic model is used to define the axial behavior of the steel rebars in
the 3D continuum model based on steel yield stress ( f y ) reported by Werner et al (1993) as listed
in Table 6-1. Figure 6-18 shows the reinforcing steel bars’ stress-strain response curve employed
in the MRO 3D continuum model.
183
400
300
200
100
Axial Stress (MPa)
0
-0.015 -0.01 -0.005 0 0.005 0.01 0.015
-100
-200
-300
-400
Axial Strain (m/m)
Figure 6-18 Reinforcing steel bars’ stress-strain response curve considered in the 3D continuum model
6.2.2 Nonlinear Material Models for Soil
Like the structural components, constitutive model for the soil is expressed in term of a stress-
strain relation in continuum mechanics. As a result, stress-strain response of the soil layers and
abutment embankments are studied to implement their nonlinearities in the MRO continuum
model. Stress-strain behavior of soils can be idealized to linear elastic, nonlinear elastic perfectly
plastic, and a hardening response model. Perfectly plastic and hardening response models are
often called conventional plasticity models (Desai and Zaman, 2014). Real and idealized soil
stress-strain relationship are shown in Figure 6-19.
184
Soil Real Response Soil Idealized Response
 
 Unloading
E Unloading
Failure E
Idealization Loading Loading

 
Residual (b) (c)
 

(a) Yield Point Yield Point
 
(d) (e)
Figure 6-19 Real and idealized stress-strain behavior of Soils (a) real response, (b) linear elastic, (c) nonlinear
elastic, (d) Perfectly plastic response, and (e) hardening response (Desai and Zaman, 2014)
As discussed by Desai and Zaman (2014), conventional soil plasticity models such as
Drucker–Prager, and Mohr–Coulomb, provide a reasonably good prediction of ultimate or failure
strength of the material but do not provide realistic predictions for the entire stress–strain response
(Desai and Zaman, 2014).
In this research, the Mohr-Coulomb (M-C) plasticity model is used in the continuum C
model of the MRO bridge to represent soil’s yield behavior. In the Mohr-Coulomb (M-C) plasticity
model, it is assumed that failure is determined by the maximum shear stress which shear stress
depends on normal stress. Cohesion and friction angle are the two key parameters in defining the
M-C plasticity model. M-C model simulates elastic-perfectly plastic behavior. The elastic behavior
is linear and when soil yields, the behavior becomes perfectly plastic. In uniaxial loading
representation, the Mohr-Coulomb can be defined by the following Equations using Mohr circle
shown in Figure 6-20 (SIMULIA, 2018).
τ = c - σ tan (  ) 0    90 (6.15)
τ = s cos (  ) (6.16)
σ = σ m + s sin (  ) (6.17)
185
Substituting Equation (6.16) and Equation (6.17) in Equation (6.15), we have:
s + σ m sin (  ) - c cos (  ) = 0 (6.18)
where,
ϕ is internal angle of friction or angle of failure envelope,
τ, σ, and c are critical shear stress, normal stress, and cohesion, respectively,
σm and s are the average of the maximum and minimum principal stresses (the normal
stress) and half of the difference between the maximum and minimum principal stresses,
respectively.
Failure Envelope
( ,  )
σ1 - σ3
s=
2
c

 1 1 3 3
σ1 + σ3
m =
2
Figure 6-20 Mohr-Coulomb model (SIMULIA, 2018)
The M-C criterion can be written for general states of stress in term of three stress
invariants: equivalent pressure stress (p), the Mises equivalent stress (q), and deviatoric stress (r)
as follow (SIMULIA, 2018):
p = -1/ 3 trace(σ) (6.19)
q= (3 / 2 (S : S )) (6.20)
1
9 3
r =  S.S : S  (6.21)
2 
where, σ is normal stress matrix and S is stress deviator and defined as per Equation (6.22).
186
S = σ + pI (6.22)
The M-C yield surface can be written as per Equation (6.23).

F = R mc q - p tan (  ) - c = 0 (6.23)
where, R mc (θ, ) is given by Equation (6.24).
1 1
R mc (θ, ) = sin ( θ + π / 3) + cos ( θ + π / 3) tan (  ) (6.24)
3sin (  ) 3
in which, θ is the deviatoric polar angle shown in Figure 6-21 and defined as per Equation (6.25)
.
( q)
3
cos ( 3θ ) = r (6.25)
θ=0
Mohr-Coulomb
(  = 20 )
o
θ=π 3
Tresca
( = 0 )
o
Rankin
(  = 90 )
o
θ = 2π 3 θ = 4π 3
Drucker-Prager
(Mises)
Figure 6-21 Mohr-Coulomb deviatoric plane. Image reproduced from B QUS user’s manua (SIMULIA,
2018)
The following Mohr-Coulomb soil plasticity model parameters should be specified for each
soil layer in the ABAQUS (SIMULIA, 2018):
a) Deviatoric eccentricity ( e ) can vary between 1/2 and 1 ( 1 2  e <1 ) and its value is given
by the Equation (6.26).
e=
(3- sin ( ) )
(3 + sin ( ) )
(6.26)
187
b) Meridional eccentricity (ε) which should be set equal to the flow potential eccentricity in
the meridional plane. The meridional eccentricity (ε) defines the rate at which the flow
potential approaches its asymptote. Its default value is 0.1.
c) Cohesion parameters: cohesion yield stress and absolute plastic strain
d) Plasticity parameters: friction angle (ϕ) and dilation angle (ψ)
Calculated Mohr-Coulomb plasticity parameters for the soil layers shown in Figure 6-22 are
listed in Table 6-4.
A
B
C
D
E
Figure 6-22 Embankment and supporting soil layers in the simulated nonlinear continuum model of the MRO
Table 6-4 Five M-C plasticity parameters of the soil layers defined in the nonlinear continuum model (C model)
Mohr-Coulomb Plasticity Parameters

Soil Layer ϕ° ψ° c e ε
Embankment 31.5 0.3 20 0.95 0.1
A 27 0.2 35.9 0.52 0.1
B 33 3 17.95 0.50 0.1
C 25 0.1 76.6 1 0.1
D 33 3 12.0 0.50 0.1
E 25 0.1 86.2 1 0.1
1) ϕ° is friction angle in degree.
Note 2) ψ° is dilation angle in degree.
3) c is cohesion yield stress in kPa.
4) e is deviatoric eccentricity and its value is calculated as per Equation (6.26).
5)  is meridional eccentricity, and it is considered 0.1.
To investigate stress-strain relationship of the soil layers, a series of Consolidated

Undrained (CU) triaxial shear numerical analysis are carried out using a unit cube element
subjected to a shear stress considering Mohr-Coulomb (M-C) Plasticity in ABAQUS. In the cube
188
model soil response is considered elasto plastic and Mohr-Coulomb plasticity parameters defined
as per the values listed in Table 6-4 for different soil layers. Result of the triaxial direct shear test
when a pre-consolidation pressure 0.6MPa is applied to the unit cube is shown in Figure 6-23.
400
350
300
Yield Stress (kPa)

Embankment
250
Soil Layer 1
200
Soil Layer 2
150 Soil Layer 3
100 Soil Layer 4

Soil Layer 5
50
0
0 0.02 0.04 0.06
Strain (m/m)
Figure 6-23 Investigation stress-strain response of soils using direct shear test on a unit cube model
6.3 Collapse-level Ground Motions
A set of ten ground motions were scaled to the collapse level corresponding to the archetype model
D4 listed in Table 4-27 in Chapter 4. Strike-slip mechanism were chosen from the NGA-West 2
ground motion PEER (2018) database to perform the nonlinear response history analyses of the C
model. A criterion similar to the one presented in Section 3.3.3 is employed in this section to
choose the ground motions for finite element analysis. In addition, a significant duration between
5% and 95% of the Arias Intensity of the motions is used as input motion in the nonlinear analyses
of the MRO continuum FE model. The selected input motions are summarized in Table 6-5 and
their time histories are shown in Appendix B.3. The total and significant duration values of the
input motions are summarized in Table 6-6.
189
Table 6-5 Collapse level ground motions used in nonlinear analyses of the MRO continuum model
Event Earthquake Name PEER Year Scale Factor M Tp (s) /Unscaled

Number RSN No (SF) PGA(g)
1 Imperial Valley-06 184 1979 1.6 6.53 0.40/0.48
2 El Mayor-Cucapah-Mexico 5836 2010 3.6 7.2 0.18/0.23
3 Hector Mine 1787 1999 1.6 7.13 0.50/0.33
4 Landers 879 1992 1.4 7.28 0.08/0.79
5 Parkfield 30 1966 1.6 6.19 0.36 / 0.44
6 Superstition Hills-02 723 1987 1.4 6.54 0.64/0.43
7 Bam-Iran 4040 2003 1.2 6.6 0.20/0.81
8 Kobe-Japan 1119 1995 1.2 6.9 0.46/0.70
9 Duzce-Turkey 1602 1999 0.8 7.14 0.4/0.81
10 Kocaeli-Turkey 1158 1999 1.6 7.51 0.38/0.36
Table 6-6 Total and significant duration values of the input time histories used in nonlinear analyses of the
MRO continuum model
Event Number Total Duration (s) Significant Duration (s)

1 39.1 6.6
2 87.5 17.8
3 45.3 9.7
4 48.1 13.8
5 44.0 15.1
6 22.3 10.6
7 66.5 8.0
8 40.9 4.4
9 55.9 9.0
10 27.2 10.9
190
6.4 Nonlinear Dynamic Analysis on the Continuum Model
6.4.1 Pushover Analysis of Pier Column and Abutment Backwall
The pushover analysis was carried out for the pier column and abutment backwall using
the ABAQUS software. To investigate the validity of the simulation, Force-Displacement curves
calculated using ABAQUS software for the pier column and abutment backwall with their
corresponding curves calculated using SeismoStruct software in Section 4.5. To calculate pushover
curve of the pier column and abutment wall, a continuously increasing static lateral force was
applied to the top of each component. As shown in Figure 6-24, the calculated pier column
pushover curve using the ABAQUS software shows a higher displacement capacity compared to
the corresponding value calculated using the SeismoStruct. In addition, the calculated pushover
curve for the abutment wall shows a higher peak force compared to the corresponding value
calculated using the SeismoStruct. There is a good agreement in plateau resisting force for pier
column and abutment components predicted by both software.
2,500 ABAQUS 800 ABAQUS

SeismoStruct 700 SeismoStruct
2,000
600
Force (kN)
Force (kN)
1,500 500
400
1,000 300
200
500
100
0 0
0.00 0.03 0.05 0.08 0.10 0.13 0.15 0.00 0.05 0.10 0.15
Displacement (m) Displacement (m)
(a) (b)
Figure 6-24 Force-displacement curve for the pier and abutment backwall component of the MRO simulated
in ABAQUS and SeismoStruct commercial software (a) Pier column with axial load=5382.4kN (b) Abutment
backwall with axial load=2691.2kN
The difference of the calculated ultimate shear and moment capacity for the pier column
of the MRO using both ABAQUS and SeismoStruct software are compared to the corresponding
values determined by Werner et al. (1993) using BIAX program. This comparison is summarized
in Table 6-7.
191
Table 6-7 Comparison of the MRO pier column ultimate moment and shear capacity using different software
Calculation Method Max. Moment Shear Capacity Difference

(kNm) (kN) (%)
BlAX program (Werner et al., 1993) 10685.2 1709.6 -
SeismoStruct (Lp/L=15%)- InfrmFBPH 10171.3 1627.4 -4.8
ABAQUS 11020.0 1763.2 3.1
As it can be seen from Table 6-7, the calculated capacities for the pier column of MRO,
have less than 5% difference with recommended values by Werner et al. (1993).
To investigate the embedded 1D wire reinforcing steel bars employed in the simulation of
the pier column and abutment backwall in the continuum model, stress in the pier column and
abutment backwall components are studied after performing a pushover analysis using ABAQUS
software. As it can be seen in Figure 6-25, the longitudinal and transversal bars in the plastic hinge
region of the pier column and abutment backwall have yielded. Yielded rebars in these members
are shown in red, where their stress is greater than the steel yield stress (312.3MPa) in Figure 6-25.
F f
f
Longitudinal
bars are yielded
(a)
f
F
Transverse bars
are yielded
Longitudinal bars
are yielded
(b)
Figure 6-25 Monitoring performance of the simulated 1D wire reinforcing steel bars using truss element (T3D2)
embedded in concrete after performing pushover analysis on the components using ABAQUS software (a) pier
column (f= 2.95 MPa) and (b) abutment backwall (f= 0. 0.57MPa)
192
6.4.2 Time History Analysis (THA)
Time History Analysis (THA) was used to investigate the capability of the detailed 3D SSI analysis
in assessing the performance of a soil+bridge system using the continuum representation of the
soil. The set of 10 ground motions is used to performed time history analyses with a 0.02s time
step. In each analysis, the input motion is applied to the bottom of the 3D model simultaneously
in the longitudinal and transvers directions (parallel and perpendicular to the deck alignment). The
significant duration of each record was used to perform these analyses.
ABAQUS’ implicit solver, known as ABAQUS/Standard is employed to perform the THA
analyses This solver employs an implicit dynamic solution with the direct time integration method.
The nonlinear solution is performed iteratively using the Newton's method. The Hilber-Hughes-
Taylor time integration method, which is an extension of Newmark-β method, is used in the
dynamic implicit analyses. Default time integration parameters are employed. The time integration
parameters α, β, and γ associated with the Hilber-Hughes-Taylor operator for moderate dissipation
applications used are: –0.41421, 0.5, and 0.91421, respectively (SIMULIA, 2018).
6.5 Results and Discussion
6.5.1 Investigation of Damage in the Main Structural Members
In concrete damage plasticity model, degradation of the elastic stiffness is characterized by two
damage variables ( d c and d t ) which are functions of the plastic strain, temperature, and other pre-
defined parameters as given in Equation (6.27) and Equation (6.28). The damage variables can
take values from zero and one corresponding to the undamaged material and total loss of strength,
respectively.
dc = f (εcpl ,θ) 0  dc  1 (6.27)
d t = g (ε plt ,θ) 0  dt  1 (6.28)

where,
θ is temperature,
ε cpl and ε pl
t
are compressive and tensile plastic strain, respectively.
193
Concrete tensile and compressive damage parameters corresponding to the abutment
backwall and pier column due to collapse level El Mayor (SF=3.6) and Imperial Valley-06
(SF=1.6) ground motions are shown in Figure 6-26 and Figure 6-27, respectively.
(a)
(b)
Figure 6-26 Predicted concrete (a) tensile damage and (b) compressive damage parameters of the MRO
continuum model at the collapsed level using El Mayor Earthquake (SF=3.6).
194
(a)
(b)
Figure 6-27 Predicted concrete (a) tensile damage and (b) compressive damage parameters of the MRO
continuum model at the collapsed level using Imperial Valley-06 Earthquake (SF=1.6).
The predicted damage parameters for other ground motions are presented in Appendix I.
The reinforcing steel bars play a key role in response of structural elements specially after
concrete gets damaged. As discussed earlier, a plastic model is considered for rebars in the pier
(including pier footing, pile cap, column and pier diaphragm) and abutment backwalls in C model
195
to define reinforcing steel bars plasticity. To investigate the state of rebars, axial stress and plastic
strain in the abutment backwalls and pier cages are studied. Maximum principal stress (tensile)
and plastic strain parameters in the steel cage of the pier column and abutment backwall subjected
to the Imperial Valley-06 Earthquake at the collapse level (Sale Factor =1.6) are shown in Figure
6-28 and Figure 6-29, respectively.
Pier Diaphragm
Abutment Backwall
Pier Column
Pier Footing
Pier Pilecap
Yielded Rebars Yielded Rebars
(a) (b)
Figure 6-28 Maximum principal stress in the reinforcing steel bars (a) abutment wall steel cage (b) Pier steel
cage due to the Imperial Valley-06 Earthquake at collapse level (SF =1.6)
196
Pier Diaphragm
Abutment Backwall
Pier Column
Pier Footing
Localization of Plastic Strain Localization of
Plastic Strain Pier Pilecap
(a) (b)
Figure 6-29 Maximum principal strain in the reinforcing steel bars (a) abutment wall steel cage (b) Pier steel
cage due to the Imperial Valley-06 Earthquake at collapse level (SF =1.6)
Yielded reinforcing steel bars and localization plastic strain areas are shown in Figure 6-28
and Figure 6-29. These figures clearly indicate formation of plastic hinges in both pier column and
abutment walls. Two plastic hinges are formed at the top and bottom of the column creating a
double curvature deformation mechanism. Also as it can be seen from Figure 6-28, the deformation
of vertical steel rebars of the pier column in the plastic hinge zone, showing a buckling mechanism.
Yielding soil layers are monitored using the AC YIELD (actively yielding) flag parameter
which is available for built-in material models in ABAQUS. This identifier is an indicator of
yielding condition (SIMULIA, 2018). The state of soil yielding around the piles in various layers
are shown in Figure 6-30 and Figure 6-31 when the continuum model C was subjected to El mayor
(Scale Factor=3.6) and Imperial Valley-06 (Scale Factor=1.6) ground motions. It can be seen that
there is a significant yielding taking place around the piles.
197
Abutment 1 Pier Abutment 2
1
Acceleration (g)
0.5
0
Soil Layer A -0.5
(Depth=0m-2.17m) -1
0 5 10 15
Time (s)
Acceleration (g)
0.5
Soil Layer B 0
-0.5
(Depth=2.17m-5.53m) -1
0 5 10 15
Time (s)
Acceleration (g)
0.5
Soil Layer C 0
-0.5
(Depth=5.53m-10.1m) -1
0 5 10 15
Time (s)
Acceleration (g)
0.5
Soil Layer D 0
(Depth=10.1m-14.67m) -0.5
-1
0 5 10 15
Time (s)
Acceleration (g)
0.5
Soil Layer E 0
(Depth=14.67m-20m) -0.5
-1
0 5 10 15
Time (s)
Figure 6-30 Yielding state of various soil layers around the piles subjected to collapse level El Mayor ground
motion (SF=3.6). Red color is an indicator of yielding state around the piles and the observed time in the
calculated acceleration time histories.

Acceleration (g)
1
Soil Layer A 0
(Depth=0m-2.17m) -1
0 2 4 6
Time (s)
Acceleration (g)
1
Soil Layer B 0
(Depth=2.17m-5.53m) -1
0 2 4 6
Time (s)
Acceleration (g)
Soil Layer C 0
(Depth=5.53m-10.1m) -1
0 2 4 6
Time (s)
Acceleration (g)
1
Soil Layer D 0
(Depth=10.1m-14.67m)
-1
0 2 4 6
Time (s)
Acceleration (g)
1
Soil Layer E
0
(Depth=14.67m-20m)
-1
0 2 4 6
Time (s)
Figure 6-31 Yielding state of various soil layers around the piles subjected to collapse level Imperial Valley-06
ground motion (SF=1.6). Red color is an indicator of yielding state around the piles and the observed time in
the calculated acceleration time histories.
198
Result of yielding state of soil layers around the piles for other earthquakes can be found
in Appendix J.
6.5.2 Comparison of Nonlinear Dynamic Analysis response from the continuum and
Discrete Models
In this section a comparison between predictions of the Continuum model (Model C) and Discrete
model D4 are presented. The D4 model is the most comprehensive discrete SSI model investigated
in this thesis which includes representation of abutments and pier piles.
The Pseudo acceleration response spectrum (for 5% damping) of the Imperial Valley-06
and El Mayor ground motions at the collapse level computed at the center of the bottom the pier
pilecap in the transverse and longitudinal direction, are compared for the D4 and C model in Figure
6-32.
3 4
C Model C Model
D4 Model 3 D4 Model
2
Sa (g)
Sa (g)
1
1
0 0
0 1 2 3 0 1 2 3
Period (s) Period (s)
4 (a) 5 (b)
C Model C Model
4 D4 Model
3 D4 Model
3
Sa (g)
Sa (g)
2
2
1
1
0 0
0 1 2 3 0 1 2 3
Period (s) Period (s)
(c) (d)
Figure 6-32 Spectra with 5% damping of the bottom pier pilecap using the collapsed level ground motions (a)
Imperial Valley-06 with 1.6 scale factor in the transverse direction (b) Imperial Valley-06 with 1.6 scale factor
in the longitudinal direction (c) El Mayor with 3.6 scale factor in the transverse direction (d) El Mayor with
3.6 scale factor in the longitudinal direction
199
As it can be seen, there is significant difference between results in the short period range,
where the first few modes of the soil+bridge system are residing. A lack of representation of the
kinematic SSI effects, including the depth effect, in the discrete model (D4) is a major contributor
to this observed difference.
The calculated drift responses using the discrete and continuum approaches to capture
different soil-structure interaction effects for the 3D MRO models was also compared. The
predicted drift of the pier column and abutment walls are presented in Figure 6-33.
1.5 2.0 C Model

Pier Column Drift (%)
Pier Column Drift (%)

D4 Model
1.5
1.0
1.0
0.5
0.5
0.0 0.0
1 2 3 4 5 6 7 8 9 10 1 2 3 4 5 6 7 8 9 10
Event No Event No
(a) (b)
Abutment Backwall Drift (%)
Abutment Backwall Drift(%)
0.6 1.0
0.5 0.8
0.4
0.6
0.3
0.4
0.2
0.1 0.2
0.0 0.0
1 2 3 4 5 6 7 8 9 10 1 2 3 4 5 6 7 8 9 10
Event No Event No
(c) (d)
Figure 6-33 Comparison of calculated drift using C and D4 models and collapse level ground motion (a) pier
column drift in the transverse direction (b) pier column drift in the longitudinal direction (c) abutment
backwall drift in the transverse direction and (d) abutment backwall drift in the longitudinal direction.
The calculated maximum drift values for the C and D4 model at the collapse level are summarized
in Table 6-8.
200
Table 6-8 Comparison of the maximum drift of the discrete and continuum model at the collapse level
Model Pier Column Drift (%) Abutment Backwall Drift (%)
Transverse Longitudinal Transverse Longitudinal

Direction Direction Direction Direction
C Model 1.4 1.58 0.6 0.95
D4 Model 1.03 0.82 0.25 0.87
As it can be seen from the Figure 6-33 and Table 6-8, the continuum model has predicted
higher drift values for both the pier column and the abutment backwalls when compared to the
discrete model. In the pier column, the discrete model predicts a drift of up to 1%, whereas the
continuum model predicts up to about 1.6% of drift. In the longitudinal direction of the abutment
backwalls, both models have about 1% drift.
One of the major differences between the continuum and discrete models, is the absence
of representation of the kinematic SSI including depth effect in the discrete model since the same
ground motion time histories are applied to all the soil springs and dashpots. Material constitutive
models employed in continuum and discrete models for structural components were different,
which also contribute to the differences observed between these two models.
A comparison of the maximum base shear of the continuum model (C) and the discrete
model (D4) for the abutment backwall and pier column at the collapse level are presented in Figure
6-34.
201
C Model
10,000 10,000
Base Shear (kN)
Base Shear (kN)

D4 Model
8,000 8,000
6,000 6,000
4,000 4,000
2,000 2,000
0 0
1 2 3 4 5 6 7 8 9 10 1 2 3 4 5 6 7 8 9 10
Event No. Event No.
(a) (b)
10,000 14,000
Base Shear (kN)

Base Shear (kN)
8,000 12,000
10,000
6,000 8,000
4,000 6,000
4,000
2,000
2,000
0 0
1 2 3 4 5 6 7 8 9 10 1 2 3 4 5 6 7 8 9 10
Event No. Event No.
(c) (d)
Figure 6-34 Comparison maximum base shear results at the collapse level (a) pier column shear in the
transverse direction (b) pier column shear in the longitudinal direction (c) abutment backwall shear in the
transverse direction (d) abutment backwall shear in the longitudinal direction
Except for the abutment backwalls in the longitudinal direction where the continuum model
consistently shows higher base shear values, in the other cases, the differences between the base
shear responses of the continuum and the discrete models are ground motion-specific.
Characteristics of each ground motion play a key role in the predicted base shear response.
To compare the sensitivity of base shear of the abutment backwall and pier column of the
continuum and discrete model to their corresponding drift, hysteresis curves of the members are
calculated and compared for the pier column and abutment backwalls in Figure 6-35 and Figure
6-36, respectively.
202
10000 12000
C Model C Model
6000 D4 Model 8000 D4 Model
Base Shear (kN)
Base Shear (kN)

2000 4000
-2000 0
-6000 -4000
-10000 -8000
-0.06 -0.04 -0.02 0 0.02 0.04 0.06 -0.06 -0.04 -0.02 0.00 0.02 0.04 0.06
(a) (b)
10000 8000
C Model C Model
6000 D4 Model D4 Model
4000
Base Shear (kN)
Base Shear (kN)

2000
0
-2000
-6000 -4000
-10000 -8000
-0.06 -0.04 -0.02 0.00 0.02 0.04 0.06 -0.10 -0.06 -0.02 0.02 0.06 0.10
(c) (d)
Figure 6-35 Calculated hysteresis curves of the pier column using the C and D4 model (a) Imperial Valley-06
with SF=1.6 in the transverse direction (b) Imperial Valley-06 with SF=1.6 in the longitudinal direction (c) El
Mayor with SF=3.6 in the transverse direction (d) El Mayor with SF=3.6 in the longitudinal direction
203
10000 12000
C Model C Model
6000 D4 Model 8000 D4 Model
Base Shear (kN)
Base Shear (kN)

2000 4000
-2000 0
-6000 -4000
-10000 -8000
-0.02 0.00 0.02 -0.04 -0.02 0.00 0.02 0.04
(a) (b)
10000
C Model 10000 C Model
D4 Model D4 Model
5000
Base Shear (kN)
Base Shear (kN) 5000
0
0
-5000 -5000
-10000 -10000
-0.02 0.00 0.02 -0.04 -0.02 0.00 0.02 0.04
(c) (d)
Figure 6-36 Calculated hysteresis curves of the abutment backwall using the C and D4 model (a) Imperial
Valley-06 with SF=1.6 in the transverse direction (b) Imperial Valley-06 with SF=1.6 in the longitudinal
direction (c) El Mayor with SF=3.6 in the transverse direction (d) El Mayor with SF=3.6 in the longitudinal
direction
As it can be seen from Figure 6-35 and Figure 6-36, there is a major difference observed
between the discrete and continuum models. It should be noted that the applied ground motions
were scaled to collapse level, leading to extreme hysteresis behavior shown in Figure 6-35 and
Figure 6-36.
204
6.6 Chapter Summary
In the direct method the entire soil-structure system is modelled together. In this study, the direct
method is used to evaluate seismic performance of a soil+bridge system. The direct method offers
the most comprehensive approach to capture SSI effects compared to other methods (e.g.
substructuring method). Both discrete and continuum representations of soil+bridge system can be
used to perform direct SSI analysis. In this chapter a comparison between the predicted response
of Meloland Road Overcrossing (MRO) using discrete and continuum approaches were made.
Discrete models were discussed in detail in Chapters 4 and 5. The continuum model was
constructed in ABAQUS software which included representation of key structural components as
well as soil layers supporting bridge foundations and abutment embankments. To capture the
material nonlinearity, appropriate material behaviors were assigned to concrete, reinforcing steel
bars and soil layers. Damage-plasticity models were applied to model nonlinearity and damage of
concrete in pier column and abutment backwalls. An elasto-plastic model was applied to the
reinforcing steel bars. And finally, plasticity models based on Mohr-Coulomb theory were applied
to the supporting soil layers and embankments. To model radiation damping, ABAQUS’
continuum infinite elements were employed at the edges of supporting soil layers.
Nonlinear time history analysis was performed using the 10 ground motions chosen from
NGA-West2 discussed in Chapter 4. The ground motions were scaled to the collapse level of the
discrete D4 model.
Comparisons between the discrete (D4) and the continuum (C) models were performed on
the following outputs: drifts and base shears pier column and abutment backwall; hysteresis loops
of pier column; pseudo spectral acceleration at the bottom of pier pilecap.
Comparing the seismic response calculated using the nonlinear analysis for the abutment
backwall and pier column of the MRO, it was observed that there is a significant difference
between the results obtained using the continuum and discrete approach. The continuum model
consistently showed higher drift envelopes for both pier column and abutment backwall when
compared to the discrete model. The drift results showed a more sensitivity to the ground motion
characteristics than the correspond base shears. The continuum model also showed higher pseudo
spectral acceleration specially in the higher frequency region where the natural frequency of
205
soil+structure resides. The main contributors to the differences observed between the continuum
and discrete models are summarized below:
- The absence of representation of the kinematic effect in the discrete model. Specifically
lack of representation of the embedment effect, wave scattering, vertically propagated
shear waves, and slab averaging.
- Difference in representation of radiation damping in soil.
- Difference between the constitutive material models used in these modeling
approaches, especially for concrete and reinforcing steel bars.
- Inherent difference between the responses of continuum versus discrete finite element
representations.
It was shown that despite the complexity and computational cost of the continuum method,
this method has a greater capability to predict SSI features that either cannot be directly or
accurately simulated in the discrete models (including kinematic SSI effect). Detailed simulation
of nonlinear behavior of soil and formation and propagation of damage in structural components
are among the benefits of the continuum modeling approach.
Advancements in computational modeling and development of powerful computer clusters
have made the use of continuum models feasible. Analytical SSI modeling requires accurate
representation of behavior of soil and structural components going thorough nonlinear regime.
Successful modeling requires high fidelity constitutive models that can predict cyclic response of
soil layers and structural components undergoing extreme nonlinear conditions and damage and
collapse.
In this study, the behavior of the material models used for concrete members and soils were
validated against monolithic loading conditions. Validating the response of structural members
and soil in cyclic loading scenarios will help to improve the level of confidence in the continuum
model predictions.
206
Chapter 7: Summary, Conclusions and Future Work
7.1 Summary
Past studies show the detrimental role of Soil-Structure Interaction (SSI) in the collapse of bridge
structures founded in soft soil. Despite the emphasis of collapse prevention of bridge structures in
the seismic codes, the provisions do not provide detailed procedures to guide bridge designers to
implement the SSI effects in their designs. In addition, the provisions do not provide a detailed
guidance on how to assess the seismic performance of bridge structures.
In this thesis, a numerical modeling framework was employed to study the effect of SSI on
response of a soil+bridge system. The Meloland Road Overcrossing (MRO) was used as the case
study in this thesis. MRO is a well-instrumented bridge that has experienced multiple earthquakes,
and it has been the subject of many studies in the past. The thesis was carried out in three phases:
a) investigation of the variation of bridge foundation motion from the free-field (Chapter 3); b)
Incremental dynamic analysis to calculate fragility curves, investigation of failure modes and
proposing a simplified methodology to assess seismic performance of RC bridges (Chapters 4 and
5); c) Nonlinear time history analyses on 3D continuum and comparing the seismic response with
a discrete model that obtained from (b)(Chapter 6).
A summary of tasks and outcomes of each chapter is provided below:
Chapter 2 of this thesis provided a literature survey on various approaches to Soil-Structure
Interaction. An overview of various design codes to SSI effects was also presented. It was
concluded that SSI is not sufficiently addressed in current design codes for bridges and there is a
need for detailed guidelines to help engineers in design and performance assessment of bridges.
In Chapter 3, the variations of bridge foundation motion from free-field motion were
investigated using a continuum FE representation of the MRO structure and its supporting soil.
The performance of the Tau-averaging method and the Elsabee and Morray’s transfer function
amplitude model in predicting the foundation motion were studied. It was shown that the Tau-
averaging method and the Elsabee and Morray’s model are not capable of capturing the
amplification in foundation motion due to their limiting assumptions.
Chapter 4 of this thesis focuses on IDA analysis of the MRO. Four different discrete
archetype models were developed with various levels of SSI representation. A total set of 22
207
ground motions were selected from the NGA-West2 PEER database to perform the IDA analysis.
The performance criteria used in this chapter was adopted from the Seismic Retrofitting manual
for the highway structures part 1 (FHWA, 2006). Fragility curves corresponding to collapse of
each archetype model were developed. It was shown that the D4 model, which includes explicit
representation of foundations’ piles, results in a fragility curve that is shifted towards left when
compared with other archetype models. This means that D4 model predicts a higher probability of
collapse given a specific spectral acceleration.
Chapter 5 proposed a simplified bridge collapse assessment method. This method, which
is inspired by FEMA P695, offers a step-by-step guideline for assessment of performance of
bridges considering soil-structure interaction. The proposed methodology is sensitive to level of
detail and reliability of the models employed. The MRO was used as a case study to demonstrate
the workflow of the proposed methodology.
In Chapter 6, a detailed continuum model of the MRO soil+structure was developed to
perform nonlinear time history analyses. Material nonlinearity of both soil and structural
components were considered in the developed 3D continuum model. A comparison of responses
of the 3D continuum model and the discrete model was presented in this chapter. It was shown
that there is a significant difference in some aspects of the response which is due to a difference in
the representation of SSI effects in these two models.
7.2 Concluding Remarks
This thesis focused on studying various aspects of Soil-Structure Interaction applicable to

reinforced Concrete Bridges. Direct Method was employed to perform numerical simulation of
soil+structure system using both Discrete and Continuum representations of the SSI system.
Meloland Road Overpassing (MRO) was used as a case study in various sections of the thesis.
This study was carried out in three phases which address various aspects of analysis and seismic
performance assessment of RC Bridges considering SSI effects.
In the first phase of study, 3D continuum representation of a soil+bridge system was used
to investigate the kinematic aspect of soil-structure interaction effect on an integral abutment
bridge. Comparison of bridge abutment footing and pier pile cap foundation motions with the free-
field motion showed zones with amplification of the excitation in the lower frequency regime and
208
zones that exhibit reduction in the higher frequency regime. It was shown that the observed
amplification takes place in the low frequency range that includes the first few natural frequencies
of the soil+bridge system. As a result, use of the free-field spectra may lead to an unsafe design.
The Tau-averaging method was originally developed for shallow slab foundations. It was shown
that this method is not capable of capturing the amplification phenomenon in the short frequency
ranges as it only focuses on slab averaging aspect and neglects other kinematic SSI effects (e.g.
embedment effect, wave scattering and vertically propagated shear waves). These other effects can
lead to amplification of the foundation motion compared to the free-field. In addition, the results
of the transfer function amplitude model developed by Elsabee and Morray is compared with the
calculated average amplitude of transmissibility function for the pier and abutment foundation
using the finite element model. As Elsabee and Morray’s model also predicts a reduction over the
entire frequency region, it is concluded that this model is not capable of prediction of variation of
the bridge pile supported foundation motion from the free field motion accurately.
In the second phase of the study, IDA was performed on four discrete archetype models
using 22 selected ground motions adopted from PEER NGA West 2 database. Outcome of the IDA
simulations were studied and compared to study the effect of structural system details as well as
ground motion characteristics. Fragility curves for all the archetype models were calculated. The
case studied here showed that the choice of SSI features representation in the archetype models
plays a significant role in both component-level and global structural collapse predictions. It also
highlights the uncertainty around the SSI features representation and the need to consider various
archetypes in collapse assessment of bridges.
The sequence of failure modes in different models was also studied. It was shown that both
model details and ground motion characteristics contribute to sequence of failure and final collapse
mode in the discrete archetype model studied in this phase. In some ground motion cases, ground
motion characteristics appear to be the dominant factor in response of the archetype models. As a
result, similar failure sequence and collapse modes were observed in all archetype models. In some
other cases, details of the archetype model, including the SSI representation, appeared to dominate
the response and failure modes. In some earthquake events, archetype models experienced a
sequence of multiple modes of failure before collapse and the structure was able to utilize its
ductility and energy absorption. Given the variability observed in this study and effect of both
209
model details and earthquakes dynamic characteristics, a statistical approach needs to be employed
to analyze the data. Special attention needs to be paid in defining the details of structural model,
including SSI representation and performing the analysis on a wide number of ground motions
which would allow for meaningful statistical analysis of results.
In this thesis, a Simplified Bridge Seismic Performance Evaluation Procedure (SBSPEP)
was introduced for RC bridges. This evaluation process which was developed based on FEMA
P695 is applicable to both design and performance assessment of RC bridges and can include SSI
effects.
The proposed methodology provides flexibility in considering soil-structure effects.
However, it is highly recommended to consider the SSI effects in the archetype models when SSI
effects are a key contributor to response of the system, e.g. bridges built on soft soils. In addition,
the methodology includes calculation of collapse margin ratio (CMR) and adjusted collapse
margin ratio (ACMR) by incorporating spectral shape effects. This methodology considers
modeling uncertainty in calculation of acceptable collapse margin ratio. Comparison of the
calculated ACMR with acceptable ACMR values provides a means of assessment for seismic
performance evaluation of RC bridges. As a case study, this evaluation process was employed to
assess the performance of the MRO bridge models. Analyses show that the bridge models did not
satisfy the requirements of the proposed methodology.
In the third phase of this study, a comparison was made between the seismic responses
calculated using discrete and continuum simulations of the MRO. It was observed that there is a
significant difference between the results obtained using the continuum and discrete approaches.
The continuum model consistently showed higher drift envelopes for both pier column and
abutment backwall when compared to the discrete model. The continuum model also showed
higher pseudo spectral acceleration specially in the higher frequency region where the natural
frequency of soil+structure resides. The absence of representation of the kinematic effects in the
discrete model and the difference between constitutive models and element types were identified
as major contributors to the difference observed between the two models. In this study, the
variation of ground motion with depth was not considered in the discrete models. In the discrete
models the same ground motion excitation is applied along the depth to all soil springs and
dashpots due to two main reasons:1) Determining ground motions at different depths requires
210
using a continuum model developed for capturing depth effect, 2) Applying specific ground
motion excitation for each depth in the discreet models is impractical, especially when performing
incremental dynamic analysis.
A comparison of the base shear of the two models for all ground motions was also
presented. It was shown that in most of the cases, the differences between the base shear responses
of continuum and discrete models are ground motion-specific where characteristics of ground
motions play a key role in the predicted base shears.
Compared to continuum approach, discrete models offer a simpler representation of the
soil+structure which makes them more convenient to use in the engineering practice. In discrete
models, effect of embankments and supporting soil is represented by a set of springs and dashpots.
The simplified representation of discrete approach has the advantage of shorter computation time.
Discrete models are unable to capture some of the key phenomena, such as effect of pore-
water pressure and radiation damping. Despite of its popularity among practitioners, the seismic
responses calculated using this type of models can be significantly different from the
corresponding results calculated using the higher fidelity continuum models as shown in Chapter
6 due to the simplifications and limitations of this approach.
On the other hand, continuum models offers a more realistic representation of system
features including soil nonlinearities, damage in structural components, pore-water pressure,
energy dissipation, and radiational damping.
In continuum approach, nonlinearities in soil and structural components can be simulated
using advanced constitutive models such as plasticity and damage models. The radiational
damping can be addressed using non-reflecting boundary conditions (modeled using infinite
element walls) to represent the far-field regions and to provide “quiet” boundaries to the FE
models. Given the level of complexity, using continuum models require an advanced knowledge
of the finite element analysis and material constitutive models. Computational cost of continuum
models is also significantly higher than the discrete representations.
As discussed, despite the complexity and computational cost of the continuum method, this
method has a greater capability to predict SSI features that either cannot be directly or accurately
simulated in the discrete models (including kinematic SSI effect). Detailed simulation of nonlinear
behavior of soil and formation and propagation of damage in structural components are among the
211
benefits of the continuum modeling approach. Advancements in computational modeling and
development of powerful computer clusters have made the use of continuum models feasible.
Analytical SSI modeling requires accurate representation of behavior of soil and structural
components going thorough nonlinear regime. Successful modeling requires high fidelity
constitutive models that can predict response of soil layers and structural components undergoing
extreme nonlinear conditions of damage and collapse.
7.3 Statement of Contributions
The followings are the original contributions in the research presented in this thesis:
a) Chapter 3 investigated the variation of foundation motion from free-field. It was shown
that for the studied case, there is an amplification of bridge foundation motion comparing
to free-field motion in the low frequency regime. As a result, ignoring foundation motion
amplification and using free-field motion in the analyses could lead to an unsafe seismic
response evaluation of the bridge structures built on the pile supported foundations. It was
also shown that simplified methods of calculating foundation motion such as tau-averaging
method is incapable of predicting amplifications in foundation motion.
b) In Chapter 4 and 5, it was shown that consideration of SSI has an impact in probability of
collapse of bridge structures like the MRO, which are built on the soft soil. In addition, it
was found that SSI effects has an important role in the predicted failure modes of the
seismic force resisting systems and redistribution of loads among the main structural
components. A simplified P695-based procedure was proposed for the seismic
performance assessment of the RC bridges to fill the gap that currently exists in seismic
evaluation procedure for bridges in the codes. The proposed methodology can be used for
the seismic performance assessment of RC bridges both in the design stage and in the
performance assessment and retrofitting of existing RC bridge structures.
c) In Chapter 6, it was shown that there is a significant difference between the abutment
backwall, and pier column’s nonlinear responses calculated using the discrete and the
continuum models. It was shown that, if a 3D continuum configuration is modeled
properly, it has a great potential capability to predict SSI features that either can’t be
212
directly or accurately simulated in the discrete models such as representing nonlinearity in
the soil and damages in the structural components.
7.4 Limitations, Recommendations and Future Work
As this research aimed only at damage assessment, loss assessment was not addressed in this study.
The results of this study were limited to the soil sub-layers, bridge type (RC Bridges), and the type
of earthquakes employed in the analyses. A similar study can be performed for other types of
bridges considering different soil classes or/and different hazards to investigate the role of SSI in
the seismic response of bridges in future.
There is an immediate need to better address the SSI effects in code provisions. This
includes guidelines for design and seismic performance assessment considering SSI, particularly
for structures built on soft soils.
Currently there is lack of an accurate method to estimate bridge pile foundation motion
from free-field motion. Considering different soil+bridge case studies, an approach needs to be
developed to quantify variation of bridge pile foundation motion from the free-field motion. Given
the amplification observed in the low frequency regime, there is a need for a model that can predict
the variation of motion.
In this research, the record ground motions were employed as the input motion to the base
of the finite element models. This is an inherent limitation of the numerical modeling due to lack
of record motion data at the bedrock. This might lead to alteration of the both free-field and
foundation responses.
The ground motions spectral shape effects play an important role in assessing collapse
performance of a bridge system. A simplified method needs to be developed to account for the
ground motions’ spectral shape effects considering prevalent hazards in different seismic regions
in Canada, especially for the West Coast.
In this study, value of total system collapse uncertainty (βTOT) is adopted from FEMA-P695
in performing bridge seismic performance evaluation. This value was originally developed for the
building system. A survey-based study to explicitly develop such values for bridge system is
required for different model qualities and period-based ductility.
213
In this research, a modified Mohr-Coulomb constitutive model was used as soil failure
criterion to simulate nonlinearity in soil components. More comprehensive continuum modeling
techniques using advanced soil constitutive models can be developed to further investigate
nonlinearity in soil components in presence of the structure. In addition, there is a need of
constitutive models in continuum modeling approach that can accurately capture cyclic response
of structural components.
Analyses of the 3D continuum model of the MRO bridge showed that there is an excessive
plastic deformation of soil around the piles, when the model was subjected to strong motions. The
relationship between the plastic deformation in the supporting soil layers of the 3D bridge
configurations and soil liquefaction phenomenon needs to be investigated.
Vertical component of the strong ground motions can affect seismic response of a
soil+bridge system. The significance of response of such system to vertical ground motions needs
to be investigated.
214
Bibliography
AASHTO. (2017). Guide Specifications for LRFD Seismic Bridge Design-8th. Edition. American
Association of State Highway and Transportation Officials, Washington D.C., USA.
Abrahamson, N. A., and Silva, W. J. (1997). "Empirical response spectral attenuation relations for
shallow crustal earthquakes. " Seismological Research Letters, 68(1):94-127.
Adebar, P. (2015). CIVL 513 Concrete Structures’ Lecture Note on Tension Stiffenning. The
University of British Columbia (UBC)-Vancouver.
Adams, J., Allen, T., Halchuk, S., Kolaj, M. (2019). "Canada's 6th Generation Seismic Hazard
Model, as Prepared for the 2020 National Building Code of Canada." 12CCEE. Québec
City, Québec, Canada.
Allam, S. M., Shoukry, M. S., Rashad, G. E., and Hassan, A. S. (2013). "Evaluation of tension
stiffening effect on the crack width calculation of flexural RC members." Alexandria
Engineering Journal, 52(2), 163–173.
Alih, S., & Khelil, A. (2012). "Tension Stiffening Parameter in Composite Concrete Reinforced
with Inoxydable Steel." Laboratory and Finite Element Analysis, World Academy of
Science, Engineering and Technology, 62, 535-540
Allotey, N. and El Naggar, M. (2008). "Generalized dynamic winkler model for nonlinear soil-
structure interaction analysis. " Canadian Geotechnical Journal, 45(4), 560–573.
Anand, V., & Satish Kumar, S. R. (2018). "Seismic Soil-structure Interaction: A State-of-the-Art
Review. " Structures, 16, 317–326.
API. (2007). "Recommended practice for planning, designing, and constructing fixed offshore
platforms." American Petroleum Institute. Section 6.8 Soil Reaction for Laterally Loaded
Piles.
ARTeMIS Modal. (2010). "Structural Vibration Solutions, Inc., " NOVI Science Park, Aalborg,
Denmark.
ASCE/SEI 7. (2010). Minimum Design Loads for Buildings and Other Structures. American
Society of Civil Engineers, Reston, Virginia.
ASCE/SEI 7. (2016). Minimum Design Loads for Buildings and Other Structures. American
Society of Civil Engineers, Reston, Virginia.
215
Aschheim, M. A.(1994). Mission Gothic Bridge, Image-NR917. Retrived October 08, 2020, from
National Earthquake Engineering Information Resources (NISEE) e-Library, University of
California, Berkeley. URL: https://nisee.berkeley.edu/elibrary.
Ashkani Zadeh, K. (2013). "Seismic Analysis of the RC integral Abutment Bridges Using
Performance Based Design Approach including Soil Structure Interaction." MASc Thesis
The University of British Columbia (UBC), Vancouver, BC, Canada.
Ashkani Zadeh, K., Ventura. (2015). "Effect of Soil-Structure Interaction on Performace Based
Design of Integral Abutment Bridges." 11CCEE. Victoria, BC, Canada.
Ashkani Zadeh K., Ventura C.E. (2017)." Investigating the Effect of Soil-Structure Interaction on
Collapse of Integral Abutment Bridges." 16WCEE. Santiago, Chile.
Ashkani Zadeh, K., Ventura, C. E., & Finn, W. D. L. (2018). "A Modified Tau-Average Method
to Evaluate The Kinematic Effect of SSI for Integral Abutment Bridges." 11NCEE. Los
Angeles. USA.
Ashkani Zadeh, K., Ventura, C. E., & Finn, W. D. L. (2019). "Effect of Kinematic Soil-Structure
Interaction on the Foundation Motion of Integral Abutment Bridges." 12CCEE. Québec
City, Québec, Canada.
Ashkani Zadeh, K., Ventura, C. E., & Finn, W. D. L. (2020). "Collapse Assessment of RC Bridges
Considering Soil Structure Interaction (SSI)." 17WCEE. Sendai, Japan.
Ashkani Zadeh, K., Ventura, C. E., & Finn, W. D. L. (2020). "Evaluation of Seismic Response of
Soil+Bridge System Using 3D Discrete and Continuum Simulation Approaches."
17WCEE. Sendai, Japan.
ATC-3. (1978). Tentative provisions for the development of seismic regulations for buildings.
ATC 3-06. California: Applied Technology Council.
Aviram, A., Mackie, K. R., and Stojadinović, B. (2008). Guidelines for Nonlinear Analysis of
Bridge Structures in California. Sacramento, Pacific Earthquake Engineering Research
Center (PEER), Sacramento, California, USA.
Aydemir, M.E. (2013). "Soil structure interaction effects on structural parameters for stiffness
degrading systems built on soft soil sites." Struct Eng Mech 2013, 45(5):655–76.
216
Badry, P., Satyam, N. (2017). "Seismic soil structure interaction analysis for asymmetrical
buildings supported on piled raft for the 2015 Nepal earthquake." Journal of Asian Earth
Science. 133:102–13.
Baker, J. W. (2007). "Measuring Bias in Structural Response Caused by Ground Motion." 8th
Pacific Conference on Earthquake Engineering (8PCEE 07), Singapore,
Baker, J. W. (2011). "Conditional Mean Spectrum: Tool for Ground-Motion Selection." Journal
of Structural Engineering, 137(3), 322–331.
Baker, J., Bronder, J., Burks, L., Ceferino, L., Chandramohan, R., Gupta, A., Hua, Y., Jayaram,
N., Lallemant, D., Lin, T., Loth, C., Markhvida, M., Yamamoto, Y. (2019). "Baker
Research Group. Ground motion model functions. Ground motion model functions. Crustal
earthquakes in active seismic regions." Accessed between 2014-2019.
https://web.stanford.edu/~bakerjw/GMPEs.html
Baker, J. W., & Cornell, C. A. (2008). "Vector-valued Intensity Measures Incorporating Spectral
Shape For Prediction of Structural Response." Journal of Earthquake Engineering, 12(4),
534–554.
Baker, J.W., and Cornell, C. A. (2006). "Spectral Shape, Epsilon and Record Selection."
Earthquake Engineering and Structural Dynamics, 35(9), 1077-1095.
Baker, J.W. (2015). "Efficient Analytical Fragility Function Fitting Using Dynamic Structural
Analysis." Earthquake Spectra, 31(1), 579-599.
Barbosa, A., Mason, B., and Romeny, K. (2014). SSI-Bridge: Soil-Bridge Interaction during Long-
Duration Earthquake Motions, Corvallis, Pacific Northwest Transportation Consortium
(Pac Trans).
Bažant, Z. P., Cedolin, L., & Bazant, Z. P. (1991). Stability of Structures: Elastic, Inelastic,
Fracture and Damage Theories. Oxford University Press.
Bažant, Z. P., & Oh, B. H. (1983). "Crack band theory for fracture of concrete." Materials and
Structures, 16(3), 155–177.
Bazant, Z. P. (1976). "Instability, ductility, and size effect in strain-softening concrete." ASCE
Journal of the Engineering Mechanics Division, 102(2), 331-344.
Bazant, Z. P., and Cedolin, L. (1983). "Finite element modeling of crack band propagation."
Journal of Structural Engineering, 109(1), 69-92.
217
Bazant, Z. P., and Cedolin, L. (1980). "Fracture mechanics of reinforced concrete." ASCE Journal
of the Engineering Mechanics Division, 106(6), 1287-1306.
BC MoTI. (2016). Volume1-Supplement to CHBDC S6-14, BC Ministry of Transportation and
Infrastructure, Vancouver, BC, Canada.
Bentz, E., and Collins, M.P. (2001). "Response-2000 user manual Version 1.1." University of
Toronto, Ontario, Canada.
Boore, D.M., and Atkinson, G.M. (2008). "Ground-motion prediction equations for the average
horizontal component of PGA, PGV, and 5%-damped PSA at spectral periods between
0:01 s and 10:0 s." Earthquake Spectra, 24(1):99-138.
Boore, D. M., Joyner, W. B., and Fumal, T. E. (1997). "Equations for estimating horizontal
response spectra and peak acceleration from western North American earthquakes: A
summary of recent work." Seismological Research Letters, 68(1):128-153.
Bycroft, G.N. (1956). "Forced vibration of a rigid circular plate on a semi-infinite elastic space
and on an elastic stratum." Philosophical Transactions of the Royal Society of London, 248
(Series A), 327–368.
Caltrans. (1969). Meloland Road Overcrossing (MRO) as built drawings, Bridge Inspection
Records Information System (BIRIS) Program, California Department of Transportation
Division of Maintenance, Sacramento, California, USA.
Caltrans. (1989). Bridge design aids 14-1, California Department of Transportation. Sacramento,
California, USA.
Caltrans. (2013). Seismic Design Criteria Version (SDC) Version 1.7, California Department of
Transportation, Sacramento, California, USA.
Caltrans. (2019a). Seismic Design Criteria (SDC) Version 2.0, California Department of
Transportation, Sacramento, California.
Caltrans. (2019b). "Caltrans ARS Online Web-based Tool Version 2.3.09." California Department
of Transportation, Sacramento, California, USA. Accessed during May 2019.
http://dap3.dot.ca.gov/ARS_Online/index.php
Campbell, K.W. (1997). "Empirical near-source attenuation relationships for horizontal and
vertical components of peak ground acceleration, peak ground velocity, and pseudo-
absolute acceleration response spectra." Seismological Research Letters, 68(1):154-179.
218
Campbell, K.W., and Bozorgnia, Y. (2008a). "Empirical ground motion model for shallow crustal
earthquakes in active tectonic environments developed for the NGA project." In
Proceedings of the 14WCEE, Beijing, China.
Campbell, K.W., and Bozorgnia, Y. (2008b). "NGA ground motion model for the geometric mean
horizontal component of PGA, PGV, PGD and 5% damped linear elastic response spectra
for periods ranging from 0:01 to 10 s." Earthquake Spectra, 24(1):139-171.
Carvajal Uribe, J.C. (2011). "Seismic Embankment-Abutment-Structure Interaction Of Integral
Abutment Bridges." PhD Thesis, The University of British Columbia (UBC), Vancouver,
BC, Canada.
Carreira, D. J., & Chu, K.-H. (1986). "Stress-Strain Relationship for Reinforced Concrete in
Tension." ACI Journal Proceedings, 83(1), 21–28.
Chang, C. Y., Power, M. S., Tseng, W. S., & Tang, Y. K. (1989). "Variations of earthquake ground
motions with depth and its effect on soil-structure interaction." Second Department of
Energy Natural Phenomena Hazards Mitigation Conference, California, USA.
Chiou, B. S.-J., and Youngs, R. R. (2008). "An NGA model for the average horizontal component
of peak ground motion and response spectra." Earthquake Spectra, 24(1):173-215.
Chopra, K.A. (2016). Dynamics of Structures - Theory and Applications to Earthquake
Engineering. 5th. Edition, Pearson, Berkeley, California, USA.
Choi, J., Brandenberg, S. J. and Kim, M. (2013). "Modeling the dynamic behavior of a single pile
in dry sand using a new p-y material model." UCLA Previously Published Works,
University of California, Los Angles, California, USA.
Ciampoli, M., Pinto, P.E. (1995). "Effects of soil-structure interaction on inelastic seismic
response of bridge piers." J Struct Eng, 121(5), 806–14.
Clough, R.W., Penzien, J. (1995). Dynamics of Structures, Berkeley, Computers & Structures,
Inc.USA.
Correia, A. A., Pecker, A., Kramer, S. L., & Pinho, R. (2012). "Nonlinear pile-head macro-element
model: SSI effects on the seismic response of a monoshaft-supported bridge." In
Proceedings, 15WCEE, Lisbon, Portugal.
Cornell, C. A. (1969). "A probability-based structural code." ACI Journal Proceedings, 66(12),
974-985.
219
Cornell, C. A., and Krawinkler, H. (2000). "Progress and Challenges in Seismic Performance
Assessment." PEER Center News, 3(2), 1-3.
CSA. (2019). S6-19 Canadian Highway Bridge Design Code. CSA Group. Toronto, Ontario,
Canada.
Canada.
Canada.
Deierlein, G. G., Liel, A. B., Haselton, C. B., & Kircher, C. A. (2008). "ATC 63 Methodology for
Evaluating Seismic Collapse Safety of Archetype Buildings." Structures Congress, 1–10.
Desai, C. S., Zaman, M. (2014). Advanced geotechnical engineering : soil-structure interaction
using computer and material models, CRC Press. Boca Raton, Florida, USA.
Douglas, B. M., Norris, G., Dodd, L., and Richardson, J. (1984). Behavior of the Meloland Road
Overcrossing during the 1979 Imperial Valley Earthquake, Seismic Research for Highway
Bridges (U.S.-Japan Program), Department of Civil Engineering, University of Pittsburgh,
Pittsburgh PA, USA.
Douglas, B. M., Maragak.is, E. A., Vrontinos, S., and Douglas, B. J. (1990). "Analytical Studies
of the Static and Dynamic response of the Meloland Road Overcrossing", Proceedings of
the 4th U. S. National Conference on Earthquake Engineering, Vol. 1, pp 987-992. Palm
Springs CA, USA.
Douglas, B. M., E. Maragakis, and S. Vrontinos (1991). "Parameter identification studies of the
Meloland Road Overcrossing." In Proc. Pacific Conference on Earthquake Engineering,
1,105–16. Wellington, New Zealand.
Douglas, B. M . Crouse, C. B., Werner, S. D., and Maragak.is, E. A. (1993). Quick Release
Dynamic Tests of the Meloland Road Overcrossing. California Department of
Transportation, Sacramento CA, Center for Civil Engineering Research, University of
Nevada. Reno NV, USA.
EC8-2. (2005). Eurocode8. Design of Structures for Earthquake Resistance-Part2: Bridges.
European Committee For Standardization. English Version.
220
Edip, K., Garevski, M., Sheshov, V., & Bojadjieva, J. (2017). "Boundary Effects in Simulation of
Soil-Structure Interaction Problems." Soil Mechanics and Foundation Engineering, 54(4),
239–243.
Elgamal, A. (2010). "Calibrated 3D Computational Modeling of Soil-Structure Systems and
Liquefaction Scenarios." Fifth International Conference on Recent Advances in
Geotechnical Earthquake Engineering and Soil Dynamics, the Missouri University of
Science and Technology. Columbia, Missouri, USA.
Elgamal, A., Yang, Z., Parra, E., & Ragheb, A. (2003). "Modeling of cyclic mobility in saturated
cohesionless soils." International Journal of Plasticity, 19(6), 883–905.
Elgamal, A., Jun, L., Yang, Y. Z. and Conte, J. P. (2008). "Three-dimensional seismic response of
Humboldt Bay bridge-foundation-ground system," Journal of Structural Engineering
134(7), 1165–1176.
Elgamal, A., Lu, J. and Yang, Z. (2005). "Liquefaction-induced settlement of shallow foundations
and remediation: 3D numerical simulation." Journal of Earthquake Engineering 9(1), 17–
45.
Elgamal, A., Yang, Z., Lai, T., Kutter, B. and Wilson, D. (2005). "Dynamic response of saturated
dense sand in laminated centrifuge container." ASCE Journal of Geotechnical and
Geoenvironmental Engineering, 131(5), 598–609.
Elgamal, A., Yang, Z. and Parra, E. (2002). "Computational modeling of cyclic mobility and post-
liquefaction site response." Soil Dynamics and Earthquake Engineering, 22, 259–271.
Elnashai, A. S., Papanikolaou, V., and Lee, D. (2002). "Zeus NL-A system for inelastic analysis
of structures." Mid-America Earthquake Center, University of Illinois at Urbana-
Champaign, Program Release Sept. 2002.
Elsabee, F., Morray, J., & Roesset, J. (1977). Dynamic behavior of embedded foundations,
Massachusetts Institute of Technology, Cambridge, Massachusetts, USA.
EN 1998-5. (2001). Eurocode 8: design of structures for earthquake resistance - part 5:
foundations, retaining structures and geotechnical aspects. EN 1998-5. Brussels: European
Committee for Standardization.
ENSOFT Inc. (2013). "LPILE v7.0," http://www.ensoftinc.com.
221
FEMA-440. (2005). Improvement of Nonlinear Static Seismic Procedures. ATC-55. Federal
Emergency Management Agency, Washington D.C., USA.
FEMA-P695. (2009). Quantification of Building Seismic Performance Factors, Federal
Emergency Management Agency, Washington, D.C., USA.
Feng, M. Q., Lee, S. C., (2009). Determining the effective system damping of highway bridges,
Department of Civil & Environmental Engineering, University of California, Irvin, USA.
Federal Highway Administration (FHWA). (2006). Seismic Retrofitting Manual for Highway
Structures: Part1-Bridges. Rep. No. FHWA-HRT-06-032, Federal Highway
Administration, US. Department of Transportation, McLean, Virginia. USA.
Finn, W. L. (2005). "A study of piles during earthquakes: issues of design and analysis," Bulletin
of Earthquake Engineering." 3(2), 141–234.
Gates, J. H. (1993). "Dynamic field response studies and earthquake instrumentation of the
Meloland Road Overcrossing. Structural Engineering in Natural Hazards Mitigation."
Proc. of papers presented at the Structures Congress 93, ASCE. 1, 343–48.
Gazetas, G. (1991). Foundation vibrations. in H. Fang, ed., Foundation Engineering Handbook.
2nd edn, Van Nostrand Reinholds, ch 15, 553–593.
Gazetas, G., Mylonakis, G. (2001). "Soil-structure interaction effects on elastic and inelastic
structures." 4th International Conference on Recent Advances in Geotechnical Earthquake
Engineering and Soil Dynamics, San Diego, USA.
Gazioglu, S. and O’Neill, M. W. (1984). "Evaluation of p-y relationships in cohesive soils." in
Proceedings of the ASCE Symposium on Analysis and Design of Pile Foundations pp. 192–
213, ASCE National Convention, San Francisco, California, USA.
Ghalibafian, H., Ventura, C.E., and Foschi, R.O. (2008). "Effects of nonlinear soil-structure
interaction on the inelastic seismic demand of pile-supported bridge piers." 14th World
Conference on Earthquake Engineering, Beijing, China.
Ghobarah, A. (2001). "Performance-Based Design in Earthquake Engineering: State of
Development," Engineering Structures, 23, 878-884.
Gičev, V., Trifunac, M. D., & Orbović, N. (2016). "Two-dimensional translation, rocking, and
waves in a building during soil-structure interaction excited by a plane earthquake P-wave
pulse." Soil Dynamics and Earthquake Engineering, 90, 454–466.
222
Grote, M.J. (2000). "Nonreflecting Boundary Condition for Time Dependant Wave Propagation.
Zurich." Seminar Fur Angevandate Mathematik, Eidgenossische Technische Hochschule,
Zurich, Switzerland.
Harmsen, S. C., Frankel, A. D., and Petersen, M. D. (2002). "Deaggregation of U.S. seismic
hazard." The 2002 update. USGS Open-File Rep. 03-440.
〈http://pubs.usgs.gov/of/2003/ofr-03-440/ofr-03-440.html〉.
Haselton, C.B., and Baker, J.W., (2006). "Ground motion intensity measures for collapse capacity
prediction: Choice of optimal spectral period and effect of spectral shape." Proceedings
8th National Conference on Earthquake Engineering, San Francisco, California, USA.
Haselton, C. B., Baker, J. W., Liel, A. B., & Deierlein, G. G. (2011). "Accounting for Ground-
Motion Spectral Shape Characteristics in Structural Collapse Assessment through an
Adjustment for Epsilon." Journal of Structural Engineering, 137(3), 332–344.
Hognestad, E. (1951). A Study of combined bending and axial load in reinforced concrete
members, University of Illinois Engineering Experiment Station, Urbana and Champaign,
Illinois, USA.
Ibarra, L. F., and Krawinkler, H. (2005). Global collapse of frame structures under seismic
excitations, John A. Blume Earthquake Engineering Center, Stanford, CA, USA.
Jarernprasert S, Bazan-Zurita E, Bielak J. (2013). "Seismic soil-structure interaction response of
inelastic structures." Soil Dynamics Earthquake Eng, 47, 132–43.
Jeremic, B., Jie, G., Preisig, M. and Tafazzoli, N. (2009). "Time domain simulation of soil-
foundation-structure interaction in non-uniform soils." Earthquake Enginearing and
Structural Dynamics, 38(5), 699–718.
JSCE 15. (2007). Standard specifications for concrete structures-design. Society of Civil
Engineers, Tokyo, Japan.
Kausel, E. (1974). Forced vibrations of circular foundations on layere dmedia, MIT Research
Report R74-11, Soils Publication No.336, Department of Civil Engineering, Massachusetts
Institute of Technology, Cambridge, USA.
Kausel, E.,Whitman, R.V., Morray, J.P., and Elsabee,F. (1978). "The spring method for embedded
foundations." Nuclear Engineering and Design, 48, 377–392.
223
Kausel, E. (2010). "Early history of soil–structure interaction." Soil Dynamics and Earthquake
Engineering, 30(9), 822–832.
Kim, J.H, and Mander, J.B. (1999). Truss Modeling of Reinforced Concrete Shear-Flexure
Behavior, Technical Report MCEER-99-0005, Multidisciplinary Center for
EarthquakeEngineering Research, University at Buffalo, Buffalo, New York, USA.
Kim, S. , and Stewart, J.P. (2003). "Kinematic soil-structure interaction from strong motion
recordings." J. Geotech . & Geoenv. Engrg., 129(4), 323-335.
Kramer, S. (1996). Geotechnical Earthquake Engineering. Prentice Hall, Upper Saddle River,
New Jersey, USA.
Kwon, O. S., & Elnashai, A. S. (2008). "Seismic analysis of Meloland Road Overcrossing using
multiplatform simulation software including SSI." Journal of Structural Engineering,
134(4), 651–660.
Lam, I. P., Law, H. K. and Martin, G. R. (2007). Bridge foundations: Modeling large pile groups
and caissons for seismic design, Technical Report MCEER 07-0018, US Multidisciplinary
Center for Earthquake Engineering Research. University at Buffalo, Buffalo, New York,
USA.
Lee, S., Feng, M. Q., Kwon, S.-J. and Hong, S.-H. (2011). "Equivalent modal damping of short-
span bridges subjected to strong motion." Journal of Bridge Engineering 16(2), 316–323.
Lemnitzer, A., Ahlberg, E. R., Nigbor, R. L., Shamsabadi, A., Wallace, J. W., and Stewart, J. P.
(2009). "Lateral performance of full-scale bridge abutment wall with granular backfill." J.
Geotech. Geoenviron. Eng., 135(4), 506–514.
Lu, J., Elgamal, A., Linjun, Y., Kincho, H. L. and Conte, J. P. (2011). "Large-scale numerical
modeling in geotechnical earthquake engineering." International Journal of
Geomechanics, 11(6), 490–503.
Luque, R., Bray, J. D. (2020). "Dynamic Soil-Structure Interaction Analyses of Two Important
Structures Affected by Liquefaction during the Canterbury Earthquake Sequence." Soil
Dynamics and Earthquake Engineering, 133(2020), 106026.
Lysmer, J. and R.L. Kuhlemeyer. (1969). "Finite dynamic model for infinite media." Journal of
the Engineering Mechnanics Division, Proc. ASCE, 95, 859 – 876.
224
Makris, N., Badoni, D., Delis, E. and Gazetas, G. (1994). "Prediction of observed bridge response
with soil-pile-structure interaction." Journal of Structural Engineering, 120(10), 2922–
3011.
Mander, J., Priestley, M. and Park, R. (1988). "Theoretical stress-strain model for confined
concrete." Journal of Structural Engineering, 114(8), 1804–1826.
Maragakis, E.A., Vrontinos, S., Douglas, B.M., Abdei-Ghaffar, S. (1991). "Soil-Structure
Interaction Effects on the Response of Meloland Bridge." Proceedings of the Second
International Conference on Recent Advances in Geotechnical Earthquake Engineering
and Soil Dynamics, Louis, Missouri, USA.
Markou, G., & Papadrakakis, M. (2012). "An efficient generation method of embedded
reinforcement in hexahedral elements for reinforced concrete simulations." Advances in
Engineering Software, 45(1), 175–187.
Martel, R.R. (1940). "Effect of foundation on earthquake motion." Civil Eng. Journal, 10 (1), 7–
10.
Massicotte, B., Elwi, A. E., & MacGregor, J. G. (1990). "Tension‐Stiffening Model for Planar
Reinforced Concrete Members." Journal of Structural Engineering, 116(11), 3039–3058.
McCallen, D. and Romstad, K. (1994). "Dynamic analyses of a skewed short-span, box-girder
overpass." Earthquake Spectra, 104, 729–756.
McKenna, F., and Fenves, G. L. (2001). The OpenSees command language manual, version 1.2,
Pacific Earthquake Engineering Research Center, Univ. of California at Berkeley,
Berkeley, California, USA.
Menglin, L., Wang, H., Chen, X., Zhai, Y. (2011). "Structure–soil–structure interaction: Literature
review." Soil Dynamics and Earthquake Engineering, 31(12), 1724–1731.
Moehle, J. P., and Deierlein, G. G. (2004). "A Framework Methodology for Performance-Based
Earthquake Engineering." 13th World Conference on Earthquake Engineering, Vancouver,
BC, Canada.
Miranda E, Bertero V. (1994). "Evaluation of strength reduction factors of earthquake-resistant
design." Earthq Spectra, 10(2), 357–79.
Mitchell, D. (2019). "CCEE2019 preconferance performance-based seismic design workshop."
Québec City, Québec, Canada.
225
Mitchell, D., Bruneau, M., Williams, M., Anderson, D., Saatcioglu, M., and Sexmith, R. (1995).
"Performance of bridges in the 1994 Northridge Earthquake." Canadian Journal of Civil
Engineering, 22(2), 415–427.
Murchison, J. and O’Neill, M. (1984). "Evaluation of p-y relationships in cohesionless soils." in
Proceedings of the ASCE Symposium on Analysis and Design of Pile Foundations pp. 174–
191, ASCE National Convention, San Francisco, California, USA.
Mylonakis, G., Nikolaou, A., and Gazetas, G. (1997). "Soil-Pile-Bridge eismic Interaction:
Kinematic and Inertial Effects." Earthquake Engineering and Structural Dynamics, 26,
337-359.
Mylonakis, G., & Gazetas, G. (2000). "Seismic Soil-Structure Interaction: Beneficial or
Detrimental?." Journal of Earthquake Engineering, 4(3), 277–301.
Mylonakis, G., Nikolaou, S. and Gazetas, G. (2006). "Footings under seismic loading: Analysis
and design issues with emphasis on bridge foundations," Journal of Soil Dyanmics and
Earthquake Engineering, 26, 824–853.
Nayal, R., & Rasheed, H. A. (2006). "Tension Stiffening Model for Concrete Beams Reinforced
with Steel and FRP Bars." Journal of Materials in Civil Engineering, 18(6), 831–841.
NEHRP. (2011). Selecting and Scaling Earthquake Ground Motions for Performing Response-
History Analyses, Rep. No. NIST GCR 11-917-15, National Institute of Standards and
Technology, U.S. Department of Commerce, Washington DC, USA.
NEHRP. (2013). Nonlinear Analysis Research and Development Program for Performance-Based
Seismic Engineering, Rep. No. NIST GCR 14-917-27, National Institute of Standards and
Technology, U.S. Department of Commerce, Washington DC, USA.
Newmark, N.M., and Hall, W.J. (1973). Seismic Design Criteria for nuclear facilities. Report No.
46, Building Practices for Disaster Mitigation, National Bureau of Standards, pp. 209-236,
U.S. Department of Commerce, Washington DC, USA.
Newmark, M., Hall, W. J. and Morgan, J. R. (1977). "Comparison of building response and free
field motion in earthquakes." Proceedings of 6th World Conference on Earthquake
Engineering, New Delhi, India
Nilson, Arthur H., "Internal Measurement of Bond Slip." ACI Journal, 69, 439-441.
226
NRCan. (2020). Natural Resources Canada (NRCan): Seismic zones in Western Canada.
Retrieved from the Natural Resources Canada website:
https://earthquakescanada.nrcan.gc.ca/zones/westcan-en.php, Date reterived: June 14,
2020.
NZS 1170.5. (2004). Structural design actions - part 5: earthquake actions, NZS 1170.5.
Wellington, New Zealand.
Ohuchi, H., Matsuda, T., Goto, Y. (1992). "A Study on Cypress Viaduct Collapse and Seismic
Performance of a Retrofitted Bent." Japan Society of Civil Engineers (Proc. of JSCE No.
446/1-19), Structural Eng. /Earthquake Eng., 9(1), 65-76.
PEER. (2019). "Open System for Earthquake Engineering Simulation."
http://opensees.berkeley.edu/ 2019).
PEER. (2019). "Pacific Earthquake Engineering Research Center (PEER), Next Generation
Attenuation Relationships for Western US (NGA-West2) Database." Accessed between
2014-2019. https://peer.berkeley.edu/research/nga-west-2.
Pandey, B. H. (2013). "Investigation of Variation of Motions Between Free Field And Foundation
in Seismic Soil-Structure Interaction of Structures with Rigid Shallow Foundation." PhD
Thesis. Vancouver, British Columbia, Canada: The University of British Columbia.
Vancouver, BC, Canada.
Priestley, M. J. N., Calvi, G. M., Kowalsky, M. J. (2007). Displacement-based seismic design of
structures. IUSS Press. Pavia, Italy.
Priestley, M. J. N., Calvi, G. M., Kowalsky, M. J., Powell, G. H. (2008). "Displacement-Based
Seismic Design of Structures." Earthquake Spectra, 24(2), 555–557.
Engineers and Geoscientists BC (EGBC). (2018). Professional Paractice Guidelines:
Performanced-Based Seismic Design of Bridges in BC Version 1.0, Engineers and
Geoscientists BC, Burnaby, BC, Canada.
Rahmani, A. (2014). "Three-dimensional nonlinear analysis of dynamic soil-pile-structure
interaction for bridge systems under earthquake shakings." PhD Thesis. British Columbia,
Canada: The University of British Columbia (UBC). Vancouver, BC, Canada.
227
Rahmani, A., Taiebat, M., and Finn, W.D.L. (2014). "Nonlinear dynamic analysis of Meloland
Road Overpass using three-dimensional continuum modeling approach," Soil Dynamics
and Earthquake Engineering, 57, 121-132.
Rahmani, A., Taiebat, M., Finn, W.D.L., Ventura, C.E. (2016). "Evaluation of substructuring
method for seismic soil-structure interaction analysis of bridges," Soil Dynamics and
Earthquake Engineering, 90, 112–127.
Rayhani, M. H. T. and El Naggar, M. H. (2008). "Numerical Modeling of Seismic Response of
Rigid Foundation on Soft Soil." International Journal Of Geomechanics, 336-346.
Romstad, K., Kutter, B., Maroney, B., Vanderbilt, E., Griggs, M., and Chai, Y. H. (1995).
Experimental measurements of bridge abutment behavior, Rep. No. UCD-STR-95-1, Dept.
of Civil and Environmental Engineering, Univ. of Calif., Davis, California, USA.
Roesset, J. M. (2013). "Soil Structure Interaction The Early Stages," Journal of Applied Science
and Engineering, 16(1), 1-8.
Scott, M.H., Fenves, G.L. (2006). "Plastic hinge integration methods for force-based beam–
column elements," ASCE Journal of Structural Engineering, 132(2), 244-252.
Seed, H.B.,Whitman, R.V.,and Lysmer, J. (1977). Soil–structure interaction effects in the design
of nuclear power plants. In:Hall,W.J. (Ed.), Structural and Geotechnical Mechanics
(Chapter13), a Volume Honoring Nathan M. Newmark. Prentice-Hall. Upper Saddle River,
New Jersey.
SeismoSoft (2020). "Earthquake Engineering Software Solution." Italy.
http://www.seismosoft.com/. Accessed between 2014-2020.
Sezawa, K. and Kanai, K. (1935). "Decay in the seismic vibration of a simple or tall structure by
dissipation of their energy into the ground." Bulletin of the Earthquake Research Institute
13, 681–696.
Shamsabadi, A. (2013). Seismi soil-foundation-structure-interaction (SFSI): practical bridge
seismic foundation specifications design and analysis. California Department of
Shamsabadi, A., Khalili-Tehran, P., Stewart, J. and Taciroglu, E. (2010). "Validated simulation
models for lateral response of bridge abutments with typical backfills." Journal of Bridge
Engineering, 15(3), 302–311.
228
Shamsabadi, A., Rollins, K. and Kapuskar, M. (2007). "Nonlinear soil-abutment-bridge structure
interaction for seismic performance-based design." Journal of Geotechnical and
Geoenvironmental Engineering, 133(4), 707–720.
Shamsabadi, A., and K. M. Rollins. (2014). "Three-dimensional nonlinear continuum seismic soil-
structure intreaction of skewed bridge abutments." 8th European Conf. on Numerical
Methods in Geotechnical Engineering, Delft, Netherlands.
Shamsabadi, A., Dasmeh, A., Nojoumi, A., Rollins, K. M., & Taciroglu, E. (2020). "Lateral
Capacity Model for Backfills Reacting against Skew-Angled Abutments under Seismic
Loading." Journal of Geotechnical and Geoenvironmental Engineering, 146(2), 1-14.
SIMULIA. (2018). "ABAQUS/Standard User's Guide, Version 6.14." Providence, RI: Dassault
Systèmes SIMULIA Corp.
Sluys, L. J. (1992). "Wave Propagation, Localization and Dispersion in Softening Solids." PhD
Thesis, Delft University of Technology, Delft, Netherlands.
Stefanidou, S.P., Sextos, A.G., Kotsoglou, A.N., Lesgidis, N. (2017). "Soil-structure interaction
effects in analysis of seismic fragility of bridges using an intensity-based ground motion
selection procedure." Eng Struct, 151, 366–80.
Steinbrugge, K. V. (1989). Cypress St. Viaduct, Image- S6130. Retrived October 08, 2020, from
National Earthquake Engineering Information Resources (NISEE) e-Library, University of
California, Berkeley. URL: https://nisee.berkeley.edu/elibrary.
Stevenson, J. (1980). "Structural damping values as a function of dynamic response stress and
deformation levels." Nuclear Engineering and Design, 60(2), 211–237.
Stewart, J. P., et al. (2007). Full scale cyclic testing of foundation support systems for highway
bridges. Part II: Abutment backwalls, Rep. No. UCLA-SGEL-2007/02, Structural and
Geotechnical Engineering. Laboratory, Univ. of Calif., Los Angeles, USA.
Stewart, J.P., Emel Seyhan, E. (2013). Semi-Empirical Nonlinear Site Amplification and its
Application in NEHRP Site Factors, Pacific Earthquake Engineering Research Center. The
University of California-Berkeley, California, USA.
Stewart, J.P., Seed, R.B., and Fenves,G.L. (1998). Empirical Evaluation of Inertial Soil-Structure
Interaction Effects, Pacific Earthquake Engineering Research Center, University of
California, Berkeley, USA.
229
Stewart, J. P., Seed, R. B., & Fenves, G. L. (1999). "Seismic Soil-Structure Interaction in
Buildings. II: Empirical Findings." Journal of Geotechnical and Geoenvironmental
Engineering, 125(1), 38–48.
The Geological Society. (2018). "Plate Tectonics. San Andreas Fault." Retrived in December
2018.
https://www.geolsoc.org.uk/Plate-Tectonics/Chap3-Plate-Margins/Conservative/San-
Andreas-Fault.
Tehrani, P., Mitchell, D. (2012). "Seismic Performance Assessment of Bridges in Montreal Using
Incremental Dynamic Analysis." 15WCEE, Lisbon, Portugal.
Tehrani, P. (2012). "Seismic Behaviour and Analysis of Continuous Reinforced Concrete
Bridges." PhD Thesis, McGill University, Montreal, Québec, Canada.
Thevaneyan, K.D. and John, P.F. (2011). "Modeling of Soil Structure Interaction of Integral
Abutment Bridges. World Academy of Sience." Engineering and Technology, 769-774.
Thorenfeldt, E., Tomaszewicz, A. and Jensen, J. J. (1987)."Mechanical Properties of High Strength
Concrete and Application to Design." Proceedings of the Symposium: Utilization of High-
Strength Concrete, Trondheim, Norway.
Todeschini, C. E., Bianchini, A. C., Kesler, C. E.(1964). "Behavior of concrete columns reinforced
with high strength steels." Journal of the American Concrete Institute, 61(6), 701-716.
Toppozada, T., & Branum, D. (2004). "California earthquake history." Annals of Geophysics,
47(2–3), 509–522.
Tsikada, A., Shimbun, Y. (1995). Collapse of Hanshin Expressway, Image- -K0013. Retrived
October 08, 2020, from National Earthquake Engineering Information Resources (NISEE)
e-Library, University of California, Berkeley. URL: https://nisee.berkeley.edu/elibrary.
Tuladhar R., Mutsuyoshi H., Maki T. (2008). "Seismic Behavior of Concrete Bridge Pier
Considering." 14WCEE, Beijing, China.
Tyapin, A. (2007). "The frequency-dependent elements in the code SASSI: A bridge between civil
engineers and the soil-structure interaction specialists." Nuclear Engineering and Design,
237, 1300–1306.
USGS. (2019). "Earthquake Hazard Programs: Unified Hazard Tool." US Geological Survey,
USA. Accessed between 2014-2019. https://earthquake.usgs.gov/hazards/interactive.
230
Vecchio, F. J., and Collins, M. P. (1986). "The Modified Compression Field Theory for Reinforced
Concrete Elements Subjected to Shear." ACI Journal, 83( 2), 219-231.
Veletsos, A. S. (1977). Dynamics of Structure-Foundation Systems. Structural 6; Geotechnical
Mech., Hall, W. J. ed. Prentice-Hall.
Veletsos, A. S., and C. E. Ventura. (1986). "Modal analysis of non-classically damped linear
systems." Earthquake Engineering and Structural Dynamics, 14, 217–43.
Veletsos, A.P. and Prasad, A.M. (1989). "Seismic interaction of structures and soils; stochastic
approach." J. Struct. Engrg., 115, 935-956.
Veletsos, A.P., Prasad, A.M. and Wu, W.H. (1997). "Transfer functions for rigid rectangular
foundations." J. Earthquake Engrg. & Struct. Dynamics, 26, 5-17.
Vamvatsikos, D., and Cornell, C. A. (2002). "Incremental Dynamic Analysis." Earthquake
Engineering and Structural Dynamics, 31, 491-514.
Ventura, C.E., Carvajal, J. C., Liam, Finn L., and Traber, J. (2011). "Ambient Vibration Testing
of the Meloland Road Overpass." 4th. International Operational Modal Analysis
Conference (IOMAC 2011). Istanbul, Turkey.
Ventura, C.E. (2012). CIVL 508 Course Notes. The University of British Columbia
(UBC),Vancouver, BC, Canada
Wahalathantri, B.L., Thambiratnam, D.P. , Chan, T.H.T. , Fawzia, S. (2011). "A material model
for flexural crack simulation in reinforced concrete elements using ABAQUS."
Proceedings of the First International Conference on Engineering, Designing and
Developing the Built Environment for Sustainable Wellbeing. Queensland University of
Technology, Queensland, Australia.
Werner, S. D., Beck, J. L., & Levine, M. B. (1987). "Seismic response evaluation of Meloland
Road Overpass using 1979 imperial Valley earthquake records." Earthquake Engineering
& Structural Dynamics, 15(2), 249–274.
Werner, S. D. and Crouse, C. B. and Katafygiotis, L. and Beck, J. L. (1993). Model Identification
and Seismic Analysis of Meloland Road Overcrossing, Report to California Department
of Transportation, California, USA.
231
Werner, S. (1994). "Study of Caltrans’ seismic evaluation procedure for short bridges."
Proceedings of the 3rd Annual Seismic Research Workshop, California Department of
Whitman, R. V., (1970). “Soil-Structure Interaction,” Seismic Design for Nuclear Power Plants,
Edited by Robert J. Hanson, The M.I.T. Press, Cambridge, Massachusetts, USA.
Wilson, J. C., and Tan, B. S. (1990). "Bridge Abutments: Formulation of Simple Model for
Earthquake Response Analysis." Journal of Engineering Mechanics, 116(8), 1828-1837.
Wight, J. K., Macgregor, J.G, (2012). Reinforced concrete : mechanics and design, 6th. Ed. Pearson
Education, Inc., Upper Saddle River, New Jersey, USA.
Wolf, O.J. (1985). Daynamic Soil-Structure Interaction. Englewood Cliffs. Prentice-Hall, Inc.
Yang, Z. and Elgamal, A. (2001). "Sand boils and liquefaction-induced lateral deformation." 15th
International Conference on Soil Mechanics and Geotechnical Engineering, Istanbul,
Turkey.
Yang, Z., Elgamal, A., and Parra, E. (2003). "Computational model for cyclic mobility and
associated shear deformation." Journal of Geotechnical and Geoenvironmental
Engineering, 129(12), 1119–1127.
Yang, Z., Lu, J. and Elgamal, A. (2008). OPENSEES soil models and solid-fluid fully coupled
elements (user’s manual), Department of Structural Engineering, University of California-
Sandiago (UCSD), Sandiago, USA.
Zafir, Z. (2002). "Seismic foundation stiffness for bridges.", Procccedings of the International
Deep Foundations Congress 2002, 1421–1437. Orlando, Florida, USA.
Zhang, J., and Makris, N. (2001). Seismic response analysis of highway overcrossings including
soil–structure interaction, Rep. No. PEER-01/02, Univ. of California, Berkeley, USA.
Zhang, J. and Makris, N. (2002). "Kinematic response functions and dynamic stiffness of bridge
embankments." Earthquake Engineering and Structural Dynamics, 31(11), 1933–1966.
Zhang, Y., Conte, J.P., Yang, Z., Elgamal, A., Bielak, J., Acero, G. (2008). "Two-dimensional
nonlinear earthquake response analysis of a bridge-foundation-ground system."
Earthquake Spectra, 24(2), 343-386.
Zienkiewicz, O. C., Emson, C., & Bettess, P. (1983). "A novel boundary infinite element."
International Journal for Numerical Methods in Engineering, 19(3), 393–404.
232
Appendices
Appendix A Meloland Road Overcrossing (MRO) Drawing Details
The Following MRO bridge drawings are the property of the California Department of
Transportation (CALTRANS) and have been generated by Bridge Records Information System
(BRIS).
A.1 MRO layout and Elevation View
(a)
(a)
(b)
(b)
Figure A- 1 Meloland Road Crossing (MRO) dimension detail (a) Lauout plan view, (b) Elvation view
(Caltrans, 1969)
233
A.2 MRO Deck and Girder Outline
(a)
(a)
(a) (a)
(b)
(b)
(b)
(b)
(c)
(c)
(c)
(c)Typical deck slab's section, (b) Girder layout (c) Deck
Figure A- 2 MRO Concrete outline and RC detail (a)
longitudinal section (Caltrans, 1969)
234
A.3 Deck and Pier RC Detail
(a) (b)
(a) (b)
(c)
(b) (c)
Figure A- 3 MRO concrete outline and RC detail (a) Sectional elevation at Pier, (b) Pier footing foundation
plan, and (c) Pier column section (Caltrans, 1969)
235
A.4 Abutment and Wingwalls Outline and RC detail
(a)
(a) (b)
(b)
(a) (b)
(c) (d) (e)

(c) (c) (d) (d) (e)(e)
Figure A- 4 MRO Abutment and Wing walls RC detail (a) Abutment plan, (b) Abutment section, (c) Wingwall
elevation, (d) Wingwall section, and (e) Wingwall corner detail (Caltrans, 1969)
236
Appendix B Time Histories of Ground Motions used for Analyses
B.1 Time History of Ground Motions are Used for Elastic Analyses of the Continuum
Model E to Investigate Variation of the Bridge Foundation Motion from the Free-Field
Motion (Phase 1 Study)
Figure B- 1 Ground motions used as input motions in response time history analyses of the elastic 3D continuum
model (E model) of the MRO
237
B.2 Time History of Ground Motions used for IDA Analyses of the Discrete Models
(Phase 2 Study)
Figure B- 2 Imperial Valley-06, Superstition Hills-02, Loma Prieta, Northridhe, Bam-Iran, Cape Mendocino,
Chi Chi-Taiwan, and Christchurch New Zealand time histories (PEER, 2018)
238
Figure B- 3 Duzce-Turkey, El Mayor-Cucapah Mexico, Gazli-USSR, Hector Mine, Imperial Valley-02, Irpinia
Italy-01, Kobe Japan, and Kocaeli turkey time histories (PEER, 2018)
239
Figure B- 4 Landers, Nahanni Canada, Parkfield, San Fernando, Spitak-Armenia and Tabas-Iran time
histories (PEER, 2018)
240
B.3 Time History of Ground Motions Used for Nonlinear Analyses of the Continuum
Model C at Collapse Level (Phase 3 Study)
Figure B- 5 Collapse level ground motions used as input motion in response time history analyses of the
nonlinear 3D continuum model (C Model) of the MRO
241
Appendix C Variation of Bridge Foundation Motion from The Free-Field Motion
C.1 MRO Abutment Footing Foundation
Bam-Iran Imperial Valley-06 Kobe Japan

7 5 4
6
4 Free-Field
5 3
Abutment Foundation
4 3
Sa (g)
Sa (g)
Sa (g)
2 Tau Averaging
3 2
2 1
1
1
0 0 0
0 1 2 3 0 1 2 3 0 1 2 3
Kocaeli Turkey Duzce-Turkey Landers

4 5 4
4
3 3
3
Sa (g)
Sa (g)
Sa (g)
2 2
2
1 1
1
0 0 0
0 1 2 3 0 1 2 3 0 1 2 3
Superstition Hills-02 Irpinia Italy-01 El Mayor

3 2 3
2 2
Sa (g)
Sa (g)
Sa (g)
1
1 1
0 0 0
0 1 2 3 0 1 2 3 0 1 2 3
Figure C- 1 Spectra of the Abutment footing foundation, free-field, and estimated foundation based on the Tau-
242
C.2 MRO Pier Pilecap Foundation
Bam-Iran Imperial Valley-06 Kobe Japan

7 5 4
6
4 Free-Field
5 3
4 3 Pier Foundation
Sa (g)
Sa (g)
Sa (g)
2
3 2 Tau Averaging
2 1
1
1
0 0 0
0 1 2 3 0 1 2 3 0 1 2 3
Kocaeli-Turkey Duzce-Turkey Landers

4 4 4
3 3 3
Sa (g)
Sa (g)
Sa (g)
2 2 2
1 1 1
0 0 0
0 1 2 3 0 1 2 3 0 1 2 3
Superstition Hills-02 Irpinia Italy-01 El Mayor
3 2 3
2 2
Sa (g)
Sa (g)
Sa (g)
1
1 1
0 0 0
0 1 2 3 0 1 2 3 0 1 2 3
Figure C- 2 Spectra of the Pier pilecap foundation, free-field, and estimated foundation based on the Tau-
243
Appendix D Calculation of Reinforced Concrete Member Capacitates
The followings are performance criteria that used to calculate bridge shear and displacement
capacities of the main structural components in the discrete models based on the Seismic Retrofitting
Manual of Highway Structures-Part 1 (FHWA, 2006).
D.1 Performance Criteria for Determining Bridge Shear Strength Capacity
The shear demand based on flexural strength is calculated using Equation (D1.1):
Vm = Mp / L (D1.1)
where Vm is the shear demand, Mp is the expected plastic moment capacity, and L is the distance to
the inflection point. If member’s shear capacity is less than the shear demand required to withstand
formation of plastic hinges, it will limit development of the plastic hinge. The Seismic Retrofitting
manual classifies this condition under two categories of Brittle Shear and Semi-Ductile Shear.
The Brittle Shear condition corresponds to the case where initial shear strength of the member
is less than shear demand (Vi ≤ Vm). In this situation, plastic hinge will not form (φp=0) and the failure
mode is a brittle shear-dominant failure.
Semi-ductile shear condition corresponds to the case that initial shear capacity (Vi) is greater
than the shear demand (Vm) but final shear strength of the member (Vf) is less than the demand (Vf ≤
Vm ≤ Vi). Shear capacity of reinforced concrete members changes during cyclic loading and is
dependent on cracking in concrete. The difference between the initial shear capacity (Vi) and the final
shear capacity (Vf) is due to formation of the cracks in concrete during cyclic loading.
Under the semi-ductile condition, plastic hinge will be initiand in the member, but it cannot
achieve its full ductility. In this case, the plastic curvature can be calculated using following equation:
  V −V  
p =  5  i m  + 2  y (D1.2)
  Vi − Vf  
where,  y is yield curvature given by Equation (D1.3).
2 y
y = (D1.3)
D
If the final shear strength of the member (Vf) is greater than the shear demand (Vm<Vf), members
rotational capacity will be determined based on flexure capacity not shear.
244
The details of calculation of initial shear strength ( Vi ), and final shear strength ( Vf ) according to
Bridge Retrofit Manual are presented in Equations (D1.4) and (D1.12), respectively.
Vi = Vs + Vp + Vci (D1.4)
where, Vs is the contribution to the shear strength provided by rebar truss action and can be calculated
as per Equation (D1.5).
D
Vs =A v f yh cot  (D1.5)
s
where, A v is the shear area provided by the transverse hoops and can be calculated as per Equation
(D1.6).
 2A bh for rectangular columns
Av =  (D1.6)
 A bh / 2 for circular columns
where,
Abh is the area of one spiral or hoop bar,
fyh is the yield stress of the transverse hoops or spirals,
D″ is the diameter of transverse hoop or spiral (measured to the centerline of the hoop),
s is the center-to-center longitudinal spacing of the transverse hoop steel,
θ is the angle of the principal crack plane (Kim and Mander, 1999) and given by the following
equation:
 1.6  v Ae 
tan  =   tan 
  A  (D1.7)
 t g 
which,
 is the fixity factor. It is equal to 2 for fixed-fixed column conditions, and 1 for fixed-pinned
conditions.
Ae is the effective shear area (assumed as 80 percent of the gross area, Ag, for rectangular and
circular sections),
 t is the ratio of the total area of longitudinal reinforcing steel bars to the gross section area
 v is the volumetric ratio of transverse steel:
 Av / bws for rectangular columns
v =  (D1.8)
 s / 2 = 2A bh / sD for circular columns
245
A v is the area of transverse shear steel,
b w is the center-to-center spacing of transverse shear steel across width of rectangular

column,
tan  is the corner-to-corner strut angle, can be obtained as per Equation (D1.9).
jd
tan  = (D1.9)
L
where,
jd is the internal lever arm of the concrete compression member, and L is length of the column.
Vp is the shear contribution provided by arch (strut) action and can be calculated as per
Equation (D1.10)

Vp = P tan  (D1.10)
2
where,
P is the axial compression load on member,  is the fixity factor and tan α is the corner-to-
corner strut angle as defined in Equation (D1.9).
Vci is the shear resistance provided by the diagonal tension field in the concrete and is
calculated as per Equation (D1.11).
Vci = 0.3 f ce A e (D1.11)
 is the unconfined concrete compressive strength and Ae is the effective shear area equal
where, f ce
to 0.8Ag. The final shear strength, V f , can be calculated using Equation (D1.12).
Vf = Vs + Vp + Vcf (D1.12)
where, Vs and Vp are as defined in Equations (D1.5) and (D1.10) above and Vcf is the final shear
strength carried by the concrete, which is reduced to allow for plasticisation, cracking, and cyclic
loading effects and can be calculated as per Equation (D1.13).
Vcf = 0.05 f ce Ae (D1.13)
246
D.2 Performance Criteria for Determining Bridge Displacement Capacity
The guideline for calculation of values of plastic curvature corresponding to various limit states in
reinforced concrete columns and beams is given on the Seismic Retrofitting Manual for Highway
Structures (FHWA, 2006), and presented in the following sections.
Compression Failure of Un-Confined Concrete
The plastic curvature corresponding to compression failure in unconfined concrete (spalling

curvature) can be calculated as per Equation (D2.1).
 cu
p = − y (D2.1)
c
where,
 cu is the ultimate concrete compression strain for concrete,
 y is the yield curvature, and
c is the depth from the extreme compression fiber to the neutral axis. The location of the
neutral axis is defined for circular sections as per Equation (D2.2).
0.725
 pe f y  1− 2 c D  
  + 0.5 t   
c 1  fc Ag f c  1 − 2 d D  
= (D2.2)
D  1.32 
 
 
where,
c is the depth to neutral axis,
D is the overall depth of section,
pe is the axial load on the section,
f c is the expected concrete strength,
f y is the expected yield strength of the longitudinal reinforcing steel bars,
Ag is the gross cross-section area,
d' is the depth from the extreme compression fiber to the center of the compression
reinforcing steel bars,
247
 t is the volumetric ratio of the longitudinal reinforcing steel bars,
 is the ratio of average concrete stress in compression zone to confined concrete strength
and can be calculated as per Equation (D2.3).
 =0.85 + 0.12 ( K −1)

0.4
(D2.3)
 is the depth of stress block and can be calculated as per Equation (D2.4).
 =0.85 + 0.13 ( K −1)

0.6
(D2.4)
where, K is the confined strength parameter given by Mander et al. for circular sections as per
Equation (D2.5).
f l f
K = 2.254 1 + 7.94 − 2 l − 1.254 (D2.5)
f ce f ce
where,
f ce is the unconfined concrete compressive strength,
f l is the effective lateral stress supplied by the transverse reinforcing steel bars at yield and
can be calculated as per Equation (D2.6).
1
f l = k e s f yh (D2.6)
2
s is the volumetric ratio of spirals or circular hoops to the core concrete and can be calculated
4A bh
s = (D2.7)
sD
k e is the confinement effectiveness coefficient for spirals and hoop steel, and can be calculated
ke =
(1 −  s D) (D2.8)
1 − cc
cc is the ratio of area of longitudinal reinforcing steel bars to area of core of section
 is the coefficient with values of 0.5 and 1.0 for spirals and hoops, respectively
s is the spacing of spirals or hoops, and
D'' is the diameter of transverse hoop or spiral (measured to the centerline of the hoop).
248
Compression Failure of Confined Concrete
For concrete confined by transverse hoops, cross-ties, or spirals, the compression strain is limited by
first fracture within the confining steel. While this type of failure depends on the cyclic load history,
a conservative estimate of the plastic curvature can be calculated using Equation (D2.9) (FHWA,
2006).
 cu
p = − (D2.9)
( c − d) y
where,
d is the distance from the extreme compression fiber of the cover concrete to the centerline
of the perimeter hoop (thus, c – d" is the depth of confined concrete under compression),
c is the depth from the extreme compression fiber of the cover concrete (which is expected to
spall) to the neutral axis, and
 cu is the ultimate compression strain of the confined core concrete, is calculated by Equation
(D2.10).
1.4 s f yh su
 cu = 0.005 + (D2.10)
f cc
which,
s is the volumetric ratio of transverse steel,
f yh is the yield stress of the transverse steel,
 su is the strain at the maximum stress of the transverse reinforcing steel bars, and
 is the confined concrete strength.
f cc
The axial load in bridge columns is typically low, and the transverse reinforcing steel bars are
mainly required to provide restraint against buckling of the longitudinal reinforcing steel bars. As a
result, low cycle fatigue is a more expected failure mode controlling mechanism than compression
failure of confined concrete for bridge pier columns that incorporate confined concrete details
(FHWA, 2006).
249
Low-cycle Fatigue of Longitudinal Reinforcing Steel Bars
Low cycle fatigue failure of the longitudinal reinforcing steel bars is likely to happen as earthquake
ground motions induce cyclic loads in bridges. This failure mode occurs particularly when the bridge
column is well confined and when other failure modes are tried to be prevented in the design. The
plastic curvature that leads to a low cycle fatigue failure is calculated using Equation (D2.11) (FHWA,
2006).
2 ap 2 ap
p = = (D2.11)
( d − d ) D
where,
D  is the distance between the outer layers of longitudinal steel in a rectangular section
( d − d ) , or the pitch circle diameter of the longitudinal reinforcing steel bars in a circular
section,
 ap is the plastic strain amplitude, can be calculated as per Equation (D2.12).
 ap = 0.08 ( 2Nf )
−0.5
(D2.12)
which, N f is the effective number of equal-amplitude cycles of loading that lead to fracture, which
can be approximated by Equation (D2.13) provided that: 2  Nf  10 .
Nf =3.5 (Tn )-1/3 (D2.13)
where, Tn is natural period of vibration of the bridge.
Buckling of Longitudinal Reinforcing Steel Bars
Local buckling of longitudinal reinforcing steel bars may take place when the rebars are not
adequately supported laterally. This can happen if the spacing (s) of the confinement reinforcing steel
bars (e.g. spirals and hoops) in the plastic hinge zone is too large (when s>6db where db is the diameter
of the longitudinal reinforcing steel bars). The plastic curvature of this failure mode can be determined
using Equation (D2.14):
b
p = − (D2.14)
( c − d ) y
where,
250
d is the distance from the extreme compression fiber to the center of the nearest compression
reinforcing steel bars, and
 b is the buckling strain in the longitudinal reinforcing steel bars. The buckling strain may be
taken as twice the yield strain of the longitudinal reinforcing steel bars as per Equation (D2.15)
if 6db < s < 30db.
2f y
b = (D2.15)
Es
Fracture of the Longitudinal Reinforcing Steel Bars
When the tensile strain reaches a critical level (  s,max ), tensile fracture occurs. This type of failure
can particularly take place under near-field impulse-type ground motions where there is a monotonic
(pushover-like) response. Longitudinal bar fracture failure can occur due to near-field impulse-type
ground motions when resulted in a monotonic response. The plastic curvature in the case of fracture
in the longitudinal bars can be calculated as per Equation (D2.16) (FHWA, 2006).
 s,max
p = − y (D2.16)
(d − c)
where,
d is the depth to the outer layer of tension steel from the extreme compression fiber, and
c is the depth to the neutral axis.
The steel fracture strain,  s,max , should be limited to a value less than or equal to 0.10. (FHWA,
2006).
Plastic Hinge Rotation, p
The plastic hinge rotation, θp , is related to the plastic curvature of a member within the hinge zone
and can be calculated using Equation (D2.17).
 p = p L p (D2.17)
where, the plastic hinge length can be calculated as per the following semi-empirical equation:
Lp = 0.08L + 4400 yd b (D2.18)
251
where,  y is yield strain of the longitudinal reinforcing steel bars, d b is the diameter of the longitudinal
reinforcing steel bars, and L is the shear span or effective height (i.e., L = M / V )
252
Appendix E AASHTO (2017) Site Class Definitions
Site class is categorized by soils stiffness as determined by the shear wave velocity and other soil
characteristics in the upper 30m as per the following Table.
Table E- 1Site Class Definitions specified in Table 3.10.3.1-1 AASHTO (AASHTO, 2017)
Site Class Soil Type and Profile
A Hard rock with measured shear wave velocity, vs30 > 5,000 ft/s (1,500m/s)
B Rock with 2,500 ft/sec < vs30 < 5,000 ft/s (760m/s < vs30 < 1,500 m/s)
C Very dense soil and soil rock with 1,200 ft/sec < vs30 < 2,500 ft/s (360m/s < vS30 < 760), or
with either N > 50 blows/ft, or su > 2,000 psf (100 kPa)
D Stiff soil with 600 ft/s < vs30 < 1,200 ft/s (180 m/s < vs30 < 360 m/s)
, or with either standard penetration resistance 15 < N < 50 blows/ft, or undrained shear
strength 1,000 < su < 2,0000 psf (50 < su < 100 kPa)
E Soil profile with vs30 < 600 ft/s (180m/s) or with either N < 15 blows/ft or su < 1,000 psf, or
any profile with more than 10.0 ft (3m) of soft clay defined as soil with PI > 20, water content
w > 40 percent and su < 500 psf (25kPa)
F Soils requiring site-specific evaluations, such as:

• Peats or highly organic clays layers more than 10.0 ft (3m) of peat or highly organic
clay
• Very high plasticity clays (PI > 75) layers more than 25.0 ft (8m) thick
• Very thick soft/medium stiff clays layers more than 120 ft (36m) thick
253
Appendix F California Mean Predicted ε0 Values for Periods of 0.2s and 1.0 s
The mean predicted epsilon values (ε0) is the expected epsilon value for a specific site and
hazard-level. The following table includes ε0 for different site classes in California. The bolded values
are averaged in calculation of the ε0 value of the MRO site, which was required for the example
application in Section 5.5.
Table F- 1 Mean Predicted ε0 values for periods of 0.2s and 1.0 s, sorted by Seismic Design Category (SDC), for
California sites and selected California cities for site class B (Haselton et al., 2011)
verage ε va ues Average Sa values
SDC ε0 (0.2s) ε0 (1.0s) Sa (0.2s) [g] Sa (1.0s) [g]
ε10/50 ε2/50 ε1/50 ε0.5/50 ε10/50 ε2/50 ε1/50 ε0.5/50 ε10/50 ε2/50 ε1/50 ε0.5/50 ε10/50 ε2/50 ε1/50 ε0.5/50
SDC B 0.14 0.42 0.49 0.55 0.31 0.8 0.94 1.04 0.06 0.18 0.26 0.39 0.02 0.06 0.08 0.11
SDC C 0.11 0.51 0.63 0.75 0.23 0.74 0.88 1.00 0.11 0.31 0.46 0.66 0.04 0.10 0.14 0.19
SDC D 0.25 0.88 1.09 1.27 0.33 0.99 1.21 1.39 0.5 1.05 1.35 1.68 0.18 0.38 0.49 0.62
SDC D, CA 0.67 1.12 1.3 1.46 0.89 1.35 1.52 1.67 0.81 1.42 1.73 2.07 0.31 0.55 0.68 0.81
SDC D, SF 0.88 1.57 1.79 1.95 0.75 1.5 1.75 1.94 1.13 1.78 2.07 2.37 0.52 0.89 1.07 1.25
SDC D, O 0.75 1.5 1.75 2.00 0.95 1.65 1.89 2.13 1.56 2.60 3.07 3.55 0.60 1.01 1.21 1.41
SDC D, BE 0.67 1.41 1.66 1.91 0.90 1.58 1.82 2.04 1.55 2.62 3.11 3.65 0.59 1.01 1.22 1.43
SDC D, SJ 1.11 1.67 1.84 1.94 0.97 1.64 1.86 2.06 1.23 1.92 2.24 2.59 0.47 0.79 0.94 1.10
SDC D, LA 0.66 1.17 1.39 1.62 0.90 1.33 1.50 1.70 1.12 1.99 2.43 2.92 0.39 0.69 0.85 1.02
SDC D, R 1.35 1.77 1.87 1.88 1.41 1.95 2.12 2.22 1.17 1.74 2.02 2.32 0.47 0.72 0.83 0.94
CA: California, SF: San Francisco, O: Oakland, BE: Berkeley, SJ: San Jose, LA: Los Angeles, and R: Riverside
254
Appendix G Relationship between ln (Sc(T1)) and ε(T1)
This appendix includes the linear regression fitting of natural logarithm of the collapse spectral
acceleration of the ground motions at the period of the first mode (Sc(T1)) as a function of epsilon
(ε(T1)).
G.1 D1 model
1.2
lnSc(T1) = 0.183 ε(T1) + 0.449
1.0
0.8
0.6
ln Sc (T1)
0.4
0.2
0.0
-1.0 0.0 1.0 2.0
-0.2
-0.4
ε(T1)
Figure G- 1 Linear regression analysis for determining β0 and β1 coefficients (intercept and slope of the linear
regression line) for the D1 archetype model
G.2 D2 model
1.4
1.2 lnSc(T1) = 0.189 ε(T1) + 0.452
1.0
ln Sc (T1)
0.8
0.6
0.4
0.2
0.0
-0.5 0.5 1.5
ε(T1)
Figure G- 2 Regression analysis for determining β0 and β1 coefficients (intercept and slope of the linear regression
line) for the D2 archetype model
255
G.3 D3 model
1.1 lnSc(T1) = 0.202 ε(T1) + 0.424

0.9
0.7
ln Sc (T1)
0.5
0.3
0.1
-0.1 -0.5 0.5 1.5
-0.3
-0.5
ε(T1)
Figure G- 3 Regression analysis for determining β0 and β1 coefficients (intercept and slope of the linear regression
line) for the D3 archetype model
256
Appendix H Collapse Failure Mode of Discrete Archetype Models D1, D2, D3, and D4
This appendix includes the complete graphs showing failure mode sequences for all four discrete
archetype models, D1 through D4.
H.1 Effect of the Numerical Models on Failure Modes and Their Sequence of Occurrence
Failure Mode:
Abutment steel rupture X X X
Pier shear capacity
1st. 2nd. 3rd. 4th. 5th. 1st. 2nd. 3rd. 4th. 5th.
(a) (b)
Failure Mode:
Pier core compression failure D1 X D1
1st. 2nd. 3rd. 4
th.
5th. 6th. 1st. 2nd. 3rd. 4th. 5th.
(c) (d)
Figure H- 1 Failure modes sequence of the index archetype models D1, D2, D3, and D4 due to the collapse level
ground motions (a) Hector Mine (SF=2, 1.6, 1.8 and 1.6), (b) Cape Mendocino (SF= 2.4, 2, 2 and 1.6), (c)
Christchurch (SF= 2.2, 2, 1.8and 2.2) and (d) El Centro (SF= 2.8, 2.4, 2.6 and 2.2)
257
Failure Mode:
th. th. th.
1st. 2nd. 3rd. 4th. 5th. 6th. 7th. 8th. 1st. 2nd. 3rd. 4th. 5 6 7
(a) (b)
Failure Mode:
Pier shear capacity
Pier core compression failure X
Pier plastic curvature D1 X D1
D3 D3
D4 D4
th.
th.
1st. 2nd. 3rd. 4 5 6th. 7th. 1st. 2nd. 3rd. 4th. 5th. 6th.
(c) (d)
ground motions (a) Kobe (SF=0.8, 1.4, 1 and 1.2), (b) Duzce (SF=1.4, 1.4, 1.2 and 0.8), (c) Kocaeli (SF=1.8, 1.8, 2
and 1.6) and (d) Bam (SF=1.2, 1.8, 1.4 and 1.2)
258
Failure Mode:
Abutment steel rupture X X X
Pier shear capacity
th. th.
1st. 2nd. 3rd. 4th. 5th. 1st. 2nd. 3rd. 4th. 5th. 6 7
(a) (b)
Failure Mode:
Abutment steel rupture X X X X X X
Pier shear capacity
Pier spalling cover failure th.
1st. 2nd. 3rd. 4th. 5th. 6th. 7th. 1st. 2nd. 3rd. 4th. 5 6th.
(c) (d)
ground motions (a) Irpinia (SF= 4.2, 4, 4 and 3.4), (b) Gazli (SF= 1.4, 1.6, 1.8 and 1.4), (c) Tabas (SF= 2, 1.8, 1.8
and 2) and (d) Nahanni (SF= 1.8, 1.6, 1.4 and 1.2)
Failure Mode:
Abutment steel rupture X X X X X X X
Pier shear capacity
th. 5th.
th.
1st. 2nd. 3rd. 4th. 5th. 6 7
th.
1st. 2nd. 3rd. 4 6th.
(a) (b)
ground motions (a) Spitak (SF= 4.4, 3.8, 3 and 3.2), and (b) Chi Chi (SF= 1, 0.8, 1and 1)
259
H.2 Effect of Earthquake Ground Motions on Failure Modes and Their Sequence of
Occurrence

Abutment steel rupture X X Hector Mine
Pier shear capacity Cape Mendocino
X
Pier plastic curvature El Centro
Sequence of Failure Sequence of Failure
(a) (b)
Pier shear capacity
(c) (d)
Figure H- 5 Failure modes and their sequence of occurrence for the index archetype models when subjected to the
Hector Min, Cape Mendocino, and El Centro earthquake ground motions at their collapse levels corresponding to
each archetype model
260
Abutment steel rupture X X X Kobe-Japan
Abutment ultimate compression strain Duzce-Turkey
Pier shear capacity X Kocaeli-Turkey
Pier plastic curvature X Bam-Iran
th.
(a) (b)
Abutment steel rupture X X X X X X X
Pier shear capacity
(c) (d)
Kobe, Duzce, Kocaeli, and Bam earthquake ground motions at their collapse levels corresponding to each
archetype model
261
Abutment steel rupture X X X Irpinia-Italy-01
Abutment ultimate compression strain Gazli-USSR
Pier shear capacity
Pier core compression failure Tabas-Iran
Nahanni-Canada
th.
(a) (b)
Pier shear capacity
(c) (d)
Irpinia, Gazli, Tabas, and Nahanni earthquake ground motions at their collapse levels corresponding to each
archetype model
262
Abutment steel rupture X X X X Spitak-Armenia
Chi-Chi-Taiwan
Pier shear capacity
Pier plastic curvature Christchurch-
Pier ultimate compression failure New Zealand
th.
(a) (b)
Pier shear capacity
(c) (d)
Spitak, Chi Chi, and Christchurch earthquake ground motions at their collapse levels corresponding to each
archetype model
263
H.3 Sequence of Failure for All the Collapsed Level Ground Motions and Index Archetype
Models
Table H- 1 Sequence of failure modes of the index archetype models D 1, D2, D3, and D4 at onset of collapse
Event Name/Model Collapse Failure Modes
D1 D2 D3 D4
Imperial Valley-06 b →d →f →g →i f →g→h →i g→i→h d→e→g→i
Superstition Hills-02 b →d →c →f →g →i b →d→c →g →i b →d→c→g →i b→d→c→ e→ h→g
→i
Kobe-Japan b →g→i→h a →b →d→e→c →f a→c→d →b→g a→c→d→e→f→g→i
→g →i →i
Hector Mine b →f →g →i f →d→g →i → h c→d → b→g →i d→b→g→i
Duzce-Turkey e →f →a →b→g →i g →i→h→f a→b →c→d→e a→b→d→c→g→i
→g →i
Kocaeli-Turkey d →b →c →f →g →i a →b →f→e →g →i d →b→f→g →i b→d→g→i
→d →f
Landers b →d →f →g →i b →c→d→f →g →i b →c→d→f →g f→b→g→i→h
→i
Bam-Iran f →g→i a →b →d→f →c→e d →b→f→g →i f→g→i
El Mayor -5836 b →a →f→g →i b →a →f→c →d →g a→b→g→i f→b→d→g→i
→i
Irpinia-Italy-01 b →f→d →g →i b →d→f →g →i b →d→f→g →i b→f→d→g→i
San Fernando b →g→i b →d→c →f →g →i b →f→g →i→h b→d→f→g→i
→c
Gazli-USSR b →d →f →g →i c →b →d→f →g →h b →c→d→f →e b→f→d→g→i→h
→g →i
Northridge-01 b →d →c →f →g →i b →d→f →e→g b →d→f →c→g f→b→c→d→g→i
→i
Cape Mendocino d →a →f→g →i a →f →d→g →i a→d →f→g →i d→c→e→g→i
Tabas-Iran b →d →f →g →i a →b →d→c →g →i b →c→d→g →i b→d→c→e→f→g→i
Nahanni-Canada b →d →f →g →i b →c→d→f →g →i b →d→f→g →i b→d→g→i
Loma Prieta b →d→c →g a →b →c →d→ b →c→d→e f→a→d→b→g
→i →h →f f →e →g→i →h →f →g→i →i
Spitak-Armenia b →d→f →c b →d→f →g → b →d→g →i a→b→d→c→e
→g →i i →g→i
Chi-Chi-Taiwan d →f→g →i d →e →f→g →i d→e→b→g e→b→h→f→g
→i →i
Christchurch-New b →d→g →i c →d →e →b→ c→a→d →e d→e→b→f→g
Zealand g →i →g →i →i
El Centro b →d→f b →d→f →g → b→f→e a→g→i
i
Parkfield b →f→c →d b →f→c →d → b→c→d→g a→b→c→d→i
→i e→i →i→f
264
Key:
a: Pier spalling cover failure

b: Pier longitudinal bars buckling
c: Pier ultimate compression strain
d: Pier plastic curvature
e: Pier core compression failure
f: Pier shear capacity
g: Abutment ultimate compression strain
h: Abutment curvature capacity
i: Abutment steel rupture
265
Appendix I Concrete Tensile and Compressive Damage Results for the Continuum Model C
Fringe plots of concrete tensile and compressive damages in abutment backwalls, wing walls and pier
columns. Each figure corresponds to a given scaled ground motion.
I.1 Tensile Damage Results
(a)
(b)
Figure I- 1 M O continuum mode ’s concrete tensile damage at the collapsed level (a) Hector Mine (Scale
Factor=1.6) (b) Landers (Scale Factor=1.4)
266
(a)
(b)
Figure I- 2 M O continuum mode ’s concrete tensi e damage at the collapsed level (a) Parkfield (Scale Factor=1.6)
(b) Superstition Hills (Scale Factor=1.4)
267
(a)
(b)
Figure I- 3 M O continuum mode ’s concrete tensi e damage at the co apsed eve (a) Bam (Sca e Factor=1.2) (b)
Kobe (Scale Factor=1.2)
268
(a)
(b)
Figure I- 4 M O continuum mode ’s concrete tensi e damage at the co apsed eve (a) uzce (Sca e Factor=0.8)
(b) Kocaeli (Scale Factor=1.6)
269
I.2 Compressive Damage Results
(a)
(b)
Figure I- 5 M O continuum mode ’s concrete compressive damage at the co apsed eve (a) Hector Mine (Sca e
Factor=1.6) (b) Landers (Scale Factor=1.4)
270
(a)
(b)
Figure I- 6 M O continuum mode ’s concrete compressive damage at the co apsed eve (a) Parkfie d (Sca e
Factor=1.6) (b) Superstition Hills (Scale Factor=1.4)
271
(a)
(b)
Figure I- 7 M O continuum mode ’s concrete compressive damage at the co apsed eve (a) Bam (Sca e
Factor=1.2) (b) Kobe (Scale Factor=1.2)
272
(a)
(b)
Figure I- 8 M O continuum mode ’s concrete compressive damage at the co apsed eve (a) uzce (Sca e
Factor=0.8) (b) Kocaeli (Scale Factor=1.6)
273
Appendix J Results of Flagged Yielded Soil Layers at Collapse Level
The following figures illustrate the yielding state of the soil layers around the abutment and pier piles.
Each figure corresponds to predictions for a given scaled ground motion.
Abutment 1 Pier Abutment 2 1
Acceleration (g)
0.5
0
Soil Layer A -0.5
-1
(Depth=0m-2.17m) -1.5
0 5 10
Time (s)
Acceleration (g)
0.5
0
Soil Layer B -0.5
-1
(Depth=2.17m-5.53m) -1.5
0 5 10
Time (s)
Acceleration (g)
0.5
Soil Layer C 0
-0.5
(Depth=5.53m-10.1m) -1
0 5 10
Time (s)
Acceleration (g)
0.5
Soil Layer D 0
(Depth=10.1m-14.67m) -0.5
-1
0 5 10
Time (s)
Acceleration (g)
0.5
Soil Layer E 0
(Depth=14.67m-20m) -0.5
-1
0 5 10
Time (s)
Figure J- 1 Yielding state of various soil layers around the piles subjected to collapse level Hector Mines ground
motion (SF=1.6). Red color is an indicator of yielding state around the piles and the observed time in the calculated
acceleration time histories.
274
1.5
Acceleration (g)
Soil Layer A 1
0.5
(Depth=0m-2.17m) 0
-0.5
-1
-1.5
0 5 10 15
Time (s)
1.5
Soil Layer B
Acceleration (g)
1
0.5
(Depth=2.17m-5.53m) 0
-0.5
-1
-1.5
0 5 10 15
Time (s)
1
Acceleration (g)
0.5
Soil Layer C 0
(Depth=5.53m-10.1m) -0.5
-1
0 5 10 15
Time (s)
1
Acceleration (g)
Soil Layer D 0.5
(Depth=10.1m-14.67m) 0
-0.5
-1
0 5 10 15
Time (s)
1
Acceleration (g)
Soil Layer E 0.5
0
(Depth=14.67m-20m) -0.5
-1
0 5 10 15
Time (s)
Figure J- 2 Yielding state of various soil layers around the piles subjected to collapse level Landers ground motion
(SF=1.4). Red color is an indicator of yielding state around the piles and the observed time in the calculated
1.5
Soil Layer A Acceleration (g) 1
0.5
0
(Depth=0m-2.17m) -0.5
-1
-1.5
0 2 4 6 8
Time (s)
1.5
Acceleration (g)
Soil Layer B 1
0.5
0
(Depth=2.17m-5.53m) -0.5
-1
-1.5
0 2 4 6 8
Time (s)
1.5
Acceleration (g)
1
Soil Layer C 0.5
0
(Depth=5.53m-10.1m) -0.5
-1
-1.5
0 2 4 6 8
Time (s)
1
Acceleration (g)
Soil Layer D 0.5

0
(Depth=10.1m-14.67m) -0.5
-1
0 2 4 6 8
Time (s)
1
Acceleration (g)
0.5
Soil Layer E 0
-0.5
(Depth=14.67m-20m) -1
0 2 4 6 8
Time (s)
Figure J- 3 Yielding state of various soil layers around the piles subjected to collapse level Parkfield ground motion
275
1
Acceleration (g)
0.5
Soil Layer A 0
(Depth=0m-2.17m) -0.5
-1
0 5 10
Time (s)
1
Acceleration (g)
Soil Layer B 0.5
(Depth=2.17m-5.53m) 0
-0.5
-1
0 5 10
Time (s)
1
Acceleration (g)
Soil Layer C 0.5
0
(Depth=5.53m-10.1m)
-0.5
-1
0 5 10
Time (s)
1
Acceleration (g)
Soil Layer D 0.5
0
(Depth=10.1m-14.67m) -0.5
-1
0 5 10
Time (s)
1
Acceleration (g)
Soil Layer E 0.5
0
(Depth=14.67m-20m)
-0.5
-1
0 5 10
Time (s)
Figure J- 4 Yielding state of various soil layers around the piles subjected to collapse level Superstition Hills
Ground Motion (SF=1.4). Red color is an indicator of yielding state around the piles and the observed time in the
calculated acceleration time histories.
1.5
Acceleration (g)
0.5
Soil Layer A
-0.5
(Depth=0m-2.17m)
-1.5
0 2 4 6 8
Time (s)
1.5
Acceleration (g)
Soil Layer B 0.5
(Depth=2.17m-5.53m) -0.5
-1.5
0 2 4 6 8
Time (s)
1.5
Acceleration (g)
Soil Layer C 0.5
(Depth=5.53m-10.1m) -0.5
-1.5
0 2 4 6 8
Time (s)
1.5
Acceleration (g)
0.5
Soil Layer D
(Depth=10.1m-14.67m) -0.5
-1.5
0 2 4 6 8
Time (s)
1.5
Acceleration (g)
Soil Layer E 0.5
(Depth=14.67m-20m) -0.5
-1.5
0 2 4 6 8
Time (s)
Figure J- 5 Yielding state of various soil layers around the piles subjected to collapse level Bam ground motion
276
Abutment 1 Pier Abutment 2 1
Acceleration (g)
0.5
0
Soil Layer A -0.5
(Depth=0m-2.17m) -1
-1.5
0 2 4
Time (s)
Acceleration (g)
0.5
Soil Layer B 0
(Depth=2.17m-5.53m) -0.5
-1
-1.5
0 2 4
Time (s)
Acceleration (g)
0.5
Soil Layer C 0
(Depth=5.53m-10.1m) -0.5
-1
-1.5
0 2 4
Time (s)
Acceleration (g)
0.5
Soil Layer D 0
(Depth=10.1m-14.67m) -0.5
-1
-1.5
0 2 4
Time (s)
Acceleration (g)
0.5
Soil Layer E 0
-0.5
(Depth=14.67m-20m) -1
-1.5
0 2 4
Time (s)
Figure J- 6 Yielding state of various soil layers around the piles subjected to collapse level Kobe ground motion
1
Acceleration (g)
0.5
Soil Layer A 0
(Depth=0m-2.17m) -0.5
-1
0 2 4 6 8 10
Time (s)
1
Acceleration (g)
0.5
Soil Layer B 0
(Depth=2.17m-5.53m) -0.5
-1
0 2 4 6 8 10
Time (s)
1
Acceleration (g)
0.5
Soil Layer C 0
(Depth=5.53m-10.1m) -0.5
-1
0 2 4 6 8 10
Time (s)
1
Acceleration (g)
Soil Layer D 0.5

0
(Depth=10.1m-14.67m)
-0.5
-1
0 2 4 6 8 10
Time (s)
1
Acceleration (g)
0.5
Soil Layer E
0
(Depth=14.67m-20m) -0.5
-1
0 2 4 6 8 10
Time (s)
Figure J- 7 Yielding state of various soil layers around the piles subjected to collapse level Duzce ground motion
277
Abutment 1 Pier Abutment 2 1.5
Acceleration (g)
0.5
Soil Layer A
-0.5
(Depth=0m-2.17m)
-1.5
0 2 4 6 8 10
Time (s)
1.5
Acceleration (g)
Soil Layer B 0.5
(Depth=2.17m-5.53m) -0.5
-1.5
0 2 4 6 8 10
Time (s)
1.5
Acceleration (g)
Soil Layer C 0.5
(Depth=5.53m-10.1m) -0.5
-1.5
0 2 4 6 8 10
Time (s)
1.5
Acceleration (g)
Soil Layer D
0.5
(Depth=10.1m-14.67m)
-0.5
-1.5
0 2 4 6 8 10
Time (s)
Soil Layer E 1.5
Acceleration (g)
(Depth=14.67m-20m) 0.5
-0.5
-1.5
0 2 4 6 8 10
Time (s)
Figure J- 8 Yielding state of various soil layers around the piles subjected to collapse level Kocaeli ground motion
278
Appendix K Execution Time of the Continuum Models (C, E, and F Model)
Table K- 1 Ground motions used in the continuum analyses

GM E and F Model C Model
1 Imperial Valley-06 Imperial Valley-06 (SF=1.6)
2 El Mayor El Mayor (SF=3.6)
3 Hector Mine Hector Mine (SF=1.6)
4 Landers Landers (SF=1.4)
5 Irpinia Italy Parkfield (SF=1.6)
6 Superstition Hills Superstition Hills (SF=1.6)
7 Bam Bam (SF=1.2)
8 Kobe Kobe (SF=1.2)
9 Duzce Duzce (SF=0.8)
10 Kocaeli Kocaeli (SF=1.6)
700
E Model
600
F Model
500 C Model
400
Hour
300
200
100
0
0 1 2 3 4 5 6 7 8 9 10
Ground Motion
Figure K- 1 Wallclock time for the continuum models using a workstation with 24 CPUs @ 3.0 GHz and 512 GB
RAM
279
Appendix L Validation of ABAQUS One-way Infinite Element CN3D8 for the Quite
Boundaries (SIMULIA, 2018)
Performance of the ABAQUS CIN3D8 infinite element for the quiet boundaries is confirmed by
comparing the responses of the infinite element and dashpot representation (SIMULIA, 2018). This
validation uses steady-state dynamic analysis of a single element model including the infinite element
(T1 Model). Then the analysis is repeated replacing the infinite element CIN3D8 by dashpots in the
model (T2 Model). The displacements and phase angles for the two models were compared.
Models Description:
A linear brick element (C3D8) used in both numerical model T1 and T2. The model configurations
used in the performance validation of the CIN3D8 for the quite boundaries are shown in Figure L- 1.
The figure shows elevation view and model configuration used in the steady-state analysis to validate
infinite element efficiency for the non-reflecting boundaries.
T1 Model T2 Model
y
D x C D
C Legend:
z Plane wave
propagation (cp)
along CD
Regular Continuum Regular Continuum
Finite Element (C3D8) Finite Element (C3D8) Shear wave
propagation
(cs)
along CD
ds ds
A B A B
dp dp
One-way Infinite Element Dashpots

(CIN3D8)
(a) (b)
Figure L- 1 (a) Continuum single element model with a CIN3D8 one-way infinite element extension, (b) Continuum
single element model with dashpots connected to the ground at the base of the model (SIMULIA, 2018).
280
Material Properties
In this validation example, the Young's modulus, density, and Poisson's ratio values equal to 1.0N/m2,
0.01kg/m3, and 0.1 were assigned to the elements in the models, respectively.
Boundary conditions
Boundary conditions were considered in account for the Plane wave and Shear wave are summarized
as follow:
– Plane wave: uy= 1.0 X 10–4m along edge CD, ux = 0 throughout the model.
– Shear wave: ux= 1.0 X 10–4m along edge CD, uy= 0 throughout the model.
Calculation of Dashpot Coefficients Satisfying Quite Boundary Condition

A total 4 numbers of the standard dashpot, which allows connecting a node to the ground in ABAQUS
are used for each type of wave in the base nodes of the T2 model. The dashpot coefficients
corresponding to the Plane wave (dp) and Shear wave (ds) were calculated using the Equation (L.1)
and Equation (L.2) to assure providing the quite boundary condition in the T2 model.
d p =  cp (L.1)
ds =  cs (L.2)
where, ρ is density, cp is the plane wave velocity and cs is the shear wave speed. Plane and shear wave
velocities can be calculated using Equation (L.3) and Equation (L.4), respectively.
 + 2G
cp = (L.3)

G
cp = (L.4)

where, λ and G are Lamé's first parameter and shear modulus, respectively, and can be calculated as
based using Equation (L.5) and Equation (L.6):
E
= (L.5)
( )(1 − 2 )
1 + 
E
G= (L.6)
2 (1 +  )
Based on the considered material properties and Equation (L.3), Equation (L.4), Equation
(L.5), and Equation (L.6), plane and shear wave velocity and damping coefficient corresponding to
the plane and shear waves were calculated. These values are summarized in the Table L- 1.
281
Table L- 1 Calculated moduli and dashpot parameters
G(Pa) λ(Pa) cp(m/s) cs(m/s) dp(Ns/m) ds(Ns/m)

0.550 0.110 11.016 7.416 0.110 0.074
A quarter value of the calculated damping coefficients corresponding to the plane and shear waves
were assigned to each dashpot in the T2 model.
Direct-Solution Steady-State Dynamic Analysis
Steady-state dynamic analysis is performed in the T1 and T2 models. This type of analysis can directly
calculate (in terms of the physical degrees of freedom of the model) the steady-state dynamic
linearized response of a system to harmonic excitation. A linear perturbation procedure is adopted in
the direct-solution steady-state dynamic analysis. In ABAQUS/Standard the direct-solution steady-
state dynamic procedure conducts a frequency sweep by applying the loading at a user-specified range
of frequencies and recording the response (SIMULIA, 2018). In the performed direct-solution steady-
state dynamic on both models, linear frequency spacing is considered. The required parameters in the
executed analyses summarized in Table L- 2.
Table L- 2 Calculated moduli and dashpot parameters
Lower frequency Range Upper frequency Range Number of points Bias

(Hz) (Hz) (-) (-)
0.05 20 20 1
In the direct-solution steady-state dynamic, the bias parameter is used to provide closer spacing of the
results points either toward the middle or toward the ends of each frequency interval.
Results and Discussion:
Calculated displacements at x and y directions for node A of the T1 and T2 models were compared
and shown in Figure L- 2 and Figure L- 3.
282
1.20E-04 T1 Model
1.00E-04 T2 Model
Harmonic Displacement (m)
8.00E-05
6.00E-05
4.00E-05
2.00E-05
0.00E+00
-2.00E-05
-4.00E-05
0 5 10 15 20 25
Ferquency (Hz)
Figure L- 2 Comparison of the calculated harmonic amplitude at x-direction at the node A for T1 and T2 Models
1.20E-04 T1 Model
1.00E-04 T2 Model
Harmonic Displacement (m)
8.00E-05
6.00E-05
4.00E-05
2.00E-05
0.00E+00
-2.00E-05
-4.00E-05
0 5 10 15 20 25
Ferquency (Hz)
Figure L- 3 Comparison of the calculated harmonic amplitude at y-direction at the node A for T1 and T2 Models
As shown in Figure L- 2 and Figure L- 3, the calculated displacement responses and phase
angles at the boundary node A show a perfect match in all frequency ranges for both models.
Comparison of the response of the two models shows that dissipation of energy in the two models is
283
identical. Considering that despots’ coefficient in the T2 model were calculated to satisfy the quite
boundary condition, it can be concluded that energy did not reflect back into the system in the T1
model. It demonstrates that the infinite element CIN3D8, can simulate the quite boundaries
transmitting the wave through the boundaries (or absorbing energy).
284

Investigation of Effects of Soil-Structure Interaction On The Seismic Response of RC Bridges Using A Performance-Based Design Approach

Uploaded by

Copyright:

Available Formats

You might also like

Investigation of Effects of Soil-Structure Interaction On The Seismic Response of RC Bridges Using A Performance-Based Design Approach

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Investigation of Effects of Soil-Structure Interaction On The Seismic Response of RC Bridges Using A Performance-Based Design Approach

Uploaded by

Copyright:

Available Formats

INVESTIGATION OF EFFECTS OF SOIL-STRUCTURE INTERACTION ON THE SEISMIC

RESPONSE OF RC BRIDGES USING A PERFORMANCE-BASED DESIGN APPROACH

Kianosh Ashkani Zadeh

B.Sc., Civil Engineering, University of Isfahan, 1993

A THESIS SUBMITTED IN PARTIAL FULFILLMENT OF

THE REQUIREMENTS FOR THE DEGREE OF

THE FACULTY OF GRADUATE AND POSTDOCTORAL STUDIES

THE UNIVERSITY OF BRITISH COLUMBIA

© Kianosh Ashkani Zadeh, 2020

Investigation of Effects of Soil-Structure Interaction on Seismic Response of RC Bridges

Using a Performance-Based Design Approach

submitted by Kianosh Ashkani Zadeh in partial fulfillment of the requirements for

the degree of Doctor of Philosophy

Despite of soil-structure interaction (SSI) potentially important role on seismic response of

This thesis titled “Investigation of effects of soil-structure interaction on seismic response of RC

Abstract ......................................................................................................................................... iii

Lay Summary ............................................................................................................................... iv

Table of Contents ......................................................................................................................... vi

List of Tables .............................................................................................................................. xiii

List of Appendix Tables............................................................................................................ xvii

List of Figures ........................................................................................................................... xviii

List of Appendix Figures ....................................................................................................... xxviii

List of Symbols ........................................................................................................................ xxxii

List of Abbreviations .................................................................................................................. xli

Acknowledgements ................................................................................................................... xliv

Dedication ................................................................................................................................... xlv

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

1.1 Overview, and Motivations ............................................................................................. 1

1.2 Design Codes and SSI Effect .......................................................................................... 3

1.3 Implementation of SSI Effects in Analysis and Design.................................................. 4

1.4 Goals, Objectives, and Tasks .......................................................................................... 7

1.5 Thesis organization .................................................................................................... 8

Chapter 2: Literature Survey .....................................................................................................12

2.1 Scientific Understanding of SSI: Early History to Current State ................................. 12

2.2 Kinematic SSI Effect .................................................................................................... 16

2.4 Summary of Key SSI Research Milestones .................................................................. 19

2.5 SSI in Design Codes ..................................................................................................... 21

2.6 Gaps in Knowledge ....................................................................................................... 25

Chapter 3: Investigation of Variation of Bridge Foundation Motion from the Free-Field

3.1 Introduction and Methodology ..................................................................................... 27

3.2 Meloland Road Overcrossing (MRO) ........................................................................... 29

3.3 Continuum Model to Investigate Kinematic Effects .................................................... 31

3.3.1 Elastic Model Description......................................................................................... 32

3.3.3 Choice of Ground Motions ....................................................................................... 37

3.3.4 Damping and Elastic Analysis .................................................................................. 39

3.4 Results ........................................................................................................................... 41

3.4.1 Effect of Kinematic Soil-Structure Interaction on Foundation Motion: Amplification

3.4.2 Applicability of Tau-averaging Method to Estimate Foundation Motion ................ 46

3.5 Chapter Summary ......................................................................................................... 55

Chapter 4: Collapse Assessment of the Meloland Road Overcrossing (MRO) ......................57

4.1 Introduction and Methodology ..................................................................................... 57

4.2 Development of nonlinear Archetype Models for Collapse Assessment ..................... 61

4.2.1 Choice of Software for Discrete Approach ............................................................... 63

4.2.2 Element Type ............................................................................................................ 64

4.2.4 Material Properties .................................................................................................... 67

4.2.5 Sectional Response of Pier Column and Abutment Backwalls ................................ 69

4.2.6 Structural Damping ................................................................................................... 71

4.2.7 SSI Features in D1, D2 and D3 Index Archetype Models .......................................... 75