Site Response, Soil-Structure Interaction and Structure-Soil-Structure Interaction For Performance Assessment of Buildings and Nuclear Structures

ISSN 1520-295X
Site Response, Soil-Structure Interaction

and Structure-Soil-Structure Interaction for
Performance Assessment of Buildings and
Nuclear Structures
by
Chandrakanth Bolisetti and Andrew S. Whittaker
Technical Report MCEER-15-0002

June 15, 2015
This research was conducted at the University at Buffalo, State University of New York, and was supported primarily by
the National Science Foundation under grant number CMMI-0830331.
NOTICE
This report was prepared by the University at Buffalo, State University of New
York, as a result of research sponsored by MCEER through a grant from the
National Science Foundation under grant number CMMI-0830331 and other
sponsors. Neither MCEER, associates of MCEER, its sponsors, University at
Buffalo, State University of New York, nor any person acting on their behalf:
a. makes any warranty, express or implied, with respect to the use of any
information, apparatus, method, or process disclosed in this report or that
such use may not infringe upon privately owned rights; or
b. assumes any liabilities of whatsoever kind with respect to the use of, or the
damage resulting from the use of, any information, apparatus, method, or
process disclosed in this report.
Any opinions, findings, and conclusions or recommendations expressed in this

publication are those of the author(s) and do not necessarily reflect the views
of MCEER, the National Science Foundation or other sponsors.
Site Response, Soil-Structure Interaction and
Structure-Soil-Structure Interaction for
Performance Assessment of Buildings and Nuclear Structures
by
Chandrakanth Bolisetti1 and Andrew S. Whittaker2
Publication Date: June 15, 2015

Submittal Date: December 15, 2014
Technical Report MCEER-15-0002
National Science Foundation Grant Number: CMMI-0830331
1 Postdoctoral Associate, Idaho National Laboratory; former Graduate Student, Department of

Civil, Structural and Environmental Engineering, University at Buffalo, State University of New
York
2 Professor, Chair, and MCEER Director, Department of Civil, Structural and Environmental
Engineering, University at Buffalo, State University of New York
MCEER
University at Buffalo, State University of New York
212 Ketter Hall, Buffalo, NY 14260
E-mail: [email protected]; Website: http://mceer.buffalo.edu
Preface
MCEER is a national center of excellence dedicated to the discovery and development of new knowl-
edge, tools and technologies that equip communities to become more disaster resilient in the face of
earthquakes and other extreme events. MCEER accomplishes this through a system of multidisciplinary,
multi-hazard research, in tandem with complimentary education and outreach initiatives.
Headquartered at the University at Buffalo, The State University of New York, MCEER was originally
established by the National Science Foundation in 1986, as the first National Center for Earthquake
Engineering Research (NCEER). In 1998, it became known as the Multidisciplinary Center for Earth-
quake Engineering Research (MCEER), from which the current name, MCEER, evolved.
Comprising a consortium of researchers and industry partners from numerous disciplines and institutions
throughout the United States, MCEER’s mission has expanded from its original focus on earthquake
engineering to one which addresses the technical and socio-economic impacts of a variety of hazards,
both natural and man-made, on critical infrastructure, facilities, and society.
The Center derives support from several Federal agencies, including the National Science Foundation,
Federal Highway Administration, National Institute of Standards and Technology, Department of
Homeland Security/Federal Emergency Management Agency, and the State of New York, other state
governments, academic institutions, foreign governments and private industry.
The studies presented in this report are a part of the NSF funded project titled, ‘Seismic Performance
Assessment of Buildings in Dense Urban Regions,’ and referred to in this report as the City Block
project. The goal of this project is to understand the significance of structure-soil-structure interac-
tion in buildings through a series of centrifuge experiments and numerical simulations, and develop
guidelines for analysts and engineers in accounting for structure-soil-structure interaction in the design
and performance assessment of buildings. A total of six centrifuge experiments were performed at
the NEES@UCDavis geotechnical centrifuge facility. These experiments tested small-scale building
models placed on dry sand (Test 1 to Test 4) and liquefiable soil (Test 5 and Test 6). The City Block
project involved a team of six universities including the University of California, Berkeley, University
of California, Davis, University of California, San Diego, University at Buffalo, and the California
Polytechnic State University.
This report presents three studies that share the common goal of better understanding the phenomena of site response,
soil-structure interaction and structure-soil-structure interaction, and advancing the numerical tools that simulate these
phenomena. The first two studies involve an assessment of the industry-standard frequency-domain and time-domain
programs used for site-response and soil-structure interaction analyses. The more recently developed nonlinear time-domain
methods are benchmarked against the traditionally used frequency-domain methods for low-intensity ground shaking,
and their predictions are compared for higher intensity shaking. The results are used to provide practical guidelines for
the usage of these programs. The third study involves the examination of structure-soil-structure interaction effects in
two of the City Block centrifuge experiments through analysis of experimental data and numerical simulations. The
numerical simulations are performed using the frequency- and time-domain programs assessed in the first two studies and
supplement the experimental data by providing a better insight into the phenomenon of structure-soil-structure interaction.
iii
ABSTRACT
During an earthquake, the flexibility of the near-surface soil affects the dynamic response of the
supported structure. This dynamic interaction is termed soil-structure interaction (SSI). Soil-structure
interaction is important for performance-based design, which has seen increasing use in earthquake
engineering in the United States in the last decade. Soil-structure interaction analysis is preceded by site-
response analysis, which is the process of calculating the response of a soil profile to rock motion at depth.
Soil flexibility also results in the interaction of adjacent structures through intermediate soil. This
interaction is termed structure-soil-structure interaction (SSSI), which can be important for the
performance assessment of buildings, especially those constructed in dense, urban areas. The studies
presented herein share the common goal of a better understanding of the phenomena of site response, SSI
and SSSI, and advancing the numerical tools that simulate these phenomena.
The state-of-the-art in site-response and SSI analysis of buildings and nuclear structures involves
the use of linear or equivalent-linear numerical methods that function in the frequency domain. However,
nonlinear time-domain methods may be more appropriate for cases involving intense earthquakes that
result in highly nonlinear soil and foundation response. Such nonlinear methods are available but rarely
used due to the higher computational requirements and lack of experience with analysts and regulators.
This report assesses industry-standard equivalent linear and nonlinear site-response and SSI analysis
programs, and provides practical guidelines on their usage. The assessment includes benchmarking the
time-domain programs against frequency-domain programs for analyses involving almost linear soil
response, and a comparison of the predictions of these programs for analyses involving highly nonlinear
soil and foundation response. The assessment also identifies the various pitfalls that can be expected when
using these programs and suggests workarounds.
Structure-soil-structure interaction has been rarely studied, mainly due to the lack of experimental
or field-based case-studies that demonstrate its effects on structural response. The US National Science
Foundation funded a research project titled ‘Seismic Performance Assessment of Buildings in Dense
Urban Environments’ to understand the significance of SSSI in buildings through a series of centrifuge
experiments and numerical simulations. The experiments involve small-scale building models with
commonly-found superstructure and foundation configurations. This report describes some of these
centrifuge experiments, and their numerical simulations, which are comprehensive and performed using
both equivalent-linear and nonlinear SSI analysis programs. The numerical and experimental results show
that wave-based SSSI does not significantly affect the global response of the buildings considered in this
study.
Y
ACKNOWLEDGEMENTS
The research presented in this report is funded by the US National Science Foundation under
Grant CMMI-0830331, ‘Seismic Performance Assessment of Dense Urban Environments’ (the City
Block project), and MCEER. The financial support is gratefully acknowledged.
The authors would like to thank the other Principal Investigators of the City Block project, Prof.
Jonathan Bray, Prof. Bruce Kutter, Prof. Greg Feigel and Prof. Tara Hutchinson, and involved graduate
students, Dr. Zhiqiang Chen, Dr. H. Benjamin Mason and Dr. Nicholas W. Trombetta.
Various professionals in the industry practice and in academia have provided valuable assistance
over the course of this research project. Ibbi Almufti, Dr. Michael Willford, Dr. Yuli Huang and Dr.
Armin Masroor of Arup, San Francisco provided much guidance on performing site-response and soil-
structure interaction analysis in LS-DYNA. Dr. Farhang Ostadan of Bechtel, San Francisco provided
counsel on the use of SASSI. Dr. Irfan Baig and Dr. Said Bolourchi of Simpson Gumpertz and Heger,
Boston provided guidance on and identified challenges with the use of SHAKE. Prof. Youssef Hashash at
University of Illinois at Urbana-Champaign provided guidance with DEEPSOIL. Prof. Boris Jeremic at
University of California at Davis, Prof. Jacobo Bielak at Carnegie Mellon University and Dr. Ushnish
Basu at Livermore Software Technology Corporation provided guidance in understanding the domain
reduction method and performing soil-structure interaction analysis using this method in LS-DYNA. The
authors are grateful for this support.
vLL
TABLE OF CONTENTS
SECTION 1 INTRODUCTION................................................................................................................. 1
1.1 Background......................................................................................................................................... 1
1.2 Motivation .......................................................................................................................................... 2
1.2.1 Assessment of numerical methods in site-response and soil-structure interaction analysis .... 2
1.2.2 Seismic structure-soil-structure interaction in dense urban environments ............................... 4
1.3 Report organization ............................................................................................................................ 5
SECTION 2 SITE-RESPONSE ANALYSIS: A REVIEW ..................................................................... 7
2.1 Dynamic behavior of soils during earthquakes .................................................................................. 7

2.1.1 Experimental response of cyclically loaded soils .................................................................... 7
2.1.2 Numerical material models for dynamic soil behavior ............................................................ 8
2.1.2.1 Equivalent-linear models ..................................................................................................... 8
2.1.2.2 Nonlinear hysteretic models .............................................................................................. 10
2.2 Site-response analysis ....................................................................................................................... 11

2.2.1 The nature of site response..................................................................................................... 12
2.2.2 One-dimensional methods of site-response analysis.............................................................. 12
2.2.2.1 Evolution of one-dimensional site-response analysis methods ......................................... 13
2.2.2.2 Equivalent-linear methods ................................................................................................. 14
2.2.2.2.1 SHAKE...................................................................................................................... 15
2.2.2.3 Nonlinear methods............................................................................................................. 16
2.2.2.3.1 DEEPSOIL ................................................................................................................ 17
2.2.2.3.2 LS-DYNA ................................................................................................................. 19
2.2.2.4 Validation of one-dimensional site-response analysis programs ....................................... 20
2.2.2.5 Comparison and limitations of various one-dimensional methods.................................... 26
2.2.3 Multi-dimensional analyses ................................................................................................... 27
[i
SECTION 3 SOIL-STRUCTURE INTERACTION: A REVIEW ....................................................... 29
3.1 Introduction ...................................................................................................................................... 29

3.2 Numerical methods in soil-structure interaction analysis ................................................................. 31
3.2.1 Substructuring method ........................................................................................................... 32
3.2.2 Direct method......................................................................................................................... 34
3.2.3 Domain reduction method ...................................................................................................... 36
3.3 Soil-structure interaction analysis using industry-standard numerical programs ............................. 38

3.3.1 SASSI ..................................................................................................................................... 38
3.3.1.1 Theory and implementation ............................................................................................... 39
3.3.1.2 Modeling and analysis procedure ...................................................................................... 42
3.3.2 LS-DYNA .............................................................................................................................. 44
3.3.2.1 Direct method .................................................................................................................... 44
3.3.2.2 The effective seismic input method ................................................................................... 45
3.3.3 The US NRC ESSI Simulator ................................................................................................ 47
3.3.4 Assessment of numerical soil-structure interaction analysis methods ................................... 47
3.4 Structure-soil-structure interaction ................................................................................................... 48
SECTION 4 AN ASSESSMENT OF SITE-RESPONSE ANALYSIS PROGRAMS ......................... 51
4.1 Introduction ...................................................................................................................................... 51

4.2 Soil profiles used for site-response analysis ..................................................................................... 51
4.3 Selection of input ground motions.................................................................................................... 52
4.4 Numerical modeling ......................................................................................................................... 58
4.4.1 Material modeling .................................................................................................................. 59
4.4.2 Site layering ........................................................................................................................... 61
4.5 Numerical analyses ........................................................................................................................... 63

4.5.1 Modal analyses....................................................................................................................... 63
4.5.2 Site-response analyses for sites E1 and E2 ............................................................................ 64
4.5.3 Site-response analyses for site W1 ......................................................................................... 64
4.5.4 Site-response analyses for site W2......................................................................................... 64
[
4.6 Observations ..................................................................................................................................... 65
4.6.1 Site E1 .................................................................................................................................... 66
4.6.2 Site E2 .................................................................................................................................... 67
4.6.3 Site W1................................................................................................................................... 69
4.6.3.1 Response to ordinary ground motions ............................................................................... 69
4.6.3.2 Response to pulse-like ground motions ............................................................................. 70
4.6.3.3 Effect of pressure-dependence of soil properties............................................................... 73
4.6.4 Site W2................................................................................................................................... 73
4.6.4.1 Response to ordinary ground motions ............................................................................... 73
4.6.4.2 Response to pulse-like ground motions ............................................................................. 75
4.6.4.3 Effect of pressure-dependence of soil properties............................................................... 77
4.7 Summary and conclusions ................................................................................................................ 78

4.7.1 General conclusions regarding nonlinear site response ......................................................... 78
4.7.2 Comparison of equivalent-linear and nonlinear responses .................................................... 79
4.7.3 Comparison between the responses from DEEPSOIL and LS-DYNA .................................. 80
4.7.4 Key issues faced during site-response analysis ...................................................................... 81
4.7.4.1 SHAKE .............................................................................................................................. 81
4.7.4.2 DEEPSOIL ........................................................................................................................ 83
4.7.4.3 LS-DYNA.......................................................................................................................... 85
4.7.4.4 Sensitivity of high-frequency response ............................................................................. 86
4.7.5 Calculation and specification of input for site-response analysis .......................................... 87
4.8 Site-response analysis for biaxial horizontal shaking ....................................................................... 89

4.9 Recommendations for future research .............................................................................................. 91
SECTION 5 AN ASSESSMENT OF SOIL-STRUCTURE INTERACTION ANALYSIS

PROGRAMS .......................................................................................................................................... 93
5.1 Overview .......................................................................................................................................... 93

5.2 Calculation of linear elastic SSI response using SASSI and LS-DYNA .......................................... 93
5.2.1 Model description .................................................................................................................. 93
5.2.2 Modeling in SASSI ................................................................................................................ 95
5.2.3 Modeling in LS-DYNA using the direct method and the effective seismic input method .... 96
xL
5.2.4 Analysis results and comparison between SASSI and LS-DYNA......................................... 99
5.2.4.1 Free-field response ............................................................................................................ 99
5.2.4.2 Fixed-base responses ....................................................................................................... 100
5.2.4.3 Response of the basemat alone ........................................................................................ 101
5.2.4.4 Response of LM1 with SSI effects .................................................................................. 102
5.2.4.5 Response of LM2 with SSI effects .................................................................................. 104
5.2.5 Observations......................................................................................................................... 106
5.3 Calculation of nonlinear SSI responses using SASSI and LS-DYNA ........................................... 108
5.3.1 Superstructure and soil modeling ......................................................................................... 108
5.3.2 Analysis results and comparison .......................................................................................... 109
5.3.2.1 MS1F_2 ........................................................................................................................... 110
5.3.2.2 MS1F_2 equipped with rigid grade beams ...................................................................... 114
5.3.2.3 MS2F ............................................................................................................................... 117
5.3.3 Observations......................................................................................................................... 120
5.4 Practical considerations for SSI analysis using SASSI and LS-DYNA ......................................... 121
5.4.1 Selection of analysis frequencies in SASSI ......................................................................... 122
5.4.2 Foundation meshing in SASSI ............................................................................................. 123
5.4.3 Column-basemat connection properties in SASSI ............................................................... 124
5.4.4 Sensitivity of rocking response to site layering in SASSI and LS-DYNA .......................... 126
5.4.5 Hourglass effects in LS-DYNA response ............................................................................ 127
5.5 Summary and conclusions .............................................................................................................. 129

5.6 Recommendations .......................................................................................................................... 130
SECTION 6 THE NEES CITY BLOCK CENTRIFUGE EXPERIMENTS: TEST 3 AND

TEST 4 ....................................................................................................................................... 131
6.1 Introduction .................................................................................................................................... 131

6.2 Geotechnical centrifuge testing ...................................................................................................... 131
6.2.1 Background .......................................................................................................................... 131
6.2.2 Advantages and limitations .................................................................................................. 136
6.2.3 The NEES@UC Davis centrifuge facility ........................................................................... 137
xLL
6.3 Test 2: Two adjacent structures on dense, dry sand ....................................................................... 139
6.3.1 Test description .................................................................................................................... 139
6.3.2 Soil profile............................................................................................................................ 140
6.3.3 Model design and instrumentation ....................................................................................... 140
6.3.3.1 Selection of prototype building parameters ..................................................................... 141
6.3.3.2 Design of experimental model structures ........................................................................ 141
6.3.3.3 Instrumentation ................................................................................................................ 146
6.3.4 Input ground motions and test chronology........................................................................... 147
6.4 Test 3: An in-plane SSSI configuration on dense, dry sand ........................................................... 149
6.4.1.1 The ‘transmitter-receiver’ configuration ......................................................................... 151
6.4.2 Soil profile............................................................................................................................ 151
6.4.3.1 Design of two-story model, MS2F_iSSSI ....................................................................... 152
6.4.3.2 Design of the one-story models, MS1F_2_Baseline and MS1F_2_iSSSI....................... 154
6.5 Test 4: Anti-plane, and combined in-plane anti-plane SSSI configurations on dense, dry sand .... 159
6.5.2 Soil profile............................................................................................................................ 164
6.5.3.1 Differences in model designs from Test 3 ....................................................................... 164
SECTION 7 EXPERIMENTAL RESULTS AND COMPARISON WITH NUMERICAL

SIMULATIONS ...................................................................................................................................... 169
7.1 Introduction .................................................................................................................................... 169

7.2 Scope of experimental data analysis and numerical simulations .................................................... 169
7.3 Examination of site response .......................................................................................................... 174
7.3.1 Calculation of the material properties .................................................................................. 174
xiLL
7.3.2 Site-response analysis in SHAKE ........................................................................................ 176
7.3.3 Site-response analysis in DEEPSOIL .................................................................................. 176
7.3.4 Site-response analysis in LS-DYNA .................................................................................... 177
7.3.5 Variability in the input ground motion and soil properties during the experiments ............ 178
7.3.6 Comparison of experimental and numerical site responses ................................................. 180
7.4 Examination of SSI and SSSI effects ............................................................................................. 182

7.4.1 Modeling in SASSI .............................................................................................................. 183
7.4.2 Modeling in LS-DYNA ....................................................................................................... 189
7.4.3 Calibration of fixed-base models ......................................................................................... 194
7.4.4 Comparison of experimental and numerical results ............................................................. 195
7.4.4.1 MS1F_2 ........................................................................................................................... 196
7.4.4.2 Footing response in MS1F_2 ........................................................................................... 199
7.4.4.2.1 Moment-rotation response ....................................................................................... 200
7.4.4.2.2 Vertical force-displacement response...................................................................... 202
7.4.4.3 MS2F ............................................................................................................................... 206
7.4.5 Effect of basement restraint on the response of MS1F_2 .................................................... 208
7.4.5.1 Pseudo-static analyses ..................................................................................................... 210
7.4.5.2 Dynamic analyses ............................................................................................................ 212
7.4.6 Effect of basement restraint on the response of MS1F_1 .................................................... 215
7.5 Summary and conclusions .............................................................................................................. 217
SECTION 8 SUMMARY, CONCLUSIONS AND RECOMMENDATIONS................................... 221
8.1 Summary......................................................................................................................................... 221

8.2 Conclusions .................................................................................................................................... 223
8.2.1 Assessment of industry-standard site-response analysis programs ...................................... 223
8.2.2 Assessment of industry-standard SSI analysis programs ..................................................... 225
8.2.3 Centrifuge experiments and numerical simulations of the NEES City Block project ......... 227
8.3 Future research ............................................................................................................................... 228
6(&7,215EFERENCES...................................................................................................................231
xiY
APPENDIX A RESULTS OF SITE-RESPONSE ANALYSES.......................................................... 245
APPENDIX B THE ORMSBY WAVELET ......................................................................................... 325
xY
LIST OF FIGURES
Figure 2-1: Calculation of equivalent-linear parameters: secant shear modulus and effective damping ratio
...................................................................................................................................................................... 8
Figure 2-2: Upper bound modulus reduction and damping curves for sand [from Seed and Idriss (1970)]. 9
Figure 2-3: Nonlinear hysteretic model with a hyperbolic backbone curve and extended Masing rules
(Hashash and Park, 2001) ........................................................................................................................... 11
Figure 2-4: One-dimensional site-response analysis................................................................................... 13
Figure 2-5: The lumped-mass and finite element models for site-response analysis [figure adopted from
Hashash et al. (2010) and modified] ........................................................................................................... 16
Figure 2-6: Ground motion input in time-domain one-dimensional site-response analysis ....................... 17
Figure 2-7: The hysteresis loop for the MAT_HYSTERETIC model for a given input strain, and the
corresponding backbone curve.................................................................................................................... 20
Figure 2-8: Sample comparisons between the numerical and observed site response from Dickenson
(1994) .......................................................................................................................................................... 22
Figure 2-9: Sample comparisons between the numerical and observed site response from Chang (1996) 22
Figure 2-10: Sample comparisons of numerical and experimental data from Borja et al. (1999) .............. 23
Figure 2-11: Section view and instrumentation of the Turkey Flat site (Kwok et al., 2008) ..................... 24
Figure 2-12: Blind predictions at locations V1 and D2 using different numerical programs and the
corresponding residuals (Kwok et al., 2008) .............................................................................................. 25
Figure 3-1: Decrease in seismic demand due to SSI effects in idealized design response spectra [adopted
from Mylonakis and Gazetas (2000) and modified] ................................................................................... 30
Figure 3-2: A typical code design spectrum and response spectra from some recorded strong ground
motions (Mylonakis and Gazetas, 2000)..................................................................................................... 31
Figure 3-3: Steps in a substructuring analysis (Kramer and Stewart, 2004) ............................................... 33
Figure 3-4: Substructure methods of SSI analysis (Ostadan, 2006a) .......................................................... 34
Figure 3-5: Finite element model for soil-structure interaction analysis using the direct method .............. 35
Figure 3-6: Illustration of the domain reduction method (Jeremic et al., 2013) ......................................... 38
Figure 3-7: Substructuring in the flexible-volume method in SASSI (Ostadan, 2006a) ............................ 40
Figure 3-8: Sample model for SSI analysis using the effective seismic input in LS-DYNA ..................... 46
Figure 3-9: Layout of the buildings considered by Çelebi (1993a) and Çelebi (1993b) for examining SSSI
.................................................................................................................................................................... 50
Figure 4-1: Soil profiles used for site-response analyses ............................................................................ 52
xvLL
Figure 4-2: Acceleration time series of the input ground motions used for WUS soil profiles .................. 55
Figure 4-3: Acceleration response spectra and Fourier amplitude spectra of the input ground motions used
for WUS soil profiles .................................................................................................................................. 56
Figure 4-4: Acceleration time series of the input ground motions used for CEUS soil profiles................. 57
Figure 4-5: Acceleration response spectra and Fourier amplitude spectra of the input ground motions used
for CEUS soil profiles ................................................................................................................................. 58
Figure 4-6: Backbone and damping curves of the materials used in site-response analysis ....................... 61
Figure 4-7: Fourier transforms of the surface accelerations for site W1 subjected to ground motion WP2
(left) and site W2 subjected to ground motion WO2 (right) ....................................................................... 81
Figure 4-8. Peak acceleration and peak strain profiles (left) and acceleration response spectra at the
surface (right) for site W2 subjected to ground motion WP1 ..................................................................... 83
Figure 4-9: Hysteresis loops of three layers in site W2 showing an abrupt change in the peak strains when
subjected to ground motion WP2 ................................................................................................................ 84
Figure 4-10: Examination of the hysteresis loops of layers involved in abrupt changes in the peak strains
in site W2 subjected to ground motion WP2 .............................................................................................. 85
Figure 4-11: Acceleration time series at 50m depth (left) and the peak acceleration profile (right) for site
W1 subjected to ground motion WP2 ......................................................................................................... 86
Figure 4-12: Outcrop and within ground motions....................................................................................... 88
Figure 4-13: Surface acceleration response spectra in the X direction for uniaxial and biaxial analysis
performed for sites W1 and W2 .................................................................................................................. 90
Figure 5-1: Lumped mass structures, LM1 and LM2, used for the verification analyses of SASSI and LS-
DYNA ......................................................................................................................................................... 94
Figure 5-2: SASSI finite element models of LM1 and LM2 ...................................................................... 96
Figure 5-3: LS-DYNA finite element models of LM1 (left) and LM2 (right) ........................................... 97
Figure 5-4: LS-DYNA finite element model for SSI analysis of LM2 using the direct method ................ 97
Figure 5-5: LS-DYNA finite element models for the effective seismic input method: site-response
analysis (left) and SSI analysis (right) ........................................................................................................ 99
Figure 5-6: 1% damped acceleration response spectra (left) and Fourier transforms of the surface free-
field accelerations computed using site-response analysis and the SSI analysis for LM2........................ 100
Figure 5-7: Response spectra for 1% damping of the roof accelerations of the fixed-base models of LM1
(left) and LM2 (right) calculated using SASSI and LS-DYNA ................................................................ 101
Figure 5-8: Response spectra for 1% damping of the horizontal acceleration at center of the basemat (left)
and the vertical acceleration at the edge of the basemat (right) for SSI analysis of the rigid basemat alone
.................................................................................................................................................................. 101
xviLL
Figure 5-9: Acceleration amplification spectra from the surface free field of the soil to the roof of LM1
.................................................................................................................................................................. 102
Figure 5-10: Deconstruction of the horizontal roof accelerations of LM1 subjected to the Ormsby wavelet
.................................................................................................................................................................. 103
Figure 5-11: Acceleration amplification spectra from the surface free field of the soil to the roof of LM2
.................................................................................................................................................................. 105
Figure 5-12: Deconstruction of the horizontal roof accelerations of LM2 subjected to the Ormsby wavelet
.................................................................................................................................................................. 106
Figure 5-13: Acceleration response spectra at the roof of MS1F_2 subjected to the JOS_L ground motion
for the preliminary linear analysis ............................................................................................................ 110
Figure 5-14: Acceleration response spectra at the roof of MS1F_2 calculated using strain-compatible soil
properties in SASSI, and nonlinear soil properties in LS-DYNA............................................................. 112
Figure 5-15: 5% damped acceleration response spectra at the roof of MS1F_2 subjected to the JOS_L
ground motion calculated using strain compatible soil properties in SASSI and nonlinear soil properties in
LS-DYNA but ignoring nonlinear soil behavior at the vicinity of the footings ....................................... 113
Figure 5-16: Acceleration response spectra at the roof of MS1F_2 equipped with grade beams, calculated
using strain-compatible soil properties in SASSI, and nonlinear soil properties in LS-DYNA ............... 115
Figure 5-17: Time series of the horizontal displacement of two adjacent footings (along the X direction)
of MS1F_2 for the PRI_H ground motion ................................................................................................ 116
Figure 5-18: Screenshots of the deformation at the footings of the MS1F_2 for the PRI_H ground motion
at about 7 sec............................................................................................................................................. 117
Figure 5-19: Acceleration response spectra at the roof of MS2F subjected to the JOS_L ground motion
calculated using strain-compatible moduli and 2% damping ratio for the soil ......................................... 117
Figure 5-20: Acceleration response spectra at the roof of MS2F calculated using strain-compatible soil
properties in SASSI, and nonlinear soil properties in LS-DYNA............................................................. 119
Figure 5-21: Response spectra of the accelerations from structural deformations calculated for MS2F
using strain-compatible properties in SASSI and nonlinear properties in LS-DYNA .............................. 120
Figure 5-22: Acceleration response spectra calculated at the roof of LM2 using SASSI for two sets of
analysis frequencies, and using LS-DYNA .............................................................................................. 122
Figure 5-23: Sensitivity of the rocking response of MS2F calculated using SASSI to the number of nodes
in the basemat ........................................................................................................................................... 124
Figure 5-24: Description of the column-basemat connection of LM2 as modeled in SASSI and the beam
cross-section properties used to model the three different connections employed in Section 5.4.3 ......... 125
xL[
Figure 5-25: SASSI response of the LM2 structure for various structure-basemat connections considered
in Section 5.4.3, and the LS-DYNA response .......................................................................................... 125
Figure 5-26: Sensitivity of the rocking accelerations of MS2F to site layering in the equivalent-linear
method ...................................................................................................................................................... 127
Figure 5-27: Sensitivity of the LS-DYNA response of LM2 to hourglass effects .................................... 128
Figure 5-28: Screenshots of the LS-DYNA response (500 times amplified) of LM2 with and without
hourglass control ....................................................................................................................................... 129
Figure 6-1: The UC Davis centrifuge ....................................................................................................... 133
Figure 6-2: Photographs of the experimental model building process. Courtesy of the City Block team 135
Figure 6-3: Rendering of the horizontal uniaxial shake table used at the UC Davis centrifuge facility
(courtesy of the staff at the Center for Geotechnical Modeling, UC Davis) ............................................. 138
Figure 6-4: FSB-2 container used in the City Block experiments (courtesy of the staff at the Center for
Geotechnical Modeling, UC Davis) .......................................................................................................... 138
Figure 6-5: Photograph of the building models in the laminar box during Test 1 .................................... 139
Figure 6-6: Photograph of the building models in the laminar box during Test 2 .................................... 140
Figure 6-7: Models used in Test 1 and Test 2 ........................................................................................... 143
Figure 6-8: Elevations of experimental models in the shaking direction, with dimensions in model scale
(Chen et al., 2010) .................................................................................................................................... 144
Figure 6-9: Designs of the three ‘fuses’ used in MS1F_1 and MS3F, dimensions in model scale (Chen et
al., 2010) ................................................................................................................................................... 145
Figure 6-10: Photograph of the building models in the laminar box during Test 3 .................................. 149
Figure 6-11: Plan view of the experimental models in the laminar box during Test 3 (dimensions in model
scale) (Mason et al., 2010c) ...................................................................................................................... 150
Figure 6-12: Elevation view of the experimental models in the laminar box during Test 3 (dimensions in
model scale) (Mason et al., 2010c) ........................................................................................................... 150
Figure 6-13: Final design of model MS2F_iSSSI (Mason et al., 2010c).................................................. 154
Figure 6-14: Dimensions of the one-story model in Test 3 (Mason et al., 2010c) ................................... 155
Figure 6-15: Photographs of fully constructed one-story model in Test 3................................................ 156
Figure 6-16: SASSI numerical models of individual buildings considered during the planning of Test 4
(arrows indicate direction of shaking) ...................................................................................................... 160
Figure 6-17: SASSI numerical models of the building arrangements, CBL-1 and CBL-2, considered
during the planning phase of Test 4 (arrows indicate direction of shaking) ............................................. 160
Figure 6-18: Numerical models of building arrangements, CBL-3 to CBL-5, considered during the
planning phase of Test 4 (arrows indicate direction of shaking) .............................................................. 161
x[
Figure 6-19: Photograph of the building models in the laminar box during Test 4 .................................. 162
Figure 6-20: Locations of the models in the laminar box during Test 4 (dimensions in model scale)
(Trombetta et al., 2011) ............................................................................................................................ 163
Figure 6-21: Differences between the Test 3 and Test 4 beam fuses (Trombetta et al., 2011) ................ 165
Figure 6-22: Photograph of the first centrifuge model on the arm shortly after the accidental oil leak ... 167
Figure 7-1: Ground motions selected, and the nomenclature used for the experimental models during data
analysis and numerical simulations performed for Tests 3 and 4 [arrows indicate direction of shaking;
renderings adopted from Trombetta (2013) and modified]....................................................................... 171
Figure 7-2: Five percent damped acceleration response spectra of the ground motions selected for data
analysis and numerical simulations........................................................................................................... 172
Figure 7-3: Acceleration time series of the ground motions selected for data analysis and numerical
simulations ................................................................................................................................................ 172
Figure 7-4: Modulus reduction and damping curves for sand calculated by Seed and Idriss (1970) ....... 176
Figure 7-5: LS-DYNA finite element model for the site-response analysis of the City Block soil profile
.................................................................................................................................................................. 178
Figure 7-6: Acceleration response spectra of the considered input ground motions (recorded at the base of
the laminar box) in Tests 1 through 4 ....................................................................................................... 179
Figure 7-7: Acceleration response spectra of the input at the base of the laminar box (left) and surface-
free-field-to-base transfer functions (right) calculated for six applications of the JOS_L ground motion
during Test 3 ............................................................................................................................................. 180
Figure 7-8: Free-field surface acceleration response spectra calculated from experimental data and
numerical simulations for the considered ground motions in Tests 3 and 4 ............................................. 182
Figure 7-9: Small-strain and strain-compatible shear moduli and damping ratios used for SSI analyses
using SASSI for simulation of Tests 3 and 4 ............................................................................................ 184
Figure 7-10: SASSI numerical models of the baseline models of MS1F_2 and MS2F............................ 186
Figure 7-11: SASSI numerical model of the in-plane SSSI arrangement in Test 3 .................................. 186
Figure 7-12: SASSI numerical model of the anti-plane SSSI arrangement in Test 4 ............................... 187
Figure 7-13: SASSI numerical model of the in-plane SSSI arrangement in Test 3 .................................. 187
Figure 7-14: LS-DYNA finite element model for soil-structure interaction analysis of the cSSSI
arrangement during Test 4 ........................................................................................................................ 190
Figure 7-15: Preliminary model of MS1F_2 in LS-DYNA with explicit modeling of the beam fuses using
shell elements (left) and the close-up of one of the beam fuses (right) .................................................... 191
Figure 7-16: Close-up isometric views of the SSI models of MS1F_2 (left) and MS2F (right) in LS-
DYNA (soil elements far away from the structure are not shown for clarity) .......................................... 192
xxL
Figure 7-17: Close-up isometric view of the SSI model for the in-plane SSSI case of Test 3 in LS-DYNA
(soil elements far away from the structures are not shown for clarity) ..................................................... 193
Figure 7-18: Close-up isometric view of the SSI model for the anti-plane SSSI case of Test 4 in LS-
DYNA (soil elements far away from the structures are not shown for clarity) ........................................ 193
Figure 7-19: Close-up isometric view of the SSI model for the combined SSSI case of Test 4 in LS-
DYNA (soil elements far away from the structures are not shown for clarity) ........................................ 194
Figure 7-20: Response spectra of the roof acceleration of MS1F_2 from numerical simulations using
SASSI and LS-DYNA, and experimental observations for JOS_L and LCN ground motions ................ 198
SASSI and LS-DYNA, and experimental observations for SCS_H and PRI_H ground motions ............ 199
Figure 7-22: Moment-rotation response of the footings of MS1F_2 in various SSSI arrangements,
calculated using LS-DYNA (separation allowed) analyses for the JOS_L ground motion ...................... 201
Figure 7-23: Moment rotation-response of the footings of MS1F_2 in various SSSI arrangements,
calculated using LS-DYNA (separation allowed) analyses for the LCN ground motion ......................... 201
calculated using LS-DYNA (separation allowed) analyses for the SCS_H ground motion ..................... 202
calculated using LS-DYNA (separation allowed) analyses for the PRI_H ground motion ...................... 202
Figure 7-26: Vertical force-displacement response of the footings of MS1F_2 in various SSSI
arrangements, calculated using LS-DYNA (separation allowed) analyses for JOS_L ground motion .... 204
arrangements, calculated using LS-DYNA (separation allowed) analyses for LCN ground motion ....... 204
arrangements, calculated using LS-DYNA (separation allowed) analyses for SCS_H ground motion ... 205
arrangements, calculated using LS-DYNA (separation allowed) analyses for PRI_H ground motion .... 205
Figure 7-30: Response spectra of the roof acceleration of MS2F from numerical simulations using SASSI
and LS-DYNA, and experimental observations for JOS_L and LCN ground motions ............................ 207
Figure 7-31: Response spectra of the roof acceleration of MS2F from numerical simulations using SASSI
and LS-DYNA, and experimental observations for SCS_H and PRI_H ground motions ........................ 208
Figure 7-32: LS-DYNA numerical model of the iSSSI_B case with MS1F_2 on the left and MS2F_B on
the right (soil elements are not shown for clarity) .................................................................................... 209
Figure 7-33: Isometric view of the LS-DYNA numerical model used for pseudo-static analysis of the
iSSSI_B case ............................................................................................................................................. 210
xxLL
Figure 7-34: Moment-rotation response and the vertical force-displacement response of the restrained
MS1F_2 footings in the baseline and iSSSI_B cases ............................................................................... 211
Figure 7-35: Lateral force-displacement response of the MS1F_2 superstructure in the baseline case and
the iSSSI_B case ....................................................................................................................................... 212
Figure 7-36: Response spectra of the roof acceleration of MS1F_2 from numerical simulations using LS-
DYNA for four ground motions................................................................................................................ 214
Figure 7-37: Moment-rotation responses of the restrained footings of MS1F_2 in the baseline and
iSSSI_B cases calculated for the four ground motions ............................................................................. 215
Figure 7-38: Response spectra of the roof acceleration of MS1F_1 calculated using numerical simulations
in LS-DYNA for four ground motions ..................................................................................................... 216
Figure A-1: Stress-strain loops at three different depths for site E1 subjected to ground motion
EO1............................................................................................................................................................246
Figure A-2: Stress-strain loops at three different depths for site E1 subjected to ground motion EO2 .... 247
Figure A-3: Stress-strain loops at three different depths for site E1 subjected to ground motion EO3 .... 248
Figure A-4: Peak acceleration and peak strain profiles for site E1 subjected to ground motion EO1 ...... 249
Figure A-7: Acceleration response spectra of the base outcrop motion, and accelerations at three different
depths for site E1 subjected to ground motion EO1 ................................................................................. 252
Figure A-10: Acceleration response histories at three different depths for site E1 subjected to ground
motion EO1 ............................................................................................................................................... 255
motion EO2 ............................................................................................................................................... 256
motion EO3 ............................................................................................................................................... 257
Figure A-13: Spectral amplification of accelerations from the base to the surface of site E1 for ground
motions EO1, EO2 and EO3 ..................................................................................................................... 258
Figure A-14: Stress-strain loops at three different depths for site E2 subjected to ground motion EO1 .. 259
xxiLL
Figure A-17: Peak acceleration and peak strain profiles for site E2 subjected to ground motion EO1 .... 262
Figure A-20: Acceleration response spectra of the base outcrop motion, and accelerations at three
different depths for site E2 subjected to ground motion EO1 ................................................................... 265
motion EO1 ............................................................................................................................................... 268
motion EO2 ............................................................................................................................................... 269
motion EO3 ............................................................................................................................................... 270
Figure A-26: Spectral amplification of accelerations in site E2 from the base to the surface for ground
motions EO1, EO2 and EO3 ..................................................................................................................... 271
Figure A-27: Stress-strain loops at three different depths for site W1 subjected to ground motion
WO1 .......................................................................................................................................................... 272
WO2 .......................................................................................................................................................... 273
WO3 .......................................................................................................................................................... 274
Figure A-30: Peak acceleration and peak strain profiles for site W1 subjected to ground motion
WO1 .......................................................................................................................................................... 275
WO2 .......................................................................................................................................................... 276
WO3 .......................................................................................................................................................... 277
different depths for site W1 subjected to ground motion WO1 ................................................................ 278
xxiY
Figure A-36: Acceleration response histories at three different depths for site W1 subjected to ground
motion WO1.............................................................................................................................................. 281
motion WO2.............................................................................................................................................. 282
motion WO3.............................................................................................................................................. 283
Figure A-39: Spectral amplification of accelerations in site W1 from the base to the surface for ground
motions WO1, WO2 and WO3 ................................................................................................................. 284
WP1 .......................................................................................................................................................... 285
WP2 .......................................................................................................................................................... 286
WP3 .......................................................................................................................................................... 287
WP1 .......................................................................................................................................................... 288
WP2 .......................................................................................................................................................... 289
WP3 .......................................................................................................................................................... 290
different depths for site W1 subjected to ground motion WP1 ................................................................. 291
motion WP1 .............................................................................................................................................. 294
motion WP2 .............................................................................................................................................. 295
motion WP3 .............................................................................................................................................. 296
xxY
motions WP1, WP2 and WP3 ................................................................................................................... 297
WO1 .......................................................................................................................................................... 298
WO2 .......................................................................................................................................................... 299
WO3 .......................................................................................................................................................... 300
WO1 .......................................................................................................................................................... 301
WO2 .......................................................................................................................................................... 302
WO3 .......................................................................................................................................................... 303
motion WO1.............................................................................................................................................. 307
motion WO2.............................................................................................................................................. 308
motion WO3.............................................................................................................................................. 309
motions WO1, WO2 and WO3 ................................................................................................................. 310
WP1 .......................................................................................................................................................... 311
WP2 .......................................................................................................................................................... 312
WP3 .......................................................................................................................................................... 313
xxvL
WP1 .......................................................................................................................................................... 314
WP2 .......................................................................................................................................................... 315
WP3 .......................................................................................................................................................... 316
Figure A-72: Acceleration response spectra of the input motion, and accelerations at three different depths
for site W2 subjected to ground motion WP1 ........................................................................................... 317
motion WP1 (unequal scales used for clarity) .......................................................................................... 320
motion WP2 (unequal scales used for clarity) .......................................................................................... 321
motion WP3 .............................................................................................................................................. 322
motions WP1, WP2 and WP3 ................................................................................................................... 323
Figure B-1: Time series of an Ormsby wavelet ........................................................................................ 326
Figure B-2: Fourier transform of an Ormsby wavelet .............................................................................. 326
Figure B-3: 1% damped response spectrum of an Ormsby wavelet..........................................................326
xxvLL
LIST OF TABLES
Table 4-1: Input ground motions used for site-response analysis ............................................................... 54
Table 4-2: Model parameters used to define the soil materials in DEEPSOIL ........................................... 61
Table 4-3: Layer thicknesses and captured maximum frequencies in each material .................................. 63
Table 4-4: Time periods of the soil profiles from modal analysis .............................................................. 64
Table 5-1: Lumped masses and beam cross section properties of the structural models used for analyses
using SASSI and LS-DYNA ....................................................................................................................... 94
Table 5-2: Analysis time in minutes for elastic SSI analyses of LM2 subjected to the Ormsby wavelet. 107
Table 6-1: Scale factors used in centrifuge testinga (Kutter, 1995) .......................................................... 132
Table 6-2: Target design parameters for model structures at prototype scale (Chen et al., 2010) ........... 141
Table 6-3: Specifications of superstructures at the model scale (Chen et al., 2010) ................................ 142
Table 6-4: Specifications of foundations at the model scale (Chen et al., 2010)...................................... 142
Table 6-5: Prototype-scale properties of the designed experimental models calculated using the numerical
models in OpenSees and hammer tests (Chen et al., 2010) ...................................................................... 145
Table 6-6: Test 2 ground motions and their sequence (only 55g ground motions included) .................... 148
Table 6-7: Dimensionless parameters governing the design of the two-story shear wall model,
MS2F_iSSSI (Mason et al., 2011) ............................................................................................................ 153
Table 6-8: Test 3 ground motions and their sequence (55g earthquake ground motions only) (Mason et
al., 2010c) ................................................................................................................................................. 158
Table 6-9: Test 4 ground motions and their sequence (only 55g ground motions applied to models with
original floor masses are included) (Trombetta et al., 2011) .................................................................... 168
Table 7-1: Soil layer properties used for site-response analysis in Tests 3 and 4 ..................................... 175
Table 7-2: DEEPSOIL curve parameters used for the Test 4 site-response analysis................................ 177
Table 7-3: Material properties used in the SASSI analyses ...................................................................... 188
Table 7-4: Description of the numerical model of MS1F_2a (dimensions in prototype scale) ................. 188
Table 7-5: Description of the numerical model of MS2Fa (dimensions in prototype scale) ..................... 189
Table 7-6: Comparison of the prototype scale fixed-base time periods calculated from SASSI and hammer
tests for all the models .............................................................................................................................. 195
xxL[
SECTION 1
INTRODUCTION
1.1 Background
Traditional structural design involves the assumption that structures are founded on rigid ground.
This assumption is reasonable in many cases, such as flexible buildings constructed on stiff soil sites.
Accounting for soil flexibility in the analysis of such buildings does not generally affect the response in a
significant way. For stiff, heavy buildings (and other structures), soil flexibility can change the response
due to soil-foundation-structure interaction (SFSI) or, as it is more commonly known, soil-structure
interaction (SSI). Soil-structure interaction effects comprise 1) changes in the dynamic behavior of the
structure, namely, an increase in the time period and damping, and 2) modification of the input ground
motion with respect to a free soil surface, or free field. The former is known as inertial interaction, and the
latter as kinematic interaction. Ignoring SSI effects in building design has traditionally been assumed to
be conservative. However, studies [e.g., Mylonakis and Gazetas (2000)] have shown that this may not
always be the case. Accounting for SSI effects in buildings is not mandatory in the United States,
although ASCE Standard 7 (ASCE, 2010) and FEMA 440 (FEMA, 2005) provide guidelines to include
SSI effects in design. Although SSI effects may not be significant for most buildings, and the cost of SSI
analysis may not be recovered by reducing the sizes of beams, columns and walls in buildings, they are of
great importance for the design of safety-related nuclear structures, including nuclear power plants.
Nuclear structures tend to have large plan dimensions, be stiff and be heavy attributes that can lead to
significant SSI. Soil-structure interaction analysis of nuclear structures is routine and generally follows
the procedures set forth in ASCE Standard 4, Seismic Analysis of Safety-Related Nuclear Structures
(ASCE, 1998; ASCE, forthcoming).
An extension of SSI is interaction of adjacent structures. During an earthquake, the presence of
one structure may affect the dynamic response of another, resulting in structure-soil-structure interaction
(SSSI). Structure-soil-structure interaction may be important for the performance assessment of buildings
in dense urban areas but little is known since there are well-documented field and/or experimental data.
Structure-soil-structure interaction is not implemented in practice and there are no related provisions in
building codes.
Site-response analysis typically involves one-dimensional analysis of a soil column composed of
multiple layers, spanning from the rock horizon at depth up to the ground surface. Site-response analysis
is a precursor to SSI analysis. It is also used to propagate rock motions at depth through soil columns to
establish free-field surface motions for analysis of surface- or near-surface-mounted structures. Site-
1
response analysis is often bypassed for building design in the United States by using ground motion
prediction (or attenuation) relationships for seismic hazard calculations that include an empirical
component associated with shear wave velocity in the upper 30m of the soil column the part of the
column thought to most influence free-field surface response.
1.2 Motivation
The earthquake engineering profession in the United States is embracing performance-based

earthquake engineering. The FEMA P-58 products (FEMA, 2012) provide the background knowledge,
implementation advice and calculation tools to enable calculations of loss in individual buildings either
for an earthquake of a specific intensity or over a period of time (based on annualized calculations of
loss). The knowledge, advice and tools rely on the correct calculation of the earthquake shaking input to
the building. Although SSI effects will likely reduce losses in buildings during earthquakes in most cases,
the percentage reduction is unknown, and indeed, in certain cases that are yet to be identified, losses may
increase due to SSI effects. An accurate calculation of loss will likely require explicit consideration of SSI
and structure-soil-structure interaction (SSSI) effects.
Seismic probabilistic risk assessment is now being routinely used to assess the adequacy of new
and existing nuclear power plants in the United States. Soil-structure interaction (and perhaps SSSI)
analysis is a key component of the risk assessment calculations because it influences demands on
structural components and floor spectra with which equipment vulnerability is judged.
This report has the goal of advancing the profession’s knowledge of SSI and SSSI effects in both
buildings in dense urban regions and in safety-related nuclear structures. The following sections present
the motivation and objectives of the studies presented in this report.
1.2.1 Assessment of numerical methods in site-response and soil-structure interaction analysis
Site response and SSI are complex nonlinear phenomena involving three-dimensional wave
propagation in soils. Site-response analyses have been traditionally performed in the linear domain
assuming one-dimensional vertically propagating shear waves in soil profiles with infinitely horizontal
soil layers. The most widely used site-response analysis program is the equivalent-linear, frequency-
domain program, SHAKE (Schnabel et al., 2012).
Soil-structure interaction analysis of buildings is often performed by modeling the soil beneath
the foundations with a set of springs that simulate the flexibility of the soil. The properties of these
springs are either calculated using established analytical solutions, or calibrated using experimental data.
SSI analyses of nuclear structures are performed using more sophisticated, yet linear, three-dimensional
2
methods. Of the several methods developed, the most widely used is the numerical program, SASSI
(Lysmer et al., 1999), which performs a linear analysis in the frequency domain. These linear methods of
site-response and SSI analyses have been used for several decades due to their simplicity and
computational efficiency. Although the use of these methods will likely result in accurate predictions for
low intensity earthquakes, where the soil behavior is essentially linear, their utility for intense ground
motions resulting in large soil strains is unproven.
Significant developments in finite-element methods as well as the availability of computational
resources have led to the development of nonlinear methods for site-response and soil-structure
interaction analyses. Comprehensive simulations of nonlinear site response and SSI, including foundation
gapping and sliding, are now possible on personal computers with commercially available programs such
as ABAQUS (Dassault Systèmes, 2005) and LS-DYNA (LSTC, 2013). However, these methods are not
yet well established and require much greater computational resources than the frequency-domain
programs. Severe earthquake shaking can produce highly nonlinear responses in soils, for which verified
and validated nonlinear analysis codes should be better suited for calculating responses than the
traditional linear codes.
Nuclear power plant structures are designed for earthquake shaking with return periods between
10,000 and 50,000 years, which result in significant ground accelerations across much of the United
States, and significant strains in soils at the sites of many plants. The accuracy of the traditional
equivalent-linear methods of site-response and SSI analysis at soil sites experiencing significant bedrock
shaking is not known. This leaves the analyst in a conundrum for SSI analysis at sites involving
significant shaking and substantially nonlinear soil behavior, namely, either 1) use traditional equivalent-
linear tools that are likely incapable of capturing the response of highly nonlinear soils or foundation
gapping and sliding that may accompany intense ground motions, or 2) use nonlinear material models and
computer codes that can capture explicitly nonlinear soil behavior and foundation gapping and sliding,
but which have seen very little use in the past and with which regulators have no substantial experience.
To improve the knowledge and advance practice in site-response and SSI analysis, a series of
studies are performed with the following objectives:
1. Benchmark nonlinear site-response and SSI analysis programs against linear programs for elastic
soils, and nonlinear soils at small strains.
2. Compare the results of linear and nonlinear site-response analysis programs for a range of soil
profiles and ground motion intensities commonly encountered in nuclear construction.
3. Compare the results of linear and nonlinear SSI analysis programs for buildings with different
foundation configurations on very soft soil, subjected to a range of ground motion intensities.
3
4. Examine the differences between the predictions of linear and nonlinear programs, and provide
guidance for practicing engineers on the use of these programs.
5. Identify the various challenges and pitfalls that can be encountered performing site-response and
SSI analysis, and provide guidance in dealing with or avoiding these challenges and pitfalls.
1.2.2 Seismic structure-soil-structure interaction in dense urban environments
The US National Science Foundation (NSF) funded a research project titled ‘Seismic
Performance Assessment in Dense Urban Areas’ to study the effects of SSI and SSSI in dense urban
regions. This project, also known as the NEES City Block project or the City Block project, involved a
series of six centrifuge experiments of small-scale model buildings at the NEES@UCDavis geotechnical
centrifuge facility. The first two experiments involved low- to mid-rise buildings designed to have
realistic, code-conforming, dynamic properties with nonlinear behavior at the global level. These
buildings were placed far apart in Test 1 and adjacent to each other in Test 2, and comparisons of the
structural responses from these tests are expected to provide insight into SSSI in such structures. Test 3
and Test 4 involved buildings that were much stiffer than those in Test 1 and Test 2, and carefully
designed to exacerbate SSSI effects. Different structural arrangements (in-plane, anti-plane and combined
in-plane anti-plane) are tested to investigate the mechanisms of interaction between these structures. Test
5 and Test 6 involved idealized linear structures founded on liquefiable soil, and the effect of interaction
on these structures is examined during liquefaction. The models in each test are subjected to several
ground motions of different intensities. These ground motions include both ordinary and pulse motions
with forward directivity. The data gathered from these tests has been carefully curated (Mason et al.,
2010b; Mason et al., 2010a; Mason et al., 2011; Trombetta et al., 2011) and analyzed by the City Block
team (Jones, 2013; Mason, 2011; Trombetta, 2013). The study presented in this report only deals with
Test 3 and Test 4 to avoid duplication of prior efforts by other researchers in the City Block project. The
objectives of this study are:
1. Analyze the experimental data from Test 3 and Test 4 to examine the site response and SSSI in
the model buildings.
2. Numerically simulate the experiments of Test 3 and Test 4 using linear, frequency-domain and
nonlinear, time-domain programs, and examine the numerical results.
3. Make qualitative comparisons of the experimental and numerical results and assess the
predictions of the numerical programs.
4
1.3 Report organization
This report is organized into nine section with contents as follows:

 Section 2 describes the commonly used numerical models that simulate the shear behavior of
soils, presents the evolution, state-of-the-art, and limitations of one-dimensional site-response
analysis, and describes some studies of three-dimensional site response.
 Section 3 presents a similar review of SSI analysis, emphasizes the latest methods developed for
SSI analyses including nonlinear time-domain methods such as the direct method and the domain
reduction method, and reviews previous studies on SSSI.
 Section 4 assesses the current, industry-standard, site-response analysis programs for a range of
soil profiles and ground motion characteristics. It provides a description of the selection of the
soil profiles and ground motions, and numerical modeling, discusses the results, makes
statements on the practical issues faced using each of the programs, provides guidelines for the
input of ground motion to site-response analysis, and studies biaxial site-response analysis.
 Section 5 presents an assessment of the current, industry-standard, SSI programs, benchmarks
time-domain programs against frequency-domain programs for linear elastic analyses, compares
results from linear, frequency-domain analyses with strain-compatible soil properties and
nonlinear time-domain analyses, describes results, and makes statements on the various practical
issues encountered using these programs.
 Section 6 presents a brief description of centrifuge testing including its advantages and
disadvantages, an introduction to the NEES@UC Davis centrifuge facility, the experimental
procedures of Tests 2, 3 and 4, including the design and construction of models, and the input
ground motion sequence for each test.
 Section 7 presents experimental results from Test 3 and Test 4 and makes comparisons with
numerical simulations performed using linear frequency-domain and nonlinear time-domain
programs. The section provides a description of the numerical models, examines the experimental
and numerical results, presents an exploratory study on local SSSI effects between two particular
types of foundation systems, and draws conclusions on the effect of SSSI on the global and local
structural response.
 Section 8 presents the overarching conclusions for each of the studies performed in this report.
The section presents recommendations for future research in the domains of site-response, SSI
and SSSI analysis.
5
SECTION 2
SITE-RESPONSE ANALYSIS: A REVIEW
2.1 Dynamic behavior of soils during earthquakes
Earthquakes induce an irregular cyclic loading in soil that results in a range of stresses and
strains. Soils close to the ground surface and sufficiently far away from built structures are known to
deform predominantly in shear during an earthquake (Kramer, 1996), and a substantial amount of
analytical and experimental research has therefore been performed to characterize the cyclic shear
behavior of soils. Studies have also been performed to understand the three-dimensional behavior of soils
for more complex situations involving three-dimensional stress fields. Numerical material models have
been developed from these studies and implemented in computer programs that are used in seismic
analyses. A few of these material models, especially those used in site-response and soil-structure
interaction analyses, are described in this section.
2.1.1 Experimental response of cyclically loaded soils
Soils are known to exhibit a nonlinear hysteretic behavior when cyclically loaded in shear. For
monotonically increasing loads, they exhibit nonlinear behavior with gradual softening, which manifests
as a decrease in shear modulus with strain (Hardin and Drnevich, 1972). Several field and laboratory
testing methods have been developed to characterize this behavior, which are classified into low-strain
tests and high-strain tests. Low-strain tests are performed at strain levels that are low enough so that no
nonlinear behavior is induced in the soil. The shear strain in these tests is typically below 0.001%
(Kramer, 1996). Many of these tests involve calculation of the wave velocities through low-amplitude
wave propagation to calculate the elastic properties (moduli and wave velocities). High-strain tests, such
as the cone penetration test (CPT), are also used to calculate elastic properties. The CPT involves a steady
penetration of the standard cone penetrometer into the soil, which has a cylindrical body with a conical tip
equipped with a load cell measuring the tip resistance. The observed tip resistance, along with established
empirical relationships, is used to calculate the shear modulus and relative density of the soil. The CPT is
a quick and inexpensive procedure that can be performed both in the field and in the laboratory, making it
one of the most widely used methods in geotechnical engineering practice (Kramer, 1996). Cyclic
nonlinear behavior of the soil is characterized through high-strain tests. The most widely used tests for
this purpose are the cyclic triaxial test and the cyclic direct simple shear test, which involve a pseudo-
static stress-controlled or strain-controlled cyclic loading of a vertically loaded soil specimen. Model tests
7
at a larger scale are also possible, and can be performed using a laminar box excited on a shaking table.
Further details regarding these tests and more testing procedures can be found in Kramer (1996).
2.1.2 Numerical material models for dynamic soil behavior
Soil materials exhibit a diverse range of complex constitutive behaviors, which makes the
development of numerical models quite challenging. Several models of varying complexity have been
developed in earthquake engineering, which are categorized by (Kramer, 1996) as: 1) equivalent-linear
models; 2) cyclic nonlinear models; and 3) advanced constitutive models. Equivalent-linear models and
cyclic nonlinear models are based on shear behavior only, and are employed for one-dimensional site-
response analyses. Advanced constitutive models incorporate the three-dimensional nonlinear behavior of
the soil, and are used for more comprehensive site-response analyses and soil-structure interaction
analyses.
2.1.2.1 Equivalent-linear models
Hardin and Drnevich (1972) observed that soils subjected to cycles of symmetrical shear loads of
constant amplitude show a hysteretic stress-strain behavior as illustrated in Figure 2-1.
Figure 2-1: Calculation of equivalent-linear parameters: secant shear modulus and effective
damping ratio
The instantaneous slope of the stress-strain curve (or the tangent shear modulus, Gtan ) decreases
with the increase in strain, indicating softening in the soil. Also a closed loop indicates the dissipation of
energy that is equal to the area of the loop for each loading cycle. The equivalent-linear model simplifies
8
this behavior by characterizing the hysteresis loop by 1) an equivalent stiffness or secant modulus ( Gsec ),
and 2) an equivalent damping ratio that is directly proportional to the hysteretic energy dissipated (Figure
2-1). The secant modulus is calculated as
c
Gsec  (2-1)
c
where  c and  c are the amplitudes of shear stress and shear strain respectively, and the equivalent
damping ratio  is given by
WD 1 Aloop
  (2-2)
4 WS 2 Gsec c2
where WD is the energy dissipated in each load cycle, WS is the maximum strain energy during loading,
Aloop is the area enclosed in the loop (Kramer, 1996), and other terms are defined above. As the
maximum strain of the loading cycles is increased, the secant shear modulus decreases and the effective
damping increases. The largest value of shear modulus corresponds to zero strain and is termed the initial
shear modulus, G0 . The variation of the shear modulus and damping with strain are represented by the
modulus reduction curve and the damping curve respectively (Figure 2-2).
Figure 2-2: Upper bound modulus reduction and damping curves for sand [from Seed and Idriss
(1970)]
The modulus reduction curve, when represented as a stress-strain curve, is called as a backbone
curve. These curves, including the equivalent shear modulus and damping, constitute the equivalent-linear
soil model, which is used in a class of site-response analysis methods known as equivalent-linear methods
9
(Section 2.2.2.2). These parameters depend on the effective mean principal stress for a given soil, and are
known to remain almost constant in the frequency range of loading considered in traditional seismic
analyses (i.e., 0-25 Hz). Given their simplicity and the ease in experimental calibration, significant efforts
have been made to characterize these parameters for various soils.
2.1.2.2 Nonlinear hysteretic models
Nonlinear hysteretic models use the actual hysteresis path during the analysis and enable a more
realistic modeling of the soil shear behavior. The hysteresis path is characterized by 1) the backbone
curve, and 2) a set of rules that govern the hysteresis path during under an irregular cyclic loading.
Various expressions have been proposed for the backbone curve [e.g. the Ramberg-Osgood model
(Ramberg and Osgood, 1943), the hyperbolic model or the Kodner-Zelasko model (Lee and Finn, 1978),
the modified hyperbolic model or the modified Kodner-Zelasko (MKZ) model (Matasovic, 1993)], each
developed to better reproduce the experimental stress-strain data. The most well-known set of hysteresis
rules are based on the Masing rules [proposed by Masing (1926), as per Kramer (1996)] and are called as
the extended Masing rules. These rules are as follows (Kramer, 1996):
1. For initial loading, the stress-strain curve follows the backbone curve [given by Fbb ( ,  ) ]
2. When a stress reversal occurs at a point (  r , r ), the path of the stress-strain curve changes to
  r   r 
 Fbb  
2  2 
which implies that the new curve has the same shape as the backbone curve (with the origin
shifted to the reversal point), but is twice enlarged. The first two rules describe the Masing
behavior (Masing, 1926), which were later extended to better describe the soil behavior under
general cyclic loading and called the extended Masing rules.
3. If an unloading or reloading curve exceeds the past maximum strain and intersects the backbone
curve, it follows the backbone curve until the next reversal.
4. If an unloading or reloading curve crosses another curve from the previous cycle, it follows the
stress-strain curve of the previous cycle.
Since they define the shape of the loops the hysteresis rules significantly affect the nonlinear
response, specifically hysteretic damping. Other sets of rules have also been proposed to better match the
hysteretic damping calculated from laboratory tests than the extended Masing rules [e.g. Pyke (2000)
Muravskii (2005)]. Modifications to the extended Masing rules have also been developed to include pore
10
water pressure effects, such as modulus degradation (decrease in the secant shear modulus in subsequent
cycles) due to excess pore water pressure [e.g. (Matasovic, 1993)].
Figure 2-3: Nonlinear hysteretic model with a hyperbolic backbone curve and extended Masing
rules (Hashash and Park, 2001)
2.2 Site-response analysis
Site-response analysis (SRA) is the process of calculating the response of a soil deposit to an
earthquake ground motion in the absence of structures (called as the free-field response). It is a nonlinear
three-dimensional wave-propagation problem that involves the excitation of a soil deposit by a wave-field
comprising body and surface waves. Seminal site-response analyses were motivated by field observations
indicating higher accelerations on the surface of a soil deposit, as compared to a rock outcrop in the
vicinity (Kanai, 1952). This suggested that the type of soil deposit (or, in general, the local characteristics
of the site) influenced the magnitude of surface acceleration. Presently the main purpose of site-response
analysis is to characterize the effects of the local site conditions such as 1) material properties of the
underlying soil layers, and 2) surface topographies such as cliffs and ridges on the surface free-field
ground motion. Site-response analysis is also a precursor to soil-structure interaction analysis, in which
the free-field site response needs to be calculated before an interaction analysis (Bolisetti, 2010; Kramer,
1996; Wolf, 1985). The present study only deals with the calculation of site response for application in
soil-structure interaction analysis without considering topographic effects.
11
2.2.1 The nature of site response
After the early site-response analysis studies such as Kanai (1952) it was well-understood that the
surface free-field acceleration during an earthquake is affected by the properties of the underlying soil
deposit. As the earthquake waves propagate upwards in the soil profile, selected frequencies in the input
motion are amplified resulting in new peaks in the surface acceleration response spectrum. For weak
ground motions that induce small soil strains (and therefore a nearly linear site response), these
frequencies correspond to the resonant frequencies of the soil deposit. Therefore a proper assessment of
the seismic hazard at a site requires a reasonable estimation of these amplifications, and the frequencies at
which these amplifications occur. As the ground motions become more intense, larger soil strains are
induced in the soil deposit, introducing nonlinear site effects. These effects include 1) decrease in the
resonant frequencies (or lengthening of the time period) of amplification due to the softening of the soil,
and 2) decrease in magnitude of amplification (or sometimes de-amplification) due to hysteretic damping.
Studies investigating the nonlinear effects in site response have also noted a spike-like waveform in the
surface acceleration indicating a high frequency response (Archuleta et al., 2000; Iai et al., 1995;
Porcella, 1980). This was later attributed to the dilatancy of sand, in which the accelerations were
recorded (Iai et al., 1995). While the preceding site-response effects are limited to simple sites resulting in
one-dimensional response, more complex soil deposits result in three-dimensional effects, which are
discussed in Section 2.2.3.
2.2.2 One-dimensional methods of site-response analysis
Most site-response analyses are performed using the one-dimensional methods, which are based
on the following assumptions: a) soil profiles comprise of horizontal layers stacked atop of each other, b)
the soil layers are homogeneous along the horizontal plane, and c) the ground motion incident on the soil
deposit is in the form of vertically propagating shear waves. These assumptions simplify the site-response
phenomenon and enable the use of simplified numerical models for numerical analysis. However the soil
deposits that do not conform to these assumptions need to be analyzed using multi-dimensional methods.
The one-dimensional method is considered reasonable for many civil engineering applications (Kramer,
1996), and is presently the most-widely used in practice. One-dimensional site-response analysis involves
the excitation of a soil profile using the horizontal component of a ground motion, and calculating the
response of the individual soil layers (Figure 2-4). An upward wave propagation analysis typically
involves the input of a rock outcrop motion (ground motion recorded at a nearby, outcropping bedrock) at
the soil-bedrock interface, while the bedrock is replaced by a transmitting boundary to enable the
radiation of outgoing waves. If the ground motion at depth is available from recorded data (termed as a
12
within-profile ground motion), it is directly input at the base of the soil profile while assuming rigid
properties for the bedrock. In a downward wave propagation analysis, the surface free-field motion is
input at the topmost layer of the soil, and the response of underlying layers is calculated. This process
termed as deconvolution (Figure 2-4).
The response of a soil deposit with linear viscoelastic properties can be computed using available
closed-form solutions (Kramer, 1996). Since these solutions are in the form of frequency-dependent
transfer functions, the computations are typically performed in the frequency domain. Nonlinear one-
dimensional site-response analyses are performed using either 1) the equivalent-linear method, or 2) the
nonlinear method. The equivalent-linear method can be employed both in the frequency as well as the
time domains, whereas the nonlinear method can only be implemented in the time domain. A description
of each of these methods is presented in Sections 2.2.2.2 and 2.2.2.3.
Figure 2-4: One-dimensional site-response analysis
2.2.2.1 Evolution of one-dimensional site-response analysis methods
Early site-response studies involved linear one-dimensional simulations to calculate the ground
motion amplification from the bedrock to the soil surface [e.g. Kanai (1952), Idriss and Seed (1967a)].
Subsequent field observations indicating nonlinear soil response led to the development nonlinear site-
response analysis methods. Idriss and Seed (1967a) proposed the use of the equivalent-linear approach,
which involves the approximation of the nonlinear response by a linear analysis with modified material
13
properties. They performed the computations in the time domain and assumed bilinear properties for the
soil. A subsequent work by Hardin and Drnevich (1972) provided substantial insight into the nonlinear
shear behavior of soil, enabling several researchers to develop more nonlinear soil models. Schnabel et al.
(1972) later implemented the equivalent-linear method in the frequency domain and developed the
numerical program SHAKE, which is now widely used in practice. Chen and Joyner (1974) performed
time-domain analyses using a multi-linear model, and in a later study (Joyner and Chen, 1975) modeled
the soil using a rheological model based on the family of material models developed by Iwan (1967).
Several other nonlinear material models, such as the Kodner-Zelasko model (Lee and Finn, 1978), and the
hyperbolic (MKZ) model (Matasovic, 1993), were later developed and incorporated into nonlinear time-
domain programs, which use different numerical integration techniques to calculate the site response.
Some of the well-known nonlinear time-domain programs include DESRA-2 (Lee and Finn, 1978), D-
MOD (Matasovic, 1993), TESS (Pyke, 2000) and DEEPSOIL (Hashash et al., 2011). DESRA-2 uses the
hyperbolic model (or the Kodner-Zelasko model) for the soil, and simulates the soil hysteretic behavior
using extended Masing rules. Matasovic (1993) improved the hyperbolic model by introducing additional
parameters and implemented it in D-MOD along with a pore-water pressure model to simulate modulus
degradation. Hashash and Park (2001) further modified this model to account for pressure dependence of
the soil properties before implementation in DEEPSOIL, and also improved the low-strain damping
behavior using various viscous damping formulations. Yang et al., (2004) developed a web-based
platform for site-response and liquefaction analysis, which uses the open-source finite element program,
OpenSees (Mazzoni et al., 2009; McKenna and Fenves, 2001), along with a plasticity-based soil
constitutive model developed by Elgamal et al. (2002). Commercial finite element programs such as LS-
DYNA (LSTC, 2013) and ABAQUS (Dassault Systèmes, 2005) can also be used to perform site-response
analysis in the time domain by using appropriate constitutive models for the soil. The numerical programs
SHAKE, DEEPSOIL and LS-DYNA are considered in this study, and are described in more detail in
Sections 2.2.2.2.1, 2.2.2.3.1, and 2.2.2.3.2, respectively.
2.2.2.2 Equivalent-linear methods
The equivalent-linear method involves the calculation of an approximate nonlinear response

using a linear analysis with the soil layer properties adjusted accordingly. These layer properties are
calculated by an iterative procedure involving a series of linear analyses that can be performed either in
the frequency or the time domains. A linear analysis is first performed using the initial values of shear
modulus and damping ratio, and the peak strains in the soil layers are computed. An effective shear strain
is calculated for each layer by multiplying the peak shear strain by an effective shear strain ratio. This
14
strain value, along with the modulus reduction and damping curves, is used to update the shear modulus
and damping of each layer. The new values are used to perform another iteration of linear analysis, and
the iterations are continued until the shear strains from consecutive analyses match within a pre-defined
tolerance. The effective shear strain ratio is a key parameter that relates the loading conditions occurring
during the earthquake to those in the laboratory tests that are used to calculate the modulus reduction and
damping curves (Kramer, 1996). A value less than 1 is typically chosen, and Idriss and Sun (1992)
suggest the following expression to calculate the effective shear strain ratio.
M 1
 (2-3)
10
where  is the effective shear strain ratio, and M is the magnitude of the earthquake. The SHAKE
user’s manual (Schnabel et al., 2012) recommends a value of 0.65, which has traditionally been used in
practice. However these recommendations are not validated by published empirical or analytical results as
to the knowledge of the author and are only based on experience matching numerical response with
recorded data. Idriss and Sun (1992) have found this value to vary between 0.4 and 0.75, while Yoshida et
al., (2002) note a much larger range between 0.2 and 1. Dickenson (1994) suggested that the response
obtained using the traditional value of 0.65 for the effective shear strain ratio can be slightly improved by
using a value between 0.35 and 0.55 for M S = 6 to 7 events, and between 0.55 and 0.7 for M S = 7 to 8
events, where M S is the surface-wave magnitude. These studies indicate that an appropriate value of 
depends largely on the characteristics of the input earthquake and not only on the model. Modifications to
the equivalent-linear model have been proposed by researchers [e.g., Assimaki et al. (2000); Yoshida et
al. (2002)], which involve frequency-dependent formulations for the effective shear strain ratio, shear
modulus and damping. However these modified versions are rarely used in practice.
2.2.2.2.1 SHAKE
SHAKE performs one-dimensional equivalent-linear site-response analysis in the frequency
domain. Site-response analysis in SHAKE requires defining 1) the soil layer properties (thickness, mass
density, initial shear modulus and damping ratio), 2) the modulus reduction and damping curves for each
material, 3) the ground motion input including the input location, 4) the analysis parameters (number of
iterations and the effective shear strain ratio), and 5) the output requests. The dynamic material properties
include the modulus reduction (backbone curve) and damping curves that are defined for each material.
The stress-strain behavior of the soil is assumed to be Kelvin-Voigt type, which follows the equation
  G   (2-4)
15
where  is the shear stress,  is the shear strain, G is the tangential shear modulus, and  is the
coefficient of viscosity (Schnabel et al., 1972), and the dot denotes the derivative. The input ground
motion can be defined at any layer interface as a rock-outcrop or within motion. The response of the soil
profile to the input ground motion can be obtained in the form of acceleration response histories and
frequency-domain transfer functions at the layer interfaces, as well as stress and strain response histories
within the layers. The iterated values of shear modulus and damping ratio can also be obtained for each
layer, and are indicative of the degree of nonlinearity induced by the earthquake.
2.2.2.3 Nonlinear methods
Nonlinear site-response analysis is performed by modeling the soil profile in two approaches: 1)
the lumped-mass approach, and 2) the finite element approach. In the lumped-mass approach the soil
layers are lumped into adjacent nodal masses, which are connected by springs that model the soil stress-
strain behavior in shear. The input ground motion is applied at the base of the soil column, and the
dynamic equations of motion are integrated to calculate the response of the soil layers. In the finite
element approach, the soil layers are modeled using solid elements stacked on top of each other, and
constrained to only move in shear (Figure 2-5).
Figure 2-5: The lumped-mass and finite element models for site-response analysis [figure adopted
from Hashash et al. (2010) and modified]
16
In these models the bedrock is modeled either as 1) a rigid material (also called as the fixed-base
condition), preventing the radiation of the waves reflected from the surface, or 2) linear viscous material
(also called as the flexible-base condition), which allows the radiation of reflected waves and is modeled
as a transmitting boundary. For the rigid bedrock assumption, the ground motion (acceleration, velocity,
or displacement time-series) is directly prescribed at the base of the soil profile. For the linear viscous
bedrock assumption, the transmitting boundary is modeled using viscous dashpots at the base of the soil
profile, and the input ground motion is prescribed at the soil-bedrock interface in the form of shear
stresses (Figure 2-6). The properties of these dashpots and the input shear stresses are calculated as
follows:
C  Vs (2-5)
 (t )  Vs  y(t ) (2-6)
where C is the damping coefficient of the viscous dashpots,  and VS are the unit mass and elastic shear
wave velocity of the bedrock material,  (t ) is the input shear stress as a function of time, and y (t ) is the
velocity history of the outcrop ground motion (Lysmer, 1978). Joyner and Chen (1975) first used this
approach, which was adopted from a transmitting boundary model proposed by Lysmer and Kuhlemeyer
(1969) for three-dimensional infinite media.
Figure 2-6: Ground motion input in time-domain one-dimensional site-response analysis
2.2.2.3.1 DEEPSOIL
DEEPSOIL is a one-dimensional site-response analysis program developed primarily for time-
domain nonlinear analysis of deep soil sites (Hashash and Park, 2001). Frequency-domain equivalent-
linear analysis is also incorporated into this program, and is performed using the same method as
followed in SHAKE. DEEPSOIL uses the lumped-mass approach for the nonlinear time-domain analysis
17
(Figure 2-5). It employs the modified hyperbolic or MKZ model (Matasovic, 1993) for the soil, which
simulates pressure-dependent soil properties and uses a modified version of the extended Masing rules to
model the hysteretic behavior. The backbone curve for this model is defined by the following equation:
G0
 (2-7)
1   (  r ) s
where  is the shear stress,  is the shear strain, G0 is the maximum shear modulus,  and s are
dimensionless parameters, and  is a reference shear strain. Pressure dependence is introduced in the
model through the following equation, which defines the variation of the reference shear strain with
pressure:
b
  
r  a  (2-8)

 ref 
where   is the effective confining pressure, a and b are the curve fitting parameters, and  ref is the
reference confining pressure. DEEPSOIL is also capable of performing an effective stress analysis and
uses the pore water pressure model developed by Matasovic (1993) for this purpose. This model captures
the pore water pressure build-up as well as the resulting modulus degradation effect.
The backbone curve of the hyperbolic model and its variants (such as the modified hyperbolic
model) are almost linear at strains smaller than 10-4% and result in negligible hysteretic damping, unlike
real soils that exhibit a finite amount of damping (Hashash and Park, 2001). This is compensated in
DEEPSOIL by incorporating a frequency-independent viscous damping formulation. It is also referred to
as low-strain damping and depends on the confining pressure as follows:
d
 1 
   ref   (2-9)
  
 v 
where  ref is the low-strain damping at a reference effective vertical stress  ref ,  v is the effective
vertical stress, and d is a curve fitting parameter (Phillips and Hashash, 2009). Phillips and Hashash
(2009) also note that using the extended Masing rules to calculate the hysteresis loops leads to an
overestimation of the hysteretic damping in high strains. DEEPSOIL provides two approaches to
circumvent this problem. The first approach involves using the experimental data to determine a best fit
either only for the modulus reduction curve (which results in a poor match in hysteretic damping), only
for the damping curve (which results in a poor match for the modulus reduction), or a fit that matches
both the curves simultaneously (which results in an optimum between both). The second approach
(termed as the MRDF-UIUC approach) introduces a damping reduction factor, F ( m ) , which is
18
multiplied to the damping curve that is calculated using extended Masing rules. The damping reduction
factor is given by the equation
p3
 G 
F ( m )  p1  p2 1  m  (2-10)
 G0 
where p1 , p2 , and p3 are dimensionless parameters to best fit the experimental damping curve,  m is the
strain from the damping curve, G m is the value of reduced shear modulus at the corresponding strain, and
G0 is the maximum shear modulus. In order to accommodate the damping reduction factor while
maintaining the compatibility between the modulus reduction and hysteretic damping curves, the
extended Masing rules are modified by changing the equations for unloading-reloading behavior. Further
details regarding the damping reduction factor and the modified unloading-reloading behavior are
presented in Phillips and Hashash (2009). DEEPSOIL also enables the inclusion of pore water pressure
effects by performing an effective stress analysis using the models developed by Matasovic (1993).
Numerical integration of the dynamic equations of motion is carried out using the implicit Newmark- 
method (Chopra, 1995).
2.2.2.3.2 LS-DYNA
LS-DYNA is a commercial finite element program that primarily uses the explicit method for
numerical integration (LSTC, 2013). It provides a large collection of material models including several
soil models that can be used in seismic analyses (e.g. Ramberg-Osgood model, Drucker-Prager model,
Mohr-Coulomb model). Analyses in this research are performed using the MAT_HYSTERETIC model
(*MAT_079 in LS-DYNA), which was developed by Arup specifically for site-response and soil-
structure interaction analyses. The MAT_HYSTERETIC model is capable of three-dimensional
constitutive behavior and can be used only with eight-node solid elements. Hysteretic behavior is
generated in this model by superposing multiple layers having elastic-perfectly plastic properties in
parallel. The value of stress at any instant is equal to the sum of the stresses in the individual elastic-
perfectly plastic layers, which results in a piecewise-linear backbone curve. Figure 2-7 below illustrates
the hysteretic behavior of the MAT_HYSTERETIC model for a given loading.
19
Figure 2-7: The hysteresis loop for the MAT_HYSTERETIC model for a given input strain, and
the corresponding backbone curve
The backbone curve is directly input by the user as a set of ten shear stress and shear strain
values. This curve is subsequently broken down into ten layers of elastic-perfectly plastic properties, each
having a value of shear modulus and yield stress. These properties are pressure dependent as described by
the following equations:
G0 ( p  p0 )b
G ( p)  (2-11)
( pref  p0 )b
 ( p) a0  a1 ( p  p0 )  a2 ( p  p0 )2
 (2-12)
 ( pref ) a0  a1 ( pref  p0 )  a2 ( pref  p0 )2
where p is the current pressure, p0 is the cut-off or datum pressure, pref is the reference pressure at
which input values are calculated, G0 is the shear modulus of the layer at the reference pressure, G( p) is
the shear modulus of the layer at pressure p ,  ( pref ) is the yield stress of the layer at the reference
pressure,  ( p) is the yield stress of the layer at pressure p , b is the exponent for pressure sensitive
moduli, and a0 , a1 , and a2 are the yield function constants. Additional effects such as dilatancy and
strain rate effects are also included in the material model, which are illustrated in the LS-DYNA keyword
manual (LSTC, 2013).
2.2.2.4 Validation of one-dimensional site-response analysis programs
One-dimensional site-response analysis programs are typically validated by comparing the

numerical response of a real soil profile to recorded field data. The soil profile is modeled using the
20
material properties calculated from site investigations or laboratory testing. The input motion for the
numerical analysis is either 1) taken from a ground motion recorded at a nearby rock outcrop and
specified as an outcrop motion, or 2) taken from a ground motion recorded at the bottom of the soil
profile in a down-hole array and specified as a within motion. Idriss and Seed (1968a) and Seed and Idriss
(1969) first compared the numerical site responses with field data recorded during the 1957 San Francisco
earthquake, the 1962 Mexico earthquake, and the 1964 Alaskan earthquake, and noted that the site
response was captured fairly well. Although the ground motions used were of low intensity, with peak
input accelerations not more than 0.07g, the observations made in these studies further motivated the use
of one-dimensional site-response analyses. The strong ground motion data recorded during the 1989
Loma Prieta earthquake led to more such studies, such as Idriss (1990), Dickenson (1994), and Chang
(1996), which involved the validation of different numerical site-response analysis programs. Idriss
(1990) used this data to assess the accuracy of the numerical program SHAKE and found that it offers a
reasonable accuracy in estimating the peak acceleration responses for low-intensity motions, but only a
fair accuracy in estimating the spectral shapes. They suggested the use of nonlinear methods for more
intense ground motions with input rock accelerations greater than 0.2g. Dickenson (1994) conducted a
similar study for the programs SHAKE04 (Youngs, 2004), which is a slightly modified version of
SHAKE, and also the nonlinear program MAR-DESRA, which is a modified version of the program
DESRA-2 (Lee and Finn, 1978). They considered a total of ten deep, soft clay sites in the Bay Area, and
found good agreements only for a few sites (Figure 2-8). They also found that the accuracy of the
numerical response was sensitive to the choice of the outcrop ground motion, in cases where multiple
rock outcrop recordings were available for a single site. Chang (1996) performed similar validation
studies for the programs SHAKE and D-MOD using the recorded data from deep, stiff clay sites during
the 1989 Loma Prieta earthquake as well as the 1994 Northridge earthquake (Figure 2-9). While the
above studies conclude that the corresponding numerical programs make reasonably good predictions of
site response, Kwok (2007) indicated that some of the studies might have involved the modification of
model parameters to arrive at a better match.
21
Figure 2-8: Sample comparisons between the numerical and observed site response from Dickenson
(1994)
Figure 2-9: Sample comparisons between the numerical and observed site response from Chang
(1996)
Unlike in the above studies that used data from a nearby rock outcrop, data from vertical down-
hole accelerometer arrays enable a more reliable validation of one-dimensional site-response analysis
programs. The Lotung Large-Scale Seismic Test (LSST) in Taiwan, which included two down-hole
22
arrays and three surface arrays for a soil-structure interaction experiment, facilitated the recordings from a
total of 11 earthquakes with peak surface accelerations ranging from 0.03g to 0.21g for the horizontal
component, and from 0.01g to 0.20g for the vertical component (Chang et al., 1991). Chang et al. (1991)
found that the body waves propagated at an angle of 6 degrees to the vertical, indicating almost one-
dimensional propagation and that the data is suitable for the validation of one-dimensional methods.
Chang et al. (1990) used this data to validate the program SHAKE, for both upward wave-propagation
analyses (convolution) as well as deconvolution analyses. The results showed that SHAKE overestimated
the high-frequency response in the deconvolution analysis as well as the resonant frequency response in
the upward wave-propagation analysis. They also used a modified version of the nonlinear program
DESRA-2 (Lee and Finn, 1978) and found that the nonlinear site-response analysis resulted in better
agreement with recorded data. Borja et al. (1999) performed a similar study using the Lotung data to
assess the programs SHAKE and SPECTRA. SPECTRA is a fully nonlinear program that employs a
constitutive model based on three-dimensional bounding-surface plasticity theory (Borja and Amies,
1994). Both the programs showed reasonable matches with the experimental data, but the nonlinear
numerical response was better (Figure 2-10).
Figure 2-10: Sample comparisons of numerical and experimental data from Borja et al. (1999)
Kwok et al. (2008) investigated the results of a blind-prediction study that required the site-
response prediction of the Turkey-Flat site before the surface acceleration data was available. Turkey-Flat
23
is a shallow stiff soil site with recording instruments available in a vertical down-hole array and also at a
nearby rock outcrop (Figure 2-11).
Figure 2-11: Section view and instrumentation of the Turkey Flat site (Kwok et al., 2008)
Their study considered six different site-response analysis programs [SHAKE04 (Youngs, 2004),
which is a modified version of SHAKE, D-MOD_2 (Matasovic, 2006), DEEPSOIL (Hashash et al.,
2011), OPENSEES (Mazzoni et al., 2009; McKenna and Fenves, 2001), TESS (Pyke, 2000), and
SUMDES (Li et al., 1992)] for the blind prediction and used a within profile ground motion for input,
which was recorded from the 2004 Parkfield earthquake and had a peak acceleration of about 0.1g. They
observed that most of the programs captured the shape of the acceleration time series reasonably well, but
significantly underestimated the measured amplitude at certain time instants. A comparison of the
acceleration response spectra calculated from different programs indicated that the responses were similar
at the elongated period of the soil profile, but they were all significantly lower than the measured response
at this period. Additionally they found a large disparity between the responses from different programs at
short periods (Figure 2-12).
After further investigations they attributed the poor prediction to a bias in the measurement of
near-surface soil properties, and identified the one-dimensional approximation of a three-dimensional
phenomenon as another possible reason. Validation of a one-dimensional site-response analysis program
ideally requires data recorded from a true one-dimensional wave field, which is not produced by real
earthquakes. Nevertheless the above studies used data from real earthquakes and, with the exception of
the studies using LSST data, it is unclear if they verified the validity of the one-dimensional method for
the sites considered.
Validation can also be performed through laboratory experiments using a laminar box for the soil.
Elgamal et al. (2005) performed a series of centrifuge tests, and used the data to calibrate the parameters
of a nonlinear finite element site-response analysis program. Simulations using the calibrated parameters
showed good agreements with the experimental data. Similar experiments are possible at a larger scale
while using a shaking table, but such tests are yet to be performed.
24
Figure 2-12: Blind predictions at locations V1 and D2 using different numerical programs and the
corresponding residuals1 (Kwok et al., 2008)
1
Residual is defined as R(T )  ln  S (T )   ln  S (T ) 
 a  data  a  predicted
25
2.2.2.5 Comparison and limitations of various one-dimensional methods
The equivalent-linear method was developed as a computationally simpler alternative to the

nonlinear method, although the latter involves a more realistic simulation of soil behavior. The
equivalent-linear method is constrained to only elastic deformations and therefore cannot simulate soil
failure and related effects such as permanent displacements and liquefaction. Furthermore the ambiguity
in the selection of the effective strain ratio makes this method even less credible compared to the
nonlinear method (Yoshida et al., 2002). Idriss and Seed (1968b) performed a nonlinear site-response
analysis assuming bilinear properties for the soil, and used the computed hysteretic stress-strain loops to
calculate the equivalent-linear properties (reduced shear modulus and equivalent damping) for each layer.
They performed a linear analysis using these properties and found that the responses of the soil layers
match reasonably well with the nonlinear responses. Although this was not a direct equivalent-linear
analysis (involving iterations to calculate the equivalent-linear properties), they demonstrated that the
results of a nonlinear analysis can be captured reasonably well by a linear analysis. Constantopoulos et
al. (1974) made detailed comparisons between the results of a nonlinear (using the Ramberg-Osgood
model) and an equivalent-linear analysis using a few hypothetical soil profile models. Their study showed
that 1) the equivalent-linear method underestimates the peak displacements and overestimates the peak
accelerations, 2) the resonant period of the soil profile calculated from the two methods match reasonably
well, and 3) the differences between the solutions from the two methods increase with the intensity of the
excitation. Joyner and Chen (1975) performed another comparison involving a deep soil profile subjected
to an intense excitation, and found that they match reasonably well for longer periods but the equivalent-
linear method does not reproduce the short-period components present in the nonlinear solution. Also the
peak accelerations were underestimated, contrary to Constantopoulos et al. (1974). Finn and Martin
(1978), in a study comparing prominent site-response analysis programs, found that SHAKE
overestimated the peak shear stresses in the soil layers, and also the acceleration amplitude at the resonant
time period of the soil profile. While Finn and Martin (1978) attributed this to resonance in a linear
analysis, Yoshida et al., (2002) contended that it is due to the mechanism of the equivalent-linear method
itself, specifically the definition of the effective shear strain ratio as a fraction of the effective shear strain.
By reexamining the work by previous researchers, Yoshida et al., (2002) summarized that the equivalent-
linear method results in 1) the overestimation of the shear stress as compared to the nonlinear method,
which leads to an overestimation of the peak accelerations under strong shaking, and 2) underestimation
of high-frequency amplifications, which sometimes leads to the underestimation of the peak accelerations.
They attributed these drawbacks to the concept of a constant value of effective shear strain ratio, and also
proposed a frequency-dependent formulation for this parameter that resulted in a better agreement
between the equivalent-linear and nonlinear analyses.
26
Despite the drawbacks the equivalent-linear method is still preferred over the nonlinear method in
practice, often due to poor understanding of nonlinear soil parameters that may not be clearly associated
with measurable soil properties (Kwok, 2007). Nonlinear methods also occasionally result in abnormal
high-frequency components in the acceleration response. This is either due to the soil material model used
in the analysis [e.g., the multi-linear material model used by Chen and Joyner (1974)], or the type of
integration technique used for solving the dynamic equations (Joyner and Chen, 1975). Chen and Joyner
(1974) investigated this issue and confirmed that these frequencies occur due to the resonance of the
individual layers of the numerical model, and suggested that the noise be removed using digital filtering.
The one-dimensional method was developed specifically for horizontally layered soil deposits
subjected to vertically propagating waves, but no definite guidelines are yet proposed that assist in
deciding the validity of this method for given site conditions. Additionally most of the numerical
programs require a uni-directional horizontal input, which entails an assumption that the horizontal
components of ground motion are uncorrelated. As this is not true, the correlation between the horizontal
components needs to be accounted for by applying them simultaneously in the input. This requires
multidimensional plasticity based material models [such as the one described in Borja et al. (1999)] that
model the correlation between the stress components. However the author is not aware of studies that
have investigated the effects of ignoring the correlation between horizontal components.
2.2.3 Multi-dimensional analyses
The assumption of vertical wave propagation does not hold in many cases that involve significant
contribution of two- or three-dimensional wave fields and require multi-dimensional site-response
analysis simulations. Some of these cases have been thoroughly investigated, which include 1) sites with
uneven surface topographies such as earth banks, cliffs and ridges, and 2) heterogeneous basins, where
the soil-bedrock interface is not horizontal. Early studies addressing these effects used discrete numerical
methods (such as the finite element, or finite difference methods) and performed two-dimensional linear
elastic analyses [e.g. Idriss and Seed (1967b) and Boore (1972)]. Although they did not account for soil
nonlinearity, these studies were notable in pointing out that the site amplifications in the presence of
topographical features were considerably different from those calculated from one-dimensional analyses.
Nonlinear two-dimensional analyses were also performed at that time: Joyner (1975), for example,
investigated the multi-dimensional effects of the presence of soft basins and earth dams using a
rheological model from the family of models developed by Iwan (1967). In more recent work, Assimaki
and Gazetas (2004) performed parametric studies involving two-dimensional site-response analyses to
investigate the contribution of site properties and the input ground motion characteristics on topographic
effects.
27
Three-dimensional linear and nonlinear site-response simulations have also been performed,
primarily in the field of computational seismology. The main application of these simulations is the
characterization of seismic hazard while accounting for large scale three-dimensional effects such as of
basins and directivity (Taborda, 2010). Xu et al. (2003) noted that the three dimensional basin effects 1)
significantly increase the amplitude of the surface acceleration response, and 2) extend the duration of the
strong phase of the ground motion, as compared to a one-dimensional analysis. Despite the significant
computational challenges involved, several such studies have been attempted, including Xu (1998), Xu et
al. (2003), Day et al. (2006), and Restrepo et al. (2011). Each of these studies has indicated the
importance of three-dimensional effects, especially basin effects, which cannot be reproduced by one-
dimensional simulations.
28
SECTION 3
SOIL-STRUCTURE INTERACTION: A REVIEW
3.1 Introduction
Earthquake waves originate from a fault rupture and propagate in the earth’s crust. In the absence
of structures at a site of interest, the resulting ground motion is called the free-field ground motion. When
a structure, or a set of structures, is present at a site of interest, the ground motion results in structural
shaking. Traditionally the response of structures to ground motion is calculated by disregarding the
flexibility of the soil below, assuming the structure is fixed to the ground, and subjecting the structure to a
free-field ground motion. This is called fixed-base analysis. In reality, soil is flexible and the dynamic
response of the soil affects the dynamic response of the structure, and vice versa. This interaction between
soil and structure is called soil-foundation-structure interaction (SFSI) or soil-structure interaction (SSI).
There are two aspects of SSI: 1) the modification of the free-field ground motion due to the presence of a
stiff foundation in the soil, and 2) modification of the structural response due to the flexibility of the soil
(flexible-base). The former is called kinematic interaction, and results in a deviation of the ground motion
input to the structure from the free-field ground motion. The latter is called inertial interaction and results
in a deviation of the response of the structure from the fixed-base response. Inertial interaction results in
1) an increase in the natural time period of the structure due to soil flexibility (period elongation), and 2)
an increase in the structural damping due to energy dissipation from the hysteretic behavior of the soil at
the foundation (hysteretic damping) and the radiation of the reflected wave-field away from the structure
(radiation damping). An accurate estimation of structural response when accounting for SSI (SSI analysis)
requires the simulation of kinematic interaction and inertial interaction, including period elongation,
hysteretic damping and radiation damping.
SSI is often neglected in the design of non-nuclear structures. The most common argument for
neglecting SSI is that it results in a negligible change in the response of the structure. Indeed, SSI effects
are small for many conventional structures such as light, flexible buildings founded on stiff soil. For these
cases, the response of the structure is very close to the fixed-base response. Another reason for ignoring
SSI is a common misconception that SSI effects are always beneficial and result in a net decrease in
structural demand. The provisions for incorporating SSI in structural design in ASCE Standard 7 (ASCE,
2010) include a reduction in the design base shear. This reduction in seismic demand is illustrated in
29
Figure 3-1, which shows the change in the design spectral acceleration S a , when the structural time
period2 increases from T to T , and the damping ratio increases from  to  due to SSI effects.
Figure 3-1: Decrease in seismic demand due to SSI effects in idealized design response spectra
[adopted from Mylonakis and Gazetas (2000) and modified]
However this decrease is due to the idealized shape of the design response spectrum, in which the
spectral acceleration decreases with the structural period in the intermediate- and long-period ranges. In
reality, spectral accelerations calculated from recorded ground motions may also increase with period in
the long-period range due to 1) amplification of the rock motion at the site period of deep soil sites, or 2)
forward fault-rupture directivity (Somerville et al., 1997), which results in a long period pulse in the
ground motion that manifests as large spectral accelerations at periods greater than 0.5 sec. Response
spectra for some such ground motions are presented in Figure 3-2, which also enables a comparison
between the response spectra with a sample design response spectrum. In addition to the increase in
spectral accelerations, Mylonakis and Gazetas (2000) also demonstrated that the increase in structural
period can result in an increase in the ductility demands for structures founded on soft soils. However,
irrespective of its beneficial or detrimental effects, there is growing consensus that SSI should still be
considered for performance assessment of buildings (FEMA, 2012; Mylonakis and Gazetas, 2000;
2
It should be noted that time periods of most buildings are in the range for which, the design spectral
acceleration is either constant or decreasing with structural period.
30
Stewart et al., 1999b). Guidelines for including SSI in the design and performance assessment of
buildings are provided in ASCE Standard 7 (ASCE, 2010), FEMA 440 (FEMA, 2005) and NIST GCR
12-917-2 (NIST, 2012).
Figure 3-2: A typical code design spectrum and response spectra from some recorded strong
ground motions (Mylonakis and Gazetas, 2000)
Unlike buildings, SSI can significantly affect the response of nuclear power plant structures.
Nuclear power plants are stiff and heavy, which are characteristics that facilitate SSI. Indeed most of the
early research in numerical SSI analyses was motivated by its significance in the design of nuclear power
plants (Kausel, 2010). Design and performance assessment of nuclear power plants in the United States
account for SSI, and guidelines for performing SSI analyses for these structures are presented in ASCE
Standard 4 (ASCE, 1998; ASCE, forthcoming).
3.2 Numerical methods in soil-structure interaction analysis
Early SSI analyses involved analytical calculations of the response of footings resting on a semi-
infinite half-space. Most of these studies [e.g., Veletsos and Wei (1971), Luco and Westman (1971),
Veletsos and Verbic (1973), Veletsos and Verbic (1974)] calculated frequency-dependent impedance
functions for rigid foundations that were limited to rectangular or circular shapes, and lay atop of elastic
31
or viscoelastic halfspaces. Amongst the studies that involved seismic analysis of buildings, a study by
Jennings and Bielak (1973) showed that the SSI response of a building with n resonant frequencies can
be found by the superposition of n  2 damped, linear oscillators subjected to modified excitations, with
the two additional oscillators corresponding to translation and rotation of the foundation. Veletsos and
Meek (1974), in an influential paper, used this result to obtain the properties of an equivalent single-
degree-of-freedom (SDOF) oscillator that had the same time period and damping ratio as an SDOF
system founded on linear elastic soil. By comprehensive numerical studies on the equivalent oscillator,
their study was also extended into identifying various system parameters that could be used to assess the
significance of SSI effects. These studies also led to the development of early guidelines in FEMA 274
(FEMA, 1997) to account for SSI effects in structural design. The advent of numerical tools such as the
finite element and boundary element methods in the 1970’s enabled researchers to take up more
complicated foundation configurations (such as embedded foundations and foundations on layered media)
to calculate impedance functions [e.g., Elsabee et al. (1977), Kausel and Roesset (1975)]. Gazetas (1991)
summarized frequency-dependent impedance functions for different foundation shapes and configurations
(including embedded foundations and arbitrarily shaped foundations), which were calculated over the
years using various analytical and numerical techniques.
Early SSI analysis involved modeling the flexibility of the soil at the foundation through
equivalent springs and dashpots, with pre-calculated impedance functions. This method of SSI analysis is
also called the spring method. The spring method had several limitations: the foundation stiffness and
damping functions calculated in the studies described above were frequency-dependent, and were only
applicable to a narrow range of frequencies (Roesset, 1998). Additionally, problems in practice involved
more complicated foundations with non-rectangular or circular shapes, deeply embedded foundations, and
soil profiles with non-uniform variation of shear modulus with depth. These problems required more
sophisticated methods, which were later developed mainly for the nuclear industry (Kausel, 2010;
Roesset, 1998). As such, the pursuit of more sophisticated SSI analyses methods resulted in two main
approaches: the substructuring method, and the direct method. These methods are also used in the SSI
analyses presented in this report, and therefore are described in the following sections.
3.2.1 Substructuring method
In the substructuring method of SSI analysis, the soil-structure system is divided into different
components (or substructures), and the response of each substructure is calculated independently (Figure
3-3). The substructures are ‘connected’ by applying equal and opposite interaction forces to each of the
substructure models. The responses of all the substructures are later combined by superposition to
32
calculate the final response of the system. The substructure technique therefore provides an advantage of
the choice of a suitable analysis method for calculating the response of each substructure. Since the
individual responses of the substructures are added by superposition, the substructure method is limited to
linear analyses. Two of the widely-used software programs that use the substructuring technique are
SASSI (A System of Analysis of Soil-Structure Interaction) (Lysmer et al., 1999) and CLASSI
(Continuum Linear Analysis of Soil-Structure Interaction), both of which, function in the frequency
domain.
Figure 3-3: Steps in a substructuring analysis (Kramer and Stewart, 2004)
Most often, the soil-structure system is subdivided into a soil medium and a structure, as
illustrated in Figure 3-3, and the responses of these substructures are calculated through site-response
analysis and structural analysis, respectively. Structural analysis can be conveniently performed by finite
element methods, once the seismic input and impedance at the foundation are calculated. The calculation
of the seismic input should account for kinematic interaction, and is called a scattering analysis. The
calculation of foundation impedance (which is the complex stiffness that includes both stiffness and
damping functions) should include radiation damping and is called the interaction analysis or the
impedance analysis. Substructuring methods can be classified into four types based on the way the
scattering problem and interaction problem are solved (Ostadan, 2006a), 1) the rigid boundary method, 2)
the flexible boundary method, 3) the flexible volume method, and 4) the subtraction method. All four
33
methods involve site-response analysis to calculate the free field ground motion, impedance analysis to
calculate the foundation impedance and the structural response analysis to calculate the structural
response. The first two methods also involve scattering analysis, which involves the calculation of
foundation input motion (motion at the foundation in the presence of a rigid, massless structure). Figure
3-4 presents a summary of these methods.
Figure 3-4: Substructure methods of SSI analysis (Ostadan, 2006a)
3.2.2 Direct method
The substructuring method is advantageous in terms of computational requirements and the

flexibility of using different analysis methods for different substructures, but it is limited to linear
analysis. A more realistic simulation of the nonlinear response of the soil-structure system is possible
through the direct approach, in which, the entire soil-structure system is analyzed in a single step,
circumventing the use of superposition. Soil-structure interaction using the direct method can be
performed in commercial finite element programs such as ABAQUS (Dassault Systèmes, 2005), ANSYS
(ANSYS Inc., 2013) and LS-DYNA (LSTC, 2013), or the open source finite element program, OpenSees
(Mazzoni et al., 2009). Figure 3-5 presents an illustration of a finite element model for SSI analysis using
the direct method. Soil-structure interaction analysis using the direct method requires significant
computational resources in comparison with the substructuring method, and has therefore been rarely
performed in practice. Consequently, nonlinear SSI analysis using the direct method is not yet well
established, and some challenges still exist including, 1) specifying the ground motion input, 2)
34
simulating an infinite domain, and 3) developing reliable three-dimensional nonlinear constitutive models
for soil.
The direct method is currently performed only for cases involving vertically propagating shear or
compressive waves. Ground motion input for these cases is applied either as a force input (for viscoelastic
bedrock) or an acceleration input (for rigid bedrock) to the nodes at the bottom of the soil profile, similar
to the specification of input in a nonlinear site-response analysis (see Section 2.2.2.3). Inclined waves and
other wave-fields cannot yet be specified as input in the direct method.
Figure 3-5: Finite element model for soil-structure interaction analysis using the direct method
Simulating an infinite domain is another major challenge in performing nonlinear SSI analysis
using the direct method. An infinite soil domain can be simulated by building a finite domain that
satisfies the following conditions: 1) effective damping of the waves radiating away from the structure so
that they do not reflect back into the soil domain from the lateral boundaries, and 2) stress equilibrium at
the lateral boundaries to account for the rest of the soil domain that is not included in the finite domain
model. The former can be achieved by using absorbing boundary models at the lateral boundaries that
absorb the incoming waves and minimize reflections. Absorbing boundary models have been developed
and implemented in commercial finite element programs. Examples include the viscous boundary model
by Lysmer and Kuhlemeyer (1969) and, a more recent, Perfectly Matched Layer (PML) model introduced
to the SSI analysis domain by Basu (2009), both of which, have been implemented in LS-DYNA.
However these boundary models are limited to waves propagating in linear elastic materials. No
absorbing boundaries have yet been developed for nonlinear materials, to the knowledge of the author.
35
Another approach to simulate radiation damping is to build a very large soil domain with
sufficient plan dimensions to dissipate the radiating waves before they reach the lateral boundaries. In this
approach, the radiating waves are dissipated completely by soil hysteresis and viscous damping in the soil
domain. Therefore a reasonable amount of viscous damping should be provided. The plan dimensions can
be determined by a trial-and-error procedure, and ensuring that the acceleration responses at the
boundaries of the soil domain are equal to the free-field acceleration, which is calculated from a separate
site-response analysis. The stress equilibrium at the lateral boundaries can be obtained by constraining the
elements at the boundaries to move in pure shear, thus simulating the free-field condition (assuming that
the input comprises vertically propagating shear waves).
3.2.3 Domain reduction method
The domain reduction method (DRM), proposed by Bielak et al. (2003), is a two-step approach to
input a complex three-dimensional wave field for the seismic analysis of a soil-structure system. This
soil-structure system could be a geological feature, a structure with neighboring soil, or even a large
landmass with numerous structures, such as an entire city. Although the DRM was primarily developed
for application in large-scale models involving earthquake simulations from fault rupture to surface
shaking at the site of interest (Taborda, 2010; Taborda and Bielak, 2011; Xu et al., 2003; Yoshimura et
al., 2003), it is also useful for SSI analyses performed at a smaller scale. Large-scale earthquake
simulations involve modeling a large soil domain including the seismic source (e.g., a ruptured region in a
fault), and the local feature of interest, at which the response needs to be calculated. The DRM provides a
way to model only the local feature with equivalent loads, which represent the loading from the seismic
source, applied at its boundaries. This method effectively circumvents the modeling of a large soil
domain, thus significantly reducing the computational requirements of the problem.
Consider a large-scale domain, with a local feature of interest, illustrated in Figure 3-6a. In this
figure,  denotes the internal region of the domain,  denotes the external region,  denotes the
boundary between the internal and external region, u e denotes the displacements of the nodes in the
external region, u b denotes the displacement of the nodes at the boundary, u i denotes the displacements
of the nodes in the internal region, and Pe  t  denotes the external force history from a seismic source.
Figure 3-6b illustrates an auxiliary model of the large-scale domain, in which, 0 denotes the free-field
0 0 0
internal region that is free of the local feature, u e , u b and u i denote the free-field displacements of the
nodes in the external region, the boundary, and the internal region, respectively.
The DRM involves two steps: 1) analysis of the auxiliary model subjected to external loads, and
2) analysis of the internal region  with the local feature of interest, with effective forces representing
36
the seismic source. The effective forces are calculated from the free-field results calculated in the first
step. These loads are given by the equation
 Pi eff   0 
 eff    0 
 P    Pb    M be u e  K be u e
eff  0
(3-1)
 
 Peeff   M eb u 0b  K eb u 0b 
 
where  P  is the effective force and M and K are the mass and stiffness matrices, respectively, of the
eff
regions described by the subscripts e, b and i, referring to the external domain, boundary and internal
0 0
domain, respectively. The superscript 0 for the displacements and accelerations (e.g., u e and u e )
indicates free-field; these displacements and accelerations are provided by the auxiliary analysis. The
derivation is presented in Bielak et al. (2003) and is not described here. The subscripts of the mass and
stiffness matrices in equation (3-1) indicate that the effective forces have non-zero values only at the
intersection of e and b nodes, namely, at the boundary  . Therefore, analysis of the internal domain can
be performed by applying the effective forces only at the boundary, and the external soil domain need not
be modeled completely. This analysis also results in scattered waves reflected from the structure and
travelling back into the halfspace. An absorbing boundary is therefore required outside  , and can be
modeled using one of the absorbing boundary models presented in the previous section.
It should be noted that although the DRM formulation requires that the external domain be
linearly elastic, the internal domain can be fully nonlinear. Therefore, the auxiliary analysis that provides
the free-field result, namely the site-response analysis, can only be performed in the linear domain.
Nonlinear site response can be accounted for by using strain-compatible properties for the external soil
domain (Bielak, 2013).
The DRM provides a computationally economical alternative to the direct method for nonlinear
SSI analysis. However, unlike in the direct method, the primary nonlinearities in the DRM can only be
accounted for using the equivalent linear approach. The DRM has been used in some studies for large-
scale simulations (Taborda, 2010; Taborda and Bielak, 2011; Yoshimura et al., 2003) but has rarely been
used for SSI analyses. It has been implemented in the US NRC ESSI simulator (Jeremic et al., 2011),
which is currently being developed for high-performance, high-fidelity SSI analyses of nuclear structures.
The DRM has not yet been implemented in commercial finite element programs. However, the effective
seismic input method, which is a predecessor of the DRM and developed by Bielak and Christiano
(1984), has been implemented in LS-DYNA, specifically for SSI analyses. The effective seismic input
method and DRM are essentially equivalent in theory, but the DRM is easier to implement in a finite
element formulation (Bielak, 2013).
37
a. A large-scale domain containing a local feature of interest
b. Auxiliary model of the large-scale domain

Figure 3-6: Illustration of the domain reduction method (Jeremic et al., 2013)
3.3 Soil-structure interaction analysis using industry-standard numerical programs
3.3.1 SASSI
SASSI is currently the most widely used numerical program in the nuclear industry of the United
States for seismic SSI analysis. Although developed by a single team, headed by Professor John Lysmer
at the University of California at Berkeley, several modified versions of the program with varying
capabilities now exist. The version SASSI2000 distributed by University of California, Berkeley and
Ostadan (2010) is used to perform the analyses presented in this report.
SASSI uses the substructuring method in the frequency domain and is capable of performing two-
and three-dimensional analyses for any foundation shape and any superstructure. Substructuring method
allows the SSI problem to be solved in parts, with each part solved using a different method of analysis.
The SASSI solution to the SSI problem is solved in three parts: 1) the site-response problem, which
involves the calculation of the free-field input to the structure and is solved using analytical methods, 2)
the impedance problem, which involves calculating the impedance functions at the foundation and solved
using the axisymmetric finite element method, and 3) the structural analysis problem, which involves the
calculation of the structural response from the impedance functions and the site-response input and solved
using the regular finite element method. Two types of substructuring techniques are available in SASSI:
38
1) the flexible-volume method, and 2) the subtraction method. Although the subtraction method requires a
smaller number of interaction nodes and is more computationally efficient, recent studies have shown that
it may not result in accurate solutions in some cases (Mertz et al., 2010). Therefore, the SSI analyses of
the studies presented in this report are performed using the flexible-volume method. The following
sections briefly describe the theory and implementation of the flexible-volume method, and the procedure
for modeling and analysis in SASSI. These descriptions are largely adopted from the SASSI user manual
(Ostadan, 2006b) and the theoretical manual (Ostadan, 2006a).
3.3.1.1 Theory and implementation
Substructuring for the flexible-volume method is performed as illustrated in Figure 3-7. The total
system is shown in panel a of the figure, and is divided into three substructures shown in panels b, c, and
d. These substructures are: 1) the free-field, labeled as substructure I and shown in panel b, 2) the
excavated soil volume, labeled as substructure II and shown in panel c, and 3) the structure-foundation
system, labeled as substructure III and shown in panel d. From Figure 3-7 it can be seen that the total
system is a combination of the substructure I and substructure III, with substructure II excluded. If the
total system, including the substructures, are discretized into nodes and elements as shown in the figure,
the equation of motion for the system can be written as follows.
 M  Uˆ   K Uˆ   Qˆ  (3-2)
where,
M = mass matrix of the total system
K = complex stiffness matrix of the total system
 Û  = displacement vector of the total system
 Q̂  = external load vector
39
(a) Total system
(b) (c) (d)

Substructure I Substructure II Substructure III
Free field Excavated soil Structure
(a) = (b) – (c) + (d)
Figure 3-7: Substructuring in the flexible-volume method in SASSI (Ostadan, 2006a)
 
If the external load is a harmonic excitation Q̂   Q  eit , where  is the excitation frequency,
 
the displacement response of the system would be Û   U  eit . Therefore, by substituting the
expressions for the displacement response and external loads, the equation of motion can be written as
 C  U    Q  (3-3)
40
where matrix  C  is the complex, frequency-dependent, dynamic stiffness matrix given by
C  K   2 M (3-4)

The nodes in each substructure shown in the figure are referred to as follows:
b - boundary of the total system
i - boundary between the soil and the structure
w - within the excavated volume
g - remaining part of the free-field
s - remaining part of the structure
f - combination of ‘i’ and ‘w’ nodes (the interaction nodes)
Additionally each substructure has its own dynamic stiffness matrix  C and displacement vector U .
For example, the stiffness matrix C  , and the displacement vector U
III III
 
for substructure III can be
written as,
CIII CisIII   Ui 
CIII    iiIII  and
CssIII 
U    U
III
 (3-5)
Csi  s
The stiffness of the soil medium (substructure I) can be represented by a matrix of impedance functions at
all the interaction nodes (f nodes). The flexible-volume method requires that the impedance functions be
calculated at all the nodes in the excavated soil (unlike the subtraction method, which involves the
calculation of impedance functions only at the nodes on the boundary of the excavated soil). Therefore, if
the impedance functions are denoted by X , the impedance matrix for the foundation nodes can be written
as
X Xiw 
 Xff  = X ii X ww 
(3-6)
 wi
Following the substructuring in Figure 3-7, and using appropriate dynamic stiffness matrices and
displacement vectors for each substructure, the equation of motion can be written as (Ostadan, 2006a)
CiiIII -CiiII +Xii II

-Ciw +Xiw CisIII   Ui   Xii Ui' +Xiw U 'w 
 II    
 -Cwi +X wi -CIIww +X ww 0   U w    X wi Ui' +X ww U 'w  (3-7)

 CsiIII 0 CssIII   Us   0 

where U i' and U i' are the free-field displacements at i and w nodes imposed by the free-field ground
motion and calculated from the site-response analysis. It can be seen that the right-hand side of equation
(3-7) denotes the external forces on the total system (there are no external forces on the s nodes) and the
41
left-hand denotes the internal forces. After the dynamic stiffness matrices, impedance matrices, and the
free-field ground motions are obtained, equation (3-7) can be used to calculate the responses of the
structure-foundation system, Us and U i . These responses are calculated separately for each frequency in a
set of frequencies input by the user. The responses for other frequencies calculated by interpolation, using
an interpolation function that is similar to the frequency-domain response of a two-degree-of-freedom
system. A more detailed description of this interpolation technique is provided in the SASSI theoretical
manual (Ostadan, 2006a).
The following steps briefly describe the algorithm for SSI analysis in SASSI. These steps are
used for both the flexible-volume method, and the subtraction method:
Step 1 - Solving the site-response problem: SASSI performs site-response analysis to calculate the
response of the soil profile. Displacements imposed at the interaction nodes by the free-
field ground motion are calculated in this step.
Step 2 - Solving the impedance problem: The impedance matrix  X ff  , which includes the
impedance functions at the interaction nodes, is calculated in this step. In the flexible-
volume method, impedance is calculated at the i and w nodes, whereas in the subtraction
method, it is calculated only at the i nodes.
Step 3 - Forming the load vector: The load vector [right-hand side in equation (3-7)] is calculated
by multiplying the impedance matrix calculated in Step 2 with the free-field displacement
calculated in Step 1.
Step 4 - Forming the complex stiffness matrix: Using material properties and nodal masses
supplied by the user, the mass matrix  M  and the complex stiffness matrix  K  are
calculated. The complex dynamic stiffness matrix is then calculated for each substructure
using equation (3-4). The matrices are them assembled as shown in the left-hand side of
equation (3-7) to calculate the dynamic stiffness matrix of the soil-structure system.
Step 5 - Solving the equations of motion: After forming the load vector in Step 3 and the dynamic
stiffness matrix in Step 4, they are substituted in the equation of motion (3-7) to calculate
the response of the system.
3.3.1.2 Modeling and analysis procedure
Since the substructuring method can be used only with linear models, the soil and structure are
assumed to have linear viscoelastic properties. Nonlinear response can be approximately calculated using
equivalent linear, strain-compatible properties for the soil. These properties are typically calculated using
42
SHAKE, but can be calculated using other equivalent linear site-response analysis programs. The soil
layers in SASSI are assumed to be horizontally infinite, and parallel to each other. Unlike the one-
dimensional site-response analysis programs, SASSI is capable of simulating three-dimensional wave
propagation, including surface waves. The input ground motion is defined as a wave-field composed of
body and surface waves, and as either an outcrop or a within motion in any layer of the soil profile. The
bedrock underlying the soil profile can be modeled either as rigid or a viscoelastic halfspace. The lateral
transmitting boundaries are intrinsically modeled in the program, and do not have to be modeled by the
user. The structure-foundation system in SASSI is modeled using two-dimensional or three-dimensional
finite elements. The element library includes two-noded beam elements, eight-noded solid elements, four-
noded thin shell elements, and two-noded spring elements, and lumped masses can be provided at the
nodes as required. SASSI also requires modeling the excavated soil elements, which are a part of the
substructuring equation (Ostadan, 2006a), when an embedded foundation is involved.
The SASSI program consists of 13 modules, of which, only five are relevant to the SSI analyses
of this report. These modules are, SITE, POINT, HOUSE, ANALYS and MOTION. Each of the SASSI
modules is an independent executable file, and the modules are run consecutively. The role of each
module and some of its main input parameters are briefly described below in an order in which, the
modules are run for an SSI analysis.
 SITE: This module performs the site-response analysis, the results of which, are used to calculate the
free-field input at the interaction nodes. The input to the SITE module includes the soil profile model
(thickness and material properties of each layer), location of the input ground motion, and the
composition of the input wave field. The analyses of this report only involve vertically propagating
shear waves, and therefore a vertically propagating SV wave field is specified in this module. The
analysis frequencies, for which the SSI analysis is performed, are also input in the SITE module. The
SASSI user manual (Ostadan, 2006b) recommends about 10 to 20 frequencies, chosen based on the
fixed-base frequencies of the superstructure.
 POINT: This module performs a point load analysis for the soil halfspace, namely, the response of
halfspace to the application of a unit point load. This analysis is performed in each co-ordinate
direction for each of the frequencies input in the SITE module. The input to the POINT module
includes the maximum number of soil layers into which, the foundation is embedded. The point load
solution is calculated for all the layers into which the structure is embedded.
43
 HOUSE: This module calculates the mass matrix  M  , and stiffness matrix  K  (and therefore the
dynamic stiffness matrix  C ), of the structure and the excavated soil. The input to the HOUSE
module is the finite element model of the structure-foundation system and of the excavated soil.
 ANALYS: This is the main module in SASSI, which assembles the impedance matrix of the
interaction nodes, and the dynamic stiffness matrix of the total system, and calculates the final
acceleration transfer functions of the response at the structural nodes relative to the input location.
 MOTION: This module reads the input acceleration (from a separate file named ‘SASSI. T14’) and
uses the transfer functions output by the ANALYS module to calculate the output accelerations. The
input to this module includes the nodes at which, the output accelerations are requested.
3.3.2 LS-DYNA
The commercial finite element program, LS-DYNA, is equipped with a large number of material
models that can be used to model soil and structure, and several contact models that can be used to
simulate the foundation-soil interface. LS-DYNA is therefore a suitable choice for comprehensive
nonlinear SSI analyses, and is employed for the SSI analyses presented in this report. Soil-structure
interaction analyses in LS-DYNA can be performed by 1) the direct method or 2) the effective seismic
input method. The procedures for these methods are briefly described in the following sections.
3.3.2.1 Direct method
The direct method is described in Section 3.2.2. The procedures for performing an SSI analysis
using the direct method in LS-DYNA are presented in this section.
The absorbing boundaries implemented in LS-DYNA do not facilitate the application of
equilibrating stresses at the lateral boundaries. Therefore the direct method in LS-DYNA can be most
efficiently performed by modeling a soil domain large enough to dissipate the scattered waves before they
reach the lateral boundaries. The soil domain can be modeling using eight-noded brick elements.
Although several soil material models are available, the analyses presented in this report use the MAT_
HYSTERETIC material, which is also used for site-response analyses and described in Section 2.2.2.3.2.
The nodes on the lateral boundaries at each elevation can be constrained using the *CONSTRAINED_
NODE_SET option, which enables the nodes to move together in the horizontal and vertical directions.
For a sufficiently large soil domain, this constraint imitates the free-field condition at the lateral boundary
44
by providing the necessary stresses for simulating an infinite soil domain. The input ground motion can
be specified as described in Section 2.2.2.3.
Secondary nonlinearities such as gapping and sliding can be simulated in LS-DYNA. These
nonlinearities involve separation at the soil-foundation interface, which can be simulated using
*CONTACT_AUTOMATIC_SURFACE_TO_SURFACE. A soil-foundation interface that does not
allow separation can be modeled using *CONTACT_TIED_SURFACE_TO_SURFACE for a flexible
foundation, or *CONSTRAINED_EXTRA_NODES for a rigid foundation. Correct simulation of
separation at the soil-foundation requires gravity loading on the structure. For the analyses (using the
direct method) presented in this report, gravity loads are applied by performing a preliminary analysis
using dynamic relaxation. A description of the dynamic relaxation method is provided in the LS-DYNA
keyword user manual (LSTC, 2013).
LS-DYNA has the option of modeling viscous damping (for both soil and structural elements)
that is independent of the frequency, unlike Rayleigh damping. This frequency-independent damping was
developed by Arup, and can be provided using the *DAMPING_FREQUENCY_RANGE option. A small
value of damping (< 2%) should be used as recommended in the LS-DYNA keyword user manual (LSTC,
2013).
3.3.2.2 The effective seismic input method
Soil-structure interaction analysis can be performed in LS-DYNA using the effective seismic
input method through a series of steps described in the keyword user manual, as well as on the LSTC
website: http://www.lstc.com/applications/soil_structure. A brief description of these steps, adapted to the
SSI analyses presented in this report, follows:
1. For cases involving vertically propagating shear wave input, perform an auxiliary site-response
analysis with the boundary  (see Figure 3-6) between the internal domain and external domain,
modeled using the *INTERFACE_SSI_AUX_EMBEDDED option. Other approaches need to be
adopted for cases involving inclined wave input, and are not discussed here. In case the free-field
internal domain 0 is not included in the site-response analysis model (namely, if the site-
response analysis model contains an excavation), the *INTERFACE_SSI_AUX should be used
(Basu, 2014). The site-response analysis results in a ground motion file named gmbin, which
0
contains the free-field ground motion at the interface ( u b ) that is to be used as input in a transient
analysis. Note that the external domain in the site-response analysis should be linearly elastic and
modeled using the MAT_ELASTIC material model.
45
2. Build the soil-structure model for transient analysis. This model includes the internal domain 
(with the structure) surrounded by a few layers of elements corresponding to the external domain,
which is again surrounded by a few layers of elements of the PML material acting as an
absorbing boundary. Figure 3-8 illustrates the cross-section and isometric views of a sample LS-
DYNA model for transient analysis in the effective seismic input method. The red elements in the
figure correspond to the structure, the green elements correspond to the soil in the internal
domain, the blue elements correspond to the external domain, and the gray elements correspond
to the PML boundary. The boundary  between the internal and external domains should be
modeled using *INTERFACE_SSI for the transient analysis. This interface model reads the free-
field ground motions from the gmbin file, and applies equivalent forces calculated using equation
(3-1) at the  boundary.
3. For cases that require gravity loading on the structure, a preliminary static analysis is needed. The
model for static analysis is similar to that of the transient analysis, except that PML boundary is
not included. The boundary  is modeled using the *INTERFACE_SSI_STATIC with the same
interface ID that will be used in the transient analysis. This interface model records the gravity
deformations on the boundary and applies them in the transient analysis. The static analysis with
gravity loads can be performed using dynamic relaxation, described in the keyword user manual
(LSTC, 2013).
a. Cross section b. Isometric view
Figure 3-8: Sample model for SSI analysis using the effective seismic input in LS-DYNA
46
As mentioned in Section 3.2.3, the external domain (blue in Figure 3-8) can be modeled using a
linear elastic material (*MAT_ELASTIC in LS-DYNA). The internal domain (green in Figure 3-8) can
be modeled with any of the soil models available in LS-DYNA. The soil-foundation interface and
damping can be modeled as in the direct method described in the previous section.
3.3.3 The US NRC ESSI Simulator
The US Nuclear Regulatory Commission Earthquake Soil-Structure Interaction Simulator (US

NRC ESSI Simulator) is a collection of hardware, software, and related documentation, specifically
developed for soil-structure interaction simulations of nuclear structures. It is under development by
Jeremic et al. (2011) through funding from the US Nuclear Regulatory Commission and the Lawrence
Berkeley National Laboratory. The primary motivation for the ESSI simulator is to perform high-
performance soil-structure interaction simulations with high fidelity that reduces modeling uncertainties
in the response3. The ESSI simulator program can perform time-domain nonlinear finite element analysis
and is equipped with various element types, material models and contact algorithms. Soil-structure
interaction analysis in the ESSI Simulator is performed using the domain reduction method described in
Section 3.2.3. The theory and application procedures are described in the accompanying documentation
(Jeremic et al., 2013), which also includes the details of verification and validation of the program. The
ESSI simulator computer uses a parallel processing framework for efficient computation, with the
hardware being under constant development. Further details regarding the ESSI simulator can be found
on the website: http://sokocalo.engr.ucdavis.edu/~jeremic/ESSI_Simulator/.
3.3.4 Assessment of numerical soil-structure interaction analysis methods
The state-of-the-art of SSI analysis in the nuclear industry of the United States involves
performing SSI analysis using the linear, frequency-domain program, SASSI with equivalent linear,
strain-compatible properties for the soil. Indeed, the linear frequency-domain programs should accurately
predict responses for low-intensity ground motions involving almost linear response in the soil. However,
for intense earthquake shaking involving large soil strains, nonlinear, time-domain methods using
hysteretic soil models are theoretically more appropriate than the frequency-domain methods. However
nonlinear time-domain methods of SSI analysis have been developed only recently, and are not yet well
understood or established.
3
Modeling uncertainties are the uncertainties in response that are a consequence of numerical modeling.
These can include uncertainties due to simplifying assumptions such as modeling a nonlinear system using
equivalent linear properties and ignoring secondary nonlinearities such as gapping and sliding of the foundation.
47
Advancement of the usage of nonlinear time-domain programs requires detailed assessments that
compare the predictions of time-domain and frequency-domain programs for a range of ground motions,
soil profiles and structural characteristics. The first step of this assessment is to benchmark the time-
domain programs against the frequency-domain programs for low-intensity ground motions, or with
linear soil properties, and affirm that they result in similar predictions for cases involving linear soil and
structural response. The predictions of these programs for cases involving nonlinear material models and
larger strains are expected to differ. For these cases the assessment studies should examine the differences
and provide guidelines to practicing engineers on the usage of these programs.
To the knowledge of the author, very few studies assessing numerical SSI programs have so far
been performed, mainly due to the contemporariness of the time-domain, nonlinear methods. In one such
study Xu et al. (2006) compared the predictions made using SASSI and LS-DYNA from the SSI analyses
of deeply embedded nuclear structures. They found that the results calculated using SASSI and LS-
DYNA differed considerably for both linear and nonlinear analyses. While the differences are expected
for nonlinear analyses, they stipulated that the results from linear analyses performed in the two programs
diverged due to the differences in the corresponding damping formulations. However similar studies
performed by Anderson et al. (2013b) and Coronado et al. (2013) showed that the linear SSI analyses of
deeply embedded nuclear structures using time-domain and frequency-domain programs resulted in very
similar structural responses. In these studies, Anderson et al. (2013b) compared the results from SAP2000
(Computers and Structures Inc., 2011) to those from SASSI2010 (Ostadan and Deng, 2011), while
Coronado et al. (2013) compared the results from the extended subtraction method in SASSI2010 to those
calculated using the commercial finite element program ANSYS (ANSYS Inc., 2013).
3.4 Structure-soil-structure interaction
A direct consequence of soil-structure interaction is the interaction of adjacent structures through

the soil around them, namely, structure-soil-structure interaction (SSSI). Although SSI has been studied
extensively, very few studies of SSSI have been attempted. These studies are mostly limited to analytical
calculations of highly simplified structures founded on idealized soil profiles. For example, the seminal
study by Luco and Contesse (1973) examined anti-plane interaction between two infinitely long shear
walls subjected to vertically incident SH waves of harmonic time-dependence. Wong and Trifunac (1975)
extended this study to an array of several structures with varying size and stiffness subjected to a shear
wave incident at an arbitrary angle. Both of these analytical studies investigated the significance of
parameters such as separation distance, foundation size, and stiffness of the structures on SSSI. Luco and
Contesse (1973) identified the factors that determine the degree of interaction between structures as a)
relative foundation sizes, b) distance between the structures, c) mass of the superstructure relative to the
48
mass of the soil excavated for the foundation, d) mass of the foundation relative to the mass of excavated
soil, and e) relative stiffness of the structures and the soil. Parametric analyses were performed and it was
concluded that SSSI effects are especially important for smaller and lighter structures situated close to
heavier structures. A similar conclusion was drawn by Wong and Trifunac (1975). Both studies noted that
the degree of interaction depends mainly on the type of wave interference (constructive or destructive)
occurring between the scattered waves from the foundations, which is a function of the spacing and
arrangement of the foundations. Some studies [e.g., Matthees and Magiera (1982), Lin et al. (1987), Qian
and Beskos (1996), Padron et al. (2009), Anderson et al. (2013a), Bolisetti and Whittaker (2011)] have
investigated SSSI between adjacent structures (buildings and nuclear power plants) using more rigorous
three-dimensional numerical methods, and most of these studies concluded that SSSI could considerably
affect structural response. A more comprehensive list of SSSI studies can be found in Menglin et al.
(2011).
A primary reason for the lack of studies of SSSI is the unavailability of case studies either from
field data or laboratory experiments. The only case study, to the knowledge of the author, was performed
by Çelebi (1993b), who analyzed data recorded from two well-instrumented neighboring buildings to
examine SSSI. The layout of these buildings and the instrumentation is presented in Figure 3-9. Çelebi
(1993b) employed spectral analysis techniques to find correlations between the responses of the buildings
to the 1987 Whittier-Narrows earthquake. The study examined the roof response, basement response, and
free-field response of the buildings, and concluded that 1) the dynamic response of the building affected
the basement response and the free-field response, and 2) there was considerable SSSI between the
structures at specific frequencies. The latter conclusion also corroborates the findings of the study by
Murakami and Luco (1977) who performed an analytical investigation of the interaction between an array
of idealized shear-wall-like structures, and found that the interaction between these structures was the
most prominent at the, so called, Rayleigh frequencies, which are determined by the properties of the soil
and sizes of the foundation (Murakami and Luco, 1977).
49
Figure 3-9: Layout of the buildings considered by Çelebi (1993a) and Çelebi (1993b) for examining
SSSI
50
SECTION 4
AN ASSESSMENT OF SITE-RESPONSE ANALYSIS PROGRAMS
4.1 Introduction
This section presents an assessment of the state-of-the-art one-dimensional site-response analysis

programs that are widely used in practice. These programs are: 1) SHAKE, which performs an
equivalent-linear site-response analysis in the frequency domain, 2) DEEPSOIL, which performs a
nonlinear site-response analysis in the time domain, and 3) LS-DYNA, which is a commercial finite
element analysis program that is capable of time-domain nonlinear analysis. A description of one-
dimensional site-response analysis using these programs is presented in Section 2. This section includes a
comparison of site responses calculated using these programs while accounting for different site
conditions and earthquake ground motions encountered in the United States. The results of these analyses
and their comparisons are used to evaluate the efficiency of these programs and also identify the issues
encountered while performing site-response analysis.
4.2 Soil profiles used for site-response analysis
Four hypothetical soil profiles are considered in this study and their properties are chosen to
reflect the site conditions encountered in the United States. Figure 4-1 presents the soil profiles in which,
sites W1, and W2 are representative of the Western United States (WUS), and sites E1, and E2 are
representative of the Central and Eastern United States (CEUS).
51
Figure 4-1: Soil profiles used for site-response analyses
Although more sites are required for a comprehensive representation, these four sites are deemed
sufficient for a high-level assessment. Each site is 100 m deep and is underlain by linear viscous bedrock.
Figure 4-1 also presents the linear material properties (small-strain shear modulus, VS , and mass density,
 ) of the different soil layers and the bedrock, and the nonlinear properties (modulus reduction and
damping curves) are presented in Section 4.4. Since the present study only involves comparisons between
the solutions of different analysis programs and does not involve recorded data, the choice of idealized,
hypothetical soil profiles (as opposed to real soil profiles) is reasonable.
4.3 Selection of input ground motions
Single component ground motions are used for input in one-dimensional site-response analyses. A total of
nine ground motions are used: six motions for the WUS soil profiles (three ordinary motions and three
near-fault pulse motions with forward directivity), and three motions for the CEUS soil profiles. Details
of these ground motions are presented in Table 4-1. The ordinary WUS motions and the CEUS motions
are chosen to cover a range of frequency contents and peak accelerations. The pulse motions for the WUS
52
soil profiles are chosen to cover a range of pulse periods. Figure 4-2 presents the acceleration time series
and Figure 4-3 presents the acceleration response spectra and Fourier response spectra of the WUS
ground motions. The amplitude of the pulse motion WP3 is significantly less than the other pulse
motions, and is presented along with the ordinary ground motions in Figure 4-3. The acceleration time
series, acceleration response spectra and the Fourier amplitude spectra of the CEUS motions are presented
in Figure 4-4 and Figure 4-5. The WUS ordinary motions and the CEUS motions are picked from a
database of ground motions assembled by Huang et al. (2009), which includes a set of ground motions for
a WUS site (Diablo Canyon Nuclear Generating Station, California) and another set for a CEUS site
(North Anna Power Station, Virginia). Two of the pulse motions (WP1 and WP2) are chosen from the
suite of ground motions used in the CityBlock project, and one pulse motion (WP3) is chosen from the
database of pulse motions created by Baker (2007). The input ground motions are applied at the base of
the soil profile through a transmitting boundary (outcrop input 4 ) for all analyses except for those
involving site W2 subjected to pulse motion, which are performed using a reflecting boundary. The
reason for choosing a reflecting boundary for these cases is described in Section 4.7.4.3.
4
The outcrop ground motion at a particular layer is the ground motion that would be recorded if there were
no layers above. The actual ground motion recorded in a particular layer is termed a within ground motion. An
outcrop input is the method of ground motion input that is used when the base of the soil profile is assumed to be a
transmitting boundary (also referred to as flexible bedrock). A within input is the method of input that is used when
the base is assumed to be a rigid boundary (also referred to as rigid bedrock). For specifying an outcrop input, the
bedrock is modeled using a set of dampers that absorb the waves reflected from the surface, which in reality, radiate
into the halfspace. The ground motion input is specified at the base of the soil profile as a shear force history that is
proportional to the velocity history of the outcrop ground motion at this depth. For a within input the soil profile is
assumed to have a fixed base and the within ground motion at this depth is directly applied as an acceleration (or
velocity or displacement) history. Therefore an outcrop ground motion is used to specify an outcrop input, and a
within ground motion is used to specify a within input. These procedures for providing the earthquake input to a soil
profile are presented in more detail in Section 2.2.2.3.
53
Table 4-1: Input ground motions used for site-response analysis
CEUS ground motions
Ground
motion Earthquake (Year) Station Component Peak acc. (g)a
ID
EO1 Mineral, VA (2011) Charlottesville 090 0.101
New Hampshire
EO2 Franklin Falls Dam 315 0.313
(1982)
Saguenay, Canada
EO3 Dickey 090 0.092
(1988)
WUS ordinary ground motions
WO1 Northridge (1994) Vasquez Rock Park 000 0.151
WO2 Northridge (1994) Wonderland Ave 185 0.172
WO3 San Fernando (1971) Lake Hughes 111 0.192
WUS pulse motions
Ground
Peak acc.
motion Earthquake (Year) Station Component Tp (sec)b
(g)a
ID
WP1 Landers (1992) Lucerne 260 0.727 5.1
WP2 Northridge (1994) Rinaldi 228 0.825 1.5
WP3 Chi Chi (1999) TCU128 128 0.187 9.0
a
Peak ground acceleration of the recorded ground motions
b
Pulse period as specified in the PEER NGA ground motion database (PEER, 2011) for WP1 and WP2,
and as calculated by Baker (2007) for WP3
54
Figure 4-2: Acceleration time series of the input ground motions used for WUS soil profiles
55
a. Acceleration response spectra
b. Fourier transforms
Figure 4-3: Acceleration response spectra and Fourier amplitude spectra of the input ground
motions used for WUS soil profiles
56
Figure 4-4: Acceleration time series of the input ground motions used for CEUS soil profiles
57
a. Acceleration response spectra
b. Fourier transforms
Figure 4-5: Acceleration response spectra and Fourier amplitude spectra of the input ground
motions used for CEUS soil profiles
4.4 Numerical modeling
General procedures for one-dimensional site-response analysis in SHAKE, DEEPSOIL and LS-
DYNA are described in Section 2 along with the corresponding soil material models. This section
presents the numerical modeling of the soil profiles and the associated material properties. To enable a
comparison of results from different programs, efforts are made to maintain the consistency between their
numerical models and material properties. Although the material models employed in the programs are
different, the corresponding model parameters are chosen such that the resultant dynamic soil properties
(modulus reduction and damping curves) are similar.
58
4.4.1 Material modeling
Three types of materials are used for the soil: Sand ( VS  300 m/sec), Soft rock ( VS  1000 m/sec)
and Hard rock ( VS  2500 m/sec). Linear viscous properties are assumed for the bedrock of the CEUS and
WUS sites along with shear wave velocities of 2500 m/s and 1000 m/s, respectively. The mass densities
and small-strain shear-wave velocities of all the materials are shown in Figure 4-1 and are input “as is”.
The dynamic soil properties are defined differently in each program based on the material model, but the
parameters of these models are chosen such that the resultant properties are similar. The backbone curves
and damping curves calculated from these parameters are presented in Figure 4-6 below. The backbone
curves used for the different programs match very closely, but small differences exist between the
damping curves.
The three numerical programs, SHAKE, DEEPSOIL and LS-DYNA employ different numerical
models to simulate the stress-strain response of the soil. These numerical models are described in Sections
2.2.2.2.1, 2.2.2.3.1 and 2.2.2.3.2, respectively. SHAKE requires a direct input of the modulus reduction
and damping curves as a set of discrete points. DEEPSOIL provides the choice of two nonlinear material
models: the ‘MRDF Pressure-Dependent Hyperbolic Model’, and the ‘Pressure-Dependent Hyperbolic
model’ (Hashash et al., 2011). The Pressure-Dependent Hyperbolic Model is used for site-response
analysis in this study and is referred to as the ‘hyperbolic model’ hereafter. Modeling in LS-DYNA is
performed using the MAT_HYSTERETIC_SOIL model. The backbone curves presented in Figure 4-6
are taken from a collection of material property data provided by Arup. Identical modulus reduction
curves are assumed for soft rock and hard rock, which is an acceptable assumption because neither rock
type experiences significant strain softening compared to the soil layers. DEEPSOIL hyperbolic curves
are fit into these data and the corresponding model parameters are calculated for each material as shown
in Table 4-2. Discrete points are then taken from the DEEPSOIL curves and are used as input for SHAKE
and LS-DYNA to maintain equivalence between the material properties used in different programs and
enable a fair comparison of the corresponding responses. DEEPSOIL also calculates and provides the
damping curves for each material, which are directly input in SHAKE. The LS-DYNA damping curves
shown in Figure 4-6 are independently calculated by integrating the stress-strain loops for each strain
amplitude. Preliminary analyses using the MAT_HYSTERETIC_SOIL model revealed an unrealistically
large contribution of higher frequencies in the acceleration response, especially in cases involving large
strains. To examine this issue an alternative approach was adopted that involved increasing the number of
points in the backbone curve to 120. This was performed by modeling each layer by superimposing 12
solid elements (12 elements sharing the same eight nodes) as opposed to using a single element in the
usual approach (see Section 2.2.2.3.2 for the description of site-response analysis using finite element
59
programs such as LS-DYNA). Each of these elements has 10 points in the backbone curve, and these
curves are chosen such that their sum equals the DEEPSOIL backbone curve of the corresponding
material. The superposition causes the elements to act in parallel, and the resultant stress value at a
particular strain equals the sum of the stresses in each element. Thus each 12-element set acts as a single
element with 120 points in the backbone curve. This approach is only adopted for the WUS sites since
very small strains are expected in the CEUS sites for the chosen ground motions. The 120-point backbone
curves for Sand and Soft rock are presented in Figure 4-6 along with the backbone curves and damping
curves. In this figure LS-DYNA 10pt represents the 10-point backbone curve, while LS-DYNA 120pt
represents the 120-point backbone curve. As seen in the figure, the curves from each program match
closely, except for slight differences in the damping curves in LS-DYNA 10pt, which is due to the multi-
linear shape of its stress-strain curve.
The analyses for WUS soil profiles are also performed with pressure-dependent properties for the
soil, to model a more realistic variation of the shear strength with confining pressure. This variation is
modeled in the MAT_HYSTERETIC_SOIL model in LS-DYNA by using the values 0, 0, 0 and 1 for the
parameters b , a0 , a1 and a2 , respectively, and a reference pressure equal to the confining pressure at a
depth of 20m [refer to equations (2-11) and (2-12) or the LS-DYNA keyword manual (LSTC, 2013) for a
description of these parameters]. The backbone curve at the reference pressure is taken as the curve
shown in Figure 4-6. From equation (2-12), incorporating pressure dependence increases the shear
strength of the layers below the reference point and decreases the shear strength of the layers above the
reference depth. An equivalent pressure-dependence of the soil properties is achieved in DEEPSOIL by
using a value of 1 for b , which results in a linear variation of  r [refer to equations (2-7) and (2-8) for the
parameters b and  r ] with confining pressure. These parameters used in the DEEPSOIL and LS-DYNA
material models simulate a linear increase in strength with the confining pressure. The material curves
calculated by DEEPSOIL using these parameters are also used in the SHAKE input.
DEEPSOIL also calculates the small-strain damping for each material from the stress-strain data.
From the stress-strain data used in this study DEEPSOIL calculates a small-strain damping ratio of about
1.1%, which used along with the ‘frequency-independent’ damping option. This damping ratio is also
incorporated into LS-DYNA using the *DAMPING_FREQUENCY_RANGE option, which provides
frequency-independent damping in the specified frequency range (2 to 50 Hz for this study). A maximum
strain increment of 0.005 for ordinary motions, and 0.001 for pulse motions, is used for the DEEPSOIL
analyses. The widely adopted effective shear strain ratio of 0.65 is used for all SHAKE analyses.
60
Figure 4-6: Backbone and damping curves of the materials used in site-response analysis
Table 4-2: Model parameters used to define the soil materials in DEEPSOIL
Sand Soft rock Hard rock

Small-strain damping ratio1,  (%) 1.09 1.13 1.13
Reference strain2,  r (%) 0.0188 0.0218 0.0218
Stress-strain curve parameter,  1.050 1.455 1.455
Stress-strain curve parameter, s 0.705 0.780 0.780
1
Pressure dependence of damping is ignored by setting the parameter d to 0
2
Pressure dependence of the stress-strain curve is ignored (included) by setting the parameter b to 0 (1)
4.4.2 Site layering
The thickness of the soil layers in a numerical model governs the shortest wavelength (which
corresponds to the highest frequency) of the shear waves that can be captured through the simulations.
61
Softer materials require a smaller layer thickness compared to stiffer materials to capture the same
maximum frequency. The maximum layer thickness is given by the equation:
min Vs
tmax   (4-1)
n n  f max
where tmax is the maximum thickness of the soil layers, min is the minimum wavelength of the shear
waves, Vs is the shear wave velocity, f max is the maximum frequency of the shear waves, and n is the
desired number of layers per shortest wave length. The inequality in equation 3-1 governs the ‘resolution’
with which the shortest wavelength is captured during wave propagation in the soil layers. For example,
when a value of 10 is used for n , the shortest wavelength is captured by at least 10 soil layers, namely 10
values of displacement. Therefore a larger value of n results in a higher resolution of the wave
propagation in the soil profile as well as higher computational requirements. The second part of equation
3-1 is the standard equation that relates wavelength and frequency. Equation (4-1) gives an accurate
maximum thickness for a linear analysis, where the shear modulus (and therefore the shear wave velocity)
remains constant through time. Nonlinear analyses involve softening of the layers in time and therefore
require a smaller maximum thickness (larger number of layers) to capture a certain maximum frequency
throughout the analysis.
Site-response analysis for traditional structures is typically performed using a maximum
frequency of 15 Hz, but a considerably higher frequency is generally needed for safety-related nuclear
applications. Therefore the layer thicknesses for this study are chosen to accommodate a maximum
frequency of at least 30 Hz (in a linear analysis), while maintaining a reasonable computing time. The
SHAKE user manual (Schnabel et al., 2009) recommends a value of five for n , but a value of 10 is
chosen for this study, to account for the higher frequencies as well as for the softening in nonlinear
analyses. The layer thicknesses calculated using this value and their corresponding maximum frequencies
(captured in a fully linear analysis) are presented in Table 4-3 for each of the three materials.
Considering the thicknesses presented in the table above, site W1 is modeled using 50 layers of
soft rock, site W2 is modeled using 100 layers of sand, site E1 is modeled using 25 layers of hard rock,
and site E2 is modeled using 28 layers of sand and 18 layers of hard rock.
62
Table 4-3: Layer thicknesses and captured maximum frequencies in each material
Maximum
Shear wave Layer
Material frequency (Hz)
velocity (m/s) thickness (m)
assuming n=10
Sand 300 1 30
Soft Rock 1000 2 50
Hard Rock 2500 4 62.5
4.5 Numerical analyses
4.5.1 Modal analyses
Prior to performing site-response analyses, modal analysis is performed for each soil profile
model. LS-DYNA is used for this purpose and a viscous damping ratio of 1.1% is assumed for all models.
The resonant time periods of a uniform soil profile (such as site E1, site W1 and site W2) can also be
calculated analytically using the equation
4H
Tn  (4-2)
(2n  1) Vs
where, Tn is the natural frequency of the nth mode, Vs is the shear-wave velocity of the soil material, and
H is the height of the soil profile (Kramer, 1996). The numerically and analytically calculated time
periods of the soil profiles are presented in Table 4-4 below, for the first two translational modes. The
time periods calculated from the LS DYNA analyses match well with the analytical calculations for the
uniform soil profiles, site E1, site W1 and site W2.
63
Table 4-4: Time periods of the soil profiles from modal analysis
Modal time periods Modal time periods

calculated using LS-DYNA (sec) from analytical calculations (sec)
Soil profile
1st mode 2nd mode 1st mode 2nd mode
Site E1 0.16 0.05 0.16 0.05
Site E2 0.39 0.14 - -
Site W1 0.40 0.13 0.40 0.13
Site W2 1.32 0.44 1.33 0.44
4.5.2 Site-response analyses for sites E1 and E2
The analyses for sites E1 and E2 are performed using SHAKE, DEEPSOIL and LS-DYNA (using
the MAT_HYSTERETIC_SOIL model with 10-point backbone curve) for each of the ground motions
EO1, EO2 and EO3. Ground motions are input at the base of the soil profile with a transmitting boundary.
An effective shear-strain ratio of 0.65 is used for the SHAKE analyses, and a maximum strain increment
of 0.005 is used for the DEEPSOIL analyses.
4.5.3 Site-response analyses for site W1
Site-response analyses for site W1 are performed using four numerical methods: SHAKE,
DEEPSOIL, LS-DYNA using the 10 point backbone curve (referred to as LS-DYNA 10pt) and LS-
DYNA using the 120 point backbone curve (referred to as LS-DYNA 120pt). Analyses are performed for
each of the ordinary ground motions WO1, WO2 and WO3, and the pulse motions WP1, WP2 and WP3.
Both pressure-independent (PI) and pressure-dependent (PD) properties are considered. Ground motions
are input at the base of the soil profile as an outcrop. An effective shear-strain ratio of 0.65 is used for the
SHAKE analyses. For the DEEPSOIL analyses, a maximum strain increment of 0.005 is used for the
ordinary ground motions and 0.001 for the pulse motions.
4.5.4 Site-response analyses for site W2
Site-response analyses for site W2 are performed using SHAKE, DEEPSOIL, LS-DYNA 10pt
and LS-DYNA 120pt for each of the ordinary ground motions WO1, WO2 and WO3, and the pulse
64
motions WP1, WP2 and WP3. Both pressure-independent (PI) and pressure-dependent (PD) properties
are considered. The ordinary ground motions are input at the base of siteW2 as an outcrop (force input)
and the pulse motions are input directly as within motions (acceleration input). An effective shear strain
ratio of 0.65 is used for the SHAKE analyses and a maximum strain increment of 0.001 is used for the
DEEPSOIL analyses for all ground motions.
4.6 Observations
Figure A-1 to Figure A-78 present the results of site-response analyses of all the soil profiles. The
results calculated from different numerical programs are presented together in each figure to enable a
comparison. Observations gathered from these results are presented in this section. Further insights into
some of these observations are presented in Section 4.7.
The results of each site-response analysis are presented in terms of 1) stress-strain loops,
acceleration response spectra and acceleration response histories at three different depths in the soil
profile (depths are indicated in each figure), 2) the peak acceleration and peak shear strain profiles and 3)
acceleration amplification spectra at the surface. The acceleration amplification spectrum at the surface is
calculated for each analysis as the ratio of the spectral acceleration at the surface to the spectral
acceleration at the base of the soil profile. For sites E1, W1 and W2, the acceleration response spectra,
acceleration response histories and the stress-strain loops are calculated at 1) the surface, 2) mid-depth,
and 3) the base. For site E2, which is the site composed of 28m of sand lying atop of 72m of hard rock,
the acceleration response spectra and response histories are calculated at 1) the surface, 2) sand-hard rock
interface, and 3) at the base. The stress-strain loops for site E2 are calculated at 1) the lowest layer of
sand, 2) the topmost layer of hard rock and 3) the base layer. While the accelerations in a particular layer
are measured at the top of the layer, the stresses and strains are calculated at the midpoint. Therefore
small differences exist between the depths at which the accelerations and the corresponding stress-strain
loops are calculated. Additionally due to some constraints in calculating the acceleration output in
DEEPSOIL, the base acceleration response spectra and response histories from all programs are
calculated one layer above the bottom of the soil profile. Since the soil profiles have different layer
thicknesses small differences exist between the depths at which the base accelerations are calculated (96m
for sites E1 and E2, 98m for site W1 and 100m for siteW2). The figures presenting the acceleration
response spectra also include the input ground motion at the base. The acceleration response spectra,
acceleration amplification spectra, peak acceleration profiles and peak strain profiles for sites W1 and W2
also include the results calculated with pressure-dependent properties for the soil. For the corresponding
figures, the pressure-independent results are indicated by PI and the pressure-dependent results are
indicated by PD.
65
4.6.1 Site E1
Site E1 is a uniform site composed of hard rock and is the stiffest of the four sites. Figure A-1 to
Figure A-13 present the site-response results of site E1 subjected to ground motions EO1, EO2 and EO3
(peak accelerations of 0.10g, 0.31g and 0.09g). Figure A-1 to Figure A-3 present the stress-strain loops
for the three ground motions, and are followed by the peak acceleration and peak shear strain profiles
(Figure A-4 to Figure A-6), acceleration response spectra (Figure A-7 to Figure A-9), acceleration
response histories (Figure A-10 to Figure A-12), and the spectral amplifications (Figure A-13). Each of
these figures includes the responses calculated using SHAKE, DEEPSOIL and LS-DYNA 10pt.
a. Hysteretic behavior and the extent of nonlinear response: The stress-strain loops presented in
Figure A-1, Figure A-2 and Figure A-3 indicate that the soil layers in site E1 remain almost linear
for the three ground motions. The peak strain profiles (Figure A-4, Figure A-5 and Figure A-6)
calculated from SHAKE, DEEPSOIL and LS-DYNA 10pt are quite similar for all the ground
motions. The maximum peak shear strain is close to 0.0003% for ground motion EO1, 0.0012%
for ground motion EO2, and 0.001% for EO3. At these strain levels the modulus reduction ratios
in hard rock are approximately 0.95, 0.89 and 0.90, respectively, which indicate slight
nonlinearity. The equivalent viscous damping ratios for these strain levels lie between 2% and 4%
of critical for the three ground motions.
b. Acceleration response: The acceleration response spectra (Figure A-7, Figure A-8 and Figure
A-9) calculated using SHAKE, DEEPSOIL and LS-DYNA 10pt are very similar for the three
ground motions. There are small differences at short periods (< 0.05 sec) that are more apparent
in the amplification spectra in Figure A-13. However the acceleration response histories (Figure
A-10 to Figure A-12) and the peak acceleration profiles (Figure A-4 to Figure A-6) are in good
agreement for the three ground motions. Although differences between the responses of SHAKE
and the time-domain programs are expected for nonlinear soils, the minor differences between
DEEPSOIL and LS-DYNA 10pt at these strain levels can be attributed to the differences in the
backbone curves and the viscous damping formulations of the two programs.
c. Site amplification: Site E1 has the first two translational modes at 0.16 sec and 0.05 sec. The
acceleration amplification spectra in Figure A-13 clearly show the peaks around the first modal
period. The amplification spectra for ground motion EO1 show another peak between 0.05 and
0.06 sec, which corresponds to the second mode. The strongest ground motion, EO2, results in
the smallest maximum amplification at the elongated site period (see Figure A-13), which is due
to the greater energy dissipation provided by the nonlinear hysteretic behavior of the soil.
Additionally the amplifications calculated from different programs match quite well, except for
slight differences at periods less than 0.05 sec.
66
d. Variation of outcrop motion through depth: The acceleration response spectrum computed at the
surface of site E1 is very close to the acceleration response spectrum of the outcrop motion at the
base for all the ground motions. This similarity of the outcrop motions at the surface and the base
indicates that the outcrop motion in site E1 does not change significantly with depth. Therefore if
site E1 was an outcropping rock, the rock outcrop motion at the surface of this site can be directly
used as an outcrop input for a nearby soil site, without performing a deconvolution analysis.
However the outcrop motion and the within motion at the base are quite different, indicating that
the presence of the hard rock soil profile significantly affects the response at the base. This
observation and its significance are discussed further in Section 4.7.5.
4.6.2 Site E2
Site E2 is non-uniform and is comprised of a 28m deep layer of sand atop of 72m of hard rock.
Figure A-14 to Figure A-26 present the site-response results of site E2 subjected to ground motions EO1,
EO2 and EO3 (peak accelerations of 0.10g, 0.31g and 0.09g). Figure A-14 to Figure A-16 present the
stress-strain loops for the three ground motions, and are followed by the peak acceleration and peak shear
strain profiles (Figure A-17 to Figure A-19), acceleration response spectra (Figure A-20 to Figure A-22),
acceleration response histories (Figure A-23 to Figure A-25), and the spectral amplifications (Figure
A-26). Each of these figures includes the responses calculated using SHAKE, DEEPSOIL and LS-DYNA
10pt.
a. Hysteretic behavior and the extent of nonlinear response: The peak strain profiles presented in
Figure A-17 to Figure A-19 show that the highest peak strains induced by the three ground
motions in site E2 are about 0.007% for EO1 and 0.025% for EO2. For ground motion EO3, the
highest peak strains calculated by the three programs vary between 0.01% and 0.018%. Given
that the largest strains always occur in the sand layers, these strain levels correspond to modulus
reduction ratios of about 0.65 and 0.44 for EO1 and EO2, respectively, and lie between 0.5 and
0.6 for EO3. The corresponding equivalent viscous damping ratios are about 8%, 13% and 10%,
respectively for the three ground motions. At these strain levels, the stress-strain loops presented
in Figure A-14 to Figure A-16 indicate that the hard rock layers remain almost linear, while the
sand layers show considerable nonlinear behavior. The peak strain profiles calculated using
SHAKE, DEEPSOIL and LS-DYNA 10pt are similar for EO1 and EO2, but differ significantly in
the sand layers for EO3. SHAKE results in the lowest values for peak strains, followed by LS-
DYNA 10pt and DEEPSOIL.
b. Acceleration response: Examination of the acceleration response spectra in Figure A-20, Figure
A-21 and Figure A-22 indicates that the responses calculated by SHAKE, DEEPSOIL and LS-
67
DYNA 10pt match well for periods above 0.1 sec. The differences at shorter periods are
considerable, and are greater than those seen in site E1 for the same ground motions. Additionally
these differences are significantly greater in the surface response than in the response at 28m
depth and at the base, which are in the hard rock layers, indicating that the differences in the
acceleration responses increase with strain. SHAKE underestimates the short-period accelerations
on the surface as compared to DEEPSOIL and LS-DYNA. Additionally the SHAKE response
spectra in these periods tend to ‘flatten out’ indicating the absence of the corresponding
frequencies in the acceleration response. This is also evident from the SHAKE acceleration
response history, which is smooth compared to that of DEEPSOIL and LS-DYNA (Figure A-23
to Figure A-25) indicating a smaller contribution of higher frequencies in the response.
Additional discussion on the high-frequency response issue in SHAKE is presented in bullet 2 of
Section 4.7.2. The peak acceleration profiles (Figure A-17 to Figure A-19) calculated from the
three programs match reasonably well for all ground motions in the hard rock layers, but are
significantly different in the sand layers for ground motion EO3. SHAKE underestimates the peak
accelerations in the sand layers for this ground motion, which is a consequence of
underestimating the short period accelerations. The differences between the peak accelerations
calculated by DEEPSOIL and LS-DYNA 10pt are also considerable, which is a result of the
differences in the hysteretic models employed in the two programs.
c. Site amplification: Site E2 has the first two translational modes at 0.39 sec and 0.14 sec. The
amplification spectra in Figure A-26 show peaks corresponding to the elongated site period,
which lies between 0.4 and 0.5 sec for ground motions EO1 and EO3, and between 0.5 and 0.6
sec for EO2. However the peaks corresponding to the second mode are seen only for ground
motions EO1 and EO2. The elongated second mode period is slightly less than 0.15 sec for EO1
and lies between 0.15 and 0.2 sec for EO2. As observed for site E1, the spectral amplification at
the site period of E2 is the smallest for the strongest ground motion, EO2. Additionally SHAKE
overestimates this amplification compared to the time-domain programs, which suggests that the
equivalent viscous damping is underestimated. This is consistent with the observation that
SHAKE underestimates the peak strains in the soil layers, which results in the underestimation of
the equivalent viscous damping.
d. Variation of outcrop motion through depth: In contrast with site E1, the outcrop motion at the
surface of site E2 is different from the outcrop at the base, with the largest differences seen in
ground motion EO2 (Figure A-21). However the acceleration response at 28m depth, which
corresponds to the sand-hard rock interface, approaches the base outcrop ground motion. This
suggests that the presence of the sand layer above the relatively stiff hard rock does not affect the
68
ground motion at the interface significantly and therefore the outcrop and within motions are
similar at this depth. This observation and its significance are discussed further in Section 4.7.5.
4.6.3 Site W1
4.6.3.1 Response to ordinary ground motions
Site W1 is uniform site composed of soft rock. Figure A-27 to Figure A-39 present the site-
response results of site W1 subjected to ground motions WO1, WO2 and WO3 (peak accelerations of
0.15g, 0.17g and 0.19g). Figure A-27 to Figure A-29 present the stress-strain loops for the three ground
motions, and are followed by the peak acceleration and peak shear strain profiles (Figure A-30 to Figure
A-32), acceleration response spectra (Figure A-33 to Figure A-35), acceleration response histories (Figure
A-36 to Figure A-38), and the spectral amplifications (Figure A-39). Each of these figures includes the
responses calculated using SHAKE, DEEPSOIL, LS-DYNA 10pt and LS-DYNA 120pt. Only the results
of the analyses performed using pressure-independent (PI) properties are discussed in this section. Results
of the analyses performed using pressure-dependent (PD) properties are discussed in Section 4.6.3.3.
a. Hysteretic behavior and the extent of nonlinear response: The peak strain profiles in Figure A-30,
Figure A-31 and Figure A-32 show that the largest strains calculated by different numerical
programs vary between 0.018% and 0.024% for WO1, 0.018% and 0.026% for WO2, and
0.006% and 0.008% for WO3. These strain ranges correspond to modulus reduction ratios that lie
between 0.38 and 0.44 for WO1, 0.37 and 0.44 for WO2, and 0.60 and 0.65 for WO3. The
equivalent viscous damping ratios corresponding to these ranges are approximately 15%, 15%
and 9% for WO1, WO2 and WO3, respectively. At these strain levels the stress-strain loops in
Figure A-27, Figure A-28 and Figure A-29 indicate that the lower layers of the soil profile show
considerable nonlinear behavior for the three ground motions. The stress-strain loops of LS-
DYNA 10pt and LS-DYNA 120pt match closely for all the ground motions indicating that
increasing the number of points in the backbone curve results in negligible changes in the
hysteretic response. The peak strain profiles show that SHAKE underestimates the peak strains
compared to the time-domain programs. Although the peak strains from LS-DYNA 10pt and LS-
DYNA 120pt match closely for all the ground motions, DEEPSOIL comparatively
underestimates the peak strains for ground motions WO2 and WO3. This can be attributed to the
differences in the hysteretic models employed in DEEPSOIL and LS-DYNA.
b. Acceleration response: The acceleration response spectra in Figure A-33, Figure A-34 and Figure
A-35 show differences in the responses calculated from SHAKE, DEEPSOIL, LS-DYNA 10pt
and LS-DYNA 120pt. These differences are greater in the short period range, especially for
69
periods less than 0.1 sec. The acceleration response spectrum calculated using SHAKE has near
constant amplitude at short periods, indicating the absence of the corresponding frequencies in the
response. This observation is consistent with the acceleration response histories presented in
Figure A-36, Figure A-37 and Figure A-38, which show that the responses calculated from
SHAKE are smoother compared to those calculated from the time-domain programs (see bullet 2
in Section 4.7.2 for additional discussion on this issue). The acceleration response spectra
calculated from the time-domain programs are similar except for periods less than 0.1 sec. The
response spectra calculated from LS-DYNA 120pt show smaller short-period accelerations
compared to LS-DYNA 10pt and DEEPSOIL, especially for ground motions WO1 and WO2.
Considerable differences exist between the peak accelerations computed by the different
programs, and these differences vary along the depth of the soil profile. A notable trend is that
LS-DYNA 120pt calculates smaller peak accelerations than LS-DYNA 10pt and DEEPSOIL in
most layers, especially for ground motions WO1 and WO2, which is consistent with the
underestimation of the short period accelerations by LS-DYNA 120pt for these ground motions.
c. Site amplification: Site W1 has the first two translational modes at 0.40 sec and 0.13 sec. From
the amplification spectra presented in Figure A-39, the elongated site period lies between 0.4 and
0.6 sec for ground motions WO1 and WO2, and between 0.4 and 0.5 sec for ground motion WO3.
The amplification spectra for ground motion WO3 also show a peak corresponding to the second
mode at about 0.15 sec. The amplifications calculated from the different programs are fairly close
at the site periods, but considerable differences arise for periods less than 0.1 sec. LS-DYNA 10pt
significantly overestimates the amplifications at periods greater than 5 sec compared to the other
programs, which is reduced in LS-DYNA 120pt.
d. Variation of outcrop motion through depth: Despite being a uniform soil profile, the acceleration
response spectrum computed at the surface of site W1 is considerably different from the
acceleration response spectrum of the outcrop motion at the base, unlike site E1. This indicates
that when larger strains are induced in the soil profile, the rock outcrop motion changes
considerably through the depth. Therefore if site W1 was an outcropping rock, the outcrop motion
calculated at the surface of this site cannot be directly used as an outcrop input at the base of a
nearby soil site. Instead, a deconvolution analysis would be required to calculate the outcrop
ground motion at depth. Detailed explanation of this observation is presented in Section 4.7.5.
4.6.3.2 Response to pulse-like ground motions
Figure A-40 to Figure A-52 present the site-response results of site W1 subjected to ground
motions WP1, WP2 and WP3. These ground motions have peak accelerations of 0.73g, 0.83g and 0.19g,
70
and pulse periods of 5.1 sec, 1.5 sec, and 9.0 sec respectively (see Table 4-1). Figure A-40 to Figure A-42
present the stress-strain loops for the three ground motions, and are followed by the peak acceleration and
peak shear strain profiles (Figure A-43 to Figure A-45), acceleration response spectra (Figure A-46 to
Figure A-48), acceleration response histories (Figure A-49 to Figure A-51), and the spectral
amplifications (Figure A-52). Each of these figures includes the responses calculated using SHAKE,
DEEPSOIL, LS-DYNA 10pt and LS-DYNA 120pt. Only the results of the analyses performed using
pressure-independent (PI) properties are discussed in this section. Results of the analyses performed using
pressure-dependent (PD) properties are discussed in Section 4.6.3.3.
Figure A-44 and Figure A-45 show that the largest strains calculated by the numerical programs
vary between 0.12% and 0.45% for WP1, 0.6% and 1.5% for WP2, and 0.05% and 0.09% for
WP3, which are an order of magnitude greater than the strains calculated for the ordinary ground
motions. These strain levels in soft rock correspond to modulus reduction ratios that lie between
0.06 and 0.15 for WP1, 0.03 and 0.05 for WP2, and 0.18 and 0.27 for WP3, which indicates
significant softening in the soil layers. The equivalent viscous damping ratios corresponding to
these ranges are between 22% and 28% for WP1, 30% and 35% for WP2, and 15% and 20% for
WP3. The stress-strain loops in Figure A-40, Figure A-41 and Figure A-42 also indicate
significant nonlinear behavior in the soil profile. The peak strain profiles show significant
differences between the peak strains calculated by SHAKE and the time-domain programs for the
three pulse motions. SHAKE underestimates the peak strains in the soil profile as compared to the
time-domain programs and the underestimation is much greater than that observed for the
ordinary motions. Of the time-domain programs, DEEPSOIL calculates the smallest peak strains;
the LS-DYNA 10pt and LS-DYNA 120pt results are similar. Additionally, Figure A-43 and
Figure A-44, which present the peak strain profiles for WP1 and WP2 respectively, show that
DEEPSOIL predicts large changes in the peak strains between consecutive layers, unlike LS-
DYNA 10pt and LS-DYNA 120pt, which predict a smoother strain profile. This DEEPSOIL
result is considered unrealistic and is examined more closely in Section 4.7.4.2.
b. Acceleration response: A comparison of the acceleration response spectra (Figure A-46 to Figure
A-48) calculated by the different programs shows larger differences for the pulse motions than
the ordinary ground motions, especially in the short period range. The acceleration response
spectra calculated by SHAKE are significantly different from those calculated by the time-
domain programs, especially for ground motion WP2. As seen for the ordinary ground motions,
the SHAKE acceleration spectra show near constant amplitude at periods less than 0.1 sec,
indicating the absence of the corresponding frequencies in the response (see bullet 2 in Section
71
4.7.2 for additional discussion on this issue). However the SHAKE response at longer periods
achieves a better match with the time-domain programs, except for WP2, where the response is
significantly overestimated. The responses calculated from the time-domain programs have
greater differences than those observed for the ordinary ground motions. The LS-DYNA 10pt
response to WP2 ground motion, which results in the largest strains, shows very high
amplification in the acceleration response for periods less than 0.3 sec, compared to the other
time-domain programs. This is also evident from the acceleration response history in Figure
A-50, which shows unrealistic spikes in the LS-DYNA 10pt response. This is resolved in LS
DYNA 120pt, which results in smoother and more realistic acceleration response histories as well
as smaller short period amplifications. This suggests that increasing the number of points in the
backbone curve circumvents the high-frequency issues associated with LS-DYNA 10pt. The
acceleration response spectra calculated by the time-domain programs are considerably different
in the long period range for ground motion WP2, which is due to the differences in hysteresis
rules followed in DEEPSOIL and LS-DYNA. The peak acceleration profiles differ significantly
for ground motions WP1 and WP2 (see Figure A-43 and Figure A-44), but have match better for
the less intense ground motion, WP3 (see Figure A-45). The peak accelerations calculated using
SHAKE match closely with those using DEEPSOIL and LS-DYNA 120pt for ground motions
WP1 and WP3. For WP2, SHAKE underestimates the accelerations at the lower layers, and
overestimates the accelerations in the upper layers. LS-DYNA 10pt results in larger peak
accelerations, especially in the middle and lower layers for ground motions WP1 and WP2, due to
the high-frequency response, which is evident from the acceleration response spectra presented in
middle panel of Figure A-47. The peak accelerations calculated by DEEPSOIL and LS-DYNA
120pt match reasonably well for all three ground motions, except for the slight overestimation by
DEEPSOIL for ground motion WP3, in layers below 60m .
c. Site amplification: The amplification spectra in Figure A-52 show that the elongated site period
lies between 0.6 and 0.8 sec for WP1, between 1.5 and 2.5 sec for WP2, and between 0.5 sec and
0.6 sec for WP3, which are significantly greater than the elongated site periods calculated for the
ordinary ground motions. The peak amplification is greatest for the weakest ground motion, WP3,
and smallest for the strongest ground motion, WP2. These observations indicate that the intensity
of the ground motion increases the elongated site period, and also the hysteretic damping, which
leads to smaller peak amplifications. The amplification spectra calculated using SHAKE are
considerably different from those calculated from the time-domain programs. SHAKE
overestimates the amplifications at the site period for all ground motions and underestimates the
elongated site period for WP1 and WP2, which are the consequences of a significant
72
underestimation of peak strains. Although amplifications calculated using the time-domain
programs are comparatively closer to each other, there are differences, especially for periods less
than 0.3 sec.
d. Variation of outcrop motion through depth: The acceleration response spectra calculated at the
surface differ significantly from the base outcrop for ground motions WP1 and WP2, and the
differences are significantly greater than those observed for ground motion WP3 and the ordinary
ground motions. This indicates that for the stronger ground motions the outcrop motion in soft
rock varies significantly through the depth. This further suggests that for stronger ground motions
that induce larger strains, the rock outcrop motion at the surface needs to be deconvolved to
calculate the outcrop motion at a depth, before using it as an input for a nearby soil site. Further
insight into this observation is presented in Section 4.7.5.
4.6.3.3 Effect of pressure-dependence of soil properties
Incorporating pressure-dependence in the soil properties changes the response of site W1 to both
ordinary as well as pulse ground motions. Since the shear strength of the soil increases when pressure-
dependence is introduced, the peak strain in the soil layers decreases. This decrease is more prominent for
the pulse motions (e.g., Figure A-43) than ordinary ground motions (e.g., Figure A-30). However the
acceleration response spectra show larger surface spectral accelerations at the site period for all the
ground motions. Consequently the spectral amplifications (Figure A-39 and Figure A-52) are
considerably larger when pressure-dependent properties are used for the soil. The peak accelerations are
also slightly larger when pressure-dependent properties are used, especially for the layers closer to the
surface. Incorporating pressure-dependence also decreases the differences between the responses
calculated from the different numerical programs. This is a consequence of the significant decrease in the
peak strains when pressure-dependence is included.
4.6.4 Site W2
4.6.4.1 Response to ordinary ground motions
Site W2 is uniform site composed of sand. Figure A-53 to Figure A-65 present the site-response
results of site W2 subjected to ground motions WO1, WO2 and WO3 (peak accelerations of 0.15g, 0.17g
and 0.19g). Figure A-53 to Figure A-55 present the stress-strain loops for the three ground motions, and
are followed by the peak acceleration and peak shear strain profiles (Figure A-56 to Figure A-58),
acceleration response spectra (Figure A-59 to Figure A-61), acceleration response histories (Figure A-62
to Figure A-64), and the spectral amplifications (Figure A-65). Each of these figures includes the
73
responses calculated using SHAKE, DEEPSOIL, LS-DYNA 10pt and LS-DYNA 120pt. Only the results
of the analyses performed using pressure-independent (PI) properties are discussed in this section. Results
of the analyses performed using pressure-dependent (PD) properties are discussed in Section 4.6.4.3.
Figure A-57 and Figure A-58 show that the largest peak strains calculated by different numerical
programs vary between 0.12% and 0.18% for WO1, 0.06% to 0.09% for WO2, and 0.03% to
0.04% for WO3. These strain ranges correspond to modulus reduction ratios that lie between 0.49
and 0.57 for WO1, 0.24 and 0.29 for WO2, and 0.36 and 0.41 for WO3. The equivalent viscous
damping ratios corresponding to these ranges are 23%, 19% and 15% for WO1, WO2 and WO3,
respectively. These strain levels and modulus reduction ratios indicate significant softening in the
soil layers due to nonlinear behavior. Nonlinear response is also evident from the stress-strain
loops presented in Figure A-53, Figure A-54 and Figure A-55. The stress-strain hysteresis of LS-
DYNA 10pt and LS-DYNA 120pt are quite similar, but differ considerably from that of
DEEPSOIL. The peak strain profiles calculated using LS-DYNA 10pt and LS-DYNA 120pt are
also very similar. SHAKE overestimates the strains in the upper layers and underestimates the
strains in the lower layers compared to the time-domain programs. Analysis using DEEPSOIL
again results in large changes in the peak strains in consecutive layers. However the peak strains
in most layers are only slightly different from those calculated using LS-DYNA. The large
changes in the peak strains are investigated in more detail in Section 4.7.4.2.
b. Acceleration response: The acceleration response spectra presented in Figure A-59, Figure A-60
and Figure A-61 show considerable differences between the responses calculated by SHAKE,
DEEPSOIL, LS-DYNA 10pt and LS-DYNA 120pt for periods less than 0.3 sec. The acceleration
response spectra calculated using SHAKE have near constant amplitudes for periods less than
about 0.1 sec. This is similar to the observations made from the responses of site W1, and
indicates the absence of high-frequency content in the response – also evident from the relatively
smooth acceleration response histories computed by SHAKE (see Figure A-62, Figure A-63 and
Figure A-64 for acceleration response histories; see bullet 2 in Section 4.7.2 for additional
discussion on this issue). Of the time-domain programs, LS-DYNA 10pt calculates the largest
responses in the short period range, followed by DEEPSOIL and LS-DYNA 120pt. Similar to the
response of site W1 to ground motion WP2, analysis using LS-DYNA 10pt results in significant
overestimations of the short period accelerations for ground motion WO2, and unrealistic spikes
in the corresponding acceleration response histories (seen in Figure A-63). The acceleration
response spectra calculated using the time-domain programs match well in the long period range,
and slightly differ from those calculated using SHAKE. The differences between the response
74
spectra calculated using DEEPSOIL and LS-DYNA 120pt can be attributed to the difference in
the corresponding hysteretic models. Figure A-56, Figure A-57 and Figure A-58 show significant
differences between the peak acceleration profiles calculated using the different programs.
Analyses using DEEPSOIL and LS-DYNA 10pt result in larger peak accelerations than LS-
DYNA 120pt and SHAKE. Large overestimation of the peak accelerations by LS-DYNA 10pt is
a consequence of the high-frequency noise in the response.
c. Site amplification: Site W2 has first two translational modes at 1.3 sec and 0.4 sec. The
acceleration amplification spectra in Figure A-65 show that the elongated site periods are about 3
sec for WO1, 2.5 sec for WO2 and 2 sec for WO3. Additionally the surface response is de-
amplified for periods less than 1 sec for all the ordinary WUS motions. Differences between the
amplifications calculated by different programs are considerable, especially for short periods.
Large differences are also seen for periods greater than 4 sec for ground motion WO1. SHAKE
overestimates the amplifications near the elongated site period, especially for WO2 and WO3,
which is a consequence of the underestimation of strains.
d. Variation of outcrop motion through depth: The acceleration response spectra at the surface (see
top panels in Figure A-59, Figure A-60 and Figure A-61) are significantly different from the base
outcrop for all ground motions and the differences are larger than those observed for site W1.
However, unlike for site W1, the acceleration response spectra at the base approach the base
outcrop, indicating that the presence of the sand layer atop the bedrock does not significantly
affect the ground motion at the sand bedrock interface. This is consistent with results of site E2,
which show that the acceleration at the interface of sand and hard rock is similar to the outcrop
input ground motion. Further insight into this observation and its significance is provided in
Section 4.7.5.
4.6.4.2 Response to pulse-like ground motions
Figure A-66 to Figure A-78 present the site-response results of site W2 subjected to ground
motions WP1, WP2 and WP3. These ground motions have peak accelerations of 0.73g, 0.83g and 0.19g,
and pulse periods of 5.1 sec, 1.5 sec, and 9.0 sec, respectively (see Table 4-1). Figure A-66 to Figure
A-68 present the stress-strain loops for the three ground motions, and are followed by the peak
acceleration and peak shear strain profiles (Figure A-69 to Figure A-71), acceleration response spectra
(Figure A-72 to Figure A-74), acceleration response histories (Figure A-75 to Figure A-77), and the
spectral amplifications (Figure A-78). Each of these figures includes the responses calculated using
SHAKE, DEEPSOIL, LS-DYNA 10pt and LS-DYNA 120pt. Only the results of the analyses performed
75
using pressure-independent properties are discussed in this section. Results of the analyses performed
using pressure-dependent properties are discussed in Section 4.6.4.3.
Figure A-70 and Figure A-71 show that the largest peak strains calculated by different numerical
programs vary between 1.5% and 4% for WP1, 2.5% and 5% for WP2, and 1.2% to 3.5% for
WO3. These strain ranges correspond to modulus reduction ratios that lie between 0.02 and 0.04
for WP1, 0.02 and 0.03 for WP2, and 0.02 and 0.05 for WP3. The equivalent viscous damping
ratios for these ranges lie between 30% and 35% for all the ground motions. Significant
differences are observed in the peak strain profiles calculated by different programs at these strain
levels, except for LS-DYNA 10pt and LS-DYNA 120pt, which produce similar stress-strain loops
(see Figure A-66 to Figure A-68) and peak strain profiles. As seen in the responses of site W2 for
ordinary ground motions, analysis using DEEPSOIL results in large changes in the peak strains in
consecutive layers and these changes are significantly greater than those observed for the
ordinary ground motions. Apart from the layers involved in these changes, the DEEPSOIL peak
strains in most layers match fairly well with LS-DYNA 10pt and LS-DYNA 120pt. Similar to the
observations made for ordinary ground motions, SHAKE results in a significant underestimation
of the peak strains as compared to the time-domain programs, especially for ground motion WP1
and WP3, which leads to the underestimation of elongated site period as well as the
overestimation of corresponding spectral amplifications.
b. Acceleration response: The acceleration response spectra in Figure A-72, Figure A-73 and Figure
A-74 are significantly different for the three programs, and the differences are greater than those
observed for the ordinary ground motions. Also, as observed for ordinary ground motions,
SHAKE predicts a near constant spectral acceleration at shorter periods (less than 0.2 sec)
indicating the absence of high-frequency components in the response (see bullet 2 in Section
4.7.2 for additional discussion on this issue). Additionally the spectral accelerations calculated
using SHAKE are significantly overestimated as compared to the time-domain programs in the
long period range, which is a consequence of underestimating the peak strains. Analyses using
LS-DYNA 10pt result in very large short period accelerations, which is also evident from the
large unrealistic spikes in the corresponding acceleration response histories (see Figure A-75,
Figure A-76 and Figure A-77). As observed from the results for ordinary ground motions, LS-
DYNA 120pt circumvents this issue and predicts smaller spectral accelerations in the short period
range, and also more realistic acceleration response histories. As a result LS-DYNA 120pt
predicts spectral accelerations that are closer to LS-DYNA 10pt at longer periods and to
DEEPSOIL at shorter periods. Significant differences are also observed in the peak acceleration
76
profiles calculated from SHAKE and the time-domain programs as seen from Figure A-69 and
Figure A-71. Amongst the time-domain programs, the peak accelerations are the greatest for LS-
DYNA 10pt, which is primarily due to the short period (high-frequency) response observed in the
corresponding acceleration response spectra. The peak accelerations obtained using LS-DYNA
120pt and DEEPSOIL match reasonably well for all the ground motions.
c. Site amplification: The acceleration amplification spectra presented in Figure A-78 show very
large elongations in the site period compared to the ordinary ground motions. Although
significant differences exist between their amplification spectra, the time-domain programs
predict an elongated site period of greater than 6 sec for all the pulse motions. This indicates
highly nonlinear behavior, given that the linear elastic time period of this site is 1.3 sec. Very
large differences exist between the amplification spectra calculated using SHAKE and the time-
domain programs, with SHAKE underestimating the elongated site period and overestimating the
amplifications at this site period. This is a consequence of the significant underestimation of the
shear strain, which is a result of using a lower value for the effective shear strain ratio (0.22) than
the traditional value of 0.65. Additionally the spectral amplification at the elongated site period is
the smallest for the strongest ground motion, WP2.
d. Effect of pulse period: A notable observation to be made is that the ground motion WP3 results in
peak strains that are in the same range as WP1 and WP2, in spite of having substantially smaller
peak accelerations (see left panel of Figure A-71). This does not happen for site W1, for which
ground motion WP3 results in significantly smaller strains than WP1 and WP2. This suggests that
subjecting site W2 to a ground motion with a pulse period close to the elongated site period
results in very large strains in the soil profile, compared to an ordinary ground motion with
similar peak ground acceleration.
e. Variation of outcrop motion through depth: The acceleration response spectra in Figure A-72,
Figure A-73 and Figure A-74 show that the base outcrop is significantly different from the
surface level acceleration response spectrum for all ground motions, but is similar to the within
motion calculated at the base. This is consistent with the observations made for the ordinary
ground motions, and suggests that the presence of the sand layer does not significantly affect the
ground motion at the sand-bedrock interface. Further insight into this observation and its
significance is provided in Section 4.7.5.
4.6.4.3 Effect of pressure-dependence of soil properties
Similar to site W1, the response of site W2 changes considerably when pressure-dependence is
incorporated in the soil properties. The peak strains in the lower layers of the site are lower while the peak
77
strains in the mid and upper layers are higher as compared with the pressure-independent (PI) case (e.g.
Figure A-56, Figure A-57). The increase in the peak strain in the upper layers is due to the weakening of
the soil layers above the reference depth as compared to the pressure-independent properties (see Section
4.4.1). However the trends in the acceleration response are similar to those seen for site W1: using
pressure-dependent properties results in larger spectral accelerations (and consequently higher spectral
amplifications) at the surface compared with the pressure-independent case (e.g., Figure A-59 and Figure
A-73). Additionally the variability in the site response is also generally smaller when pressure-dependent
properties are used. This is a consequence of the significantly smaller strains in the lower layers of the
soil profile.
4.7 Summary and conclusions
Site-response analysis is performed for four different soil profiles using industry-standard
numerical programs, SHAKE, DEEPSOIL and LS-DYNA. Results of site-response analyses from all the
programs are presented in Appendix A and the observations gathered from all analyses are presented in
Section 4.6. These observations are summarized in this section, and are used to explain some of the
practical issues faced in site-response analysis. Section 4.7.1 presents conclusions regarding nonlinear site
response in general, and Sections 4.7.2 and 4.7.3 present conclusions derived from the comparisons
between equivalent-linear and nonlinear site responses, and nonlinear site responses calculated from
different time-domain programs, respectively. Section 4.7.4 presents the key issues faced while
performing site-response analysis in different programs, along with their implications in practice and
some possible solutions. The observations made in Section 4.6 are used to make recommendations for
calculation and specification of input for one-dimensional site-response analysis. These recommendations
are presented in Section 4.7.5, followed by a description of some prospects for future research in Section
4.9.
4.7.1 General conclusions regarding nonlinear site response
The following conclusions can be made regarding site-response analysis using the soil profiles
and ground motions considered in this study:
1. As the intensity of the ground motion increases, the elongation in the site period (period of
maximum spectral amplification) increases, and the amplification at this period decreases. This is
due to the nonlinear behavior of the soil, which involves reduction of the shear modulus (which
increases the elongation) and increase in hysteretic damping (which decreases the amplification).
78
The spectral amplification here refers to the amplification of the acceleration from the within
motion at the base (and not the outcrop input) to the surface of the soil profile at each period.
2. Pulse motions with long pulse periods can induce very large strains in deep, soft soil profiles
compared to ordinary ground motions of similar peak ground acceleration. This is observed when
the pulse period of the input is close to the elongated period of the soil profile, such as site W2
subjected to ground motion WP3, which resulted in strains similar to those calculated for
significantly more intense pulse motions, WP1 and WP2.
4.7.2 Comparison of equivalent-linear and nonlinear responses
The difference in responses calculated using SHAKE and the time-domain programs, DEEPSOIL
and LS-DYNA, varies by the soil profile and ground motion. The WUS soil profiles, site W1 and site
W2, show larger differences than the CEUS profiles site E1 and site E2. For the CEUS profiles, the
responses are the closest in the hard rock layers of site E1 and site E2 since they are subject to very small
strains, but are considerably different in the sand layers of site E2. For the WUS soil profiles the
responses are closer for site W1 than site W2, and for ordinary ground motions than pulse-like ground
motions. The following conclusions can be drawn from the observations made in Section 4.6:
1. The difference between the equivalent-linear and nonlinear response increases with strain in the
soil layers and thus with the nonlinearity induced in the soil profile.
2. The elongated site periods calculated by SHAKE and the nonlinear analyses match fairly well in
most cases.
3. The surface spectral accelerations calculated using SHAKE and the nonlinear programs are
increasingly different at shorter periods. Additionally the short-period spectral accelerations
calculated using SHAKE are almost constant, especially for sites W1 and W2, which undergo
larger strains. This observation is further explained in Section 4.7.4.1.
4. SHAKE underestimates the peak shear strains as compared with the nonlinear programs for
intense ground motions, especially in the lower layers of the soil profile where the strains are the
largest. This is because the equivalent-linear method accounts for nonlinear behavior in an
‘averaged’ sense by decreasing the shear modulus based on the effective shear strain, which can
be seen as a measure of the extent of nonlinearity induced by the earthquake. In contrast,
nonlinear analysis accounts explicitly for softening and load-history effects.
5. SHAKE overestimates the acceleration amplifications at the site period compared with the
nonlinear programs, especially for intense ground motions. This is a result of underestimating the
damping ratio in the soil layers, which stems from the underestimation of the peak shear strains.
79
6. Results of the analyses with pressure-dependent properties for sites W1 and W2 show similar
trends to those with pressure-independent properties. However, the differences between the
equivalent-linear and nonlinear responses for the pressure-dependent analyses are smaller,
because the soil layers sustain much lower strains.
4.7.3 Comparison between the responses from DEEPSOIL and LS-DYNA
Despite using very similar backbone curves and damping curves to model the soil materials in
DEEPSOIL and LS-DYNA, the site responses calculated using these programs can be significantly
different for analyses involving large strains. The following conclusions can be made from the
observations made in Section 4.6:
1. The peak strain profiles as well as the acceleration response spectra match well for site E1.
However, differences in responses increase for softer soil profiles, sites W1 and W2, and also in
the sand layer of site E2. This indicates that the responses calculated using the nonlinear
programs are increasingly different for larger strains induced in the soil layers.
2. The peak strain profiles calculated from LS-DYNA 10pt and LS-DYNA 120pt are similar for
most cases indicating that the increase in the number of points in the backbone curve causes
negligible changes in the hysteretic behavior. However, significant differences arise between the
peak strains calculated using DEEPSOIL and LS-DYNA, which is a result of using different
hysteresis rules in the two programs.
3. Analyses using DEEPSOIL sometimes predict abrupt changes in the peak strains of consecutive
layers. This observation is further examined in Section 4.7.4.2.
4. The differences between the acceleration response spectra at the surface are more prominent at
shorter periods, especially for sites W1 and W2. LS-DYNA 10pt sometimes overestimates the
short period response compared with DEEPSOIL and LS-DYNA 120pt.
5. The acceleration response spectra calculated using DEEPSOIL and LS-DYNA 120pt are
reasonably close for most cases. In some cases DEEPSOIL overestimates the short period
response compared with LS-DYNA 120pt. This is attributed to the abrupt changes in the peak
strains of consecutive layers.
6. The differences between the calculated nonlinear responses are smaller when pressure-
dependence is introduced, because smaller strains are induced in the layers.
80
4.7.4 Key issues faced during site-response analysis
4.7.4.1 SHAKE
Analysis using SHAKE results in near constant spectral accelerations in the short period range for
sites W1 and W2. This observation is also reported in the study by Stewart and Kwok (2008). The
spectral acceleration for site W1 (PI) subjected to ground motion WP2 is effectively constant at periods
less (frequencies greater) than 0.15 sec (6.7 Hz). Similarly the spectral accelerations for site W2 (PI)
subjected to ground motion WO2 remain constant at periods less than 0.1 sec. A comparison of the
Fourier transforms for these cases (Figure 4-7) indicates that these short-period spectral accelerations
correspond to very small amplitudes in the frequency domain, indicating that the equivalent-linear
procedure is unable to reproduce the higher frequency response.
Figure 4-7: Fourier transforms of the surface accelerations for site W1 subjected to ground motion
WP2 (left) and site W2 subjected to ground motion WO2 (right)
This is because the equivalent-linear procedure cannot capture some of the higher modes of
vibration of the soil profile that are excited due to the continuous changes in the soil material properties
through time. Nonlinear procedures can capture these higher modes of vibration, since they simulate the
changes in material properties in through time. However, the nonlinear procedures are also susceptible to
numerical noise based on the material model and the integration algorithms used (Chen and Joyner, 1974;
Joyner and Chen, 1975). It should also be noted that the equivalent-linear method does not consistently
overestimate or underestimate the short period (high frequency) response as compared with the nonlinear
methods, although the corresponding Fourier amplitude is always underestimated. This can be seen in
Figure A-60 and Figure A-46, where SHAKE overestimates the short period response as compared with
81
DEEPSOIL and LS-DYNA 120pt (LS-DYNA 10pt results in relatively higher responses due to numerical
noise at these periods) for site W2 subjected to ground motion WO2, while it underestimates the response
for site W1 subjected to ground motion WP1.
Analysis using SHAKE does not converge for site W2 subjected to pulse motions in spite of
using the numerical models and analysis parameters that are identical to those used for the ordinary
ground motions. The nonlinear procedures estimate peak strains close to 4% for the W2 site subjected to
pulse motions, which substantially larger than the 1% peak shear strain that is universally accepted as
SHAKE’s limit. Modifications to the SHAKE analysis, namely, extrapolation of the modulus reduction
curves to a peak strain greater than 10% and decreasing the layer thickness, do not enable convergence.
Notably, analysis for the same soil profile and ground motions converges when the equivalent-linear
option of the program DEEPSOIL (referred to as DEEPSOIL-EL hereafter) is used. The SHAKE analysis
converges when the effective shear strain ratio is reduced to 0.22 from a traditionally recommended value
of 0.65, but the resultant responses are very different from the corresponding DEEPSOIL-EL responses.
Specifically, when compared with the nonlinear analyses (Figure 4-8), the SHAKE analysis with an
effective shear strain ratio of 0.22 significantly underestimates the peak strains and overestimates the
spectral acceleration at the site period. The SHAKE analysis also converges for an effective shear strain
ratio at 0.65 if the maximum analysis frequency is reduced to 45Hz from the value of 100Hz used for
ordinary ground motions 5 . The resultant responses after this modification match exactly with the
corresponding DEEPSOIL-EL results [note that the DEEPSOIL-EL and SHAKE (   0.65 ) results
overlap in Figure 4-8], and show smaller differences from the nonlinear results.
5
Issues with convergence are not encountered with a different version of SHAKE (Baig and Bolourchi,
2013), suggesting that the problem is intrinsic to the source code.
82
Figure 4-8. Peak acceleration and peak strain profiles (left) and acceleration response spectra at the
surface (right) for site W2 subjected to ground motion WP1
4.7.4.2 DEEPSOIL
Analysis using DEEPSOIL occasionally resulted in abrupt changes in the peak strains of
consecutive layers, despite the layers having the same material properties. These changes were associated
with the peak strains in the soil profile greater than 0.1%. SHAKE and LS-DYNA resulted in relatively
smooth peak strain profiles for the corresponding analyses. An example of this behavior is site W2
subjected to ground motion WP2. The DEEPSOIL peak strain profile for this case is presented in Figure
A-70, which shows many “spikes” through the depth of the soil profile. The largest spike in this peak
strain profile is formed by the strains in three consecutive layers at depths 94.5m, 95.5m and 96.5m,
where the peak strains are 2.3%, 7% and 2.6%, respectively. A comparison of the hysteresis loops
calculated by DEEPSOIL for these layers is presented in Figure 4-9 below. Arrows in this figure indicate
the path of hysteresis. The large cycle in this figure corresponds to the pulse of the ground motion, and
the smaller cycles correspond to the rest of the ground motion.
83
Figure 4-9: Hysteresis loops of three layers in site W2 showing an abrupt change in the peak strains
when subjected to ground motion WP2
In this figure, the hysteresis loops of the three layers match closely in the smaller cycles in the
initial part of the earthquake, while the second layer at 95.5m diverges significantly from the other two
adjacent layers in the large cycle. This divergence results in a large difference between the peak strains of
the consecutive layers and thus a spike in the peak strain profile. The large cycle in Figure 4-9 is isolated
and more clearly reproduced in Figure 4-10, with panel a and panel b representing the hysteresis loops of
the three layers from 0 to 2.05 sec, and 0 to 2.3 sec respectively. Panel a shows that the initial parts of the
hysteresis paths (before 2.05 sec) are close to each other, with the path of the layer at 95.5m depth
approaching the backbone curve (dotted line). This hysteresis path then follows the backbone curve
according to rule 3 of the extended Masing rules (see Section 2.1.2.2 for a description of the extended
Masing rules). However, the other two layers miss the backbone curve and then trace a completely
different path resulting in large differences in the final peak strains. This indicates that small changes in
the initial paths of the hysteresis loops might lead to very large differences in the final peak strains, which
is unrealistic. This behavior is not seen in LS-DYNA since the MAT_HYSTERETIC_SOIL model is not
governed by the extended Masing rules. Although DEEPSOIL results in unrealistic peak strain profiles,
the corresponding acceleration responses are reasonably close to those calculated from LS-DYNA 120pt
indicating that this behavior does not affect the acceleration response. However caution has to be
exercised for applications where the peak strain is important, such as predicting permanent displacements
in the soil profile.
84
a. Hysteresis loop from 0 to 2.05 sec b. Hysteresis loop from 0 to 2.3 sec
Figure 4-10: Examination of the hysteresis loops of layers involved in abrupt changes in the peak
strains in site W2 subjected to ground motion WP2
4.7.4.3 LS-DYNA
Significant high-frequency content was seen in the LS-DYNA response when the 10 point
backbone curve was used. Examination of the corresponding acceleration response histories (e.g. Figure
A-76) indicated that these components were unrealistic and were a consequence of the 10-point model.
This hypothesis is substantiated by the significant reduction in the high-frequency content when the 120-
point backbone curve was used (compare acceleration response histories in Figure A-76). This indicates
that the high-frequency noise in the LS-DYNA 10pt response is primarily due to the significant changes
in stiffness in its multilinear backbone curve. Increasing the number of points in the backbone curve
significantly decreases the stiffness increments and therefore results in smoother accelerations. An
alternative approach (Huang, 2013) to achieve a smoother acceleration response is to incorporate strain-
rate effects in LS-DYNA 10pt. Figure 4-11 also presents the results when strain-rate effects6 are included
in the model. This observation could play a critical role in estimating the seismic demands for stiff
structures such as nuclear power plants since higher frequencies are most important in these structures.
6
Strain-rate effects are included in the MAT_HYSTERETIC_SOIL model by defining the curve, LCSR,
which specifies the scale factor for the yield strength as a function of strain-rate (LSTC, 2013)
85
Figure 4-11: Acceleration time series at 50m depth (left) and the peak acceleration profile (right)
for site W1 subjected to ground motion WP2
LS-DYNA 120pt results in numerical instabilities for analyses involving site W2 subjected to
pulse motions, when the transmitting boundary is employed. However these numerical instabilities are not
encountered when a reflecting boundary is employed for the same cases. Using a transmitting boundary
involves the application of an input force (a ‘natural’ boundary condition 7) at the bedrock, while the
reflecting boundary involves the application of an acceleration input (an ‘essential’ boundary condition 7).
Instabilities with the transmitting boundary condition arise because finite element methods directly
comply with the essential boundary conditions but only indirectly comply with the natural boundary
conditions. It is theorized that the errors due to the natural boundary condition at the base, where the
strains are very large for the pulse motions, lead to numerical instabilities. Further investigation is
required to fully understand this problem.
4.7.4.4 Sensitivity of high-frequency response
The site responses calculated from various numerical programs were most different in the short
period (high frequency) range, which is critical for stiff structures such as nuclear power plants. Analyses
using SHAKE resulted in a near constant response, while the responses calculated using the time-domain
programs showed considerable variability, in the short periods. Noticeable differences in the short period
spectral accelerations were observed even in the analyses for site E1 (see Figure A-8), in which the soil
7
In the terminology used in finite element methods, the boundary conditions involving displacements,
accelerations, etc., are termed ‘essential’ boundary conditions, and the boundary conditions involving tractions,
forces, etc., are termed ‘natural’ boundary conditions (Bathe, 1996)
86
layers remained almost linear. However the range of periods at which these variations occur was different
for each site: for example, the spectral accelerations calculated by the different programs at the surface of
site E1 (with a first mode period of 0.16 sec) when subjected to ground motion EO2 deviated from each
other at periods less than 0.04 sec (see Figure A-8), whereas for site W1 (first mode period of 0.40 sec)
subjected to ground motion WO2 the deviations occurred at periods less than 0.20 sec. The site-
dependence of these period ranges indicates that the variability in the short period responses might be due
to a poor match in the higher mode responses calculated from the different programs. Kwok et al. (2008)
observed the variability in the short period responses calculated by different time-domain programs and
attributed it to differences in the corresponding viscous damping formulations. However the DEEPSOIL
and LS-DYNA analyses performed in this study consider almost similar viscous damping formulations
(frequency-independent damping; damping ratio of 1.1%). While viscous damping could be an issue, the
variability also arises from the differences in the hysteretic material models (such as differences in the
hysteresis rules). Furthermore these observations indicate that the high frequency response is more
sensitive to the changes in the viscous damping and hysteretic material models than the response at lower
frequencies, which are typically focused on in traditional civil engineering structures. Validation efforts
therefore require much more attention to the high frequency response in order to improve the credibility
of site-response analysis programs for nuclear applications.
4.7.5 Calculation and specification of input for site-response analysis
Earthquake ground motions are most often recorded at the surface of an outcropping rock (rock
outcrop motion), but not at depth (within profile motion). However the concept of ‘outcrop input’ allows
the usage of a nearby recorded rock outcrop ground motion as the input for site-response analysis (see
Section 2.2.2) when a transmitting boundary is assumed for the base of the soil profile. Consider the
illustration in Figure 4-12 with a hypothetical soil profile (right) and outcropping bedrock (left). In this
figure, VB is the low-strain shear wave velocity of the bedrock and VA is the low-strain shear wave velocity
of the material in the soil profile. Let ORS be the ground motion recorded at the surface of the outcropping
rock, ORB and WRB be the outcrop and within motions at the base of the soil profile (calculated using a
deconvolution analysis), and OSS be the surface response of the soil profile. Assume that the soil profile
and bedrock are horizontally infinite and do not interact with each other.
Most often in practice, site-response analysis is performed by using the ground motion ORS to
calculate the outcrop input forces at the base, and not ORB , which is the outcrop ground motion at the base.
The analyses of site E1 showed that the response at the surface of the hard rock profile is very close to the
87
outcrop motion at the base, indicating that the outcrop motion does not change significantly through the
depth for very stiff sites. Therefore it can be said that the recorded ground motion, ORS , is close to the base
outcrop, ORB , and can be used as an outcrop input at the base of the soil profile, avoiding the need for a
deconvolution analysis. However this is not true for the analysis of site W1, which involves considerably
larger strains. The outcrop ground motion in this case differs through the depth soil profile, which is
evident from the surface response of site W1 differing considerably from the input outcrop even for
ordinary ground motions. In these cases the base outcrop ground motion ORB should be calculated by a
deconvolution analysis, and this ground motion can be specified as outcrop input to the soil profile under
consideration.
From the results of site E2 and site W2 it can be seen that the presence of a very soft soil layer on
the top of a relatively hard rock does not influence the ground motion at the soil-rock interface
significantly. Namely, if VA is significantly smaller than VB , the outcrop ground motion at the base, ORB , is
almost equal to the within motion, WRB , indicating that the bedrock behaves like a rigid base to the soil
profile. Therefore the rigid base assumption can be used in these cases and the calculated base outcrop
motion, ORB , can be input to the soil profile as a within motion, which is conveniently provided as direct
acceleration input (as opposed to the outcrop input, which is provided as a series of shear forces).
Figure 4-12: Outcrop and within ground motions
The following statements can be made regarding the calculation of input from a recorded rock
outcrop for the site-response analysis of a nearby soil profile:
88
1. If a ground motion is recorded on a very stiff outcropping rock (such as site E1 with a low-strain
shear wave velocity of 2500 m/s) where mostly linear strains are expected, the recorded ground
motion can be directly used as an input at the base of a nearby soil site.
2. If some nonlinearity is expected in the response of the outcropping rock, the recorded
acceleration should be deconvolved to calculate the ground motion at the base level of the nearby
soil site. This ground motion should be used as the input for site-response analysis.
The following statements can be made regarding the specification of input and the boundary
condition at the base for the site-response analysis for a soil profile, when the ground motion at the base
of the soil profile is available:
1. For soil sites that are very soft compared to the underlying bedrock, the bedrock can be assumed
as rigid and ground motion calculated in bullets 1 or 2 can be directly input as a within motion,
namely, in the form of accelerations (rigid boundary assumption).
2. If the stiffness of the soil site is comparable to the underlying bedrock, the ground motion
calculated in bullets 1 or 2 should be specified as an outcrop input at the base of the soil site
(transmitting boundary assumption).
4.8 Site-response analysis for biaxial horizontal shaking
One-dimensional site-response analysis is traditionally performed independently for the two

horizontal components of the ground motion, with the assumption that site response in orthogonal
directions is uncorrelated. However, since soil is a nonlinear material, the shear stresses in the principal
directions are generally correlated (Hadjian, 1981). The MAT_HYSTERETIC_SOIL material model
accounts for this correlation by defining the failure surface of each of the nested surface layers by the
equation
4 2
J 2i   max (4-3)
3
i
where, J 2i is the second invariant of the stress tensor in layer i , and  maxi is its yield stress. The material
models in SHAKE and DEEPSOIL only account for one-dimensional shear behavior and therefore cannot
simulate biaxial site response. To investigate the interaction between the horizontal components of the
site response, biaxial analyses are performed for sites W1 and W2 using LS-DYNA 120pt. These sites are
subjected to simultaneous excitation in two horizontal directions (X and Y); ground motion WP2 is used
in the X direction and the corresponding perpendicular component (WP2y) is used in the Y direction. The
correlation coefficient for the two horizontal components is 0.32 (Hadjian, 1981). These ground motions
are applied at three different intensities: 50%, 100% and 150% of the recorded motion. Figure 4-13 below
presents the acceleration response spectra calculated from these analyses at the surface of site W1. To
89
enable a comparison between the results of uniaxial and biaxial shaking, the responses are presented for
site W1 subjected to ground motions WP2 and WP2y simultaneously (biaxial), as well as ground motion
WP2 alone (uniaxial). The spectra are calculated for the responses in the X direction, and only the
pressure-independent (PI) cases are considered.
Figure 4-13: Surface acceleration response spectra in the X direction for uniaxial and biaxial
analysis performed for sites W1 and W2
Slight differences are seen between the responses of site W1, with the uniaxial analysis predicting
higher spectral accelerations than the biaxial analysis, especially in the short period range. The differences
between the uniaxial and biaxial results are greater for site W2 and increase with the ground motion
intensity. The response sensitivity to ground motion intensity is expected because each nested surface of
the MAT_HYSTERETIC_SOIL material undergoes yielding earlier in a biaxial analysis (recall that both
of the stress components are included in the stress invariant; see equation 4). Larger strains are induced in
the biaxial analysis, leading to higher hysteretic damping and smaller accelerations. From a practical
perspective, the differences between the uniaxial and biaxial site responses are minor for site W1 (as
compared with those observed between the results from different numerical programs) but are significant
for site W2. Further study using other material models is needed to more completely characterize the
differences between uniaxial and biaxial site response.
90
4.9 Recommendations for future research
Although a number of studies have been performed to develop numerical models for one-
dimensional site-response analysis, significant differences still exist in the responses calculated from the
industry-standard numerical programs especially 1) in the short period range, and 2) for the cases
involving large strains. Robust validation of these programs is therefore needed to increase the credibility
of these programs, especially for soft soil profiles subjected to intense ground motions. These validations
need to be in the form of large-scale laboratory experiments since the one-dimensional wave field is not
produced by a real earthquake. Particular attention needs to be paid to the short-period (high-frequency)
response of the soil profiles since it is critical for calculating the seismic demands in stiff structures such
as nuclear power plants. Additionally, the present material models need to be extended and validated for
very large strains (greater than 1%) in order to accommodate site-response analyses for nuclear
applications, which involve the calculation of seismic demands for ground motions with very long return
periods.
91
SECTION 5
AN ASSESSMENT OF SOIL-STRUCTURE INTERACTION ANALYSIS PROGRAMS
5.1 Overview
This section presents an assessment of the industry-standard soil-structure interaction (SSI)

analysis programs, SASSI (Lysmer et al., 1999) and LS-DYNA (LSTC, 2013). SASSI is a frequency-
domain program for linear analyses, and LS-DYNA is a time-domain program capable of linear and
nonlinear analyses. The assessment involves comparisons of the SSI responses calculated using the two
programs for 1) simple models involving uniform elastic soil profiles and lumped mass structures on a
surface basemat (Section 5.2), and 2) nonlinear models involving the soil profile and experimental model
structures used in the City Block project (Section 5.3). The model structures MS1F_2 and MS2F used in
Tests 3 and 4 are considered for the nonlinear analyses. Nonlinearities in the soil, foundation-soil
interface, and the structure (only MS1F_2) are modeled. The process of performing these comparisons
reveal some practical challenges in performing SSI analysis, and these challenges are identified in Section
5.4.
5.2 Calculation of linear elastic SSI response using SASSI and LS-DYNA
Soil-structure interaction analysis using SASSI and LS-DYNA can be a complicated task prone to
modeling errors. Before assessing these programs for the solution of nonlinear problems, they are
compared and assessed for linear elastic problems. The SASSI analyses are performed using the flexible-
volume method (Ostadan, 2006a) and the LS-DYNA analyses are performed using both the direct method
and the effective seismic input method, which is a predecessor of the domain reduction method (DRM).
These methods are described in Section 3.2.
5.2.1 Model description
A uniform soil profile with a depth of 29.47m (equal to the depth of the soil profile used in the
City Block project) is chosen for the elastic analyses presented in this section. The soil material is
assumed to be linear viscoelastic with a mass density of 1700 kg/m3, shear wave velocity of 164.59 m/s,
Poisson’s ratio of 0.3, and a damping ratio of 0.02. Two lumped mass structures are considered for
analysis: 1) a 10m tall structure (LM1) with one lumped mass at the top, and 2) a 25.6m tall structure
(LM2) with lumped masses at the top, mid-height and the bottom. Both the structures are founded on a
basemat identical to that of the experimental model MS2F (Section 6.4.3.1), designed for the centrifuge
experiments of the City Block project. This basemat has a plan area of 13.8m  13.8m with a thickness of
93
1.4m, and embedded with an embedment depth equal to the mat thickness. Both structures are assumed to
be made of Aluminum, with a mass density of 2700 kg/m3, elastic modulus of 69 GPa, and a Poisson’s
ratio of 0.3. The lumped masses and beam cross section properties of the structures are presented in Table
5-1. The structures are shown in Figure 5-1. LM1 is designed to produce minimal SSI effects, and LM2 is
designed as a lumped-mass representation of MS2F, which shows significant SSI effects.
Table 5-1: Lumped masses and beam cross section properties of the structural models used for
analyses using SASSI and LS-DYNA
Lumped mass Beam cross section properties

Structure
(Metric tons) A (m2)a SA (m2)b Iyy (m4)c
LM1 166 0.213 0.114 1.06
Top 311.35
LM2 Mid-height 518.92 0.213 10.02 70.63
Base 207.57
a
Area of cross section
b
Shear area
c
Moment of inertia for bending in the Y direction (ground motion is along the X direction)
Figure 5-1: Lumped mass structures, LM1 and LM2, used for the verification analyses of SASSI
and LS-DYNA
94
Prior to performing SSI analyses, the fixed-base models of the structures are analyzed using
SASSI and LS-DYNA and results are compared to ensure that the structural models in the two programs
are equivalent. SSI analyses are then performed with the rigid basemat alone and the responses are
compared. Following these preliminary analyses, each of the structures, LM1 and LM2, is analyzed for 1)
rigid superstructure on a rigid basemat, and 2) flexible superstructure on a rigid basemat, and 3) flexible
superstructure on a flexible basemat. The analyses are performed using SASSI and LS-DYNA and the
results are compared in Section 5.2.4. A single ground motion is used for the comparisons made in this
section. This ground motion is an Ormsby wavelet with a peak acceleration of 0.5g, and frequency
parameters, 0Hz, 0.2Hz, 20Hz, and 25Hz (refer to Appendix B for information on the Ormsby wavelet).
The ground motion is applied at the base of the soil profile assuming rigid bedrock beneath the soil.
5.2.2 Modeling in SASSI
General modeling procedures in SASSI are described in Section 3.3.1 and the modeling details
corresponding to the SSI analysis of LM1 and LM2 are presented in this section.
The soil profile in SASSI is modeled with 15 layers, all with the same shear wave velocity
(164.59 m/s), compressive wave velocity (307.91 m/s) and damping ratio (0.02). The analysis is
performed for 44 frequencies, ranging from 0.02Hz to 25Hz. Although the SASSI user manual
recommends only 10 to 20 frequencies for analysis, a larger number of frequencies are used for reasons
described in Section 5.4.1. The basemat and excavated soil elements are modeled using eight-noded brick
elements, and the superstructure is modeled using beam elements with the properties specified in Table
5-1. Lumped masses specified in Table 5-1 are assigned to the superstructure nodes. The basemat is just
embedded below the soil surface, with an embedment depth equal to its thickness. The superstructure is
connected to the basemat with rigid beam elements. Since SASSI does not have a rigid material model,
rigid elements are usually modeled by using a very large elastic modulus or large cross-section properties.
A damping ratio of 0.02 is used for the soil, the basemat, and the superstructure. The control motion is
applied at the base of the soil profile as a vertically propagating shear wave. A reflecting boundary is
modeled at the base since the bedrock is assumed to be rigid. The SASSI models of LM1 and LM2 are
presented in Figure 5-2. The lumped masses in this figure are indicated by red solid circles, and the
ground surface is indicated by a shaded brown area.
95
a. LM1 b. LM2
Figure 5-2: SASSI finite element models of LM1 and LM2
5.2.3 Modeling in LS-DYNA using the direct method and the effective seismic input method
General modeling procedures for the LS-DYNA direct method as well as the effective seismic
input method are explained in Section 3.2 and the details corresponding to the SSI modeling of LM1 and
LM2 are presented in this section.
The basemat and the superstructure in LS-DYNA are modeled using eight-noded brick elements
and two-noded beam elements, respectively. The beams elements are of the Belytschko-Schwer resultant
type (LSTC, 2013), with the input properties presented in Table 5-1. Lumped masses specified in Table
5-1 are assigned to the superstructure beam element nodes. A rigid connection between the superstructure
and the basemat is modeled using the *CONSTRAINED_NODAL_RIGID_BODY option, which creates
a rigid connection between a specified set of nodes. Figure 5-3 presents the finite element models of LM1
and LM2. The contact interface between the soil and the basemat is modeled differently for the rigid
basemat and the flexible basemat cases. When the basemat is flexible, the time-step of the analysis is
controlled by the solid elements of the basemat. This time-step can be much smaller than the time-step
corresponding to the soil elements and can therefore significantly increase the analysis time. In order to
have a larger time-step (and a smaller analysis time) the basemat is meshed to be considerably coarser
96
than the surrounding soil. Therefore, the basemat-soil interface in this case is created using the
*CONTACT_ TIED_SURFACE_TO_SURFACE option available in LS-DYNA. This ties the soil to the
basemat allowing no separation at the basemat-soil interface. For the rigid basemat, the rigid elements are
not included in the time-step calculation, and therefore the basemat is meshed identically to the
surrounding soil to avoid the use of contact elements.
a. LM1 b. LM2
Figure 5-3: LS-DYNA finite element models of LM1 (left) and LM2 (right)
Figure 5-4: LS-DYNA finite element model for SSI analysis of LM2 using the direct method
97
A soil domain of size 219m  118m (219m in the shaking direction, X) in plan and 29.47m deep,
comprising eight-noded solid elements, is built for analysis using the direct method in LS-DYNA. No
transmitting boundary is employed at the boundaries of the soil domain. Instead, the soil domain is made
large enough such that the scattered waves are almost damped before reaching the boundaries. The lateral
boundaries are constrained in such a way that the perimeter nodes at each elevation move together in the
horizontal direction (see Section 3.2.2), creating a ‘free-field’ condition at the boundaries. The adequacy
of the soil domain size is checked by comparing the surface free-field ground motion in a soil-structure
interaction analysis to the free-field surface ground motion calculated using a separate one-dimensional
site-response analysis performed in LS-DYNA. Section 5.2.4.1 presents the results for these analyses and
demonstrates that the soil domain is large enough for linear elastic SSI analyses. The input ground motion
is applied directly at the base of the soil profile as an acceleration input. This method of input is consistent
with the rigid bedrock assumption made for the analyses in this section. Elastic material is modeled using
the *MAT_ELASTIC material model with an elastic modulus of 1.085  108 Pa, which is equivalent8 to
the shear wave velocity and Poisson’s ratio specified in Section 5.2.1. Damping is modeled using the
*DAMPING_FREQUENCY_RANGE_DEFORM option, which provides an approximately frequency-
independent damping in a given frequency range. A damping ratio of 0.02 is specified in the frequency
range of 1 to 25 Hz, which is adequate for the analyses presented in this section.
The effective seismic input method involves an initial site-response analysis that calculates the
input forces for a subsequent SSI analysis. These forces are input at boundary of the near-field soil (refer
to Section 3.3.2.2 for description of the effective seismic input method in LS-DYNA) for an SSI analysis.
The site-response model and the SSI model used in the effective seismic input method are presented in
Figure 5-5. In the SSI model, the basemat is indicated in green, the near-field soil is indicated in light
green, the far-field soil is indicated navy blue, and the PML material at the boundaries is indicated in
orange. The basemat-soil interface is modeled as in the direct method. A damping ratio of 0.02 is also
applied to the soil and the structure as in the direct method.
8
The elastic modulus is slightly decreased to account for the stiffening by the *DAMPING_
FREQUENCY_RANGE_DEFORM.
98
Figure 5-5: LS-DYNA finite element models for the effective seismic input method: site-response
analysis (left) and SSI analysis (right)
In the vertical direction the element sizes in the soil domain of the direct method and the effective
seismic input method are identical. This element size is limited to 1m, which allows the propagation of
frequencies up to 15Hz for the shear wave velocity of 164.59 m/s when a value 10 is used for n in
equation (4-1) (refer to Section 4.4.2 for a description of the site layering procedure). This maximum
frequency of propagation is adequate for the analyses presented in this section.
5.2.4 Analysis results and comparison between SASSI and LS-DYNA
5.2.4.1 Free-field response
The free-field response of the soil domain to the Ormsby wavelet, calculated using SASSI, LS-
DYNA one-dimensional site-response analysis, and the LS-DYNA SSI analysis, is presented in Figure
5-6. The left panel presents 1% damped response spectra, and the right panel presents the Fourier
transforms of the acceleration at the surface of the soil domain. The figure clearly shows that the free-
field responses calculated from the two programs are almost identical. Small differences in the peak
amplitudes exist due to the differences in the damping formulations adopted in the two programs. While
SASSI uses a formulation that models a truly frequency-independent damping ratio, the damping
formulation used in LS-DYNA only models an approximately frequency-independent damping ratio.
Additionally the LS-DYNA formulation results in a small increase the stiffness of the model, which is
99
compensated by a reduction in the elastic stiffness of the materials as recommended in the LS-DYNA
keyword manual (LSTC, 2013). This compensation can also result in the differences in the SASSI and
LS-DYNA responses. A notable feature of these results, however, is the peaks in the response spectra and
the Fourier transforms. These peaks correspond to the natural frequencies of the soil profile, which can be
calculated using the expression,  2n  1VS 4H , where n is the vibration mode number, VS is the shear
wave velocity of the material, and H is the total depth of the soil profile (Kramer, 1996). The peaks in
the Fourier transforms presented in Figure 5-6 show that the vibration modes occur at 1.4Hz, 4.2Hz, 7Hz,
9.8Hz, and so on. These frequencies are identical to those calculated analytically using the above
expression.
Figure 5-6: 1% damped acceleration response spectra (left) and Fourier transforms of the surface
free-field accelerations computed using site-response analysis and the SSI analysis for LM2
5.2.4.2 Fixed-base responses
Figure 5-7 presents 1% damped response spectra of the horizontal acceleration at the roof of the
fixed-base models of LM1 and LM2 calculated using SASSI and LS-DYNA. These response spectra are
almost identical, indicating that the fixed-base models of the structures in SASSI and LS-DYNA are
equivalent. Slight differences in the peak spectral accelerations exist, and are unavoidable given the
difference in the damping formulations used in the programs as explained in Section 5.2.4.1. The figure
shows that LM1 and LM2 have fixed base time periods of 0.23 sec and 0.14 sec, respectively.
100
Figure 5-7: Response spectra for 1% damping of the roof accelerations of the fixed-base models of
LM1 (left) and LM2 (right) calculated using SASSI and LS-DYNA
5.2.4.3 Response of the basemat alone
Figure 5-8 below presents the 1% damped response spectra of the horizontal and vertical
accelerations of the basemat. The figure shows that the horizontal acceleration responses calculated from
the different programs are almost identical, but small differences exist between the vertical acceleration
spectra. The differences between the vertical accelerations can be attributed to the damping formulations
as described in 5.2.4.1. The peaks in the response spectra and the Fourier transforms correspond to the
vibration modes of the soil profile.
Figure 5-8: Response spectra for 1% damping of the horizontal acceleration at center of the
basemat (left) and the vertical acceleration at the edge of the basemat (right) for SSI analysis of the
rigid basemat alone
101
5.2.4.4 Response of LM1 with SSI effects
Figure 5-9 presents 1% damped amplification spectra of the horizontal acceleration at the roof of
LM1 from the surface free field of the soil for three cases: rigid structure on a rigid basemat, flexible
structure on a rigid basemat, and flexible structure on a flexible basemat. For each case, the amplification
is calculated using SASSI, LS-DYNA direct method and LS-DYNA DRM.
Figure 5-9: Acceleration amplification spectra from the surface free field of the soil to the roof of
LM1
The figure shows that the amplification spectra calculated using the three methods are almost
identical for all cases. Notably, the responses calculated using the direct method and DRM in LS-DYNA
overlap, indicating that the methods are equivalent for LM1. A comparison of the responses in the lower
two panels of the figure indicates that including the flexibility of the basemat decreases the peak structural
amplification for LM1. Figure 5-9 also shows that the flexible-base period of LM1 is about 0.6 sec, which
is much greater than the fixed-base period of 0.23 sec, indicating period lengthening due to SSI effects.
102
Figure 5-10 presents a deconstruction of the roof accelerations of LM1 for each analysis case into
1) acceleration relative to the basemat, 2) acceleration due to rocking of the basemat, and 3) acceleration
due to structural deformations. One percent damped response spectra for these components are presented.
Since the direct method and DRM produce identical results, the deconstruction is presented only for
SASSI and the LS-DYNA direct method. The three rows in Figure 5-10 correspond to the three analysis
cases.
Figure 5-10: Deconstruction of the horizontal roof accelerations of LM1 subjected to the Ormsby
wavelet
103
The figure shows that the rocking component clearly dominates the response when the
superstructure is rigid (top row), and the structural deformations are negligible. In contrast, for the
flexible structure cases (middle and bottom rows), the structural deformations dominate the response of
LM1 and the rocking response is negligible, indicating that the SSI effects in LM1 are minimal. A
comparison of the responses in the middle panel and the bottom panel shows that the flexibility of the
basemat has a small, but measurable, effect on the rocking accelerations and structural deformations. It
can be seen that the flexibility of the basemat slightly increases the rocking acceleration and decreases the
accelerations from structural deformations in the first mode. A comparison of the SASSI (left panels) and
LS-DYNA (right panels) results shows minor differences in the peak spectral accelerations in all cases.
These differences can be attributed to the material damping formulations in the programs, as discussed in
Section 5.2.4.1.
5.2.4.5 Response of LM2 with SSI effects
Figure 5-11 presents 1% damped amplification spectra of the horizontal acceleration at the roof of
structure LM2 from the surface free field of the soil for the three analysis cases. For each case, the
acceleration is calculated using SASSI, LS-DYNA direct method and LS-DYNA DRM. The top panel of
the figure shows that spectral amplifications are almost identical for the analysis case with a rigid
structure on a rigid basemat. However the bottom two panels show that the peak amplifications calculated
using SASSI are considerably larger than those calculated using LS-DYNA. These differences are larger
than those seen for LM1. This, again, can be attributed to the damping formulations of the two programs,
and specifically to the fact that the damping ratio in LS-DYNA is not truly frequency-independent. The
responses calculated using the LS-DYNA direct and DRM are almost identical for all cases, which is also
observed in LM1. This indicates that the methods are equivalent for the structures considered in this
section. In contrast with LM1, accounting for the basemat flexibility slightly increases the peak structural
amplifications, indicating that the impact of basemat flexibility on the structural response depends on the
flexible-base period of the structure. Figure 5-11 also shows that the flexible-base period of LM2 is about
0.42 sec, which is much longer than the fixed-base period of 0.14 sec indicating significant period
lengthening due to SSI effects.
Figure 5-12 presents a similar deconstruction of the roof accelerations as in Figure 5-10, but for
the responses of LM2. The three rows in Figure 5-12 correspond to the three analysis cases. Similar to
LM1, the rocking component dominates the response when the structure is rigid, and the structural
accelerations are almost zero. The absence of structural deformations in this case verifies the rigidity of
the structure-basemat connection. Unlike the response of LM1, the rocking component dominates the
104
response of LM2 even when the superstructure is flexible, as seen from the middle and bottom rows of
Figure 5-12. This is due to the much higher stiffness of LM2 in the shaking direction. Additionally a
comparison of the responses in the second and third rows shows that the basemat flexibility has only a
minor effect on the rocking as well as the structural deformation components. A comparison of the SASSI
(left panels) and LS-DYNA (right panels) results shows minor differences in the peak spectral
accelerations in all cases. These differences can be attributed to the material damping formulations in the
programs, as discussed in Section 5.2.4.1.
Figure 5-11: Acceleration amplification spectra from the surface free field of the soil to the roof of
LM2
105
Figure 5-12: Deconstruction of the horizontal roof accelerations of LM2 subjected to the Ormsby
wavelet
5.2.5 Observations
The following observations can be made from the linear elastic analyses presented in this section:
1. The free-field soil responses calculated using SASSI and LS-DYNA are almost identical, with
small differences in the peak spectral accelerations. This is due to the difference in the damping
106
formulations of the two programs, as well as errors in the adjustment of the elastic modulus in
LS-DYNA, which is required for the *DAMPING_FREQUENCY_RANGE_DEFORM option.
2. Analysis of the fixed-base models in SASSI and LS-DYNA result in almost identical responses
indicating that the models are equivalent. Small differences in the responses result from the
difference in the damping formulations.
3. The structural responses calculated from the LS-DYNA direct method and DRM are essentially
identical for both LM1 and LM2 in all cases. The methods are equivalent for elastic SSI analyses.
4. Basemat flexibility has a small but measurable effect on the overall response of LM1 and LM2.
For LM1, accounting for basemat flexibility decreases the peak spectral amplification at the roof,
whereas it increases for LM2. This indicates that the effect of basemat flexibility depends on the
flexible-base time period of the structure.
5. SASSI and LS-DYNA produce very similar responses for both structures for the cases considered
in this section. Small differences in the peak spectral amplifications at the roof are attributed to
differences in the damping formulations in the programs. The programs also produce almost
identical responses for the rocking, and structural deformation components of the response. This
verifies that the SSI analysis programs SASSI and LS-DYNA are equivalent in the linear elastic
range.
6. The analysis times for SASSI, LS-DYNA direct method, and LS-DYNA DRM are summarized in
Table 5-2. SSI analysis in SASSI is faster than both the LS-DYNA direct method and the DRM.
7. The LS-DYNA DRM uses a smaller soil domain than the direct method and so is faster for all the
analysis cases. The DRM and the direct method results are identical.
Table 5-2: Analysis time in minutes for elastic SSI analyses of LM2 subjected to the Ormsby
wavelet
Analysis case
Rigid basemat Rigid basemat Flexible basemat

Rigid structure Flexible structure Flexible structure
SASSI 0.5 0.5 1.0
LS-DYNA Direct 6.5 31.5 127.1
LS-DYNA DRM 6.1 27.5 29
107
5.3 Calculation of nonlinear SSI responses using SASSI and LS-DYNA
Nonlinear SSI responses calculated using SASSI and LS-DYNA are compared in this section.
The comparisons are made for the building models, MS1F_2 and MS2F, considered in the City Block
project. The soil profile used in the City Block project is adopted for the analyses of this section. The
numerical models of the soil profile and the buildings are presented in Section 7.3 and Section 7.4,
respectively. MS1F_2 is a one-story moment-frame structure based on spread footings, and MS2F is a
two-story shear wall structure founded on a basemat. The design of the models MS1F_2 and MS2F are
based on realistic superstructure and foundation designs, and using these models for the numerical
analyses presented in this will provide an insight into nonlinear SSI effects in buildings commonly
encountered in practice. This section only considers SSI effects in the standalone structures (no SSSI). No
comparison is made with experimental results in this section. The ground motions used for the SSI
analyses of the City Block project are also used for the analyses of this section. These ground motions,
identified as JOS_L, LCN, SCS_H and PRI_H, are presented in Figure 7-2 and Figure 7-3 using 5%
damped acceleration response spectra and acceleration time series, respectively.
The SASSI and LS-DYNA responses of MS1F_2 and MS2F are compared for two cases: 1)
linear elastic analyses in both SASSI and LS-DYNA using strain-compatible shear moduli and 2%
damping ratio for the soil, and 2) equivalent-linear analyses in SASSI using strain-compatible shear
moduli and damping for the soil, and nonlinear analyses in LS-DYNA. Analyses for the former case are
performed for the low-intensity ground motion, JOS_L. Analyses for the latter case are performed for all
four ground motions. The analyses in the latter case are also performed for MS1F_2 equipped with rigid
grade beams to demonstrate the effect of restraining the rocking of footings.
5.3.1 Superstructure and soil modeling
The numerical models of the soil profile, as well as of the structures MS1F_2 and MS2F, in
SASSI and LS-DYNA are described in detail in Sections 7.3 and 7.4, and are briefly described here.
Since SASSI is a linear analysis program, the soil profile in SASSI is modeled using strain-compatible
properties that approximate the nonlinear soil response. These properties are dependent on the ground
motion input and are calculated using the site-response analysis program, SHAKE, for each ground
motion. The strain-compatible properties for the ground motions considered in this section are presented
in Figure 7-9. The process of calculating the strain-compatible properties in SHAKE, and the soil profile
modeling in SASSI are presented in Section 7.3.2 and Section 7.4.1, respectively. The SASSI numerical
models of MS1F_2 and MS2F are presented in Section 7.4.1. For the equivalent-linear case, the soil
material in LS-DYNA is modeled using the *MAT_ELASTIC model with strain-compatible moduli. For
108
the nonlinear analyses, the soil is modeled using the *MAT_HYSTERETIC material model. Damping is
modeled using the *DAMPING_FREQUENCY_RANGE_DEFORM option for both the equivalent-
linear and nonlinear analyses. This option is used in nonlinear analyses to model only the low-strain
damping in the soil domain and structural damping. Soil damping in a nonlinear analysis is primarily
achieved through the hysteretic response of the soil. The footings of MS1F_2, and the basemat of MS2F
are modeled as rigid, for the reasons given in Section 7.4.2. The foundation-soil interface in LS-DYNA is
modeled as 1) a ‘tied’ interface with no loss of contact between the soil and the foundation, and 2) a more
realistic interface that allows for loss of contact between the soil and the foundation, resulting in sliding
and gapping. The tied interface is modeled using the *CONSTRAINED_EXTRA_NODES option, which
constrains the soil nodes at the soil-foundation interface to move along with the rigid foundation. If the
foundation is flexible, the contact model, *CONTACT_TIED_SURFACE_TO_SURFACE, should be
employed. To allow separation at the foundation-soil interface, the *CONTACT_AUTOMATIC
_SURFACE_TO_SURFACE option, which is also used in the City Block simulations described in
Section 7.4. Further details regarding the LS-DYNA numerical models of the soil profile, and the
structures MS1F_2 and MS2F, are presented in Section 7.4.
5.3.2 Analysis results and comparison
Nonlinear behavior in a soil-structure system can be classified into 1) nonlinear site response
(primary nonlinearities), which results in differences in the free-field soil response and therefore the
foundation-level input to the structures, and 2) nonlinear behavior at the foundation (soil hysteresis,
gapping, sliding) and nonlinear structural behavior (secondary nonlinearities). Although SASSI can
approximately account for the primary nonlinearities when strain-compatible soil properties are used, it
cannot directly account for secondary nonlinearities. LS-DYNA can directly account for all the nonlinear
behavior in the soil-structure system. For a linear model, the responses calculated using SASSI and LS-
DYNA should be very close, as seen in Section 5.2. However, differences can be expected between the
nonlinear responses calculated using SASSI and LS-DYNA, even when identical material properties are
used. These differences primarily arise from the extent of nonlinear behavior captured by the two
programs. As seen in the site-response comparisons presented in Section 4, it is expected that the SASSI
and LS-DYNA responses are increasingly different as the response of the system is more nonlinear.
The differences in the total acceleration at the roof of a structure calculated using SASSI and LS-
DYNA reflect both the primary and secondary nonlinearities in the soil-structure system. Primary
nonlinearities are investigated in Section 4 through an assessment of site-response analysis programs. To
address secondary nonlinearities, which is the focus of this section, the roof accelerations relative to base
are considered, which do not include the free-field acceleration. Therefore the roof acceleration of
109
MS1F_2 is calculated in terms of 1) acceleration relative to the base (one of the footings), and 2) total
acceleration. Similarly, since MS2F predominantly responds in the rocking mode, the roof acceleration of
this structure is calculated in terms of 1) acceleration due to rocking, and 2) total acceleration. The former
component in the roof acceleration in both the structures is assumed to be primarily influenced by
secondary nonlinearities. The usage of contact models in LS-DYNA introduces high-frequency noise in
the acceleration response of the structures. Therefore all the LS-DYNA acceleration responses are filtered
using a low-pass filter with a cut-off frequency of 10 Hz. Since the anticipated flexible-base frequencies
of the structures, and the fundamental frequency of the soil profile (all around 1.7 Hz) are much smaller
than 10 Hz, this cut-off frequency is deemed appropriate for the analyses presented in this section. The
following sections present the responses of MS1F_2 and MS2F for various cases and compare the
responses calculated using SASSI and LS-DYNA.
5.3.2.1 MS1F_2
Before comparing the equivalent-linear response in SASSI and the nonlinear response in LS-
DYNA, a linear elastic analysis is performed using strain-compatible moduli and 2% damping ratio in
both the programs, and for the JOS_L ground motion input. Figure 5-13 presents the 5% damped
response spectra of the accelerations relative to the base, and the total accelerations at the roof of MS1F_2
for this linear elastic analysis. The figure shows that the results are almost identical, demonstrating the
equivalency of the SASSI and LS-DYNA SSI models of MS1F_2 in the linear domain.
Figure 5-13: Acceleration response spectra at the roof of MS1F_2 subjected to the JOS_L ground
motion for the preliminary linear analysis
110
Nonlinear SSI response of MS1F_2 is calculated by changing the material properties in the linear
models of SASSI and LS-DYNA, and modeling the foundation-soil interface in LS-DYNA using the
appropriate contact models. The equivalent-linear, strain-compatible moduli and damping ratios are used
in SASSI, and the nonlinear hysteretic material model, *MAT_HYSTERETIC, is used in LS-DYNA to
model the soil.
The analysis results for the four ground motions are presented in Figure 5-14. In this figure ‘LS-
DYNA (tied)’ and ‘LS-DYNA (separation allowed)’ represent the two foundation-soil interface
conditions mentioned previously. Contrary to the results of the linear analysis in Figure 5-13, this figure
shows significant differences in the SASSI and LS-DYNA results for both relative and total acceleration
responses. The differences are significant even for the JOS_L ground motion, which results in similar
free-field accelerations in SASSI and LS-DYNA (see Section 7.3.4 for free-field responses). The peaks
in the relative acceleration spectra calculated using LS-DYNA (separation allowed) occur at longer
periods than those calculated using SASSI, indicating that the differences arise from increased flexibility
either at the foundation or in the superstructure, due to nonlinear behavior. Since the fuses of MS1F_2 did
not yield for the first three ground motions, the differences in the structural responses are attributed to
nonlinear behavior in the soil near the footings. This hypothesis is confirmed by performing a separate
nonlinear analysis in LS-DYNA, with the soil in the vicinity of the footings modeled with linear elastic
material, and with a tied foundation condition. Figure 5-15 presents the results of this analysis as well as
the results for the SASSI analysis for the JOS_L ground motion.
111
Figure 5-14: Acceleration response spectra at the roof of MS1F_2 calculated using strain-
compatible soil properties in SASSI, and nonlinear soil properties in LS-DYNA
112
Figure 5-15: 5% damped acceleration response spectra at the roof of MS1F_2 subjected to the
JOS_L ground motion calculated using strain compatible soil properties in SASSI and nonlinear
soil properties in LS-DYNA but ignoring nonlinear soil behavior at the vicinity of the footings
The SASSI and LS-DYNA responses shown in this figure virtually identical, and confirm that
differences in the structural responses for the JOS_L ground motion in Figure 5-14 are due to nonlinear
soil behavior in the vicinity of the footings. Figure 5-14 also shows that preventing separation at the
foundation-soil interface in the LS-DYNA model by using a tied foundation condition significantly
changes the response. Specifically, the peaks of the relative acceleration spectra occur at a shorter period
(which is still longer than the period corresponding to the SASSI responses), and the amplitude of these
peaks is smaller as compared with the LS-DYNA case that allows separation at the foundation. This
indicates that a significant amount of nonlinear behavior at the foundation is due to the gapping and
sliding of footings, and partly due to the hysteretic behavior of the soil. However, difference between the
LS-DYNA results for the tied interface and the SASSI results indicates that the nonlinear response due to
soil hysteresis is still significant. In summary, the results presented in Figure 5-14 show the significant
impact of secondary nonlinearities, including sliding and gapping of the footings and hysteretic behavior
of soil at the footing vicinity, on the MS1F_2 response. These nonlinearities result in larger peak spectral
accelerations that occur at longer periods as compared with the equivalent-linear response.
Yielding of the beam fuses is observed in the LS-DYNA response to the PRI_H ground motion
when separation at the foundation-soil interface is allowed. However, no yielding is observed in the
corresponding LS-DYNA response for the tied foundation case, indicating that gapping and sliding of the
footings result in increased ductility demands in the roof beams. No yielding is observed in the beams for
113
the other ground motions for both the LS-DYNA cases. SASSI cannot directly capture yielding in a
superstructure.
5.3.2.2 MS1F_2 equipped with rigid grade beams
Considering the observed significance of footing displacement in the response of MS1F_2, the
impact on the response of restraining the footings through grade beams is investigated in this section.
Grade beams that connect the footings are used in practice to prevent the differential horizontal
displacement of footings. Grade beams can also restrain the independent rocking of footings. Rigid grade
beams are simulated in LS-DYNA by constraining all the rigid footings to move together, acting as a
single rigid body. Rigid grade beams in SASSI are explicitly modeled as very stiff beams connecting the
bases of the columns. Figure 5-16 presents the SASSI and LS-DYNA responses of MS1F_2 equipped
with rigid grade beams. The LS-DYNA response accounts for separation at the foundation-soil interface.
The results presented in this figure are markedly different from the responses of MS1F_2 without grade
beams. Firstly, the responses calculated from SASSI and LS-DYNA for the modified structure with grade
beams are much closer than those calculated for the original structure (Figure 5-14). This observation
holds for both the relative accelerations and the total accelerations at the roof, and indicates that using
grade beams reduces nonlinear behavior at the footings. Secondly, although the period of maximum
spectral acceleration (relative to the base) remains almost unchanged for the SASSI responses, it
decreases significantly for the LS-DYNA responses. Thirdly, the peak spectral accelerations of modified
MS1F_2 are slightly smaller than those calculated for the original structure for all but the JOS_L ground
motion. These differences are all primarily due to the reduced gapping and sliding of the footings when
grade beams are used, which also leads to lesser hysteretic behavior in the soil at the vicinity of the
footings. Figure 5-16 also shows a significant reduction in the peak spectral acceleration at the roof of the
modified structure for the PRI_H ground motion. The SSI period of the original structure (about 1.2 sec)
coincides with the spectral peak in the free-field response for this ground motion. Using grade beams
reduces the SSI period to about 0.45 sec, and therefore reduces the peak spectral acceleration at the roof
of the modified structure. No yielding is observed in the beams of MS1F_2 equipped with grade beams,
even for the PRI_H ground motion. The grade beams prevent differential horizontal movement of the
footings: footing spreading. The footings in the original structure slide and rock independently, resulting
in footing spreading, which increases the bending moment in the beam fuses at the roof level. Grade
beams restrict the footing spreading and result in smaller moment demands in the beams.
114
Figure 5-16: Acceleration response spectra at the roof of MS1F_2 equipped with grade beams,
calculated using strain-compatible soil properties in SASSI, and nonlinear soil properties in LS-
DYNA
115
Figure 5-17 presents the displacement of two adjacent footings along the X-direction for both the original
and modified MS1F_2 subjected to the PRI_H ground motion. Footings 1 and 2 in the figure correspond
to the left-hand side and right-hand side footings of the structure, when facing in the positive Y-direction.
Footing 2 shows larger residual displacements than footing 1 for the original structure, indicating lateral
spreading. The footings in modified MS1F_2 have identical displacements. Figure 5-18 presents
screenshots of the deformed LS-DYNA models of the original and modified structures. The figure clearly
shows the reduction in footing displacements when grade beams are used.
Figure 5-17: Time series of the horizontal displacement of two adjacent footings (along the X
direction) of MS1F_2 for the PRI_H ground motion
116
a. original MS1F_2 b. MS1F_2 with grade beams
Figure 5-18: Screenshots of the deformation at the footings of the MS1F_2 for the PRI_H ground
motion at about 7 sec
5.3.2.3 MS2F
Similar to MS1F_2, a linear elastic analysis is first performed for MS2F using strain-compatible
moduli and 2% damping ratio for the soil in both SASSI and LS-DYNA. This linear elastic analysis is
performed for the JOS_L ground motion only. Results for this analysis are presented in Figure 5-19.
Figure 5-19: Acceleration response spectra at the roof of MS2F subjected to the JOS_L ground
motion calculated using strain-compatible moduli and 2% damping ratio for the soil
The figure shows that the SASSI and LS-DYNA responses are very close, but not as close as the
responses calculated for MS1F_2. This is because the rocking response of MS2F is particularly sensitive
117
to the layer thickness and properties in the vicinity of the foundation. Small differences exist between the
thicknesses of the soil layers near the surface of the SASSI and LS-DYNA models. The LS-DYNA
structural model is founded on a finely-meshed soil to account for the large strains expected in the
nonlinear analysis. However the SASSI structural model is founded directly on the soil layers, which are
not meshed with brick elements, unlike in LS-DYNA. The sensitivity of the MS2F response to the soil
layer thicknesses and material properties is further examined in Section 5.4.4. The close match of the
SASSI and LS-DYNA responses in Figure 5-19 still confirms that the corresponding numerical models
are equivalent in the linear domain.
Following the verification of the numerical models for linear analysis, nonlinear analyses in LS-
DYNA and equivalent-linear analyses in SASSI are performed for MS2F subjected to the four ground
motions. Similar to MS1F_2, two contact models (tied and separation-allowed) are employed for the
foundation-soil interface. Figure 5-20 presents the 5% damped response spectra of the roof accelerations
(rocking accelerations in the left panels and total accelerations in the right panels) calculated using SASSI
and LS-DYNA. The figure shows that the SASSI and LS-DYNA responses are closer than those
calculated for MS1F_2 without grade beams. The LS-DYNA responses for the two contact conditions are
also closer than those calculated for MS1F_2. There is a small increase in the peak spectral accelerations
calculated using LS-DYNA when the foundation-soil interface enables separation. This increase is due to
the sliding of the basemat from loss of contact with the surrounding soil. This increase is much smaller
than that observed for MS1F_2 (without grade beams), indicating that the degree of sliding and gapping
in MS2F is smaller. Figure 5-20 shows that although the SASSI and LS-DYNA responses are close for
the JOS_L ground motion, the peak spectral responses of the total acceleration are smaller than those
calculated using SASSI for the other three ground motions. This is primarily because of the
underestimation of the free-field response by LS-DYNA as compared with SASSI as seen in Figure 7-8
(this figure presents the free-field responses calculated from SHAKE, which are generally identical to
those calculated using SASSI). The peak spectral acceleration of the rocking response calculated using
LS-DYNA is also slightly smaller than the corresponding SASSI response for the SCS_H and the PRI_H
ground motions. This is because of the hysteretic response of the soil in the vicinity of the basemat
(secondary nonlinearities) that involves softening of the soil and damping of the rocking response of
MS2F. However the comparatively small difference between the SASSI and LS-DYNA responses for
MS2F, as compared with MS1F_2, indicates that the secondary nonlinearities for the basemat are smaller
than those observed for the footings.
118
Figure 5-20: Acceleration response spectra at the roof of MS2F calculated using strain-compatible
soil properties in SASSI, and nonlinear soil properties in LS-DYNA
Figure 5-21 presents the 5% damped spectral responses of the roof accelerations from structural
deformation in MS2F. These accelerations are calculated by subtracting the rocking acceleration at the
roof from the basemat input acceleration (horizontal acceleration at the center of the basemat). A
comparison of the magnitude of spectral accelerations in this figure with those in Figure 5-20 shows that
119
the acceleration component from structural deformation is significantly smaller than the rocking
acceleration for all ground motions. This is expected since MS2F is designed to primarily respond in the
rocking mode. The figure also shows that the spectral accelerations calculated using LS-DYNA are much
larger than those calculated using SASSI at periods shorter than the rocking period (~0.6 sec for JOS_L,
and ~1 sec for the other ground motions). This is mainly due to the high-frequency (short period) noise
introduced by the contact elements used in LS-DYNA. Since the fixed-base time period of MS2F (0.14
sec) in the range of the frequencies corresponding to the noise, this high-frequency response is amplified
in the structure. The difference between the amplitude of the short-period response calculated using the
two contact models in LS-DYNA also indicates that this response might be a result of using contact
elements. However since this component of the acceleration is negligible as compared with the rocking
accelerations, the effects of these short-period components can be ignored for this structure.
Figure 5-21: Response spectra of the accelerations from structural deformations calculated for
MS2F using strain-compatible properties in SASSI and nonlinear properties in LS-DYNA
5.3.3 Observations
The following observations can be made from the equivalent-linear and nonlinear analyses
presented in Section 5.3.2:
120
1. The responses calculated using SASSI and LS-DYNA for the linear analyses match very closely
for both MS1F_2 and MS2F, indicating that the soil and structural models built in SASSI and LS-
DYNA are equivalent in the linear domain.
2. Significant differences are seen in the equivalent-linear (SASSI) and nonlinear (LS-DYNA)
responses of MS1F_2 even for the low-intensity ground motion, JOS_L. This is due to the highly
nonlinear behavior of the soil in the vicinity of the footings (secondary nonlinearities).
3. Accounting for separation at the foundation-soil interface significantly increases the response of
MS1F_2 due to the contribution from sliding and gapping of the footings. Neglecting footing
sliding and gapping can result in unconservative estimates of demands for MS1F_2.
4. Restraining the rocking of the footings using rigid grade beams in MS1F_2 leads to a marked
difference in response. The SASSI and LS-DYNA responses of modified MS1F_2 are much
closer due to the significant reduction in the secondary nonlinearities.
5. Retrofitting MS1F_2 with rigid grade beams affects the yielding of the roof beams for the PRI_H
ground motion. The grade beams compel the footings to respond together, preventing lateral
spreading of the footings and reducing the moment demand in the beams.
6. The equivalent-linear (SASSI) and nonlinear (LS-DYNA) responses of MS2F are much less
different than the responses calculated for MS1F_2 without grade beams. Most of the differences
in the total response are attributed to the difference in the free-field input to the structure.
Differences in the rocking responses are smaller, and are due to the nonlinear response of the soil
in the vicinity of the basemat.
7. Accounting for sliding and gapping of the basemat slightly increases the peak spectral
accelerations of MS2F. However this increase is much smaller than that observed for MS1F_2.
8. The secondary nonlinearities in MS2F are much less pronounced than those in MS1F_2 without
grade beams.
9. The roof accelerations from structural deformation of MS2F calculated using LS-DYNA contains
a significant contribution of short-period accelerations as compared with SASSI. This is due to
the amplification of high-frequency noise (introduced by the contact models) by the
superstructure.
5.4 Practical considerations for SSI analysis using SASSI and LS-DYNA
The process of comparing the SASSI and LS-DYNA responses in the linear and nonlinear
domain revealed several challenges that could be faced by analysts performing SSI analyses using these
programs. These challenges are unlikely to be identified if only one of the programs is used for the
121
analysis. To avoid these challenges some practical considerations are to be made while performing the
SSI analyses. This section outlines these considerations for both SASSI and LS-DYNA.
5.4.1 Selection of analysis frequencies in SASSI
The SASSI input includes a set of frequencies at which the analysis is performed. The SASSI
user manual (Ostadan, 2006b) mentions that a reasonable response can usually be achieved when 10 to 20
frequencies are used. Additionally, the manual recommends that these frequencies be chosen based on the
peaks in the transfer functions observed in the fixed-base response of the structure. Figure 5-22 presents
the spectral accelerations at the roof of the LM2 structure calculated using SASSI for a set of 28
frequencies and a set of 44 frequencies. The LS-DYNA response is also presented in the plots for
comparison. The frequencies used in the analysis are indicated with solid black circles on the SASSI
responses. Both the frequency sets have an adequate number of frequencies around the fixed-base period
(0.14 sec).
Figure 5-22: Acceleration response spectra calculated at the roof of LM2 using SASSI for two sets
of analysis frequencies, and using LS-DYNA
However, the significant difference in the SASSI response resulting from the use of more
frequencies between 0.2 sec and 1 sec is clear in the figure. It appears that this improvement primarily
comes from including more frequencies at the three peaks occurring at about 0.25 sec, 0.4 sec, and 0.7
sec. The peak at 0.4 sec corresponds to the flexible-base period of the structure, and the other two peaks
correspond to the natural periods of the soil profile. This indicates that the input set of frequencies should
122
be complete enough to characterize the peaks in the transfer function corresponding to 1) flexible-base
period of the structure, and 2) the natural periods of the site. Additionally, the frequency set should also
include the dominant frequencies of the input ground motion. Although the site periods can be estimated
by a separate site-response analysis, the flexible-base period of the structure can be difficult to identify
before the analysis, especially for complex structures. This makes it difficult to choose the analysis
frequencies a priori.
5.4.2 Foundation meshing in SASSI
The SSI analysis in SASSI involves calculation of the foundation impedance functions at the
interaction nodes. For the analysis of MS2F presented in this section, the interaction nodes include all of
the nodes in the basemat. SASSI calculates the impedance functions only for these nodes and forms the
impedance matrix, which is used to calculate the transfer functions. Therefore the meshing of the basemat
(which determines the number of nodes in the basemat) affects the foundation impedance matrix, and
therefore the overall foundation impedance. This is unlike the finite element method, in which the
foundation impedance is sensitive to the meshing of the soil around the basemat, and not the meshing of
the basemat itself, which is modeled as a rigid body. Figure 5-23 presents the rocking response of MS2F
calculated using SASSI for the two different basemat meshes shown in the figure. The corresponding LS-
DYNA response is also provided for comparison. The figure shows significant differences between the
SASSI responses calculated for the two basemat meshes, and that the response for basemat 2, which has a
finer mesh, is almost identical to the LS-DYNA response9. Therefore care must be taken when meshing
the foundation for a SASSI analysis.
9
The structure-basemat connections employed in the two basemat models are also different. However since
the basemat is modeled to be much stiffer than the surrounding soil, the structure-basemat connection is assumed to
have no effect on the structural response.
123
Figure 5-23: Sensitivity of the rocking response of MS2F calculated using SASSI to the number of
nodes in the basemat
5.4.3 Column-basemat connection properties in SASSI
SASSI does not have a provision to directly model a rigid connection between the beam elements
of the structure and the solid elements of the foundation. These connections can be modeled using beam
elements with a stiffness much larger than the stiffness of the beams elements in the superstructure (see
Figure 5-24). However, if the stiffness of these connection beams is too large, SASSI produces
anomalous results, as shown in Figure 5-25. The figure shows the SASSI responses (1% damped
acceleration amplification spectra from the surface free-field), calculated at the roof of the LM2 structure
when subjected to the Ormsby wavelet, for three different values of stiffness of the beams constituting the
column-basemat connection. The column-basemat connection in LS-DYNA is modeled using the
*CONSTRAINED_ NODAL_RIGID_BODY option, which provides a fully rigid connection. Figure
5-25 also includes the corresponding LS-DYNA response. There are large differences between the SASSI
responses calculated using the different value of stiffness for the column-basemat connection beams. Only
the SASSI response with connection 3 matches closely with the LS-DYNA response.
124
Connection 1: EI = 6.9  1022 kN-m2
Figure 5-24: Description of the column-basemat connection of LM2 as modeled in SASSI and the
beam cross-section properties used to model the three different connections employed in Section
5.4.3
Figure 5-25: SASSI response of the LM2 structure for various structure-basemat connections
considered in Section 5.4.3, and the LS-DYNA response
It is difficult to pinpoint the exact reason for the anomalous responses in SASSI without access to
the source code. On the basis of this limited study, the rigid beams should be modeled using a stiffness
that is at most 104 times greater than that of the superstructure beam elements. Using a stiffness that is too
large can result in anomalous responses, which can be taken as a true response without a benchmark
solution such as the one provided by LS-DYNA in this case.
125
5.4.4 Sensitivity of rocking response to site layering in SASSI and LS-DYNA
The soil layer thicknesses in a soil profile model (for both site-response and soil-structure
interaction analyses) are chosen based on the maximum frequency (minimum wavelength) that propagates
through the soil profile. The SHAKE and the SASSI user manuals recommend at least five soil layers per
minimum wavelength [ n  5 in equation (4-1)], whereas Jeremić et al. (2009) recommend 10 layers per
minimum wavelength for time-domain nonlinear analyses. However these recommendations are primarily
intended for site-response predictions, and do not consider the impact on SSI response of a structure.
Figure 5-26, for example, presents the SSI response of MS2F subjected to the JOS_L ground motion,
calculated using SASSI and LS-DYNA for two site layering procedures. Layering scheme 1 corresponds
to a soil profile with layer thicknesses calculated using n  5 , and layering scheme 2 corresponds to the
layer thicknesses calculated using n  10 . The figure also includes the corresponding strain compatible
shear wave velocities calculated using SHAKE, and the site responses calculated using SASSI and LS-
DYNA. Although the site responses for these profiles are identical, the corresponding rocking responses
vary considerably. The variation in the rocking response is significantly greater for SASSI than for LS-
DYNA. This is because the strain-compatible properties in SASSI are calculated from a site-response
analysis, and do not account for the stresses imposed by a structure. LS-DYNA, however, performs a
fully nonlinear analysis and accounts for the soil stresses imposed by the structure. The small variation in
the LS-DYNA response indicates that the thickness of the soil layers near the foundation have a minor
effect on the rocking response of MS2F.
126
Figure 5-26: Sensitivity of the rocking accelerations of MS2F to site layering in the equivalent-
linear method
5.4.5 Hourglass effects in LS-DYNA response
Preliminary analyses of LM2 with a flexible basemat and a flexible superstructure revealed large
differences between the SASSI and LS-DYNA responses. Figure 5-27 presents the 1% damped
acceleration amplification spectra at the roof of LM2. The dotted blue line in this figure presents the LS-
DYNA response of LM2 in this analysis. This figure clearly shows that the SASSI and LS-DYNA
responses are very different. The peak amplification in the LS-DYNA response occurs at a longer period
127
and is smaller than that of the SASSI response, indicating significant softening in either the structure or
the foundation of the LS-DYNA model. An examination of the deformed shape calculated using LS-
DYNA (panel a in Figure 5-28) reveals that the deformation in the solid elements at the perimeter of the
basemat is comparable to the deformation in the adjacent soil elements. This result is unrealistic, given
that the basemat is significantly stiffer than the soil, and is likely a result of hourglass effects. To further
investigate this result the LS-DYNA analysis is performed again using hourglass control [IHQ = 5 in
*CONTROL_HOURGLASS; refer to the LS-DYNA keyword manual (LSTC, 2013)], that prevents the
formation of hourglass modes. The solid blue line in Figure 5-27 presents the amplification spectra for the
LS-DYNA analysis with hourglass control. The SASSI and LS-DYNA results are now much closer,
indicating that the softening in the LS-DYNA model is due to hourglass effects. The corresponding
deformed shape (panel b in Figure 5-28) also shows a significant smaller deformation in the basemat
elements that the adjacent soil elements. Hourglass control should therefore be used for SSI analyses in
LS-DYNA. An alternative method to avoid hourglass effects is to use full integration in the solid
elements. However this requires much longer computation times.
Figure 5-27: Sensitivity of the LS-DYNA response of LM2 to hourglass effects
128
a. Without hourglass control b. With hourglass control
Figure 5-28: Screenshots of the LS-DYNA response (500 times amplified) of LM2 with and without
hourglass control
Linear elastic and fully nonlinear SSI analyses are performed in SASSI and LS-DYNA. The
process of the performing these analyses and a comparison of results revealed some potential challenges
for SSI analysts. The following conclusions can be drawn from the studies presented in this section:
1. SASSI and LS-DYNA produce almost identical responses for SSI analysis of linear elastic
models. Small differences in peak spectral accelerations can be expected, which arise from the
differences in the damping formulations.
2. The nonlinear response predictions of SASSI and LS-DYNA for MS1F_2 are very different even
for low-intensity ground motions, due to nonlinear soil behavior in the immediate vicinity of the
footings. Restraining footing rocking using grade beams reduces this nonlinear behavior and
brings the SASSI and LS-DYNA responses much closer. Equipping a structure with grade beams
also reduces ductility demands in the floor beams of moment frames by eliminating relative
horizontal displacement in the footings (footing spreading).
3. The nonlinear response predictions of SASSI and LS-DYNA are closer for the two-story shear
wall structure on a basemat (MS2F) than those of the one-story moment frame structure on spread
footings (MS1F_2) because the secondary nonlinearities are much smaller. The response
predictions from the two programs become increasingly different for more intense ground
129
motions due to 1) the differences in the site response, and 2) increasing nonlinear behavior of the
soil near the foundation.
4. The responses of both the structures can be significantly affected by foundation sliding and
gapping caused by separation at the foundation-soil interface. Neglecting these effects can result
in unconservative acceleration demands in the superstructure.
5.6 Recommendations
SSI effects in buildings are typically evaluated using simplified methods such as those prescribed
in ASCE 7-10 (ASCE, 2010) and FEMA 440 (FEMA, 2005), which are based on linear analytical
solutions. The study presented in this section shows that nonlinear SSI effects can be significant in
buildings, especially for structures founded on spread footings. Ignoring these effects can result in
unconservative estimates of demand in the superstructure. In the context of performance-based design,
nonlinear soil-structure interaction effects should be considered if accurate estimates of loss are required.
Soil-structure interaction analysis using either SASSI or LS-DYNA can be complicated, both in
terms of execution and interpretation of the results. The possible anomalous results presented in Section
5.4 could be interpreted as the correct response. Indeed the issues presented in this section may not have
been identified if only one of the programs had been used for analysis. For example, the SASSI response
with an inadequate number of frequencies presented in Figure 5-22 could be assumed to be the exact
result, if not compared with a reference result calculated using LS-DYNA. The use of both programs and
preliminary analyses with simplified models may be the only practical way to ensure that the results are
accurate.
130
SECTION 6
THE NEES CITY BLOCK CENTRIFUGE EXPERIMENTS: TEST 3 AND TEST 4
6.1 Introduction
The experimental program of the NEES City Block project included six centrifuge tests, titled
Test 1 through Test 6, that examined the phenomena of soil-structure interaction (SSI) and structure-soil-
structure interaction (SSSI). A preliminary test, Test 0, was performed on the soil alone to calibrate the
ground motions to be input to the centrifuge shake table during Tests 1 through 6. Test 1 is a baseline test
that involves two building models (a one-story model supported on footings, and a three-story model on
an embedded basement), representing a low-rise and a mid-rise SMRF building at the prototype scale,
placed sufficiently far away from each other to eliminate significant interaction. These structures are
placed directly adjacent to each other in Test 2 to observe SSSI effects between the structures. Test 3 and
Test 4 involve stiffer, heavier structures (a one-story model supported on footings and a two story shear
wall model) designed to exacerbate the interaction between the structures in such a way that the dynamic
behavior of one structure affects the behavior of the other structure. These structures are placed in
different arrangements to understand the effects of their relative location on SSSI. The study presented in
this report is only confined to Test 3 and Test 4. Descriptions of the plan and execution of Test 3 and Test
4 are presented in Sections 6.4, and 6.5, respectively. Experimental results are presented in the next
section, along with the description of the comprehensive numerical simulations of Test 3 and Test 4. Test
6 and Test 7 examine liquefaction effects, along with SSSI, and are not presented in this report. Data and
documentation for all the NEES City Block experiments are available for open access in the project
website in the NEEShub Project Warehouse (https://nees.org/warehouse /project/639).
6.2 Geotechnical centrifuge testing
6.2.1 Background
Physical phenomena such as site effects and soil-structure interaction (SSI) occur at a very large
dimensional scale and are almost impossible to physically simulate in the laboratory at a full scale. Scale
models can be built, but owing to the complicated material behavior of soil it is desirable that these
models have an identical strength and stiffness as the prototype. To maintain identical strength and
stiffness in the model and prototype, similitude laws are used to establish scale factors for all the other
parameters in the model, given the factor with which the model is scaled down from the prototype. These
scale factors are listed in Table 6-1 below. Derivations for these scale factors are presented in Kutter
(1995) and Bolisetti (2010). The acceleration scale factor in the table indicates that the model acceleration
131
is scaled up the same amount as the model dimensions are scaled down. This indicates that for a model
that is smaller than the prototype by a factor of 50, the accelerations should be scaled up 50 times,
including acceleration due to gravity, which is only possible in a centrifuge.
Table 6-1: Scale factors used in centrifuge testinga (Kutter, 1995)
Quantity Scaling Factor

Stress 1
Strain 1
Length 1N
Acceleration N
Dynamic Time 1N
Period 1N
Frequency N
3
Mass 1N
2
Force 1N
a
N is the centrifuge scale factor, and a value of 55 is used for
the City Block project
A geotechnical centrifuge consists of a swing bucket that hangs freely on one end of a horizontal
arm, which is attached to a vertical axle that supports the rotation of the arm-bucket assembly (see Figure
6-1). The swing bucket contains a laminar box mounted on a shake table that simulates earthquake
shaking. The swing bucket is equipped with a data acquisition system. The laminar box contains the
instrumented experimental model including the soil and structures. The role of the laminar box is to
simulate a transmitting boundary that absorbs the waves radiating from the soil surface and the structures
(if any) thus simulating an infinite soil halfspace. When the centrifuge arm rotates around the vertical
axle, it creates a centrifugal force acting in the vertical direction relative to the soil-structure model in the
laminar box. By maintaining a pre-determined angular velocity, a vertical force equivalent to the required
gravitational force in the model can be simulated.
132
a. Photograph of the UC Davis Geotechnical Centrifuge showing the
horizontal arm, the swing bucket, and the laminar box
b. Photograph of the centrifuge swing bucket when fully equipped showing

the laminar box in the bucket and the data acquisition system attached to the
outside walls of the bucket
Figure 6-1: The UC Davis centrifuge
133
A centrifuge test is conceived by choosing the prototype structures and soil site characteristics
such as topography and soil profile. The centrifuge scale factor is the most crucial parameter in the test
plan and is determined by the size of the laminar box, size of the prototype structures, and the level of
detail desired in the experimental models. For the City Block project, a scale factor of 55 was chosen.
Once the scale factor is chosen the model structures are designed and the test ground motions are selected.
Model structures are then fabricated and instrumented with strain gages, accelerometers, etc. The laminar
box is filled with the soil using a technique that maintains a desired in-situ relative density. This
technique is known as pluviation, and the device used for this purpose is known as the pluviator (Figure
6-2). The pluviator is an elevated soil container that gradually pours the soil into the laminar box placed
below. It is moved back and forth along the length of the laminar box at a specific speed, so as to
maintain uniformity in the soil profile. The relative density of the soil in the laminar box is determined by
the rate at which the soil is discharged from the pluviator. This rate depends on the speed of the pluviator
and its elevation, which are pre-calibrated in order to achieve the intended relative density in the laminar
box. The instruments used to record soil responses (e.g., accelerometers, pore pressure transducers) are
placed in the soil during pluviation. Instrumented structural models are also carefully placed in the soil
during pluviation (or after pluviation in case of surface mounted structures). The laminar box containing
the soil-structure model is then carefully transferred onto the swing bucket on the centrifuge arm (Figure
6-2). The soil and structural instruments are then connected to the data acquisition system before the
centrifuge is spun for testing.
134
a. A building model before instrumentation b. Laminar box before pluviation
c. Pluviation d. Laminar box right after pluviation
e. Laminar box and building models after instrumentation f. Laminar box in the swing bucket
Figure 6-2: Photographs of the experimental model building process. Courtesy of the City Block
team
135
6.2.2 Advantages and limitations
Centrifuge testing can be a very useful tool for gaining insight into several geotechnical
phenomena. Field observations from reconnaissance missions and data analysis from instrumented
structures can be helpful but are limited in terms of the types of structures and ground motions. Centrifuge
testing is therefore an economical alternative that can physically simulate complex geotechnical
phenomena in a laboratory. By the nature of laboratory testing, these phenomena can be repeatedly
simulated for a wide range of situations. The repeatability also enables the analysis of simpler problems at
first and the gradual increase of complexity (e.g., by adding additional structures) thereby allowing
complex problems to be broken down and understood incrementally.
Centrifuge testing has several limitations that must be considered before interpreting test data.
Predictably most of these limitations arise from the fact that centrifuge tests are 1) scaled down by large
factors (scale effects), and 2) performed in the finite domain of a laminar box as opposed to an infinite
soil halfspace (boundary effects). Soil in a centrifuge model, being a much smaller specimen than the
corresponding prototype, may exhibit different stress-strain behavior from that expected in the prototype.
This is called the ‘particle size’ effect (Kutter, 1995), which is also observed with other geotechnical
testing techniques such as triaxial tests, where specimens of different sizes may exhibit different strain
softening rates. A similar argument can be made for structural models, which are typically tested at near
full scale unlike centrifuge models. Although soil models exhibit particle size effects, it is incorrect to
employ the simplistic assumption that a ‘sand specimen in the centrifuge represents gravel in the
prototype’ (Kutter, 1995). An important consequence of particle size effects lies in the shear behavior at
the soil-structure interface. Soil shear behavior is localized to a ‘shear band’ of soil particles and its
thickness depends on the soil properties (Kutter, 1995). Since the soil properties in centrifuge testing are
independent of scale, the shear band thickness at the soil-structure interface is the same in the model and
in the prototype. This leads to a scaling inconsistency in the shear behavior at the contact interface since
the foundations of the prototype and the (much smaller) model experience the shear band of the same
thickness. Time scaling also results in rate effects in the centrifuge. Since the centrifuge model
experiences the same dynamic stresses and strains as the prototype in a much smaller time span, strain
rate effects can be significant.
One of the goals of using a laminar box is to produce a one-dimensional wave field in the soil
comprising of vertically propagating shear waves. An ideal laminar box needs to have the same stiffness
and damping ratio as the contained soil profile, and should have a mechanism to create the correct shear
stresses at the sides of the soil to reproduce a one-dimensional wave field. Such a laminar box is almost
impossible to build given the complexity of nonlinear stress-strain behavior of the soil. Several laminar
boxes have been developed to better address this problem but none creates a perfect one-dimensional
136
wave field (Kutter, 1995). Boundary effects are therefore expected in centrifuge tests and are partly
avoided by placing the structures away from the walls of the laminar box.
The mass of the centrifuge model, including the soil, structures and the laminar box, is usually of
the same order as that of the shake table and the swing bucket itself. Centrifuge models are therefore
susceptible to soil-container-shaker interaction, which leads to an interaction between the soil response
and the input to the laminar box. This interaction also acts as an energy dissipation mechanism at the base
of the laminar box. Ilankatharan and Kutter (2010) found that the soil-container-shaker interaction
considerably affects the model response and should be accounted for in numerical simulations.
Given the above limitations, centrifuge testing may not be suitable for directly modeling a
specific prototype event, but is useful in discovering mechanisms of behavior, or comparing the
performance of similar structures under different geotechnical conditions (Kutter, 1995). Centrifuge test
data can also be used to validate numerical tools, provided the above stated effects are carefully
accounted for in the numerical simulations.
6.2.3 The NEES@UC Davis centrifuge facility
The NEES City Block (NCB) project involves a series of centrifuge experiments that aim to
understand and characterize the effects of structure-soil-structure interaction on buildings (also called as
‘city block effects’ or ‘site-city effects’) in the context of performance assessment of buildings in dense
urban environments. These experiments are performed on scale models of buildings placed on soil, and
the recorded data is used to understand the mechanisms and consequences of SSSI effects on the
corresponding prototype buildings. The City Block experiments require a sufficiently large centrifuge to
accommodate 6 building models and a robust data acquisition system capable of recording the response of
these structures in significant detail. The Center for Geotechnical Modeling (CGM) at UC Davis, an
equipment site of the George E. Brown, Jr. Network for Earthquake Engineering Simulation (NEES),
operates the largest geotechnical centrifuge in the United States and was used for the NEES City Block
experiments. The UC Davis centrifuge (Figure 6-1) has a 9m radius arm and can accommodate payloads
of 4.5 tons at a rotational speed of 90 rpm, which corresponds to a centrifugal acceleration of 75g.
The centrifuge facility owns both uniaxial (horizontal only) and biaxial (horizontal and vertical)
shake tables. Only the uniaxial shake table is used for the City Block project. This shake table is capable
of producing peak accelerations of up to 30g and a velocity of up to 0.8m/s, depending upon the mass and
stiffness of the experimental model. It is approximately 6.6ft  3.3ft in plan and holds containers of 5.8ft
 2.3ft  2.0ft in size. Figure 6-3 below is a 3D rendering of the uniaxial shake table. The CGM is
equipped with 1) two Flexible Shear Beam containers (FSB-1 and FSB-2) designed to move along with
the soil profile, 2) a Rigid Container (RC-1) with transparent side walls providing a clear view of below
137
grade soil behavior, and 3) a Hinged Plate Container (HPC-1), which is also flexible but retains the
residual lateral deformation of the soil profile. The FSB-2 container is used in the first two City Block
experiments. This container has the inner dimensions 65in  31in  23in (65in in the shaking direction)
and is built from aluminum rings separated by thin strips of neoprene as shown in Figure 6-4. Further
information regarding the specifications of the centrifuge can be found at the NEES at UC Davis website
(www.nees.ucdavis.edu).
Figure 6-3: Rendering of the horizontal uniaxial shake table used at the UC Davis centrifuge facility
(courtesy of the staff at the Center for Geotechnical Modeling, UC Davis)
Figure 6-4: FSB-2 container used in the City Block experiments (courtesy of the staff at the Center
for Geotechnical Modeling, UC Davis)
138
6.3 Test 2: Two adjacent structures on dense, dry sand
6.3.1 Test description
Test 2 was the second of the two baseline tests (Test 1 being the first) that involved two steel
moment resisting frame (SMRF) model structures placed on dense, dry Nevada sand. The main goal of
Test 1 was to examine the soil-structure interaction (SSI) effects on a one-story building model supported
on spread footings and a three-story building model supported on a basement (Bolisetti, 2010). Test 2 was
performed to examine the structure-soil-structure interaction (SSSI) between these models by placing
them adjacent to each other in a line parallel to the shaking direction (termed as in-plane SSSI or iSSSI)
and comparing their seismic responses to those observed during Test 1. The Test 1 models therefore serve
as baseline to the Test 2 models. Figure 6-5 and Figure 6-6 present photographs of the arrangement of the
models in the laminar box during Test 1 and Test 2, respectively.
MS3F_Basel
ine
MS1F_1_Ba
seline
Shaking
Direction
Figure 6-5: Photograph of the building models in the laminar box during Test 1
139
MS1F_1 MS3F_i
_iSSSI SSSI
Shaking
Direction
6.3.2 Soil profile
Dense, dry Nevada sand was used in Test2 (and Test 0 and Test 1). This sand was fine and
uniformly graded, with a coefficient of uniformity ( Cu ) equal to 1.06 and a mean grain diameter ( D50 ) of
0.15mm. Nevada sand was delivered to the UC Davis centrifuge facility in batches. The soil has a specific
gravity ( GS ) of 2.65 and void ratios, emin and emax , of 0.510 and 0.748 respectively, based on the
laboratory tests performed on the batch of sand used in Test 0 and Test 1. No material tests were
performed on batches that arrived after Test 1 because the differences between material properties of
different batches of Nevada sand were assumed to be small. The laminar box was filled to a depth of
21.1in during Test 2 [identical to Test 0 and Test 1; see Bolisetti (2010)], which corresponds to a depth of
97ft at the prototype scale. A relative density of 80% was maintained during pluviation.
6.3.3 Model design and instrumentation
Chen et al. (2010) designed and developed the model buildings used in Test 1 and Test 2. The
one-story model (herein referred to as MS1F_1) represents a three-story steel moment resisting frame
(SMRF) building at the prototype scale, and the three-story model (herein referred to as MS3F) represents
a nine-story SMRF building at the prototype scale. The designs procedures for the models are presented
140
in Bolisetti (2010) as a part of the description of Test 1, and are reproduced here10. Slight modifications
were made to the one-story model for Test 2 as explained in Section 6.3.3.2.
6.3.3.1 Selection of prototype building parameters
The prototype building designs were developed by Ganuza (2006). These designs assumed a
fixed-base, and were developed for a class D site in Los Angeles. As such, they were deemed suitable for
the City Block project. After consultation with the project practice committee, Chen et al. (2010) adopted
a three-story SMRF building, lam123, and a nine-story SMRF building, lam19, as prototypes to develop
models for centrifuge testing. Key design parameters for the buildings at the prototype scale are presented
in Table 6-2 below. In this table, Tn is the time period of the nth mode of vibration, Vy W is the strength
(base shear at yield) to weight ratio, and  y is the yield drift ratio in the uppermost story.
Table 6-2: Target design parameters for model structures at prototype scale (Chen et al., 2010)
Prototype structure Description Tn (sec) Vy W  y (%)

lam19 SMRF, 9-story 2.6, 0.8, 0.5 a 0.2 1.4
lam123 SMRF, 3-story 1.1 a 0.3 1.2
a
The rationale for targeting only the lower modes is provided in Section 6.3.3.2 below
6.3.3.2 Design of experimental model structures
The dimensions of a centrifuge model are generally determined by geometrically scaling down
the prototype dimensions by a scale factor N [equal to 55 in this project; refer to Section 3.2 in Bolisetti
(2010)]. This was not practical for the prototype buildings since it would have resulted in a model with
numerous beams and columns. As such, the models were designed to capture only the first few modes of
corresponding prototype buildings. Model MS1F_1 was designed to capture the first mode of the
prototype building, lam123, and model MS3F was designed to capture the first three modes of the
prototype building, lam19. These modes contributed almost 95% of the total reactive masses (Chen et al.,
2010). Additionally, the models targeted the yield base shear ratio ( Vy W ) and yield drift ratio (  y ) of
the prototypes. The experimental models were also equipped with foundations. The one-story model,
MS1F_1, was built atop spread footings, whereas the three-story model, MS3F, was built atop an
10
The one-story and three story models are referred to as MS1F_SF80 and MS3F_B in the description of
Test 1 in Bolisetti (2010).
141
embedded one-story (in model scale, and three-story in prototype scale) basement. The models were
symmetric in plan and very stiff in the direction orthogonal to the uniaxial shaking. Figure 6-7 presents
photographs of the models equipped with strain gages, prior to testing. Unlike Test 1, these models were
placed adjacent to each other with a gap of 5 mm between the foundations (Figure 6-6). Sketches
detailing the frame elevations along the shaking direction are presented in Figure 6-8. Each model
structure consisted of two parallel frames connected to each other using stiff steel plates. Further
information on the models is presented in Table 6-3 and Table 6-4.
Table 6-3: Specifications of superstructures at the model scale (Chen et al., 2010)
Plan dimensionsa (in) Elevationa,b (in)

Structure Floor Mass (lb)
L B L   H1, H 2 , H 3 
MS3F 7.9  7.9 7.9  (9.1, 7.9, 7.9) 8.75 (M2), 6.88 (M1)
MS1F_1 7.1  7.9 7.1  9.1 9.26 (M3)
a
L is the length in the shaking direction, and B is the length in the orthogonal direction
b
H n is the height of the nth story
Table 6-4: Specifications of foundations at the model scale (Chen et al., 2010)
Dimensionsa (in) Depth Foundation mass

Structure Foundation type
Lf  Bf  H f (in) (lb)
MS3F Basement below mat 8.4  8.4  7.9b 7.9 18.74d

MS1F_1 Isolated spread footings 11/8  11/8  7/8c 0.8 6.92e
a
L f is the length in shaking direction, H f is the height and B f is the length in orthogonal direction
b
Dimensions of the basement
c
Dimensions of each individual footing
d
Combined mass of basement and mat
e
Combined mass of all footings
The beams and columns in the models were constructed using 1/2”  1/2”  1/16” steel tubing. The
beams spanning in the direction orthogonal to the shaking direction were replaced by ‘rigid plates’ made
of steel having a rectangular cross section of 1/8” wide and 2” deep. These plates provided high stiffness
in the orthogonal-to-shaking direction so that the drift response in that direction was minimal. Rigid
connections were provided at the beam-column joints. Steel plates of known mass (denoted by M1, M2
and M3) were supported between the beams at each story of both structures to provide the required
142
reactive masses. Table 6-3 presents the values of these masses and Figure 6-8 shows their locations in the
superstructure. Columns of both the structures were rigidly connected to the foundations. The spread
footings of MS1F_1 were constructed of steel and placed just below the soil surface. The basement walls
of MS3F_iSSSI were made of 1/2” thick aluminum and the basement was completely embedded in the
soil. Table 6-3 and Table 6-4 present the configurations and important dimensions of the models and
Figure 6-8 provides their elevations.
a) MS1F_1 b) MS3F
Figure 6-7: Models used in Test 1 and Test 2
143
(a) MS3F (b) MS1F_1
Figure 6-8: Elevations of experimental models in the shaking direction, with dimensions in model
scale (Chen et al., 2010)
Nonlinearity in these structures was introduced by constructing ‘fuses’ at locations of potential

plastic hinges; see Figure 6-8 and Figure 6-9. This was achieved by reducing the cross-section area of the
member for a certain length, thereby decreasing the plastic moment at that location and promoting the
formation of a plastic hinge. Three types of fuses (shown in Figure 6-9) were designed and detailed with
different lengths in the structures to achieve the required design of the models.
The Phase I beams used in MS1F_1 (see panel b of Figure 6-8) for Test 1 responded in an
unexpected way, namely, plastic hinges were formed at the two ends of each of the 1-inch long fuses.
Details regarding this failure mechanism are presented in the Test 1 data report (Mason et al., 2010a).
This mode of failure was avoided in later tests by reducing the size of Fuse 3 to 0.5 in (Phase II beams)
from 1 in (Phase I beams). Model MS1F_1 of Test 2 was tested using both the Phase I and Phase II beams
to maintain consistency with Test 1. Section 6.3.4 presents the test sequence showing the ground motions
for which the Phase I and Phase II beams were used.
144
Figure 6-9: Designs of the three ‘fuses’ used in MS1F_1 and MS3F, dimensions in model scale
(Chen et al., 2010)
Chen et al.,(2010) also developed numerical models of these models to predict their behavior
during Test 1. Two-dimensional models were developed in the finite element program, OpenSees [e.g.,
Mazzoni et al. (2009)]. The nonlinear beams and columns were modeled using beam with hinges
elements, and foundation flexibility was modeled using the beam-on-nonlinear-Winkler-foundation
(BNWF) model available in OpenSees. Details regarding these simulations can be found in Chen et
al.,(2010). However this procedure cannot be used to simulate the SSSI effects that were expected during
Test 2, and no pre-test numerical simulations were carried out to predict these effects. Hammer tests were
conducted to calculate the fixed base time period of the experimental models. The dynamic properties of
the models calculated by OpenSees analyses and calculated by analysis of hammer test data are
summarized in Table 6-5.
Table 6-5: Prototype-scale properties of the designed experimental models calculated using the
numerical models in OpenSees and hammer tests (Chen et al., 2010)
OpenSees Analysis Hammer Test
Fixed-base Base shear

Drift ratio Flexible-base Fixed-base
Structure time period ratio
y time period time period
Vy W
T fix (sec) (fixed-base) Tssi (sec) T fix (sec)
(fixed-base)
MS3F 2.26, 0.67, 0.35 1.66 0.21 2.36, 0.71, 0.39 2.14, 0.66, 0.35
MS1F_1 1.06 1.72 0.59 1.14 0.93
145
6.3.3.3 Instrumentation
A total of 164 transducers were used in Test 2. One hundred and thirteen of these transducers
were used in the structure-foundation system, which included 14 linear displacement potentiometers, 32
uniaxial accelerometers, and 67 strain-gages. An additional 41 transducers were used to capture the soil
response, and three accelerometers were used to monitor the rack accelerations. Seven accelerometers
were used in unloaded foundations placed in the soil to track the attenuation of surface acceleration away
from the structures. Only details regarding the instrumentation of the structure-foundation system are
included in this report. Information regarding the instrumentation used in the soil, rack and the unloaded
foundations is presented in the Test 2 data report (Mason et al., 2010b).
Standard elongation uniaxial strain gages were used at the top and bottom ends of the columns to
measure the bending moments and axial loads. Triaxial shear rosettes were used at the mid-height of each
column in each story to calculate shear forces. High elongation strain gages were used on the top and
bottom flanges of the beam fuses.
Three accelerometers on each floor mass measured the floor acceleration in the three principal
directions: shaking direction (north-south); orthogonal-to-shaking direction (east-west); and vertical
direction. To check the (unintended) torsional response of the superstructure, a fourth accelerometer was
placed at the mid-span of one beam that measured horizontal acceleration of the floor eccentric to the
shaking direction. Horizontal and vertical accelerometers were placed on each of the spread footings of
MS1F_1 to measure the acceleration input to the superstructure and the rocking response of the
foundations. Nine accelerometers were placed on the basement-mat system of MS3F to capture its
translational and rotational response.
Displacement transducers were placed throughout the structure-foundation system and attached to
a specially designed rack, which also served as a reference frame for the displacement measurements. The
horizontal displacement of each floor was measured by a linear potentiometer. One horizontal
potentiometer was installed for each footing of the one-story structure to measure the transient and
residual sliding displacements in the shaking direction. Two horizontal potentiometers at the soil surface
level measured the translational response of the basement-mat foundation. Settlement and rocking
response were recorded by vertically-installed potentiometers on each of two footings of the one-story
structure and the basement of the three-story structure. Additional details and sketches of instrumentation
plans can be found in Mason et al. (2010b).
146
6.3.4 Input ground motions and test chronology
The main goal of Test 2 was to calculate SSSI effects by benchmarking the model responses
against those observed in Test 1, where the model structures were isolated from each other. To maintain
consistency between the tests, the ground motions used for Test1 were also used in Test 2, and were input
in the same order. Details regarding the selection of ground motions for Test 1 are presented in Bolisetti
(2010). Test 2 was performed over three days involving three spins (each spin consisting of a spin-up of
the centrifuge, shaking of the models at a desired g-level, and spin-down) with a total of 25 ground
motions and 10 step-waves. Step-waves are very low amplitude ground motions (0.01 g) that were
applied at the beginning and end of each spin to ‘heat-up’ the shaker, ensure that all the sensors were
functioning, and check the elastic time period of the soil profile. Of the 25 ground motions, 21 were
applied at 55g, and two each were applied at 40g and 20g. Only the first 21 ground motions are relevant
to this study. Table 6-6 below presents the final schedule of ground motions for Test2, excluding the step-
waves. The peak ground accelerations (PGA) observed at the container base and soil surface (measured at
free-field) are provided in the table at the prototype scale. Additional details regarding the test sequence
can be found in the Test 2 data report (Mason et al., 2010b).
147
Table 6-6: Test 2 ground motions and their sequence (only 55g ground motions included)
Spin #
--
Date GM Amp.
ID Earthquake Station PGAb
-- # factora
MS1F_1
beam design
1 JOS_L_1 92 Landers Joshua Tree 090 0.22 0.05
Spin 1
-- 2 TCU_L_1 99 Chi Chi TCU078 270 (E) 0.55 0.10
12-09-2009 3 RRS_1 94 Northridge Rinaldi R Sta 228 0.42 0.37
--
Phase I beam 4 PTS_1 87 Sup Hills Parachute T S 315 0.49 0.13
Fuse 3 5 SCS_L_1 94 Northridge Sylmar Conv Sta 052 0.20 0.18
1” length
6 LCN_1 92 Landers Lucerne 260 0.70 0.25
Spin 2
-- 8 TCU_L_2 99 Chi Chi TCU078 270 (E) 0.55 0.11
12-09-2009 9 RRS_2 94 Northridge Rinaldi R Sta 228 0.42 0.36
--
Phase II beam 10 PTS_2 87 Sup Hills Parachute T S 315 0.47 0.12
Fuse 3 11 SCS_L_2 94 Northridge Sylmar Conv Sta 052 0.20 0.19
0.5” length
12 LCN_2 92 Landers Lucerne 260 0.70 0.26
14 WVC_L 89 Loma Prieta Saratoga WV Col 270 0.80 0.28
Spin 3 15 SCS_H 94 Northridge Sylmar Conv Sta 052 1.00 0.57
--
16 JOS_H 92 Landers Joshua Tree 090 1.00 0.35
12-10-2009
-- 17 WPI 94 Northridge Newhall W Pico C046 0.80 0.42
Phase II beam
18 PRI_mod 95 Kobe Port Island-Mod-79m 5.10 0.64
Fuse 3
0.5” length 19 TCU_H 99 Chi Chi TCU078 270 (E) 1.00 0.30
20 WVC_H 89 Loma Prieta Saratoga WV Col 270 1.50 0.35
a
Factor applied to the input ground motion
b
Measured at the base of the laminar box
148
6.4 Test 3: An in-plane SSSI configuration on dense, dry sand
The results of Test 1 and Test 2 showed some kinematic SSI effects in the response of the models
but little inertial SSI and SSSI effects (Jones, 2013; Mason, 2011; Trombetta, 2013). In view of the lack
of SSI and SSSI in Test 1 and Test 2, the main goals of Test 3 were to 1) promote significant inertial SSI
and 2) generate interaction between adjacent structures (SSSI). Two structural models were designed for
Test 3: 1) a one-story SMRF model on footings, which is a modified version of MS1F_1 used in Test 1
and Test 2, and herein termed as MS1F_2 and 2) a two-story shear wall model on basemat herein termed
as MS2F. These models were designed to have significant inertial SSI, and to act as a ‘transmitter-
receiver’ couple (described in Section 6.4.1.1) when placed adjacent to each other. Test 3 included two
cases: 1) MS1F_2 placed far enough from other structures so that there is no SSSI, and 2) MS1F_2 and
MS2F placed adjacent to each other along the direction of shaking causing in-plane SSSI (iSSSI). The
former is intended to serve as a baseline case for MS1F_2 in the calculation of SSSI effects in the latter
case. To avoid confusion between the one-story models of the two cases the models in the baseline and
iSSSI cases are named as MS1F_2_Baseline and MS1F_2_iSSSI, respectively. Accordingly, the two-
story structure is named as MS2F_iSSSI. Figure 6-10 shows a photograph of the fully instrumented
models placed in the laminar box, along with the nomenclature of the models used in this section. Figure
6-11 and Figure 6-12 present sketches of the plan view and the elevation view, respectively, showing the
locations of these models in the laminar box.
MS2F_iSSSI
MS1F_2_Baseline
MS1F_2_iSSSI
149
Figure 6-11: Plan view of the experimental models in the laminar box during Test 3 (dimensions in
model scale) (Mason et al., 2010c)
Figure 6-12: Elevation view of the experimental models in the laminar box during Test 3
(dimensions in model scale) (Mason et al., 2010c)
150
6.4.1.1 The ‘transmitter-receiver’ configuration
The basic hypothesis of Test 3 was that placing a heavy, stiff structure against a light, flexible
structure having equal flexible-base time periods would amplify the SSSI effects due to resonance
between these structures. The heavier structure, termed here as the ‘transmitter’, was expected to transmit
energy by driving the lighter structure, which is termed here as the ‘receiver’. Due to the transmitted
energy the response of the receiver structure was expected to be amplified compared to a control
structure, which was identical to the receiver but was isolated from the other structures. During Test 3 the
two-story shear wall model, MS2F_iSSSI and the one-story SMRF model, MS1F_2_iSSSI, were placed
adjacent to each other and were expected to behave as transmitter and receiver structures, respectively, by
exhibiting in-plane SSSI. The baseline one-story model, MS1F_2_Baseline, was placed away from these
two models and the laminar box boundary, thereby serving as a control structure for observing SSSI
effects in the seismic response of MS1F_2_iSSSI.
6.4.2 Soil profile
The soil profile in Test 3 was identical to that in the previous tests (see Section 6.3.2 for the
description of Test 2 soil profile). Dry, dense Nevada sand was used at a relative density of 80%, and the
laminar box was filled to a depth of 21.1 in corresponding to a prototype depth of 96.7 ft.
The experimental models of Test 3 were designed to maximize the SSSI effects between the
transmitter and receiver structures, especially during strong shaking events. Unlike Test 1 and Test 2, no
specific prototype structures were used in this test to calculate the target design properties. The model
MS2F_iSSSI was designed to be stiff and heavy to exhibit significant inertial interaction, and also have a
flexible-base time period close to that of the soil profile. The design of MS1F_2_Baseline (and the
identical model, MS1F_2_iSSSI) was based on the one-story model of Test 1 and Test 2, but was
modified to match the flexible-base time period of MS2F_iSSSI. Brief descriptions of the designs of the
two-story model and the one-story model are presented in Sections 6.4.3.1 and 6.4.3.2 respectively.
Numerical models of both model structures were developed in OpenSees. However no pre-test
numerical simulations were performed on the complete experimental model to predict the SSSI effects.
Complete details of the design, construction and numerical modeling can be found in the Test 3 data
report (Mason et al., 2010c).
151
6.4.3.1 Design of two-story model, MS2F_iSSSI
The degree of inertial interaction in the seismic response of an elastic structure can be evaluated
using dimensionless parameters that are functions of the properties of the soil and structure. Some of
these parameters that have been widely used by previous researchers (Bielak, 1975; Stewart et al., 1999a;
Stewart et al., 1999b; Veletsos and Meek, 1974) are: 1) the soil-to-structure stiffness ratio (  ); 2)
structure-to-soil mass ratio (  ); 3) ratio of effective structural height to the effective foundation radius
( heff r ). Assuming that the model remains elastic, these parameters, including a few others, were used to
guide the design of MS2F_iSSSI. Table 6-7 lists all the predictive parameters and the corresponding
ranges that were targeted to make the model more susceptible to inertial interaction. Also indicated in the
table are the values that were achieved with the final design. Based on these predictive parameters and the
need to keep the model elastic throughout testing, a tall rocking shear wall structure was considered the
ideal choice for the transmitter structure. A prototype scale rocking time period of 0.6 sec was targeted,
which was the approximate natural time period of the soil profile. This was expected to induce resonance
between the structure and the soil profile and further increase inertial interaction. A macro-level
numerical model of MS2F_iSSSI was developed in OpenSees. Base flexibility was modeled using Beam
on Nonlinear Winkler Foundations (BNWF), which has been implemented in the program. Specific
details regarding the numerical modeling are provided in Mason et al. (2010c) and are not discussed here.
The dimensions of the model were selected so the predictive parameters were in the target ranges, and the
required rocking time period was achieved. The final model (in the stiffer direction) had a fixed-base 1st
mode time period of 0.13 sec and a flexible-base 1st mode time period of 0.60 sec at the prototype scale.
The dimensions of the designed model are illustrated in panels a and b of Figure 6-13. A photograph of
the fully constructed model is shown in panel c of the figure.
152
Table 6-7: Dimensionless parameters governing the design of the two-story shear wall model,
MS2F_iSSSI (Mason et al., 2011)
Parameter Definition Range to maximize SSI Values calculated for final design
1 Vs
>0.2 0.775
 Tfb  heff
heff >1.0
- 2.597
 r <4.0

mstructure

δ
 reff 2 heff
soil  >0.15 0.272
Df
- <0.25 0.224
reff
m fnd
- minimize 0.644
 mstructure
L
- minimize 0.337
2heff

Vs = Shear wave velocity of the soil
 ρsoil = Mass density of the soil
Tfb = Fixed-base time period of the structure
heff = Effective height of the structure calculated using the 1st flexible-base mode shape
mstructure = Total superstructure mass
mfnd = Total foundation mass
reff = Effective foundation radius (translation)
rθ = Effective foundation radius (rocking)
L = Foundation length in direction of translation
153
b. Elevation 2 (orthogonal-to-shaking c. Fully constructed
a. Elevation 1 (shaking direction)
direction) model
Figure 6-13: Final design of model MS2F_iSSSI (Mason et al., 2010c)
The entire model (walls, floor masses and foundation) was constructed using Aluminum 6061
plate stock of various thicknesses. The superstructure shear walls were continuous through the two floors
and were welded to a mat foundation. Each shear wall was 19.25 in tall, 6 in wide and 0.5 in thick. The
floor masses were welded to a connection plate that was bolted to the shear walls at the floor levels
(Figure 6-13). The story height was 9.125 in. The floor masses measured 6 in  5 in in plan (6 in along the
shaking direction) and were 0.5 in thick. Each floor mass weighed 2 lb, including the weight of the
connection plates. The mat foundation was square in plan, measured 9.875 in on side, and had a thickness
of 1 in. A foundation embedment of 1.25 in below the soil surface was assumed in the numerical
modeling, and this embedment was targeted during the experiment.
6.4.3.2 Design of the one-story models, MS1F_2_Baseline and MS1F_2_iSSSI
The basic design concept for the one-story model of Test 3 was adopted from the MS1F_1 used
in Tests 1 and 2. Design specifications and construction procedures of this model are provided in Section
6.3.3.2 as well as Mason et al. (2010a) and Bolisetti (2010). The target flexible-base time period of the
one-story model was 0.6 sec unlike the original model, which had a time period of 1 sec. To meet this
target, the story height and the floor mass of the original model were decreased until the required flexible-
base time period was achieved. To avoid the ‘shear-like’ failure observed during Test 1 [see Test 1 data
154
report by Mason et al. (2010a)], a shorter fuse length (3/8”) was used with the Fuse 3 design (see Figure
6-9).
Grade 6061 Aluminum was used for the construction of the floor mass and footings, thereby
making them lighter than the original model, which used steel. Lighter floor mass resulted in a shorter
time period, whereas lighter footings were expected to exacerbate SSI effects in the response. The floor
mass in the modified model was 3.86 lb whereas the original model had a floor mass of 9.06 lb (see Table
6-3 and Table 6-4 for specifications of the original model). The column height was also reduced to 7.125
inches from 9.125 inches in the original model. Eigen analysis of the modified model in OpenSees
showed a flexible-base 1st mode time period of 0.65 sec, which was considered sufficiently close to the
target time period of 0.60 sec. The fixed-base 1st mode time period was 0.471 sec, and the flexible-base
2nd mode time period was 0.151 sec (Mason et al., 2010c). Beam and column members in the models
were constructed using 1/2”  1/2”  1/16” A513 steel tubing as in the original model, MS1F_1. The
foundation dimensions, beam lengths and column fuse design of this model were not changed. Figure
6-14 presents the dimensions of the model in elevation and plan views, and Figure 6-15 presents
photographs of the fully constructed model equipped with strain gages.
a. Elevation view (shaking direction) b. Plan view (foundations omitted)

Figure 6-14: Dimensions of the one-story model in Test 3 (Mason et al., 2010c)
155
a. Elevation view (shaking direction) b. Plan view
Figure 6-15: Photographs of fully constructed one-story model in Test 3
A total of 176 transducers were used in Test 3 to instrument the experimental setup. One hundred
and thirty six of these transducers were used in the structure-foundation system, which included 26 linear
displacement potentiometers, 46 uniaxial accelerometers, and 64 strain-gages. Various sensors failed
during different spins and were either discarded or replaced. The instrumentation list for all of the spins
are documented in the Test 3 data report (Mason et al., 2010c), along with a chronological list of sensors
that failed throughout the experiment.
Standard elongation uniaxial strain gages were used at the top and bottom ends of the columns in
the one-story models to measure bending moments and axial loads. Triaxial shear rosettes were used at
the mid-height of each column to calculate shear forces. High elongation strain gages were used on the
top and bottom flanges of the beams in the fuses. The strains in model MS2F_iSSSI were not measured.
Two accelerometers on each floor mass measured the story acceleration in the shaking direction
(north-south) direction. One of these accelerometers was placed at the mid-point of a beam to measure
unintended story acceleration eccentric to the shaking direction. Two additional vertical accelerometers
were placed at the edges of each floor mass to monitor the rocking in the structures. Horizontal and
vertical accelerometers were placed on each of the spread footings of one-story models to measure the
acceleration input to the superstructure, and the rocking response of the footings. Six accelerometers were
placed on the mat foundation of the MS2F_iSSSI model that measured three translational accelerations,
acceleration in the torsional direction, and rocking.
156
horizontal displacement of each floor was measured by a linear potentiometer. One horizontal
potentiometer was installed for each footing of the one-story models to measure the transient and residual
sliding displacements in the shaking direction. One horizontal potentiometer at the soil surface level
measured the translational response of the mat foundation of model MS2F_iSSSI. Settlement and rocking
responses were recorded by vertically-installed potentiometers on the footings of the one-story models
and the mat foundation of model MS2F_iSSSI. An additional 40 transducers were used to measure the
acceleration and displacement responses of the soil and the container. Accelerometers were installed
through the depth of the soil profile and on the surface. An accelerometer placed at the soil surface
between the models MS2F_iSSSI and MS1F_Baseline was assumed to measure the free field surface
acceleration. Further details regarding the Test 3 instrumentation are documented by (Mason et al.,
2010c).
The ground motions used in Test 3 were identical to those used in the previous tests. A total of
eight spins were conducted to perform the ground motion simulations and the cone penetration tests
(CPT). The CPTs were performed during the first and last spins to evaluate in-flight soil properties, and
ground motions were applied during the other spins. During spin-down after the seventh spin, ground
motions were again applied at 42.8g to collect data at a different centrifuge scaling factor. Various
problems occurred in a few of these spins and the centrifuge had to be spun down before applying the
scheduled number of ground motions (Mason et al., 2010c).
A total of 40 shaking events were completed through these eight spins, which included 13 step
waves at 55g, two step waves at 42.8g, 23 earthquake ground motions at 55g, and two earthquake ground
motions at 42.8g. After spin 5 it was determined that the beams of the one-story model had accumulated
significant plastic strains and they were replaced with a new set of beams for the next spin. A list of
ground motions completed during each spin is presented in Table 6-8, excluding step waves and the 42.8g
earthquake ground motions. Only the spins consisting of earthquake ground motions are included. A
complete list of ground motions and a detailed test chronology is presented in the Test 3 data report
(Mason et al., 2010c).
157
Table 6-8: Test 3 ground motions and their sequence (55g earthquake ground motions only)
(Mason et al., 2010c)
Spin # GM Ground Motion Amp.
Earthquake Station PGAb
Date # ID Factora
Spin 2
7-12-2010
Spin 3 2 JOS_L_2 92 Landers Joshua Tree 090 0.22 0.06
7-15-2010 3 TCU_L_1 99 Chi Chi TCU078 270 (E) 0.55 0.13
Spin 4
7-16-2010
6 TCU_L_2 99 Chi Chi TCU078 270 (E) 0.55 0.12
7 RRS 94 Northridge Rinaldi R Sta 228 0.42 0.40
8 PTS 87 Sup Hills Parachute T S 315 0.49 0.10
Spin 5 9 SCS_L_1 94 Northridge Sylmar Conv Sta 052 0.20 0.15
7-16-2010 10 LCN_1 92 Landers Lucerne 260 0.70 0.26
12 SCS_H_1 94 Northridge Sylmar Conv Sta 052 1.00 0.58
13 PRI_H_1 95 Kobe Port Island-Mod-79m 5.10 0.64
Damaged one-story model beams replaced with new beams
17 WPI 94 Northridge Newhall W Pico C 046 0.80 0.42
18 TCU_H 99 Chi Chi TCU078 270 (E) 1.00 0.33
Spin 7
7-21-2010
20 SCS_H_2 94 Northridge Sylmar Conv Sta 052 1.00 0.60
21 PRI_2 95 Kobe Port Island-Mod-79m 5.10 0.60
22 SCS_H_3 94 Northridge Newhall W Pico C 046 1.00 0.63
a
Factor applied to the input ground motion
b
Measured at the base of the laminar box
158
6.5 Test 4: Anti-plane, and combined in-plane anti-plane SSSI configurations on dense, dry sand
After Test 3, which examined the effects of in-plane SSSI (or iSSSI) between two structures
arranged in a line parallel to the direction of shaking, the preliminary goal for Test 4 was to simulate a
‘complete city block’ by including five structures on dry, dense sand. A series of numerical SSSI
simulations were performed in the program SASSI (Lysmer et al., 1999) considering various
combinations of buildings, two of them with five buildings clustered together. These arrangements
included the building models used in the prior tests, as well as their modified versions such as 1) one-
story model in Test 3 on mat foundations instead of spread footings (MS1F_mat), and 2) the three-story
model with a reduced (shallower) basement (MS3F_reduced). Numerical simulations were performed at
the prototype scale for eight arrangements (CBL-1 to CBL-8) of building models (city blocks), with two
of the arrangements consisting of five buildings clustered together. The SASSI numerical models of the
individual structures are presented in Figure 6-16, and the numerical models of the arrangements are
presented in Figure 6-17 and Figure 6-18. Global responses of the structures were calculated for each
arrangement and compared to examine the effect of the surrounding buildings. Results from the larger
arrangements, CBL-1 and CBL-2 showed some SSSI effects, but these arrangements were too
complicated to draw definitive conclusions. Results from the smaller arrangements, CBL-3, CBL-4 and
CBL-5, also showed some SSSI effects and prompted the need to first examine smaller arrangements with
two or three buildings before considering the larger city blocks. This experience revealed the complicated
nature of SSSI in practice and guided the City Block team to adopt a bottom-up approach to investigating
this phenomenon. Test 4 was thus configured as a direct extension of Test 3, and examined two aspects of
SSSI: 1) the effects of anti-plane SSSI (or aSSSI) between two structures arranged in a line perpendicular
to the direction of shaking, and 2) the effects of SSSI in an ‘L’ shaped configuration achieved by
combining the in-plane and anti-plane configurations (herein referred to as cSSSI).
159
a. MS1F_2 b. MS1F_mat c. MS2F d. MS3F_reduced
Figure 6-16: SASSI numerical models of individual buildings considered during the planning of
Test 4 (arrows indicate direction of shaking)
a. CBL-1 b. CBL-2
Figure 6-17: SASSI numerical models of the building arrangements, CBL-1 and CBL-2, considered
during the planning phase of Test 4 (arrows indicate direction of shaking)
160
c. CBL-3 d. CBL-4 e. CBL-5
c. CBL-3 d. CBL-4 e. CBL-5

Figure 6-18: Numerical models of building arrangements, CBL-3 to CBL-5, considered during the
planning phase of Test 4 (arrows indicate direction of shaking)
To maintain continuity with Test 3, the Test 4 models were designed to be as similar as possible
to the Test 3 models (see Section 6.5.3.1 for the changes in model design from Test 3). No new models
were designed for Test 4. The first configuration examining anti-plane SSSI consisted of two models
placed at the north end of the laminar box arranged in a line perpendicular to the shaking direction (see
Figure 6-19 and Figure 6-20): a one-story SMRF model and a two-story shear wall model, which are
herein termed as MS1F_2_aSSSI, and MS2F_aSSSI respectively. The second configuration was placed at
the south end of the box, and consisted of a one-story SMRF model with two adjacent two-story shear-
wall models. This resulted in an ‘L’ shaped configuration with the one-story model at its corner (see
Figure 6-19 and Figure 6-20). The one-story model of the second configuration is termed herein as
MS1F_2_cSSSI, and the two-story models are termed as MS2F_ciSSSI and MS1F_caSSSI. Figure 6-19
presents a photograph of the Test 4 experimental models when fully instrumented and arranged in the
161
laminar box. Figure 6-20 presents sketches showing the exact locations of these models in the laminar
box.
MS2F_caSSSI
MS2F_aSSSI MS2F_ciSSSI
MS1F_2_aSSSI
MS1F_2_cSSSI
162
a. Plan view
b. Elevation view
Figure 6-20: Locations of the models in the laminar box during Test 4 (dimensions in model scale)
(Trombetta et al., 2011)
163
Seminal analytical studies that examined SSSI (Luco and Contesse, 1973; Wong and Trifunac,
1975) dealt with anti-plane configurations. Specifically, they examined the interaction effects between
infinitely long elastic shear wall structures with semi-circular foundations subjected to SH-waves
exhibiting particle motions parallel to the length of the shear walls. Although the structures used in Test 4
are different from these analytical counterparts, it was expected that the results from the first portion of
this test examining anti-plane SSSI serve to corroborate analytical observations. A comparison of the
responses of this configuration in Test 4 with those of the in-plane configuration in Test 3 was expected to
provide insight into the relative effects of iSSSI and aSSSI.
Observations made by Bray (2011) during earthquake reconnaissance visits suggest that corner
buildings in dense urban areas undergo the most severe damage when struck by strong earthquakes. The
second portion of Test 4 therefore attempts to simulate SSSI effects in a corner building of a City Block
by superposing the in-plane and anti-plane configurations. This was also a progression towards the
overarching goal of the project, which is to simulate a City Block with a two-dimensional array of
structures clustered together.
6.5.2 Soil profile
Since Test 4 was an extension of Test 3, the soil profile in Test 4 was identical to that of Test 3.
Dense, dry Nevada sand was used at a relative density of 80%, and the laminar box was filled to a depth
of 21.1 in, which corresponds to a prototype depth of 96.7 ft. Details of the soil material are presented in
Section 6.3.2, as well as the Test 4 data report (Trombetta et al., 2011).
Two types of structural models were used during Test 4: the one-story SMRF model with shallow
embedded footings and the two-story rocking shear wall model with a shallow embedded mat foundation.
These models, designed for Test 3, acted as receiver and transmitter structures respectively, and were
expected to exacerbate the in-plane SSSI effects. Therefore the same designs were used during Test 4.
The design procedures and specifications of these models are presented in Section 6.4.3 and in the Test 3
data report (Mason et al., 2010c). Only the modifications made to these designs for Test 4 are presented
below.
6.5.3.1 Differences in model designs from Test 3
The two-story model used in Test 4 is identical to that in Test 3 and is not presented here. Slight
differences in the one-story models arose due to unexpected problems (Trombetta et al., 2011) in the
164
manufacturing of the beams and columns. After their delivery from the machine shop, it was noticed that
the beams and columns had longer fuses than specified. While this difference was negligible for the
columns, the beam fuses were 1.5 times longer than planned (Figure 6-21). The OpenSees numerical
analysis of the model with the longer beam fuses showed a flexible-base 1st mode time period of 0.62 sec,
rather than the period of 0.61 sec in Test 3 model. The yield drift ratios and the base shear ratios were
similar to the Test 3 model. These differences were deemed insignificant and models were assumed to be
sufficiently similar to enable a comparison of the results of Test 3 and Test 4.
a. As-built Test 3 beam fuse b. As-built Test 4 beam fuse

Figure 6-21: Differences between the Test 3 and Test 4 beam fuses (Trombetta et al., 2011)
The instrumentation strategy for Test 4 was quite similar to Test 3 since the models were almost
identical. Since Test 4 involved five structures (unlike Test 3, which had three structures) the redundancy
in the instrumentation had to be reduced for this test owing to the limited number of channels available
for use. A total of 165 transducers were used during Test 4, which included uniaxial accelerometers,
linear displacement potentiometers, and regular and high elongation strain gages. One hundred and
twenty six of these transducers were used in the structure-foundation system, which consisted of 48 strain
gages, 49 accelerometers, and 29 linear displacement potentiometers.
Similar to Test 3, standard elongation uniaxial strain gages were used at the top and bottom ends
of the columns in the one-story models to measure bending moments and axial loads. The shear strains in
the columns of the one-story models were not measured in this test, and no strains were measured in the
two-story models.
The accelerometers in the structures were also placed in a similar manner to Test 3.
Accelerometers placed on the floor masses and beams measured the translational acceleration in the two
165
horizontal directions, as well as the torsional acceleration. Since the two-story models were expected to
respond mostly in a rocking mode, only the second story accelerations were measured. Three footings in
each one-story structure were instrumented with horizontal and vertical accelerometers to measure the
acceleration input to the superstructure, and the translational, vertical and rocking accelerations of the
individual footings. The mat foundations were also similarly instrumented to measure the acceleration
input to the superstructure, and the translational, torsional and rocking responses of the foundation.
horizontal displacement of each floor was measured by a linear potentiometer. Only the second floor
displacements were measured in the two-story models. One horizontal potentiometer was installed for
each footing of the one-story models to measure the transient and residual sliding displacements in the
shaking direction. Settlement and rocking responses were recorded by vertically-installed potentiometers
on the footings of the one-story models. Although translational displacements were measured for the
foundations of all the three two-story models, rocking was only measured in the MS2F_aSSSI mat
foundation.
The soil and laminar box were instrumented with 36 accelerometers and three displacement
potentiometers. Accelerometers were installed through the depth of the soil profile and on the surface. An
accelerometer placed at the soil surface exactly in between the two configurations was assumed to
measure the free field surface acceleration. Further details regarding the Test 4 instrumentation are
documented in the data report (Trombetta et al., 2011).
During the first attempt at testing of the experimental model, the set up was damaged due to an
accidental oil leak while the centrifuge was spinning up. A significant portion of the oil in the hydraulic
system of the shake table was spilled onto the laminar box and its surroundings. It was later noticed that
the oil had seeped into as much as 3 inches into the soil, and also under the foundations (Figure 6-22).
This led to the test being abandoned and the model was rebuilt, starting from pluviation of the soil. The
model that was tested during Test 4 was therefore the second centrifuge model.
166
Hydraulic oil on sand
Sand
Figure 6-22: Photograph of the first centrifuge model on the arm shortly after the accidental oil
leak
The rebuilt centrifuge model was spun for a total of five times and involved five CPTs, four
bender element tests, and 47 shaking events. The goals of the original test plan were achieved after 26
shaking events in a span of four spins. However the in-test data processing results showed that the one-
story model beams did not accumulate residual strains to the same extent as observed during Test 3 for
the same ground motions. Despite the elastic strains being close to the Test 3 values, the differences in the
residual strains of the beams were large enough that it was deemed to be due to an increase in the yield
strength of the steel material, since a different batch of steel tubes was used after Test 3 (Trombetta et al.,
2011). Due to these differences, the ground motions were repeated after increasing the floor mass of the
one-story models by 2.2 lb each, with an intention to further yield the beams and capture some nonlinear
response. This part of the test is not relevant to this thesis and is not presented here. Complete details
regarding this modification can be found in the Test 4 data report (Trombetta et al., 2011).
The first 26 shaking events included eight step waves and 18 earthquake ground motions at 55g.
The last 21 shaking events with increased one-story model floor masses included two step waves and 15
earthquake ground motions at 55g, and two step waves and two earthquake ground motions at 43g. Table
6-9 lists the 18 earthquake ground motions applied to the original model (with smaller floor masses) at
167
55g with their respective spin numbers. A detailed test chronology and a complete list of shaking events
is presented in the Test 4 data report (Trombetta et al., 2011).
Table 6-9: Test 4 ground motions and their sequence (only 55g ground motions applied to models
with original floor masses are included) (Trombetta et al., 2011)
Spin # GM Amp.
GM ID Earthquake Station PGAb
Date # Factora
Spin 1
2 TCU 99 Chi Chi TCU078 270 (E) 0.55 0.13
6-16-2011
3 RRS 94 Northridge Rinaldi R Sta 228 0.42 0.37
5 PTS 87 Sup Hills Parachute T S 315 0.49 0.12
6 SCS_L 94 Northridge Sylmar Conv Sta 052 0.20 0.19
7 LCN 92 Landers Lucerne 260 0.70 0.26
Spin 3
6-17-2011
9 SCS_H 94 Northridge Sylmar Conv Sta 052 1.00 0.58
10 PRI_mod 95 Kobe Port Island-Mod-79m 5.10 0.64
11 PRI_mod_55 95 Kobe Port Island-Mod-79m 7.00 0.58
Spin 4 15 WPI 94 Northridge Newhall WPico C046 0.80 0.39

6-20-2011 16 TCU_H 99 Chi Chi TCU078 270 (E) 1.00 0.31
a
Factor multiplied to the input ground motion
b
Peak acceleration measured at the base of the laminar box
168
SECTION 7
EXPERIMENTAL RESULTS AND COMPARISON WITH NUMERICAL
SIMULATIONS
7.1 Introduction
This section presents an analysis of the data recorded in the centrifuge experiments described in
Section 6, and a qualitative comparison of this data with responses calculated using numerical
simulations. The site response of the soil profile, soil-structure interaction (SSI) and structure-soil-
structure interaction (SSSI) effects in the building models are examined. Numerical site-response analyses
are performed using SHAKE, DEEPSOIL and LS-DYNA. Soil-structure interaction analyses are
performed using SASSI and LS-DYNA. The data analysis and numerical simulations are only performed
for a subset of the ground motions used during the experiments. Sections 7.2 presents the scope of
experimental data analysis and numerical simulations performed in this section. Section 7.3 describes the
numerical simulations of site response using the three programs, and compares the experimental and
numerical site responses. Section 7.4 presents a description of the soil-structure interaction simulations of
the experiments of Tests 3 and 4, and examines SSI and SSSI effects observed in the experiments and
numerical simulations.
7.2 Scope of experimental data analysis and numerical simulations
A large amount of data was recorded during the NEES City Block centrifuge experiments from
strain gages, accelerometers and displacement transducers placed on, in and around the experimental
models. These data are used to gain insight into the effects of SSI and SSSI on both global and local
structural behavior. From the City Block project team, Trombetta et al. (2013a) and Mason et al. (2013)
used data from Test 1 and Test 2 to study the effects of SSSI on the nonlinear response of superstructures
and foundations. Importantly, they showed that the footings of the one-story model showed a stiffer
moment-rotation response when placed against the basement of the three-story model, resulting in larger
demands in the columns of the superstructure. Chen et al. (2013) developed a seismic system-
identification framework to estimate SSI parameters from centrifuge test data, and successfully applied
this framework to data recorded during Test 1 and Test 2. In a separate study, Trombetta et al. (2013b)
utilized data from Tests 3 and 4 and examined SSSI effects on structural response through spectral
analysis. They also studied the effects of SSSI on footing response and hysteretic damping of the one-
story structure.
169
The studies presented in the following sections examine SSSI effects on the structures in Tests 3
and 4, which involved the two building models, MS1F_2 and MS2F, being tested in various
arrangements. These arrangements included, 1) in-plane SSSI (or iSSSI) 11 between MS1F_2 and MS2F
during Test 3, 2) anti-plane SSSI (or aSSSI)11 between MS1F_2 and MS2F during Test 4, and 3)
combined in-plane anti-plane SSSI or (cSSSI)11 between two MS2F models and MS1F_2 in an ‘L’ shaped
arrangement, as shown in Figure 7-1. MS1F_2 was also tested alone with no SSSI effects. This case of
MS1F_2 provides a baseline to examine SSSI effects in other configurations. However the baseline case
for MS2F was not established by testing. Numerical simulations are performed for all the cases
considered in Tests 3 and 4. Additionally, the baseline case for MS2F is simulated for completeness. The
different arrangements used in Tests 3 and 4, and the nomenclature used for the individual models in
these arrangements is summarized in Figure 7-1.
The soil and structural specimens used in the different experiments were built from different
batches of the same materials. There are minor differences in the construction of MS1F_2 across different
tests, as explained in Section 6.5.3.1. Sequential testing in the centrifuge results in variations in the
structural and soil properties over the course of each test as a consequence of nonlinear behavior and
accumulation of permanent strains in the soil and the structure. For these reasons, only a few of the
ground motions used in the experiments are selected for the comparison of structural and soil responses
across different tests. These ground motions are selected such that 1) the specifications of MS1F_2 used
in Tests 3 and 4 are almost identical, 2) they follow the same order in the different tests, thus maintaining
similar initial conditions, and 3) the structures are almost elastic and have little or no damage at the
beginning of the first ground motion in each test. The selected ground motions are listed in Figure 7-1,
and their response spectra and acceleration time series are presented in Figure 7-2 and Figure 7-3,
respectively.
11
In-plane interaction refers to the interaction between structures arranged along the axis of ground
shaking. Anti-plane interaction refers to the interaction between structure arranged in a direction perpendicular to
the direction of shaking. Combined in-plane anti-plane interaction refers to the interaction between structures placed
in an ‘L’ shaped arrangement as shown in Figure 6-19.
170
Chosen ground motions
1. JOS_L
2. LCN
3. SCS_H
4. PRI_H
Figure 7-1: Ground motions selected, and the nomenclature used for the experimental models
during data analysis and numerical simulations performed for Tests 3 and 4 [arrows indicate
direction of shaking; renderings adopted from Trombetta (2013) and modified]
171
Figure 7-2: Five percent damped acceleration response spectra of the ground motions selected for
data analysis and numerical simulations
Figure 7-3: Acceleration time series of the ground motions selected for data analysis and numerical
simulations
Centrifuge models do not exactly represent a real soil-structure system due to boundary effects
and box-shaker-container interaction (Ilankatharan, 2008; Ilankatharan and Kutter, 2010; Kutter, 1995),
as described in Section 6.2.2. Numerical simulation of a centrifuge experiment for validation purposes
should ideally address all of these limitations (Oberkampf and Trucano, 2008) to enable a robust
comparison of experimental and numerical results. Addressing these limitations in numerical simulations
172
is a challenging task and requires a good understanding of the underlying mechanisms (e.g., shaker-
container interaction) and the involved physical properties (e.g., the friction coefficient between the soil
and the walls of the container). Additionally numerical programs such as SHAKE and SASSI are limited
to linear analyses and therefore cannot explicitly capture nonlinear aspects of the soil-structure system
such as nonlinear site response, yielding of the beam fuses, sliding, gapping and uplift of the foundations,
and softening of the soil in the immediate vicinity of the foundation. Aside from these challenges, the
City Block project did not involve comprehensive soil testing, requiring the use of calibration and
empirical relations and databases from the literature to calculate soil properties for the numerical
simulations. Accordingly, direct numerical simulations of the centrifuge tests are not performed in the
study presented in this section. Instead, numerical models are built as follows:
1. The numerical models are built at the prototype scale and not at the centrifuge model scale.
2. The numerical models are built assuming almost complete dissipation of the radiated waves, thus
neglecting lateral boundary effects in the laminar box.
3. The numerical models are built with a fixed base, assuming no energy dissipation at the base of
the soil profile, as opposed to the energy dissipation that occurs at the base of the laminar box due
to shaker-container interaction (Section 6.2.2).
4. The foundation embedment in the numerical models is slightly smaller than those adopted in the
experiments, for both structures. The foundations in the numerical model are embedded to a depth
equal to the thickness of the foundation, simplifying the finite element meshing of the
foundations and the surrounding soil.
5. The numerical simulations are performed separately for each ground motion and not
consecutively as in the experiments, setting aside cumulative soil strain, cumulative structural
damage, and changing soil properties due to consolidation.
These modifications might lead to a considerable deviation of the numerical response of the
models from the experimental response, but they are required given the aforementioned limitations. It is
assumed that they do not considerably affect the interaction between the structures (i.e., the comparison of
structural responses when it is tested alone and in the presence of neighboring structures). As a
consequence of these modifications, it is more appropriate to make only qualitative comparisons between
the numerical and experimental results and not a direct comparison. The following sections describe the
numerical simulations of site response and SSI, and compare experimental and numerical results. Section
7.3 presents a description and results of site-response simulations, and Section 7.4 presents a description
and results of SSI and SSSI simulations.
173
7.3 Examination of site response
Site-response analysis is performed at the prototype scale using SHAKE (Schnabel et al., 1972)
and DEEPSOIL (Hashash et al., 2011) for the ground motions listed in Figure 7-1. Numerical modeling
in these programs includes layering the soil profile and specifying the material properties, input ground
motion, and analysis parameters. The ground motion recorded at the base of the laminar box during the
experiments is used as the input at the base of the soil profile model assuming rigid bedrock. Since the
same set of input ground motions were used in all the City Block experiments, site-response analyses are
only performed for the input ground motions recorded during Test 3. Slight differences exist between the
same input ground motions recorded in the different tests, but it is assumed that these differences have
only a minor effect on the strain-compatible shear moduli and damping ratios of the soil. The inter-test
and intra-test variability in the ground motions recorded at the base of the laminar box are investigated in
Section 7.3.4.
7.3.1 Calculation of the material properties
For Tests 1 through 4, the laminar box was filled with dry Nevada sand filled to a depth of 21.1
inches (equivalent to 29.5 m at the prototype scale using the scale factor N  55 ). This soil profile is
modeled at the prototype scale with 14 layers lying atop a bedrock halfspace. An initial damping ratio of
0.05, and a density of 16.66 kN/m3 is assumed for all the layers in the soil profile. The small-strain shear
modulus, Gmax , is evaluated for each layer using the following empirical relationship developed for sands
(Kramer, 1996; Seed and Idriss, 1970):

Gmax  1000K2,max ( m )0.5 (7-1)
In this equation, K 2,max is a parameter determined from the void ratio or relative density of the sand, and
 m' is the mean principal effective stress at the corresponding depth. The depth of a layer is taken as the
distance from the soil surface to the center of that layer. A value of 50 is chosen for K 2,max , based on a
calibration analysis12 performed using the JOS_L_1 motion of Test 4. The mean principal effective stress,
 m , is calculated as
12
The maximum shear moduli calculated using equation (7-1) are only estimates and not based on
laboratory testing. Due to a lack of reliable test data to calculate the small-strain shear modulus, the SHAKE
numerical model is calibrated to reproduce the response of the soil profile measured during the experiment. This is
done by modifying the parameter K 2,max (and therefore the input shear moduli) so that the response calculated at the
surface of the soil profile for the ground motion JOS_L_1 matches the surface free-field response recorded during
the experiment.
174
 1  2K0 
 m'   v'   (7-2)
 3 
where  v is the effective vertical pressure, and K 0 is the coefficient of earth pressure at rest. Since the
sand used in the experiment is completely dry,  v at a particular depth in the soil profile is calculated as
the vertical pressure at that point. A constant value of 0.35, which is suitable for dense sands, is chosen
for K0 following Craig (1997). Using these parameters, the mean principal effective stress and therefore
Gmax is calculated for each layer. These values are presented in Table 7-1. The individual layer
thicknesses are chosen such that a maximum frequency of 8 Hz can pass through the soil column
according to equation (4-1). The layer thicknesses are also presented in Table 7-1.
Table 7-1: Soil layer properties used for site-response analysis in Tests 3 and 4
Layer Thickness (m) Unit Weight (kN/m3) Gmax (MPa)
1 1.80 16.66 31.9
2 2.00 16.66 56.3
3 2.10 16.66 74.0
4 2.10 16.66 88.6
5 2.10 16.66 101.1
6 2.10 16.66 112.2
7 2.10 16.66 122.3
8 2.10 16.66 131.7
9 2.10 16.66 140.4
10 2.10 16.66 148.6
11 2.10 16.66 156.4
12 2.10 16.66 163.8
13 2.10 16.66 170.9
14 2.57 16.66 178.4
175
7.3.2 Site-response analysis in SHAKE
The following properties are input for each layer for a SHAKE analysis: 1) layer thickness, 2)
mass density,  , 3) small-strain shear modulus, G , 4) small-strain damping ratio,  , and 5) dynamic
material properties (modulus reduction and damping curves). The values presented in Table 7-1 are
directly input into SHAKE, along with a small-strain damping ratio of 5% for all the layers. The modulus
reduction and damping curves developed by Seed and Idriss (1970) for sands are used in the analyses.
The lower bound modulus reduction curve (panel a, Figure 7-4) and the upper bound damping ratio curve
(panel b, Figure 7-4), both of which are available in the SHAKE database, are employed. The bedrock is
modeled using a linear elastic material with no modulus reduction and a constant damping ratio of 1%.
The traditionally recommended value of 0.65 is used for the effective shear strain ratio. Analyses are
performed for the four ground motions of Figure 7-1. The free-field site responses are presented in
Section 7.3.4.
Figure 7-4: Modulus reduction and damping curves for sand calculated by Seed and Idriss (1970)
7.3.3 Site-response analysis in DEEPSOIL
DEEPSOIL performs time-domain nonlinear site-response analysis using the modified hyperbolic
model, or the Modified Kodner-Zelasko (MKZ) model (Matasovic, 1993), to simulate the shear behavior
of soil. This model is described in detail in Section 2.2.2.3.1. Each layer in the DEEPSOIL is defined by
same properties as in SHAKE. However DEEPSOIL uses the MKZ model instead of modulus reduction
and damping curves. The layer thickness, unit weight and the small strain shear modulus of the soil layers
are listed in Table 7-1. The model parameters that define the MKZ model are listed in Table 7-2. These
parameters are chosen such that the resulting modulus reduction and strain compatible damping curves
176
are almost equal to those specified in the SHAKE analysis, thus maintaining equivalence between the
material properties. The MKZ model parameters are calculated using a curve-fitting tool included in
DEEPSOIL that calculates the appropriate values for the parameters for a given pair of shear modulus and
damping curves. A detailed description of the calculation of the model parameters corresponding to a
given set of modulus reduction and damping curves is presented in Phillips and Hashash (2009) and the
DEEPSOIL user manual (Hashash et al., 2011). Additionally, a value of 0.001 is used for the maximum
strain increment for all ground motions. This parameter affects the time-step calculation in the Newmark-
 method (Hashash and Park, 2001; Hashash et al., 2011).
Table 7-2: DEEPSOIL curve parameters used for the Test 4 site-response analysis
Small-strain damping ratio1,  (%) 0.42
Reference strain2, r (%) 0.0378
Stress-strain curve parameter,  1.56
Stress-strain curve parameter, s 0.765

1
Pressure dependence of damping is ignored by setting the parameter d to 0
2
Pressure dependence of the stress-strain curve is ignored by setting the
parameter b to 0
7.3.4 Site-response analysis in LS-DYNA
Site-response analyses for the City Block soil profile are also performed using the nonlinear finite
element program, LS-DYNA, for the ground motions listed in Figure 7-1. Section 2.2.2.3.2 presents a
description of nonlinear site-response analysis using LS-DYNA. Only the details specific to the analyses
of the City Block soil profile are presented in this section. The layering of the soil profile in LS-DYNA is
identical to SHAKE and DEEPSOIL. Each of the 14 soil layers, with properties presented in Table 7-1, is
modeled using a single eight-noded solid element in LS-DYNA. All these elements are constrained to
move in pure lateral shear. The acceleration input is applied directly at the base of the soil column, which
is consistent with a rigid bedrock assumption beneath the soil profile. Figure 7-5 presents the LS-DYNA
model of the soil profile used for site-response analysis.
177
Figure 7-5: LS-DYNA finite element model for the site-response analysis of the City Block soil
profile
The soil material is modeled using the MAT_HYSTERETIC model (LSTC, 2013), described in
Section 2.2.2.3.2. This model simulates the shear behavior of soil with a piece-wise linear stress-strain
curve such as that shown in Figure 2-7. These stress-strain curves are calculated from the modulus
reduction curves using the following equation,
G
 ( )  ( )  Gmax  
(7-3)
Gmax
where  and  are the instantaneous shear stress, and shear strain, respectively, and G Gmax is the
modulus reduction ratio. The modulus reduction curve used for SHAKE is used for the calculation of the
stress-strain curves of all the layers.
7.3.5 Variability in the input ground motion and soil properties during the experiments
The primary goal of this section is the examination of SSSI effects, which involves comparison of
the responses of MS1F_2 and MS2F recorded in the various arrangements of Tests 3 and 4. Factors such
as variability in the soil properties in different tests and the variability in the input ground motions
contribute to the differences between the structural responses across the tests. Therefore, it is important to
examine these factors before making observations on the effects of SSSI on structural response in Section
178
7.4.4. Two factors are examined in this section: 1) the inter-test and intra-test variability in the input
ground motion, and 2) the inter-test and intra-test variability in site response, which identifies the
variability in soil properties.
Figure 7-6 presents the acceleration response spectra of the input ground motions achieved by the
centrifuge shaker during Tests 1 through 4 for nine ground motions, including the four ground motions
considered in this section. The figure shows significant inter-test variability in the input for four ground
motions: JOS_L, TCU_L, PTS and SCS_L. However the inputs achieved during Tests 3 and 4 are similar
for all the ground motions: even more so for the ground motions, JOS_L, LCN, SCS_H and PRI_H,
which are the ground motions selected for analysis in this section. The reasons for the inter-test variability
in the ground motions is not relevant to this discussion. However, to objectively examine the SSSI effects
during the experiment for these ground motions, the variability in inputs must be considered.
Figure 7-6: Acceleration response spectra of the considered input ground motions (recorded at the
base of the laminar box) in Tests 1 through 4
179
The ground motion JOS_L was input seven times at various instances during Test 3. Data from
six inputs are used to examine the variability in site response during the test: intra-test variability. Figure
7-7 presents the acceleration response spectra of the input at the base of the laminar box, and the spectral
amplifications of the free-field acceleration at the surface of the soil. Small differences are evident in the
input acceleration, but the variability in the amplifications is considerable. The intra-test variability in site
response, although not significant, shows that progressive changes in soil properties will result in a
change in site response and therefore adds to the variability in the acceleration input to the structure. This
variability must be considered in the analysis of data from a centrifuge test.
Figure 7-7: Acceleration response spectra of the input at the base of the laminar box (left) and
surface-free-field-to-base transfer functions (right) calculated for six applications of the JOS_L
ground motion during Test 3
7.3.6 Comparison of experimental and numerical site responses
The experimental and numerical site responses for the ground motions considered for data
analysis (see Figure 7-1) are presented in this section. Comparisons are made between the experimental
site responses and the numerical site responses calculated using SHAKE, DEEPSOIL and LS-DYNA.
Figure 7-8 presents the response spectra of the free-field surface accelerations for the ground motions
selected for detailed analysis in this section. The plots in the left column present spectra calculated from
experimental recordings in Tests 3 and 4; the plots in the right column present spectra calculated from
numerical simulations in SHAKE, DEEPSOIL and LS-DYNA. The figure shows differences between the
response spectra of each ground motion recorded in Tests 3 and 4. These differences are a result of the
variation in the input ground motion, as discussed earlier, and different mechanical properties for the soil
specimens used for Tests 3 and 4. However, these differences are smaller than those in the results of the
180
numerical simulations, which show that the nonlinear programs, DEEPSOIL and LS-DYNA, yield similar
results, but that these are significantly different from the equivalent-linear results produced by SHAKE. In
particular, SHAKE predicts larger spectral accelerations than the nonlinear programs for all ground
motions, especially at the site period (period of maximum spectral acceleration). This overprediction is a
consequence of the equivalent-linear method accounting for nonlinear behavior in an ‘averaged’ sense13.
Figure 7-8 also shows significant differences between the experimental and numerical results for
all ground motions, except the JOS_L ground motion. The results for the JOS_L ground motion match
better since the soil profile properties input to the numerical models are calculated through calibration
using data from this ground motion. For the other ground motions, the shapes of the experimental and
numerical response spectra are qualitatively similar, albeit having different magnitudes, especially at
periods smaller than the site period. In addition, the numerical simulations also reasonably predict the site
period for the JOS_L, LCN and PRI_H ground motions. Figure 7-8 shows that the site period is about 0.6
sec during the JOS_L ground motion, and about 1 sec during the LCN and PRI_H ground motions. The
site period is not clear from the surface response spectra for the SCS_H ground motion. The numerical
programs reasonably capture the elongation of the site period for increasing intensity of ground motions
(JOS_L < LCN < SCS_H < PRI_H). The primary reason for the large differences between the
experimental and numerical results is the lack of measured material properties. Whereas the numerical
programs reasonably predict the almost linear response during the low-intensity JOS_L ground motion,
the predictions of the measured responses for the more intense ground motions are poor, reinforcing the
need for estimates of nonlinear properties (or modulus reduction and damping curves) for validation of
soil models and computer codes.
13
The variability in the results of the numerical programs was examined in more detail in Section 4.
181
Figure 7-8: Free-field surface acceleration response spectra calculated from experimental data and
numerical simulations for the considered ground motions in Tests 3 and 4
7.4 Examination of SSI and SSSI effects
The experimental data is analyzed in this section to examine the soil-structure interaction and
structure-soil-structure interaction effects observed in Tests 3 and 4. Numerical simulations of the
182
centrifuge tests are also performed at the prototype scale. Linear analyses are performed using SASSI and
nonlinear analyses are performed using LS-DYNA, for all structural arrangements for the ground motions
listed in Figure 7-1. Numerical modeling in these programs is described in the next sections, and
qualitative comparisons between the experimental and numerical results are made in Section 7.4.4.
7.4.1 Modeling in SASSI
SASSI modeling for the simulation of Tests 3 and 4 is described in this section. A more detailed
description of SSI analysis in SASSI is presented in Section 3.3.1 as well as in Bolisetti (2010). The
SASSI numerical model comprises 1) the soil profile model, 2) the structural model, and 3) the excavated
soil model for embedded foundations. Since SASSI performs a linear analysis, strain-compatible
properties are used for the soil. The properties are usually calculated using SHAKE, which, apart from
equivalent-linear site response, outputs strain-compatible shear moduli and damping ratios of the soil
layers for each ground motion. The SASSI soil profile is therefore different for each ground motion input.
Although the strain-compatible properties from the SHAKE analyses presented in Section 7.3.2 could be
used in this section, the properties are calculated through separate SHAKE analyses because the soil
profile developed for the SSI analyses of this section is discretized with 29 layers (as opposed to 14 for
the site-response analyses) for reasons given in Section 5.4.4. However, the layer properties at each depth
remain the same as in Table 7-1. The small-strain shear modulus and damping ratio through the depth of
the soil column for the more finely discretized soil profile are presented in Figure 7-9. This figure also
includes the strain-compatible shear modulus and damping ratio profiles calculated for each ground
motion using SHAKE and input in the SASSI analyses of this section. A Poisson’s ratio of 0.3 (Kulhawy
and Mayne, 1990; Mason, 2014) is used for the soil in the analyses described in this section.
183
Figure 7-9: Small-strain and strain-compatible shear moduli and damping ratios used for SSI
analyses using SASSI for simulation of Tests 3 and 4
SASSI only models horizontally infinite soil layers, and therefore the boundary effects from the
laminar box must be neglected in the numerical analyses. The earthquake input is modeled as a vertically
propagating shear wave-field to be consistent with the experiments. Additionally, the base of the soil
profile is fixed, consistent with the rigid bedrock assumption. The structure-foundation system is modeled
using two-dimensional or three-dimensional finite elements in SASSI. For the numerical simulation of the
City Block experiments, three-dimensional finite element models of the experimental model buildings are
developed at the prototype scale. The model parameters are scaled to the prototype level by multiplication
using the scale factors presented in Table 6-1. Each beam and column of MS1F_2 is modeled using a
beam element with the corresponding material and cross-section properties. The coordinates of the nodes
of the beam elements are taken as the centers of the cross-sections of the beams and columns in the
experimental model. The floor mass is simulated by assigning appropriate lumped masses on the floor
beams. MS2F is modeled using two beam elements as a two-story lumped mass structure, with each beam
modeled as the aggregate of two shear walls. The masses of the shear walls and the floors are lumped at
the floor levels. Miscellaneous additions to the floor mass, such as bolts, instrumentation wires, etc., are
included in the lumped masses. The SASSI numerical models of MS1F_2 and MS2F are presented in
Figure 7-10. The lumped masses in this figure are indicated as red dots in the beam elements, and the
184
ground surface is indicated in brown. Since SASSI does not allow for explicit modeling of the beam
cross-sections, the beam and column fuses cannot be modeled explicitly. The fuses are therefore modeled
in an averaged sense by decreasing the moment of inertia of the beam element cross-section through a
calibration of the fixed-base numerical models to the data from hammer tests on the experimental models.
This calibration process is described in Section 7.4.3. Details of the numerical models of MS1F_2 and
MS2F are presented in Table 7-4 and Table 7-5, respectively.
The footings of MS1F_2 and the mat foundation of MS2F are modeled using eight-noded solid
elements. The foundations are embedded to a depth equal to their thickness, namely, the top of the
foundation is at the same elevation as the ground surface. This embedment depth is slightly smaller than
that used in the experiments for both structures. The column-foundation connections of both buildings are
modeled using stiff beam elements as shown in Figure 7-10. The stiff beams are modeled with a
negligible mass and a very large stiffness compared to the rest of the structure. These beams ensure that
the column-foundation connections are almost rigid. The excavated soil, for the embedded footings and
the basemat, is modeled using eight-noded solid elements and has the same dimensions as the
corresponding foundations. The solid elements of the excavated soil have the same material properties as
the corresponding layers in the soil profile, as required by SASSI.
SASSI analyses are performed for the arrangements listed in Figure 7-1. The numerical models
for these arrangements are built by combining the numerical models of the individual buildings, and
changing their co-ordinates according to their new position in the box. A 0.5m gap is maintained between
the foundations of the adjacent buildings, which is equal to the gap used in the experiment, at the
prototype scale. The numerical models for these building arrangements are presented in Figure 7-10. Each
of the building arrangements is subjected to the ground motions of Figure 7-1.
185
a. MS1F_2 b. MS2F
Figure 7-10: SASSI numerical models of the baseline models of MS1F_2 and MS2F
Figure 7-11: SASSI numerical model of the in-plane SSSI arrangement in Test 3
186
Figure 7-12: SASSI numerical model of the anti-plane SSSI arrangement in Test 4
Figure 7-13: SASSI numerical model of the in-plane SSSI arrangement in Test 3
187
Table 7-3: Material properties used in the SASSI analyses
Property Steel Aluminum
Elastic modulus (GPa) 200 69
Poisson’s ratio 0.3 0.3
Unit weight (kN/m3) 76.84 26.48
Damping ratio (%) 2 2
Table 7-4: Description of the numerical model of MS1F_2a (dimensions in prototype scale)
Nodal masses on each

Moment of Inertia (m4)
Plan dimensionsb beamc (kg)
Height (m)
L  B (m)
Center End Beamsc Columns
9.96  11.00 12.76 145578 1820 0.0106 0.0141
Footing dimensions (m) 4.37  4.37
Footing thickness (m) 0.875

a
The area of the cross section, shear area, and torsional inertia are the same for all elements, and are
calculated from the cross section of the experimental model
b
L is along the direction of shaking; see Figure 7-10
c
Only for the beams parallel to the direction of shaking
188
Table 7-5: Description of the numerical model of MS2Fa (dimensions in prototype scale)
Moment of Inertiaa,c (m4)

b
Floor height (m)
Y-bending X-bending
12.58 70.63 0.51
Mat dimensions (m) 13.80  13.80
Mat thickness (m) 1.34

a
The area of the cross section, shear area, and torsional inertia are the same for all
elements, and are calculated from the cross section of the experimental model
b
Story heights are equal
c
X is the direction of shaking; Z is the vertical direction; see Figure 7-10
7.4.2 Modeling in LS-DYNA
Nonlinear soil-structure interaction analysis for Tests 3 and 4 using LS-DYNA is described in this
section. General modeling procedures for soil-structure interaction analysis in LS-DYNA are presented in
Section 3.3.2. Only the details corresponding to the simulations of Tests 3 and 4 are presented here.
The LS-DYNA numerical analyses are performed using the direct method (see Section 3.2 for the
description of the direct method and the domain reduction method), which involves modeling a large soil
domain to simulate an infinite soil halfspace. The soil domain in the current analyses is modeled using
eight-noded solid elements. The linear soil properties (small-strain shear modulus, small-strain damping,
unit weight and Poisson’s ratio) for these analyses are the same as used in SHAKE and presented in
Figure 7-9. The soil layer thicknesses are the same as in SASSI. Each soil layer in SASSI corresponds to
one solid element layer in LS-DYNA. The soil material is modeled using the MAT_HYSTERETIC
material with the same material properties as used for the site-response analyses in Section 7.3.4. For the
simulations in this section, a soil domain of size 219m  118m in plan (219m in the shaking direction, X)
and 29.47m deep is built, which is about 15 times the width of the basemat of MS2F. This soil domain is
equal in size to that used in the verification analyses of Section 5.2. The lateral boundaries for this soil
domain are modeled as described in Section 5.2.3. Figure 7-14 presents a sample LS-DYNA model used
for the analysis of the combined SSSI arrangement in Test 4. The large dimensions of the soil domain
relative to the plan dimensions of the structures are apparent in this figure. The soil layers, indicated by
different colors, are also seen in the figure. The input ground motion for all the analyses is applied
189
directly at the base as an acceleration input, which is consistent with the fixed-base assumption for the
soil profile.
Figure 7-14: LS-DYNA finite element model for soil-structure interaction analysis of the cSSSI
arrangement during Test 4
The building models MS1F_2 and MS2F are modeled as described previously for the SASSI
analyses: Table 7-4 for MS1F_2, and Table 7-5 for MS2F. Nonlinear behavior in MS1F_2 can be
simulated in LS-DYNA. Although the beam fuses in the experimental models can be modeled explicitly
in LS-DYNA, preliminary analyses using one such model (in which the fuses were modeled using shell
elements; see Figure 7-15) required very long runtimes. Since only qualitative comparisons are made
between the experimental and numerical results, the beam fuses are not modeled explicitly and lumped
plastic hinges are used instead. Nonlinearity in the beams is modeled using the material model,
*MAT_SEISMIC_BEAM, available in the LS-DYNA material database (LSTC, 2013). This material
model simulates nonlinear behavior at one of the nodes of the beam element in the form of a plastic hinge,
which follows a plastic moment-rotation curve input by the user. The plastic moment-rotation curve is
calculated from a pseudo-static analysis of one of the beam fuses shown in Figure 7-15. To account for
the interaction between the shear force in the fuse and the moment-rotation behavior, a shear force equal
190
to a quarter of the roof mass is applied on the fuse during the pseudo-static analysis. The effects of axial
force, and changes in the shear force during the earthquake on the moment-rotation behavior are
neglected. Additionally, the fuses in the columns are not modeled since no yielding of the column fuses
was observed in the experiments.
Figure 7-15: Preliminary model of MS1F_2 in LS-DYNA with explicit modeling of the beam fuses
using shell elements (left) and the close-up of one of the beam fuses (right)
The footings of MS1F_2 and the basemat of MS2F are modeled using solid elements. Both
foundations are embedded such that the top of the foundation is at grade. The dimensions of the
foundations are the same as used in SASSI, and are presented in Table 7-4 and Table 7-5. Both the
footings and the basemat are modeled using the rigid material model available in LS-DYNA, which uses
rigid body dynamics to simulate a completely rigid material. Preliminary analyses using an elastic
material for the foundations revealed that the solid elements of the footing or those of the basemat (for the
analysis case of stand-alone MS2F model) control the analysis time-step. Using the rigid material model
avoids including the footing and basemat elements in the time-step calculations, resulting in an increase in
the time step and a reduction in the analysis time. For these analyses, the use of a rigid material model for
the foundations reduced the analysis time by about two-thirds.
The LS-DYNA models used for the various analyses cases are presented in Figure 7-16 (baseline
cases of MS1F_2 and MS2F), Figure 7-17 (in-plane SSSI case), Figure 7-18 (anti-plane SSSI case), and
Figure 7-19 (combined SSSI case). The figures only present the superstructures, foundations, and soil
elements close to the foundations. The rest of the soil domain is shown in Figure 7-14 and is not shown in
these figures for clarity. The figures also show that a much finer mesh is used for the soil close to the
191
foundations as compared with the rest of the soil domain. This is to accommodate the large soil strains
that are expected near the foundations. The interface between the finer mesh and the coarser mesh is
modeled using the contact model, *CONTACT_TIED_SURFACE_TO_SURFACE. This model
simulates a fully tied contact at the interface and does not allow separation. The foundation-soil interface
for both the footings and the basemat can modeled as either 1) a ‘tied’ interface with no loss of contact
between the soil and the foundation, or 2) a more realistic interface that allows for loss of contact between
the soil and the foundation, resulting in sliding, gapping and uplift. The analyses in this section are
performed for both the interface conditions. The tied interface is modeled using the *CONSTRAINED_
EXTRA_NODES option, which constrains the soil nodes at the interface to move along with the rigid
footing or basemat. The second type of interface is modeled using the contact model,
*CONTACT_AUTOMATIC_SURFACE_TO_ SURFACE, that allows sliding and separation. This
contact model adopts the Coulomb friction model to simulate sliding, and a friction coefficient of 0.84 is
used for the analyses presented in this section. This friction coefficient is calculated as tan  , where  is
the friction angle of Nevada sand, which is estimated to be between 37 and 42 degrees (Mason et al.,
2011). A friction angle of 40 degrees is used for the analyses presented in this section.
Structural damping is modeled using the *DAMPING_FREQUENCY_RANGE option, which
simulates an approximately frequency-independent damping in a given frequency range. This damping
option is also used to simulate the low-strain damping in the soil. A 2% damping ratio is used for the
structures, and a 0.4% damping is used for the soil in the frequencies between 1 Hz and 25 Hz. The 0.4%
ratio for the soil is the same as used in DEEPSOIL, as described in Section 7.3.3.
Figure 7-16: Close-up isometric views of the SSI models of MS1F_2 (left) and MS2F (right) in LS-
DYNA (soil elements far away from the structure are not shown for clarity)
192
Figure 7-17: Close-up isometric view of the SSI model for the in-plane SSSI case of Test 3 in LS-
DYNA (soil elements far away from the structures are not shown for clarity)
Figure 7-18: Close-up isometric view of the SSI model for the anti-plane SSSI case of Test 4 in LS-
193
Figure 7-19: Close-up isometric view of the SSI model for the combined SSSI case of Test 4 in LS-
7.4.3 Calibration of fixed-base models
Prior to performing the soil-structure interaction analyses, the structural models of MS1F_2 and
MS2F in SASSI and LS-DYNA are calibrated to achieve their target dynamic properties, which are the
fixed-base time periods of the experimental model structures in the prototype scale. Calibration of the
numerical models is required to account for the miscellaneous masses such as nuts, bolts and
instrumentation, and also the reduction in structural stiffness due to the beam and column fuses. These
factors cannot be directly accounted for in the numerical models. Therefore the moment of inertia of the
beam elements and the lumped masses are adjusted until the target time period is achieved, and these
adjusted values are used in the SSI analysis.
The SASSI fixed-base model of each experimental model is built by modeling the superstructure
on a very stiff soil profile. This soil profile is modeled using very thin layers having a shear modulus
much greater than the shear modulus of steel used in the superstructure model. This ensures that the soil
profile is very stiff compared to the structure, imitating an almost fixed-base condition. The soil profile
consists of three horizontal layers, each with a thickness of 0.003 m. A shear wave velocity of 3.06 107
m/s, and a unit weight of 0.48 kN/m3 are specified for each layer including the bedrock. Seismic
excitation is applied to this model at the base of the soil profile. Excitation for the LS-DYNA model is
applied by specifying an acceleration input at the base of the structure. The time period of the structure is
194
calculated from the acceleration amplification spectrum at the roof of the structure. The calibrated values
of moments of inertia are presented in Table 7-4 and Table 7-5. The fixed-base time periods calculated
from the SASSI and LS-DYNA analyses are presented in Table 7-6, together with the time periods of the
individual structures calculated from hammer tests.
Table 7-6: Comparison of the prototype scale fixed-base time periods calculated from SASSI and
hammer tests for all the models
Time period(s) (sec)

Superstructure
SASSI LS-DYNA Hammer test
MS1F_2 0.49 0.48 0.47
MS2F 0.13 0.14 0.13
7.4.4 Comparison of experimental and numerical results
The experimental and numerical responses of MS1F_2 and MS2F are presented in this section for
the baseline cases and the SSSI configurations illustrated in Figure 7-1. SSSI effects are examined by
comparing the structural response in the baseline case with the responses in various SSSI configurations.
Structural response is measured using acceleration response spectra at the roof of the structures.
Figure 7-20 and Figure 7-21 present the experimental and numerical responses of MS1F_2 for the
baseline case and the SSSI cases illustrated in Figure 7-1. Figure 7-20 presents the responses for the
JOS_L and LCN ground motions. Figure 7-21 presents the responses for the SCS_H and PRI_H ground
motions. Similarly, the experimental and numerical responses of MS2F for the baseline and SSSI cases
are presented in Figure 7-30 (JOS_L and LCN ground motions) and Figure 7-31 (SCS_H and PRI_H
ground motions). The numerical responses of MS2F also include the baseline case. The numerical
responses for each case are calculated using SASSI, LS-DYNA with a tied soil-foundation interface, and
LS-DYNA with separation allowed at the soil-foundation interface. Observations made from the
experimental and numerical results regarding SSSI effects on the response of each building model are
presented in Sections 7.4.4.1 and 7.4.4.2. Qualitative comparisons between the experimental and
numerical results with regards to SSSI are also presented.
195
7.4.4.1 MS1F_2
After the completion of Test 4, and during the dismantling of the experimental setup, it was
noticed that MS1F_2 in the anti-plane SSSI arrangement (aSSSI model of MS1F_2) was inadequately
assembled. Consequently, the experimental results presented in Figure 7-20 and Figure 7-21 (indicated in
light blue) show that the response of this structure is erratic and very different from the responses of
MS1F_2 in other SSSI arrangements. The MS1F_2 response in the aSSSI arrangement is therefore
ignored in this study.
The experimental results in Figure 7-20 and Figure 7-21 show that the response of MS1F_2 for a
given ground motion is similar in all the arrangements illustrated in Figure 7-1. The one exception is the
peak spectral acceleration in the combined SSSI case for the JOS_L ground motion. However this
reduction in response is not seen in the experimental results from the other ground motions, and does not
indicate a general trend. Small differences are evident in the experimental responses for the more intense
ground motions, SCS_H and PRI_H, but these could be attributed to changes in material properties and
variability in input ground motion, as described in Section 7.3.5. The global response of MS1F_2 in these
experiments is not affected by SSSI. The same observation is made from examination of the numerical
results calculated using SASSI and LS-DYNA. Figure 7-20 and Figure 7-21 show that the variability in
the numerical responses of MS1F_2 for a given ground motion for different arrangements is minor. Given
that the variabilities in input ground motion and the material properties can be ruled out for the numerical
simulations, these minor differences can be attributed to the interaction between the structures. The
interaction between the structures that are in the arrangements illustrated in Figure 7-1 do not affect the
global response of MS1F_2 sufficiently to warrant consideration of SSSI effects.
The results show that the numerical predictions from SASSI and LS-DYNA agree reasonably
well with the experimental results, with regards to the magnitude of SSSI effects in MS1F_2. However
the numerical and experimental responses are quite different for all ground motions, especially in terms of
the spectral accelerations. Although some differences are expected for the reasons given in Section 7.2,
the differences seen in the results are particularly large and warrant investigation. A similar magnitude of
differences between experimental and numerical results is seen in the results calculated by another City
Block researcher (Jones, 2013) who performed numerical simulations of Tests 3 and 4 using the program
FLAC (Itasca Consulting Group, 2005). Jones (2013) found that the peak spectral accelerations calculated
using the numerical programs were much smaller than those from the experiments. To explain the
differences between the experimental and numerical results, Jones (2013) conducted a series of sensitivity
analyses in FLAC to understand the effect of parameters such as the shear-wave velocity profile of the
soil, fixed-base time period of the structure, and modulus reduction and damping and curves, on the
response of the structure. These sensitivity analyses did not adequately explain the under-prediction of the
196
numerical simulations and Jones (2013) speculated that the differences might be due to the combined
resonance of the soil profile and the structure. However the nonlinear systems (soil, structure) considered
here do not resonate. Further studies are required to understand the differences between the experimental
and numerical results, which is not the focus of this study.
In spite of the differences in the magnitude of the spectral accelerations, the shape of the response
spectra from the numerical simulations and the experimental data are qualitatively similar for the JOS_L
ground motion. Both SASSI and LS-DYNA result in reasonable predictions of the flexible-base time
period for this ground motion. However the response spectra calculated by SASSI and LS-DYNA (with
the tied interface) have a very different shape from the experimental response spectra for the other three,
more intense ground motions, especially LCN. For these ground motions, SASSI and LS-DYNA (tied)
result in rather poor predictions of the period of maximum spectral acceleration. The other LS-DYNA
(separation allowing interface) analyses, on the other hand, result in response spectra that look similar in
shape to the experimental responses for the SCS_H and PRI_H ground motion, and a reasonable
prediction of the period of maximum spectral acceleration. This indicates that the LS-DYNA analyses
that allow gapping and sliding of the footings result in more reasonable predictions than SASSI and LS-
DYNA (tied). This result is expected given that the soil around the footings undergoes highly nonlinear
response as described in Section 5.3.2.1. All the numerical programs result in poor predictions of the
period of the spectral peaks in the LCN ground motion, which is a consequence of the poor prediction of
site response around the flexible-base time period of MS1F_2, which is about 0.6 to 0.8 sec (see Figure
7-8). The experimental site response for this ground motion shows a peak at 0.6 sec, which being close to
the flexible-base time period of MS1F_2, results in a dominant peak in the structural response. The site
response calculated from the numerical simulations only show a peak at 1 sec.
197
SASSI and LS-DYNA, and experimental observations for JOS_L and LCN ground motions
198
SASSI and LS-DYNA, and experimental observations for SCS_H and PRI_H ground motions
7.4.4.2 Footing response in MS1F_2
The results presented in Section 7.4.3 show that SSSI in Tests 3 and 4 has no meaningful effect
on the global response of MS1F_2. However SSSI effects can still affect the local response of the
199
footings due to restraint provided by adjacent substructures. Analysis of data from Tests 1 and 2 by
Mason (2011) and Trombetta (2013) showed that the footings of MS1F_1 underwent smaller
displacements and rotations, when restrained by the basement of MS3F. Mason (2011) concluded from
Tests 3 and 4 that SSSI did not have a significant effect on the displacements of the footings closest to the
basemat, except that they had residual upward displacements due to the rocking of the adjacent basemat.
Trombetta (2013) examined the data from Tests 3 and 4 and found that the demands on the columns of
MS1F_2 supported on restrained footings (footings closest to the basemat) increased for most of the
ground motions. In this section, the numerical results from the nonlinear LS-DYNA analyses (with
separation-allowing interface) are examined to understand the local SSSI effects on the footings of
MS1F_2 adjacent to the MS2F basemat. Specifically, the moment-rotation response in rocking and the
force-displacement response in the vertical direction are examined. The same four ground motions
considered in the previous sections are used in this section. The footing responses calculated from
numerical results are not compared with the experimental data.
7.4.4.2.1 Moment-rotation response

Figure 7-22, Figure 7-23, Figure 7-24 and Figure 7-25 present the moment-rotation responses of
two footings in MS1F_2 during the ground motions, JOS_L, LCN, SCS_H and PRI_H, respectively, for
the baseline case as well as the SSSI arrangements considered in this section. The top row in each figure
presents the responses of the footing that is always the farthest from the adjacent structures (termed ‘far’
footing hereafter), and the bottom row presents the responses of the footing that is always closest to the
adjacent structures (termed called ‘near’ footing hereafter). The location of the footing in the plan view of
each arrangement is identified by a solid black square in an inset in the top left corner of each plot.
The plots presented in the first row of each figure show that the moment-rotation behavior of the
far footing is barely affected by the presence of the adjacent structures. However, the moment-rotation
behavior of the near footing is affected, although not considerably, for all the ground motions. Notice that
the plots corresponding to the iSSSI and cSSSI cases (second and fourth in each row) of the near footing
show larger negative rotations on average (the plots are skewed slightly to the left), than the baseline and
aSSSI cases, for all four ground motions. For the co-ordinate system used in these analyses, negative
rotation denotes that the footing rotates towards basemat (i.e., the side of the footing closest to the
basemat is moving downwards) in the iSSSI arrangement. This indicates that the footing, when close to
the basemat of the iSSSI and ciSSSI models of MS2F, undergoes more negative, asymmetric, rotation and
rotates towards the basemat when compared with its response in the baseline and aSSSI cases. This is
expected since the soil between the basemat and the footing experiences rocking from two foundations
resulting and larger strains and consequent softening, compared with the baseline case in which the soil
200
experiences rocking only from the footing. However, this softening is not significant enough to result in a
consistent change in the peak footing moment or rotation in the near footing in comparison with the far
footing.
calculated using LS-DYNA (separation allowed) analyses for the JOS_L ground motion
Figure 7-23: Moment rotation-response of the footings of MS1F_2 in various SSSI arrangements,
calculated using LS-DYNA (separation allowed) analyses for the LCN ground motion
201
calculated using LS-DYNA (separation allowed) analyses for the SCS_H ground motion
calculated using LS-DYNA (separation allowed) analyses for the PRI_H ground motion
7.4.4.2.2 Vertical force-displacement response

Figure 7-26, Figure 7-27, Figure 7-28 and Figure 7-29 present the vertical force-displacement
responses for the ground motions JOS_L, LCN, SCS_H, and PRI_H, respectively, for the far footing and
the near footing. The first row in each figure presents the responses of the far footing, and the second row
202
presents the responses of the near footing. The location of the footings relative to the adjacent structures is
indicated by a black solid square at the top left corner of each plot. The vertical forces in these results are
calculated as the axial earthquake-induced compressive forces in the columns above each footing. The
vertical displacement of each footing is calculated as the downward displacement of the center of the
footing. The results do not include the vertical forces and displacements from gravity loading, and the
earthquake-induced forces are calculated by subtracting the gravity loadings from the corresponding
transient results.
The first row in each figure shows that the vertical forces and displacements in the far footing are
virtually identical for the baseline and SSSI cases, indicating that the response of the far footing is
unaffected by the presence of an adjacent structure. The vertical displacement of the near footing is
significantly greater for the SSSI cases. Specifically, the displacements are the greatest for the cSSSI case,
for which the peak displacement of the near footing is almost twice that of the baseline case, for the LCN,
SCS_H and PRI_H ground motions. This is due to softening of the soil between the adjacent structures
due to the presence of two foundations. Rocking of both the basemat and the footings deforms the soil
more than in the baseline case, resulting in larger plastic strains and softer response of the soil. This
observation is different from that by Mason (2011) who noted that the footings adjacent to the MS2F
basemat had residual upward displacements, likely from the upheaval of the soil around MS2F due to the
rocking of the basemat. The main reason for the difference between the experimental and numerical
observations could be the inability of the soil material model to capture residual displacements, which is
challenging and requires the use of sophisticated material models. Note that the numerical models were
built with the objective of simulating the transient responses of the soil-structure system and not
necessarily to predict the residual displacements.
203
arrangements, calculated using LS-DYNA (separation allowed) analyses for JOS_L ground motion
arrangements, calculated using LS-DYNA (separation allowed) analyses for LCN ground motion
204
arrangements, calculated using LS-DYNA (separation allowed) analyses for SCS_H ground motion
arrangements, calculated using LS-DYNA (separation allowed) analyses for PRI_H ground motion
The results show no considerable change due in the vertical forces on the footings due to SSSI:
the peak vertical forces in the SSSI cases are similar to the peak vertical forces in the baseline case. This
observation is contrary to that made by Trombetta (2013) who noted that restraining the footings of
205
MS1F_2 led to an increase in the peak demands in the columns above these footings for a majority of the
input ground motions.
7.4.4.3 MS2F
The shear wall structure, MS2F, was designed to act as a ‘transmitter’ structure that affects the
adjacent MS1F_2 response by transmitting energy through the soil. However, the numerical and
experimental responses of MS1F_2 show negligible SSSI effects, even for the combined SSSI
arrangement, which comprised two MS2F models adjacent to the MS1F_2, indicating that MS2F did not
perform the expected ‘transmitting’ function.
Unlike MS1F_2, Figure 7-30 shows significant variability in the experimental responses of MS2F
in different arrangements for the JOS_L and LCN ground motions. Specifically, the response of the
model in the anti-plane combined SSSI arrangement (caSSSI) is different from response of the other
models, which are similar for the JOS_L ground motion. For the LCN ground motion, the ciSSSI and
caSSSI models of MS2F respond differently from the other two models. This indicates that MS2F is more
sensitive to the presence of the adjacent MS1F_2 in the ciSSSI and caSSSI arrangements, than in the
ciSSSI and caSSSI arrangements during the JOS_L and LCN ground motions. However this behavior of
the ciSSSI and caSSSI models of MS2F is not apparent from Figure 7-31, which shows that the
experimental responses of MS2F in various arrangements are very similar during the SCS_H and PRI_H
ground motions. This shows that the SSSI effects seen in the ciSSSI and caSSSI arrangements for the first
two ground motions may be a result of erratic behavior and not interaction between the structures. The
results of the numerical simulations presented in Figure 7-30 and Figure 7-31 also show that the
responses of MS2F in the different arrangements are very similar for each ground motion. There are small
differences in the responses for a given ground motion, but these are too small to warrant consideration.
Identical to MS1F_2, the SSSI arrangements illustrated in Figure 7-1 have little to no effect on the global
structural response of MS2F.
Figure 7-30 also shows that the numerical simulations result in reasonable predictions of the
spectral accelerations for the JOS_L ground motion, in terms of the peak spectral acceleration as well as
the corresponding period, with the exception of the caSSSI model. The predictions are not as good for the
LCN ground motion, which is due to the inaccurate prediction of the site response at the flexible-base
time period of the structures (around 0.6 to 0.8 sec). For the more intense ground motions, SCS_H and
PRI_H, the nonlinear analyses in LS-DYNA result in spectral shapes that are closer to those observed in
the experiments, than the linear analyses in SASSI. However the magnitude of the numerically calculated
spectral accelerations are much smaller than those calculated from experimental results for these ground
motions.
206
Figure 7-30: Response spectra of the roof acceleration of MS2F from numerical simulations using
SASSI and LS-DYNA, and experimental observations for JOS_L and LCN ground motions
207
Figure 7-31: Response spectra of the roof acceleration of MS2F from numerical simulations using
SASSI and LS-DYNA, and experimental observations for SCS_H and PRI_H ground motions
7.4.5 Effect of basement restraint on the response of MS1F_2
The footing responses calculated using numerical simulations and presented in Section 7.4.4.2
showed that footing restraint by the adjacent basemat did not result in considerable changes in column
demands. However it is possible that a larger neighboring foundation, such as a deeply embedded
208
basement, provides more restraint resulting in a considerable change in near footing response and near
column demands, as observed by Mason (2011) and Trombetta (2013). The effects of such a restraint,
between spread footings and a deep basement, is investigated numerically in this section using nonlinear
SSI analyses in LS-DYNA. The two superstructure models, MS1F_2 and MS2F, are used for this
purpose. However the basemat of MS2F is replaced with a basement of plan dimensions 15m  17m, and
is fully embedded to a depth of 10.1m. The basement walls are 0.5m thick and are numerically modeled
using shell elements and a rigid material model with the unit weight of aluminum, which was used in the
construction of the MS3F basement in Tests 1 and 2 (see Table 7-3). However the unit weight of the
basement is not likely to affect the restraint on adjacent footings, which is the focus of this study. The
interface between the basement and soil is modeled using the contact properties used for the foundation-
soil interfaces of MS2F and MS1F_2. No changes are made in the footings of MS1F_2. The effect of
basement restraint is investigated using 1) a pseudo-static analysis involving horizontal forces
proportional to roof masses applied at roof levels, and 2) a dynamic analysis with the four ground motions
used in the previous sections. These analyses are performed for two cases: 1) a baseline case with
MS1F_2 alone, and 2) an in-plane SSSI case with MS1F_2 placed adjacent to MS2F with a deep
basement (hereafter referred to as MS2F_B): the iSSSI_B case. The baseline case is the same as in the
previous sections, with its LS-DYNA numerical model presented in Figure 7-16. The LS-DYNA
numerical model of the iSSSI_B case is presented in Figure 7-32.
Figure 7-32: LS-DYNA numerical model of the iSSSI_B case with MS1F_2 on the left and MS2F_B
on the right (soil elements are not shown for clarity)
209
7.4.5.1 Pseudo-static analyses
Pseudo-static analyses are performed for the baseline and iSSSI_B cases to investigate SSSI
effects on MS1F_2 due to an adjacent basement. Global SSSI effects are examined from the lateral force-
displacement response at the roof. Local SSSI effects due to footing restraint are examined using 1) the
moment-rotation response, and 2) the vertical force-displacement response of the restrained footings.
Given that dynamic effects in this analysis are negligible, a much smaller soil domain is built,
significantly reducing the computational cost. The numerical model used for the pseudo-static analysis is
presented in Figure 7-33. The lateral force is applied to the roof nodes of MS1F_2 and the forces are
proportional to the values of the corresponding lumped masses. No lateral forces are applied to the MS2F
superstructure and the peak value of each lateral force at each roof mass is equal to the corresponding
mass times acceleration due to gravity. Additionally the forces are applied in the negative X direction,
pushing the superstructure and footings of MS1F_2 towards MS2F. This is assumed to provide more
restraint to the footings, compared with that when the force is applied in the direction away from MS2F,
resulting in more SSSI. The foundation-soil interface for both the structures is modeled to allow for
separation. Details of these contact models are presented in Section 7.4.2.
Figure 7-33: Isometric view of the LS-DYNA numerical model used for pseudo-static analysis of the
iSSSI_B case
Figure 7-34 presents the moment-rotation response, and the vertical force-vertical displacement
response of one of the two MS1F_2 footings closest to the basement. The responses are presented for the
210
baseline and iSSSI cases. The figure shows that the slope of the moment-rotation and vertical force-
displacement responses are considerably larger for the iSSSI_B case, for which the footing is restrained
by a rigid basement. The difference in the slopes is significant in the vertical force-displacement response,
indicating that this response is more sensitive to footing restraint than the moment-rotation response. The
results of the pseudo-static analyses therefore show a considerable increase in both the rocking stiffness
and vertical stiffness of the restrained footings due to SSSI. As a result of the increased footing stiffness,
the global lateral stiffness of MS1F_2 is also greater for the iSSSI_B case. This is evident from Figure
7-35, which presents the lateral force-displacement response of MS1F_2 for the baseline and iSSSI_B
cases. The increase in the lateral stiffness of MS1F_2 for the iSSSI_B case is clear from this figure.
Foundation restraint due to adjacent basement(s) may result in a meaningful change in global structural
response.
Figure 7-34: Moment-rotation response and the vertical force-displacement response of the
restrained MS1F_2 footings in the baseline and iSSSI_B cases
211
Figure 7-35: Lateral force-displacement response of the MS1F_2 superstructure in the baseline case
and the iSSSI_B case
7.4.5.2 Dynamic analyses
The pseudo-static analyses showed an increase in the rocking and vertical stiffnesses of the near
footings of MS1F_2 when they are restrained by a deep basement. This also led to an increase in the
global lateral stiffness of the superstructure. This section investigates the effects of this interaction on the
seismic response of MS1F_2. Since MS2F_B is founded on a deep basement, it is assumed that the SSSI
effects on this structure are negligible and the response of MS2F_B is not examined. The dynamic
analyses are performed for the four ground motions, JOS_L, LCN, SCS_H and PRI_H, used previously in
this section. The structural response is measured using roof acceleration response spectra in MS1F_2. The
SSSI effects are examined by comparing the response of the structure in the baseline case to that in the
iSSSI_B case. The moment-rotation response and the vertical force-displacement responses of the
footings closest to the basement are also examined.
Figure 7-36 presents the responses of MS1F_2 calculated using LS-DYNA for the baseline and
iSSSI_B cases for the four ground motions. The panels in the left column in the figure present the
response calculated using a tied soil-foundation interface; the panels in the right column present the
response calculated using an interface that allows separation. The figure shows that the responses in the
baseline and iSSSI_B cases are similar when simulated with a tied interface, indicating that the SSSI
effects are negligible. However there are significant differences in the baseline and iSSSI responses
212
calculated using the separation-allowing interface for the LCN, SCS_H and PRI_H ground motions. The
differences are primarily in the peak spectral accelerations, which are much smaller for the iSSSI_B case
than in the baseline case. Additionally, the period of peak spectral acceleration is slightly smaller in the
iSSSI_B case indicating stiffening in the response. To further investigate the reason for these differences,
the moment-rotation responses of the restrained footing are calculated for the baseline and iSSSI_B cases
from the results of the analyses performed with separation-allowing interface. These responses are
presented in Figure 7-37. Two important observations can be made from this figure: 1) the average slope
of the moment-rotation curves is higher for the iSSSI_B case than in the baseline case, and 2) the
moment-rotation response of the iSSSI_B case shows only a small negative slope portion (at large
positive rotations) unlike for the baseline case where the slope of the moment-rotation curve is negative
beyond a rotation of 0.0004 radians. The former is consistent with the findings of the pseudo-static
analyses, which showed stiffening in the moment-rotation behavior of the restrained footing. The negative
slope portion in the moment-rotation response corresponds to partial uplift (or gapping) of the footing,
which results in a sudden drop in the moment with gradual increase in the rotation. The moment-rotation
responses show that gapping in the restrained footings is almost completely absent in the iSSSI_B case,
while there is significant gapping in the footings in the baseline case. This explains the large differences
between the roof response of the baseline and iSSSI_B cases: since secondary nonlinearities such as
sliding and gapping, contribute significantly to the response of MS1F_2 (see Section 5.3.2.1 for a detailed
description of the SSI response of MS1F_2), reduction of gapping in the iSSSI_B case reduces the
superstructure response as seen in Figure 7-36. The reduction of gapping in this case is due to the increase
in the rocking stiffness, which results in smaller footing rotations. In the case of the restrained footings of
MS1F_2 the rotations are smaller than the threshold required for gapping between the soil and the
footings. Additionally the absence of SSSI effects in the results calculated using a tied soil-foundation
interface supports the conclusion that the significant SSSI in the iSSSI_B case in Figure 7-36 is due to a
reduction in gapping.
213
LS-DYNA for four ground motions
214
Figure 7-37: Moment-rotation responses of the restrained footings of MS1F_2 in the baseline and
iSSSI_B cases calculated for the four ground motions
7.4.6 Effect of basement restraint on the response of MS1F_1
The analyses presented in Section 7.4.5 showed that the presence of a basement significantly
affects the response of MS1F_2. However, MS1F_2 is founded on light footings, designed to intensify the
SSI effects comprising rocking and sliding of the footings, unlike MS1F_1, which is designed to have
realistic, code-conforming dynamic properties. This section investigates the effect of an adjacent
basement on the response of a more realistic footing structure, MS1F_1.
MS1F_1 is a heavier and more flexible model than MS1F_2 and is described in Section 6.3.3.
Specifically, MS1F_1 is taller, and has heavier roof masses and heavier footings that are made of steel (as
opposed to aluminum used for MS1F_2). The numerical models of the baseline and iSSSI_B cases of
MS1F_1 are built by making modifications to the numerical models used in the previous section for
MS1F_2. No changes are made to the superstructure or the basement of MS2F_B. The foundation-soil
interfaces of both the structures are modeled to allow separation, and analyses with tied interfaces are not
215
performed. The ground motions used previously, JOS_L, LCN, SCS_H and PRI_H, are used for the
analyses in this section. Structure-soil-structure interaction effects in MS1F_1 are examined by
comparing acceleration response spectra at the roof for the baseline case and the iSSSI_B cases. Figure
7-38 presents these acceleration response spectra and enables such a comparison. The figure shows that
roof responses of MS1F_1 are virtually identical for the baseline and iSSSI_B cases. This observation is
different from that of the responses of MS1F_2, which show significant SSSI effects when a separation-
allowing interface is used. The smaller SSSI effects in MS1F_1 are due to the flexible superstructure,
which results in less rocking in the footings and consequently no gapping for both the baseline and the
iSSSI_B cases. Additionally, the footings in MS1F_1 are much heavier and have a larger moment of
inertia than those of MS1F_2 and require larger moments to produce the same rotation. This further
decreases the footing rotations and avoids gapping under the footings. Figure 7-38 also shows slightly
greater peak spectral accelerations for the iSSSI_B case than in the baseline case. This is because
restraining the footings in the iSSSI_B case decreases the footing rotations from those observed in the
baseline case, resulting in smaller hysteretic energy dissipation in the soil at the vicinity of the footings.
Therefore the hysteretic damping is slightly smaller, causing an increase in the peak spectral
accelerations.
Figure 7-38: Response spectra of the roof acceleration of MS1F_1 calculated using numerical
simulations in LS-DYNA for four ground motions
216
Four centrifuge experiments were conducted at the NEES@UCDavis centrifuge facility to

understand the effects of SSSI on buildings in dense urban environments. Test 1 involved two baseline
SMRF structures (MS1F_1 and MS3F) placed sufficiently far away from each to not introduce any SSSI
effects. These structures were placed directly adjacent to each other in Test 2 to isolate and understand the
impact of SSSI. Tests 3 and 4 involved a two-story shear wall structure (MS2F) and a one-story SMRF
structure (MS1F_2) designed to act as a ‘transmitter-receiver’ couple. The baseline model of MS1F_2 and
an in-plane SSSI configuration were studied in Test 3, and an anti-plane SSSI and a combined in-plane
anti-plane SSSI configuration were studied in Test 4. Numerical simulations were performed to calculate
the site response and SSI response. The frequency-domain equivalent-linear program, SHAKE, and the
nonlinear time-domain programs, DEEPSOIL and LS-DYNA, were used for site-response analysis. The
SSI analyses were performed using frequency-domain linear program SASSI, and the direct method in
LS-DYNA. The LS-DYNA analyses were performed for two cases: using the tied interface, and a
separation-allowing interface. In addition to the SSSI arrangements considered in the City Block
experiments and illustrated in Figure 7-1, numerical simulations were performed to investigate the SSSI
effects in MS1F_2 due to the presence of an adjacent structure with a basement (MS2F_B). This
investigation was performed through pseudo-static and dynamic analyses of the SSI models. To
understand the effects of an adjacent basement on the response of a structure with more realistic footings,
the dynamic analyses were rerun after replacing MS1F_2 with MS1F_1. The following conclusions can
be drawn from the experimental and numerical results:
1. Considerable inter-test variability and intra-test variability exist in the input ground motion and
the site response. This variability needs to be taken into account when assessing SSSI effects, so
as to not overstate the effects, if any.
2. The site period elongates and hysteretic damping increases as the ground motion intensity
increases due to nonlinear behavior in the soil. The numerical programs capture this period
elongation reasonably well.
3. The equivalent-linear program, SHAKE, overpredicts spectral accelerations at the site period in
comparison with the nonlinear programs, which result in predictions that are similar to each other.
All programs predict smaller spectral accelerations at periods less than the elongated site period
than those calculated from the experimental results.
4. The experimental results show that the interaction between MS1F_2 and MS2F in the SSSI
arrangements considered in the City Block experiments have no considerable effect on global
structural responses. Some variability in the response of MS2F in different arrangements is
217
observed for the JOS_L and LCN ground motions but this is deemed erratic because the
variability is not seen in either the remaining experiments or the numerical simulations.
5. The numerical results agree with the experimental results in terms of the (small) magnitude of
SSSI effects. The global structural responses calculated in different SSSI arrangements are
virtually identical for a given ground motion, indicating that the interaction between MS1F_2 and
MS2F in these arrangements is insignificant.
6. There are large differences between the numerical and experimental results in terms of the
magnitude of spectral accelerations, especially for MS1F_2. Similar differences were also
observed by another City Block researcher (Jones, 2013), who performed numerical simulations
using FLAC. The reason for these differences warrants further investigation, which is outside the
purview of this study.
7. Of the different numerical analysis methods, only LS-DYNA (with a separation-allowing
interface) reasonably captures the shape of the roof response spectra of MS1F_2 for the more
intense ground motions, SCS_H and PRI_H. This indicates that simulating foundation
nonlinearities, such as footing sliding and gapping, are important to accurately predict structural
response for surface- or near-surface-mounted structures subjected to intense ground motions.
8. Large differences between the experimental and numerical predictions of spectral accelerations
are seen for MS2F during the more intense ground motions, SCS_H and PRI_H, which were also
observed by Jones (2013). However all of the numerical programs make reasonable predictions of
spectral shape and the flexible-base time period of MS2F.
9. The ‘transmitter-receiver’ hypothesis, which was presumed for the design of Tests 3 and 4, was
negated because MS1F_2 and MS2F responded almost independently of each other.
10. The response of the footing close to the basemat of the adjacent MS2F (in the iSSSI and ciSSSI
positions) showed more asymmetric behavior than the footings away from the basemat. This
asymmetric behavior could be addressed when modeling the footings by spring-based methods of
SSI analysis.
11. The experimental and numerical results from Tests 3 and 4 showed that the structure-soil-
structure interaction due to the dynamic behavior of the structures (which can also be referred to
as wave-based SSSI) is not significant for the SSSI arrangements considered in the City Block
project. Since these structures are rather generic, it can be said that wave-based SSSI effects are
not important for calculating the global response of low- to medium-rise structures.
12. Pseudo-static analysis provides a convenient and computationally economical alternative to
dynamic analysis to investigate SSSI effects resulting from foundation restraints. Such an
investigation of the interaction between MS1F_2 and MS2F_B (MS2F with a deep basement)
218
showed that the rocking and vertical stiffnesses of the footings restrained by an adjacent basement
increased considerably. Additionally, the vertical stiffness of the footings was more sensitive to
restraint by the adjacent basement than the rocking stiffness. The increase in the footing stiffness
also increased the global lateral stiffness of the MS1F_2.
13. Dynamic analyses showed that the presence of an adjacent deep basement considerably changes
the global structural response of MS1F_2, when the soil-footing interface was modeled to allow
separation. This is primarily due to the reduction in gapping between the soil and the footings,
which contributes significantly to the MS1F_2 response as shown in Section 5, by increasing the
rocking stiffness of the restrained footings. The SSSI effects in MS1F_2 involving a tied footing-
soil interface were much smaller since gapping was not simulated.
14. The response of MS1F_2 exhibits a situation where local SSSI effects due to footing restraint
from an adjacent basement results in considerable changes in the global structural response. SSSI
effects from local foundation restraint should therefore be accounted for in stiff structures
founded on spread footings where considerable gapping, or uplift, is expected in design basis
shaking.
15. The heavier, more flexible MS1F_1 showed much smaller SSSI effects when placed adjacent to a
structure with a deep basement than MS1F_2. This is because the superstructure of MS1F_1 is
more flexible, and its footings are heavier than those of MS1F_2; thus, gapping is less likely to
occur beneath the footings of MS1F_1.
16. The numerical analyses do not show any consistent changes in the column demands due to
restraints provided by an adjacent basement. Rather, the results of the numerical analyses show
that an increase in footing stiffness due to restraint reduces the transient rotations.
17. SSSI is a complicated phenomenon that needs to be investigated with ‘bottom-up’ approach by
starting with very simple, elastic, experimental models. Experimental data must be supplemented
by comprehensive nonlinear numerical simulations to gain insight into the underlying physical
mechanisms.
219
SECTION 8
SUMMARY, CONCLUSIONS AND RECOMMENDATIONS
8.1 Summary
Soil-structure interaction (SSI) analysis is the process of calculating structural response

accounting for the flexibility of the underlying soil. Site-response analysis is a precursor to SSI analysis
and involves the calculation of the response of a soil deposit to a bedrock motion. Site-response analysis
is also used to calculate free-field surface ground motion for input to surface or near-surface mounted
structures. Structure-soil-structure interaction (SSSI) is an extension of SSI, and refers to the interaction
of neighboring structures through the soil surrounding them. The larger goal of this report is to improve
the current understanding of site response, SSI and SSSI, and to advance the numerical tools that simulate
these phenomena for deployment in performance assessment of buildings and nuclear structures.
Section 2 reviews the numerical methods for site-response analysis, including their evolution, and
the state-of-the-art. It describes site-response analysis using the equivalent-linear frequency-domain
program SHAKE (Schnabel et al., 2012), and the nonlinear time-domain programs DEEPSOIL (Hashash
et al., 2011) and LS-DYNA (LSTC, 2013). Section 3 presents a similarly structured review of numerical
methods in SSI analyses and describes SSI analysis using the linear frequency-domain program SASSI
(Lysmer et al., 1999), and LS-DYNA. It presents a review of past studies of SSSI in buildings and
nuclear structures.
Site response, SSI and SSSI are complex nonlinear problems involving three-dimensional wave
propagation and sometimes, very large strains in the soil. Traditionally, site-response analysis is bypassed
for most buildings in lieu of using attenuation relationships that empirically account for the influence of
the underlying soil deposit on the surface ground motion. Soil-structure interaction is usually ignored for
buildings that are flexible and constructed on stiff soil sites, but when important, is accounted for using
simplified methods that model the underlying soil using equivalent springs and dashpots. Nuclear
structures, generally being stiff and heavy, elicit significant SSI, and are analyzed with more rigorous
methods. Site-response analysis for nuclear applications is performed using one-dimensional equivalent-
linear methods in the frequency domain, typically using SHAKE. Soil-structure interaction analysis for
these applications is typically performed using SASSI, which is a linear, frequency-domain program that
is capable of three-dimensional SSI analyses. These frequency-domain methods, developed several
decades ago, are now very well established and are used extensively in both the civil and nuclear industry
because of their simplicity and computational efficiency. There is consensus that the linear frequency-
domain methods will likely result in accurate predictions for low intensity earthquakes, where soil strains
221
are small and the behavior is essentially linear. However the utility of these methods for intense
earthquakes, resulting in large soil strains, is unproven. Nonlinear time-domain methods, which have been
developed more recently, potentially enable a more realistic modeling of soil-structure systems, and can
capture nonlinearities near the foundation such as gapping and sliding that are much more likely with
intense earthquake shaking. Well-known nonlinear time-domain programs are DEEPSOIL, used for site-
response analysis, FLAC (Itasca Consulting Group, 2005), used for site-response and SSI analysis, and
the commercial finite element programs, ABAQUS (Dassault Systèmes, 2005) and LS-DYNA, used for
site-response and SSI analyses. The nonlinear methods are computationally more expensive than the
linear frequency-domain methods and their use has been limited to high-value buildings, bridges and
petrochemical facilities.
Sections 4 and 5 of this report present detailed assessments of current, industry-standard site-
response and SSI analysis programs, respectively, and provide guidelines to practicing engineers on the
usage of these programs. Section 4 presents an assessment of the site-response analysis programs,
SHAKE, DEEPSOIL and LS-DYNA. The assessment includes calculation of responses of four idealized,
but representative, soil sites (two sites representing Central and Eastern United States, and two sites
representing Western United States) subjected to various bedrock ground motions of different intensities
and frequency contents. The responses calculated using the different programs are compared for all sites
and ground motions, and challenges and pitfalls with the use of each of these programs are identified.
Biaxial site-response analysis is also studied to understand the effect on site response of simultaneously
applying two horizontal ground motion components on site response. Guidelines are provided on the use
of site-response analysis programs for different levels of expected peak shear strains in the soil profile.
Section 5 presents an assessment of industry-standard SSI analysis programs, SASSI and LS-DYNA. The
first step in the assessment is to benchmark the time-domain program LS-DYNA (using the domain
reduction method as well as the direct method) against the frequency-domain program SASSI, for linear
SSI analyses. This is done by performing linear SSI analyses of two idealized, lumped mass structures
using SASSI and LS-DYNA, and verifying that the responses calculated using the two programs are
similar. The second part of the assessment involves calculating nonlinear SSI responses of a one-story
structure founded on spread footings and a two-story structure founded on a basemat for four input
ground motions of different intensities. Nonlinear responses are calculated in SASSI using strain-
compatible moduli and damping ratios for the soils. LS-DYNA analyses are performed using two soil-
foundation interfaces: 1) an interface that is tied and does not allow separation (gapping and sliding)
between the soil and the foundation, and 2) an interface that allows separation between the soil and the
foundation. The responses calculated using SASSI and LS-DYNA are compared for all the ground
motions, and challenges and pitfalls in performing SSI analyses using these programs are identified.
222
Unlike for SSI, which has been studied for many decades and is currently well understood,
studies of SSSI are small in number and mostly limited to analytical studies involving highly idealized
structures, foundations and soil profiles. A better understanding of SSSI is important for performance
assessment of buildings in dense urban areas, which are often constructed in ‘city blocks’ and may
interact with each other during an earthquake. The US NSF funded project titled ‘Seismic Performance
Assessment in Dense Urban Environments’, and called the ‘City Block project’ in this report, attempts to
remedy the lack of SSSI research through a series of centrifuge experiments and numerical simulations.
Six centrifuge experiments were conducted as a part of this project at the NEES@UC Davis centrifuge
facility, on scaled-down models of buildings. The first two experiments, Tests 1 and 2, examined SSI and
SSSI in buildings designed with code-conforming properties. Test 3 and 4 involved stiffer buildings that
were designed to experience more SSI and SSSI. Tests 5 and 6 examined SSI and SSSI in idealized
structures founded on liquefiable soils. Section 6 describes Tests 2, 3 and 4 of the City Block centrifuge
experiments. It includes descriptions of the test plan, design and construction of the model structures, and
the chronology of the input ground motions used during the tests. Section 7 presents numerical
simulations of site response in Tests 3 and 4 using SHAKE, DEEPSOIL and LS-DYNA, and numerical
simulations of SSI and SSSI using SASSI and LS-DYNA. The results of the simulations are qualitatively
compared to the experimental results to make observations regarding the effects of SSSI on the global
responses of the model structures. Section 7 also presents an exploratory study on the effects of SSSI
through foundation restraint from adjacent substructures.
8.2 Conclusions
The overarching conclusions of each study in this report are presented in the sections that follow.
8.2.1 Assessment of industry-standard site-response analysis programs
The following conclusions are drawn from the assessment of the site-response analysis programs,
SHAKE, DEEPSOIL and LS-DYNA, in Section 4:
 Equivalent-linear and nonlinear programs result in very similar predictions of response for stiff sites
subjected to low intensity ground motions. The predictions become increasingly different for analyses
involving higher strains, and at higher frequencies. The predictions from equivalent-linear programs
start diverging from those of the nonlinear programs when the equivalent-linear peak shear strains in
the soil are greater than about 0.1%. The predictions of the equivalent-linear and nonlinear programs
are very different at peak shear strains greater than 1%.
223
 The differences between the predictions of the equivalent-linear and nonlinear programs for cases
involving large shear strains are as follows :
o The spectral accelerations calculated using the equivalent-linear and nonlinear programs are
increasingly different at shorter periods (periods smaller than the elongated site period). The
equivalent-linear programs tend to result in almost constant spectral accelerations at these
periods, which is a manifestation of near-zero frequency content in the response at the
corresponding frequencies. This is a direct consequence of modeling the soil response using a
secant shear modulus, as opposed to modeling the changes in the shear modulus with time, as in a
nonlinear analysis. Modeling cyclic changes in shear modulus results in high-frequency response
when the shear modulus is high, and low-frequency response when the shear modulus is low.
Equivalent-linear programs do not capture the high-frequency content since they simulate the soil
response with a constant shear modulus, which is equal to the iterated modulus.
o The equivalent-linear programs tend to underestimate the peak strains in the soil layers in
comparison with the nonlinear programs.
o The equivalent-linear programs tend to overestimate the acceleration amplifications at the site
period in comparison with the nonlinear programs.
o The equivalent-linear and nonlinear programs predict almost the same elongated site periods for
rock and firm soil sites. For others, involving very soft soil sites subjected to intense ground
motions, the equivalent-linear programs predict slightly smaller site periods than the nonlinear
programs.
 The site responses calculated using all three programs contain anomalous responses for peak strains
greater than 1%. Of particular importance is the response at higher frequencies (> 15Hz), which is
important for nuclear structures: SHAKE tends to underestimate the site response at these frequencies
when compared with the responses calculated using the nonlinear programs due to reasons given
earlier. LS-DYNA analysis using the MAT_HYSTERETIC material model tends to result in an
unusual number of spikes in the acceleration response when compared with the other programs. These
spikes result in short-period spectral accelerations that are larger than those calculated using
DEEPSOIL. This anomalous response from LS-DYNA is due to sudden changes in the slope of the
backbone curve used in the MAT_HYSTERETIC material model, which currently accepts a
maximum of 10 stress strain points as input. Increasing the number of points significantly (such as
120, in the study presented in this report) will ‘smoothen’ the backbone curve, reduce the spikes in
the acceleration response, and result in a response that is much closer to that calculated using
DEEPSOIL. DEEPSOIL also results in anomalous responses for analyses involving large strains:
there are large differences in the peak strains of consecutive layers, despite the layers having identical
224
material properties. An investigation into this anomalous response revealed that it is due to the usage
of extended Masing rules of soil hysteresis (see Section 2.1.2.2 for a description of extended Masing
rules).
 There is significant variability in the site-response predictions from different programs for high-
intensity ground motions, despite using almost identical input material properties, and very simple
soil profiles. Greater deviations between the predictions are expected for the more complex soil
profiles encountered in practice.
 Equivalent-linear methods should provide a reasonable solution if soil strains are small and can be
used with high confidence if the peak soil shear strains are expected to be less than 0.1%. Nonlinear
analysis should be performed and its results substantially weighted for design, when peak shear
strains are significantly greater than 0.1%.
 Site-response analysis is typically performed using only one component of ground motion.
Performing a biaxial analysis by simultaneously using two horizontal components of ground motion
can considerably affect the site response for intense ground motions resulting in large shear strains.
8.2.2 Assessment of industry-standard SSI analysis programs

The following conclusions are drawn from the assessment of the SSI analysis programs, SASSI
and LS-DYNA, in Section 5:
 The frequency-domain program, SASSI and the time-domain program, LS-DYNA, result in almost
identical responses for SSI analyses of linear models. This is an important result in the benchmarking
of time-domain programs against the frequency-domain programs for linear analyses. However the
frequency-domain programs are much more computationally efficient: for one analysis case
considered in this study, SASSI was about 120+ times faster than the direct method in LS-DYNA.
The domain reduction method in LS-DYNA also offers a computationally efficient alternative to the
direct method, while producing almost identical results: for one analysis case, the domain reduction
method was about four times faster than the direct method.
 Nonlinear SSI predictions can be significantly different from those made using linear frequency-
domain programs. The differences are greatest for cases with significant nonlinearities, such as
nonlinear site response (primary nonlinearities) and nonlinear behavior at the foundation (secondary
nonlinearities), namely, soil hysteresis, and gapping and sliding underneath the foundation. In the
study presented in this report, the responses of a one-story footing structure calculated using SASSI
and LS-DYNA differed considerably, even for a very low intensity ground motion due to nonlinear
soil behavior at the vicinity of the footings.
225
 Soil-structure interaction analyses in both the frequency domain in SASSI and time domain in LS-
DYNA can produce anomalous responses, which can go unnoticed if only one of the programs is
used. Some practical considerations should be made to avoid these anomalous responses. Some of
these considerations include:
o Structural response calculated using SASSI can be particularly sensitive to the number of analysis
frequencies at which the transfer functions are calculated. Although the SASSI user manual
(Ostadan, 2006b) recommends 10 to 20 frequencies, which are to be selected based on the fixed-
base natural frequencies of the structure, it was found that a larger number of frequencies is
required, and that the input frequency set should be sufficient enough to characterize the peaks in
the transfer function corresponding to 1) the flexible-base (and not the fixed-base) frequency of
the structure, and 2) the natural frequencies of the soil profile. However since the flexible-base
frequencies of the structure may not be available before the analysis, it is difficult to choose the
analysis frequencies a priori.
o Structural response calculated using SASSI can be sensitive to the number of elements (i.e., the
meshing) of the foundation, in spite of the foundation being significantly stiffer than the
surrounding soil. The number of elements in the foundation governs the number of interaction
nodes: the nodes for which SASSI calculates the foundation impedance functions. However, it is
difficult to identify the adequacy of the number of foundation elements unless an analysis result
from a different program, such as LS-DYNA, is available for comparison.
o The rocking response of a structure calculated using SASSI can be considerably different for two
different site layering schemes, although these schemes result in identical site responses at the
surface. Smaller layer thicknesses are required near the foundation of the structure to produce a
structural response that is similar to that calculated using LS-DYNA. The site layering scheme
used for site-response analysis (see Section 4.4.2 for description of site layering for site-response
analysis), may not be adequate for SSI analysis in SASSI.
o The responses of linear or nonlinear models calculated using LS-DYNA can be affected by
hourglass modes in the elements of the foundation, and of the soil. These modes can be mitigated
using hourglass control in LS-DYNA.
 Nonlinear response at the vicinity of the foundations (secondary nonlinearities) can be significant in
buildings. This nonlinear response can lead to an increase in superstructure demands. A considerable
increase in the roof accelerations of the one-story structure with footings is observed when it is
modeled with a soil-footing interface that simulates gapping and sliding, as opposed to an interface
that does not allow separation between the soil and the footing. Ignoring secondary nonlinearities
226
such as gapping and sliding can therefore result in an unconservative prediction of superstructure
response.
 Structures with spread footings are susceptible to an increase in framing ductility demands due to
footing spreading: residual displacements of the footings away from each other. Connecting the
footings with grade beams will eliminate footing spreading and reduce ductility demands in the
superstructure.
8.2.3 Centrifuge experiments and numerical simulations of the NEES City Block project
The following conclusions are drawn from the centrifuge experiments examining SSSI and the
corresponding numerical simulations presented in Sections 6 and 7:
 Accurate numerical modeling of centrifuge experiments is very difficult. It requires modeling of the
centrifuge itself, and careful determination of the soil properties when the centrifuge is in flight.
 Analysis of the data recorded from a centrifuge test to examine SSSI involves the comparison of the
data from two different models: usually a baseline model involving a single structure sufficiently far
away from neighboring structures, and an SSSI model involving multiple structures that interact with
each other during an earthquake. For an accurate assessment of SSSI effects through this comparison,
inherent variabilities in input ground motion and material properties should be accounted for. The
variability in the input ground motion and soil properties for the City Block experiments are
considerable.
 The site-response predictions calculated using SHAKE, DEEPSOIL and LS-DYNA are very different
from the experimental results, especially in the short period range, but capture the elongated time
period reasonably well.
 Structure-soil-structure interaction analyses can be performed in the frequency domain using SASSI
and the time domain using LS-DYNA, with explicit modeling of the soil and structures, and without
using soil springs. However some practical considerations for SSI analysis, noted in the previous
section, should be made to avoid anomalous responses and produce reasonable predictions.
 Structure-soil-structure interaction does not result in any significant change in the global structural
responses of the structures considered in Tests 3 and 4 of the City Block project. There is reasonable
agreement between the experimental and numerical results in this regard.
 There are significant differences between the roof responses calculated from numerical simulations
and those from experiments. Numerical simulations by another City Block researcher (Jones, 2013)
using FLAC (Itasca Consulting Group, 2005) also showed similar differences. The causes of the
227
differences between numerical and experimental responses is beyond the scope of this research
project.
 Structure-soil-structure interaction effects due to dynamic response of the structures, namely, wave-
based effects, have a negligible effect on the responses of low- to medium-rise structures such as
those considered in Tests 3 and 4 of the City Block project.
 Structure-soil-structure interaction through foundation restraint by an adjacent substructure or
foundation may affect the local response of the foundation, such as its moment-rotation response or
its vertical force-vertical displacement response. Such effects of SSSI, which do not involve dynamic
effects, can be conveniently examined through pseudo-static analyses. The pseudo static analyses
performed in this study revealed that the presence of an adjacent basement increases the rocking
stiffness and vertical stiffness of footings, with the vertical stiffness being more sensitive to the
restraint provided by the basement. For the structure considered in this study, the increase in footing
stiffness also resulted in an increase in the global lateral stiffness of the structure.
 The dynamic SSSI analyses presented in Section 7 showed that restraint on footings by an adjacent
basement considerably reduced the transient rotations and transient vertical displacement of the
footings. In one case, involving a stiff superstructure and very light footings prone to gapping and
uplift, the effects of footing restraint by the adjacent basement resulted in significant changes in the
global response of the superstructure. Further investigation revealed that this change was due smaller
footing rotations, which resulted in a reduction in gapping beneath the footings.
 The numerical investigations presented in Section 7 revealed no significant changes in the demands
on the columns above the restrained footings due to the presence of an adjacent substructure or
foundation.
 Structure-soil-structure interaction due to restraint from an adjacent foundation can affect global
superstructure response in some cases, and nonlinear analysis that simulates separation at the
foundation-soil interface is needed to accurately compute the response. Accurate simulation of SSSI
requires nonlinear analysis with robust contact models.
8.3 Future research
 Investigation of the mechanisms of structure-soil-structure interaction: The study performed in this

research project, as well as by other City Block researchers, showed that wave-based SSSI, namely,
the effect of the dynamic response of one structure on the dynamic response of an adjacent structure,
is negligible for the low- to medium-rise buildings considered in Test 1 through Test 4 of the City
Block project. Interaction between neighboring structures can result from 1) foundation restraint due
228
to the presence of an adjacent foundation, 2) modification of the ground motion due to kinematic
interaction, and 3) shielding of ground motion by adjacent structures. The first mechanism was
observed in the spread footings of a one-story structure considered in this study, when placed against
a basement. The second mechanism can be important in cases involving significant kinematic
interaction, such as very deep basements or buildings adjacent to underground tunnels. Shielding will
be important for buildings in dense urban regions involving deeply embedded foundations or
underground tunnels, subjected to horizontal and inclined wave-fields. These mechanisms should be
explored as a part of an investigation of SSSI in dense urban areas.
 Effect of SSI and SSSI in performance assessment of buildings using the second generation
performance-based earthquake engineering (FEMA-P58) procedures: The effects of SSI and SSSI
can be interpreted in terms of acceleration response spectra, spectral amplifications, or engineering
demand parameters such as peak roof acceleration, peak story drift and residual story drift. Although
these are valid interpretations of interaction effects, they are not sufficiently tangible to make
objective decisions regarding the significance of SSI and SSSI in the overall design of a structure. An
alternate way to judge the importance of SSI and SSSI is to estimate the change in earthquake loss
with its explicit considerations, where losses can be expressed in terms of repair cost, downtime, and
casualties, and calculated for intensities of ground motion or on an annualized basis. The FEMA-P58
methodology (FEMA, 2012) could be used for these calculations.
 Improvement of the numerical predictions of site response and SSI: The numerical methods for site-
response and SSI analysis can be improved by:
o Validation: Validation of the industry-standard numerical programs is a critical challenge for site-
response and SSI analysis. The validation exercises should be performed with an emphasis on
large strains (> 1%) and high frequencies (> 10 Hz).
o Accounting for three-dimensional effects: Earthquakes produce complex three-dimensional wave-
fields involving body and surface waves. Three-dimensional effects are typically ignored, and
further studies are required to evaluate their significance.
 Soil-structure interaction effects in base-isolated nuclear power plants: Base isolation can
significantly reduce mean seismic demands in structures, systems and components of fixed base
nuclear power plants. Base-isolation may also reduce SSI effects in nuclear power plants but it is
unknown whether this reduction is sufficient to warrant neglecting the effects altogether when
calculating distributions of seismic demand, as is required by ASCE Standard 4 (ASCE, 1998; ASCE,
forthcoming). Further study is required to examine the influence of base isolation on SSI effects in
nuclear structures. This study should investigate whether a) a fixed-base analysis using a free-field
ground motion (ignoring SSI effects) can be performed in lieu of an SSI analysis, to calculate
229
distributions of seismic demands in base isolated nuclear structures, and b) rocking effects are
significant on sites considered appropriate for use of horizontal base isolation.
230
SECTION 9
REFERENCES
Anderson, L. M., Carey, S., Amin, J., and Ostadan, F. (2013a). "Effect of Separation Distance and Soil
Parameters on the Structure-Soil-Structure Interaction Response of Adjacent Deeply Embedded
Structures." 22nd International Conference in Structural Mechanics in Reactor Technology (SMiRT-22),
San Francisco, California.
Anderson, L. M., Elkhoraibi, T., and Ostadan, F. (2013b). "Validation of SASSI2010 Solution Methods
through Independent Verification Using SAP2000 for Deeply Embedded Structures with Large
Footprints." 22nd International Conference in Structural Mechanics in Reactor Technology - SMiRT 22,
August 2013, San Francisco, California.
ANSYS Inc. (2013). Computer Program ANSYS 13.0, ANSYS Inc., Canonsburg, Pennsylvania.
Archuleta, R. J., Bonilla, L. F., and Lavallée, D. (2000). "Nonlinearity in Observed and Computed
Accelerograms." Proceedings: 12th World Conference in Earthquake Engineering, Auckland, New
Zealand.
American Society of Civil Engineers (ASCE). (1998). "Seismic Analysis of Safety-Related Nuclear
Structures and Commentary." ASCE 4-98, American Society of Civil Engineers, Reston, Virginia.
American Society of Civil Engineers (ASCE). (2010). "Minimum Design Loads for Buildings and Other
Structures." ASCE/SEI 7-10, American Society of Civil Engineers, Reston, Virginia.
American Society of Civil Engineers (ASCE). (forthcoming). "Seismic Analysis of Safety-Related

Nuclear Structures and Commentary." ASCE 4-**, American Society of Civil Engineers, Reston,
Virginia.
Assimaki, D., Kausel, E., and Whittle, A. (2000). "Model for Dynamic Shear Modulus and Damping for
Granular Soils." ASCE Journal for Geotechnical and Geoenvironmental Engineering, 126, 859-869.
Assimaki, D., and Gazetas, G. (2004). "Soil and Topographic Amplification on Canyon Banks and the
1999 Athens Earthquake." Journal of Earthquake Engineering, 8(1), 1-43.
Baig, I., and Bolourchi, S. (2013). Personal Communication
Baker, J. W. (2007). "Quantitative Classification of Near-Fault Ground Motions Using Wavelet

Analysis." Bulletin of the Seismological Society of America, 97(5), 1486-1501.
231
Basu, U. (2009). "Explicit Finite Element Perfectly Matched Layer for Transient Three-Dimensional
Elastic Waves." International Journal for Numerical Methods in Engineering, 77, 151-176.
Basu, U. (2014). Personal Communication
Bathe, K.-J. (1996). "Finite Element Procedures." Prentice Hall, Upper Saddle River, New Jersey.
Bielak, J. (1975). "Dynamic Behaviour of Structures with Embedded Foundations." Earthquake

Engineering & Structural Dynamics, 3, 259-274.
Bielak, J., and Christiano, P. (1984). "On the Effective Seismic Input for Nonlinear Soil-Structure
Interaction Systems." Earthquake Engineering & Structural Dynamics, 12(1), 107-119.
Bielak, J., Loukakis, K., Hisada, Y., and Yoshimura, C. (2003). "Domain Reduction Method for Three-
Dimensional Earthquake Modeling in Localized Regions, Part-I: Theory." Bulletin of the Seismological
Society of America, 93(2), 817-824.
Bielak, J. (2013). Personal Communication
Bolisetti, C. (2010). "Numerical and Physical Simulations of Soil-Structure Interaction." M.S. Thesis,
University at Buffalo, The State University of New York, Buffalo, New York.
Bolisetti, C., and Whittaker, A. S. (2011). "Seismic Structure-Soil-Structure Interaction in Nuclear Power
Plant Structures." 21st International Conference in Structural Mechanics in Reactor Technology (SMiRT-
21), New Delhi, India.
Boore, D. M. (1972). "A Note on the Effect of Simple Topography on Seismic SH Waves." Bulletin of the
Seismological Society of America, 62(1), 275-284.
Borja, R., and Amies, A. (1994). "Multiaxial Cyclic Plasticity Model for Clays." Journal of Geotechnical
Engineering, 120(6), 1051-1070.
Borja, R. I., Chao, H. Y., Montáns, F. J., and Lin, C. H. (1999). "Nonlinear Ground Response at Lotung
LSST Site." Journal of Geotechnical and Geoenvironmental Engineering, 125(3), 187-197.
Bray, J. (2011). Personal Communication
Çelebi, M. (1993a). "Seismic Responses of Two Adjacent Buildings. I: Data and Analyses." Journal of
Structural Engineering, 119(8), 2461-2476.
Çelebi, M. (1993b). "Seismic Responses of Two Adjacent Buildings. II: Interaction." Journal of
Structural Engineering, 119(8), 2477-2492.
232
Chang, C.-Y., Mok, C. M., Power, M. S., Tang, Y. K., Tang, H. T., and Stepp, J. C. (1990). "Equivalent
Linear and Nonlinear Ground Response Analyses at Lotung Seismic Experiment Site." Proceedings:
Fourth U.S. National Conference on Earthquake Engineering Earthquake Engineering Research Institute,
Palm Springs, California.
Chang, C.-Y., Mok, C. M., and Power, M. S. (1991). "Analysis of Ground Response Data at Lotung
Large-Scale Soil-Structure Interaction Experiment Site." Report NP-7306-M, Electric Power Research
Institute, Palo Alto, California.
Chang, S. W. (1996). "Seismic Response of Deep Stiff Soil Deposits." Ph.D. Dissertation, Unviersity of
California, Berkeley, California.
Chen, A. T. F., and Joyner, W. B. (1974). "Multi-linear Analysis for Ground Motion Studies of Layered
Systems." Report USGS-GD-74-020, United States Department of the Interior Geological Survey, Menlo
Park, California.
Chen, Z., Hutchinson, T. C., Trombetta, N. W., Mason, H. B., Bray, J. D., Jones, K. C., Bolisetti, C.,
Whittaker, A. S., Choy, B. Y., Kutter, B. L., Feigel, G. L., Montgomery, J., Patel, R. J., and Reitherman,
R. D. (2010). "Seismic Performance Assessment in Dense Urban Areas: Evaluation of Nonlinear
Building-Foundation Systems using Centrifuge Tests." Proceedings: Fifth International Conference on
Recent Advances in Geotechnical Earthquake Engineering and Soil Dynamics, May 24-29, 2010, San
Diego, California.
Chen, Z., Trombetta, N. W., Hutchinson, T. C., Mason, H. B., Bray, J. D., and Kutter, B. L. (2013).
"Seismic System Identification Using Centrifuge-Based Soil-Structure Interaction Test Data." Journal of
Earthquake Engineering, 17(4), 469-496.
Chopra, A. K. (1995). "Dynamics of Structures." Prentice Hall.
Computers and Structures Inc. (2011). Computer Program SAP2000 - Structural Analysis Program,
Version 11.0.0, Computers and Structures, Inc., Berkeley, California.
Constantopoulos, I. V., Roesset, J. M., and Christian, J. T. (1974). "A Comparison of Linear and Exact
Nonlinear Analyses of Soil Amplification." Proceedings: The Fifth World Conference in Earthquake
Engineering, Rome.
Coronado, C., Anderson, L. M., and Ostadan, F. (2013). "Verification of SASSI Extended Subtraction
Method for Computing the Seismic Response of Deeply Embedded Structures." 22nd International
Conference in Structural Mechanics in Reactor Technology - SMiRT 22, August 2013, San Francisco,
California.
233
Craig, R. F. (1997). "Soil Mechanics." E & FN Spon, New York.
Dassault Systèmes (2005). Computer Program ABAQUS - Finite Element Analysis Software, Dassault
Systèmes, Providence, Rhode Island.
Day, S. M., Bielak, J., Dreger, D., Graves, R., Larsen, S., Olsen, K. B., Pitarka, A., and Guzman, L. R.
(2006). "Numerical Simulation of Basin Effects on Long Period Ground Motion." Proceedings:
Proceedings of the 8th U.S. National Conference on Earthquake Engineering, San Francisco, California.
Dickenson, S. E. (1994). "Dynamic Response of Soft and Deep Cohesive Soils During the Loma Prieta
Earthquake of October 17, 1989." Ph.D. Dissertation, University of California, Berkeley, California.
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.
Elgamal, A., Yang, Z., Lai, T., Kutter, B. L., and Wilson, D. W. (2005). "Dynamic Response of Saturated
Dense Sand in Laminated Centrifuge Container." Journal of Geotechnical and Geoenvironmental
Engineering, 131(5), 598-609.
Elsabee, F., Kausel, E., and Roesset, J. M. (1977). "Dynamic Stiffness of Embedded Foundations."
Proceedings: Second Annual Engineering Mechanics Division Specialty Conference, ASCE, North
Carolina State University, Rayleigh, North Carolina, 40-43.
Federal Emergency Management Agency (FEMA). (1997). "NEHRP Commentary on the Guidelines for
Seismic Rehabilitation of Buildings." FEMA 274, Federal Emergency Management Agency, Washington,
District of Columbia.
Federal Emergency Management Agency (FEMA). (2005). "Improvement of Nonlinear Static Seismic
Analysis Procedures." FEMA 440, Federal Emergency Management Agency, Washington, District of
Columbia.
Federal Emergency Management Agency (FEMA). (2012). "Seismic Performance Assessment of

Buildings." FEMA P-58, Federal Emergency Management Agency, Washington, District of Columbia.
Finn, W. D. L., and Martin, G. R. (1978). "Comparison of Dynamic Analyses for Saturated Sand."
Proceedings: Earthquake Engineering and Soil Dynamics: Proceedings of the ASCE Geotechnical
Engineering Division Speicialy Conference, Pasadena, California, 472-491.
Ganuza, E. A. B. (2006). "Seismic Behavior of Hybrid Lateral-Force-Resisting Systems." M.S.

Dissertation, University at Buffalo, The State University of New York, Buffalo, New York.
234
Gazetas, G. (1991). "Foundation Vibrations." Foundation Engineering Handbook, Ed. H. Fang, Kluwer
Academic Publishers, Norwell, Massachusetts.
Hadjian, A. H. (1981). "On the Correlation of the Components of Strong Ground Motion - Part 2."
Bulletin of the Seismological Society of America, 71(4), 1323-1331.
Hardin, B. O., and Drnevich, V. P. (1972). "Shear Modulus and Damping in Soils: Measurement and
Parameter Effects." Journal of the Soil Mechanics and Foundations Division, ASCE, 98(8), 473-496.
Hashash, Y. M. A., and Park, D. (2001). "Nonlinear One-Dimensional Seismic Ground Motion
Propagation in the Mississippi Embayment." Engineering Geology, 62, 185-206.
Hashash, Y. M. A., Phillips, C., and Groholski, D. R. (2010). "Recent Advances in Nonlinear Site
Response Analysis." Fifth International Conference on Recent Advances in Geotechnical Earthquake
Engineering and Soil Dynamics, May 24-29, San Diego, California.
Hashash, Y. M. A., Groholski, D. R., Phillips, C. A., Park, D., and Musgrove, M. (2011). "DEEPSOIL
5.0 User Manual and Tutorial." 107 p.
Huang, Y.-N., Whittaker, A. S., Kennedy, R. P., and Mayes, R. L. (2009). "Assessment of Base-Isolated
Nuclear Structures for Design and Beyond-Design Basis Earthquake Shaking." Multidisciplinary Center
for Earthquake Engineering (MCEER), Buffalo, New York.
Huang, Y. (2013). Personal Communication
Iai, S., Morita, T., Kameoka, T., Matsunaga, Y., and Abiko, K. (1995). "Response of a Dense Sand
Deposit During 1993 Kushiro-Oki Earthquake." Soils and Foundations, Japanese Society of Soil
Mechanics and Foundation Engineering, 35(1), 115-131.
Idriss, I. M., and Seed, H. B. (1967a). "Response of Horizontal Soil Layers During Earthquakes." Soil
Mechanics and Bituminous Materials Research Laboratory, University of California, Berkeley, Berkeley,
California.
Idriss, I. M., and Seed, H. B. (1967b). "Response of Earth Banks During Earthquakes." Journal of the Soil
Mechanics and Foundations Division, ASCE, 93(SM3), 61-82.
Idriss, I. M., and Seed, H. B. (1968a). "An Analysis of Ground Motions During the 1957 San Francsico
Earthquake." Bulletin of the Seismological Society of America, 58(6), 2013-2032.
Idriss, I. M., and Seed, H. B. (1968b). "Seismic Response of Horizontal Soil Layers." Proceedings of
ASCE, Journal of Soil Mechanics and Foundations Division, 94, 1003-1031.
235
Idriss, I. M. (1990). "Response of Soft Soil Sites During Earthquakes." Proceedings: Proceedings
Memorial Symposium to Honor Porfessor Harry Bolton Seed, Berkeley, California.
Idriss, I. M., and Sun, J. I. (1992). "User's Manual for SHAKE91: A Computer Program for Conducting
Equivalent Linear Seismic Response Analyses of Horizontally Layered Soil Deposits."
Ilankatharan, M. (2008). "Centrifuge Modeling for Soil-Pile-Bridge Systems with Numerical Simulations
Accounting for Soil-Container-Shaker Interaction." Ph.D. Dissertation, University of California, Davis,
California.
Ilankatharan, M., and Kutter, B. (2010). "Modeling Input Motion Boundary Conditions for Simulations of
Geotechnical Shaking Table Tests." Earthquake Spectra, 26(2), 349-369.
Itasca Consulting Group (2005). Computer Program FLAC: Advanced, Two Dimensional Continuum
Modeling for Geotechnical Analysis of Rock, Soil, and Structural Support, Itasca Consulting Group, Inc.,
Minneapolis, Minnesota.
Iwan, W. D. (1967). "On a Class of Models for the Yielding Behavior of Continuous and Composite
Systems." Journal of Applied Mechanics, 34, 612-617.
Jennings, P. C., and Bielak, J. (1973). "Dynamics of Building-Soil Interaction." Bulletin of the
Jeremic, B., Kammerer, A., Tafazzoli, N., and Kamrani, B. (2011). "The NRC ESSI Simulator." 21st
International Conference in Structural Mechanics and Reactor Technology: SMiRT 21, New Delhi, India.
Jeremic, B., Jie, G., Preisig, M., and Tafazzoli, N. (2013). "The NRC ESSI Simulator Notes." Technical
Report, , University of California at Davis, and Lawrence Berkeley National Laboratory, Davis,
California.
Jeremić, B., Jie, G., Preisig, M., and Tafazzoli, N. (2009). "Time Domain Simulation of Soil-Foundation-
Structure Interaction in Non-Uniform Soils." Earthquake Engineering & Structural Dynamics, 38(5),
699-718.
Jones, K. C. (2013). "Dynamic Soil-Structure Interaction Analysis of Structures in Dense Urban

Environments." Ph.D. Dissertation, University of California, Berkeley, California.
Joyner, W. B. (1975). "A Method for Calculating Nonlinear Seismic Response in Two Dimensions."
Joyner, W. B., and Chen, A. T. F. (1975). "Calculation of Nonlinear Ground Response in Earthquakes."
236
Kanai, K. (1952). "Relation between the Nature of Surface Layer and the Amplitudes of Earthquake
Motions." Bulletin of Earthquake Research Institute, 30, 31-37.
Kausel, E., and Roesset, J. M. (1975). "Dynamic Stiffness of Circular Foundations." Journal of the
Engineering Mechanics Division, 101, EM6, 771-748.
Kausel, E. (2010). "Early History of Soil-Structure Interaction." Soil Dynamics and Earthquake
Engineering, 30, 822-832.
Kramer, S. L. (1996). "Geotechnical Earthquake Engineering." Prentice Hall, Upper Saddle River, New
Jersey.
Kramer, S. L., and Stewart, J. P. (2004). "Geotechnical Aspects of Seismic Hazards." Earthquake
Engineering From Engineering Seismology to Performance-Based Engineering, eds. Y. Bozorgnia, and
V. V. Bertero, CRC Press.
Kulhawy, F. H., and Mayne, P. W. (1990). "Manual on Estimating the Soil Properties for Foundation
Design." Report EL-6800, Electric Power Research Institute, Palo Alto, California.
Kutter, B. L. (1995). "Recent Advances in Centrifuge Modeling of Seismic Shaking." Proceedings: Third
International Conference on Recent Advances in Geotechnical Earthquake Engineering and Soil
Dynamics, April 2-7, St. Louis, Missouri.
Kwok, A. O. L., Stewart, J. P., and Hashash, Y. M. A. (2008). "Nonlinear Ground-Response Analysis of
Turkey Flat Shallow Stiff-Soil Site to Strong Ground Motion." Bulletin of the Seismological Society of
America, 98(1), 331-343.
Kwok, O. L. (2007). "Application of One-Dimensional Geotechnical Modeling for Site Response

Predictions." Ph.D. Dissertation, University of California, Los Angeles, California.
Lee, M. K. W., and Finn, W. (1978). "DESRA-2: Dynamic Effective Stress Response Analysis of Soil
Deposits with Energy Transmitting Boundary Including Assessment of Liquefaction Potential." The
University of British Columbia, Faculty of Applied Science, Vancouver, British Columbia.
Li, X. S., Z., W., and Shen, C. K. (1992). "SUMDES: A Nonlinear Procedure for Response Analysis of
Horizontally-Layered Sites Subjected to Multi-Directional Earthquake Loading, Manual." Department of
Civil Engineering, University of California, Davis, California.
Lin, H., -T., Roesset, J. M., and Tassoulas, J. L. (1987). "Dynamic Interaction Between Adjacent
Foundations." Earthquake Engineering and Structural Dynamics, 15, 323-343.
237
Livermore Software Technology Corporation (LSTC). (2013). "LS-DYNA Keyword User's Manual -
Version R 7.0." Livermore Software Technology Corporation, Livermore, California.
Luco, J. E., and Westman, R. A. (1971). "Dynamic Response of Circular Footings." Journal of the
Engineering Mechanics Division, 97, EM5, 1381-1395.
Luco, J. E., and Contesse, L. (1973). "Dynamic Structure-Soil-Structure Interaction." Bulletin of the
Lysmer, J., and Kuhlemeyer, R. L. (1969). "Finite Dynamic Model for Infinite Media." Journal of
Engineering Mechanics Division, ASCE, 95, EM4, 859-877.
Lysmer, J. (1978). "Analytical Procedures in Soil Dynamics." Report UCB/EERC-78/29, Earthquake

Engineering Research Center, University of California, Berkeley, California.
Lysmer, J., Ostadan, F., and Chin, C. (1999). Computer Program SASSI2000 - A System for Analysis of
Soil-Structure Interaction, University of California, Berkeley, California.
Masing, G. (1926). "Eignespannungen und Verfestigung beim Messing." Proceedings: 2nd International
Congress on Applied Mechanics, Zurich, Switzerland, 332-335.
Mason, H. B., Trombetta, N. W., Gille, S., Lund, J., Zupan, J., Jones, K. C., Puangnak, H., Bolisetti, C.,
Bray, J., Hutchinson, T., Fiegel, G., Kutter, B. L., and Whittaker, A. S. (2010a). "Seismic Performance
Assessment in Dense Urban Environments: Centrifuge Data Report for HBM02 (Test 1)." University of
California at Davis Center for Geotechnical Modeling, Davis, California.
Bray, J., Hutchinson, T., Fiegel, G., Kutter, B. L., and Whittaker, A. S. (2010b). "Seismic Performance
Assessment in Dense Urban Environments: Centrifuge Data Report for HBM03 (Test 2)." University of
California at Davis Center for Geotechnical Modeling, Davis, California.
Bray, J., Hutchinson, T., Fiegel, G., Kutter, B. L., and Whittaker, A. S. (2010c). "Seismic Performance
Assessment in Dense Urban Environments: Centrifuge data report for HBM04 (Test 3)." Report in
preparation, University of California at Davis Center for Geotechnical Modeling, Davis, California.
Mason, H. B. (2011). "Seismic Performance Assessment in Dense Urban Environments." Ph.D.

Dissertation, University of California, Berkeley, California.
Bray, J., Hutchinson, T., Fiegel, G., Kutter, B. L., and Whittaker, A. S. (2011). "Seismic Performance
238
Assessment in Dense Urban Environments: Centrifuge Data Report for HBM04 (Test 3)." Report No.
UCD/CGMDR-03/11, University of California at Davis Center for Geotechnical Modeling, Davis,
California.
Mason, H. B., Trombetta, N. W., Chen, Z., Bray, J. D., Hutchinson, T. C., and Kutter, B. L. (2013).
"Seismic Soil–Foundation–Structure Interaction Observed in Geotechnical Centrifuge Experiments." Soil
Dynamics and Earthquake Engineering, 48, 162-174.
Mason, H. B. (2014). Personal Communication
Matasovic, N. (1993). "Seismic Response of Composite Horizontally-Layered Soil Deposits." Ph. D.

Dissertation, University of California, Los Angeles, California.
Matasovic, N. (2006). "D-MOD_2: A Computer Program for Seismic Response Analysis of Horizontally
Layered Soil Deposits, Earthfill Dams and Solid Waste Landfills." Geomotions, LLC, Lacey, WA.
Matthees, W., and Magiera, G. (1982). "A Sensitivity Study of Seismic Structure-Soil-Structure
Interaction Problems for Nuclear Power Plants." Nuclear Engineering and Design, 73, 343-363.
Mazzoni, S., McKenna, F., Scott, M. H., and Fenves, G. L. (2009). Computer Program OpenSees: Open
System for Earthquake Engineering Simulation, Pacific Earthquake Engineering Research Center
(PEER), University of California, Berkeley, California.
McKenna, F., and Fenves, G. L. (2001). "The OpenSees Command Language Manual: Version 1.2."
Pacific Earthquake Engineering Research (PEER) Center, University of California, Berkeley, California.
Menglin, L., Huaifeng, W., Xi, C., and Yongmei, Z. (2011). "Structure-Soil-Structure Interaction:
Literature Review." Soil Dynamics and Earthquake Engineering, 31, 1724-1731.
Mertz, G., Cuesta, I., Maham, A., and Costantino, M. (2010). "Seismic Response of Embedded Facilities
Using the SASSI Subtraction Method." Los Alamos National Laboratory, Los Alamos, New Mexico.
Murakami, H., and Luco, J. E. (1977). "Seismic Response of a Periodic Array of Structures." Journal of
the Engineering Mechanics Division, ASCE SM5, 103, 965-976.
Muravskii, G. (2005). "On Description of Hysteretic Behaviour of Materials." International Journal of

Solids and Structures, 42(9–10), 2625-2644.
Mylonakis, G., and Gazetas, G. (2000). "Seismic Soil-Structure Interaction: Beneficial or Detrimental?."
Journal of Earthquake Engineering, 4(3), 277-301.
239
National Institute of Standards and Technology (NIST). (2012). "Soil-Structure Interaction for Building
Structures." NIST GCR 12-917-21, National Institute of Standards and Technology, Gaithersburg,
Maryland.
Oberkampf, W. L., and Trucano, T. G. (2008). "Verification and Validation Benchmarks." Nuclear
Engineering and Design, 238(3), 716-743.
Ostadan, F. (2006a). "SASSI2000: A System for Analysis of Soil Structure Interaction - Theoretical
Manual." University of California, Berkeley, California.
Ostadan, F. (2006b). "SASSI2000: A System for Analysis of Soil Structure Interaction - User's Manual."
University of California, Berkeley, California.
Ostadan, F. (2010). Personal Communication
Ostadan, F., and Deng, N. (2011). Computer Program SASSI2010 - A System for Analysis of Soil-
Structure Interaction, Version 1.1, Geotechnical and Hydraulic Engineering Services, Bechtel National
Inc., San Francisco, California.
Padron, L. A., Aznarez, J. J., and Maeso, O. (2009). "Dynamic Structure-Soil-Structure Interaction
between Nearby Piled Buildings under Seismic Excitation by BEM-FEM Model." Soil Dynamics and
Earthquake Engineering, 29, 1084-1096.
PEER (2011). "PEER Ground Motion Database." <http://peer.berkeley.edu/peer_ground_motion

_database>. (25th November, 2012).
Phillips, C., and Hashash, Y. M. A. (2009). "Damping Formulation for Nonlinear 1D Site Response
Analyses." Soil Dynamics & Earthquake Engineering, 29, 1143-1158.
Porcella, R. L. (1980). "Atypical Accelerograms Recorded During Recent Earthquakes." Geological

Survey Circular, Report 854-B, United States Department of the Interior, Menlo Park, California, 1-7.
Pyke, R. M. (2000). "TESS Users' Manual." TAGA Engineering Software Services, Lafayette, California.
Qian, J., and Beskos, D. E. (1996). "Harmonic Wave Response of Two 3-D Rigid Surface Foundations."
Soil Dynamics and Earthquake Engineering, 15, 95-110.
Ramberg, W., and Osgood, W. (1943). "Description of Stress-Strain Curves by Three Parameters."
Technical Note No. 902, National Advisory Committee of Aeronautics, Washington, District of
Columbia.
240
Restrepo, D., Taborda, R., and Bielak, J. (2011). "Effects of Soil Nonlinearity on Ground Response in 3D
Simulations - An Application to the Salt Lake City Basin." Proceedings: 4th IASPEI / IAEE International
Symposium, University of California, Santa Barbara.
Roesset, J. M. (1998). "Seismic Design of Nuclear Power Plants - Where Are We Now?" Nuclear
Engineering and Design, 182, 3-15.
Ryan, H. (1994). "Ricker, Ormsby, Klander, Butterworth - A Choice of Wavelets." Canadian Society of
Exploration Geophysicists Recorder, 19(7), 8-9.
Schnabel, P. B., Lysmer, J., and Seed, H. B. (1972). "SHAKE: A Computer Program for Earthquake
Ground Response Analysis for Horizontally Layered Sites." Earthquake Engineering Research Center,
University of California, Berkeley, California.
Schnabel, P. B., Lysmer, J., and Seed, H. B. (2009). Computer Program SHAKE2000: A Computer
Program for the 1D Analysis of Geotechnical Earthquake Engineering Problems, University of California,
Berkeley, California.
Schnabel, P. B., Lysmer, J., and Seed, H. B. (2012). Computer Program SHAKE: A Computer Program
for Earthquake Response Analysis of Horizontally Layered Sites, University of California, Berkeley,
California.
Seed, H. B., and Idriss, I. M. (1969). "Influence of Soil Conditions on Ground Motions during
Earthquakes." Proceedings of the ASCE, Journal of the Soil Mechanics and Foundations Division, SM1,
99-137.
Seed, H. B., and Idriss, M. (1970). "Soil Moduli and Damping Factors for Dynamic Response Analysis."
Report UCB/EERC 70/10, Earthquake Engineering Research Center, University of California, Berkeley,
California.
Somerville, P. G., Smith, R. W., Graves, R. W., and Abrahamson, N. A. (1997). "Modification of
Empirical Strong Ground Motion Attentuation Relations to Include the Amplitude and Duration Effects
of Rupture Directivity." Seismological Research Letters, 68, 199-222.
Stewart, J. P., Fenves, G. L., and Seed, R. B. (1999a). "Seismic Soil-Structure Interaction in Buildings. I:
Analytical Aspects." Journal of Geotechnical and Geoenvironmental Engineering, 125, GEOBASE, 26-
37.
Stewart, J. P., Seed, R. B., and Fenves, G. L. (1999b). "Seismic Soil-Structure Interaction in Buildings. II:
Empirical Findings." Journal of Geotechnical and Geoenvironmental Engineering, 125, GEOBASE, 38-
48.
241
Stewart, J. P., and Kwok, A. O. L. (2008). "Nonlinear Seismic Ground Response Analysis: Code Usage
Protocols and Verification Against Vertical Array Data." Proceedings: Geotechnical Engineering and
Soil Dynamics IV, 18-22 May, Sacramento, California.
Taborda, R. (2010). "Three Dimensional Nonlinear Soil and Site-City Effects in Urban Regions." Ph.D.
Dissertation, Carnegie Mellon University.
Taborda, R., and Bielak, J. (2011). "Large-Scale Earthquake Simulation: Computational Seismology and
Complex Engineering Systems." Computing in Science and Engineering, 13(4), 14-27.
Trombetta, N. W., Zupan, J., Bolisetti, C., Puangnak, H., Jones, K. C., Tran, J. K., Bassal, P., Bray, J.,
Hutchinson, T., Fiegel, G., Kutter, B. L., and Whittaker, A. S. (2011). "Seismic Performance Assessment
in Dense Urban Environments: Centrifuge Data Report for NWT01 (Test 4)." Report in preparation,
University of California at Davis Center for Geotechnical Modeling, Davis, California.
Trombetta, N. W. (2013). "Seismic Soil-Foundation Structure Interaction in Urban Environments." Ph.D.

Dissertation, University of California, San Diego, California.
Trombetta, N. W., Mason, H. B., Chen, Z., Hutchinson, T. C., Bray, J. D., and Kutter, B. L. (2013a).
"Nonlinear Dynamic Foundation and Frame Structure Response Observed in Geotechnical Centrifuge
Experiments." Soil Dynamics and Earthquake Engineering, 50, 117-133.
Trombetta, N. W., Mason, H. B., Hutchinson, T. C., Zupan, J. D., Bray, J. D., and Kutter, B. L. (2013b).
"Nonlinear Soil-Foundation-Structure and Structure-Soil-Structure Interaction: Centrifuge Test
Observations." ASCE Journal for Geotechnical and Geoenvironmental Engineering, In Review.
Veletsos, A. S., and Wei, Y. T. (1971). "Lateral and Rocking Vibration of Footings." Journal of the ASCE
Soil Mechanics and Foundations Division 97, SM9, 1227-1248.
Veletsos, A. S., and Verbic, B. (1973). "Vibration of Viscoelastic Foundations." Earthquake Engineering
and Structural Dynamics, 2(1), 87-105.
Veletsos, A. S., and Meek, J. W. (1974). "Dynamic Behavior of Building Foundation Systems."
Earthquake Engineering and Structural Dynamics, 3(2), 121-138.
Veletsos, A. S., and Verbic, B. (1974). "Basic Response Functions for Elastic Foundations." Journal of
the ASCE Engineering Mechanics Division, 100, EM2, 189-205.
Wolf, J. P. (1985). "Dynamic Soil-Structure Interaction." Prentice Hall, Englewood Cliffs, New Jersey.
242
Wong, H. L., and Trifunac, M. D. (1975). "Two-Dimensional, Antiplane, Building-Soil-Building
Interaction for Two or More Buildings and for Incident Plane SH Waves." Bulletin of the Seismological
Society of America, 65(6), 1863-1885.
Xu, J. (1998). "Three-Dimensional Simulation of Wave Propagation in Inelastic Media on Parallel

Computers with Application to Seismic Response." Ph.D. Dissertation, Carnegie Mellon University,
Pittsburgh, Pennsylvania.
Xu, J., Bielak, J., Ghattas, O., and Wang, J. (2003). "Three-Dimensional Nonlinear Seismic Ground
Motion Modeling in Basins." Physics of the Earth and Planetary Interiors, 137, 81-95.
Xu, J., Miller, C., Costantino, C., and Hofmayer, C. (2006). "Assessment of Seismic Analysis
Methodologies for Deeply Embedded Nuclear Power Plant Structures." U.S. Nuclear Regulatory
Commission, Washington, District of Columbia
Yang, Z., Lu, J., and Elgamal, A. (2004). "A Web-Based Platform for Computer Simulation of Seismic
Ground Response." Advances in Engineering Software, 35, 249-259.
Yoshida, N., Kobayashi, S., Suetomi, I., and Muira, K. (2002). "Equivalent Linear Method Considering
Frequency Dependent Characteristics of Stiffness and Damping." Soil Dynamics & Earthquake
Engineering, 22, 205-222.
Yoshimura, C., Bielak, J., and Hisada, Y. (2003). "Domain Reduction Method for Three-Dimensional
Earthquake Modeling in Localized Regions, Part-II: Verification and Examples." Bulletin of the
Youngs, R. R. (2004). "Software Validation Report for SHAKE04." Geomatrix Consultants, Oakland,
California.
243
APPENDIX A
RESULTS OF SITE-RESPONSE ANALYSES
Results of the site-response analyses performed using SHAKE, DEEPSOIL and LS-DYNA are
presented in this appendix. The analysis models and procedures are described in Section 4. Responses are
presented in the form of acceleration response spectra, acceleration response histories, stress-strain
hysteresis loops of the soil layers, and the peak acceleration and peak strain profiles. The outcrop motion
at the base, which is used to calculate the input shear forces for the soil profile, is also included in the
acceleration response spectra for comparison. Acceleration amplification spectra from the base to the
surface of the soil profile are presented for each ground motion. The acceleration response spectra,
response histories and the stress strain loops are calculated at three different depths in the soil profile, at
the depths indicated in each figure. Since the acceleration responses are calculated at the interfaces of the
soil layers, and stress-strain loops are calculated at the mid-points of the soil layers, the depths at which
the corresponding data are presented are slightly different. In the following pages, the results for site E1,
site E2, site W1 and site W2 are presented in Figure A-1 to Figure A-13, Figure A-14 to Figure A-26,
Figure A-27 to Figure A-52, and Figure A-53 to Figure A-78, respectively. The results for each site begin
with the stress-strain loops for all ground motions, followed by the peak acceleration and peak strain
profiles, acceleration response spectra, acceleration response histories and the spectral amplifications at
the surface.
245
Figure A-1: Stress-strain loops at three different depths for site E1 subjected to ground motion EO1
246
247
248
Figure A-4: Peak acceleration and peak strain profiles for site E1 subjected to ground motion EO1
249
250
251
different depths for site E1 subjected to ground motion EO1
252
253
254
Figure A-10: Acceleration response histories at three different depths for site E1 subjected to
ground motion EO1
255
ground motion EO2
256
ground motion EO3
257
Figure A-13: Spectral amplification of accelerations from the base to the surface of site E1 for
ground motions EO1, EO2 and EO3
258
EO1
259
EO2
260
EO3
261
262
263
264
265
266
267
ground motion EO1
268
ground motion EO2
269
ground motion EO3
270
Figure A-26: Spectral amplification of accelerations in site E2 from the base to the surface for
ground motions EO1, EO2 and EO3
271
WO1
272
WO2
273
WO3
274
WO1
275
WO2
276
WO3
277
different depths for site W1 subjected to ground motion WO1
278
279
280
Figure A-36: Acceleration response histories at three different depths for site W1 subjected to
ground motion WO1
281
ground motion WO2
282
ground motion WO3
283
Figure A-39: Spectral amplification of accelerations in site W1 from the base to the surface for
ground motions WO1, WO2 and WO3
284
WP1
285
WP2
286
WP3
287
WP1
288
WP2
289
WP3
290
different depths for site W1 subjected to ground motion WP1
291
292
293
ground motion WP1
294
ground motion WP2
295
ground motion WP3
296
ground motions WP1, WP2 and WP3
297
WO1
298
WO2
299
WO3
300
WO1
301
WO2
302
WO3
303
304
305
306
ground motion WO1
307
ground motion WO2
308
ground motion WO3
309
ground motions WO1, WO2 and WO3
310
WP1
311
WP2
312
WP3
313
WP1
314
WP2
315
WP3
316
Figure A-72: Acceleration response spectra of the input motion, and accelerations at three different
depths for site W2 subjected to ground motion WP1
317
318
319
ground motion WP1 (unequal scales used for clarity)
320
ground motion WP2 (unequal scales used for clarity)
321
ground motion WP3
322
ground motions WP1, WP2 and WP3
323
APPENDIX B
THE ORMSBY WAVELET
Earthquake ground motions comprise a spectrum of frequencies. The contribution of each

frequency to the total energy carried by the ground motion is different, and most ground motions have a
dominant frequency range. These dominant frequencies depend on several factors, such as the site where
the ground motion is recorded, directivity of the earthquake, etc. Typical ground motions therefore may
not have a considerable contribution from all the frequencies that are relevant to a dynamic system, and
may result in overlooking some of the frequencies.
In order to examine and investigate the response of a complex dynamic system, such as a soil-
foundation-structure system, it is desirable to use an input ground motion that has a considerable
contribution from all frequencies in the desired range. The Ormsby wavelet (Ryan, 1994) provides a
convenient alternative to real ground motions for understanding the response of complex dynamic
systems. The Ormsby wavelet is a wavelet that features a Fourier transform with a constant magnitude in
a specified frequency range. This wavelet is defined by the expression (Ryan, 1994)
  f 42  f32 2
 sinc( f 4 (t - ts ))   sinc( f3 (t - ts ))  
2
f (t )  
 f 4  f3 f 4  f3 
(B-1)
  f 22  f12 2
 sinc( f 2 (t - ts ))   sinc( f1 (t - ts )) 
2

 f 2  f1 f 2  f1 
where, f1 and f 2 define the lower frequency band, f 3 and f 4 define the upper frequency band, t s is the
time of the maximum amplitude, f (t ) is the amplitude of the wave at time t , and sinc( x) is defined by
sin( x)
sinc( x)  (B-2)
x
Figure B-1 presents a sample Ormsby wavelet with the frequency parameters, f1 , f 2 , f 3 and
f 4 , equal to 0 Hz, 0.2 Hz, 20 Hz and 25 Hz, respectively, the time of maximum amplitude equal to 2 sec,
and normalized to a maximum amplitude of 1. Figure B-2 presents the Fourier transform of this wavelet.
The Fourier amplitude of this wavelet increases linearly from f1 to f 2 , remains constant from f 2 to f 3 ,
and decreases linearly from f 3 to f 4 . Figure B-3 presents the 1% damped acceleration response
spectrum for this wavelet.
325
Figure B-1: Time series of an Ormsby wavelet
Figure B-2: Fourier transform of an Ormsby wavelet
Figure B-3: 1% damped acceleration response spectrum of an Ormsby wavelet
326
MCEER Technical Reports
MCEER publishes technical reports on a variety of subjects written by authors funded through MCEER. These reports are
available from both MCEER Publications and the National Technical Information Service (NTIS). Requests for reports should
be directed to MCEER Publications, MCEER, University at Buffalo, State University of New York, 133A Ketter Hall, Buffalo,
New York 14260. Reports can also be requested through NTIS, P.O. Box 1425, Springfield, Virginia 22151. NTIS accession
numbers are shown in parenthesis, if available.
NCEER-87-0001 "First-Year Program in Research, Education and Technology Transfer," 3/5/87, (PB88-134275, A04, MF-
A01).
NCEER-87-0002 "Experimental Evaluation of Instantaneous Optimal Algorithms for Structural Control," by R.C. Lin, T.T.
Soong and A.M. Reinhorn, 4/20/87, (PB88-134341, A04, MF-A01).
NCEER-87-0003 "Experimentation Using the Earthquake Simulation Facilities at University at Buffalo," by A.M. Reinhorn
and R.L. Ketter, not available.
NCEER-87-0004 "The System Characteristics and Performance of a Shaking Table," by J.S. Hwang, K.C. Chang and G.C.
Lee, 6/1/87, (PB88-134259, A03, MF-A01). This report is available only through NTIS (see address given
above).
NCEER-87-0005 "A Finite Element Formulation for Nonlinear Viscoplastic Material Using a Q Model," by O. Gyebi and G.
Dasgupta, 11/2/87, (PB88-213764, A08, MF-A01).
NCEER-87-0006 "Symbolic Manipulation Program (SMP) - Algebraic Codes for Two and Three Dimensional Finite Element
Formulations," by X. Lee and G. Dasgupta, 11/9/87, (PB88-218522, A05, MF-A01).
NCEER-87-0007 "Instantaneous Optimal Control Laws for Tall Buildings Under Seismic Excitations," by J.N. Yang, A.
Akbarpour and P. Ghaemmaghami, 6/10/87, (PB88-134333, A06, MF-A01). This report is only available
through NTIS (see address given above).
NCEER-87-0008 "IDARC: Inelastic Damage Analysis of Reinforced Concrete Frame - Shear-Wall Structures," by Y.J. Park,
A.M. Reinhorn and S.K. Kunnath, 7/20/87, (PB88-134325, A09, MF-A01). This report is only available
NCEER-87-0009 "Liquefaction Potential for New York State: A Preliminary Report on Sites in Manhattan and Buffalo," by
M. Budhu, V. Vijayakumar, R.F. Giese and L. Baumgras, 8/31/87, (PB88-163704, A03, MF-A01). This
report is available only through NTIS (see address given above).
NCEER-87-0010 "Vertical and Torsional Vibration of Foundations in Inhomogeneous Media," by A.S. Veletsos and K.W.
Dotson, 6/1/87, (PB88-134291, A03, MF-A01). This report is only available through NTIS (see address
given above).
NCEER-87-0011 "Seismic Probabilistic Risk Assessment and Seismic Margins Studies for Nuclear Power Plants," by Howard
H.M. Hwang, 6/15/87, (PB88-134267, A03, MF-A01). This report is only available through NTIS (see
address given above).
NCEER-87-0012 "Parametric Studies of Frequency Response of Secondary Systems Under Ground-Acceleration Excitations,"
by Y. Yong and Y.K. Lin, 6/10/87, (PB88-134309, A03, MF-A01). This report is only available through
NTIS (see address given above).
NCEER-87-0013 "Frequency Response of Secondary Systems Under Seismic Excitation," by J.A. HoLung, J. Cai and Y.K.
Lin, 7/31/87, (PB88-134317, A05, MF-A01). This report is only available through NTIS (see address given
above).
NCEER-87-0014 "Modelling Earthquake Ground Motions in Seismically Active Regions Using Parametric Time Series
Methods," by G.W. Ellis and A.S. Cakmak, 8/25/87, (PB88-134283, A08, MF-A01). This report is only
available through NTIS (see address given above).
NCEER-87-0015 "Detection and Assessment of Seismic Structural Damage," by E. DiPasquale and A.S. Cakmak, 8/25/87,
(PB88-163712, A05, MF-A01). This report is only available through NTIS (see address given above).
327
NCEER-87-0016 "Pipeline Experiment at Parkfield, California," by J. Isenberg and E. Richardson, 9/15/87, (PB88-163720,
A03, MF-A01). This report is available only through NTIS (see address given above).
NCEER-87-0017 "Digital Simulation of Seismic Ground Motion," by M. Shinozuka, G. Deodatis and T. Harada, 8/31/87,
(PB88-155197, A04, MF-A01). This report is available only through NTIS (see address given above).
NCEER-87-0018 "Practical Considerations for Structural Control: System Uncertainty, System Time Delay and Truncation of
Small Control Forces," J.N. Yang and A. Akbarpour, 8/10/87, (PB88-163738, A08, MF-A01). This report is
only available through NTIS (see address given above).
NCEER-87-0019 "Modal Analysis of Nonclassically Damped Structural Systems Using Canonical Transformation," by J.N.
Yang, S. Sarkani and F.X. Long, 9/27/87, (PB88-187851, A04, MF-A01).
NCEER-87-0020 "A Nonstationary Solution in Random Vibration Theory," by J.R. Red-Horse and P.D. Spanos, 11/3/87,
(PB88-163746, A03, MF-A01).
NCEER-87-0021 "Horizontal Impedances for Radially Inhomogeneous Viscoelastic Soil Layers," by A.S. Veletsos and K.W.
Dotson, 10/15/87, (PB88-150859, A04, MF-A01).
NCEER-87-0022 "Seismic Damage Assessment of Reinforced Concrete Members," by Y.S. Chung, C. Meyer and M.
Shinozuka, 10/9/87, (PB88-150867, A05, MF-A01). This report is available only through NTIS (see address
given above).
NCEER-87-0023 "Active Structural Control in Civil Engineering," by T.T. Soong, 11/11/87, (PB88-187778, A03, MF-A01).
NCEER-87-0024 "Vertical and Torsional Impedances for Radially Inhomogeneous Viscoelastic Soil Layers," by K.W. Dotson
and A.S. Veletsos, 12/87, (PB88-187786, A03, MF-A01).
NCEER-87-0025 "Proceedings from the Symposium on Seismic Hazards, Ground Motions, Soil-Liquefaction and Engineering
Practice in Eastern North America," October 20-22, 1987, edited by K.H. Jacob, 12/87, (PB88-188115, A23,
MF-A01). This report is available only through NTIS (see address given above).
NCEER-87-0026 "Report on the Whittier-Narrows, California, Earthquake of October 1, 1987," by J. Pantelic and A.
Reinhorn, 11/87, (PB88-187752, A03, MF-A01). This report is available only through NTIS (see address
given above).
NCEER-87-0027 "Design of a Modular Program for Transient Nonlinear Analysis of Large 3-D Building Structures," by S.
Srivastav and J.F. Abel, 12/30/87, (PB88-187950, A05, MF-A01). This report is only available through NTIS
(see address given above).
NCEER-87-0028 "Second-Year Program in Research, Education and Technology Transfer," 3/8/88, (PB88-219480, A04, MF-
A01).
NCEER-88-0001 "Workshop on Seismic Computer Analysis and Design of Buildings With Interactive Graphics," by W.
McGuire, J.F. Abel and C.H. Conley, 1/18/88, (PB88-187760, A03, MF-A01). This report is only available
NCEER-88-0002 "Optimal Control of Nonlinear Flexible Structures," by J.N. Yang, F.X. Long and D. Wong, 1/22/88, (PB88-
213772, A06, MF-A01).
NCEER-88-0003 "Substructuring Techniques in the Time Domain for Primary-Secondary Structural Systems," by G.D.
Manolis and G. Juhn, 2/10/88, (PB88-213780, A04, MF-A01).
NCEER-88-0004 "Iterative Seismic Analysis of Primary-Secondary Systems," by A. Singhal, L.D. Lutes and P.D. Spanos,
2/23/88, (PB88-213798, A04, MF-A01).
NCEER-88-0005 "Stochastic Finite Element Expansion for Random Media," by P.D. Spanos and R. Ghanem, 3/14/88, (PB88-
213806, A03, MF-A01).
328
NCEER-88-0006 "Combining Structural Optimization and Structural Control," by F.Y. Cheng and C.P. Pantelides, 1/10/88,
(PB88-213814, A05, MF-A01).
NCEER-88-0007 "Seismic Performance Assessment of Code-Designed Structures," by H.H-M. Hwang, J-W. Jaw and H-J.
Shau, 3/20/88, (PB88-219423, A04, MF-A01). This report is only available through NTIS (see address given
above).
NCEER-88-0008 "Reliability Analysis of Code-Designed Structures Under Natural Hazards," by H.H-M. Hwang, H. Ushiba
and M. Shinozuka, 2/29/88, (PB88-229471, A07, MF-A01). This report is only available through NTIS (see
NCEER-88-0009 "Seismic Fragility Analysis of Shear Wall Structures," by J-W Jaw and H.H-M. Hwang, 4/30/88, (PB89-
102867, A04, MF-A01).
NCEER-88-0010 "Base Isolation of a Multi-Story Building Under a Harmonic Ground Motion - A Comparison of
Performances of Various Systems," by F-G Fan, G. Ahmadi and I.G. Tadjbakhsh, 5/18/88, (PB89-122238,
A06, MF-A01). This report is only available through NTIS (see address given above).
NCEER-88-0011 "Seismic Floor Response Spectra for a Combined System by Green's Functions," by F.M. Lavelle, L.A.
Bergman and P.D. Spanos, 5/1/88, (PB89-102875, A03, MF-A01).
NCEER-88-0012 "A New Solution Technique for Randomly Excited Hysteretic Structures," by G.Q. Cai and Y.K. Lin,
5/16/88, (PB89-102883, A03, MF-A01).
NCEER-88-0013 "A Study of Radiation Damping and Soil-Structure Interaction Effects in the Centrifuge," by K. Weissman,
supervised by J.H. Prevost, 5/24/88, (PB89-144703, A06, MF-A01).
NCEER-88-0014 "Parameter Identification and Implementation of a Kinematic Plasticity Model for Frictional Soils," by J.H.
Prevost and D.V. Griffiths, not available.
NCEER-88-0015 "Two- and Three- Dimensional Dynamic Finite Element Analyses of the Long Valley Dam," by D.V.
Griffiths and J.H. Prevost, 6/17/88, (PB89-144711, A04, MF-A01).
NCEER-88-0016 "Damage Assessment of Reinforced Concrete Structures in Eastern United States," by A.M. Reinhorn, M.J.
Seidel, S.K. Kunnath and Y.J. Park, 6/15/88, (PB89-122220, A04, MF-A01). This report is only available
NCEER-88-0017 "Dynamic Compliance of Vertically Loaded Strip Foundations in Multilayered Viscoelastic Soils," by S.
Ahmad and A.S.M. Israil, 6/17/88, (PB89-102891, A04, MF-A01).
NCEER-88-0018 "An Experimental Study of Seismic Structural Response With Added Viscoelastic Dampers," by R.C. Lin, Z.
Liang, T.T. Soong and R.H. Zhang, 6/30/88, (PB89-122212, A05, MF-A01). This report is available only
NCEER-88-0019 "Experimental Investigation of Primary - Secondary System Interaction," by G.D. Manolis, G. Juhn and
A.M. Reinhorn, 5/27/88, (PB89-122204, A04, MF-A01).
NCEER-88-0020 "A Response Spectrum Approach For Analysis of Nonclassically Damped Structures," by J.N. Yang, S.
Sarkani and F.X. Long, 4/22/88, (PB89-102909, A04, MF-A01).
NCEER-88-0021 "Seismic Interaction of Structures and Soils: Stochastic Approach," by A.S. Veletsos and A.M. Prasad,
7/21/88, (PB89-122196, A04, MF-A01). This report is only available through NTIS (see address given
above).
NCEER-88-0022 "Identification of the Serviceability Limit State and Detection of Seismic Structural Damage," by E.
DiPasquale and A.S. Cakmak, 6/15/88, (PB89-122188, A05, MF-A01). This report is available only through
NTIS (see address given above).
NCEER-88-0023 "Multi-Hazard Risk Analysis: Case of a Simple Offshore Structure," by B.K. Bhartia and E.H. Vanmarcke,
7/21/88, (PB89-145213, A05, MF-A01).
329
NCEER-88-0024 "Automated Seismic Design of Reinforced Concrete Buildings," by Y.S. Chung, C. Meyer and M.
Shinozuka, 7/5/88, (PB89-122170, A06, MF-A01). This report is available only through NTIS (see address
given above).
NCEER-88-0025 "Experimental Study of Active Control of MDOF Structures Under Seismic Excitations," by L.L. Chung,
R.C. Lin, T.T. Soong and A.M. Reinhorn, 7/10/88, (PB89-122600, A04, MF-A01).
NCEER-88-0026 "Earthquake Simulation Tests of a Low-Rise Metal Structure," by J.S. Hwang, K.C. Chang, G.C. Lee and
R.L. Ketter, 8/1/88, (PB89-102917, A04, MF-A01).
NCEER-88-0027 "Systems Study of Urban Response and Reconstruction Due to Catastrophic Earthquakes," by F. Kozin and
H.K. Zhou, 9/22/88, (PB90-162348, A04, MF-A01).
NCEER-88-0028 "Seismic Fragility Analysis of Plane Frame Structures," by H.H-M. Hwang and Y.K. Low, 7/31/88, (PB89-
131445, A06, MF-A01).
NCEER-88-0029 "Response Analysis of Stochastic Structures," by A. Kardara, C. Bucher and M. Shinozuka, 9/22/88, (PB89-
174429, A04, MF-A01).
NCEER-88-0030 "Nonnormal Accelerations Due to Yielding in a Primary Structure," by D.C.K. Chen and L.D. Lutes,
9/19/88, (PB89-131437, A04, MF-A01).
NCEER-88-0031 "Design Approaches for Soil-Structure Interaction," by A.S. Veletsos, A.M. Prasad and Y. Tang, 12/30/88,
(PB89-174437, A03, MF-A01). This report is available only through NTIS (see address given above).
NCEER-88-0032 "A Re-evaluation of Design Spectra for Seismic Damage Control," by C.J. Turkstra and A.G. Tallin, 11/7/88,
(PB89-145221, A05, MF-A01).
NCEER-88-0033 "The Behavior and Design of Noncontact Lap Splices Subjected to Repeated Inelastic Tensile Loading," by
V.E. Sagan, P. Gergely and R.N. White, 12/8/88, (PB89-163737, A08, MF-A01).
NCEER-88-0034 "Seismic Response of Pile Foundations," by S.M. Mamoon, P.K. Banerjee and S. Ahmad, 11/1/88, (PB89-
145239, A04, MF-A01).
NCEER-88-0035 "Modeling of R/C Building Structures With Flexible Floor Diaphragms (IDARC2)," by A.M. Reinhorn, S.K.
Kunnath and N. Panahshahi, 9/7/88, (PB89-207153, A07, MF-A01).
NCEER-88-0036 "Solution of the Dam-Reservoir Interaction Problem Using a Combination of FEM, BEM with Particular
Integrals, Modal Analysis, and Substructuring," by C-S. Tsai, G.C. Lee and R.L. Ketter, 12/31/88, (PB89-
207146, A04, MF-A01).
NCEER-88-0037 "Optimal Placement of Actuators for Structural Control," by F.Y. Cheng and C.P. Pantelides, 8/15/88,
(PB89-162846, A05, MF-A01).
NCEER-88-0038 "Teflon Bearings in Aseismic Base Isolation: Experimental Studies and Mathematical Modeling," by A.
Mokha, M.C. Constantinou and A.M. Reinhorn, 12/5/88, (PB89-218457, A10, MF-A01). This report is
available only through NTIS (see address given above).
NCEER-88-0039 "Seismic Behavior of Flat Slab High-Rise Buildings in the New York City Area," by P. Weidlinger and M.
Ettouney, 10/15/88, (PB90-145681, A04, MF-A01).
NCEER-88-0040 "Evaluation of the Earthquake Resistance of Existing Buildings in New York City," by P. Weidlinger and M.
Ettouney, 10/15/88, not available.
NCEER-88-0041 "Small-Scale Modeling Techniques for Reinforced Concrete Structures Subjected to Seismic Loads," by W.
Kim, A. El-Attar and R.N. White, 11/22/88, (PB89-189625, A05, MF-A01).
NCEER-88-0042 "Modeling Strong Ground Motion from Multiple Event Earthquakes," by G.W. Ellis and A.S. Cakmak,
10/15/88, (PB89-174445, A03, MF-A01).
330
NCEER-88-0043 "Nonstationary Models of Seismic Ground Acceleration," by M. Grigoriu, S.E. Ruiz and E. Rosenblueth,
7/15/88, (PB89-189617, A04, MF-A01).
NCEER-88-0044 "SARCF User's Guide: Seismic Analysis of Reinforced Concrete Frames," by Y.S. Chung, C. Meyer and M.
Shinozuka, 11/9/88, (PB89-174452, A08, MF-A01).
NCEER-88-0045 "First Expert Panel Meeting on Disaster Research and Planning," edited by J. Pantelic and J. Stoyle, 9/15/88,
(PB89-174460, A05, MF-A01).
NCEER-88-0046 "Preliminary Studies of the Effect of Degrading Infill Walls on the Nonlinear Seismic Response of Steel
Frames," by C.Z. Chrysostomou, P. Gergely and J.F. Abel, 12/19/88, (PB89-208383, A05, MF-A01).
NCEER-88-0047 "Reinforced Concrete Frame Component Testing Facility - Design, Construction, Instrumentation and
Operation," by S.P. Pessiki, C. Conley, T. Bond, P. Gergely and R.N. White, 12/16/88, (PB89-174478, A04,
MF-A01).
NCEER-89-0001 "Effects of Protective Cushion and Soil Compliancy on the Response of Equipment Within a Seismically
Excited Building," by J.A. HoLung, 2/16/89, (PB89-207179, A04, MF-A01).
NCEER-89-0002 "Statistical Evaluation of Response Modification Factors for Reinforced Concrete Structures," by H.H-M.
Hwang and J-W. Jaw, 2/17/89, (PB89-207187, A05, MF-A01).
NCEER-89-0003 "Hysteretic Columns Under Random Excitation," by G-Q. Cai and Y.K. Lin, 1/9/89, (PB89-196513, A03,
MF-A01).
NCEER-89-0004 "Experimental Study of `Elephant Foot Bulge' Instability of Thin-Walled Metal Tanks," by Z-H. Jia and R.L.
Ketter, 2/22/89, (PB89-207195, A03, MF-A01).
NCEER-89-0005 "Experiment on Performance of Buried Pipelines Across San Andreas Fault," by J. Isenberg, E. Richardson
and T.D. O'Rourke, 3/10/89, (PB89-218440, A04, MF-A01). This report is available only through NTIS (see
NCEER-89-0006 "A Knowledge-Based Approach to Structural Design of Earthquake-Resistant Buildings," by M. Subramani,

P. Gergely, C.H. Conley, J.F. Abel and A.H. Zaghw, 1/15/89, (PB89-218465, A06, MF-A01).
NCEER-89-0007 "Liquefaction Hazards and Their Effects on Buried Pipelines," by T.D. O'Rourke and P.A. Lane, 2/1/89,
(PB89-218481, A09, MF-A01).
NCEER-89-0008 "Fundamentals of System Identification in Structural Dynamics," by H. Imai, C-B. Yun, O. Maruyama and
M. Shinozuka, 1/26/89, (PB89-207211, A04, MF-A01).
NCEER-89-0009 "Effects of the 1985 Michoacan Earthquake on Water Systems and Other Buried Lifelines in Mexico," by
A.G. Ayala and M.J. O'Rourke, 3/8/89, (PB89-207229, A06, MF-A01).
NCEER-89-R010 "NCEER Bibliography of Earthquake Education Materials," by K.E.K. Ross, Second Revision, 9/1/89,
(PB90-125352, A05, MF-A01). This report is replaced by NCEER-92-0018.
NCEER-89-0011 "Inelastic Three-Dimensional Response Analysis of Reinforced Concrete Building Structures (IDARC-3D),
Part I - Modeling," by S.K. Kunnath and A.M. Reinhorn, 4/17/89, (PB90-114612, A07, MF-A01). This
NCEER-89-0012 "Recommended Modifications to ATC-14," by C.D. Poland and J.O. Malley, 4/12/89, (PB90-108648, A15,
MF-A01).
NCEER-89-0013 "Repair and Strengthening of Beam-to-Column Connections Subjected to Earthquake Loading," by M.

Corazao and A.J. Durrani, 2/28/89, (PB90-109885, A06, MF-A01).
NCEER-89-0014 "Program EXKAL2 for Identification of Structural Dynamic Systems," by O. Maruyama, C-B. Yun, M.
Hoshiya and M. Shinozuka, 5/19/89, (PB90-109877, A09, MF-A01).
331
NCEER-89-0015 "Response of Frames With Bolted Semi-Rigid Connections, Part I - Experimental Study and Analytical
Predictions," by P.J. DiCorso, A.M. Reinhorn, J.R. Dickerson, J.B. Radziminski and W.L. Harper, 6/1/89,
not available.
NCEER-89-0016 "ARMA Monte Carlo Simulation in Probabilistic Structural Analysis," by P.D. Spanos and M.P. Mignolet,
7/10/89, (PB90-109893, A03, MF-A01).
NCEER-89-P017 "Preliminary Proceedings from the Conference on Disaster Preparedness - The Place of Earthquake
Education in Our Schools," Edited by K.E.K. Ross, 6/23/89, (PB90-108606, A03, MF-A01).
NCEER-89-0017 "Proceedings from the Conference on Disaster Preparedness - The Place of Earthquake Education in Our
Schools," Edited by K.E.K. Ross, 12/31/89, (PB90-207895, A012, MF-A02). This report is available only
NCEER-89-0018 "Multidimensional Models of Hysteretic Material Behavior for Vibration Analysis of Shape Memory Energy
Absorbing Devices, by E.J. Graesser and F.A. Cozzarelli, 6/7/89, (PB90-164146, A04, MF-A01).
NCEER-89-0019 "Nonlinear Dynamic Analysis of Three-Dimensional Base Isolated Structures (3D-BASIS)," by S.

Nagarajaiah, A.M. Reinhorn and M.C. Constantinou, 8/3/89, (PB90-161936, A06, MF-A01). This report has
been replaced by NCEER-93-0011.
NCEER-89-0020 "Structural Control Considering Time-Rate of Control Forces and Control Rate Constraints," by F.Y. Cheng
and C.P. Pantelides, 8/3/89, (PB90-120445, A04, MF-A01).
NCEER-89-0021 "Subsurface Conditions of Memphis and Shelby County," by K.W. Ng, T-S. Chang and H-H.M. Hwang,
7/26/89, (PB90-120437, A03, MF-A01).
NCEER-89-0022 "Seismic Wave Propagation Effects on Straight Jointed Buried Pipelines," by K. Elhmadi and M.J. O'Rourke,
8/24/89, (PB90-162322, A10, MF-A02).
NCEER-89-0023 "Workshop on Serviceability Analysis of Water Delivery Systems," edited by M. Grigoriu, 3/6/89, (PB90-
127424, A03, MF-A01).
NCEER-89-0024 "Shaking Table Study of a 1/5 Scale Steel Frame Composed of Tapered Members," by K.C. Chang, J.S.
Hwang and G.C. Lee, 9/18/89, (PB90-160169, A04, MF-A01).
NCEER-89-0025 "DYNA1D: A Computer Program for Nonlinear Seismic Site Response Analysis - Technical
Documentation," by Jean H. Prevost, 9/14/89, (PB90-161944, A07, MF-A01). This report is available only
NCEER-89-0026 "1:4 Scale Model Studies of Active Tendon Systems and Active Mass Dampers for Aseismic Protection," by
A.M. Reinhorn, T.T. Soong, R.C. Lin, Y.P. Yang, Y. Fukao, H. Abe and M. Nakai, 9/15/89, (PB90-173246,
A10, MF-A02). This report is available only through NTIS (see address given above).
NCEER-89-0027 "Scattering of Waves by Inclusions in a Nonhomogeneous Elastic Half Space Solved by Boundary Element
Methods," by P.K. Hadley, A. Askar and A.S. Cakmak, 6/15/89, (PB90-145699, A07, MF-A01).
NCEER-89-0028 "Statistical Evaluation of Deflection Amplification Factors for Reinforced Concrete Structures," by H.H.M.
Hwang, J-W. Jaw and A.L. Ch'ng, 8/31/89, (PB90-164633, A05, MF-A01).
NCEER-89-0029 "Bedrock Accelerations in Memphis Area Due to Large New Madrid Earthquakes," by H.H.M. Hwang,
C.H.S. Chen and G. Yu, 11/7/89, (PB90-162330, A04, MF-A01).
NCEER-89-0030 "Seismic Behavior and Response Sensitivity of Secondary Structural Systems," by Y.Q. Chen and T.T.
Soong, 10/23/89, (PB90-164658, A08, MF-A01).
NCEER-89-0031 "Random Vibration and Reliability Analysis of Primary-Secondary Structural Systems," by Y. Ibrahim, M.
Grigoriu and T.T. Soong, 11/10/89, (PB90-161951, A04, MF-A01).
332
NCEER-89-0032 "Proceedings from the Second U.S. - Japan Workshop on Liquefaction, Large Ground Deformation and
Their Effects on Lifelines, September 26-29, 1989," Edited by T.D. O'Rourke and M. Hamada, 12/1/89,
(PB90-209388, A22, MF-A03).
NCEER-89-0033 "Deterministic Model for Seismic Damage Evaluation of Reinforced Concrete Structures," by J.M. Bracci,
A.M. Reinhorn, J.B. Mander and S.K. Kunnath, 9/27/89, (PB91-108803, A06, MF-A01).
NCEER-89-0034 "On the Relation Between Local and Global Damage Indices," by E. DiPasquale and A.S. Cakmak, 8/15/89,
(PB90-173865, A05, MF-A01).
NCEER-89-0035 "Cyclic Undrained Behavior of Nonplastic and Low Plasticity Silts," by A.J. Walker and H.E. Stewart,
7/26/89, (PB90-183518, A10, MF-A01).
NCEER-89-0036 "Liquefaction Potential of Surficial Deposits in the City of Buffalo, New York," by M. Budhu, R. Giese and
L. Baumgrass, 1/17/89, (PB90-208455, A04, MF-A01).
NCEER-89-0037 "A Deterministic Assessment of Effects of Ground Motion Incoherence," by A.S. Veletsos and Y. Tang,
7/15/89, (PB90-164294, A03, MF-A01).
NCEER-89-0038 "Workshop on Ground Motion Parameters for Seismic Hazard Mapping," July 17-18, 1989, edited by R.V.
Whitman, 12/1/89, (PB90-173923, A04, MF-A01).
NCEER-89-0039 "Seismic Effects on Elevated Transit Lines of the New York City Transit Authority," by C.J. Costantino,
C.A. Miller and E. Heymsfield, 12/26/89, (PB90-207887, A06, MF-A01).
NCEER-89-0040 "Centrifugal Modeling of Dynamic Soil-Structure Interaction," by K. Weissman, Supervised by J.H. Prevost,
5/10/89, (PB90-207879, A07, MF-A01).
NCEER-89-0041 "Linearized Identification of Buildings With Cores for Seismic Vulnerability Assessment," by I-K. Ho and
A.E. Aktan, 11/1/89, (PB90-251943, A07, MF-A01).
NCEER-90-0001 "Geotechnical and Lifeline Aspects of the October 17, 1989 Loma Prieta Earthquake in San Francisco," by
T.D. O'Rourke, H.E. Stewart, F.T. Blackburn and T.S. Dickerman, 1/90, (PB90-208596, A05, MF-A01).
NCEER-90-0002 "Nonnormal Secondary Response Due to Yielding in a Primary Structure," by D.C.K. Chen and L.D. Lutes,
2/28/90, (PB90-251976, A07, MF-A01).
NCEER-90-0003 "Earthquake Education Materials for Grades K-12," by K.E.K. Ross, 4/16/90, (PB91-251984, A05, MF-
A05). This report has been replaced by NCEER-92-0018.
NCEER-90-0004 "Catalog of Strong Motion Stations in Eastern North America," by R.W. Busby, 4/3/90, (PB90-251984, A05,
MF-A01).
NCEER-90-0005 "NCEER Strong-Motion Data Base: A User Manual for the GeoBase Release (Version 1.0 for the Sun3)," by
P. Friberg and K. Jacob, 3/31/90 (PB90-258062, A04, MF-A01).
NCEER-90-0006 "Seismic Hazard Along a Crude Oil Pipeline in the Event of an 1811-1812 Type New Madrid Earthquake,"
by H.H.M. Hwang and C-H.S. Chen, 4/16/90, (PB90-258054, A04, MF-A01).
NCEER-90-0007 "Site-Specific Response Spectra for Memphis Sheahan Pumping Station," by H.H.M. Hwang and C.S. Lee,
5/15/90, (PB91-108811, A05, MF-A01).
NCEER-90-0008 "Pilot Study on Seismic Vulnerability of Crude Oil Transmission Systems," by T. Ariman, R. Dobry, M.
Grigoriu, F. Kozin, M. O'Rourke, T. O'Rourke and M. Shinozuka, 5/25/90, (PB91-108837, A06, MF-A01).
NCEER-90-0009 "A Program to Generate Site Dependent Time Histories: EQGEN," by G.W. Ellis, M. Srinivasan and A.S.
Cakmak, 1/30/90, (PB91-108829, A04, MF-A01).
NCEER-90-0010 "Active Isolation for Seismic Protection of Operating Rooms," by M.E. Talbott, Supervised by M.
333
NCEER-90-0011 "Program LINEARID for Identification of Linear Structural Dynamic Systems," by C-B. Yun and M.
NCEER-90-0012 "Two-Dimensional Two-Phase Elasto-Plastic Seismic Response of Earth Dams," by A.N. Yiagos, Supervised
by J.H. Prevost, 6/20/90, (PB91-110197, A13, MF-A02).
NCEER-90-0013 "Secondary Systems in Base-Isolated Structures: Experimental Investigation, Stochastic Response and
Stochastic Sensitivity," by G.D. Manolis, G. Juhn, M.C. Constantinou and A.M. Reinhorn, 7/1/90, (PB91-
110320, A08, MF-A01).
NCEER-90-0014 "Seismic Behavior of Lightly-Reinforced Concrete Column and Beam-Column Joint Details," by S.P.
Pessiki, C.H. Conley, P. Gergely and R.N. White, 8/22/90, (PB91-108795, A11, MF-A02).
NCEER-90-0015 "Two Hybrid Control Systems for Building Structures Under Strong Earthquakes," by J.N. Yang and A.
Danielians, 6/29/90, (PB91-125393, A04, MF-A01).
NCEER-90-0016 "Instantaneous Optimal Control with Acceleration and Velocity Feedback," by J.N. Yang and Z. Li, 6/29/90,
(PB91-125401, A03, MF-A01).
NCEER-90-0017 "Reconnaissance Report on the Northern Iran Earthquake of June 21, 1990," by M. Mehrain, 10/4/90, (PB91-
125377, A03, MF-A01).
NCEER-90-0018 "Evaluation of Liquefaction Potential in Memphis and Shelby County," by T.S. Chang, P.S. Tang, C.S. Lee
and H. Hwang, 8/10/90, (PB91-125427, A09, MF-A01).
NCEER-90-0019 "Experimental and Analytical Study of a Combined Sliding Disc Bearing and Helical Steel Spring Isolation
System," by M.C. Constantinou, A.S. Mokha and A.M. Reinhorn, 10/4/90, (PB91-125385, A06, MF-A01).
This report is available only through NTIS (see address given above).
NCEER-90-0020 "Experimental Study and Analytical Prediction of Earthquake Response of a Sliding Isolation System with a
Spherical Surface," by A.S. Mokha, M.C. Constantinou and A.M. Reinhorn, 10/11/90, (PB91-125419, A05,
MF-A01).
NCEER-90-0021 "Dynamic Interaction Factors for Floating Pile Groups," by G. Gazetas, K. Fan, A. Kaynia and E. Kausel,
9/10/90, (PB91-170381, A05, MF-A01).
NCEER-90-0022 "Evaluation of Seismic Damage Indices for Reinforced Concrete Structures," by S. Rodriguez-Gomez and
A.S. Cakmak, 9/30/90, PB91-171322, A06, MF-A01).
NCEER-90-0023 "Study of Site Response at a Selected Memphis Site," by H. Desai, S. Ahmad, E.S. Gazetas and M.R. Oh,
10/11/90, (PB91-196857, A03, MF-A01).
NCEER-90-0024 "A User's Guide to Strongmo: Version 1.0 of NCEER's Strong-Motion Data Access Tool for PCs and
Terminals," by P.A. Friberg and C.A.T. Susch, 11/15/90, (PB91-171272, A03, MF-A01).
NCEER-90-0025 "A Three-Dimensional Analytical Study of Spatial Variability of Seismic Ground Motions," by L-L. Hong
and A.H.-S. Ang, 10/30/90, (PB91-170399, A09, MF-A01).
NCEER-90-0026 "MUMOID User's Guide - A Program for the Identification of Modal Parameters," by S. Rodriguez-Gomez
and E. DiPasquale, 9/30/90, (PB91-171298, A04, MF-A01).
NCEER-90-0027 "SARCF-II User's Guide - Seismic Analysis of Reinforced Concrete Frames," by S. Rodriguez-Gomez, Y.S.
Chung and C. Meyer, 9/30/90, (PB91-171280, A05, MF-A01).
NCEER-90-0028 "Viscous Dampers: Testing, Modeling and Application in Vibration and Seismic Isolation," by N. Makris
and M.C. Constantinou, 12/20/90 (PB91-190561, A06, MF-A01).
NCEER-90-0029 "Soil Effects on Earthquake Ground Motions in the Memphis Area," by H. Hwang, C.S. Lee, K.W. Ng and
T.S. Chang, 8/2/90, (PB91-190751, A05, MF-A01).
334
NCEER-91-0001 "Proceedings from the Third Japan-U.S. Workshop on Earthquake Resistant Design of Lifeline Facilities and
Countermeasures for Soil Liquefaction, December 17-19, 1990," edited by T.D. O'Rourke and M. Hamada,
2/1/91, (PB91-179259, A99, MF-A04).
NCEER-91-0002 "Physical Space Solutions of Non-Proportionally Damped Systems," by M. Tong, Z. Liang and G.C. Lee,
1/15/91, (PB91-179242, A04, MF-A01).
NCEER-91-0003 "Seismic Response of Single Piles and Pile Groups," by K. Fan and G. Gazetas, 1/10/91, (PB92-174994,
A04, MF-A01).
NCEER-91-0004 "Damping of Structures: Part 1 - Theory of Complex Damping," by Z. Liang and G. Lee, 10/10/91, (PB92-
197235, A12, MF-A03).
NCEER-91-0005 "3D-BASIS - Nonlinear Dynamic Analysis of Three Dimensional Base Isolated Structures: Part II," by S.
Nagarajaiah, A.M. Reinhorn and M.C. Constantinou, 2/28/91, (PB91-190553, A07, MF-A01). This report
has been replaced by NCEER-93-0011.
NCEER-91-0006 "A Multidimensional Hysteretic Model for Plasticity Deforming Metals in Energy Absorbing Devices," by
E.J. Graesser and F.A. Cozzarelli, 4/9/91, (PB92-108364, A04, MF-A01).
NCEER-91-0007 "A Framework for Customizable Knowledge-Based Expert Systems with an Application to a KBES for
Evaluating the Seismic Resistance of Existing Buildings," by E.G. Ibarra-Anaya and S.J. Fenves, 4/9/91,
(PB91-210930, A08, MF-A01).
NCEER-91-0008 "Nonlinear Analysis of Steel Frames with Semi-Rigid Connections Using the Capacity Spectrum Method,"
by G.G. Deierlein, S-H. Hsieh, Y-J. Shen and J.F. Abel, 7/2/91, (PB92-113828, A05, MF-A01).
NCEER-91-0009 "Earthquake Education Materials for Grades K-12," by K.E.K. Ross, 4/30/91, (PB91-212142, A06, MF-
A01). This report has been replaced by NCEER-92-0018.
NCEER-91-0010 "Phase Wave Velocities and Displacement Phase Differences in a Harmonically Oscillating Pile," by N.
Makris and G. Gazetas, 7/8/91, (PB92-108356, A04, MF-A01).
NCEER-91-0011 "Dynamic Characteristics of a Full-Size Five-Story Steel Structure and a 2/5 Scale Model," by K.C. Chang,
G.C. Yao, G.C. Lee, D.S. Hao and Y.C. Yeh," 7/2/91, (PB93-116648, A06, MF-A02).
NCEER-91-0012 "Seismic Response of a 2/5 Scale Steel Structure with Added Viscoelastic Dampers," by K.C. Chang, T.T.
Soong, S-T. Oh and M.L. Lai, 5/17/91, (PB92-110816, A05, MF-A01).
NCEER-91-0013 "Earthquake Response of Retaining Walls; Full-Scale Testing and Computational Modeling," by S.
Alampalli and A-W.M. Elgamal, 6/20/91, not available.
NCEER-91-0014 "3D-BASIS-M: Nonlinear Dynamic Analysis of Multiple Building Base Isolated Structures," by P.C.
Tsopelas, S. Nagarajaiah, M.C. Constantinou and A.M. Reinhorn, 5/28/91, (PB92-113885, A09, MF-A02).
NCEER-91-0015 "Evaluation of SEAOC Design Requirements for Sliding Isolated Structures," by D. Theodossiou and M.C.
Constantinou, 6/10/91, (PB92-114602, A11, MF-A03).
NCEER-91-0016 "Closed-Loop Modal Testing of a 27-Story Reinforced Concrete Flat Plate-Core Building," by H.R.
Somaprasad, T. Toksoy, H. Yoshiyuki and A.E. Aktan, 7/15/91, (PB92-129980, A07, MF-A02).
NCEER-91-0017 "Shake Table Test of a 1/6 Scale Two-Story Lightly Reinforced Concrete Building," by A.G. El-Attar, R.N.
White and P. Gergely, 2/28/91, (PB92-222447, A06, MF-A02).
NCEER-91-0018 "Shake Table Test of a 1/8 Scale Three-Story Lightly Reinforced Concrete Building," by A.G. El-Attar, R.N.
NCEER-91-0019 "Transfer Functions for Rigid Rectangular Foundations," by A.S. Veletsos, A.M. Prasad and W.H. Wu,
7/31/91, not available.
335
NCEER-91-0020 "Hybrid Control of Seismic-Excited Nonlinear and Inelastic Structural Systems," by J.N. Yang, Z. Li and A.
Danielians, 8/1/91, (PB92-143171, A06, MF-A02).
NCEER-91-0021 "The NCEER-91 Earthquake Catalog: Improved Intensity-Based Magnitudes and Recurrence Relations for
U.S. Earthquakes East of New Madrid," by L. Seeber and J.G. Armbruster, 8/28/91, (PB92-176742, A06,
MF-A02).
NCEER-91-0022 "Proceedings from the Implementation of Earthquake Planning and Education in Schools: The Need for
Change - The Roles of the Changemakers," by K.E.K. Ross and F. Winslow, 7/23/91, (PB92-129998, A12,
MF-A03).
NCEER-91-0023 "A Study of Reliability-Based Criteria for Seismic Design of Reinforced Concrete Frame Buildings," by
H.H.M. Hwang and H-M. Hsu, 8/10/91, (PB92-140235, A09, MF-A02).
NCEER-91-0024 "Experimental Verification of a Number of Structural System Identification Algorithms," by R.G. Ghanem,
H. Gavin and M. Shinozuka, 9/18/91, (PB92-176577, A18, MF-A04).
NCEER-91-0025 "Probabilistic Evaluation of Liquefaction Potential," by H.H.M. Hwang and C.S. Lee," 11/25/91, (PB92-
143429, A05, MF-A01).
NCEER-91-0026 "Instantaneous Optimal Control for Linear, Nonlinear and Hysteretic Structures - Stable Controllers," by J.N.
Yang and Z. Li, 11/15/91, (PB92-163807, A04, MF-A01).
NCEER-91-0027 "Experimental and Theoretical Study of a Sliding Isolation System for Bridges," by M.C. Constantinou, A.
Kartoum, A.M. Reinhorn and P. Bradford, 11/15/91, (PB92-176973, A10, MF-A03).
NCEER-92-0001 "Case Studies of Liquefaction and Lifeline Performance During Past Earthquakes, Volume 1: Japanese Case
Studies," Edited by M. Hamada and T. O'Rourke, 2/17/92, (PB92-197243, A18, MF-A04).
NCEER-92-0002 "Case Studies of Liquefaction and Lifeline Performance During Past Earthquakes, Volume 2: United States
Case Studies," Edited by T. O'Rourke and M. Hamada, 2/17/92, (PB92-197250, A20, MF-A04).
NCEER-92-0003 "Issues in Earthquake Education," Edited by K. Ross, 2/3/92, (PB92-222389, A07, MF-A02).
NCEER-92-0004 "Proceedings from the First U.S. - Japan Workshop on Earthquake Protective Systems for Bridges," Edited
by I.G. Buckle, 2/4/92, (PB94-142239, A99, MF-A06).
NCEER-92-0005 "Seismic Ground Motion from a Haskell-Type Source in a Multiple-Layered Half-Space," A.P. Theoharis, G.
Deodatis and M. Shinozuka, 1/2/92, not available.
NCEER-92-0006 "Proceedings from the Site Effects Workshop," Edited by R. Whitman, 2/29/92, (PB92-197201, A04, MF-
A01).
NCEER-92-0007 "Engineering Evaluation of Permanent Ground Deformations Due to Seismically-Induced Liquefaction," by

M.H. Baziar, R. Dobry and A-W.M. Elgamal, 3/24/92, (PB92-222421, A13, MF-A03).
NCEER-92-0008 "A Procedure for the Seismic Evaluation of Buildings in the Central and Eastern United States," by C.D.
Poland and J.O. Malley, 4/2/92, (PB92-222439, A20, MF-A04).
NCEER-92-0009 "Experimental and Analytical Study of a Hybrid Isolation System Using Friction Controllable Sliding
Bearings," by M.Q. Feng, S. Fujii and M. Shinozuka, 5/15/92, (PB93-150282, A06, MF-A02).
NCEER-92-0010 "Seismic Resistance of Slab-Column Connections in Existing Non-Ductile Flat-Plate Buildings," by A.J.
Durrani and Y. Du, 5/18/92, (PB93-116812, A06, MF-A02).
NCEER-92-0011 "The Hysteretic and Dynamic Behavior of Brick Masonry Walls Upgraded by Ferrocement Coatings Under
Cyclic Loading and Strong Simulated Ground Motion," by H. Lee and S.P. Prawel, 5/11/92, not available.
NCEER-92-0012 "Study of Wire Rope Systems for Seismic Protection of Equipment in Buildings," by G.F. Demetriades,
M.C. Constantinou and A.M. Reinhorn, 5/20/92, (PB93-116655, A08, MF-A02).
336
NCEER-92-0013 "Shape Memory Structural Dampers: Material Properties, Design and Seismic Testing," by P.R. Witting and
F.A. Cozzarelli, 5/26/92, (PB93-116663, A05, MF-A01).
NCEER-92-0014 "Longitudinal Permanent Ground Deformation Effects on Buried Continuous Pipelines," by M.J. O'Rourke,
and C. Nordberg, 6/15/92, (PB93-116671, A08, MF-A02).
NCEER-92-0015 "A Simulation Method for Stationary Gaussian Random Functions Based on the Sampling Theorem," by M.
Grigoriu and S. Balopoulou, 6/11/92, (PB93-127496, A05, MF-A01).
NCEER-92-0016 "Gravity-Load-Designed Reinforced Concrete Buildings: Seismic Evaluation of Existing Construction and
Detailing Strategies for Improved Seismic Resistance," by G.W. Hoffmann, S.K. Kunnath, A.M. Reinhorn
and J.B. Mander, 7/15/92, (PB94-142007, A08, MF-A02).
NCEER-92-0017 "Observations on Water System and Pipeline Performance in the Limón Area of Costa Rica Due to the April
22, 1991 Earthquake," by M. O'Rourke and D. Ballantyne, 6/30/92, (PB93-126811, A06, MF-A02).
NCEER-92-0018 "Fourth Edition of Earthquake Education Materials for Grades K-12," Edited by K.E.K. Ross, 8/10/92,
(PB93-114023, A07, MF-A02).
NCEER-92-0019 "Proceedings from the Fourth Japan-U.S. Workshop on Earthquake Resistant Design of Lifeline Facilities
and Countermeasures for Soil Liquefaction," Edited by M. Hamada and T.D. O'Rourke, 8/12/92, (PB93-
163939, A99, MF-E11).
NCEER-92-0020 "Active Bracing System: A Full Scale Implementation of Active Control," by A.M. Reinhorn, T.T. Soong,
R.C. Lin, M.A. Riley, Y.P. Wang, S. Aizawa and M. Higashino, 8/14/92, (PB93-127512, A06, MF-A02).
NCEER-92-0021 "Empirical Analysis of Horizontal Ground Displacement Generated by Liquefaction-Induced Lateral

Spreads," by S.F. Bartlett and T.L. Youd, 8/17/92, (PB93-188241, A06, MF-A02).
NCEER-92-0022 "IDARC Version 3.0: Inelastic Damage Analysis of Reinforced Concrete Structures," by S.K. Kunnath, A.M.
Reinhorn and R.F. Lobo, 8/31/92, (PB93-227502, A07, MF-A02).
NCEER-92-0023 "A Semi-Empirical Analysis of Strong-Motion Peaks in Terms of Seismic Source, Propagation Path and
Local Site Conditions, by M. Kamiyama, M.J. O'Rourke and R. Flores-Berrones, 9/9/92, (PB93-150266,
A08, MF-A02).
NCEER-92-0024 "Seismic Behavior of Reinforced Concrete Frame Structures with Nonductile Details, Part I: Summary of
Experimental Findings of Full Scale Beam-Column Joint Tests," by A. Beres, R.N. White and P. Gergely,
9/30/92, (PB93-227783, A05, MF-A01).
NCEER-92-0025 "Experimental Results of Repaired and Retrofitted Beam-Column Joint Tests in Lightly Reinforced Concrete
Frame Buildings," by A. Beres, S. El-Borgi, R.N. White and P. Gergely, 10/29/92, (PB93-227791, A05, MF-
A01).
NCEER-92-0026 "A Generalization of Optimal Control Theory: Linear and Nonlinear Structures," by J.N. Yang, Z. Li and S.
Vongchavalitkul, 11/2/92, (PB93-188621, A05, MF-A01).
NCEER-92-0027 "Seismic Resistance of Reinforced Concrete Frame Structures Designed Only for Gravity Loads: Part I -
Design and Properties of a One-Third Scale Model Structure," by J.M. Bracci, A.M. Reinhorn and J.B.
Mander, 12/1/92, (PB94-104502, A08, MF-A02).
NCEER-92-0028 "Seismic Resistance of Reinforced Concrete Frame Structures Designed Only for Gravity Loads: Part II -
Experimental Performance of Subassemblages," by L.E. Aycardi, J.B. Mander and A.M. Reinhorn, 12/1/92,
(PB94-104510, A08, MF-A02).
NCEER-92-0029 "Seismic Resistance of Reinforced Concrete Frame Structures Designed Only for Gravity Loads: Part III -
Experimental Performance and Analytical Study of a Structural Model," by J.M. Bracci, A.M. Reinhorn and
J.B. Mander, 12/1/92, (PB93-227528, A09, MF-A01).
337
NCEER-92-0030 "Evaluation of Seismic Retrofit of Reinforced Concrete Frame Structures: Part I - Experimental Performance
of Retrofitted Subassemblages," by D. Choudhuri, J.B. Mander and A.M. Reinhorn, 12/8/92, (PB93-198307,
A07, MF-A02).
NCEER-92-0031 "Evaluation of Seismic Retrofit of Reinforced Concrete Frame Structures: Part II - Experimental
Performance and Analytical Study of a Retrofitted Structural Model," by J.M. Bracci, A.M. Reinhorn and
J.B. Mander, 12/8/92, (PB93-198315, A09, MF-A03).
NCEER-92-0032 "Experimental and Analytical Investigation of Seismic Response of Structures with Supplemental Fluid
Viscous Dampers," by M.C. Constantinou and M.D. Symans, 12/21/92, (PB93-191435, A10, MF-A03). This
NCEER-92-0033 "Reconnaissance Report on the Cairo, Egypt Earthquake of October 12, 1992," by M. Khater, 12/23/92,
(PB93-188621, A03, MF-A01).
NCEER-92-0034 "Low-Level Dynamic Characteristics of Four Tall Flat-Plate Buildings in New York City," by H. Gavin, S.
Yuan, J. Grossman, E. Pekelis and K. Jacob, 12/28/92, (PB93-188217, A07, MF-A02).
NCEER-93-0001 "An Experimental Study on the Seismic Performance of Brick-Infilled Steel Frames With and Without
Retrofit," by J.B. Mander, B. Nair, K. Wojtkowski and J. Ma, 1/29/93, (PB93-227510, A07, MF-A02).
NCEER-93-0002 "Social Accounting for Disaster Preparedness and Recovery Planning," by S. Cole, E. Pantoja and V. Razak,
2/22/93, (PB94-142114, A12, MF-A03).
NCEER-93-0003 "Assessment of 1991 NEHRP Provisions for Nonstructural Components and Recommended Revisions," by
T.T. Soong, G. Chen, Z. Wu, R-H. Zhang and M. Grigoriu, 3/1/93, (PB93-188639, A06, MF-A02).
NCEER-93-0004 "Evaluation of Static and Response Spectrum Analysis Procedures of SEAOC/UBC for Seismic Isolated
Structures," by C.W. Winters and M.C. Constantinou, 3/23/93, (PB93-198299, A10, MF-A03).
NCEER-93-0005 "Earthquakes in the Northeast - Are We Ignoring the Hazard? A Workshop on Earthquake Science and
Safety for Educators," edited by K.E.K. Ross, 4/2/93, (PB94-103066, A09, MF-A02).
NCEER-93-0006 "Inelastic Response of Reinforced Concrete Structures with Viscoelastic Braces," by R.F. Lobo, J.M. Bracci,
K.L. Shen, A.M. Reinhorn and T.T. Soong, 4/5/93, (PB93-227486, A05, MF-A02).
NCEER-93-0007 "Seismic Testing of Installation Methods for Computers and Data Processing Equipment," by K. Kosar, T.T.
Soong, K.L. Shen, J.A. HoLung and Y.K. Lin, 4/12/93, (PB93-198299, A07, MF-A02).
NCEER-93-0008 "Retrofit of Reinforced Concrete Frames Using Added Dampers," by A. Reinhorn, M. Constantinou and C.
Li, not available.
NCEER-93-0009 "Seismic Behavior and Design Guidelines for Steel Frame Structures with Added Viscoelastic Dampers," by
K.C. Chang, M.L. Lai, T.T. Soong, D.S. Hao and Y.C. Yeh, 5/1/93, (PB94-141959, A07, MF-A02).
NCEER-93-0010 "Seismic Performance of Shear-Critical Reinforced Concrete Bridge Piers," by J.B. Mander, S.M. Waheed,
M.T.A. Chaudhary and S.S. Chen, 5/12/93, (PB93-227494, A08, MF-A02).
NCEER-93-0011 "3D-BASIS-TABS: Computer Program for Nonlinear Dynamic Analysis of Three Dimensional Base Isolated
Structures," by S. Nagarajaiah, C. Li, A.M. Reinhorn and M.C. Constantinou, 8/2/93, (PB94-141819, A09,
MF-A02).
NCEER-93-0012 "Effects of Hydrocarbon Spills from an Oil Pipeline Break on Ground Water," by O.J. Helweg and H.H.M.
Hwang, 8/3/93, (PB94-141942, A06, MF-A02).
NCEER-93-0013 "Simplified Procedures for Seismic Design of Nonstructural Components and Assessment of Current Code
Provisions," by M.P. Singh, L.E. Suarez, E.E. Matheu and G.O. Maldonado, 8/4/93, (PB94-141827, A09,
MF-A02).
NCEER-93-0014 "An Energy Approach to Seismic Analysis and Design of Secondary Systems," by G. Chen and T.T. Soong,
8/6/93, (PB94-142767, A11, MF-A03).
338
NCEER-93-0015 "Proceedings from School Sites: Becoming Prepared for Earthquakes - Commemorating the Third
Anniversary of the Loma Prieta Earthquake," Edited by F.E. Winslow and K.E.K. Ross, 8/16/93, (PB94-
154275, A16, MF-A02).
NCEER-93-0016 "Reconnaissance Report of Damage to Historic Monuments in Cairo, Egypt Following the October 12, 1992
Dahshur Earthquake," by D. Sykora, D. Look, G. Croci, E. Karaesmen and E. Karaesmen, 8/19/93, (PB94-
142221, A08, MF-A02).
NCEER-93-0017 "The Island of Guam Earthquake of August 8, 1993," by S.W. Swan and S.K. Harris, 9/30/93, (PB94-
141843, A04, MF-A01).
NCEER-93-0018 "Engineering Aspects of the October 12, 1992 Egyptian Earthquake," by A.W. Elgamal, M. Amer, K.
Adalier and A. Abul-Fadl, 10/7/93, (PB94-141983, A05, MF-A01).
NCEER-93-0019 "Development of an Earthquake Motion Simulator and its Application in Dynamic Centrifuge Testing," by I.
Krstelj, Supervised by J.H. Prevost, 10/23/93, (PB94-181773, A-10, MF-A03).
NCEER-93-0020 "NCEER-Taisei Corporation Research Program on Sliding Seismic Isolation Systems for Bridges:
Experimental and Analytical Study of a Friction Pendulum System (FPS)," by M.C. Constantinou, P.
Tsopelas, Y-S. Kim and S. Okamoto, 11/1/93, (PB94-142775, A08, MF-A02).
NCEER-93-0021 "Finite Element Modeling of Elastomeric Seismic Isolation Bearings," by L.J. Billings, Supervised by R.
Shepherd, 11/8/93, not available.
NCEER-93-0022 "Seismic Vulnerability of Equipment in Critical Facilities: Life-Safety and Operational Consequences," by
K. Porter, G.S. Johnson, M.M. Zadeh, C. Scawthorn and S. Eder, 11/24/93, (PB94-181765, A16, MF-A03).
NCEER-93-0023 "Hokkaido Nansei-oki, Japan Earthquake of July 12, 1993, by P.I. Yanev and C.R. Scawthorn, 12/23/93,
(PB94-181500, A07, MF-A01).
NCEER-94-0001 "An Evaluation of Seismic Serviceability of Water Supply Networks with Application to the San Francisco
Auxiliary Water Supply System," by I. Markov, Supervised by M. Grigoriu and T. O'Rourke, 1/21/94,
(PB94-204013, A07, MF-A02).
Experimental and Analytical Study of Systems Consisting of Sliding Bearings, Rubber Restoring Force
Devices and Fluid Dampers," Volumes I and II, by P. Tsopelas, S. Okamoto, M.C. Constantinou, D. Ozaki
and S. Fujii, 2/4/94, (PB94-181740, A09, MF-A02 and PB94-181757, A12, MF-A03).
NCEER-94-0003 "A Markov Model for Local and Global Damage Indices in Seismic Analysis," by S. Rahman and M.
Grigoriu, 2/18/94, (PB94-206000, A12, MF-A03).
NCEER-94-0004 "Proceedings from the NCEER Workshop on Seismic Response of Masonry Infills," edited by D.P. Abrams,
3/1/94, (PB94-180783, A07, MF-A02).
NCEER-94-0005 "The Northridge, California Earthquake of January 17, 1994: General Reconnaissance Report," edited by
J.D. Goltz, 3/11/94, (PB94-193943, A10, MF-A03).
NCEER-94-0006 "Seismic Energy Based Fatigue Damage Analysis of Bridge Columns: Part I - Evaluation of Seismic
Capacity," by G.A. Chang and J.B. Mander, 3/14/94, (PB94-219185, A11, MF-A03).
NCEER-94-0007 "Seismic Isolation of Multi-Story Frame Structures Using Spherical Sliding Isolation Systems," by T.M. Al-
Hussaini, V.A. Zayas and M.C. Constantinou, 3/17/94, (PB94-193745, A09, MF-A02).
NCEER-94-0008 "The Northridge, California Earthquake of January 17, 1994: Performance of Highway Bridges," edited by
I.G. Buckle, 3/24/94, (PB94-193851, A06, MF-A02).
NCEER-94-0009 "Proceedings of the Third U.S.-Japan Workshop on Earthquake Protective Systems for Bridges," edited by
I.G. Buckle and I. Friedland, 3/31/94, (PB94-195815, A99, MF-A06).
339
NCEER-94-0010 "3D-BASIS-ME: Computer Program for Nonlinear Dynamic Analysis of Seismically Isolated Single and
Multiple Structures and Liquid Storage Tanks," by P.C. Tsopelas, M.C. Constantinou and A.M. Reinhorn,
4/12/94, (PB94-204922, A09, MF-A02).
NCEER-94-0011 "The Northridge, California Earthquake of January 17, 1994: Performance of Gas Transmission Pipelines,"
by T.D. O'Rourke and M.C. Palmer, 5/16/94, (PB94-204989, A05, MF-A01).
NCEER-94-0012 "Feasibility Study of Replacement Procedures and Earthquake Performance Related to Gas Transmission
Pipelines," by T.D. O'Rourke and M.C. Palmer, 5/25/94, (PB94-206638, A09, MF-A02).
NCEER-94-0013 "Seismic Energy Based Fatigue Damage Analysis of Bridge Columns: Part II - Evaluation of Seismic
Demand," by G.A. Chang and J.B. Mander, 6/1/94, (PB95-18106, A08, MF-A02).
Experimental and Analytical Study of a System Consisting of Sliding Bearings and Fluid Restoring
Force/Damping Devices," by P. Tsopelas and M.C. Constantinou, 6/13/94, (PB94-219144, A10, MF-A03).
NCEER-94-0015 "Generation of Hazard-Consistent Fragility Curves for Seismic Loss Estimation Studies," by H. Hwang and
J-R. Huo, 6/14/94, (PB95-181996, A09, MF-A02).
NCEER-94-0016 "Seismic Study of Building Frames with Added Energy-Absorbing Devices," by W.S. Pong, C.S. Tsai and
G.C. Lee, 6/20/94, (PB94-219136, A10, A03).
NCEER-94-0017 "Sliding Mode Control for Seismic-Excited Linear and Nonlinear Civil Engineering Structures," by J. Yang,
J. Wu, A. Agrawal and Z. Li, 6/21/94, (PB95-138483, A06, MF-A02).
NCEER-94-0018 "3D-BASIS-TABS Version 2.0: Computer Program for Nonlinear Dynamic Analysis of Three Dimensional
Base Isolated Structures," by A.M. Reinhorn, S. Nagarajaiah, M.C. Constantinou, P. Tsopelas and R. Li,
6/22/94, (PB95-182176, A08, MF-A02).
NCEER-94-0019 "Proceedings of the International Workshop on Civil Infrastructure Systems: Application of Intelligent
Systems and Advanced Materials on Bridge Systems," Edited by G.C. Lee and K.C. Chang, 7/18/94, (PB95-
252474, A20, MF-A04).
NCEER-94-0020 "Study of Seismic Isolation Systems for Computer Floors," by V. Lambrou and M.C. Constantinou, 7/19/94,
(PB95-138533, A10, MF-A03).
NCEER-94-0021 "Proceedings of the U.S.-Italian Workshop on Guidelines for Seismic Evaluation and Rehabilitation of
Unreinforced Masonry Buildings," Edited by D.P. Abrams and G.M. Calvi, 7/20/94, (PB95-138749, A13,
MF-A03).
Experimental and Analytical Study of a System Consisting of Lubricated PTFE Sliding Bearings and Mild
Steel Dampers," by P. Tsopelas and M.C. Constantinou, 7/22/94, (PB95-182184, A08, MF-A02).
NCEER-94-0023 “Development of Reliability-Based Design Criteria for Buildings Under Seismic Load,” by Y.K. Wen, H.
Hwang and M. Shinozuka, 8/1/94, (PB95-211934, A08, MF-A02).
NCEER-94-0024 “Experimental Verification of Acceleration Feedback Control Strategies for an Active Tendon System,” by
S.J. Dyke, B.F. Spencer, Jr., P. Quast, M.K. Sain, D.C. Kaspari, Jr. and T.T. Soong, 8/29/94, (PB95-212320,
A05, MF-A01).
NCEER-94-0025 “Seismic Retrofitting Manual for Highway Bridges,” Edited by I.G. Buckle and I.F. Friedland, published by
the Federal Highway Administration (PB95-212676, A15, MF-A03).
NCEER-94-0026 “Proceedings from the Fifth U.S.-Japan Workshop on Earthquake Resistant Design of Lifeline Facilities and
Countermeasures Against Soil Liquefaction,” Edited by T.D. O’Rourke and M. Hamada, 11/7/94, (PB95-
220802, A99, MF-E08).
340
NCEER-95-0001 “Experimental and Analytical Investigation of Seismic Retrofit of Structures with Supplemental Damping:
Part 1 - Fluid Viscous Damping Devices,” by A.M. Reinhorn, C. Li and M.C. Constantinou, 1/3/95, (PB95-
266599, A09, MF-A02).
NCEER-95-0002 “Experimental and Analytical Study of Low-Cycle Fatigue Behavior of Semi-Rigid Top-And-Seat Angle
Connections,” by G. Pekcan, J.B. Mander and S.S. Chen, 1/5/95, (PB95-220042, A07, MF-A02).
NCEER-95-0003 “NCEER-ATC Joint Study on Fragility of Buildings,” by T. Anagnos, C. Rojahn and A.S. Kiremidjian,
1/20/95, (PB95-220026, A06, MF-A02).
NCEER-95-0004 “Nonlinear Control Algorithms for Peak Response Reduction,” by Z. Wu, T.T. Soong, V. Gattulli and R.C.
Lin, 2/16/95, (PB95-220349, A05, MF-A01).
NCEER-95-0005 “Pipeline Replacement Feasibility Study: A Methodology for Minimizing Seismic and Corrosion Risks to
Underground Natural Gas Pipelines,” by R.T. Eguchi, H.A. Seligson and D.G. Honegger, 3/2/95, (PB95-
252326, A06, MF-A02).
NCEER-95-0006 “Evaluation of Seismic Performance of an 11-Story Frame Building During the 1994 Northridge
Earthquake,” by F. Naeim, R. DiSulio, K. Benuska, A. Reinhorn and C. Li, not available.
NCEER-95-0007 “Prioritization of Bridges for Seismic Retrofitting,” by N. Basöz and A.S. Kiremidjian, 4/24/95, (PB95-
252300, A08, MF-A02).
NCEER-95-0008 “Method for Developing Motion Damage Relationships for Reinforced Concrete Frames,” by A. Singhal and
A.S. Kiremidjian, 5/11/95, (PB95-266607, A06, MF-A02).
Part II - Friction Devices,” by C. Li and A.M. Reinhorn, 7/6/95, (PB96-128087, A11, MF-A03).
NCEER-95-0010 “Experimental Performance and Analytical Study of a Non-Ductile Reinforced Concrete Frame Structure
Retrofitted with Elastomeric Spring Dampers,” by G. Pekcan, J.B. Mander and S.S. Chen, 7/14/95, (PB96-
137161, A08, MF-A02).
NCEER-95-0011 “Development and Experimental Study of Semi-Active Fluid Damping Devices for Seismic Protection of
Structures,” by M.D. Symans and M.C. Constantinou, 8/3/95, (PB96-136940, A23, MF-A04).
NCEER-95-0012 “Real-Time Structural Parameter Modification (RSPM): Development of Innervated Structures,” by Z.

Liang, M. Tong and G.C. Lee, 4/11/95, (PB96-137153, A06, MF-A01).
Part III - Viscous Damping Walls,” by A.M. Reinhorn and C. Li, 10/1/95, (PB96-176409, A11, MF-A03).
NCEER-95-0014 “Seismic Fragility Analysis of Equipment and Structures in a Memphis Electric Substation,” by J-R. Huo and
H.H.M. Hwang, 8/10/95, (PB96-128087, A09, MF-A02).
NCEER-95-0015 “The Hanshin-Awaji Earthquake of January 17, 1995: Performance of Lifelines,” Edited by M. Shinozuka,
11/3/95, (PB96-176383, A15, MF-A03).
NCEER-95-0016 “Highway Culvert Performance During Earthquakes,” by T.L. Youd and C.J. Beckman, available as
NCEER-96-0015.
NCEER-95-0017 “The Hanshin-Awaji Earthquake of January 17, 1995: Performance of Highway Bridges,” Edited by I.G.
Buckle, 12/1/95, not available.
NCEER-95-0018 “Modeling of Masonry Infill Panels for Structural Analysis,” by A.M. Reinhorn, A. Madan, R.E. Valles, Y.
Reichmann and J.B. Mander, 12/8/95, (PB97-110886, MF-A01, A06).
NCEER-95-0019 “Optimal Polynomial Control for Linear and Nonlinear Structures,” by A.K. Agrawal and J.N. Yang,
12/11/95, (PB96-168737, A07, MF-A02).
341
NCEER-95-0020 “Retrofit of Non-Ductile Reinforced Concrete Frames Using Friction Dampers,” by R.S. Rao, P. Gergely and
R.N. White, 12/22/95, (PB97-133508, A10, MF-A02).
NCEER-95-0021 “Parametric Results for Seismic Response of Pile-Supported Bridge Bents,” by G. Mylonakis, A. Nikolaou
and G. Gazetas, 12/22/95, (PB97-100242, A12, MF-A03).
NCEER-95-0022 “Kinematic Bending Moments in Seismically Stressed Piles,” by A. Nikolaou, G. Mylonakis and G. Gazetas,
12/23/95, (PB97-113914, MF-A03, A13).
NCEER-96-0001 “Dynamic Response of Unreinforced Masonry Buildings with Flexible Diaphragms,” by A.C. Costley and
D.P. Abrams,” 10/10/96, (PB97-133573, MF-A03, A15).
NCEER-96-0002 “State of the Art Review: Foundations and Retaining Structures,” by I. Po Lam, not available.
NCEER-96-0003 “Ductility of Rectangular Reinforced Concrete Bridge Columns with Moderate Confinement,” by N. Wehbe,
M. Saiidi, D. Sanders and B. Douglas, 11/7/96, (PB97-133557, A06, MF-A02).
NCEER-96-0004 “Proceedings of the Long-Span Bridge Seismic Research Workshop,” edited by I.G. Buckle and I.M.
Friedland, not available.
NCEER-96-0005 “Establish Representative Pier Types for Comprehensive Study: Eastern United States,” by J. Kulicki and Z.
Prucz, 5/28/96, (PB98-119217, A07, MF-A02).
NCEER-96-0006 “Establish Representative Pier Types for Comprehensive Study: Western United States,” by R. Imbsen, R.A.
Schamber and T.A. Osterkamp, 5/28/96, (PB98-118607, A07, MF-A02).
NCEER-96-0007 “Nonlinear Control Techniques for Dynamical Systems with Uncertain Parameters,” by R.G. Ghanem and
M.I. Bujakov, 5/27/96, (PB97-100259, A17, MF-A03).
NCEER-96-0008 “Seismic Evaluation of a 30-Year Old Non-Ductile Highway Bridge Pier and Its Retrofit,” by J.B. Mander,
B. Mahmoodzadegan, S. Bhadra and S.S. Chen, 5/31/96, (PB97-110902, MF-A03, A10).
NCEER-96-0009 “Seismic Performance of a Model Reinforced Concrete Bridge Pier Before and After Retrofit,” by J.B.
Mander, J.H. Kim and C.A. Ligozio, 5/31/96, (PB97-110910, MF-A02, A10).
NCEER-96-0010 “IDARC2D Version 4.0: A Computer Program for the Inelastic Damage Analysis of Buildings,” by R.E.
Valles, A.M. Reinhorn, S.K. Kunnath, C. Li and A. Madan, 6/3/96, (PB97-100234, A17, MF-A03).
NCEER-96-0011 “Estimation of the Economic Impact of Multiple Lifeline Disruption: Memphis Light, Gas and Water
Division Case Study,” by S.E. Chang, H.A. Seligson and R.T. Eguchi, 8/16/96, (PB97-133490, A11, MF-
A03).
NCEER-96-0012 “Proceedings from the Sixth Japan-U.S. Workshop on Earthquake Resistant Design of Lifeline Facilities and
Countermeasures Against Soil Liquefaction, Edited by M. Hamada and T. O’Rourke, 9/11/96, (PB97-
133581, A99, MF-A06).
NCEER-96-0013 “Chemical Hazards, Mitigation and Preparedness in Areas of High Seismic Risk: A Methodology for
Estimating the Risk of Post-Earthquake Hazardous Materials Release,” by H.A. Seligson, R.T. Eguchi, K.J.
Tierney and K. Richmond, 11/7/96, (PB97-133565, MF-A02, A08).
NCEER-96-0014 “Response of Steel Bridge Bearings to Reversed Cyclic Loading,” by J.B. Mander, D-K. Kim, S.S. Chen and
G.J. Premus, 11/13/96, (PB97-140735, A12, MF-A03).
NCEER-96-0015 “Highway Culvert Performance During Past Earthquakes,” by T.L. Youd and C.J. Beckman, 11/25/96,
(PB97-133532, A06, MF-A01).
NCEER-97-0001 “Evaluation, Prevention and Mitigation of Pounding Effects in Building Structures,” by R.E. Valles and
A.M. Reinhorn, 2/20/97, (PB97-159552, A14, MF-A03).
NCEER-97-0002 “Seismic Design Criteria for Bridges and Other Highway Structures,” by C. Rojahn, R. Mayes, D.G.
Anderson, J. Clark, J.H. Hom, R.V. Nutt and M.J. O’Rourke, 4/30/97, (PB97-194658, A06, MF-A03).
342
NCEER-97-0003 “Proceedings of the U.S.-Italian Workshop on Seismic Evaluation and Retrofit,” Edited by D.P. Abrams and
G.M. Calvi, 3/19/97, (PB97-194666, A13, MF-A03).
NCEER-97-0004 "Investigation of Seismic Response of Buildings with Linear and Nonlinear Fluid Viscous Dampers," by
A.A. Seleemah and M.C. Constantinou, 5/21/97, (PB98-109002, A15, MF-A03).
NCEER-97-0005 "Proceedings of the Workshop on Earthquake Engineering Frontiers in Transportation Facilities," edited by
G.C. Lee and I.M. Friedland, 8/29/97, (PB98-128911, A25, MR-A04).
NCEER-97-0006 "Cumulative Seismic Damage of Reinforced Concrete Bridge Piers," by S.K. Kunnath, A. El-Bahy, A.
Taylor and W. Stone, 9/2/97, (PB98-108814, A11, MF-A03).
NCEER-97-0007 "Structural Details to Accommodate Seismic Movements of Highway Bridges and Retaining Walls," by R.A.
Imbsen, R.A. Schamber, E. Thorkildsen, A. Kartoum, B.T. Martin, T.N. Rosser and J.M. Kulicki, 9/3/97,
(PB98-108996, A09, MF-A02).
NCEER-97-0008 "A Method for Earthquake Motion-Damage Relationships with Application to Reinforced Concrete Frames,"
by A. Singhal and A.S. Kiremidjian, 9/10/97, (PB98-108988, A13, MF-A03).
NCEER-97-0009 "Seismic Analysis and Design of Bridge Abutments Considering Sliding and Rotation," by K. Fishman and
R. Richards, Jr., 9/15/97, (PB98-108897, A06, MF-A02).
NCEER-97-0010 "Proceedings of the FHWA/NCEER Workshop on the National Representation of Seismic Ground Motion
for New and Existing Highway Facilities," edited by I.M. Friedland, M.S. Power and R.L. Mayes, 9/22/97,
(PB98-128903, A21, MF-A04).
NCEER-97-0011 "Seismic Analysis for Design or Retrofit of Gravity Bridge Abutments," by K.L. Fishman, R. Richards, Jr.
and R.C. Divito, 10/2/97, (PB98-128937, A08, MF-A02).
NCEER-97-0012 "Evaluation of Simplified Methods of Analysis for Yielding Structures," by P. Tsopelas, M.C. Constantinou,
C.A. Kircher and A.S. Whittaker, 10/31/97, (PB98-128929, A10, MF-A03).
NCEER-97-0013 "Seismic Design of Bridge Columns Based on Control and Repairability of Damage," by C-T. Cheng and
J.B. Mander, 12/8/97, (PB98-144249, A11, MF-A03).
NCEER-97-0014 "Seismic Resistance of Bridge Piers Based on Damage Avoidance Design," by J.B. Mander and C-T. Cheng,
12/10/97, (PB98-144223, A09, MF-A02).
NCEER-97-0015 “Seismic Response of Nominally Symmetric Systems with Strength Uncertainty,” by S. Balopoulou and M.
Grigoriu, 12/23/97, (PB98-153422, A11, MF-A03).
NCEER-97-0016 “Evaluation of Seismic Retrofit Methods for Reinforced Concrete Bridge Columns,” by T.J. Wipf, F.W.
Klaiber and F.M. Russo, 12/28/97, (PB98-144215, A12, MF-A03).
NCEER-97-0017 “Seismic Fragility of Existing Conventional Reinforced Concrete Highway Bridges,” by C.L. Mullen and
A.S. Cakmak, 12/30/97, (PB98-153406, A08, MF-A02).
NCEER-97-0018 “Loss Asssessment of Memphis Buildings,” edited by D.P. Abrams and M. Shinozuka, 12/31/97, (PB98-
144231, A13, MF-A03).
NCEER-97-0019 “Seismic Evaluation of Frames with Infill Walls Using Quasi-static Experiments,” by K.M. Mosalam, R.N.
NCEER-97-0020 “Seismic Evaluation of Frames with Infill Walls Using Pseudo-dynamic Experiments,” by K.M. Mosalam,
R.N. White and P. Gergely, 12/31/97, (PB98-153430, A07, MF-A02).
NCEER-97-0021 “Computational Strategies for Frames with Infill Walls: Discrete and Smeared Crack Analyses and Seismic
Fragility,” by K.M. Mosalam, R.N. White and P. Gergely, 12/31/97, (PB98-153414, A10, MF-A02).
343
NCEER-97-0022 “Proceedings of the NCEER Workshop on Evaluation of Liquefaction Resistance of Soils,” edited by T.L.
Youd and I.M. Idriss, 12/31/97, (PB98-155617, A15, MF-A03).
MCEER-98-0001 “Extraction of Nonlinear Hysteretic Properties of Seismically Isolated Bridges from Quick-Release Field
Tests,” by Q. Chen, B.M. Douglas, E.M. Maragakis and I.G. Buckle, 5/26/98, (PB99-118838, A06, MF-
A01).
MCEER-98-0002 “Methodologies for Evaluating the Importance of Highway Bridges,” by A. Thomas, S. Eshenaur and J.
Kulicki, 5/29/98, (PB99-118846, A10, MF-A02).
MCEER-98-0003 “Capacity Design of Bridge Piers and the Analysis of Overstrength,” by J.B. Mander, A. Dutta and P. Goel,
6/1/98, (PB99-118853, A09, MF-A02).
MCEER-98-0004 “Evaluation of Bridge Damage Data from the Loma Prieta and Northridge, California Earthquakes,” by N.
Basoz and A. Kiremidjian, 6/2/98, (PB99-118861, A15, MF-A03).
MCEER-98-0005 “Screening Guide for Rapid Assessment of Liquefaction Hazard at Highway Bridge Sites,” by T. L. Youd,
6/16/98, (PB99-118879, A06, not available on microfiche).
MCEER-98-0006 “Structural Steel and Steel/Concrete Interface Details for Bridges,” by P. Ritchie, N. Kauhl and J. Kulicki,
7/13/98, (PB99-118945, A06, MF-A01).
MCEER-98-0007 “Capacity Design and Fatigue Analysis of Confined Concrete Columns,” by A. Dutta and J.B. Mander,
7/14/98, (PB99-118960, A14, MF-A03).
MCEER-98-0008 “Proceedings of the Workshop on Performance Criteria for Telecommunication Services Under Earthquake
Conditions,” edited by A.J. Schiff, 7/15/98, (PB99-118952, A08, MF-A02).
MCEER-98-0009 “Fatigue Analysis of Unconfined Concrete Columns,” by J.B. Mander, A. Dutta and J.H. Kim, 9/12/98,
(PB99-123655, A10, MF-A02).
MCEER-98-0010 “Centrifuge Modeling of Cyclic Lateral Response of Pile-Cap Systems and Seat-Type Abutments in Dry
Sands,” by A.D. Gadre and R. Dobry, 10/2/98, (PB99-123606, A13, MF-A03).
MCEER-98-0011 “IDARC-BRIDGE: A Computational Platform for Seismic Damage Assessment of Bridge Structures,” by
A.M. Reinhorn, V. Simeonov, G. Mylonakis and Y. Reichman, 10/2/98, (PB99-162919, A15, MF-A03).
MCEER-98-0012 “Experimental Investigation of the Dynamic Response of Two Bridges Before and After Retrofitting with
Elastomeric Bearings,” by D.A. Wendichansky, S.S. Chen and J.B. Mander, 10/2/98, (PB99-162927, A15,
MF-A03).
MCEER-98-0013 “Design Procedures for Hinge Restrainers and Hinge Sear Width for Multiple-Frame Bridges,” by R. Des
Roches and G.L. Fenves, 11/3/98, (PB99-140477, A13, MF-A03).
MCEER-98-0014 “Response Modification Factors for Seismically Isolated Bridges,” by M.C. Constantinou and J.K. Quarshie,
11/3/98, (PB99-140485, A14, MF-A03).
MCEER-98-0015 “Proceedings of the U.S.-Italy Workshop on Seismic Protective Systems for Bridges,” edited by I.M. Friedland
and M.C. Constantinou, 11/3/98, (PB2000-101711, A22, MF-A04).
MCEER-98-0016 “Appropriate Seismic Reliability for Critical Equipment Systems: Recommendations Based on Regional
Analysis of Financial and Life Loss,” by K. Porter, C. Scawthorn, C. Taylor and N. Blais, 11/10/98, (PB99-
157265, A08, MF-A02).
MCEER-98-0017 “Proceedings of the U.S. Japan Joint Seminar on Civil Infrastructure Systems Research,” edited by M.
Shinozuka and A. Rose, 11/12/98, (PB99-156713, A16, MF-A03).
MCEER-98-0018 “Modeling of Pile Footings and Drilled Shafts for Seismic Design,” by I. PoLam, M. Kapuskar and D.
Chaudhuri, 12/21/98, (PB99-157257, A09, MF-A02).
344
MCEER-99-0001 "Seismic Evaluation of a Masonry Infilled Reinforced Concrete Frame by Pseudodynamic Testing," by S.G.
Buonopane and R.N. White, 2/16/99, (PB99-162851, A09, MF-A02).
MCEER-99-0002 "Response History Analysis of Structures with Seismic Isolation and Energy Dissipation Systems:
Verification Examples for Program SAP2000," by J. Scheller and M.C. Constantinou, 2/22/99, (PB99-
162869, A08, MF-A02).
MCEER-99-0003 "Experimental Study on the Seismic Design and Retrofit of Bridge Columns Including Axial Load Effects,"
by A. Dutta, T. Kokorina and J.B. Mander, 2/22/99, (PB99-162877, A09, MF-A02).
MCEER-99-0004 "Experimental Study of Bridge Elastomeric and Other Isolation and Energy Dissipation Systems with
Emphasis on Uplift Prevention and High Velocity Near-source Seismic Excitation," by A. Kasalanati and M.
C. Constantinou, 2/26/99, (PB99-162885, A12, MF-A03).
MCEER-99-0005 "Truss Modeling of Reinforced Concrete Shear-flexure Behavior," by J.H. Kim and J.B. Mander, 3/8/99,
(PB99-163693, A12, MF-A03).
MCEER-99-0006 "Experimental Investigation and Computational Modeling of Seismic Response of a 1:4 Scale Model Steel
Structure with a Load Balancing Supplemental Damping System," by G. Pekcan, J.B. Mander and S.S. Chen,
4/2/99, (PB99-162893, A11, MF-A03).
MCEER-99-0007 "Effect of Vertical Ground Motions on the Structural Response of Highway Bridges," by M.R. Button, C.J.
Cronin and R.L. Mayes, 4/10/99, (PB2000-101411, A10, MF-A03).
MCEER-99-0008 "Seismic Reliability Assessment of Critical Facilities: A Handbook, Supporting Documentation, and Model
Code Provisions," by G.S. Johnson, R.E. Sheppard, M.D. Quilici, S.J. Eder and C.R. Scawthorn, 4/12/99,
(PB2000-101701, A18, MF-A04).
MCEER-99-0009 "Impact Assessment of Selected MCEER Highway Project Research on the Seismic Design of Highway
Structures," by C. Rojahn, R. Mayes, D.G. Anderson, J.H. Clark, D'Appolonia Engineering, S. Gloyd and
R.V. Nutt, 4/14/99, (PB99-162901, A10, MF-A02).
MCEER-99-0010 "Site Factors and Site Categories in Seismic Codes," by R. Dobry, R. Ramos and M.S. Power, 7/19/99,
(PB2000-101705, A08, MF-A02).
MCEER-99-0011 "Restrainer Design Procedures for Multi-Span Simply-Supported Bridges," by M.J. Randall, M. Saiidi, E.
Maragakis and T. Isakovic, 7/20/99, (PB2000-101702, A10, MF-A02).
MCEER-99-0012 "Property Modification Factors for Seismic Isolation Bearings," by M.C. Constantinou, P. Tsopelas, A.
Kasalanati and E. Wolff, 7/20/99, (PB2000-103387, A11, MF-A03).
MCEER-99-0013 "Critical Seismic Issues for Existing Steel Bridges," by P. Ritchie, N. Kauhl and J. Kulicki, 7/20/99,
(PB2000-101697, A09, MF-A02).
MCEER-99-0014 "Nonstructural Damage Database," by A. Kao, T.T. Soong and A. Vender, 7/24/99, (PB2000-101407, A06,
MF-A01).
MCEER-99-0015 "Guide to Remedial Measures for Liquefaction Mitigation at Existing Highway Bridge Sites," by H.G.
Cooke and J. K. Mitchell, 7/26/99, (PB2000-101703, A11, MF-A03).
MCEER-99-0016 "Proceedings of the MCEER Workshop on Ground Motion Methodologies for the Eastern United States,"
edited by N. Abrahamson and A. Becker, 8/11/99, (PB2000-103385, A07, MF-A02).
MCEER-99-0017 "Quindío, Colombia Earthquake of January 25, 1999: Reconnaissance Report," by A.P. Asfura and P.J.
Flores, 10/4/99, (PB2000-106893, A06, MF-A01).
MCEER-99-0018 "Hysteretic Models for Cyclic Behavior of Deteriorating Inelastic Structures," by M.V. Sivaselvan and A.M.
Reinhorn, 11/5/99, (PB2000-103386, A08, MF-A02).
345
MCEER-99-0019 "Proceedings of the 7th U.S.- Japan Workshop on Earthquake Resistant Design of Lifeline Facilities and
Countermeasures Against Soil Liquefaction," edited by T.D. O'Rourke, J.P. Bardet and M. Hamada,
11/19/99, (PB2000-103354, A99, MF-A06).
MCEER-99-0020 "Development of Measurement Capability for Micro-Vibration Evaluations with Application to Chip
Fabrication Facilities," by G.C. Lee, Z. Liang, J.W. Song, J.D. Shen and W.C. Liu, 12/1/99, (PB2000-
105993, A08, MF-A02).
MCEER-99-0021 "Design and Retrofit Methodology for Building Structures with Supplemental Energy Dissipating Systems,"
by G. Pekcan, J.B. Mander and S.S. Chen, 12/31/99, (PB2000-105994, A11, MF-A03).
MCEER-00-0001 "The Marmara, Turkey Earthquake of August 17, 1999: Reconnaissance Report," edited by C. Scawthorn;
with major contributions by M. Bruneau, R. Eguchi, T. Holzer, G. Johnson, J. Mander, J. Mitchell, W.
Mitchell, A. Papageorgiou, C. Scaethorn, and G. Webb, 3/23/00, (PB2000-106200, A11, MF-A03).
MCEER-00-0002 "Proceedings of the MCEER Workshop for Seismic Hazard Mitigation of Health Care Facilities," edited by
G.C. Lee, M. Ettouney, M. Grigoriu, J. Hauer and J. Nigg, 3/29/00, (PB2000-106892, A08, MF-A02).
MCEER-00-0003 "The Chi-Chi, Taiwan Earthquake of September 21, 1999: Reconnaissance Report," edited by G.C. Lee and
C.H. Loh, with major contributions by G.C. Lee, M. Bruneau, I.G. Buckle, S.E. Chang, P.J. Flores, T.D.
O'Rourke, M. Shinozuka, T.T. Soong, C-H. Loh, K-C. Chang, Z-J. Chen, J-S. Hwang, M-L. Lin, G-Y. Liu,
K-C. Tsai, G.C. Yao and C-L. Yen, 4/30/00, (PB2001-100980, A10, MF-A02).
MCEER-00-0004 "Seismic Retrofit of End-Sway Frames of Steel Deck-Truss Bridges with a Supplemental Tendon System:
Experimental and Analytical Investigation," by G. Pekcan, J.B. Mander and S.S. Chen, 7/1/00, (PB2001-
100982, A10, MF-A02).
MCEER-00-0005 "Sliding Fragility of Unrestrained Equipment in Critical Facilities," by W.H. Chong and T.T. Soong, 7/5/00,
(PB2001-100983, A08, MF-A02).
MCEER-00-0006 "Seismic Response of Reinforced Concrete Bridge Pier Walls in the Weak Direction," by N. Abo-Shadi, M.
Saiidi and D. Sanders, 7/17/00, (PB2001-100981, A17, MF-A03).
MCEER-00-0007 "Low-Cycle Fatigue Behavior of Longitudinal Reinforcement in Reinforced Concrete Bridge Columns," by
J. Brown and S.K. Kunnath, 7/23/00, (PB2001-104392, A08, MF-A02).
MCEER-00-0008 "Soil Structure Interaction of Bridges for Seismic Analysis," I. PoLam and H. Law, 9/25/00, (PB2001-
105397, A08, MF-A02).
MCEER-00-0009 "Proceedings of the First MCEER Workshop on Mitigation of Earthquake Disaster by Advanced
Technologies (MEDAT-1), edited by M. Shinozuka, D.J. Inman and T.D. O'Rourke, 11/10/00, (PB2001-
105399, A14, MF-A03).
MCEER-00-0010 "Development and Evaluation of Simplified Procedures for Analysis and Design of Buildings with Passive
Energy Dissipation Systems, Revision 01," by O.M. Ramirez, M.C. Constantinou, C.A. Kircher, A.S.
Whittaker, M.W. Johnson, J.D. Gomez and C. Chrysostomou, 11/16/01, (PB2001-105523, A23, MF-A04).
MCEER-00-0011 "Dynamic Soil-Foundation-Structure Interaction Analyses of Large Caissons," by C-Y. Chang, C-M. Mok,
Z-L. Wang, R. Settgast, F. Waggoner, M.A. Ketchum, H.M. Gonnermann and C-C. Chin, 12/30/00,
(PB2001-104373, A07, MF-A02).
MCEER-00-0012 "Experimental Evaluation of Seismic Performance of Bridge Restrainers," by A.G. Vlassis, E.M. Maragakis
and M. Saiid Saiidi, 12/30/00, (PB2001-104354, A09, MF-A02).
MCEER-00-0013 "Effect of Spatial Variation of Ground Motion on Highway Structures," by M. Shinozuka, V. Saxena and G.
Deodatis, 12/31/00, (PB2001-108755, A13, MF-A03).
MCEER-00-0014 "A Risk-Based Methodology for Assessing the Seismic Performance of Highway Systems," by S.D. Werner,
C.E. Taylor, J.E. Moore, II, J.S. Walton and S. Cho, 12/31/00, (PB2001-108756, A14, MF-A03).
346
MCEER-01-0001 “Experimental Investigation of P-Delta Effects to Collapse During Earthquakes,” by D. Vian and M.
Bruneau, 6/25/01, (PB2002-100534, A17, MF-A03).
MCEER-01-0002 “Proceedings of the Second MCEER Workshop on Mitigation of Earthquake Disaster by Advanced
Technologies (MEDAT-2),” edited by M. Bruneau and D.J. Inman, 7/23/01, (PB2002-100434, A16, MF-
A03).
MCEER-01-0003 “Sensitivity Analysis of Dynamic Systems Subjected to Seismic Loads,” by C. Roth and M. Grigoriu,
9/18/01, (PB2003-100884, A12, MF-A03).
MCEER-01-0004 “Overcoming Obstacles to Implementing Earthquake Hazard Mitigation Policies: Stage 1 Report,” by D.J.
Alesch and W.J. Petak, 12/17/01, (PB2002-107949, A07, MF-A02).
MCEER-01-0005 “Updating Real-Time Earthquake Loss Estimates: Methods, Problems and Insights,” by C.E. Taylor, S.E.
Chang and R.T. Eguchi, 12/17/01, (PB2002-107948, A05, MF-A01).
MCEER-01-0006 “Experimental Investigation and Retrofit of Steel Pile Foundations and Pile Bents Under Cyclic Lateral
Loadings,” by A. Shama, J. Mander, B. Blabac and S. Chen, 12/31/01, (PB2002-107950, A13, MF-A03).
MCEER-02-0001 “Assessment of Performance of Bolu Viaduct in the 1999 Duzce Earthquake in Turkey” by P.C. Roussis,
M.C. Constantinou, M. Erdik, E. Durukal and M. Dicleli, 5/8/02, (PB2003-100883, A08, MF-A02).
MCEER-02-0002 “Seismic Behavior of Rail Counterweight Systems of Elevators in Buildings,” by M.P. Singh, Rildova and
L.E. Suarez, 5/27/02. (PB2003-100882, A11, MF-A03).
MCEER-02-0003 “Development of Analysis and Design Procedures for Spread Footings,” by G. Mylonakis, G. Gazetas, S.
Nikolaou and A. Chauncey, 10/02/02, (PB2004-101636, A13, MF-A03, CD-A13).
MCEER-02-0004 “Bare-Earth Algorithms for Use with SAR and LIDAR Digital Elevation Models,” by C.K. Huyck, R.T.
Eguchi and B. Houshmand, 10/16/02, (PB2004-101637, A07, CD-A07).
MCEER-02-0005 “Review of Energy Dissipation of Compression Members in Concentrically Braced Frames,” by K.Lee and
M. Bruneau, 10/18/02, (PB2004-101638, A10, CD-A10).
MCEER-03-0001 “Experimental Investigation of Light-Gauge Steel Plate Shear Walls for the Seismic Retrofit of Buildings”
by J. Berman and M. Bruneau, 5/2/03, (PB2004-101622, A10, MF-A03, CD-A10).
MCEER-03-0002 “Statistical Analysis of Fragility Curves,” by M. Shinozuka, M.Q. Feng, H. Kim, T. Uzawa and T. Ueda,
6/16/03, (PB2004-101849, A09, CD-A09).
MCEER-03-0003 “Proceedings of the Eighth U.S.-Japan Workshop on Earthquake Resistant Design f Lifeline Facilities and
Countermeasures Against Liquefaction,” edited by M. Hamada, J.P. Bardet and T.D. O’Rourke, 6/30/03,
(PB2004-104386, A99, CD-A99).
MCEER-03-0004 “Proceedings of the PRC-US Workshop on Seismic Analysis and Design of Special Bridges,” edited by L.C.
Fan and G.C. Lee, 7/15/03, (PB2004-104387, A14, CD-A14).
MCEER-03-0005 “Urban Disaster Recovery: A Framework and Simulation Model,” by S.B. Miles and S.E. Chang, 7/25/03,
(PB2004-104388, A07, CD-A07).
MCEER-03-0006 “Behavior of Underground Piping Joints Due to Static and Dynamic Loading,” by R.D. Meis, M. Maragakis
and R. Siddharthan, 11/17/03, (PB2005-102194, A13, MF-A03, CD-A00).
MCEER-04-0001 “Experimental Study of Seismic Isolation Systems with Emphasis on Secondary System Response and
Verification of Accuracy of Dynamic Response History Analysis Methods,” by E. Wolff and M.
Constantinou, 1/16/04 (PB2005-102195, A99, MF-E08, CD-A00).
MCEER-04-0002 “Tension, Compression and Cyclic Testing of Engineered Cementitious Composite Materials,” by K. Kesner
and S.L. Billington, 3/1/04, (PB2005-102196, A08, CD-A08).
347
MCEER-04-0003 “Cyclic Testing of Braces Laterally Restrained by Steel Studs to Enhance Performance During Earthquakes,”
by O.C. Celik, J.W. Berman and M. Bruneau, 3/16/04, (PB2005-102197, A13, MF-A03, CD-A00).
MCEER-04-0004 “Methodologies for Post Earthquake Building Damage Detection Using SAR and Optical Remote Sensing:
Application to the August 17, 1999 Marmara, Turkey Earthquake,” by C.K. Huyck, B.J. Adams, S. Cho,
R.T. Eguchi, B. Mansouri and B. Houshmand, 6/15/04, (PB2005-104888, A10, CD-A00).
MCEER-04-0005 “Nonlinear Structural Analysis Towards Collapse Simulation: A Dynamical Systems Approach,” by M.V.
Sivaselvan and A.M. Reinhorn, 6/16/04, (PB2005-104889, A11, MF-A03, CD-A00).
MCEER-04-0006 “Proceedings of the Second PRC-US Workshop on Seismic Analysis and Design of Special Bridges,” edited
by G.C. Lee and L.C. Fan, 6/25/04, (PB2005-104890, A16, CD-A00).
MCEER-04-0007 “Seismic Vulnerability Evaluation of Axially Loaded Steel Built-up Laced Members,” by K. Lee and M.
Bruneau, 6/30/04, (PB2005-104891, A16, CD-A00).
MCEER-04-0008 “Evaluation of Accuracy of Simplified Methods of Analysis and Design of Buildings with Damping Systems
for Near-Fault and for Soft-Soil Seismic Motions,” by E.A. Pavlou and M.C. Constantinou, 8/16/04,
(PB2005-104892, A08, MF-A02, CD-A00).
MCEER-04-0009 “Assessment of Geotechnical Issues in Acute Care Facilities in California,” by M. Lew, T.D. O’Rourke, R.
Dobry and M. Koch, 9/15/04, (PB2005-104893, A08, CD-A00).
MCEER-04-0010 “Scissor-Jack-Damper Energy Dissipation System,” by A.N. Sigaher-Boyle and M.C. Constantinou, 12/1/04
(PB2005-108221).
MCEER-04-0011 “Seismic Retrofit of Bridge Steel Truss Piers Using a Controlled Rocking Approach,” by M. Pollino and M.
Bruneau, 12/20/04 (PB2006-105795).
MCEER-05-0001 “Experimental and Analytical Studies of Structures Seismically Isolated with an Uplift-Restraint Isolation
System,” by P.C. Roussis and M.C. Constantinou, 1/10/05 (PB2005-108222).
MCEER-05-0002 “A Versatile Experimentation Model for Study of Structures Near Collapse Applied to Seismic Evaluation of
Irregular Structures,” by D. Kusumastuti, A.M. Reinhorn and A. Rutenberg, 3/31/05 (PB2006-101523).
MCEER-05-0003 “Proceedings of the Third PRC-US Workshop on Seismic Analysis and Design of Special Bridges,” edited
by L.C. Fan and G.C. Lee, 4/20/05, (PB2006-105796).
MCEER-05-0004 “Approaches for the Seismic Retrofit of Braced Steel Bridge Piers and Proof-of-Concept Testing of an
Eccentrically Braced Frame with Tubular Link,” by J.W. Berman and M. Bruneau, 4/21/05 (PB2006-
101524).
MCEER-05-0005 “Simulation of Strong Ground Motions for Seismic Fragility Evaluation of Nonstructural Components in
Hospitals,” by A. Wanitkorkul and A. Filiatrault, 5/26/05 (PB2006-500027).
MCEER-05-0006 “Seismic Safety in California Hospitals: Assessing an Attempt to Accelerate the Replacement or Seismic
Retrofit of Older Hospital Facilities,” by D.J. Alesch, L.A. Arendt and W.J. Petak, 6/6/05 (PB2006-105794).
MCEER-05-0007 “Development of Seismic Strengthening and Retrofit Strategies for Critical Facilities Using Engineered
Cementitious Composite Materials,” by K. Kesner and S.L. Billington, 8/29/05 (PB2006-111701).
MCEER-05-0008 “Experimental and Analytical Studies of Base Isolation Systems for Seismic Protection of Power
Transformers,” by N. Murota, M.Q. Feng and G-Y. Liu, 9/30/05 (PB2006-111702).
MCEER-05-0009 “3D-BASIS-ME-MB: Computer Program for Nonlinear Dynamic Analysis of Seismically Isolated
Structures,” by P.C. Tsopelas, P.C. Roussis, M.C. Constantinou, R. Buchanan and A.M. Reinhorn, 10/3/05
(PB2006-111703).
MCEER-05-0010 “Steel Plate Shear Walls for Seismic Design and Retrofit of Building Structures,” by D. Vian and M.
Bruneau, 12/15/05 (PB2006-111704).
348
MCEER-05-0011 “The Performance-Based Design Paradigm,” by M.J. Astrella and A. Whittaker, 12/15/05 (PB2006-111705).
MCEER-06-0001 “Seismic Fragility of Suspended Ceiling Systems,” H. Badillo-Almaraz, A.S. Whittaker, A.M. Reinhorn and
G.P. Cimellaro, 2/4/06 (PB2006-111706).
MCEER-06-0002 “Multi-Dimensional Fragility of Structures,” by G.P. Cimellaro, A.M. Reinhorn and M. Bruneau, 3/1/06
(PB2007-106974, A09, MF-A02, CD A00).
MCEER-06-0003 “Built-Up Shear Links as Energy Dissipators for Seismic Protection of Bridges,” by P. Dusicka, A.M. Itani
and I.G. Buckle, 3/15/06 (PB2006-111708).
MCEER-06-0004 “Analytical Investigation of the Structural Fuse Concept,” by R.E. Vargas and M. Bruneau, 3/16/06
(PB2006-111709).
MCEER-06-0005 “Experimental Investigation of the Structural Fuse Concept,” by R.E. Vargas and M. Bruneau, 3/17/06
(PB2006-111710).
MCEER-06-0006 “Further Development of Tubular Eccentrically Braced Frame Links for the Seismic Retrofit of Braced Steel
Truss Bridge Piers,” by J.W. Berman and M. Bruneau, 3/27/06 (PB2007-105147).
MCEER-06-0007 “REDARS Validation Report,” by S. Cho, C.K. Huyck, S. Ghosh and R.T. Eguchi, 8/8/06 (PB2007-106983).
MCEER-06-0008 “Review of Current NDE Technologies for Post-Earthquake Assessment of Retrofitted Bridge Columns,” by
J.W. Song, Z. Liang and G.C. Lee, 8/21/06 (PB2007-106984).
MCEER-06-0009 “Liquefaction Remediation in Silty Soils Using Dynamic Compaction and Stone Columns,” by S.
Thevanayagam, G.R. Martin, R. Nashed, T. Shenthan, T. Kanagalingam and N. Ecemis, 8/28/06 (PB2007-
106985).
MCEER-06-0010 “Conceptual Design and Experimental Investigation of Polymer Matrix Composite Infill Panels for Seismic
Retrofitting,” by W. Jung, M. Chiewanichakorn and A.J. Aref, 9/21/06 (PB2007-106986).
MCEER-06-0011 “A Study of the Coupled Horizontal-Vertical Behavior of Elastomeric and Lead-Rubber Seismic Isolation
Bearings,” by G.P. Warn and A.S. Whittaker, 9/22/06 (PB2007-108679).
MCEER-06-0012 “Proceedings of the Fourth PRC-US Workshop on Seismic Analysis and Design of Special Bridges:
Advancing Bridge Technologies in Research, Design, Construction and Preservation,” Edited by L.C. Fan,
G.C. Lee and L. Ziang, 10/12/06 (PB2007-109042).
MCEER-06-0013 “Cyclic Response and Low Cycle Fatigue Characteristics of Plate Steels,” by P. Dusicka, A.M. Itani and I.G.
Buckle, 11/1/06 06 (PB2007-106987).
MCEER-06-0014 “Proceedings of the Second US-Taiwan Bridge Engineering Workshop,” edited by W.P. Yen, J. Shen, J-Y.
Chen and M. Wang, 11/15/06 (PB2008-500041).
MCEER-06-0015 “User Manual and Technical Documentation for the REDARSTM Import Wizard,” by S. Cho, S. Ghosh, C.K.
Huyck and S.D. Werner, 11/30/06 (PB2007-114766).
MCEER-06-0016 “Hazard Mitigation Strategy and Monitoring Technologies for Urban and Infrastructure Public Buildings:
Proceedings of the China-US Workshops,” edited by X.Y. Zhou, A.L. Zhang, G.C. Lee and M. Tong,
12/12/06 (PB2008-500018).
MCEER-07-0001 “Static and Kinetic Coefficients of Friction for Rigid Blocks,” by C. Kafali, S. Fathali, M. Grigoriu and A.S.
Whittaker, 3/20/07 (PB2007-114767).
MCEER-07-0002 “Hazard Mitigation Investment Decision Making: Organizational Response to Legislative Mandate,” by L.A.
Arendt, D.J. Alesch and W.J. Petak, 4/9/07 (PB2007-114768).
MCEER-07-0003 “Seismic Behavior of Bidirectional-Resistant Ductile End Diaphragms with Unbonded Braces in Straight or
Skewed Steel Bridges,” by O. Celik and M. Bruneau, 4/11/07 (PB2008-105141).
349
MCEER-07-0004 “Modeling Pile Behavior in Large Pile Groups Under Lateral Loading,” by A.M. Dodds and G.R. Martin,
4/16/07(PB2008-105142).
MCEER-07-0005 “Experimental Investigation of Blast Performance of Seismically Resistant Concrete-Filled Steel Tube
Bridge Piers,” by S. Fujikura, M. Bruneau and D. Lopez-Garcia, 4/20/07 (PB2008-105143).
MCEER-07-0006 “Seismic Analysis of Conventional and Isolated Liquefied Natural Gas Tanks Using Mechanical Analogs,”
by I.P. Christovasilis and A.S. Whittaker, 5/1/07, not available.
MCEER-07-0007 “Experimental Seismic Performance Evaluation of Isolation/Restraint Systems for Mechanical Equipment –
Part 1: Heavy Equipment Study,” by S. Fathali and A. Filiatrault, 6/6/07 (PB2008-105144).
MCEER-07-0008 “Seismic Vulnerability of Timber Bridges and Timber Substructures,” by A.A. Sharma, J.B. Mander, I.M.
Friedland and D.R. Allicock, 6/7/07 (PB2008-105145).
MCEER-07-0009 “Experimental and Analytical Study of the XY-Friction Pendulum (XY-FP) Bearing for Bridge
Applications,” by C.C. Marin-Artieda, A.S. Whittaker and M.C. Constantinou, 6/7/07 (PB2008-105191).
MCEER-07-0010 “Proceedings of the PRC-US Earthquake Engineering Forum for Young Researchers,” Edited by G.C. Lee
and X.Z. Qi, 6/8/07 (PB2008-500058).
MCEER-07-0011 “Design Recommendations for Perforated Steel Plate Shear Walls,” by R. Purba and M. Bruneau, 6/18/07,
(PB2008-105192).
MCEER-07-0012 “Performance of Seismic Isolation Hardware Under Service and Seismic Loading,” by M.C. Constantinou,
A.S. Whittaker, Y. Kalpakidis, D.M. Fenz and G.P. Warn, 8/27/07, (PB2008-105193).
MCEER-07-0013 “Experimental Evaluation of the Seismic Performance of Hospital Piping Subassemblies,” by E.R. Goodwin,
E. Maragakis and A.M. Itani, 9/4/07, (PB2008-105194).
MCEER-07-0014 “A Simulation Model of Urban Disaster Recovery and Resilience: Implementation for the 1994 Northridge
Earthquake,” by S. Miles and S.E. Chang, 9/7/07, (PB2008-106426).
MCEER-07-0015 “Statistical and Mechanistic Fragility Analysis of Concrete Bridges,” by M. Shinozuka, S. Banerjee and S-H.
Kim, 9/10/07, (PB2008-106427).
MCEER-07-0016 “Three-Dimensional Modeling of Inelastic Buckling in Frame Structures,” by M. Schachter and AM.
Reinhorn, 9/13/07, (PB2008-108125).
MCEER-07-0017 “Modeling of Seismic Wave Scattering on Pile Groups and Caissons,” by I. Po Lam, H. Law and C.T. Yang,
9/17/07 (PB2008-108150).
MCEER-07-0018 “Bridge Foundations: Modeling Large Pile Groups and Caissons for Seismic Design,” by I. Po Lam, H. Law
and G.R. Martin (Coordinating Author), 12/1/07 (PB2008-111190).
MCEER-07-0019 “Principles and Performance of Roller Seismic Isolation Bearings for Highway Bridges,” by G.C. Lee, Y.C.
Ou, Z. Liang, T.C. Niu and J. Song, 12/10/07 (PB2009-110466).
MCEER-07-0020 “Centrifuge Modeling of Permeability and Pinning Reinforcement Effects on Pile Response to Lateral
Spreading,” by L.L Gonzalez-Lagos, T. Abdoun and R. Dobry, 12/10/07 (PB2008-111191).
MCEER-07-0021 “Damage to the Highway System from the Pisco, Perú Earthquake of August 15, 2007,” by J.S. O’Connor,
L. Mesa and M. Nykamp, 12/10/07, (PB2008-108126).
MCEER-07-0022 “Experimental Seismic Performance Evaluation of Isolation/Restraint Systems for Mechanical Equipment –
Part 2: Light Equipment Study,” by S. Fathali and A. Filiatrault, 12/13/07 (PB2008-111192).
MCEER-07-0023 “Fragility Considerations in Highway Bridge Design,” by M. Shinozuka, S. Banerjee and S.H. Kim, 12/14/07
(PB2008-111193).
350
MCEER-07-0024 “Performance Estimates for Seismically Isolated Bridges,” by G.P. Warn and A.S. Whittaker, 12/30/07
(PB2008-112230).
MCEER-08-0001 “Seismic Performance of Steel Girder Bridge Superstructures with Conventional Cross Frames,” by L.P.
Carden, A.M. Itani and I.G. Buckle, 1/7/08, (PB2008-112231).
MCEER-08-0002 “Seismic Performance of Steel Girder Bridge Superstructures with Ductile End Cross Frames with Seismic
Isolators,” by L.P. Carden, A.M. Itani and I.G. Buckle, 1/7/08 (PB2008-112232).
MCEER-08-0003 “Analytical and Experimental Investigation of a Controlled Rocking Approach for Seismic Protection of
Bridge Steel Truss Piers,” by M. Pollino and M. Bruneau, 1/21/08 (PB2008-112233).
MCEER-08-0004 “Linking Lifeline Infrastructure Performance and Community Disaster Resilience: Models and Multi-
Stakeholder Processes,” by S.E. Chang, C. Pasion, K. Tatebe and R. Ahmad, 3/3/08 (PB2008-112234).
MCEER-08-0005 “Modal Analysis of Generally Damped Linear Structures Subjected to Seismic Excitations,” by J. Song, Y-L.
Chu, Z. Liang and G.C. Lee, 3/4/08 (PB2009-102311).
MCEER-08-0006 “System Performance Under Multi-Hazard Environments,” by C. Kafali and M. Grigoriu, 3/4/08 (PB2008-
112235).
MCEER-08-0007 “Mechanical Behavior of Multi-Spherical Sliding Bearings,” by D.M. Fenz and M.C. Constantinou, 3/6/08
(PB2008-112236).
MCEER-08-0008 “Post-Earthquake Restoration of the Los Angeles Water Supply System,” by T.H.P. Tabucchi and R.A.
Davidson, 3/7/08 (PB2008-112237).
MCEER-08-0009 “Fragility Analysis of Water Supply Systems,” by A. Jacobson and M. Grigoriu, 3/10/08 (PB2009-105545).
MCEER-08-0010 “Experimental Investigation of Full-Scale Two-Story Steel Plate Shear Walls with Reduced Beam Section
Connections,” by B. Qu, M. Bruneau, C-H. Lin and K-C. Tsai, 3/17/08 (PB2009-106368).
MCEER-08-0011 “Seismic Evaluation and Rehabilitation of Critical Components of Electrical Power Systems,” S. Ersoy, B.
Feizi, A. Ashrafi and M. Ala Saadeghvaziri, 3/17/08 (PB2009-105546).
MCEER-08-0012 “Seismic Behavior and Design of Boundary Frame Members of Steel Plate Shear Walls,” by B. Qu and M.
Bruneau, 4/26/08 . (PB2009-106744).
MCEER-08-0013 “Development and Appraisal of a Numerical Cyclic Loading Protocol for Quantifying Building System
Performance,” by A. Filiatrault, A. Wanitkorkul and M. Constantinou, 4/27/08 (PB2009-107906).
MCEER-08-0014 “Structural and Nonstructural Earthquake Design: The Challenge of Integrating Specialty Areas in Designing
Complex, Critical Facilities,” by W.J. Petak and D.J. Alesch, 4/30/08 (PB2009-107907).
MCEER-08-0015 “Seismic Performance Evaluation of Water Systems,” by Y. Wang and T.D. O’Rourke, 5/5/08 (PB2009-
107908).
MCEER-08-0016 “Seismic Response Modeling of Water Supply Systems,” by P. Shi and T.D. O’Rourke, 5/5/08 (PB2009-
107910).
MCEER-08-0017 “Numerical and Experimental Studies of Self-Centering Post-Tensioned Steel Frames,” by D. Wang and A.
Filiatrault, 5/12/08 (PB2009-110479).
MCEER-08-0018 “Development, Implementation and Verification of Dynamic Analysis Models for Multi-Spherical Sliding
Bearings,” by D.M. Fenz and M.C. Constantinou, 8/15/08 (PB2009-107911).
MCEER-08-0019 “Performance Assessment of Conventional and Base Isolated Nuclear Power Plants for Earthquake Blast
Loadings,” by Y.N. Huang, A.S. Whittaker and N. Luco, 10/28/08 (PB2009-107912).
351
MCEER-08-0020 “Remote Sensing for Resilient Multi-Hazard Disaster Response – Volume I: Introduction to Damage
Assessment Methodologies,” by B.J. Adams and R.T. Eguchi, 11/17/08 (PB2010-102695).
MCEER-08-0021 “Remote Sensing for Resilient Multi-Hazard Disaster Response – Volume II: Counting the Number of
Collapsed Buildings Using an Object-Oriented Analysis: Case Study of the 2003 Bam Earthquake,” by L.
Gusella, C.K. Huyck and B.J. Adams, 11/17/08 (PB2010-100925).
MCEER-08-0022 “Remote Sensing for Resilient Multi-Hazard Disaster Response – Volume III: Multi-Sensor Image Fusion
Techniques for Robust Neighborhood-Scale Urban Damage Assessment,” by B.J. Adams and A. McMillan,
11/17/08 (PB2010-100926).
MCEER-08-0023 “Remote Sensing for Resilient Multi-Hazard Disaster Response – Volume IV: A Study of Multi-Temporal
and Multi-Resolution SAR Imagery for Post-Katrina Flood Monitoring in New Orleans,” by A. McMillan,
J.G. Morley, B.J. Adams and S. Chesworth, 11/17/08 (PB2010-100927).
MCEER-08-0024 “Remote Sensing for Resilient Multi-Hazard Disaster Response – Volume V: Integration of Remote Sensing
Imagery and VIEWSTM Field Data for Post-Hurricane Charley Building Damage Assessment,” by J.A.
Womble, K. Mehta and B.J. Adams, 11/17/08 (PB2009-115532).
MCEER-08-0025 “Building Inventory Compilation for Disaster Management: Application of Remote Sensing and Statistical
Modeling,” by P. Sarabandi, A.S. Kiremidjian, R.T. Eguchi and B. J. Adams, 11/20/08 (PB2009-110484).
MCEER-08-0026 “New Experimental Capabilities and Loading Protocols for Seismic Qualification and Fragility Assessment
of Nonstructural Systems,” by R. Retamales, G. Mosqueda, A. Filiatrault and A. Reinhorn, 11/24/08
(PB2009-110485).
MCEER-08-0027 “Effects of Heating and Load History on the Behavior of Lead-Rubber Bearings,” by I.V. Kalpakidis and
M.C. Constantinou, 12/1/08 (PB2009-115533).
MCEER-08-0028 “Experimental and Analytical Investigation of Blast Performance of Seismically Resistant Bridge Piers,” by
S.Fujikura and M. Bruneau, 12/8/08 (PB2009-115534).
MCEER-08-0029 “Evolutionary Methodology for Aseismic Decision Support,” by Y. Hu and G. Dargush, 12/15/08.
MCEER-08-0030 “Development of a Steel Plate Shear Wall Bridge Pier System Conceived from a Multi-Hazard Perspective,”
by D. Keller and M. Bruneau, 12/19/08 (PB2010-102696).
MCEER-09-0001 “Modal Analysis of Arbitrarily Damped Three-Dimensional Linear Structures Subjected to Seismic
Excitations,” by Y.L. Chu, J. Song and G.C. Lee, 1/31/09 (PB2010-100922).
MCEER-09-0002 “Air-Blast Effects on Structural Shapes,” by G. Ballantyne, A.S. Whittaker, A.J. Aref and G.F. Dargush,
2/2/09 (PB2010-102697).
MCEER-09-0003 “Water Supply Performance During Earthquakes and Extreme Events,” by A.L. Bonneau and T.D.
O’Rourke, 2/16/09 (PB2010-100923).
MCEER-09-0004 “Generalized Linear (Mixed) Models of Post-Earthquake Ignitions,” by R.A. Davidson, 7/20/09 (PB2010-
102698).
MCEER-09-0005 “Seismic Testing of a Full-Scale Two-Story Light-Frame Wood Building: NEESWood Benchmark Test,” by
I.P. Christovasilis, A. Filiatrault and A. Wanitkorkul, 7/22/09 (PB2012-102401).
MCEER-09-0006 “IDARC2D Version 7.0: A Program for the Inelastic Damage Analysis of Structures,” by A.M. Reinhorn, H.
Roh, M. Sivaselvan, S.K. Kunnath, R.E. Valles, A. Madan, C. Li, R. Lobo and Y.J. Park, 7/28/09 (PB2010-
103199).
MCEER-09-0007 “Enhancements to Hospital Resiliency: Improving Emergency Planning for and Response to Hurricanes,” by
D.B. Hess and L.A. Arendt, 7/30/09 (PB2010-100924).
352
MCEER-09-0008 “Assessment of Base-Isolated Nuclear Structures for Design and Beyond-Design Basis Earthquake Shaking,”
by Y.N. Huang, A.S. Whittaker, R.P. Kennedy and R.L. Mayes, 8/20/09 (PB2010-102699).
MCEER-09-0009 “Quantification of Disaster Resilience of Health Care Facilities,” by G.P. Cimellaro, C. Fumo, A.M Reinhorn
and M. Bruneau, 9/14/09 (PB2010-105384).
MCEER-09-0010 “Performance-Based Assessment and Design of Squat Reinforced Concrete Shear Walls,” by C.K. Gulec and
A.S. Whittaker, 9/15/09 (PB2010-102700).
MCEER-09-0011 “Proceedings of the Fourth US-Taiwan Bridge Engineering Workshop,” edited by W.P. Yen, J.J. Shen, T.M.
Lee and R.B. Zheng, 10/27/09 (PB2010-500009).
MCEER-09-0012 “Proceedings of the Special International Workshop on Seismic Connection Details for Segmental Bridge
Construction,” edited by W. Phillip Yen and George C. Lee, 12/21/09 (PB2012-102402).
MCEER-10-0001 “Direct Displacement Procedure for Performance-Based Seismic Design of Multistory Woodframe
Structures,” by W. Pang and D. Rosowsky, 4/26/10 (PB2012-102403).
MCEER-10-0002 “Simplified Direct Displacement Design of Six-Story NEESWood Capstone Building and Pre-Test Seismic
Performance Assessment,” by W. Pang, D. Rosowsky, J. van de Lindt and S. Pei, 5/28/10 (PB2012-102404).
MCEER-10-0003 “Integration of Seismic Protection Systems in Performance-Based Seismic Design of Woodframed

Structures,” by J.K. Shinde and M.D. Symans, 6/18/10 (PB2012-102405).
MCEER-10-0004 “Modeling and Seismic Evaluation of Nonstructural Components: Testing Frame for Experimental
Evaluation of Suspended Ceiling Systems,” by A.M. Reinhorn, K.P. Ryu and G. Maddaloni, 6/30/10
(PB2012-102406).
MCEER-10-0005 “Analytical Development and Experimental Validation of a Structural-Fuse Bridge Pier Concept,” by S. El-
Bahey and M. Bruneau, 10/1/10 (PB2012-102407).
MCEER-10-0006 “A Framework for Defining and Measuring Resilience at the Community Scale: The PEOPLES Resilience
Framework,” by C.S. Renschler, A.E. Frazier, L.A. Arendt, G.P. Cimellaro, A.M. Reinhorn and M. Bruneau,
10/8/10 (PB2012-102408).
MCEER-10-0007 “Impact of Horizontal Boundary Elements Design on Seismic Behavior of Steel Plate Shear Walls,” by R.
Purba and M. Bruneau, 11/14/10 (PB2012-102409).
MCEER-10-0008 “Seismic Testing of a Full-Scale Mid-Rise Building: The NEESWood Capstone Test,” by S. Pei, J.W. van de
Lindt, S.E. Pryor, H. Shimizu, H. Isoda and D.R. Rammer, 12/1/10 (PB2012-102410).
MCEER-10-0009 “Modeling the Effects of Detonations of High Explosives to Inform Blast-Resistant Design,” by P. Sherkar,
A.S. Whittaker and A.J. Aref, 12/1/10 (PB2012-102411).
MCEER-10-0010 “L’Aquila Earthquake of April 6, 2009 in Italy: Rebuilding a Resilient City to Withstand Multiple Hazards,”
by G.P. Cimellaro, I.P. Christovasilis, A.M. Reinhorn, A. De Stefano and T. Kirova, 12/29/10.
MCEER-11-0001 “Numerical and Experimental Investigation of the Seismic Response of Light-Frame Wood Structures,” by
I.P. Christovasilis and A. Filiatrault, 8/8/11 (PB2012-102412).
MCEER-11-0002 “Seismic Design and Analysis of a Precast Segmental Concrete Bridge Model,” by M. Anagnostopoulou, A.
Filiatrault and A. Aref, 9/15/11.
MCEER-11-0003 ‘Proceedings of the Workshop on Improving Earthquake Response of Substation Equipment,” Edited by
A.M. Reinhorn, 9/19/11 (PB2012-102413).
MCEER-11-0004 “LRFD-Based Analysis and Design Procedures for Bridge Bearings and Seismic Isolators,” by M.C.
Constantinou, I. Kalpakidis, A. Filiatrault and R.A. Ecker Lay, 9/26/11.
353
MCEER-11-0005 “Experimental Seismic Evaluation, Model Parameterization, and Effects of Cold-Formed Steel-Framed
Gypsum Partition Walls on the Seismic Performance of an Essential Facility,” by R. Davies, R. Retamales,
G. Mosqueda and A. Filiatrault, 10/12/11.
MCEER-11-0006 “Modeling and Seismic Performance Evaluation of High Voltage Transformers and Bushings,” by A.M.
Reinhorn, K. Oikonomou, H. Roh, A. Schiff and L. Kempner, Jr., 10/3/11.
MCEER-11-0007 “Extreme Load Combinations: A Survey of State Bridge Engineers,” by G.C. Lee, Z. Liang, J.J. Shen and
J.S. O’Connor, 10/14/11.
MCEER-12-0001 “Simplified Analysis Procedures in Support of Performance Based Seismic Design,” by Y.N. Huang and
A.S. Whittaker.
MCEER-12-0002 “Seismic Protection of Electrical Transformer Bushing Systems by Stiffening Techniques,” by M. Koliou, A.
Filiatrault, A.M. Reinhorn and N. Oliveto, 6/1/12.
MCEER-12-0003 “Post-Earthquake Bridge Inspection Guidelines,” by J.S. O’Connor and S. Alampalli, 6/8/12.
MCEER-12-0004 “Integrated Design Methodology for Isolated Floor Systems in Single-Degree-of-Freedom Structural Fuse
Systems,” by S. Cui, M. Bruneau and M.C. Constantinou, 6/13/12.
MCEER-12-0005 “Characterizing the Rotational Components of Earthquake Ground Motion,” by D. Basu, A.S. Whittaker and
M.C. Constantinou, 6/15/12.
MCEER-12-0006 “Bayesian Fragility for Nonstructural Systems,” by C.H. Lee and M.D. Grigoriu, 9/12/12.
MCEER-12-0007 “A Numerical Model for Capturing the In-Plane Seismic Response of Interior Metal Stud Partition Walls,”
by R.L. Wood and T.C. Hutchinson, 9/12/12.
MCEER-12-0008 “Assessment of Floor Accelerations in Yielding Buildings,” by J.D. Wieser, G. Pekcan, A.E. Zaghi, A.M.
Itani and E. Maragakis, 10/5/12.
MCEER-13-0001 “Experimental Seismic Study of Pressurized Fire Sprinkler Piping Systems,” by Y. Tian, A. Filiatrault and
G. Mosqueda, 4/8/13.
MCEER-13-0002 “Enhancing Resource Coordination for Multi-Modal Evacuation Planning,” by D.B. Hess, B.W. Conley and
C.M. Farrell, 2/8/13.
MCEER-13-0003 “Seismic Response of Base Isolated Buildings Considering Pounding to Moat Walls,” by A. Masroor and G.
Mosqueda, 2/26/13.
MCEER-13-0004 “Seismic Response Control of Structures Using a Novel Adaptive Passive Negative Stiffness Device,” by
D.T.R. Pasala, A.A. Sarlis, S. Nagarajaiah, A.M. Reinhorn, M.C. Constantinou and D.P. Taylor, 6/10/13.
MCEER-13-0005 “Negative Stiffness Device for Seismic Protection of Structures,” by A.A. Sarlis, D.T.R. Pasala, M.C.
Constantinou, A.M. Reinhorn, S. Nagarajaiah and D.P. Taylor, 6/12/13.
MCEER-13-0006 “Emilia Earthquake of May 20, 2012 in Northern Italy: Rebuilding a Resilient Community to Withstand
Multiple Hazards,” by G.P. Cimellaro, M. Chiriatti, A.M. Reinhorn and L. Tirca, June 30, 2013.
MCEER-13-0007 “Precast Concrete Segmental Components and Systems for Accelerated Bridge Construction in Seismic
Regions,” by A.J. Aref, G.C. Lee, Y.C. Ou and P. Sideris, with contributions from K.C. Chang, S. Chen, A.
Filiatrault and Y. Zhou, June 13, 2013.
MCEER-13-0008 “A Study of U.S. Bridge Failures (1980-2012),” by G.C. Lee, S.B. Mohan, C. Huang and B.N. Fard, June 15,
2013.
MCEER-13-0009 “Development of a Database Framework for Modeling Damaged Bridges,” by G.C. Lee, J.C. Qi and C.
Huang, June 16, 2013.
354
MCEER-13-0010 “Model of Triple Friction Pendulum Bearing for General Geometric and Frictional Parameters and for Uplift
Conditions,” by A.A. Sarlis and M.C. Constantinou, July 1, 2013.
MCEER-13-0011 “Shake Table Testing of Triple Friction Pendulum Isolators under Extreme Conditions,” by A.A. Sarlis,
M.C. Constantinou and A.M. Reinhorn, July 2, 2013.
MCEER-13-0012 “Theoretical Framework for the Development of MH-LRFD,” by G.C. Lee (coordinating author), H.A
Capers, Jr., C. Huang, J.M. Kulicki, Z. Liang, T. Murphy, J.J.D. Shen, M. Shinozuka and P.W.H. Yen, July
31, 2013.
MCEER-13-0013 “Seismic Protection of Highway Bridges with Negative Stiffness Devices,” by N.K.A. Attary, M.D. Symans,
S. Nagarajaiah, A.M. Reinhorn, M.C. Constantinou, A.A. Sarlis, D.T.R. Pasala, and D.P. Taylor, September
3, 2014.
MCEER-14-0001 “Simplified Seismic Collapse Capacity-Based Evaluation and Design of Frame Buildings with and without
Supplemental Damping Systems,” by M. Hamidia, A. Filiatrault, and A. Aref, May 19, 2014.
MCEER-14-0002 “Comprehensive Analytical Seismic Fragility of Fire Sprinkler Piping Systems,” by Siavash Soroushian,
Emmanuel “Manos” Maragakis, Arash E. Zaghi, Alicia Echevarria, Yuan Tian and Andre Filiatrault, August
26, 2014.
MCEER-14-0003 “Hybrid Simulation of the Seismic Response of a Steel Moment Frame Building Structure through
Collapse,” by M. Del Carpio Ramos, G. Mosqueda and D.G. Lignos, October 30, 2014.
MCEER-14-0005 “Seismic Performance of Steel Plate Shear Walls Considering Various Design Approaches,” by R. Purba and
M. Bruneau, October 31, 2014.
MCEER-14-0006 “Air-Blast Effects on Civil Structures,” by Jinwon Shin, Andrew S. Whittaker, Amjad J. Aref and David
Cormie, October 30, 2014.
MCEER-14-0007 “Seismic Performance Evaluation of Precast Girders with Field-Cast Ultra High Performance Concrete
(UHPC) Connections,” by G.C. Lee, C. Huang, J. Song, and J. S. O’Connor, July 31, 2014.
MCEER-14-0008 “Post-Earthquake Fire Resistance of Ductile Concrete-Filled Double-Skin Tube Columns,” by Reza Imani,
Gilberto Mosqueda and Michel Bruneau, December 1, 2014.
MCEER-14-0009 “Cyclic Inelastic Behavior of Concrete Filled Sandwich Panel Walls Subjected to In-Plane Flexure,” by Y.
Alzeni and M. Bruneau, December 19, 2014.
MCEER-14-0010 “Analytical and Experimental Investigation of Self-Centering Steel Plate Shear Walls,” by D.M. Dowden
and M. Bruneau, December 19, 2014.
MCEER-15-0001 “Seismic Analysis of Multi‐story Unreinforced Masonry Buildings with Flexible Diaphragms,” by J.
Aleman, G. Mosqueda and A.S. Whittaker, June 12, 2015.
MCEER-15-0002 “Site Response, Soil-Structure Interaction and Structure-Soil-Structure Interaction for Performance
Assessment of Buildings and Nuclear Structures, by C. Bolisetti and A.S. Whittaker, June 15, 2015.
355
ISSN 1520-295X
Characterizing the Rotational Components of Earthquake Ground Motion

Characterizing the Rotational
Components of Earthquake
Ground Motion
by
Dhiman Basu, Andrew S. Whittaker
and Michael C. Constantinou
University at Buffalo, The State University of New York

133A Ketter Hall Buffalo, New York 14260-4300 Technical Report MCEER-12-0005
Phone: (716) 645-3391 Fax: (716) 645-3399
Email: [email protected] Web: http://mceer.buffalo.edu June 15, 2012
ISSN 1520-295X
MCEER-12-0005
This research was conducted at the University at Buffalo, State University of New York and was
supported by MCEER Thrust Area 3, Innovative Technologies.

Site Response, Soil-Structure Interaction and Structure-Soil-Structure Interaction For Performance Assessment of Buildings and Nuclear Structures

Uploaded by

Document Information

Original Description:

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

Site Response, Soil-Structure Interaction and Structure-Soil-Structure Interaction For Performance Assessment of Buildings and Nuclear Structures

Uploaded by

Copyright:

Available Formats

ISSN 1520-295X

Site Response, Soil-Structure Interaction

Technical Report MCEER-15-0002

Any opinions, findings, and conclusions or recommendations expressed in this

Chandrakanth Bolisetti1 and Andrew S. Whittaker2

Publication Date: June 15, 2015

Technical Report MCEER-15-0002

National Science Foundation Grant Number: CMMI-0830331

1 Postdoctoral Associate, Idaho National Laboratory; former Graduate Student, Department of

1.3 Report organization ............................................................................................................................ 5

SECTION 2 SITE-RESPONSE ANALYSIS: A REVIEW ..................................................................... 7

2.1 Dynamic behavior of soils during earthquakes .................................................................................. 7

2.2 Site-response analysis ....................................................................................................................... 11

3.1 Introduction ...................................................................................................................................... 29

3.3 Soil-structure interaction analysis using industry-standard numerical programs ............................. 38

3.4 Structure-soil-structure interaction ................................................................................................... 48

SECTION 4 AN ASSESSMENT OF SITE-RESPONSE ANALYSIS PROGRAMS ......................... 51

4.1 Introduction ...................................................................................................................................... 51

4.5 Numerical analyses ........................................................................................................................... 63

4.7 Summary and conclusions ................................................................................................................ 78

4.8 Site-response analysis for biaxial horizontal shaking ....................................................................... 89

SECTION 5 AN ASSESSMENT OF SOIL-STRUCTURE INTERACTION ANALYSIS

5.1 Overview .......................................................................................................................................... 93

5.5 Summary and conclusions .............................................................................................................. 129

SECTION 6 THE NEES CITY BLOCK CENTRIFUGE EXPERIMENTS: TEST 3 AND

6.1 Introduction .................................................................................................................................... 131

SECTION 7 EXPERIMENTAL RESULTS AND COMPARISON WITH NUMERICAL

7.1 Introduction .................................................................................................................................... 169

7.4 Examination of SSI and SSSI effects ............................................................................................. 182

7.5 Summary and conclusions .............................................................................................................. 217

SECTION 8 SUMMARY, CONCLUSIONS AND RECOMMENDATIONS................................... 221

8.1 Summary......................................................................................................................................... 221

8.3 Future research ............................................................................................................................... 228

APPENDIX B THE ORMSBY WAVELET ......................................................................................... 325

The earthquake engineering profession in the United States is embracing performance-based

1.2.1 Assessment of numerical methods in site-response and soil-structure interaction analysis

1.2.2 Seismic structure-soil-structure interaction in dense urban environments

This report is organized into nine section with contents as follows:

2.1 Dynamic behavior of soils during earthquakes

2.1.1 Experimental response of cyclically loaded soils

2.1.2 Numerical material models for dynamic soil behavior

2.1.2.1 Equivalent-linear models

damping ratio  is given by

2.1.2.2 Nonlinear hysteretic models

2.2 Site-response analysis

2.2.2 One-dimensional methods of site-response analysis

Figure 2-4: One-dimensional site-response analysis

2.2.2.1 Evolution of one-dimensional site-response analysis methods

2.2.2.2 Equivalent-linear methods

The equivalent-linear method involves the calculation of an approximate nonlinear response

2.2.2.3 Nonlinear methods

 (t )  Vs  y(t ) (2-6)

Figure 2-6: Ground motion input in time-domain one-dimensional site-response analysis

2.2.2.4 Validation of one-dimensional site-response analysis programs

One-dimensional site-response analysis programs are typically validated by comparing the

The equivalent-linear method was developed as a computationally simpler alternative to the

2.2.3 Multi-dimensional analyses

3.2 Numerical methods in soil-structure interaction analysis

3.2.1 Substructuring method

Figure 3-3: Steps in a substructuring analysis (Kramer and Stewart, 2004)

Figure 3-4: Substructure methods of SSI analysis (Ostadan, 2006a)

3.2.2 Direct method