Electronic Thesis and Dissertations

UCLA

Peer Reviewed

Title:
Dynamic Soil-Structure Interaction of Instrumented Buildings and Test Structures

Author:
Givens, Michael James

Acceptance Date:
2013

Series:
UCLA Electronic Theses and Dissertations

Degree:
D.Env., Civil Engineering 0300UCLA

Advisor:
Stewart, Jonathan P

Committee:
Brandenberg, Scott, Taciroglu, Ertugrul, Yin, An, Mylonakis, George

Permalink:
http://escholarship.org/uc/item/3rh1w1hr

Abstract:

Copyright Information:
All rights reserved unless otherwise indicated. Contact the author or original publisher for any
necessary permissions. eScholarship is not the copyright owner for deposited works. Learn more
at http://www.escholarship.org/help_copyright.html#reuse

eScholarship provides open access, scholarly publishing

services to the University of California and delivers a dynamic
research platform to scholars worldwide.
 UNIVERSITY OF CALIFORNIA
Los Angeles

Dynamic Soil-Structure Interaction of

Instrumented Buildings and Test Structures

A dissertation submitted in partial satisfaction of the requirements for the

degree of Doctor of Philosophy in
Civil Engineering

by

Michael James Givens

2013
i
 ABSTRACT OF THE DISSERTATION

Dynamic Soil-Structure Interaction of

Instrumented Buildings and Test Structures

by

Michael James Givens

Doctor of Philosophy in Civil Engineering

University of California, Los Angeles, 2013

Professor Jonathan P. Stewart, Chair

The effects of soil-structure interaction (SSI) are investigated through careful interpretation of

available data from instrumented buildings and recently performed forced vibration experiments

on instrumented buildings and test structures. Conventional engineering practice typically

ignores soil-structure interaction (SSI) during evaluation of the seismic demand on buildings

based on the perception that consideration of SSI will reduce demands on structures and ignoring

SSI effects will cause seismic demands to be conservatively biased. I show that it is not always

conservative to ignore SSI effects. Analysis of field performance data is undertaken to provide

deeper insights into SSI phenomena ranging from kinematic effects on foundation ground

ii
motions to mobilized foundation stiffness and damping across a wide range of frequencies and

loading levels. These data are interpreted to evaluate strengths and limitations of engineering

analysis procedures for SSI.

Foundation damping incorporates the combined effects of energy loss due to waves

propagating away from the vibrating foundation in translational and rotational modes (radiation

damping), as well as hysteretic action in the soil (material damping). Previous foundation

damping models were developed for rigid circular foundations on homogenous halfspace and

were often expressed using confusing or incomplete functions. Starting from first principles, we

derive fundamental expressions for foundation damping in which foundation impedance

components representing radiation damping and the soil hysteretic damping ratio appear as

variables, providing maximum flexibility to the analyst. We utilize these general expressions

with impedance solutions for rectangular-footprint foundations to: (1) compare predicted

foundation damping levels with those from field case histories and (2) develop new foundation

damping relationships for application in the building code (NEHRP Provisions).

Ground motions at the foundation levels of structures differ from those in the free-field as

a result of inertial and kinematic interaction effects. Inertial interaction effects tend to produce

narrow-banded ground motion modification near the fundamental period of the soil-structure

system, whereas kinematic effects are relatively broad-banded but most significant at high

frequencies. Kinematic interaction effects can be predicted using relatively costly finite element

analyses with incoherent input or simplified models. The simplified models are semi-empirical in

nature and derived from California data. These simplified models are the basis for seismic design

guidelines used in the western United States, such as ASCE-41 and NIST (2012). We compile

iii
some available data from building and ground instrumentation arrays in Japan for comparison to

these two sets of models. We demonstrate that the model predictions for the sites under

consideration are very similar to each other for modest foundation sizes (equivalent radii under

about 50 m). However, the data show that both approaches overestimate the transfer function

ordinates relative to those from a subset of the Japanese buildings having pile foundations. The

misfit occurs at frequencies higher than the first-mode resonant frequency and appears to result

from pile effects on kinematic interaction that are not accounted for in current models.

A complete model of a soil-foundation-structure system for use in response history

analysis requires modification of input motions relative to those in the free-field to account for

kinematic interaction effects, foundation springs and dashpots to represent foundation-soil

impedance, and a structural model. The recently completed NIST (2012) report developed

consistent guidelines for evaluation of kinematic interaction effects and foundation impedance

for realistic conditions. We implement those procedures in seismic response history analyses for

two instrumented buildings in California, one a 13-story concrete-moment frame building with

two levels of basement and the other a 10-story concrete shear wall core building without

embedment. We develop three-dimensional baseline models (MB) of the building and foundation

systems (including SSI components) that are calibrated to reproduce observed responses from

recorded earthquakes. SSI components considered in the MB model include horizontal and

vertical springs and dashpots that represent the horizontal translation and rotational impedance,

kinematic ground motion variations from embedment and base slab averaging, and ground

motion variations over the embedment depth of basements. We then remove selected

components of the MB models one at a time to evaluate their impact on engineering demand

iv
parameters (EDPs) such as inter-story drifts, story shear distributions, and floor accelerations.

We find that a bathtub model that retains all features of the MB approach except for depth-

variable motions provides for generally good above-ground superstructure responses, but biased

demand assessments in subterranean levels. Other common approaches using a fixed-based

representation can produce poor results.

To expand the data inventory of response histories to evaluate SSI, we performed forced

vibration testing of a well-instrumented steel and reinforced concrete structure that has

removable bracing. The testing was performed at three sites with varying soil conditions. I

describe testing at one of the sites located in Garner Valley, California. During testing at this site

an adjacent structure and local concrete slab were also instrumented in addition to the test

structure. I describe the testing setup, schemes, instrumentation, and data processing techniques.

The data are analyzed to evaluate the stiffness and damping associated with the foundation-soil

interaction, revealing linear-elastic behavior at low forcing levels characterized by smaller

stiffnesses and both lower and higher damping than is predicted by classical models. Nonlinear

behavior at stronger shaking levels includes pronounced reductions of stiffness and changes in

damping. Interestingly, kinematic interaction effects observed on an adjacent slab excited

principally by surface waves were of a similar character to expectations from analytical models

for body wave excitation from earthquakes. If verified, these results could lead to site- and

foundation-specific test methods for evaluating kinematic interaction effects.

v
The dissertation of Michael James Givens is approved:

__________________________________
Scott Brandenberg

__________________________________
Ertugrul Taciroglu

__________________________________
An Yin

__________________________________
George Mylonakis

__________________________________
Jonathan P. Stewart, Committee Chair

University of California, Los Angeles

2013

vi
 To My Loving Family
With them everything is better

vii
 TABLE OF CONTENTS

PAGE

ABSTRACT OF THE DISSERTATION ................................................................ ii

TABLE OF CONTENTS ..................................................................................... viii
FIGURES .............................................................................................................. xii
TABLES ............................................................................................................ xxvi
SYMBOLS......................................................................................................... xxix
ACKNOWLEDGMENTS ............................................................................... xxxiv
VITA .............................................................................................................. xxxvi
1 INTRODUCTION..................................................................................................1
1.1 SOIL-STRUCTURE INTERACTION IN ENGINEERING PRACTICE ........................1
1.2 ALTERNATIVE FRAMEWORKS FOR SSI ANALYSIS .........................................2
1.3 RESEARCH MOTIVATIONS AND ORGANIZATION ............................................6
2 FOUNDATION DAMPING ................................................................................12
2.1 INTRODUCTION ...........................................................................................12
2.2 APPLICATIONS OF FOUNDATION DAMPING .................................................13
2.2.1 Foundation Damping in Force-Based Methods ...................13
2.2.2 Foundation Damping in Displacement-Based
Methods ...............................................................................16
2.3 GENERAL SOLUTION FOR FOUNDATION DAMPING ......................................18
2.3.1 Impedance Solutions for Shallow Foundations ...................19
2.3.2 Considerations on the Use of Viscous Structural
Damping...............................................................................23
2.3.3 Foundation Damping Derivations ........................................25
2.3.3.1 Derivation from Imaginary Component ..................27
2.3.3.2 Derivation from Complex-Valued Impedance
Expressions ..............................................................31
2.3.3.3 Comparison of Alternate Solutions for Foundation
Damping ..................................................................34
2.4 COMPARISON TO FOUNDATION DAMPING SOLUTIONS IN
LITERATURE ...............................................................................................36
2.4.1 Bielak (1971) Solution .........................................................36
2.4.2 Veletsos and Nair (1975) Solution.......................................38
2.4.3 Roesset (1980) and Wolf (1985) Solutions..........................41
2.4.4 Maravas et al. (2007) Solution .............................................41
2.4.5 Relating Foundation Damping to Period
Lengthening .........................................................................44
2.5 SOLUTIONS FOR DAMPING OF RECTANGULAR FOUNDATIONS .....................46

viii
 2.6 EVALUATION OF FOUNDATION DAMPING SOLUTIONS.................................55
2.6.1 Idealized Solutions Adapted for Realistic
Conditions ............................................................................55
2.6.2 Comparison to Recorded Data .............................................57
2.7 CONCLUSIONS AND RECOMMENDATIONS....................................................67
3 KINEMATIC SOIL-STRUCTURE INTERACTION EFFECTS FROM
BUILDING AND FREE-FIELD SEISMIC ARRAYS IN JAPAN .................69
3.1 INTRODUCTION ...........................................................................................69
3.2 STRUCTURAL-GROUND ARRAYS IN JAPAN .................................................71
3.3 MODELS FOR PREDICTION OF FOUNDATION TO FREE-FIELD
TRANSFER FUNCTIONS ................................................................................78
3.3.1 Spatial Coherency Model Implemented in SASSI ..............78
3.3.2 Semi-Empirical Approach ...................................................79
3.4 EVALUATION AND COMPARISON OF TRANSFER FUNCTIONS .......................81
3.4.1 Transfer Function Calculation from Data ............................81
3.4.2 Application of Models .........................................................82
3.4.3 Results for Subject Buildings ..............................................82
3.5 INTERPRETATION AND CONCLUSIONS .........................................................86
4 ASSESSMENT OF ALTERNATIVE SSI MODELING STRATEGIES .......88
4.1 INTRODUCTION ...........................................................................................88
4.2 SIMILAR STUDIES IN THE LITERATURE ........................................................89
4.3 MODEL TYPES.............................................................................................91
4.3.1 Baseline Model (MB) ..........................................................91
4.3.2 Simplified and Alternative Models ......................................92
4.4 BUILDINGS CONSIDERED ............................................................................96
4.5 SHERMAN OAKS BUILDING .........................................................................98
4.5.1 Site Description....................................................................99
4.5.2 Ground Motion Recordings ...............................................102
4.5.3 Foundation Conditions .......................................................105
4.5.4 Development of Foundation Springs and Dashpots ..........107
4.6 WALNUT CREEK BUILDING.......................................................................116
4.6.1 Site Description..................................................................117
4.6.2 Ground Motion Recordings ...............................................119
4.6.3 Foundation Conditions .......................................................122
4.6.4 Development of Foundation Springs and Dashpots ..........124
4.7 BASELINE MODEL AND COMPARISON TO RECORDINGS.............................131
4.8 MODEL TO MODEL COMPARISON ..............................................................142
4.8.1 Sherman Oaks Analysis Results ........................................142
4.8.2 Walnut Creek Analysis Results .........................................151
4.9 INTERPRETATION AND CONCLUSIONS .......................................................154
5.0 FORCED VIBRATION TESTING OF A SHALLOW FOUNDATION
FIELD TEST STRUCTURE ............................................................................156

ix
 5.1 INTRODUCTION .........................................................................................156
5.2 TEST SETUP AND INSTRUMENTATION........................................................158
5.2.1 Garner Valley Downhole Array (GVDA) Test Site ..........158
5.2.2 Test Structures ...................................................................163
5.2.3 Sensors ...............................................................................166
5.2.4 Forced Vibration Shakers ..................................................168
5.2.5 Data Acquisition Setup ......................................................169
5.3 TRIALS AND DATA PROCESSING PROCEDURES ..........................................173
5.3.1 Field Test Logs ..................................................................173
5.3.2 Data Processing..................................................................181
5.3.2.1 Structural Response Data Processing ....................181
5.3.2.2 Mighty Mouse Shaker Data Processing ................183
5.3.3 Cross-Correlation of the Data Acquisition
Systems ..............................................................................186
5.3.4 Unprocessed Data Files .....................................................191
5.3.4.1 Unprocessed Structural Response Data Files ........191
5.3.4.2 Unprocessed Mighty Mouse Shaker Data Files ....192
5.3.5 Converted Data Files .........................................................193
5.3.5.1 Converted Structural Response Data Files ............194
5.3.5.2 Converted Mighty Mouse Shaker Data Files ........195
5.3.6 Corrected Data Files ..........................................................195
5.3.6.1 Corrected Structural Response Data Files .............195
5.3.6.2 Corrected Mighty Mouse Shaker Data Files .........196
5.3.6.3 Corrected Synchronized Structural Response and
Mighty Mouse Shaker Data Files ..........................197
5.4 STRUCTURAL RESPONSE OF TEST STRUCTURE ..........................................199
5.5 CONCLUSIONS ...........................................................................................221
6 EVALUATION OF SOIL-STRUCTURE INTERACTION EFFECTS
FROM FORCED VIBRATION TESTING.....................................................223
6.1 INTRODUCTION .........................................................................................223
6.2 PREVIOUS FORCED VIBRATION TESTING OF FIELD TEST
STRUCTURES TO EVALUATE IMPEDANCE FUNCTIONS ...............................224
6.3 INVERSION PROCEDURES FOR EVALUATION OF IMPEDANCE
ORDINATES FROM FORCED VIBRATION TEST DATA..................................237
6.4 FORCED VIBRATION AND INERTIAL RESPONSES OF TEST
STRUCTURE AT GVDA .............................................................................241
6.5 IMPEDANCE FUNCTIONS FROM FORCED VIBRATION TESTING ...................247
6.5.1 Small Amplitude Forced Vibration Testing.......................247
6.5.2 Large Amplitude Forced Vibration Testing.......................257
6.6 PARAMETRIC SYSTEM IDENTIFICATION OF FORCED VIBRATION
TESTING ....................................................................................................272
6.6.1 System Identification Method ............................................272

x
 6.6.2 System Identification of Instrumented Field Test
Structures ...........................................................................274
6.7 KINEMATIC INTERACTION ANALYSIS FROM FORCED VIBRATION
TESTING AT GVDA ..................................................................................280
6.8 STRUCTURE-SOIL-STRUCTURE INTERACTION FROM FORCED
VIBRATION TESTING .................................................................................291
6.9 INTERPRETATION OF SOIL-STRUCTURE INTERACTION EFFECTS
FROM FORCED VIBRATION TESTING .........................................................294
6.9.1 Small Amplitude Forced Vibration Testing.......................294
6.9.2 Large Amplitude Forced Vibration Testing.......................297
6.9.3 Small Amplitude to Large Amplitude
Comparisons ......................................................................301
6.9.4 GVDA Results and Previous Studies.................................304
6.9.5 Kinematic Interaction and Structure-Soil-Structure
Interaction Interpretation ...................................................308
6.10 CONCLUSIONS ...........................................................................................309
7 CONCLUSIONS AND RECOMMENDATIONS ...........................................311
7.1 SCOPE OF RESEARCH ................................................................................311
7.2 RESEARCH FINDINGS AND RECOMMENDATIONS .......................................314
7.2.1 Chapter 2; Foundation Damping........................................314
7.2.2 Chapter 3; Kinematic Soil-Structure Interaction
Effects from Building and Free-Field Seismic
Arrays in Japan ..................................................................315
7.2.3 Chapter 4; Assessment of Alternative SSI
Modeling Strategies ...........................................................316
7.2.4 Chapter 5; Forced Vibration Testing of a Test
Structure .............................................................................317
7.2.5 Chapter 6; Evaluation of Soil-Structure Interaction
Effects from Forced Vibration Testing ..............................318
7.3 RECOMMENDATIONS FOR FUTURE RESEARCH ..........................................321
8 REFERENCES...................................................................................................323

xi
 FIGURES

PAGE

Figure 1-1 Schematic illustration of direct approach to analysis of the soil-

structure interaction (SSI) problem using continuum modeling (NIST,
2012). .......................................................................................................... 3
Figure 1-2 Schematic illustration of substructure approach to analysis of soil-
foundation-structure interaction (SFSI) problem using either (i) rigid
foundation or (ii) flexible foundation assumption (NIST, 2012). ............... 5
Figure 2-1 Illustration of effects of period lengthening and change of damping on
fundamental mode spectral acceleration from which base shear is
evaluated. (NIST, 2012) ............................................................................ 14
Figure 2-2 Schematic illustration of deflections caused by force applied to (a)
fixed-base structure and (b) structure with horizontal and rotational
compliance. (NIST, 2012)......................................................................... 15
Figure 2-3 Schematic illustration of analysis to develop pushover curve (NIST,
2012) ......................................................................................................... 17
Figure 2-4 Schematic illustration of SFSI effects on displacement-based method
of evaluating nonlinear structural performance. ....................................... 18
Figure 2-5 Sketch showing the geometry adapted for the rectangular foundation
(L B ) ...................................................................................................... 21
Figure 2-6 Shows the comparison of the dynamic stiffness modifiers and
radiation damping ratios as determined by the numerical data
(Veletsos and Wei 1971) and the approximate method (Veletsos and
Verbic 1973). Conditions for plots are a rigid, massless, circular disc
supported on an elastic homogeneous isotropic halfspace with
hysteretic soil damping s = 0. = 0.45 for a nearly saturated clay and
= 0.33 for sand. ...................................................................................... 22
Figure 2-7 Comparison of foundation damping solutions based on the two
approaches from Section 2.3.3, plotted against structure-to-soil
stiffness ratio ( h (VsT ) ). Conditions for the plots are a rigid, massless,
circular disc supported on an homogeneous isotropic halfspace with
hysteretic soil damping a) s = 0% and b) s = 10%. = 0.33.
Impedance functions from Veletsos and Wei (1971) used to derive the
foundation damping. ................................................................................. 36
Figure 2-8 Comparison of foundation damping models (Eq. 2-35 and Eq. 2-38)
accounting for soil damping (s = 10%) differently, plotted against
structure-to-soil stiffness ratio ( h (VsT ) ). Conditions for the plot are a

xii
 rigid, massless, circular disc supported on an elastic homogeneous
isotropic halfspace with hysteretic soil damping s = 10 %. = 0.33.
Impedance functions from Veletsos and Verbic (1973) were used to
derive the foundation damping. Eq. 2-35 used Veletsos and Verbic
(1973) elastic impedance solution with additive soil damping,
whereas Eq. 2-38 uses an alternative impedance solution in which soil
damping for a visco-elastic medium is incorporated into the solution. .... 40
Figure 2-9 Comparison of the foundation damping models for the two proposed
formulations and the exact solution of Maravas et al. (2007). All
results shown in the figure are based on the Veletsos and Wei (1971)
impedance model for a rigid, massless, circular disc supported on an
elastic homogeneous isotropic halfspace with = 0.33 and a) s = 0%
and b) s = 10%......................................................................................... 44
Figure 2-10 Foundation damping as a function of period lengthening from a)
ASCE 7-05, 2006 and FEMA 450, 2003 and b) ASCE 7-10, 2010 and
FEMA P-750, 2009. Foundation damping computed using Veletsos
and Verbic (1973) impedance functions for rigid circular foundation.
= 0.33 and D = 0 for plots. ..................................................................... 46
Figure 2-11 Foundation damping as a function of h (Vs T ) using Eqs. 2-25 and 2-
35. Top plots show contributions from radiation damping only (s =
0) for circular and square foundations. Bottom plots include soil
hysteretic damping (s = 0.1). = 0.33 and D = 0 for plots. Veletsos
and Wei (1971) impedance functions were used for the circular case.
The h/B input for the Pais and Kausel (1988) impedance functions for
the square case was modified to represent equivalent areas per Eq. (2-
47) and are denoted by the equivalent h/r input. ...................................... 51
Figure 2-12 Foundation damping as a function of period lengthening using Eqs.
(2-25) and (2-35). Foundation damping computed for square
foundations using a) only radiation damping terms (s = 0) and b) soil
hysteretic damping (s = 0.1). L/B = 1, = 0.33 and D = 0 used for
plots. .......................................................................................................... 52
Figure 2-13 Foundation damping as a function of foundation aspect ratios (L/B)
and period lengthening using Eqs. (2-25) and (2-35). Top plots show
foundation damping for h/B= 1; bottom plots are for h/B = 2. s = 0,
= 0.33 and D = 0 for plots. ........................................................................ 54
Figure 2-14 Histograms of h/(VsT) and h/B in data sets used to evaluate SSI
effects from instrumented structures. Top row corresponds to
instrumented structures shaken by earthquakes and analyzed by
Stewart et al. (1999b); bottom row corresponds to forced vibration
tests results from this study and Tileylioglu et al., (2011). ....................... 59

xiii
Figure 2-15 Period lengthening and foundation damping residuals from combined
data sets. Foundation damping residuals from model in Eq. (2-25). ........ 66
Figure 2-16 Period lengthening and foundation damping residuals from combined
data sets. Foundation damping residuals from model in Eq. (2-35). ........ 67
Figure 3-1. Building layout and site conditions at Sendai University site (J1).
Data source: OYO Corporation, 2007. ..................................................... 75
Figure 3-2. Building layout and site conditions at Onagawa Nuclear Power Plant,
Reactor 1. Data sources: Tohoku Electric Power Company, 2009 and
2011........................................................................................................... 76
Figure 3-3. Building layout and site conditions at Nagoya Office Building (J4).
Data sources: CRBOMLIT, 2003 and 2005. ............................................ 77
Figure 3-4 Sendai site horizontal transfer functions and coherence for the 2003
Off Miyagi earthquake. ............................................................................. 84
Figure 3-5 Sendai site horizontal transfer functions and coherence for the 2011
Tohoku earthquake.................................................................................... 84
Figure 3-6 Onagawa site horizontal transfer functions and coherence for the 2011
Tohoku earthquake.................................................................................... 85
Figure 3-7 Onagawa site rocking transfer functions and coherence for the 2011
Tohoku earthquake.................................................................................... 85
Figure 3-8 Nagoya site horizontal transfer functions and coherence for the 2004
Off Kii Peninsula earthquake. ................................................................... 86
Figure 4-1 Schematic illustration of simulation used in Naeim et al. (2008) and
Tileylioglu et al. (2010). ........................................................................... 90
Figure 4-2 Schematic illustration of baseline model considered in simulations. ....... 92
Figure 4-3 Schematic illustration of simplified models 1 to 4 considered in
simulations. ............................................................................................... 94
Figure 4-4 Photographs of the example applications: 13-story Sherman Oaks
building (left) and the 10-story Walnut Creek (right). Photographs
obtained from CESMD. ............................................................................ 97
Figure 4-5 Plan view of Sherman Oaks building site showing locations of
borehole and geophysical logs used for site characterization ................. 100
Figure 4-6 Subsurface characteristics based on adjacent site exploration at sites
south and northwest (NW) of the local intersection of the Sherman
Oaks building: a) shear wave velocity profile and b) material profile.
1 ft = 0.3048 m ........................................................................................ 101
Figure 4-7 CSMIP Station No. 24322: Sherman Oaks 13-Story commercial
building, cross-sectional view and sensor location sketch...................... 102

xiv
Figure 4-8 Response spectra of uFIM (recorded) and ug (inferred) for Sherman
Oaks building, 1994 Northridge earthquake ........................................... 104
Figure 4-9 Calculated variation of peak acceleration and velocity over the
embedment depth for Sherman Oaks building, 1994 Northridge
earthquake ............................................................................................... 105
Figure 4-10 Sherman Oaks building foundation plan, based on construction
drawings. 1 ft = 0.3048 m ....................................................................... 107
Figure 4-11 Vertical spring and dashpot intensities and areas distinguished by
color. Solid lines represent tributary area boundaries in which
stiffness and dashpot values were distributed to the column node of
each area. Solid circles represent column nodes. 1 ft = 0.3048 m .......... 112
Figure 4-12 Distribution of nodes on foundation and calculation of individual
spring stiffnesses ..................................................................................... 115
Figure 4-13 Site plan of Walnut Creek office building showing locations of
borings and refraction lines ..................................................................... 118
Figure 4-14 Subsurface characteristics based site and adjacent site explorations of
the Walnut Creek building: a) shear wave velocity profile and b)
material profile ........................................................................................ 119
Figure 4-15 CSMIP Station No. 58364: Walnut Creek 10-Story commercial
building, sensor location sketch .............................................................. 120
Figure 4-16 Response spectra of the uFIM and ug for Walnut Creek building, 1989
Loma Prieta earthquake .......................................................................... 122
Figure 4-17 Overall north-south elevation view of the Walnut Creek building
(from Stewart and Stewart, 1997). .......................................................... 123
Figure 4-18 Walnut Creek building foundation plan (from construction drawings) .. 124
Figure 4-19 Vertical spring and dashpot intensities and areas distinguished by
color. Solid lines represent tributary area boundaries in which
stiffness and dashpot values were distributed to each area. 1 ft =
0.3048 m ................................................................................................. 130
Figure 4-20 Schematic illustration of modeling approaches for modeling Walnut
Creek building. ........................................................................................ 133
Figure 4-21 Comparison of recorded displacement histories at roof, 8th floor,
ground level and base level with those predicted using the Baseline
Model (MB). Northridge earthquake. ..................................................... 135
Figure 4-22 Comparison of recorded peak displacements over the height (X and Y
direction) with those predicted using the baseline model (MB).
Northridge earthquake. ........................................................................... 136

xv
Figure 4-23 Comparison of recorded lateral acceleration and relative displacement
histories at roof, 8th floor, 3rd floor and ground level with those
predicted using the Baseline Model (MB). Loma Prieta earthquake. ..... 138
Figure 4-24 Comparison of maximum floor relative displacement and absolute
acceleration between model and recorded response. Loma Prieta
earthquake. .............................................................................................. 139
Figure 4-25 High-pass filtering of Walnut Creek building vertical motions at the
base level. Figure shows the effect of the corner frequency on
displacement histories and rocking history for the Walnut Creek
building. .................................................................................................. 140
Figure 4-26 Comparison of the foundation rotation from base rocking of the
baseline model and recorded data for the Walnut Creek building. ......... 141
Figure 4-27 Comparison of the baseline model roof displacement from base
rocking compared to the total roof displacement from the recorded
data for the Walnut Creek building. ........................................................ 141
Figure 4-28 Comparison of displacement histories, drift ratios, and story shears
obtained from MB and Model 4 (bathtub) for the Sherman Oaks
building. .................................................................................................. 144
Figure 4-29 Comparison of displacement histories, drift ratios, and story shears
obtained from MB and Model 1 (fixed at ground line) for the
Sherman Oaks building. .......................................................................... 145
Figure 4-30 Comparison of displacement histories, drift ratios, and story shears
obtained from MB and Model 2 (fixed at base) for the Sherman Oaks
building. .................................................................................................. 147
Figure 4-31 Comparison of displacement histories, drift ratios, and story shears
obtained from MB and Model 3 (fixed horizontal springs) for the
Sherman Oaks building. .......................................................................... 148
Figure 4-32 Comparison of displacement histories, drift ratios, and story shears for
MB, MB.1 (rigid foundations), and MB.2 (gaps) for the Sherman
Oaks building. ......................................................................................... 149
Figure 4-33 Comparison of peak displacement over height, drift ratios, and story
shears in transverse direction from all models of the Sherman Oaks
building. .................................................................................................. 150
Figure 4-34 Comparison of the roof acceleration and displacement histories for the
reference model (MB, equivalent to Model 4) and Models 1 and 2 of
the Walnut Creek building. ..................................................................... 152
Figure 4-35 Comparison of profiles of maximum displacement, maximum
interstory drift, and peak floor acceleration for the reference model

xvi
 (MB, equivalent to Model 4) and Models 1 and 2 of the Walnut Creek
building. .................................................................................................. 153
Figure 5-1 Plan view of the GVDA site showing the approximate locations of the
SASW test locations. .............................................................................. 160
Figure 5-2 Subsurface characteristics based on site exploration of the GVDA
site: a) shear wave velocity profile and b) material profile. ................... 160
Figure 5-3 Soil shear wave velocity profiles based on crosshole seismic surveys
performed on May 26, 2011.................................................................... 161
Figure 5-4 Construction plan cross-section of the test structure Mini-Me. .............. 164
Figure 5-5 Sketch showing the geometry adapted for the rectangular foundation
(L B ) . .................................................................................................... 165
Figure 5-6 Photograph showing the test structure Mini-Me (right) and the
previously constructed structure Dr. Evil (left) located at GVDA ......... 165
Figure 5-7 Sketch showing the structures location and sensor arrays for testing at
GVDA. .................................................................................................... 167
Figure 5-8 Flowchart describing the workflow and associated data files for the
structural response data files. The graphical example is for experiment
3.4.19a as described in Table 5-6. The solid boxes in the flowchart
describe the data file types with file name descriptions in brackets,
while the dashed boxes describe the effort taken to achieve the
underlying processed data file. ............................................................... 183
Figure 5-9 Flowchart describing the workflow and associated data files for the
Mighty Mouse data files and corresponding structural response data
files (described in Figure 8). The solid boxes in the flowchart describe
the data file types with file name descriptions in brackets, while the
dashed boxes describe the effort taken to achieve the underlying
processed data file. .................................................................................. 185
Figure 5-10 Experiment 3.4.1a cross-correlations analysis components: (a)
structural response data channel 15 accelerometer, (b) parallel Mighty
Mouse shaker data channel 2, (c) cross-correlation graphical result,
and (d) short time window showing required time-shift for
correlation. .............................................................................................. 188
Figure 5-11 Experiment 3.4.1a results following application of time-lag: (a)
structural response data channel 15 accelerometer, (b) parallel Mighty
Mouse shaker data channel 2 adjusted by t=0.8907, and (c) short
time window showing correlation of adjusted Mighty Mouse shaker
data to structural response data. .............................................................. 189

xvii
Figure 5-12 Experiment 3.4.19a cross-correlations analysis components: (a)
structural response data channel 15 accelerometer, (b) parallel Mighty
Mouse shaker data channel 2, (c) cross-correlation graphical result,
and (d) short time window showing required time-shift for
correlation. .............................................................................................. 190
Figure 5-13 Experiment 3.4.19a results following application of time-lag: (a)
structural response data channel 15 accelerometer, (b) parallel Mighty
Mouse shaker data channel 2 adjusted by t=0.8907, and (c) short
time window showing correlation of adjusted Mighty Mouse shaker
data to structural response data. .............................................................. 191
Figure 5-14 Shaker force loading histories that represent the small, moderate and
large forces used during forced vibration testing at GVDA. The titles
of the shaker forces correspond to Table 5-7. ......................................... 205
Figure 5-15a The horizontal structural responses of the foundation are shown for
Exp. 3.3.5a (Unbraced condition, Atom Ant shaker loading on the
roof in the y-direction). The local axes along with channel
identification are shown in the middle, while the structural responses
are shown in the peripheral. .................................................................... 206
Figure 5-15b The horizontal structural responses of the roof are shown for Exp.
3.3.5a (Unbraced condition, Atom Ant shaker loading on the roof in
the y-direction). The local axes along with channel identification are
shown in the middle, while the structural responses are shown in the
peripheral. ............................................................................................... 207
Figure 5-15c The vertical structural responses are shown for Exp. 3.3.5a (Unbraced
condition, Atom Ant shaker loading on the roof in the y-direction).
The local axes along with channel identification are shown in the
middle, while the structural responses are shown on at the sides. .......... 208
Figure 5-15e The horizontal structural responses of the adjacent structure (Dr. Evil)
are shown in comparison to the free-field and Mini-Me responses for
Exp. 3.3.5a (Unbraced condition, Atom Ant shaker loading on the
roof in the y-direction). The local axes of Mini-Me is shown along
with channel identifications of the responses. Dr. Evil foundation plan
view is shown at the top left and the cross-section relationship of the
structures and free-field is shown at the top right. .................................. 210
Figure 5-16a The horizontal structural responses of the foundation are shown for
Exp. 3.4.1a (Unbraced condition, Mighty Mouse shaker with 5%
eccentricity loading on the roof in the y-direction). The local axes
along with channel identification are shown in the middle, while the
structural responses are shown in the peripheral. ................................... 211

xviii
Figure 5-16b The horizontal structural responses of the roof are shown for Exp.
3.4.1a (Unbraced condition, Mighty Mouse shaker with 5%
eccentricity loading on the roof in the y-direction). The local axes
along with channel identification are shown in the middle, while the
structural responses are shown in the peripheral. ................................... 212
Figure 5-16c The vertical structural responses are shown for Exp. 3.4.1a (Unbraced
condition, Mighty Mouse shaker with 5% eccentricity loading on the
roof in the y-direction). The local axes along with channel
identification are shown in the middle, while the structural responses
are shown on at the sides. ....................................................................... 213
Figure 5-16e The horizontal structural responses of the adjacent structure (Dr. Evil)
are shown in comparison to the free-field and Mini-Me responses for
Exp. 3.4.1a (Unbraced condition, Mighty Mouse shaker with 5%
eccentricity loading on the roof in the y-direction). The local axes of
Mini-Me is shown along with channel identifications of the responses.
Dr. Evil foundation plan view is shown at the top left and the cross-
section relationship of the structures and free-field is shown at the top
right. ........................................................................................................ 215
Figure 5-17a The horizontal structural responses of the foundation are shown for
Exp. 3.4.19a (Unbraced condition, Mighty Mouse shaker with 100%
eccentricity loading on the roof in the y-direction). The local axes
along with channel identification are shown in the middle, while the
structural responses are shown in the peripheral. ................................... 216
Figure 5-17b The horizontal structural responses of the roof are shown for Exp.
3.4.19a (Unbraced condition, Mighty Mouse shaker with 100%
eccentricity loading on the roof in the y-direction)The local axes
along with channel identification are shown in the middle, while the
structural responses are shown in the peripheral. ................................... 217
Figure 5-17c The vertical structural responses are shown for Exp. 3.4.19a
(Unbraced condition, Mighty Mouse shaker with 100% eccentricity
loading on the roof in the y-direction). The local axes along with
channel identification are shown in the middle, while the structural
responses are shown on at the sides. ....................................................... 218
Figure 5-17e The horizontal structural responses of the adjacent structure (Dr. Evil)
are shown in comparison to the free-field and Mini-Me responses for
Exp. 3.4.19a (Unbraced condition, Mighty Mouse shaker with 100%
eccentricity loading on the roof in the y-direction). The local axes of
Mini-Me is shown along with channel identifications of the responses.
Dr. Evil foundation plan view is shown at the top left and the cross-
section relationship of the structures and free-field is shown at the top
right. ........................................................................................................ 220

xix
Figure 6-1 Impedance functions derived from forced vibration testing of the
Millikan Library Building (Wong et al., 1988). The figure shows the
impedance functions in the north-south direction for a) rocking and b)
horizontal and in the east-west for c) rocking and d) horizontal. ........... 229
Figure 6-2 Impedance functions for Cholame station in the horizontal (a-b) and
rocking (c-d) modes of vibration. (Crouse et al., 1990).......................... 230
Figure 6-3 Impedance functions for Station 6 in the horizontal (a-b) and rocking
(c-d) modes of vibration. (Crouse et al., 1990). ...................................... 230
Figure 6-4 Normalized impedance functions for Hualien containment model in
the horizontal (a-b) and rocking (c-d) modes of vibration. (de Barros
and Luco, 1995). ..................................................................................... 233
Figure 6-5 Normalized impedance functions for testing of Dr. Evil as GVDA for
horizontal (top) and rocking (bottom) modes of vibration. (Tileylioglu
et al., 2011). ............................................................................................ 234
Figure 6-6 Impedance functions for testing of Mini-Me at WLA for horizontal
(top) and rocking (bottom) modes of vibration using small shaker.
Exp 2.1.7 (Unbraced condition, Atom Ant shaker loading on the roof
in the x-direction). (Star, 2011)............................................................... 235
Figure 6-7 Impedance functions for testing of Mini-Me at WLA for horizontal
(top), rocking (middle) modes of vibration using a large shaker. The
bottom graphic shows a representation of the soil nonlinearity during
testing. Exp. 2.2.224b (Unbraced condition, Mighty Mouse shaker
(100% ecc.) loading on the roof in the y-direction). (Star, 2011). .......... 236
Figure 6-8 Sketch showing the soil-structure interaction system used for
determination of impedance functions from forced vibration testing
based on Tileylioglu et al (2011). ........................................................... 237
Figure 6-9 Frequency domain presentation of the a) responses and b) force
components of Exp. 3.3.5a (Unbraced condition, Atom Ant shaker
loading on the roof in the y-direction). Refer to Figure 6-8 for
definitions. .............................................................................................. 244
Figure 6-10 Response correlation of the shear versus translation and moment
versus rotation of Exp. 3.3.5a (Unbraced condition, Atom Ant shaker
loading on the roof in the y-direction). ................................................... 244
Figure 6-11 Frequency domain presentation of the a) responses and b) force
components of Exp. 3.4.1a (Unbraced condition, Mighty Mouse
shaker (5% ecc.) loading on the roof in the y-direction). Refer to
Figure 6-8 for definitions. ....................................................................... 245

xx
Figure 6-12 Response correlation of the shear versus translation and moment
versus rotation of Exp. 3.4.1a (Unbraced condition, Mighty Mouse
shaker (5% ecc.) loading on the roof in the y-direction). ....................... 245
Figure 6-13 Frequency domain presentation of the a) responses and b) force
components of Exp. 3.4.19a (Unbraced condition, Mighty Mouse
shaker (100% ecc.) loading on the roof in the y-direction). Refer to
Figure 6-8 for definitions. ....................................................................... 246
Figure 6-14 Response correlation of the shear versus translation and moment
versus rotation of Exp. 3.4.19a (Unbraced condition, Mighty Mouse
shaker (100% ecc.) loading on the roof in the y-direction). ................... 246
Figure 6-15 Photographs showing the Atom Ant shaker on the left, secured in-
place with rope during forcing tests at the roof slab and on the right
free-standing during forcing test at the foundation slab. ........................ 248
Figure 6-16 Inverted impedance function results from analysis of Exp. 3.3.5
(Unbraced condition, Atom Ant shaker loading on the roof in the y-
direction). The plots show a) the translation stiffness, b) rocking
stiffness, c) translation damping ratio, and d) rocking damping ratio of
the SSI system in the direction of loading. ............................................. 249
Figure 6-17 Inverted impedance function results from analysis of Exp. 3.3.1
(Unbraced condition, Atom Ant shaker loading on the roof in the x-
direction). The plots show a) the translation stiffness, b) rocking
stiffness, c) translation damping ratio, and d) rocking damping ratio of
the SSI system in the direction of loading. ............................................. 250
Figure 6-18 Inverted impedance function results from analysis of Exp. 3.1.6
(Braced condition, Atom Ant shaker loading on the roof in the y-
direction). The plots show a) the translation stiffness, b) rocking
stiffness, c) translation damping ratio, and d) rocking damping ratio of
the SSI system in the direction of loading. ............................................. 251
Figure 6-19 Inverted impedance function results from analysis of Exp. 3.1.1
(Braced condition, Atom Ant shaker loading on the roof in the x-
direction). The plots show a) the translation stiffness, b) rocking
stiffness, c) translation damping ratio, and d) rocking damping ratio of
the SSI system in the direction of loading. ............................................. 252
Figure 6-20 Inverted impedance function results from analysis of Exp. 3.3.9
(Unbraced condition, Atom Ant shaker loading on the foundation in
the y-direction). The plots show a) the translation stiffness, b) rocking
stiffness, c) translation damping ratio, and d) rocking damping ratio of
the SSI system in the direction of loading. Note: the shaker was not
externally mounted to the structure......................................................... 253

xxi
Figure 6-21 Inverted impedance function results from analysis of Exp. 3.3.7
(Unbraced condition, Atom Ant shaker loading on the foundation in
the x-direction). The plots show a) the translation stiffness, b) rocking
stiffness, c) translation damping ratio, and d) rocking damping ratio of
the SSI system in the direction of loading. Note: the shaker was not
externally mounted to the structure......................................................... 254
Figure 6-22 Inverted impedance function results from analysis of Exp. 3.1.8
(Braced condition, Atom Ant shaker loading on the foundation in the
x-direction). The plots show a) the translation stiffness, b) rocking
stiffness, c) translation damping ratio, and d) rocking damping ratio of
the SSI system in the direction of loading. Note: the shaker was not
externally mounted to the structure......................................................... 255
Figure 6-23 Inverted impedance function results from analysis of Exp. 3.1.11
(Braced condition, Atom Ant shaker loading on the foundation in the
y-direction). The plots show a) the translation stiffness, b) rocking
stiffness, c) translation damping ratio, and d) rocking damping ratio of
the SSI system in the direction of loading. Note: the shaker was not
externally mounted to the structure......................................................... 256
Figure 6-24 Inverted impedance function results from analysis of Exp. 3.4.1a
(Unbraced condition, Mighty Mouse shaker (5% ecc.) loading on the
roof in the y-direction). The plots show impedance functions in the a)
translational and b) rocking modes of vibration. The bottom plots
represent the soil nonlinearity for each mode. ........................................ 259
Figure 6-25 Inverted impedance function results from analysis of Exp. 3.4.3a
(Unbraced condition, Mighty Mouse shaker (10% ecc.) loading on the
roof in the y-direction). The plots show impedance functions in the a)
translational and b) rocking modes of vibration. The bottom plots
represent the soil nonlinearity for each mode. ........................................ 260
Figure 6-26 Inverted impedance function results from analysis of Exp. 3.4.5a
(Unbraced condition, Mighty Mouse shaker (25% ecc.) loading on the
roof in the y-direction). The plots show impedance functions in the a)
translational and b) rocking modes of vibration. The bottom plots
represent the soil nonlinearity for each mode. ........................................ 261
Figure 6-27 Inverted impedance function results from analysis of Exp. 3.4.17a
(Unbraced condition, Mighty Mouse shaker (50% ecc.) loading on the
roof in the y-direction). The plots show impedance functions in the a)
translational and b) rocking modes of vibration. The bottom plots
represent the soil nonlinearity for each mode. ........................................ 262
Figure 6-28 Inverted impedance function results from analysis of Exp. 3.4.19a
(Unbraced condition, Mighty Mouse shaker (100% ecc.) loading on
the roof in the y-direction). The plots show impedance functions in

xxii
 the a) translational and b) rocking modes of vibration. The bottom
plots represent the soil nonlinearity for each mode. ............................... 263
Figure 6-29 Inverted impedance function results from analysis of Exp. 3.4.7a
(Unbraced condition, Mighty Mouse shaker (5% ecc.) loading on the
roof in the x-direction). The plots show impedance functions in the a)
translational and b) rocking modes of vibration. The bottom plots
represent the soil nonlinearity for each mode. ........................................ 264
Figure 6-30 Inverted impedance function results from analysis of Exp. 3.4.9b
(Unbraced condition, Mighty Mouse shaker (10% ecc.) loading on the
roof in the x-direction). The plots show impedance functions in the a)
translational and b) rocking modes of vibration. The bottom plots
represent the soil nonlinearity for each mode. ........................................ 265
Figure 6-31 Inverted impedance function results from analysis of Exp. 3.4.13a
(Unbraced condition, Mighty Mouse shaker (50% ecc.) loading on the
roof in the x-direction). The plots show impedance functions in the a)
translational and b) rocking modes of vibration. The bottom plots
represent the soil nonlinearity for each mode. ........................................ 266
Figure 6-32 Inverted impedance function results from analysis of Exp. 3.2.7a
(Braced condition, Mighty Mouse shaker (5% ecc.) loading on the
roof in the x-direction). The plots show impedance functions in the a)
translational and b) rocking modes of vibration. The bottom plots
represent the soil nonlinearity for each mode. ........................................ 267
Figure 6-33 Inverted impedance function results from analysis of Exp. 3.2.6a
(Braced condition, Mighty Mouse shaker (10% ecc.) loading on the
roof in the x-direction). The plots show impedance functions in the a)
translational and b) rocking modes of vibration. The bottom plots
represent the soil nonlinearity for each mode. ........................................ 268
Figure 6-34 Inverted impedance function results from analysis of Exp. 3.2.4a
(Braced condition, Mighty Mouse shaker (25% ecc.) loading on the
roof in the x-direction). The plots show impedance functions in the a)
translational and b) rocking modes of vibration. The bottom plots
represent the soil nonlinearity for each mode. ........................................ 269
Figure 6-35 Inverted impedance function results from analysis of Exp. 3.2.1a
(Braced condition, Mighty Mouse shaker (10% ecc.) loading on the
roof in the y-direction). The plots show impedance functions in the a)
translational and b) rocking modes of vibration. The bottom plots
represent the soil nonlinearity for each mode. ........................................ 270
Figure 6-36 Inverted impedance function results from analysis of Exp. 3.2.2a
(Braced condition, Mighty Mouse shaker (25% ecc.) loading on the
roof in the y-direction). The plots show impedance functions in the a)

xxiii
 translational and b) rocking modes of vibration. The bottom plots
represent the soil nonlinearity for each mode. ........................................ 271
Figure 6-37 Transmissibility functions from parametric and non-parametric
system identification using response history data from forced
vibration testing at GVDA with Atom Ant shaker. a) Exp. 3.3.1a
(Unbraced condition, loading on the roof in the x-direction) and b)
Exp. 3.3.5a (Unbraced condition, loading on the roof in the y-
direction). f1= first-mode frequency and 1= corresponding first-mode
damping................................................................................................... 278
Figure 6-38 Transmissibility functions from parametric and non-parametric
system identification using response history data from forced
vibration testing at GVDA with Atom Ant shaker. a) Exp. 3.1.1a
(Braced condition, loading on the roof in the x-direction) and b) Exp.
3.1.6a (Braced condition, loading on the roof in the y-direction). f1=
first-mode frequency and 1= corresponding first-mode damping. ........ 279
Figure 6-39 Sketch showing the kinematic interaction test setup with an
instrumented concrete slab near the location of the forced vibration
testing of Mini-Me. The sensor IDs are shown along with their
distance from the center of the Mini-Me test structure. .......................... 284
Figure 6-40a 3D depiction of particle motions at each accelerometer (A-E) at 5 Hz
frequency excitation as shown in the bottom right corner (Mighty
Mouse shaker frequency). ....................................................................... 285
Figure 6-40b 3D depiction of particle motion at each accelerometer (A-E) at 10 Hz
frequency excitation 10 Hz as shown in the bottom right corner
(Mighty Mouse shaker frequency). Based on Exp. 3.5.1 data. ............... 286
Figure 6-41 Experimental vertical and horizontal transfer functions (Hu) shown in
blue and the predicted horizontal Hu shown in red. Seismic source
was from the harmonic forced vibration testing of Mini-Me structure
with shaking from the Mighty Mouse shaker at 100% ecc. in the y-
direction. Based on Exp. 3.5.1 data. ....................................................... 287
Figure 6-42 Experimental vertical and horizontal transfer functions (Hu) shown in
blue and the predicted horizontal Hu shown in red. Seismic source
was from the harmonic forced vibration testing of Mini-Me structure
with shaking from the Mighty Mouse shaker at 100% ecc. in the y-
direction. Based on Exp. 3.5.2 data. ....................................................... 288
Figure 6-43 Experimental vertical and horizontal transfer functions (Hu) shown in
blue and the predicted horizontal Hu shown in red. Seismic source
was from the harmonic forced vibration testing of Mini-Me structure
with shaking from the Mighty Mouse shaker at 100% ecc. in the y-
direction. Based on Exp. 3.5.3 data. ....................................................... 289

xxiv
Figure 6-44 Experimental vertical and horizontal transfer functions (Hu) shown in
blue and the predicted horizontal Hu shown in red. Seismic source
was from the harmonic forced vibration testing of Mini-Me structure
with shaking from the Mighty Mouse shaker at 100% ecc. in the y-
direction. Based on Exp. 3.5.4 data. ....................................................... 290
Figure 6-45 Transfer function and coherence for Dr. Evil and Mini-Me
foundations. Exp. 3.3.5a (Unbraced structure, shaking in y-direction
with Atom Ant). ...................................................................................... 293
Figure 6-46 Transfer function and coherence for Dr. Evil and Mini-Me
foundations. Exp. 3.4.1a (Unbraced structure, shaking in y-direction
with Mighty Mouse 5% ecc). .................................................................. 293
Figure 6-47 Transfer function and coherence for Dr. Evil and Mini-Me
foundations. Exp. 3.4.19a (Unbraced structure, shaking in y-direction
with Mighty Mouse 100% ecc). .............................................................. 294
Figure 6-48 Hysteretic response secant stiffnesses (depicted by the dotted lines)
and their relationship to the load-deformation loops for the system.
Exp. 3.4.1a (Unbraced structure, shaking in the y-direction with
Mighty Mouse at 5% eccentricity).......................................................... 300
Figure 6-49 Hysteretic response secant stiffnesses (depicted by the dotted lines)
and their relationship to the load-deformation loops for the system.
Exp. 3.4.19a (Unbraced structure, shaking in the y-direction with
Mighty Mouse at 100% eccentricity)...................................................... 301
Figure 6-50 Normalized impedance function results from analysis of the Mini-Me
test structure (unbraced condition) loaded on the roof in the y-
direction by various loading amplitudes. The plots show impedance
functions in the a) translational and b) rocking modes of vibration.
The bottom plots represent the soil nonlinearity for each mode. ............ 303
Figure 6-51 Normalized impedance function results from analysis of the Mini-Me
test structure (unbraced condition) loaded on the roof in the y-
direction with Atom Ant shaker at GVDA (Exp. 3.3.5a) and WLA
(Exp. 2.1.7a). The plots show the a) horizontal translational stiffness,
b) rotational stiffness, c) horizontal translational damping ratio and d)
rotational damping ratios. Included are modeled impedance functions
based on Table 2.1 for = 0.33 and = 0.33. ........................................ 306
Figure 6-52 Normalized impedance function results from previous forced
vibration studies. The plots show the a) horizontal translational
stiffness, b) rotational stiffness, c) horizontal translational damping
ratio and d) rotational damping ratios. Included are modeled
impedance functions based on Table 2.1 for L/B=1 using = 0.33 and
D=0. ........................................................................................................ 307

xxv
 TABLES

PAGE

Table 2-1a Equations for Radiation Damping Ratios & Corresponding Terms for
Surface Foundations (Pais and Kausel, 1988) .......................................... 48
Table 2-1b Equations for Radiation Damping Ratios & Corresponding Terms for
Embedded Foundations (Pais and Kausel, 1988) ..................................... 49
Table 2-2 Values of Shear Modulus Reduction for Various Site Classes and
Shaking Amplitudes .................................................................................. 57
Table 2-3 Attributes of data set for instrumented structures with earthquake
recordings (Stewart et al., 1999b) showing period lengthening and
foundation damping from system identification and models. ................... 63
Table 2-4 Attributes of data set for field test structures subject to forced
vibration testing and earthquake shaking, including period
lengthening and foundation damping from system identification and
models. ...................................................................................................... 65
Table 3-1 List of currently compiled instrumented structure and ground arrays in
Japan. ........................................................................................................ 72
Table 4-1 Summary of the Properties of the MB and the Simplified Models
Considered ................................................................................................ 95
Table 4-2 Summary of Recorded Earthquake Events and Peak Accelerations (g)
at Three Levels ........................................................................................ 103
Table 4-3 Effective Profile Depths and Velocities for Calculation of Static
Stiffness. 1 ft=0.3048 m .......................................................................... 109
Table 4-4 Evaluation of the Impedance Functions For Sherman Oaks Building.
Static Stiffness Results Shown are for Case of Northridge Earthquake,
Which Uses a Shear Modulus Reduction Factor of G/G0=0.5. Shear
Moduli (G) and Static Stiffnesses (Ksur) for Other Earthquakes are
Doubled Relative to Those Shown In This Table (G/G0=1 for Those
Events) .................................................................................................... 111
Table 4-5 Calculation of Pile Group Stiffnesses for Horizontal and Vertical
Vibration. Modulus Applies for Case of Northridge Earthquake with
G/G0=0.5 ................................................................................................. 114
Table 4-6 Summary of Recorded Earthquake Events and Peak Accelerations at
Three Levels............................................................................................ 121
Table 4-7 Evaluation of Soil-to-Foundation Stiffness and Foundation Flexibility.
1 ft=0.3048 m .......................................................................................... 126

xxvi
Table 4-8 Effective Profile Depths and Velocities for Calculation of Static
Stiffness................................................................................................... 126
Table 4-9 Velocity Correction for Structural Overburden Weight ......................... 127
Table 4-10 Evaluation of the Impedance Functions for Walnut Creek Building.
Static Stiffness Results Shown are for Case of Loma Prieta
Earthquake .............................................................................................. 129
Table 4-11 Comparison of 1st Mode Periods for Alternative Model
Configurations (for both models)............................................................ 142
Table 5-1 GVDA Sensor Locations plan for all forced vibration testing located
at Garner Valley Downhole Array site in June and August of 2011. ..... 171
Table 5-2a Mighty Mouse Shaker Data plan for vibration testing using Mighty
Mouse shaker located at Garner Valley Downhole Array site in June
of 2011. ................................................................................................... 172
Table 5-2b Mighty Mouse Shaker Data plan for vibration testing using Mighty
Mouse shaker located at Garner Valley Downhole Array site in
August of 2011. ....................................................................................... 172
Table 5-3 GVDA forced vibration testing logs for testing on June 22, 2011
(Trial 1: Mini-Me in braced condition with forcing from Atom Ant
shaker). .................................................................................................... 175
Table 5-4a GVDA forced vibration testing logs for testing on June 23 and 24,
2011 (Trial 2: Mini-Me in braced condition with forcing from Mighty
Mouse shaker). ........................................................................................ 176
Table 5-4b Mighty Mouse shaker data logs for testing on June 23 and 24, 2011
(Trial 2: Mini-Me in braced condition with forcing from Mighty
Mouse shaker). ........................................................................................ 177
Table5-5 GVDA forced vibration testing logs for testing on August 1 and 2,
2011 (Trial 3: Mini-Me in unbraced condition with forcing from
Atom Ant shaker). ................................................................................... 178
Table 5-6a GVDA forced vibration testing logs for testing on August 2, 2011
(Trial 4: Mini-Me in unbraced condition with forcing from Mighty
Mouse shaker). ........................................................................................ 179
Table 5-6b Mighty Mouse shaker data logs for testing on August 2, 2011 (Trial 4:
Mini-Me in unbraced condition with forcing from Mighty Mouse
shaker). .................................................................................................... 180
Table 5-7 Test Experiment Used for Presentation of Structural Responses Due
to Forced Vibration Testing. ................................................................... 199
Table 6-1 Summary of Field Forced Vibration Tests Used to Evaluate
Foundation Impedance Functions (NIST, 2012) .................................... 225

xxvii
Table 6-2 Input-Output Pairs To Evaluate Flexible-Base and Fixed-Base Modal
Parameters. (Tileylioglu, 2008) .............................................................. 274
Table 6-3 System Identification of Instrumented Field Test Structures ................. 276
Table 6-4 Test Experiments of Kinematic Interaciton Due to Forced Vibration
Testing..................................................................................................... 281

xxviii
 SYMBOLS

LATIN SYMBOLS
A footing area, m2 (ft2)
ao dimensionless frequency for rectangular footings, B Vs
a0r dimensionless frequency for circular footings, r Vs
B foundation half-width (small plan dimension), m (ft)

B eA half-dimension of an equivalent square footing matching the area of the actual footing, m
(ft)

BeI half-dimension of an equivalent square footing matching the moment of inertia of the
actual footing, m (ft)
cj dashpot coefficient for translation along the j-axis, kN-sec/m (kipf-sec/ft)
cjj dashpot coefficient for rotation about the j-axis, kN-m/rad (kipf-ft/rad)
cs structural dashpot coefficient, kN-sec/m (kipf-sec/ft)
czi dashpot intensity, kN-sec/m3 (kipf-sec/ft3)
C dashpot matrix, , kN-sec/m3 (kipf-sec/ft3)
Cs seismic coefficient
C% s seismic coefficient read from the design spectrum at elongated period T%
D embedment depth, m (ft)
DL internal node spacing, m (ft)
dw height of effective side wall contact of embedded foundation, m (ft)
Ef foundation Youngs modulus, kPa (ksi)
Es Youngs modulus of the soil, kPa (ksi)
F force, kN (lbf)
fL limiting frequency (Hz)
f1 first-mode frequency from parametric system identification (Hz)
G soil shear modulus, kPa (ksi)
G0 maximum soil shear modulus (at small strains), kPa (ksi)
h height of the center of mass for the first-mode shape, m (ft)
hf height of foundation, m (ft)

xxix
H Horizo
ontal force
Hu kinematic transfer function
f
area moment
m of ineertia of soil-fooundation conntact, m4 (ft4);; i denotes thee axis about which
w
the surrface is taken
If mass moment
m of ineertia about thhe foundation,, m4 (ft4)
j index denoting moddes of translattion displacem
ment or rotatiion
Jt polar moment
m of inertia of soil-ffoundation contact surface,, m4 (ft4)
k spring stiffness, trannslation, kN/m
m (lbf/ft), andd rotation, kN
N-m/rad (lbf-fft/rad)
kj spring stiffness, trannslation, alonng the j-axis, N/m
N (psi)
ks structu
ural stiffness (structural
( compliance), cm
m (in)
K stiffneess matrix, Mgg (lbs)
Kj foundaation static stiiffness for moode j, translattion, N/m (lbff/ft), and rotattion, N-m/radd (lbf-
ft /rad))
k *j complex-valued im mpedance funcction for modee j, translationn, N/m (lbf/ftt), and rotatioon,
N-m/raad (lbf-ft/rad))
kjj spring stiffness, rotation, about the
t j-axis, kN-m/rad (lbf-ftt/rad)
k zi m3 (lbf/ft3)
stiffneess intensity (ccoefficient off subgrade reaaction), kN/m
L foundaation half-length, m (ft)
Lend o foundation with increasee stiffness, m (ft)
end region length of
m mass, Mg (lbs)
M mass matrix,
m Mg (lbbs)
mf mass of
o foundation, Mg (lbs)
ms mass of
o superstructture, Mg (lbs))
n,, ns, nx, nyy exponents
e rellated with fouundation dampping
Rc dashpo
ot intensities scalar (Equattion 2-21 in NIST
N report)
Re end len
ngth ratio
Rk spring stiffness inteensities scalarr (Equation 2--21 in NIST report)
r
r foundaation radius, m (ft)
rrr rotatio
onal mode aboout axis of a circular
c founddation

xxx
Sa, Sa-FIM spectral acceleration, fixed base, and with FIM effects, respectively, m/s2, (ft/s2), g

S ae equivalent spectral acceleration, m/sec2 (ft/sec2), g

S%a spectral acceleration for a flexible-base structure, m/sec2, (ft/sec2), g
Sd, S de spectral displacement and equivalent spectral displacement, m (ft)
T undamped natural vibration period, sec
T% flexible-base period, sec
tf foundation thickness, m (ft)
Ti fixed-base period for a case history (Sections 2.6.2 and 6.6), sec;
~ flexible-base period for a modeled prediction (Sections 2.6.2 and 6.6), sec
Ti
Tp predominant period of ground motion, sec
Tx, Tyy fictitious vibration periods, calculated as if the only source of the vibration was
translation or rotation, sec
T x , T yy complex-valued fictitious vibration periods, calculated as if the only source of the
vibration was translation or rotation, sec

T% T period lengthening ratio due to SSI

T%i Ti period lengthening ratio from case history data (Sections 2.6.2 and 6.6)
T%i Ti period lengthening ratio predicted from model (Sections 2.6.2 and 6.6)
uf measured foundation horizontal deflection, cm (in)
uFIM foundation input motion, acceleration, m/sec2, velocity, m/sec or displacement, m
ug free-field ground motion, acceleration, m/sec2, velocity, m/sec or displacement, m
us deflection amount of structural mass associated with the structural compliance, cm (in)
U&& ,U& , & U horizontal and rotational accelerations, velocity and displacements,
respectively, of each degree of freedom
Vapp apparent wave propagation velocity, m/sec (ft/sec)
V base shear force, kN (lbf)
Vs shear wave velocity over a representative depth, m/sec (ft/sec); used generally in report
for either measured or averaged result
Vs,d time-averaged shear wave velocity over the embedment depth (D) of the foundation
Vs0 time-averaged shear wave velocity in free-field, over depth interval zp, m/sec (ft/sec)

xxxi
Vs,f overburden-corrected shear wave velocity under footing, m/sec (ft/sec)
zp depth interval, m (ft)

GREEK SYMBOLS
j dynamic (frequency) stiffness modifier for surface foundation for mode j
f foundation damping ratio
f ,i foundation damping ratio for a case history (Sections 2.6.2 and 6.6)
f ,i foundation damping ratio for a modeled prediction (Sections 2.6.2 and 6.6)
i fixed base structural damping ratio
j dynamic damping ratio for mode j
0 flexible-base system damping ratio
rj radiation damping ratio for mode j
s soil hysteretic damping ratio
1 corresponding first-mode damping ratio from parametric system identification
shear strain
m soil-structure mass ratio = m hA ; typical range of 0.1 to 0.2 for flexible to stiff
structures, respectively, based on Veletsos (1977)
deflection, m (ft)
% total deflection with respect to free-field at the top of a SDOF structure, cm (in)
v increment of vertical stress at depth z from the structural weight, kPa (ksi)
j, jj embedment correction factor for rigid footing spring constants for mode j
rotation deflection amount in a SDOF structure due to base moment (rad)
f measured foundation rotation (rad)
a dimensionless parameter accounting for effects of incoherence from wave passage and
stochastic processes, as well as Fourier amplitude variability
soil Poissons ratio
f foundation Poissons ratio
s soil mass density, Mg/m3 (lbs/ft3)
m effective confining stress, kPa (ksi)

xxxii
 v 0 effective stress from the self-weight of the soil, kPa (ksi)
undamped natural vibration frequency (rad/sec)

xxxiii
 ACKNOWLEDGMENTS

I would like to express my great appreciation and gratitude to my advisor Jonathan P.

Stewart for his guidance and support during my research. He was instrumental in shaping

my doctoral program and providing me the opportunities to get a well-rounded breath of

knowledge of soil-structure interaction. Thank you for selecting me to carry out these

tasks and in turn helping me become a better engineer. Our almost weekly meetings

discussing research and various related topics were invaluable to me.

Thank you to my committee members Scott Brandenberg, An Yin, Ertugrul

Taciroglu, and George Mylonakis for their valuable input into this document. I appreciate

the time you have spent reading, discussing and critiquing my work. I am honored to

have had my work reviewed by such well-accomplished people. A special thanks to

George Mylonakis for his countless hours of collaboration with me.

Support for my work to produce Chapter 4 was provided by the NEHRP

Consultants Joint Venture, a partnership of the Applied Technology Council and

Consortium of Universities for Research in Earthquake Engineering, through Contract

SB134107CQ0019, Earthquake Structural and Engineering Research, issued by the

National Institute of Standards and Technology (NIST).

Support for my work to produce Chapters 5 and 6 was provided by the National

Science Foundation under Grant No. CMMI-0618804 through the George E. Brown Jr.

Network for Earthquake Engineering Simulations (NEES), PI: Dr. Jack Moehle.

xxxiv
 This research could have not been possible without the love and support of my

family and friends. My wife Debbie has been my biggest supporter and confidante during

my studies. The satisfaction that she brings to my life makes me better at everything.

xxxv
 VITA

2003-2005 Engineering Technician / Geologist-in-Training

Geo-Environmental, Inc.
Irvine, California

2006 B.S., Geology

California State University, Long Beach

2006-2009 Staff Geologist

Geo-Environmental, Inc.
Irvine, California

2010 M.S., Civil Engineering

University of California, Los Angeles

2010-2011 Project Graduate Student

Consortium of Universities for Research in
Earthquake Engineering (CUREE)
Richmond, California

2011-2013 Graduate Research Student

Department of Civil and Environmental Engineering
University of California, Los Angeles

2013 Teaching Assistant

Department of Civil and Environmental Engineering
University of California, Los Angeles
Awarded: UCLA ASCE TA of the Year 2012-2013

PUBLICATIONS

Givens, M.J., Haselton, C., Mazzoni, S., and Stewart, J.P., 2012. Assessment of Soil-
Structure Interaction Modeling Strategies for Response History Analysis of
Buildings, Proc. 15th World Conf. on Eqk. Engrg, September 24-28, Lisbon
Congress Center, Lisbon.
Givens, M.J., Mikami, A., Kashima, T., and Stewart, J.P, 2012. Kinematic Soil-Structure
Interaction Effects from Building and Free-Field Seismic Arrays in Japan, Proc.
8th Inter. Conf. on Urban Eqk. Engrg & 4th Asia Conf on Eqk. Engrg/, March 6-8,
Tokyo Institute of Technology, Tokyo.
National Institute of Standards and Technology, (NIST), 2012. Soil-Structure Interaction
for Building Structures, Report NIST/GCR 12-917-21, presented by NEHRP
Consultants Joint Venture, J.P. Stewart (project director), September; Working
Group Member.

xxxvi
1 INTRODUCTION

1.1 Soil-Structure Interaction in Engineering Practice

In this Ph.D. research, I seek to perform analysis of field performance data to provide

deeper insights into SSI phenomena ranging from kinematic effects on foundation ground

motions to mobilized foundation stiffness and damping across a wide range of

frequencies and loading levels. These data are interpreted to evaluate strengths and

limitations of engineering analysis procedures for SSI. The profession seeks to provide

accurate evaluations of facility performance in response to seismic excitation. This

includes formal consideration of uncertainties in the representations of seismic demand

and the engineering evaluation of response to that demand. For that reason, a seismic

response analysis operates most effectively when the model of the structural system in

question is as realistic as is reasonable possible. Currently, however, conventional design

practice typically ignores SSI based on the perception that the SSI components would

provide beneficial effects to the structure during seismic events. It is assumed that the

increase in flexibility of the structure due to SSI would decrease the demand on the

structure. Conventional engineering practice is not performance-based, but is based on

certain prescriptive requirements for demand specification and structural detailing that is

intended to achieve a single performance goal collapse prevention.

1
 What is problematic about this approach is that, because it is assumed that SSI

effects reduce the demands on structures, ignoring SSI effects is considered the

equivalent of erring conservatively. In reality, it is not always conservative to ignore SSI

effects, as will be described in Chapter 2 of this research. Moreover, because

conventional design practice seeks to evaluate accurately the expected building response

and its dispersion, deliberately erring an analysis result (conservatively or not) is counter-

productive. I will argue that comprehensive design analysis should consider SSI, and

indeed there is strong evidence that the profession is moving in this direction at an

increasing rate.

1.2 Alternative Frameworks for SSI Analysis

SSI can be incorporated into seismic response analysis using direct or substructure

approaches. The direct approach represents the SSI system globally as a continuum

during the analysis. Figure 1-1 depicts the direct approach using a finite element method

where the conventional analysis is expanded beyond just detailed modeling of the

structure and foundation elements to include the foundation-soil interface elements, soil

elements, and the seismic incident wave interactions. The direct approach of modeling

the entire SSI system in one step can be computationally costly.

Figure11 Schematic illustration of direct approach to analysis of the soilstructure

interaction(SSI)problemusingcontinuummodeling(NIST,2012).

The substructure approach breaks up the SSI system into several parts (to simplify

the analysis) as shown in Figure 1-2. The substructure approach is described below:

a) Complete System: The complete soil-foundation-structure system is excited by

an incident wave field represented by a free-field ground motion (ug). Motion ug refers to

ground motion that would occur in the absence of a foundation at the site.

b) Kinematic Interaction: A non-embedded rigid foundation excited by vertically

propagating seismic incident waves would theoretically experience the motion ug.

However, in reality, incident waves are incoherent and do not propagate vertically.

Kinematic interaction refers to the deviation of the foundation motion from the ug as a

result of the foundations geometry and embedment in the absence of the building and

foundation inertial responses (which is considered in Section c below). Kinematic

3
interaction results from base slab averaging, wave scattering and embedment effects. As a

result, the substructure approach uses the modified motion termed a Foundation Input

Motion (FIM) for analysis.

c) Foundation-Soil Flexibility and Damping: The seismic loading at the

foundation level introduces an inertial response into the building which in turn produces

base moment, shear, and torsion. These demands produce foundation displacements and

rotations that result in a redistribution of forces within the building relative to a fixed-

base case (no SSI), altering the buildings inertial response. This process is known as

inertial interaction and is captured in the substructure approach by using frequency-

dependent soil springs and dashpots. Classical solutions for impedance functions (Section

2.2) were developed for the rigid foundation case represented in Figure 1-2 by and

for translation and rocking, respectively. These springs and dashpots can be applied

in a simple manner for a rigid foundation (Figure 1-2c.i) or dispersed in a series of

springs and dashpots for flexible foundations (Figure 1-2c.ii).

4

Figure12 Schematicillustrationofsubstructureapproachtoanalysisofsoilfoundation
structure interaction (SFSI) problem using either (i) rigid foundation or (ii)
flexiblefoundationassumption(NIST,2012).

d) Excitation with FIM of Structure with Foundation Flexibility/Damping:

Response history analysis is performed through superposition of the inertial interaction

and kinematic interaction effects. The system is excited by the FIM at the end of the

5
springs, rather than using the ug. As opposed to the rigid foundation case (Figure 1-2d.i)

where springs are excited by a single uFIM and FIM, the flexible embedded foundation

case (Figure 1-2d.ii) requires multi-support excitation. The multi-support excitation

results from site response effects to the uFIM and is commonly determined by a simple

equivalent linear ground response analysis.

The substructure approach represents the SSI problem in a simplistic manner. It

entails evaluating the SSI effects as two separate components before performing the

seismic response analysis. First, the soil elements are represented with frequency-

dependent springs and dashpots. Second, the seismic incident wave motion is introduced

as a FIM. This approach allows for the two components (representing inertial interaction

and kinematic interaction, respectively) to be applied in structural analysis computer

software to perform relatively realistic response history analyses of buildings (as

compared to fixed-base analysis). The substructure approach does not account for soil

nonlinearities introduced by the inertial response (i.e., the spring deflection), which may

affect the incoherent demand distribution and impedance. Nonlinearity can be

incorporated in an approximate way through the site response (often called primary

nonlinearities).

1.3 Research Motivations and Organization

I was a working group member for a National Institute of Standards and Technology

(NIST) Task Order 69221 entitled Improved Procedures for Characterizing and

Modeling Soil-Structure Interaction for Performance-Based Seismic Engineering. The

task order resulted in the NIST/GCR 12-917-21 report, entitled Soil-Structure

6
Interaction for Building Structures, referred to in this document as NIST report. The

NIST report provides a detailed description of analysis methods reflecting the current

state of SSI knowledge. My contributions to the NIST report included assistance to the

Project Director in SSI literature review and development of publication quality figures

presenting SSI concepts, becoming the first user of the NIST report to develop soil

springs and dashpots and variable ground motions inputs (FIM and ug) for problem

focused studies (PFS) outlined in the NIST report, evaluation of EDPs that resulted from

the PFS, and implementation of review comment changes produced from practitioners

and researches at a NIST workshop after a 90% completion of the NIST report.

The SSI system behavior is often represented by period lengthening and damping

of a single-degree-of-freedom (SDOF) system, which is used in the design provisions

(e.g., FEMA-440, 2005). Whereas the period lengthening effect is relatively

straightforward, there is little consistency in the literature on foundation damping.

Currently, design provisions such as ASCE-7 (2010) present graphical correlations of

period lengthening versus foundation damping that are based on rigid, circular

foundations on homogenous soils. There is a need to understand the reasons for the

different foundation damping models and then to provide more representative models that

account for rectangular shaped foundations. Guidance on how to account for the effects

of flexible foundations and heterogeneous soils in the analysis of foundation damping is

also needed.

Starting from first principles, we derive fundamental expressions for foundation

damping in which foundation impedance components representing radiation damping and

7
the soil hysteretic damping ratio appear as variables, providing maximum flexibility to

the analyst. We utilize these general expressions with impedance solutions for

rectangular-footprint foundations to: (1) compare predicted foundation damping levels

with those from field case histories and (2) develop new foundation damping

relationships for application in the building code (NEHRP Provisions).

Kinematic interaction effects are commonly expressed as transfer functions to

represent the foundation to free-field motions. Simplified models for these transfer

functions have been distilled from the literature and presented in the NIST report (2012)

and ASCE-41. As a result, these models have become widely used, though they have

recently been questioned by some researchers. The critiques have argued that the models

have predicted unconservative results compared to recorded seismic arrays and therefore

should be used with caution. The simplified models were developed from western

United States sites with foundation base dimensions of =15-40 m (where is the

square root of the foundation area divided by 4). Data from Japan has recently become

available for comparative studies to the existing models. Chapter 3 presents a comparison

of the aforementioned models to Japanese data and presents conclusions as to the validity

of those models for use in the substructure approach to SSI analysis. Some issues related

to kinematical modeling of foundations with heavy pile foundations are identified.

During the development of the NIST (2012) report, an internal baseline

knowledge report was completed presenting the state of knowledge for SSI modeling,

including proposed procedures for substructure modeling. Problem focused studies (PFS)

were undertaken to apply the procedures. A primary goal of the PFS was to shed light on

8
the relative significance of different elements of the modeling process in the predictions

of structural Engineering Demand Parameters (EDPs). PFS of two buildings were

completed as part of the NIST report. In an effort to evaluate the effects of SSI

components on the dynamic structural response, the PFS included multiple modeling

schemes of varying complexity to evaluate the relative significance of various modeling

details on the computed EDPs.

I was tasked with developing the soil springs/dashpots and input ground motions

for the PFS. The development of the springs/dashpots required a background review of

construction drawings and reports containing local geotechnical boring logs and seismic

surveys at or close to the sites modeled. Following the review, I was faced with several

difficulties including appropriate selection of shear modulus for stiffness values in the

presence of embedded structures, how to deal with hybrid foundation systems that have

both shallow and deep components, evaluation of the flexibility of the foundation,

determination of increased stiffness values at the edges of the foundation, and soil

damping modeling. Additionally, the input ground responses were modeled as both the ug

and uFIM. The issue for the PFS was reversed from typical practice in that the uFIM was

known and the ug was back-calculated. During the development of the multi-excitation

motions, I encountered phasing issues from the site response analysis (performed with

DeepSoil computer software) compared to the original FIM. The PFS are discussed in

detail in Chapter 4.

Chapter 5 describes the forced vibration testing of a steel and reinforced concrete

structure that was performed to evaluate soil-structure interaction (SSI) systems. Testing

9
was performed at multiple George E. Brown, Jr. Network for Earthquake Engineering

Simulations (NEES) sites. This work has been completed as part of a multi-institutional

National Science Foundation (NSF) grand challenge (NSF 06-504 Program) project

called, Mitigation of Collapse Risk in Vulnerable Concrete Buildings. This project has

been documented on the Network for Earthquake Engineering Simulations Research

(NEESR) website, known as NEEShub, as project NEES-2008-0637. The SSI test series

consisted of three test sites with varying degress of base flexibility: University of

California, Los Angeles Laboratory (LAB), Wildlife Liquefaction Array (WLA), and

Garner Valley Downhole Array (GVDA). The LAB represents an effectively fixed-base

condition with structural columns attached to the existing laboratory strong-floor. The

WLA site, with a soft clay layer overlying loose, liquefiable sands, and the GVDA site,

with firm lake-bed alluvium material, represent flexible base conditions having variable

levels of soil stiffness. The three sites were chosen to represent a range of SSI effects

based on the foundation conditions with practically no SSI effect expected at the LAB to

moderate effects at the GVDA site to a strong SSI influence at the WLA. Chapter 5

presents the details of the experiment performed at the GVDA site, which was my

principal contribution to the NEES testing sequence.

Chapter 6 presents an evaluation of SSI effects from forced vibration testing at the

GVDA site. As will be shown, a limited number of full-scale field tests have been

performed to specifically to evaluate SSI effects for comparison to the existing models.

Many full-scale forced vibration tests have been performed; however, when evaluation of

SSI effects is not the motivation of the work the responses needed to properly identify

10
SSI effects are commonly missing. The evaluation of SSI effects on buildings requires

the horizontal accelerometers at the free-field, foundation, and roof and two vertical

accelerometers on the foundation to evaluate rocking response (Stewart and Fenves,

1998). The structure that was forced with shakers at GVDA was well-instrumented along

with a concrete slab and adjacent structure. Chapter 6 presents analysis and

interpretations of SSI effects that were ascertained from the field testing at GVDA.

Chapter 7 reviews the scope of the research undertaken in my doctoral studies. I

then summarize the principle findings and recommendations of the technical work

described in Chapters 2-6. Finally, I provide recommendations for future work based on

lessons learned during this research.

11
2 FOUNDATIONDAMPING

2.1 Introduction

The contribution of foundation damping to the overall damping of a soil-foundation-

structure system was introduced in the 1970s in seminal studies by Bielak, Veletsos and

colleagues. The work was predicated on the need to evaluate the effects of soil-structure

interaction (SSI) on the seismic response of nuclear power plants. Based on that need,

alternative sets of equations were developed to predict foundation damping of a rigid

circular foundation resting on a uniform elastic halfspace. Due in part to the convenience

of its application in the specification of seismic demands using response spectrum (force-

based) or pushover (displacement-based) methods of analysis, foundation damping has

more recently appeared in several seismic design guidelines for building structures (e.g.,

ASCE-7; ASCE-41). In this chapter, we investigate the alternate assumptions used to

derive foundation damping equations. Moreover, considering most foundations are

neither circular nor situated on uniform elastic soils, we reevaluate the foundation

damping equations to consider various site-specific conditions such as a rectangular

footprint, embedment and non-uniform soil conditions.

12
2.2 Applications of Foundation Damping

In this section, we review the application of foundation damping in force-based and

displacement-based methods of analyzing seismic demands in structures.

2.2.1 Foundation Damping in Force-Based Methods

Force-based methods of analysis for seismic demand take the seismic base shear force as

the product of pseudo spectral acceleration (PSA) at the first-mode period (normalized by

g) and the building weight. Within this context, the effect of inertial SSI on the base shear

for a building structure is illustrated in Figure 2-1. The PSA for a flexible-based structure
~
( S a ) is obtained by entering the spectrum drawn for flexible-base damping ratio 0 at the

~
corresponding flexible base period T . As indicated in Figure 2-1, both the damping ratio

and period are modified from the fixed-base case in which inertial SSI effects are
~
ignored; period is elongated from T (fixed-base) to T , while damping 0 can decrease or

increase from the fixed-base value, i, although damping increase is more common. For

the common case of buildings with relatively long periods on the descending portion of
~
the spectrum, using S a in lieu of Sa typically reduces base shear demand, whereas in

relatively short-period structures SSI can increase the base shear. These modifications of

base shear are related to the slope of the spectrum: base shear tends to increase when the

slope is positive and decrease when the slope is negative.

13


Figure21 Illustration of effects of period lengthening and change of damping on

fundamentalmodespectralaccelerationfromwhichbaseshearisevaluated.
(NIST,2012)

Since the effects of SSI on base shear is related to its impact on fundamental-

mode period and damping ratio, simplified procedures have been developed to predict

both of these quantities (Veletsos 1977, Stewart et al 2003). Figure 2-2 shows a single-

degree-of-freedom (SDOF) oscillator that is useful to visualize the link between

foundation compliance and period lengthening. Figure 2-2(a) depicts the fixed-base

system having height h and period T = 2 m k , which neglects SSI. Figure 2-2(b)

depicts the deformable nature of the soil-foundation system using translational and

rotational springs, which enable the base of the structure to displace horizontally and

rotate in response to base shear and moment, respectively.

14
Figure22 Schematicillustrationofdeflectionscausedbyforceappliedto(a)fixedbase
structure and (b) structure with horizontal and rotational compliance. (NIST,
2012)

The springs in Figure 2-2(b) are the real component of complex numbers. Spring

kx represents the horizontal spring stiffness and kyy represents the rotational stiffness

against foundation rocking in the x-z plane. For simplicity, throughout this chapter

equations will be adapted to these subscript notations in which horizontal translation is

along the y-axis (y) and rotation is about the x-axis (xx). The period of the flexible-base

structure can be readily computed from the approximate formula (Veletsos and Meek,

1974):

~
T k kh 2
= 1+ + (2-1)
T k x k yy

where is referred to as the period lengthening ratio.

Foundation damping represents the second necessary component of the SSI

modeling to modify the base shear, as depicted in Figure 2-1. As discussed further in

15
Section 2.2, foundation damping f can be related to fixed-base damping i and flexible-

base damping 0 as:

1
f = o i (2-2)
(T% T )
n

where n is an exponent related to details of the foundation damping derivation, as

described in Section 2.2. In practice, f is computed using procedures in Section 2.2, and

then combined with the i term in Eq. (2-2) to compute 0.

2.2.2 Foundation Damping in Displacement-Based Methods

In displacement-based methods, the system performance is represented by a lateral force-

displacement relationship calculated using pushover analyses. As illustrated in Figure 2-

3, pushover analyses involve application of static lateral loads distributed over the height

of the structure and calculation of the resulting displacements from a model of the SSI

system. The cumulative lateral load (resultant H in Figure 2-3) is related to a reference

displacement from the nonlinear pushover curve. A point on this diagram defines a

damage state in the building, since the deformation of all of the structural components is

assumed to be related to the reference displacement. For some applications, the pushover

curve is modified to an acceleration-displacement response spectrum (ADRS) by

converting H to an equivalent spectral acceleration and by converting to an equivalent

spectral displacement (e.g, Chopra and Goel, 1999; Powell, 2006).

16


Figure23 Schematicillustrationofanalysistodeveloppushovercurve(NIST,2012)

The expected seismic structural performance is assessed by combining the

capacity diagram with an Acceleration-Displacement Response Spectra (ADRS) as

illustrated in Figure 2-4. The ADRS represents the seismic demand, and consists of a plot

of elastic spectral acceleration (Sa) vs. spectral displacement (Sd). The free-field demand

spectrum is reduced twice, first for kinematic interaction effects (Chapter 3), then for the

effects of inelastic deformation of the structure and potential damping increase from i to

0. The reduced ADRS is compared to the capacity diagram. The capacity and demand

spectra meet at the performance point, which represents the expected structural

performance given the seismic demand.

17
Figure24 SchematicillustrationofSFSIeffectsondisplacementbasedmethodof
evaluatingnonlinearstructuralperformance.

SSI enters displacement-based analysis procedures through three components: (1)

springs used in pushover analysis; (2) reduction of free-field design spectrum for

kinematic interaction; and (3) reduction of design spectrum for SSI system damping

ratios 0 greater than i. Hence, the role of foundation damping is similar to force-based

methods, in that it is used to scale the response spectrum representing the seismic

demand. Procedures similar to these appear in ATC-40, FEMA-440, and ASCE-41

(ATC, 1996; FEMA, 2005; ASCE, 2007, respectively).

2.3 General Solution for Foundation Damping

Most seismic design guidelines utilize the analogy of a replacement oscillator with period

and an adjusted flexible-base damping ratio 0 in the analysis of seismic demand (e.g.,

base shear). In the previous section, it was shown how the analysis of foundation

damping is integral to evaluating 0.

18
 Whereas the analysis of period lengthening is relatively straightforward,

formulating an analytical solution for foundation damping is more complex, as it requires

assessing the relative contributions of hysteretic and radiation damping in multiple modes

of foundation vibration. Analytical models for foundation damping of circular

foundations have been presented, among others, by Veletsos and Nair (1975), Bielak

(1971, 1975, 1976), Roesset (1980), Wolf (1985), and Maravas et al. (2007). These

foundation damping models are discussed in the remainder of this section.

2.3.1 Impedance Solutions for Shallow Foundations

The complex-valued impedance functions ( k j ) of a general foundation system in

vibration mode j can be expressed as function of the frequency-dependent foundation

stiffness (kj) and dashpot coefficient (cj) as:

k j = k j + ic j (2-3)

where =cyclic frequency (rad/s) and j subscript denotes either the translational (x) or

rotational (yy) vibration mode. For circular foundations, translation is often indicated

with subscript r and rotation as r.

The real part of the complex-valued springs are denoted kj (no overbar) and

represent ordinary springs. The imaginary part (cj) accounts for the phase difference (lag)

between excitation and response in a harmonically excited damped system, and can be

interpreted as viscous damper (cj). An alternative form for Eq. 2-3 is

k j = k j (1 + 2i j ) (2-4)

where

19
 c j
j = (defined for kj > 0) (2-5)
2k j

An advantage of using j over cj is that at resonance of the SSI system the former can be

interpreted as a percentage of critical damping in the classical sense (Clough and Penzien,

1993). The stiffness coefficient (kj) is a function of the soil shear modulus (G), Poissons

ratio (), dynamic stiffness modifier (j) and foundation dimension:

k j = K j j (2-6a)

Circular foundation: K j = Gr m f ( ) j = f (a0r , ) (2-6b)

Rectangular foundation: K j = GB m f ( B L , ) j = f (a0 , B L) (2-6c)

r
Circular foundation: a0r = (2-7a)
Vs

B
Rectangular foundation: a0 = (2-7b)
Vs

where Kj is the static foundation stiffness at zero frequency for mode j, a0r and a0 are

dimensionless frequency parameters, exponent m is 1 for translation (x or r) and 3 for

rotation (yy or r), and B and L are foundation plan half-dimensions, as indicated in

Figure 2-5. The aforementioned equations are described for circular and rectangular

foundations.

20
Figure25 Sketchshowingthegeometryadaptedfortherectangularfoundation (L B ) .

Solutions for circular foundation impedance terms in Eq. (2-3) to (2-7) were

derived by Luco and Westmann (1971) and Veletsos and Wei (1971) for the condition of

rigid circular discs resting on elastic halfspace. The solutions were formulated from

numerical analysis of the mixed boundary value problem (i.e., without assuming a

particular distribution of tractions or displacement at the soil-footing interface). They

worked independently and produced similar results.

Veletsos and Verbic (1973) presented simplified approximate solutions for

impedance functions involving circular rigid foundations resting on an elastic or

viscoelastic halfspace. Figure 2-6 shows those approximate solutions relative to the

elastic solutions for j and j by Veletsos and Wei (1971) (for Poissons ratios of 0.45

and 0.33). Results similar to those shown in Figure 2-6 were obtained by Luco (1976),

who solved the circular foundation impedance problem for a viscoelastic halfspace,

which extends his previous (1971) results for an elastic halfspace. The Veletsos and

Verbic (1973) solutions are convenient in that they present closed-form equations that

21
can directly account for dynamic effects in the impedance functions including soil

damping.

Figure26 Shows the comparison of the dynamic stiffness modifiers and radiation
dampingratiosasdeterminedbythenumericaldata(VeletsosandWei1971)
andtheapproximatemethod(VeletsosandVerbic1973).Conditionsforplots
are a rigid, massless, circular disc supported on an elastic homogeneous
isotropic halfspace with hysteretic soil damping s = 0. = 0.45 for a nearly
saturatedclayand=0.33forsand.

The numerical solution and approximate solution trends in Figure 2-6 match fairly

well for both soil types from a0r = 0 2. This is a practical range of analysis, since most

foundation excited by an earthquake will have the frequency range of interest fall below

22
a 0r = 2 . Solutions for rectangular foundations are presented subsequently in Section 2.5

as part of our proposed model for foundation damping.

2.3.2 Considerations on the Use of Viscous Structural Damping

As shown in Figure 2-2, structural damping is represented here with damping ratio i.

The damping ratio is an expression of the decay of oscillations in a system and several

models for damping exist. Damping is often taken as either viscous (linearly frequency-

dependent) or hysteretic (frequency-independent).

If a viscous damping ratio is assumed, then a modification must be made to the

equations in which it appears. A viscous damping ratio for a vibrating system with

natural frequency n is defined as (e.g., Chopra, 2011):

civis
ivis = (2-8)
2n m

where ivis = viscous damping ratio, civis is the viscous dashpot coefficient, and other

variables are as given above. The complex-valued structural stiffness can be expressed as

follows:

c vis
k = k + icivis = k 1 + 2i i (2-9)
2k

From structural dynamics, the undamped natural vibration frequency, n , and period, T,

of the structure are given by Clough and Penzien (1993) as:

k 2 m
n = , T= = 2 (2-10)
m n k

23
Substituting Eq. (2-8) into Eq. (2-9) and recognizing the n definition in Eq. (2-10)

provides an expression for the complex-valued structural stiffness with a viscous

damping ratio:

vis
k = k 1 + 2i (2-11)
n i

Recall that in this expression is the excitation frequency. Recognizing that

k = k (1 + 2i i ) and comparing to Eq. (2-11), we see that the general damping ratio i

can be related to its viscous counterpart as follows:

vis
i = (2-12a)
n i

At resonance, the excitation frequency is equivalent to the system frequency, % , and

the above expression can be expressed as:

% vis 1
i = i = % ivis (2-12b)
n T T

Hence, the use of a viscous damping ratio introduces a dependence on period lengthening

into the imaginary part of the structural stiffness.

24
2.3.3 Foundation Damping Derivations

Foundation damping as a distinct component of the damping of a structural system was

introduced as part of Bielaks (1971, 1973) derivation of the replacement SDOF system

and was later refined by Veletsos and Nair (1975), Roesset (1980) and others. According

to this approach, the overall damping of the soil-foundation-structure systems () can be

partitioned into components associated with soil-foundation interaction (including

hysteretic and radiation damping), f, and damping within the structural system, i, by re-

arranging Eq. (2-2) as follows:

1
o = f + ~ i (2-13)
(T T )n
Equation 2-13 appears in many seismic design guidelines with n = 3. As shown in the

previous section, the exponent n is dependent on the type of damping (i.e., viscous or

general, including hysteretic). Historical precedent and system identification studies often

dictate values of i in the range of 2-5% in structural design practice for buildings

responding in the elastic range (FEMA 440, 2005). Since i can be established from such

guidelines and period lengthening analysis is relatively straightforward as described in

Section 2.2.1, the most challenging aspect of deriving the system damping 0 is the

evaluation of foundation damping f.

We consider the SDOF replacement oscillator shown in Figure 2-2(b). The total
~
deflection ( ) of a flexible-base oscillator mass (m) at height (h) from the application of

a horizontal force (F) results from horizontal deflection of the structural spring and rigid

25
body displacement of the structural mass from horizontal foundation displacement (uf)

and base rotation ( ).

F
% =+ u f + h
k
(2-14)
F F F h
% = + + h
k k x k yy

~
Considering that the replacement oscillator has deflection in response to force F, the
~
stiffness of the replacement oscillator k% can be taken as the ratio of to F, which leads

directly to the period lengthening expression given in Eq. (2-1).

The formulation of the damping component of the equivalent oscillator requires

consideration of the imaginary part. Damping results from the phase difference between

the real and imaginary parts. We present two derivations for foundation damping. Both

begin by dividing the right and left sides of Eq. (2-14) by force F, recognizing that % F

is equivalent to an effective flexible stiffness of the replacement oscillator k% , and

generalizing each term for dynamic loading through the introduction of complex-valued

stiffness terms (indicated by an overbar), as follows:

1 1 1 h2
= + + (2-15a)
k% k k x k yy

We present two ways to proceed from Eq. (2-15a) to derive expressions for foundation

damping. The first approach (Section 2.3.3.1), which is similar in some respects to prior

work by Bielak (1971) and Maravas et al. (2007), separates Eq. (2-15a) into its real and

complex parts, then operates exclusively on the imaginary part to evaluate the effective

damping of the replacement oscillator. The foundation damping is then readily derived

26
from the system damping. The second approach (Section 2.3.3.2), which is similar in

some respect to prior work by Veletsos and Nair (1975), retains both the real and

complex parts of Eq. (2-15a) in the evaluation of expressions for the dynamic properties

of the replacement oscillator. The resulting expressions for foundation damping retain

complex-valued terms, which are not present in the first approach. Differences between

foundation damping predictions derived from the two approaches are presented in

Section 2.3.3.3.

2.3.3.1 Derivation from Imaginary Component

The first approach proceeds from (Eq. 2-15a) by expanding each complex stiffness term

according to Eq. (2-4):

1 1 1 h2
= + + (2-15b)
k% (1 + 2i 0 ) k (1 + 2i i ) k x (1 + 2i x ) k yy (1 + 2i yy )

Note that hysteretic soil damping effects are not considered at this stage, but are

introduced subsequently. Multiplying and dividing each term by its complex conjugate,

neglecting the higher-order damping term (i.e., 2 0 ), and multiplying both sides by k,

we obtain:

kh 2
k
~ (1 2i 0 ) = (1 2i i ) +
k
kx
(1 2i x ) + (
k yy
1 2i yy ) (2-16)
k

The equality in Eq. (2-16) requires that both the real and imaginary parts of the

expressions on the right and lefts sides of the equal sign be equivalent. Accordingly, we

consider the equivalence of the imaginary parts exclusively as follows:

27
 k% k% % 2
kh
0 = i + x + yy (2-17a)
k kx k yy
~ ~
k k k kh 2
0 = i + x + yy (2-17b)
k k k x k yy

Eq. (2-17b) is convenient in the sense that the flexible-base system damping components

are proportional to the stiffness ratio for flexible-base and fixed-base oscillators ( k% k ),

which can be related to the period lengthening (Eq. 2-1) when foundation mass is

ignored. Using the definitions in Eq. (2-10), period lengthening can be recovered from

the stiffness ratio as follows:

~ ~
k k m ~ 2 ~ 2 2 1
= = 2 = 2 = (2-18)
k m k 2 ~
T T ( ) 2

Using Eq. (2-18), the flexible-base system damping in Eq. (2-17b) can now be presented

as a function of period lengthening:

1 k
1 2
0 = ~ + ~ x + kh yy (2-19)
(T T ) 2 i
( )2
T T kx k yy

To further simplify, we introduce the following fictitious vibration periods, calculated as

if the only source of the vibration was foundation translation or rotation, as follows:

m mh 2
T x = 2 T yy = 2 (2-20)
kx k yy

k k x = ( Tx T )
2
Using Eq. (2-20) and recalling T = 2 m k , term and

kh 2 k yy = ( T yy T ) . With these substitutions, Eq. (2-19) becomes:

2

28
 1 1 1
0 = i + x + yy (2-21)
(T% T ) (T% Tx ) (T% Tyy )
2 2 2

Eq. (2-19) and (2-21) have been developed based on the general impedance

functions (Eq. 2-3) that considers the damping terms to be non-viscous. Frequency-

independent (i.e., hysteretic) soil damping ( s ) can be included in the system by simply

adding it to the translational and rotational damping terms (Rosset, 1980, and Wolf,

1985). When this is applied to the damping formulation in Eq. (2-19), we obtain:

kh 2
0 =
1
i +
k
( + ) + yy + s ( ) (2-22a)
( )
~ 2 x s
T T kx k yy

Rearranging Eq. (2-22a) based on the damping terms gives:

1 k kh 2 k kh 2
0 = i + x + yy + + s (2-22b)
(T~ T )2 kx k yy k x k yy

The soil damping dependent terms can be further reduced using the real parts of Eq. (2-

15a):

1 1 1 h2
~ = + + (2-23a)
k k k x k yy

Factoring out the structural stiffness from the right-side produces:

1 1 k kh 2

~ = 1 + + (2-23b)
k k k x k yy

Eq. (2-23b) can be re-written to isolate the terms associated with the soil damping that

appear in Eq. (2-22b):

29
 k k kh 2
~ 1 = + (2-23c)
k k x k yy

Using the relation between k% k and period lengthening given in Eq. (2-18) and

the soil damping dependent terms identified on the right side of Eq. (2-23c), we re-write

Eq. (2-22b) as:

0 =
1
(T% T )
2
k
i + x +
kx
kh 2
k yy
(
yy + ( T% T ) 1 s
2

) (2-24a)

Recognition of the steps taken from Equations 2-19 through 2-21 produces the following

expression for flexible-base system damping:

1 1 1 1
0 = i + 1 + x + yy (2-24b)
( )
~
T T
2

~
T T ( ) 2 s

~
(
T Tx ) 2
( ~
T T yy )2
The foundation damping then becomes:

1 1 1
f = 1 + x + yy (2-25)

( )
~ 2 s
T T
~
T Tx ( ) 2
(
~
T T yy )2
This is the same expression given for general application in a pending national guideline

developed in the course of this research (NIST, 2012). The advantage of the last

expression over earlier formulations in the literature lies in the nature of the

dimensionless multipliers of damping terms, which can be interpreted as weight factors.

Indeed neglecting the additional contributions of radiation damping, the sum of the two

factors multiplying i and s terms is unity.

30
2.3.3.2 Derivation from Complex-Valued Impedance Expressions

Our second derivation of foundation damping retains the complex-valued form of (Eq. 2-

15a), but also utilizes the principal described in the prior section that both the real and

imaginary parts of the expression must hold. Specifically, this derivation utilizes the

equality of the real-valued terms as shown previously in Eq. (2-23). Minor re-

arrangement of the real term (Eq. 2-23) and the complex form (Eq. 2-15a) gives:

1 1 1 h2
~ = + (2-26a)
k k k x k yy

1 1 1 h2
= + (2-26b)
k% k k x k yy

The left-side of Eq. (2-26b) can be expanded to include the real and imaginary terms (per

Eq. 2-4), and then multiplied by the complex conjugate, to produce:

1 1 1 2i 0 1 2ii
= (2-27)
k% k k% k

Eq. (2-27) is re-written by isolating the complex terms on the right side and using the

relations in Eq. (2-26) as:

1 1 1 1 2
2i %0 i = + h (2-28a)
k k k x k x k yy k yy

Reducing the right side of Eq. (2-28a) for common denominators provides:

1 1 1 1 2 kx kx k yy k yy 2
+ h = + h (2-28b)
k x k x k yy k yy kx kx k yy k yy

31
The right side of Eq. (2-28b) can be re-written by expanding the complex-valued
impedance terms in the numerator to include their real and complex parts per Eq. (2-4)
as:

k x k x k yy k yy 2
+ h =
k k
x x yy yy k k
(2-28c)
k x k x (1 + 2i x ) k yy k yy (1 + 2i yy ) 2 2i x 2i yy h 2
+ h =

+
kx kx k yy k yy kx k yy

Equating the left side of Eq. (2-28a) to the right side of Eq. (2-28c) produces:

h2
2i %0 i = 2i x + yy (2-29a)
k k kx k yy

Dividing through by -2i, multiplying through by k% , and moving the i term to the right

side produces:

k% k% % 2
kh k% k% k kh 2
0 = i + x + yy = i + x + yy (2-29b)
k kx k yy k k k x k yy

Recalling the relationship between k% k and ( T% T ) in Eq. (2-18), we obtain:

2

1 1 k kh 2
0 = i + 2
x + yy (2-30)
(T% T ) (T% T ) k x
2
k yy

Eq. (2-30) matches Eq. (2-19) with the exception that the horizontal and rotational

impedance terms within the brackets have gone from real-valued to complex-valued. As

with the first derivation, we introduce hysteretic soil damping at this stage:

1 1 k kh 2
0 = i + ( + ) + ( + ) (2-31)
(T% T ) (T% T ) k x
2 2 x s yy s
k yy

32
It should be noted that the impedance functions as written in Eqs. 2-3 and 2-4 do not

include soil hysteretic damping. As we have shown, the soil damping effect can be

accounted for by adding s to the respective radiation damping ratios x and yy. This

addition should also be performed within the imaginary term of the complex-valued

impedance functions as follows:

[
k j = k j 1 + 2i j + s( )] (2-32)

As with the first derivation, we introduce fictitious vibration periods, which are

now complex-valued:

m m
Tx = 2 Tyy = 2 (2-33)
kx k yy

k k x = ( Tx T )
2
Using Eq. (2-33) and recalling T = 2 m k , term and

kh 2 k yy = ( Tyy T ) . With these substitutions, Eq. (2-30) (the case without hysteretic soil
2

damping) becomes:

1 1 1
0 = i + x + yy (2-34a)
(T% T ) (T% T ) (T% T )
2 2 2
x yy

Likewise, Eq. (2-31) (the case with hysteretic soil damping) becomes:

1 1 1
0 = i + ( x + s ) + ( + s ) (2-34b)
(T% T ) (T% Tx ) (T% Tyy )
2 2 2 yy

A general expression for foundation damping can then be written as:

33
 1 1
f = ( x + s ) + ( + s ) (2-35)
(T% Tx ) (T% Tyy )
2 2 yy

The solution of foundation damping is complex-valued; the amplitude of the complex

number is usually taken for application (Veletsos and Nair, 1975).

2.3.3.3 Comparison of Alternate Solutions for Foundation Damping

On theoretical grounds, there is no clear benefit of one of the aforementioned foundation

damping solutions over the other. The two expressions for foundation damping are given

in Eqs. (2-25) and (2-35). A practical benefit of the first solution is that it is expressed

entirely in terms of real-valued variables, whereas the second includes complex variables

that produce a complex-valued foundation damping that is relatively difficult to

understand.

In Figure 2-7, we plot foundation damping derived from the two solutions against

the ratio ( h (VsT ) ) , which is often called the wave parameter (Veletsos, 1977). The wave

parameter can be thought of as a structure-to-soil stiffness ratio because (h T ) represents

the stiffness of a structures lateral force resisting system in velocity units whereas soil

shear wave velocity Vs is related to the soil shear stiffness. In Figure 2-7, foundation

damping solutions are given for circular foundations and various structure height aspect

ratios (h/r). Translation modes of foundation vibration dominate for small height aspect

ratios (< 1), and rocking dominates for larger height aspect ratios (> 2). Veletsos and Wei

(1971) impedance functions were used for the two sets of solutions shown in Figure 2-7.

Figure 2-7 shows foundation damping results for the cases of radiation damping

only (Figure 2-7a; s=0) and combined radiation and hysteretic soil damping (Figure 2-

34
7b; s=0.1). The solution from the first approach (derived from the imaginary

component) produces higher damping, particularly for small height aspect ratios.

Otherwise, the solutions show well-known patterns of behavior, in particular:

As h/(VsT) increases, the significance of inertial SSI increases, causing increased

foundation damping;

As h/r increases, rotational modes of foundation vibration become more

dominant, which reduces foundation damping because foundation rotation

produces less radiation damping than foundation translation;

The effects of hysteretic soil damping scale with the significance of inertial SSI,

as measured for example by h/(VsT). For low h/(VsT), hysteretic damping has little

effect (zero at h/(VsT) = 0), whereas at high h/(VsT) the foundation damping is

nearly the sum of foundation damping from radiation damping and s.

35
Figure27 Comparison of foundation damping solutions based on the two approaches
from Section 2.3.3, plotted against structuretosoil stiffness ratio ( h (VsT ) ).
Conditions for the plots are a rigid, massless, circular disc supported on an
homogeneousisotropichalfspacewithhystereticsoildampinga) s=0%and
b) s=10%.=0.33.ImpedancefunctionsfromVeletsosandWei(1971)used
toderivethefoundationdamping.

2.4 Comparison to Foundation Damping Solutions in Literature

2.4.1 Bielak (1971) Solution

Bielaks (1971) derived an expression for foundation damping by identifying the

dynamic properties of a replacement fixed-based oscillator to match those of the flexible-

base oscillator shown in Figure 2-2(b). In the derivation, the foundation mass and mass

moment of inertia were taken as negligible (as above), the structural damping was

considered to be viscous, the foundation was assumed circular with a radius r, and the

36
higher-order damping terms were neglected. The viscous damping assumption for the

structure was made for the sake of computational efficiency.

Bielaks derivation is similar to the approach presented in Section 2.3.3.1 in

which the damping is derived by equating the imaginary parts of the stiffness terms in the

replacement oscillator and flexible-base system. As such, the Bielak derivation mirrors

that in Section 2.3.3.1 up to Eq. (2-19). However Bielak makes the specific assumption of

viscous structural damping, which requires modification of the structural damping ratio

as given in Eq. (2-12). With the substitution of ivis for i in Eq. (2-19), we obtain:

1 1 k kh 2
0 = vis + 2
+ yy (2-36)
(T% T ) (T% T ) k x
3 i x
k yy

Eq. (2-36) matches Eq. (3.66c) in Bielak (1971), except the nomenclature has been

adapted to be consistent with this work and periods instead of frequencies have been

used. The foundation damping ( f ) is then taken from the damping components

unrelated to structural damping as follows:

1k kh 2
f = + (2-37)
(T% T ) k x
2 x yy
k yy

System damping solutions incorporating viscous structural damping (Eq. 2-36)

are used in current force-based seismic provisions (e.g., BSSC, 2009). To the extent that

actual structural damping is not purely viscous, the expression in Eq. (2-36) can be

considered as an approximation to the true flexible-base system damping (Eq. 2-24b or 2-

37
34b). If the exponent n on the period lengthening terms is taken as 2 instead of 3, the

more general expressions given in Section 2.3.3 are recovered.

2.4.2 Veletsos and Nair (1975) Solution

Veletsos and Nair (1975) derived an expression for foundation damping by equating

amplitudes of responses between the real parts of the flexible-base system and those of

the replacement oscillator. Their derivation utilized the full complex form of stiffness

terms in equating the stiffness of the replacement oscillator to that of the flexible-base

oscillator, which is similar to the second approach described in Section 2.3.3.2 above.

Assumptions similar to those of Bielak (1971) were made regarding the properties of the

replacement oscillator (i.e., viscous damping, negligible foundation mass, etc.). The

Veletsos and Nair damping terms can be derived using a process matching that in Section

2.3.3.2 with the exception of two deviations.

The first deviation is that Veletsos and Nair (1975) used viscous structural

damping. As in the prior section, in the equation of system damping (e.g. Eq. 2-36), this

converts to 3 the exponent on the period lengthening applied to structural damping.

The second deviation concerns the incorporation of hysteretic soil damping into

the solution. The simple addition of s to radiation damping terms applied in Section

2.3.3 represents an approximate solution to mathematically complex problem of how

these damping terms interact. For example, an integral equation approach to this problem

is presented by Apsel and Luco (1987) (their result is a numerical solution, not a closed-

form equation). Veletsos and Nair (1975) use an approximate solution to this problem by

Veletsos and Verbic (1973) (the approximation is in the fitting of the dynamic impedance

38
coefficients for the case of zero soil damping with simple closed for expressions, along

with an assumption of real-valued Poissons ratio). In the Veletsos and Verbic solution,

the soil damping appears as a term in a series of equations used to derive dynamic

modifiers in the general equations for foundation impedance (j and j in Eq. 2-6 and 2-

4, respectively).

These two deviations have little impact on the damping solution given in Section

2.3.3.2, and the derived system damping, given below, is very similar to that in Eq. (2-

30):

1 1 k 2
0 = ~ vis + x + kh yy (2-38)
(T T ) 3 i
(T~ T )2 kx
k yy

The only differences between Eq. (2-38) and Eq. (2-30) are in the structural damping

terms (due to the use of viscous damping) and in the form of impedance function terms

k x and k yy . Note that Eq. (2-38) is used with or without soil damping; the effect of

including soil damping is confined to its effect on the impedance.

In Figure 2-8 we show the effect of the different approaches for incorporating

hysteretic soil damping s into the solution. As a baseline case, we show the predicted

foundation damping for s = 0.1 using the second approach given in Section 2.3.3.2 (Eq.

2-35). For the baseline calculation, radiation damping terms were computed using the

closed form expressions given in Veletsos and Verbic (1973) for the elastic medium (i.e.,

radiation damping only, or s = 0), and hysteretic damping was added to radiation

damping ratios as described in Section 2.3.3. We also show the foundation damping

39
obtained using Eq. (2-38) with the Veletsos and Verbic (1973) solution for impedance

terms on a visco-elastic medium. Similar results are found in Figure 2-8 for both

approaches of accounting for soil damping. The results slightly diverge as h (VsT )

increases, with the exception of the largest height aspect ratio shown (h r = 5) .

Figure28 Comparisonoffoundationdampingmodels(Eq.235andEq.238)accounting
for soil damping (s = 10%) differently, plotted against structuretosoil
stiffnessratio( h (VsT ) ).Conditionsfortheplotarearigid,massless,circular
discsupportedonanelastichomogeneousisotropichalfspacewithhysteretic
soil damping s = 10 %. = 0.33. Impedance functions from Veletsos and
Verbic (1973) were used to derive the foundation damping. Eq. 235 used
Veletsos and Verbic (1973) elastic impedance solution with additive soil
damping, whereas Eq. 238 uses an alternative impedance solution in which
soildampingforaviscoelasticmediumisincorporatedintothesolution.

40
2.4.3 Roesset (1980) and Wolf (1985) Solutions

Roessets (1980) presented a foundation damping solution in which the imaginary

component of the replacement oscillator stiffness is matched to that of the flexible-base

system. He also used a general (non-viscous) damping formulation for the structural

stiffness and similar assumptions to other investigators, so the derivation matches that

given in Section 2.3.3.1. Using our nomenclature, Roesset expressed the foundation

damping as:

1 1 k 2
f = 1 ~ s + ~ x + kh yy (2-39)
(T T )2 (T T )2 kx
k yy

The derivation of Wolf (1985) matches that of Roesset (1980), except that the

fictitious vibration periods given in Section 2.3.3.1 ( Tx = 2 m k x and

T yy = 2 mh 2 k yy ) are introduced to re-write the foundation damping in the form given

in Eq. (2-25).

2.4.4 Maravas et al. (2007) Solution

Recognizing the previous solutions as approximate, Maravas et al. (2007) developed an

exact solution for foundation damping of rigid circular foundations. The derivation

begins by using the complex-valued impedance functions of the replacement oscillator

represented by Eq. (2-15). Multiplying each term by the complex-conjugate, without

ignoring the higher-order damping terms results in:

41
 (1 2i 0 ) (1 2i i ) (1 2i x ) (
h 2 1 2i yy )
= + +
~
(
k 1 + 4 02 ) k (1 + 4 ) i
2
(
k x 1 + 4 x2 ) ( 2
k yy 1 + 4 yy ) (2-40)

Eq. (2-40) can be separated into real and imaginary parts as follows:

1 1 1 h2
= + +
( ) k (1 + 4 ) ( ) ( )
~ (2-41a)
k 1 + 4 02 i
2
k x 1 + 4 x2 2
k yy 1 + 4 yy

0 i x h 2 yy
= + + (2-41b)
~
(
k 1 + 4 02 ) (
k 1 + 4 i2 k x 1 + 4 x2 ) 2
(
k yy 1 + 4 yy ) ( )
~
( )
Recognizing that the term k 1 + 4 02 exists in both the real and imaginary part of the

solution, the foundation damping can be established by first rearranging the real part (Eq.

2-41a) as

~
(
k 1 + 4 02 = ) 1
(2-42a)
1 1 h2
+ +
(
k 1 + 4 i2 ) (
k x 1 + 4 x2 ) ( 2
k yy 1 + 4 yy )
~
Secondly, both sides of the imaginary part (Eq. 2-41b) are multiplied by k 1 + 4 02 to get ( )
an expression for the flexible-base system damping:

2
~
(
0 = k 1 + 4 02 )i
+
x
+
h yy

( ) ( ) ( )
(2-42b)
k 1 + 4 i2 k x 1 + 4 x2 2
k yy 1 + 4 yy

The flexible-base system damping is then formulated by inserting the right side of Eq. (2-

42a) into Eq. (2-42b) and multiplying the numerator and denominator by k:

42
 i k x kh 2 yy
+ +
0 =
(1 + 4 )
i
2
(
k x 1 + 4 x2 ) ( 2
k yy 1 + 4 yy ) (2-44)
1 k 1 kh 2 1
+ +
(1 + 4 )
i
2
(
k x 1 + 4 x2 ) ( 2
k yy 1 + 4 yy )
It should be noted that soil hysteretic damping s can be directly added into the radiation

damping terms x and yy in Eq. (2-43). Foundation damping can be determined from the

Maravas et al. (2007) system damping (Eq. 2-43) from a back-calculation of Eq. (2-13)

with n=2.

The exact solution for the period lengthening can be established as a function of

the system damping (Eq. 2-42) based on a re-expression of the real part of the derivation

(Eq. 2-41a) as:

(1 + 4 ) 1 + k (1 + 4 ) + kh (1 + 4 )
1
~ 2 2 2 2
T i i i
=
T (1 + 4 ) k (1 + 4 ) k (1 + 4 )
0
2
x x
2
yy
2
yy
(2-44)

Figure 2-9 shows the aforementioned foundation damping model and the two

approaches discussed in Section 2.3.3 compared to the structure-to-soil stiffness ratio

(h VsT ) . The cases of radiation damping only (Figure 2-9a; s=0) and combined

radiation and hysteretic soil damping (Figure 2-9b; s=0.1) are shown for rigid circular

foundations. Figure 2-9 uses Veletsos and Wei (1971) impedance functions for an elastic

halfspace, with the soil damping being added to the radiation damping terms. At small

structure-to-soil ratios (h Vs T 0.3) the exact solution of Maravas et al. (2007) tends to

match Eq. (2-35). As the ratio increases (h Vs T > 0.3) for the small height aspect ratios

43
(h r < 2) the damping trend follows Eq. (2-25) and yields more foundation damping than
Eq. (2-35).

Figure29 Comparison of the foundation damping models for the two proposed
formulationsandtheexactsolutionofMaravasetal.(2007).Allresultsshown
inthefigurearebasedontheVeletsosandWei(1971)impedancemodelfora
rigid, massless, circular disc supported on an elastic homogeneous isotropic
halfspacewith=0.33anda)s=0%andb)s=10%.

2.4.5 Relating Foundation Damping to Period Lengthening

Because both period lengthening and foundation damping are related to similar

impedance terms, Veletsos (1977) proposed to simplify the analysis of foundation

damping to a relationship between T% T and f. This simplified approach has been

retained in more recent seismic guidelines (e.g., ASCE 7, 2006 & 2010, FEMA 450, 2003

and FEMA P-750, 2009) with the solution always derived for the case of circular

foundations using the Veletsos and Nair (1975) foundation damping equations with the

44
Veletsos and Verbic (1973) approximate impedance functions. In the most recent

provisions (ASCE 7-10, 2010 and FEMA P-750, 2009), the Veletsos and Nair (1975)

solution (Eq. 24 of their paper) was implemented with a programming bug in which an a0

term was taken at the fixed-base frequency instead of the flexible-base frequency, which

caused the foundation damping to be underestimated for large values of h (VsT ) or T% T .

This has been corrected in the present formulation.

The seismic provisions include the effects of soil hysteretic damping and strain-

reduction by relating it to the peak short period design spectral response acceleration

parameter, SDS. SDS is the peak spectral response acceleration value used for earthquake

design loads. The SDS parameter is based on site conditions and the maximum considered

earthquake (MCE) ground motion. SDS is roughly proportional to the expected strain

demands and hence damping in the underlying foundation soils. Recognition of this

relationship led to its use in seismic provisions for modification of a buildings system

damping to include the approximate effects of hysteretic soil damping.

Figure 2-10 shows the graphical solutions provided in previous (Figure 2-10a)

and current (Figure 2-10b) seismic guidelines. The graphical solutions present unique

solutions for specific height aspect ratios and SDS relationships using a Poissons ratio of

0.33. Linear interpretation is recommended for determination of values between height

aspect ratios and SDS values. To error conservatively, seismic provisions limit the amount

of overall damping for analysis to not be less than 5% and not greater than 20%.

Comparing the previous provisions (Figure 2-10a) to current provisions (Figure 2-10b)

shows reduced foundation damping for low aspect ratio structures in the current

45
provisions. New relationships for foundation damping are recommended in a subsequent

section of this chapter.

Figure210 Foundation damping as a function of period lengthening from a) ASCE 705,

2006 and FEMA 450, 2003 and b) ASCE 710, 2010 and FEMA P750, 2009.
Foundation damping computed using Veletsos and Verbic (1973) impedance
functionsforrigidcircularfoundation.=0.33andD=0forplots.

2.5 Solutions for Damping of Rectangular Foundations

In this section, we apply the general solutions for foundation damping derived in Section

2.3 to foundations having a rectangular geometry in plan view. Specifically, we apply the

foundation damping in Eqs. (2-25) and (2-35), which includes terms x and yy for

radiation damping from foundation translation and rotation. The geometric shape of the

foundation enters the foundation damping formulation through those terms as given in

46
Table 2-1a from the formulation of Pais and Kausel (1988), as adapted by NIST (2012).

If the foundation is embedded, the corresponding impedance terms are given in Table 2-

1b. While the circular foundation geometry has been prevalent in prior foundation

damping solutions, it is not representative of the geometry of most building structures.

The present extension to rectangular geometries is therefore motivated in large part by a

practical need to update design guidelines and codes for a more representative, while still

simplified, representation of foundation geometry.

47
Table 2-1a Equations for Radiation Damping Ratios & Corresponding Terms for Surface Foundations (Pais and Kausel, 1988)

Translation along y-axis

Translation along x-axis
4(L B ) a0
4(L B ) a0 y =
x =
K x GB 2 x K y GB 2 y

GB L 0.65 GB L 0.65 L
Kx = 6.8 + 2.4 Ky = 6.8 + 0.8 + 1.6
2 B 2 B B

x =1 y =1
Rocking about the x-axis Rocking about the y-axis


(4 3)(L B )a0 2 a0 (4 3)(L B )3 a02 a0
xx = yy =

2 xx K yy 1.8 2 yy
K xx 0 . 4 + a0
2
2.2 + a 02
GB 3 1 + 1.75(L B 1)
3
(L B ) 3
GB
GB 3 L
2.4
GB 3 L K yy = 3.73 + 0.27
K xx = 3.2 B + 0.8 1
1 B

xx

= 1.0
(
0.55 + 0.01 L B 1 a02 ) yy

= 1.0
0.55a02

0.4 0.6 1.4 + a 2
2.4 + a0
2
3
(L B ) (L B )3
0

where = ( )
2(1 v ) (1 2v ) 2.5 , L = half-length and B = half-width (see Figure 2-5)

48
Table 2-1b Equations for Radiation Damping Ratios & Corresponding Terms for Embedded Foundations (Pais and Kausel, 1988)
Translation along x-axis Translation along y-axis
4[L B + (D B )( + L B )] a0 4[L B + (D B )(1 + L B )] a0
x = y =
(K x x ) GB 2 x ( )
K y y GB 2 y
1.34 D
0.8 1.34 D
0.8
x = 1.0 + 0.33 + y = 1.0 + 0.33 +

1 + L B B 1 + L B B
Rocking about the x-axis
D D 3 3
D L D L L 3
(4 3) + + + 3 + a 02 4 L + 1 D

B B
B B B B B 3 B B a 0
xx = +
K xx xx 1 .8 2 K xx xx 2 xx

+ a0

GB 3
1 + 1.75(L B 1) GB 3

D 1.6 D
2
xx = 1.0 + +
B 0.35 + L B B
Rocking about the x-axis
L 3 D 3
D L D
3
D L
2
L
3 3
(4 3) + + + 3 + a 02 4 L + D

B B B B B B B B 3 B B a0
yy = +
K yy yy 1 .8 K yy yy
2 yy
2
GB 3 1 + 1.75(L B 1) + a 0 GB 3

D 1.6 D 2
yy = 1.0 + +
4 B
B 0.35 + (L B )
where = ( )
2(1 v ) (1 2v ) 2.5 , D = embedment, L = half-length and B = half-width (see Figure 2-5)

49
 Figure 2-11 shows foundation damping computed using the above impedance

solutions for square foundations along with solutions based on the impedance of circular

foundations (Veletsos and Wei, 1971). In both cases, foundation damping is computed

with Eqs. (2-25) and (2-35). Foundation damping is very sensitive to height aspect ratio,

so it is important that h/B and h/r for these plots reflect similar foundation conditions. For

low aspect ratios where foundation damping is dominated by the translational mode, this

is accomplished by selecting h/B and h/r to represent common foundation areas:

h h 4
= (2-45a)
B r

At larger height aspect ratios (h r & h B > 2) , foundation damping is controlled by the

rocking mode, so we match the moment of inertia rather than foundation contact area as

follows:

h h 4 12
= (2-45b)
B r 3

Figure 2-11 shows good agreement between the circular and square impedance functions

when equivalent foundation properties are accounted for as done using Eq. (2-45).

50


Figure211 Foundation damping as a function of h (Vs T ) using Eqs. 225 and 235. Top
plotsshowcontributionsfromradiationdampingonly(s=0)forcircularand
squarefoundations.Bottomplotsincludesoilhystereticdamping(s=0.1).
=0.33andD=0forplots.VeletsosandWei(1971)impedancefunctionswere
used for the circular case. The h/B input for the Pais and Kausel (1988)
impedancefunctionsforthesquarecasewasmodifiedtorepresentequivalent
areasperEq.(245)andaredenotedbytheequivalenth/rinput.

51
 Figures 2-12 and 2-13 show the relationship between period lengthening and

foundation damping using the impedance functions for rectangular foundations in Table

2.1. These are candidate relationships for replacing the T% T vs f plots used in current

seismic guidelines (e.g., ASCE 7, 2010 and FEMA P-750, 2009: See Figure 2-10 and

Section 2.4.5). The rectangular foundation case introduces an additional design

parameter, the foundation aspect ratio (L B ) that introduces a dependency on the shaking

direction.

Figure212 FoundationdampingasafunctionofperiodlengtheningusingEqs.(225)and
(235). Foundation damping computed for square foundations using a) only
radiationdampingterms(s=0)andb)soilhystereticdamping(s=0.1).L/B=
1,=0.33andD=0usedforplots.

Figure 2-12 shows a comparison of the two approaches (Eqs. 2-25 and 2-35)

discussed in Section 2.3.3 for square foundations (L B = 1) with different height aspect

52
ratios (h B ) . Cases of radiation damping only (Figure 2-12a; s=0) and combined

radiation and hysteretic soil damping (Figure 2-12b; s=0.1) are shown. Both the

translational and rotational terms are affected by the introduction of imaginary terms in

Eq. (2-35) (relative to the terms in Eq. 2-25) by amounts ranging from 0-10% for the

translational term and 0-16% for the rotational term. These differences have negligible

influence at high height aspect ratios (e.g., h B 2 ), but substantial influence as the

height aspect ratio is reduced. In all cases, Eq. (2-25) results in more foundation damping

than does Eq. (2-35).

Figure 2-13 shows the effect of different foundation aspect ratios ( L = B ,

L B = 2 , and L B = 4 ). The solutions have been calculated for height aspect ratios of

h B = 1 and h B = 2 . The effects of changes in foundation aspect ratio (L/B) and height

aspect ratio (h/B) can be summarized as follows:

For L B > 1 , shaking in the long direction (x-direction) produces higher

foundation damping than does shaking in the short (y) direction. This occurs

because rocking will be less pronounced for shaking in the long direction, in turn

causing foundation translation to dominate, which produces more foundation

damping.

Eq. (2-25) has more sensitivity to L B than does Eq. (2-35). For example, a

structure with h B = 1 has a foundation damping differential between L = B and

L B = 2 of about 4% using Eq. (2-25) and 1% using Eq. (2-35).

53


Figure213 Foundationdampingasafunctionoffoundationaspectratios(L/B)andperiod
lengtheningusingEqs.(225)and(235).Topplotsshowfoundationdamping
forh/B=1;bottomplotsareforh/B=2.s=0,=0.33andD=0forplots.

54
2.6 Evaluation of Foundation Damping Solutions

For practical use, we recommend the use of closed form equations (such as Eq. 2-25 or 2-

35) over design charts such as those shown in Figures 2-10, 2-12 and 2-13. We consider

both Eq. (2-25) and Eq. (2-35) to be theoretically correct, yet they produce differences in

foundation damping that can be significant. We turn to available empirical data from

Stewart et al. (1999a,b), Tileylioglu et al. (2011) and recently performed forced vibration

testing to guide the selection of a single solution for foundation damping. Since empirical

data applies for realistic conditions that do not match those in idealized solutions for

foundation impedance, we begin by describing how effects such as soil nonhomogeneity

and nonlinearity are considered in the analysis of foundation damping. We then apply

both damping formulations to the buildings for which damping is known from prior

system identification work, to facilitate selection of the solution producing the best fit to

the data.

2.6.1 Idealized Solutions Adapted for Realistic Conditions

The foundation damping solutions given in Eqs. (2-25) and (2-35) are fully general, but

their typical application will be with solutions for foundation impedance that apply for

rigid foundations of either circular or rectangular geometries resting on a homogenous

half-space. This section summarizes procedures for adapting those solutions for

conditions involving non-uniform soil profiles and nonlinear soil behavior. Both of these

considerations affect the shear modulus that should be used with the impedance models

(Table 2-1).

55
 Non-uniform soil stiffnesses can be addressed by selecting an effective average

velocity across a depth interval extending from the base of the foundation to a depth of zp

below the foundation depth. The effective velocity profile is established by computing

the ratio of zp to the travel time (tt) for shear waves to travel over the depth interval of zp

to the base of the foundation. The effective velocity profile is zp /tt. The depth interval has

been related to the foundation area and moment of inertia as shown below (NIST, 2012):

Horizontal ( x & y ) : z p = BeA , BeA = A 4 = BL (2-46a)

Rocking ( xx & yy ) : z p B ei , xx : B eI = 4 0.75 I x = 4 B 3 L ,

(2-46b)
yy : B eI = 4 0.75 I y = 4 BL 3

where A is the foundation-soil contact area and Ix and Iy are the foundation moment of

inertia about the x and y axis, respectively. It should be noted that Eq. (2-46) has been

reduced for the most common foundation geometry (rectangular, with half-width B and

half-length L dimensions). Eq. 2-46 can easily be adapted for circular foundations of

radius r with the following relations:

r
Horizontal ( x & y ) : z p = BeA , BeA = A4= (2-47a)
2

r4
Rocking ( rr ) : z p = BeI , BeI = 4 0.75I rr = 3 (2-47b)
4

where subscript rr refers to the rotational mode. Shear modulus reduction factors for

different site categories and ground motion levels are provided in design guidelines (such

as IBC 2012, NIST 2012 (Table 4-1), and FEMA P-750) as shown in Table 2-2.

However, the preferred method is to complete a site-specific equivalent-linear ground

56
response analysis. Shear modulus reduction factors from such analyses or from Table 2-2

can be used to adjust the shear modulus in the static stiffness equations in Table 2-1.

Table 2-2 Values of Shear Modulus Reduction for Various Site Classes and Shaking
Amplitudes
Value of Vs/Vs0 Value of G/G0
Site Site Class
PGA PGA
Class Soil Vs,30 (m/s)
0.1 0.4 0.8 0.1 0.4 0.8
A Vs > 1500 1.00 1.00 1.00 1.00 1.00 1.00
B 1500 > Vs > 760 1.00 0.97 0.95 1.00 0.95 0.90
C 760 > Vs > 360 0.97 0.87 0.77 0.95 0.75 0.60
D 360 > Vs > 180 0.95 0.71 0.32 0.90 0.50 0.10
E 180 > Vs 0.77 0.22 * 0.60 0.05 *
F * * * * * * *
Note: Use straight line interpolation for intermediate values of PGA
* = should be evaluated from site-specific analysis

2.6.2 Comparison to Recorded Data

This section investigates how well Eqs. (2-25) and (2-35) predict the levels of foundation

damping observed empirically from field tests and case histories. Foundation damping is

not measured directly from field performance data; rather system identification analyses

are performed to evaluate fixed- and flexible-base properties of structures, from which

foundation damping is computed using Eq. (2-2).

System identification refers to the process of mathematically evaluating the

unknown properties (i.e., period and damping characteristics) of a dynamic system. Non-

parametric and parametric methods exist for evaluating the fixed-base and flexible-base

period and damping ratios. Non-parametric studies consist of simply examining ratios of

output/input motions in the frequency domain (e.g., Ljung, 1987; Pandit, 1991; Mikami

et al., 2008). Parametric system identification requires regression analyses in which the

57
properties of a parametric model of the system are evaluated for a given output/input pair

of motions (e.g., Stewart and Fenves, 1998). Whether parametric or non-parametric

analyses are to be performed, a crucial element of system identification for SSI purposes

is the selection of appropriate input-output motions to evaluate fixed-base and flexible-

base properties of the structure.

Parametric system identification of several buildings shaken by earthquakes with

varying degrees of SSI effects was performed by Stewart et al. (1999b) to evaluate their
~
first mode period and damping ratio for the fixed-base ( T , i ) and flexible-base ( T , 0 )

conditions. Based on the identified fixed-base and flexible-base damping ratios, Eq. (2-2)

is used to back-calculate the foundation damping ratio of the system. Similar analyses

were performed by Tileylioglu et al. (2011) using forced vibration and earthquake

loading of a field test structure in Garner Valley, CA (referred to as Dr. Evil in Chapter

5). To expand the data inventory, additional data sets from forced vibration testing of a

field test structure (referred to as Mini-Me in Chapter 5) have been analyzed using

parametric system identification procedures. Mini-Me was cyclically loaded with varying

degrees of force and structural stiffness at two sites (as discussed in Chapter 5) and the

response history data from both sites have been evaluated for the fixed-base and flexible

base periods and damping ratios.

To place the value of these forced vibration tests in context, Figure 2-14 shows

histograms of the available data set with respect to h/(VsT) and h/B. The data derived

from earthquake shaking of instrumented buildings (Stewart et al., 1999b) is dominated

by low values of the wave parameter (h (Vs T ) 0.1) for which inertial SSI effects are

58
small. The forced vibration tests (Tileylioglu et al., 2011; this study) occupy a range of

h (VsT ) = 0.1-0.5 with an average value of approximately 0.27. Height aspect ratios (h/B)

range from about 0-6 in the instrumented structures data set and 2-3 in the forced

vibration test data set.

Figure214 Histograms of h/(VsT) and h/B in data sets used to evaluate SSI effects from
instrumented structures. Top row corresponds to instrumented structures
shaken by earthquakes and analyzed by Stewart et al. (1999b); bottom row
correspondstoforcedvibrationtestsresultsfromthisstudyandTileyliogluet
al.,(2011).

59
 Table 2-3 presents critical attributes of the first data set, including meta data

[earthquake sources, lateral force resisting systems, foundation condition (i.e., whether on

piles, rock, soil, etc.], calculated period lengthening and foundation damping from first-

mode periods and damping ratios evaluated from parametric system identification, and

predicted period lengthening and damping based on Eqs. (2-25) and (2-35). All of these

results are as given in Stewart et al. (1999b), with the exception of foundation damping

(re-computed using Eq. 2-2 with n=2) and model predictions.

Table 2-4 summarizes attributes of the second data set associated with test

structures used for forced vibration testing (this study and Tileylioglu et al., 2011). We

list applicable metadata [test structure, site, source of shaking (earthquake or forced

vibration), type of shaker], period lengthening and foundation damping identified from

first-mode vibration properties established from system identification (details in Section

6.6, and period lengthening and foundation damping from the models presented earlier in

this chapter (period lengthening from Eq. 2-1; damping from Eqs. 2-25 and 2-35).

We denote the foundation damping observed from data analysis as f ,i , with

subscript i serving as an index for the case history under consideration. The

corresponding foundation damping from the model is denoted f ,i , and the residual (data

minus model) is f ,i f ,i . Residuals are plotted against the wave parameter for both

data sets in Figures 2-15 and 2-16 (using Eq. 2-25 and 2-35 for f ,i , respectively). With

regard to period lengthening, Ti denotes fixed-base period for case history i, T%i Ti is

60
period lengthening from data, T%i Ti is predicted period lengthening, which is identical

(
for both approaches, and the residual is T%i T%i ) T.
i

Period lengthening residuals, shown in both figures, generally range between -0.2

and 0.2. Binned means for ranges of h/(VsT) = 0-0.05, 0.05-0.1, 0.1-0.2, 0.2-0.3, and >

0.3 show a tendency of Eq. (2-1) with Table 2.1 impedance functions to underpredict the

case studies period lengthening as h/(VsT) increases. Generally the confidence intervals

on the binned means encompass zero, indicating that the misfits are not statistically

significant. Note that the under-prediction of period lengthening corresponds to computed

impedances that are too stiff. Similar misfits have been observed for impedance functions

inferred from field testing, as described in Chapter 6.

Foundation damping residuals in Figures 2-15 and 2-16 mostly fall within a range

of -5% to 5%. There are a number of significant outliers, particularly in the case of Figure

2-15, including site A17 from the Stewart et al. (1999b) data set. The residuals are more

nearly centered at zero and have fewer outliers in Figure 2-16. The differences between

Figures 2-15 and 2-16 are principally between h/(VsT) > 0.1, where SSI effects are most

significant. The sites within that range of the wave parameter show a tendency for Eq. (2-

25) to overpredict damping. For site A46, which is at the upper end of the h/(VsT) range

(1.54), Eq. (2-25) overpredicts foundation damping, while Eq. (2-35) underpredicts.

These results do not necessarily indicate that the formulation in Eq. (2-35) is

correct and Eq. (2-25) is incorrect, because there are likely compensating errors

related to the impedance functions. In particular, halfspace models for impedance tend to

61
over-predict translational radiation damping and under-predict rotational radiation

damping (Tileylioglu et al., 2011 and this study). If impedance functions were developed

that removed these sources of bias in the predicted radiation damping, we could very well

find that Eq. (2-25) is unbiased and (2-35) is biased. All that can be said at present is that

given the limitations of halfspace-based impedance functions, Eq. (2-35) provides a

better fit to the available empirical data than does Eq. (2-25).

62
Table 2-3 Attributes of data set for instrumented structures with earthquake recordings (Stewart et al., 1999b) showing period
lengthening and foundation damping from system identification and models.
Traverse(ydirection) Longitudinal(xdirection)

d
Confidence
b
Lateral

Rock/Soil
SystemID Eq.225 Eq.235 SystemID Eq.225 Eq.235

a
Piles
D>0
a a a a
Site# EQK Force h/(VsT) h/(VsT)
b ~ a f ,i c ~ f ,i ~ f ,i ~ a f ,i c ~ f ,i ~ f ,i
Resistance T T Ti Ti Ti Ti T T Ti Ti Ti Ti
A1 PT SW N N S A 0.29 1.57 13.1 1.39 21.4 1.39 13.8 0.2 1.09 8.2 1.14 14.7 1.14 9.1
PT 0.08 1.04 10.0 1.06 8.6 1.06 4.7 0.1 1.08 24.0 1.09 15.7 1.09 7.3
A2 SW N N S L
PTA 0.08 1.05 3.6 1.06 8.6 1.06 4.7 0.1 1.03 9.0 1.09 15.7 1.09 7.3
A4 LP CF Y N S A 0.2 1.02 5.9 1.19 2.9 1.19 2.8 0.18 1.00 6.9 1.15 2.2 1.15 2.2
A5 LP DWF Y N R A 0.03 1.04 0.6 1.01 0.0 1.01 0.0 0.04 1.03 0.0 1.00 0.0 1.00 0.0
A7 LP SW N N S A 0.08 1.03 9.0 1.04 3.4 1.04 2.4
A8 LP SW Y N R A 0.09 1.16 3.2 1.04 2.0 1.04 1.7 0.06 1.00 1.9 1.03 1.6 1.03 1.4
A9 CGA DWF N Y S L 0.09 1.00 3.0 1.02 1.0 1.02 0.8
A10 LP DWF N Y S A 0.16 1.06 5.7 1.11 11.4 1.11 6.5 0.16 1.03 13.1 1.08 11.3 1.08 6.2
A11 LP SF N N R A 0.02 1.00 0.0 1.00 0.0 1.00 0.0 0.02 1.00 0.0 1.00 0.0 1.00 0.0
A12 IMP DWF Y N S A 0.23 1.47 5.2 1.23 9.4 1.23 7.4 0.09 1.00 0.0 1.03 1.0 1.03 0.9
A13 LD DWF Y Y S A 0.12 1.05 3.5 1.05 2.2 1.05 1.7 0.13 1.03 0.6 1.04 2.5 1.04 1.9
A14 WT SW Y N S A 0.14 1.00 1.0 1.10 6.8 1.10 4.9
A15 NR SF Y N S A 0.06 1.06 3.3 1.02 0.9 1.02 0.8 0.06 1.02 0.0 1.02 1.0 1.02 0.9
A16 NR SF N N S L 0.18 1.28 0.0 1.15 0.8 1.15 0.8 0.2 1.34 1.3 1.18 1.0 1.18 1.0
A17 NR SW N N R A 0.14 1.17 10.8 1.18 34.6 1.18 13.2 0.12 1.09 5.4 1.14 24.5 1.14 10.3
A18 WT SF Y N S A 0.08 1.00 0.0 1.02 0.4 1.02 0.4
A20 NR SW Y Y R A 0.17 1.13 0.0 1.09 3.2 1.09 2.6 0.16 1.05 0.8 1.08 2.7 1.08 2.2
SM 0.03 1.01 0.0 1.01 0.2 1.01 0.2 0.03 1.00 3.8 1.01 0.2 1.01 0.2
A21 LD BI N N R L 0.02 1.00 0.0 1.00 0.1 1.00 0.1 0.02 1.00 0.0 1.00 0.1 1.00 0.1
NR 0.02 1.00 0.0 1.00 0.1 1.00 0.1 0.03 1 0.0 1.01 0.2 1.01 0.2
A23 NR SF N Y S L 0.11 1.08 0.0 1.04 0.6 1.04 0.6
A24 NR SW Y N S A 0.09 1.04 1.0 1.07 8.5 1.07 4.8 0.23 1.60 3.6 1.25 12.4 1.25 8.8
LD 0.05 1.00 0.0 1.01 0.2 1.01 0.2 0.05 1.00 0.0 1.01 0.2 1.01 0.2
A25 BI N N R A
NR 0.05 1.00 0.0 1.01 0.3 1.01 0.3 0.05 1.01 0.0 1.01 0.3 1.01 0.3
A26 NR DWF N Y S L 0.19 1.04 0.0 1.12 3.0 1.12 2.7 0.11 1.00 0.0 1.03 0.7 1.03 0.6
LD 0.05 1.00 0.9 1.01 0.1 1.01 0.1 0.05 1.00 0.3 1.01 0.1 1.01 0.1
A27 SF N N R A
NR 0.05 1.00 5.7 1.01 0.1 1.01 0.1 0.05 1.00 6.8 1.01 0.1 1.01 0.1
WT 0.06 1.01 3.8 1.02 0.2 1.02 0.2
A29 CF Y N S L
NR 0.05 1.02 3.5 1.02 0.2 1.02 0.2 0.15 1.06 0.4 1.04 1.1 1.04 1.0
A30 NR SF Y Y S A 0.09 1.08 1.3 1.04 3.2 1.04 2.3
LD 0.13 1.19 2.7 1.10 0.9 1.10 0.8
A31 SW N N S A
NR 0.13 1.16 0.0 1.10 1.0 1.10 1.0 0.14 1.14 0.5 1.06 1.2 1.06 1.1
A32 WT SF Y Y S A 0.06 1.00 1.1 1.01 0.4 1.01 0.4
WT 0.07 1.00 0.2 1.02 1.1 1.02 0.9 0.07 1.00 0.5 1.02 1.1 1.02 0.9
A33 SF Y Y S A
NR 0.06 1.00 1.4 1.02 0.7 1.02 0.6 0.07 1.00 1.2 1.02 1.0 1.02 0.8
A34 NR SW N N S A 0.12 1.66 9.8 1.09 6.7 1.09 4.5 0.09 1.22 9.1 1.04 4.0 1.04 2.9
*Table continued on next page. 1 of 2
63
 Traverse(ydirection) Longitudinal(xdirection)

d
Confidence
b
Lateral

Rock/Soil
SystemID Eq.225 Eq.235 SystemID Eq.225 Eq.235

a
Piles
D>0
a a a a
Site# EQK Force h/(VsT) h/(VsT)
b ~ a
f ,i c ~ f ,i ~ f ,i ~ a
f ,i c ~ f ,i ~ f ,i
Resistance T T Ti Ti Ti Ti T T Ti Ti Ti Ti
WT 0.09 1.02 3.4 1.03 2.5 1.03 1.8 0.09 1.02 0.0 1.03 2.5 1.03 1.8
A35 CF N Y S A
UP 0.06 1.01 4.4 1.03 1.9 1.03 1.4 0.06 1 0.0 1.03 1.9 1.03 1.4
A36 LD CF N Y S L 0.04 1.17 0.0 1.01 0.1 1.01 0.1 0.05 1.39 4.9 1.01 0.1 1.01 0.1
RD 0.06 1.03 0.8 1.03 1.6 1.03 1.2 0.06 1.00 1.4 1.02 1.6 1.02 1.3
WT 0.06 1.02 0.0 1.03 1.6 1.03 1.3 0.07 1.01 2.4 1.02 1.1 1.02 0.9
A37 UP BI N Y S A 0.07 1.00 0.0 1.02 1.2 1.02 1.0 0.07 1.00 0.0 1.02 1.2 1.02 1.0
LD 0.06 1.01 0.0 1.02 0.8 1.02 0.7 0.06 1.02 0.5 1.01 0.8 1.01 0.6
NR 0.07 1.02 0.0 1.02 1.1 1.02 0.9 0.06 1.02 0.0 1.01 0.7 1.01 0.6
A38 LD SF N N S L 0.06 1.07 0.0 1.02 1.3 1.02 1.1 0.06 1.03 0.0 1.02 1.4 1.02 1.1
A39 NR SW N Y S L 0.06 1.00 3.9 1.01 0.6 1.01 0.5 0.08 1.00 1.3 1.02 1.4 1.02 1.1
A40 LD SF Y N S A 0.04 1.00 0.0 1.01 0.1 1.01 0.1 0.04 1.00 1.4 1.01 0.1 1.01 0.1
A41 NR SF Y N S A 0.07 1.01 0.0 1.03 2.1 1.03 1.6 0.04 1.02 0.7 1.01 0.4 1.01 0.4
A42 NR SF N N R L 0.04 1.02 4.6 1.02 0.0 1.02 0.0 0.04 1.00 3.1 1.01 0.0 1.01 0.0
LD 0.07 1.02 0.0 1.02 0.4 1.02 0.4 0.06 1.04 0.0 1.02 0.6 1.02 0.5
A43 BI Y Y S A
NR 0.06 1.03 0.0 1.02 0.7 1.02 0.6 0.09 1.03 0.9 1.02 0.9 1.02 0.8
WT 0.16 1.10 0.9 1.12 10.9 1.12 7.4 0.17 1.17 3.2 1.14 12.4 1.14 8.2
A44 SW N N S A
NR 0.15 1.04 1.0 1.11 10.6 1.11 6.9 0.21 1.29 12.9 1.20 22.7 1.20 12.8
A45 NR SW Y N S A 0.15 1.34 0.8 1.12 1.7 1.12 1.5
A46 LO7 N Y S A 1.45 4.14 30.4 3.20 33.3 3.20 25.1 1.54 4.01 30.8 3.36 33.6 3.36 25.1
B1 LP SW N N S A 0.17 1.06 0.0 1.04 0.5 1.04 0.5
B2 LP SW N N S A 0.07 1.13 3.6 1.02 0.4 1.02 0.4
B3 LP SF Y Y S A 0.17 1.03 0.0 1.19 1.4 1.19 1.4
B4 LP SF N Y S A
B5 LP SW Y N S A 0.27 1.64 0.0 1.35 7.6 1.35 6.2
B6 LP SF N Y S A 0.07 1.01 0.0 1.02 0.2 1.02 0.2 0.07 1.01 0.6 1.02 0.2 1.02 0.2
B7 LP DWF Y N R A 0.1 1.17 0.0 1.05 0.3 1.05 0.3
B9 NR SF N N S A
B10 NR DWF N Y S A 0.08 1.00 2.5 1.03 0.4 1.03 0.4 0.1 1.03 0.0 1.03 0.6 1.03 0.6
LD 0.09 1.13 1.4 1.12 0.3 1.12 0.3
B11 SW Y N R A
NR 0.09 1.17 2.1 1.12 0.4 1.12 0.4
B12 NR DWF N Y R A 0.12 1.06 1.3 1.05 0.3 1.05 0.3
B13 NR SF N Y R A 0.06 1.02 0.0 1.02 0.1 1.02 0.1
B14 NR SW N N S A 0.12 1.14 2.2 1.15 1.3 1.15 1.2
*EQK:CGA:ColingaAftershock,IMP:ImpericalValley,LD:Landers,LO7:LotungEvent7,NR:Northridge,PTA:PetoliaAftershock,PT:Petrolia
SM:SierraMadre,UP:Upland,WT:Whittier
*Lateralforceresistingsystem:BI=BaseIsolation,CF=Concreteframe,DWF=dualwall/framesystem,SF=steelframe,SW=shaerwall,masonryorconc.
a b c d
DatafromStewartetal.1999; DatafromStewartandStewart1997, DerivedfromStewartetal.1999data, A=acceptableconfidence&L=lowconfidenceinStewartetal.'99results

2 of 2

64
Table 2-4 Attributes of data set for field test structures subject to forced vibration testing and earthquake shaking, including period
lengthening and foundation damping from system identification and models.
ObservedData PredictedData
Direction SystemID Eq.225 Eq.235
Structure Structure Shake
s
b d e
a DataSource of c h/B L/B h/(VsT)
&Site Condition Source ~ a ~ ~
Shaking T T f ,i c Ti Ti f ,i Ti Ti f ,i

TEA(2011) Unbraced NA EQML=4.2 1.20 0.15 1.15 0.9 1.12 0.8 1.12 0.8
DE
TEA(2011) Unbraced NA EQMw=5.4 1.20 0.15 1.15 3.4 1.12 0.8 1.12 0.8
@ 2.15 1.00
GVDA TEA(2011) Unbraced NA LM 1.20 0.14 1.09 0.4 1.10 0.7 1.10 0.7
TEA(2011) Braced NA LM 1.20 0.28 1.29 0.0 1.38 3.0 1.38 2.8
Exp.3.3.1a Unbraced X AA 1.20 0.17 1.13 2.4 1.10 1.5 1.10 1.4
Exp.3.4.9b Unbraced X MME10% 9.80 0.26 1.28 4.3 1.22 7.1 1.22 6.2
Exp.3.4.13a Unbraced X MME50% 13.70 0.29 1.25 6.4 1.27 10.2 1.27 8.6
Exp.3.4.15a Unbraced X MME100% Errorintest
Exp.3.3.5a Unbraced Y AA 1.20 0.17 1.31 2.4 1.19 1.2 1.19 1.1
Exp.3.4.3a Unbraced Y MME10% 7.00 0.23 1.61 8.2 1.33 4.5 1.33 4.3
Exp.3.4.17a Unbraced Y MME50% 12.10 0.27 1.76 8.7 1.43 8.2 1.43 7.5
MM
Exp.3.4.19a Unbraced Y MME100% 12.90 0.26 1.78 11.1 1.41 8.3 1.41 7.6
@ 2.68 2.00
Exp.3.1.1a Braced X AA 1.20 0.24 1.12 0.8 1.19 3.5 1.19 3.2
GVDA
Exp.3.2.6a Braced X MME10% 9.00 0.33 1.33 10.0 1.34 10.5 1.34 9.1
Exp.3.2.12a Braced X MME50% 13.90 0.45 1.63 9.1 1.59 19.5 1.59 15.8
Exp.3.2.14 Braced X MME100% 14.90 0.31 1.12 13.5 1.31 11.9 1.31 9.9
Exp.3.1.6a Braced Y AA 1.20 0.22 1.38 2.5 1.30 1.9 1.30 1.8
Exp.3.2.1a Braced Y MME10% 7.10 0.31 1.65 5.3 1.55 6.6 1.55 6.1
Exp.3.2.11 Braced Y MME50% 12.40 0.27 1.63 7.8 1.43 8.4 1.43 7.7
Exp.3.2.15 Braced Y MME100% 12.40 0.34 2.27 14.8 1.64 10.6 1.64 9.6
Exp.2.1.9a Unbraced X AA 2.00 0.33 1.44 8.8 1.3 7.1 1.30 6.4
MM
Exp.2.1.7a Unbraced Y AA 2.00 0.29 1.53 6.0 1.42 3.3 1.42 3.0
@ 2.68 2.00
Exp.2.1.17a Braced X AA 2.00 0.41 1.53 11.8 1.45 10.6 1.45 9.4
WLA
Exp.2.1.15a Braced Y AA 2.00 0.39 1.81 6.7 1.69 4.8 1.69 4.4
a
GVDAGarnerValleyDownholeArraySite;WLAWildlifeLiquefactionArraySite;DEDr.EvilStructure;MMMiniMeStructure(SeeChap.5)
b
TEA=Tileyliogluetal(2011);Datawith"Exp"nameswerecompiledaspartofaUCLAfieldtestingseries,informationcanbefoundathttps://nees.org/warehouse/project/637/andChapter5.
c
AA=AtomAntshaker;MME#%=MightyMouseshakerwith#%eccentricityforce;LMLinearmassshaker,similartoAA
d
Centerofmassheightoftheroofslab.
e 1/2
BasedonMenq(2003)modelusing65%ofapproximatedstrainvalues(fondationvelocitypersoilshearwaveveolicty)atflexiblebasefrequency,Note:Vs /Vs0=(Gs /Gs0) andWLAsiteapproximateds =0.2

65


Figure215 Period lengthening and foundation damping residuals from combined data
sets.FoundationdampingresidualsfrommodelinEq.(225).

66


Figure216 Period lengthening and foundation damping residuals from combined data
sets.FoundationdampingresidualsfrommodelinEq.(235).

2.7 Conclusions and Recommendations

Closed-form solutions to determine the period lengthening and foundation damping have

been presented. Eqs. (2-25) and (2-35) are both general expressions for foundation

damping that can incorporate the Pais and Kausel (1988) closed-form impedance function

solutions (Table 2-1) for either surface or embedded rectangular shaped foundations.

Both models allow for a direct calculation of the foundation damping of rectangular

foundations and obviate the use of traditional models for circular footings. The models

can be adapted to accommodate site-specific soil nonhomogeneity and design-levels of

shear strain using approximate methods from the literature (NIST, 2012).

67
 While either solution (Eqs. 2-25 and 2-35) is a valid formulation of foundation

damping on a conceptual level, the levels of damping from Eq. (2-25) are higher than

those from Eq. (2-35) for foundations of modest height aspect ratios. The lower damping

from Eq. (2-35) is generally more consistent with field case histories either re-analyzed or

presented for the first time in this chapter. Because of known biases in radiation damping

formulations for translation and rocking used in the foundation damping equations, it

cannot be concluded on the basis of these comparisons that one equation is fundamentally

better than the other. However, given the limitations of current radiation damping models

for halfspace representations of the foundation medium, Eq. (2-35) appears to be the

preferred formulation at the present time.

68
3 KINEMATICSOILSTRUCTURE
INTERACTIONEFFECTSFROM
BUILDINGANDFREEFIELDSEISMIC
ARRAYSINJAPAN

This Chapter presents work that was completed in collaboration with our Japanese

colleagues for a joint conference proceeding of the 9th International Conference on Urban

Earthquake Engineering and 4th Asia Conference on Earthquake Engineering that will be

held at the Tokyo Institute of Technology in Tokyo Japan on March 6-8, 2012. The

coauthors of this work include Atsushi Mikami of the University of Tokushima in Japan,

Toshihide Kashima of the International Institute of Seismology and Earthquake

Engineering, Building Research Institute in Japan and Jonathan P. Stewart of the

University of California, Los Angeles.

3.1 Introduction

Foundation-level and free-field seismic ground motions are identical only for vertically

propagating incident waves and non-embedded foundations. In reality, incident waves are

neither vertically propagating nor coherent, which gives rise to spatially variable ground

69
motions even on sites with relatively uniform site conditions. In the presence of such

incident wave fields, motions on foundations will be reduced in amplitude relative to

free-field motions, an effect termed base slab averaging. Base slab averaging is one type

of kinematic soil-structure interaction effect. Other kinematic effects result from

foundation embedment below the ground surface (e.g., buildings with basements) and the

presence of pile foundations.

The spatial variability of seismic ground motions in the free-field (i.e., away from

building foundations), has been examined by a number of investigators (e.g.,

Abrahamson, 1991; Zerva and Zervas, 2002) who have derived empirical functions that

express the variability of phase angle in terms of a coherency function. Those functions

were derived largely using dense array records from Lotung, Taiwan, although they have

since been validated and adjusted using data from other arrays (Ancheta et al., 2011).

Recently, Abrahamsons ground motion coherency model was implemented into the

Computer Program SASSI (a System for Analysis of Soil-Structure Interaction; Ostadan,

2005). SASSI computes the relative motion of the foundation to the free-field (input

motion) in terms of a transfer function expressing the frequency-dependent ratio of

foundation motion amplitude to free-field motion amplitude. The result is one solution to

the kinematic interaction problem, including base slab averaging.

Another approach that has been utilized to predict foundation/free-field transfer

functions from base slab averaging is based on a theoretical solution by Veletsos and co-

workers (1989, 1997) coupled with empirical calibration of an incoherence parameter by

Kim and Stewart (2003). This approach is inherently calibrated for California sites, but

70
the degree to which its application is appropriate for building types not represented in the

database and regions other than California remains unknown. This approach has been

implemented for engineering application, in combination with a simple transfer function

for embedment effects, in two design standards: ASCE-41 (2013) and NIST (2012).

In this chapter, we apply the above two models for predicting transfer functions to

a subset of the available data from Japan for several particularly interesting structures.

Both models are applied under the assumption of vertically propagating, incoherent

incident wave fields, with the implicit understanding that bias could result from this

assumption under some circumstances. The predicted transfer functions are compared to

each other and to transfer functions derived from the data.

3.2 Structural-Ground Arrays in Japan

As part of an ongoing project, we compiled a list of instrumented structures in Japan with

neighboring ground stations, which can be used for the evaluation of foundation/free-

field ground motion variations of the type described in the Introduction. This list of

structures is given in Table 3-1.

71
 Table 3-1 List of currently compiled instrumented structure and ground arrays in Japan.
Two base
# Stories
Station vertical # Basement Foundation Instrumented Site # Earthquakes
Location Array owner Lat/Long
code sensors
above
levels type stories condition recorded
ground
(Y/N)
Sendai, Miyagi Tohoku Institute of Technology 3814'44"N, 14051'12"E J1 no 4 0 pile GL,4,FF 100
Hachinohe, Aomori Hachinone National College of Tech. 4029'32"N, 14126'57"E J5 no 0 pile GL,FF
Toda Corp., Tsukuba? Toda Corporation 3606'37"N, 14004'36"E J6 6 0 pile GL,6,FF 13
Tokushima Univ. The University of Tokushima 3404'39"N, 13433'46"E J7 no 6 0 pile GL,6,FF sand 2
Nihon Univ. Bldg. #14 Nihon University 3543'31"N, 14003'28"E 5 1 pile GL,4,RF,FF many
Nagoya Univ. Nagoya University 3509'20"N, 13658'01"E
Hachinohe Institute of Technology 4028'44"N, 14133'36"E
3500'45"N, 13546'06"E 7 1 B1,7,4,FF
Nikken Sekkei 3541'59"N, 13945'04"E
Shimizu Corporation 3539'46"N, 13947'52"E
Kanagawa University 3529'07"N, 13937'13"E 8 2 B2,B1,3,6,8
NPS
NPS Kashiwazaki Tokyo Electric Power Company 3725'17"N, 13835'44"E
NPS Fukushima Tokyo Electric Power Company 3725'14"N, 14101'59"E
NPS Onagawa Tohoku Electric Power Company 3824'04"N, 14129'59"E
BRI
Takamatsu, Kagawa BRI 3420'20"N, 13402'48"E J2, TKM no 8 1 pile? B1,RF,FF sand 23
Fukuoka, Fukuoka BRI 3335'16"N, 13025'27"E J3 no 10 1 pile B1,10,FF sand 14
Nagoya, Aichi BRI 3510'51"N, 13654'11"E J4, FKO no 12 2 pile B2,GL,12 sand 55
Miyako BRI 3938'29"N, 14157'25"E MYK 7 0 GL,7,FF 224
Hachinohe BRI 4030'42"N, 14129'20"E HCN2 10 1 B1, GL,10,FF 467
Yachiyo BRI 3543'20"N, 14005'59"E YCY 6 1 B1,GL,7 23
Nippon Inst. Tech. BRI 3601'38"N, 13942'41"E NIT 6 1 GL,6,FF 327
Misato City Hall BRI 3549'48"N, 13952'20"E MST no 7 1 GL,7,FF 14
Funabashi BRI 3542'26"N, 14000'20"E FNB 8 0 GL,8,FF 37
Chiba BRI 3536'24"N, 14007'22"E CHB 8 1 B1,GL,8 23
Toda City Hall BRI 3549'03"N, 13940'40"E TDS 8 1 B1,GL,8 27
The National Museum of
BRI 3542'55"N, 13946'32"E NMW 3 1 B1,GL,4,FF 264
Western Art
Univ. of Tokyo, Bldg.11 BRI 3542'49"N, 13945'34"E UTK 9 0 GL,7,FF 49
Marine Sci. and Tech. BRI 3537'42"N, 13944'53"E TUF 7 0 GL,7,FF 37
College of MLIT BRI 3543'08"N, 13928'53"E KDI 3 0 GL,3,FF 69
Yamanashi Pref. Hall BRI 3539'50"N, 13834'06"E YMN 8 1 B1,GL,8,FF 15
Kushiro BRI 4259'11"N, 14422'40"E KGC 9 1 B1,GL,9,FF 253
Olumpic BRI 3540'36"N, 13941'39"E YYG 4 1 B1,GL,4 16
BRI 3607'58"N, 14004'34"E NCTD 7 0 GL,8,FF 2
BRI 3540'42"N, 13944'39"E NDLM 5 4 GL,17 48
BRI 3540'44"N, 13944'38"E NDLA 4 8 B8,B4,GL,4 37
BRI 3540'41"N, 13944'42"E NDLG 242
B1,GL,2,5,8,FF,
BRI 3607'55"N, 14004'24"E ANX 8 1 1142
ground array
B# = Basement floor level #, BRI = Building Research Institute, FF = Free-field, GL = Ground level, Ground array = sensors at depth in soil, NPS = Nuclear power station, and RF = Roof level

72
 The Building Research Institute (BRI) in Japan owns most of the structural-

ground arrays and distributes seismic data at the web site: http://smo.kenken.go.jp/. Most

of the BRI instrumented buildings range from 3 to 10 stories in height above ground.

Foundation types are not provided on the site. Base rocking cannot be directly estimated

for most of these buildings since BRI sensor array configurations typically have only one

vertical accelerometer at its lowest level.

The Electric Power Company in Japan is another major owner of structural-

ground arrays from in nuclear power stations. When released for public use, the data is

distributed by the Japan Association for Earthquake Engineering (JAEE) at their web site:

http://www.jaee.gr.jp/stack/sta05.html.

From the list in Table 3-1, we select three buildings for consideration in this

manuscript: Nagoya office building (hereafter referred to as Nagoya), Onagawa nuclear

power plant (Onagawa), and Sendai university building (Sendai).

The Sendai site consists of a four story building above ground and contains nine

accelerometers (including three free-field), as shown in Figure 3-1. Lateral loads are

resisted by reinforced concrete frames and shear walls in the longitudinal and transverse

directions, respectively. The foundation consists of end-bearing concrete piles connected

by grade beams. Figure 3-1 shows the underlying soil stratigraphy and geophysical data

that were determined from boring operations which included suspension logging at the

site (OYO Corporation, 2007). The Sendai site was investigated previously by Mikami et

al. (2006) using data from the 2003 Off Miyagi earthquake. This data is considered again

here along with recordings from the recent 2011 Tohoku earthquake, which produced

73
much stronger ground motions. The site peak ground acceleration (PGA) for the Off

Miyagi and Tohoku earthquakes are 0.232g and 0.813g, respectively. The purpose for

selecting this site was to examine possible differences in the kinematic response for the

variable strengths of shaking.

The Onagawa site consists of a five story reactor above ground with two sub-

terranean levels (embedment of 16 m) and contains 22 accelerometers (including three

free-field), as shown in Figure 3-2. The foundation consists of a 3.5 m thick concrete slab

with dimensions of 55.3 by 51.8 m. Figure 3-2 shows the underlying soil lithology and

geotechnical data. The Onagawa structure was selected because it has the only available

array that enables evaluation of base rocking. The site is investigated for the 2011

Tohoku earthquake that had a PGA of 0.533g at the site.

The Nagoya site consists of a 10 story building above ground with two basement

levels (embedment of 12.83 m) and contains 9 accelerometers (including three free-field),

as shown in Figure 3-3. The foundation consists of a 2.53 m thick concrete slab with

dimensions of 79.6 by 23.2 m overlying reinforced concrete piles. Figure 3-3 shows the

underlying soil lithology and geotechnical data that were determined from boring

operations which included suspension logging 440 m west of the site (CRBOMLIT,

2003). The Nagoya structure was selected because it has a large footprint area, which is

of interest because its dimensions are near the limit of the calibration range for the Kim

and Stewart (2003) semi-empirical model. The site was investigated for the 2004 Off Kii

Pensinsula earthquake that had a PGA of 0.014g. To our knowledge, neither the Onagawa

nor the Nagoya structures have been examined in prior studies.

74


Figure31. BuildinglayoutandsiteconditionsatSendaiUniversitysite(J1).Datasource:
OYOCorporation,2007.

75


Figure32. BuildinglayoutandsiteconditionsatOnagawaNuclearPowerPlant,Reactor
1.Datasources:TohokuElectricPowerCompany,2009and2011.

76


Figure33. Building layout and site conditions at Nagoya Office Building (J4). Data
sources:CRBOMLIT,2003and2005.

77
3.3 Models for Prediction of Foundation to Free-Field Transfer Functions

3.3.1 Spatial Coherency Model Implemented in SASSI

The spatial variation of phase of strong ground motion is quantified by coherency. Using

recordings from dense arrays in Lotung, Taiwan Abrahamson (1991) derived empirical

functions that describe coherency as a function of separation distance and frequency.

Coherency is unity at zero frequency and reduces strongly with increasing frequency and

relatively weakly with increasing distance. Coherency less than unity results both from

deterministic phase lag and relatively random (stochastic) wave scattering effects.

Abrahamson (1991) present a model for lagged coherency, which represents the

stochastic component only (wave passage removed). The model is considered applicable

to frequencies greater than 1 Hz and for separation distances of 6 to 85 meters

(Abrahamson, 1991). That coherency model has been implemented into SASSI by

Ostadan (2005) and ACS-SASSI by Ghiocel (2006).

SASSI was originally developed at the University of California, Berkeley

(Lysmer et al., 1981, 1999). SASSI utilizes the substructuring approach in which the

linear SSI problem is divided into sub problems based on the principle of superposition;

thus the analysis is performed using linear material properties. Soil is assumed to consist

of horizontal layers overlying either a rigid base or an elastic half-space. The structure

and foundation are modeled by finite elements. Foundations are modeled as rigid,

massless slabs to exclude inertial effects. Piles are modeled as hollow elements using

representative stiffness of prestressed high-strength concrete.

78
3.3.2 Semi-Empirical Approach

Veletsos and co-workers (1989, 1997) developed useful models for theoretical base slab

averaging that combine an analytical representation of the spatial variation of ground

motion with rigorous treatment of foundation-soil contact.

Kim and Stewart (2003) calibrated Veletsos analysis procedure against observed

foundation/free-field ground motion variations as quantified by frequency dependent

transfer functions. Two types of transfer functions can be obtained from well

instrumented structures as follows:

uf L
H u ( ) = H = (3-1)
ug ug
where uf denotes foundation translation, ug ground motion translation in the same

direction, denotes kinematic rotation about an axis normal to the direction of uf and ug,

and L is the foundation half dimension in the same direction as uf and ug. The Kim and

Stewart calibration considered the horizontal translation transfer function only (Hu), and

resulted in apparent values (denoted a) for each structure/data set combination. Those

a values reflect not only incoherence effects, but necessarily also include average

foundation flexibility and wave inclination effects for the calibration data set.

The data set considered by Kim and Stewart (2003) consisted of buildings with

mat, footing and grade beam, and grade beam and friction pile foundations, generally

with base dimensions in the range of BeA =15-40 m (where BeA is the square root of

foundation area divided by four). Although the Veletsos models strictly apply to rigid

foundations, the semi-empirical model applies to the more realistic foundation conditions

79
present in the calibration data set, which consist principally of shallow foundations that

are inter-connected (i.e., continuous mats or footings interconnected with grade beams).

Errors could occur when the model is applied to conditions beyond the calibration data

set. In particular, the effects of incoherence in the Veletsos models is taken as

proportional to wavelength, thus implying strong scaling with frequency and distance. As

mentioned previously, array data indicates the distance scaling is much weaker than the

frequency scaling (Abrahamson et al., 1991; Ancheta et al., 2011), so the model would be

expected to over-predict the effects of incoherence (under-predict Hu) for very large

foundations (opposite for small foundations). Even within the parameter space of the

calibration data set, it should be recognized that the empirical model fits the data set in an

average sense but would not be expected to match any particular observation. Example

applications showing fits to data are given in subsequently in the paper.

Two of the three buildings considered in this research have foundation

embedment (Nagoya and Onagawa). We utilize a model originally presented by Kausel et

al. (1978) in combination with the Kim and Stewart model for these buildings. As

presented by NIST (2012), the model is given as follows:

D D
H u = cos , < 1.1 (3-2a)
V V
s , d s , d

D
H u = 0.45, > 1 .1 (3-2b)
V s ,d
D D
H ( ) = 0.261 cos , < (3-3a)
V V 2
s ,d s , d

D
H ( ) = 0.26, > (3-3b)
V s ,d 2

80
where D = embedment depth and Vs,d is a time-averaged shear wave velocity over the

embedment depth of the foundation.

3.4 Evaluation and Comparison of Transfer Functions

3.4.1 Transfer Function Calculation from Data

Transfer functions are evaluated from the recordings using procedures described in

Mikami et al. (2008). In particular, frequency domain smoothing is applied to spectral

density functions for the input (denominator in Eq. 3-1, denoted y) and output

(numerator in Eq. 3-1, denoted x) as follows:

S xx
H ( ) = (3-4)
S yy

where Sxx and Syy are auto-spectral density functions for the input and output,

respectively. The smoothing is applied using a 7-point Hamming window, which

provides an equivalent frequency bandwidth of Be = 0.207 Hz. In addition, the coherence

(square of coherency) of the data is calculated as:

S xy ( )
2

( ) =
2
(3-5)
S xx ( )S yy ( )

where Sxy is the cross spectral density function. The coherence is used the judge the

effects of noise in the data. Frequency ranges in the transfer function that are dominated

by noise will have low coherence. The average coherence of pairs of white noise signals

for our frequency bandwidth is 0.25 0.04 (Mikami et al., 2008).

81
3.4.2 Application of Models

The models describes in Section 3 are applied to the conditions at the three building sites.

Equivalent linear shear wave velocity profiles were developed using deconvolution

analysis with the free-field ground motions and velocity profiles shown in Figures 3-1 to

3-3. Nonlinear properties were developed based on the soil conditions using the modulus

reduction and damping models from Menq (2003) for sands and Darendeli (2001) for

other soil and weathered rock materials

The properties developed through the above process are the direct input for the

SASSI analysis, which is a linear analysis with no iteration on strain-dependent soil

properties. The model for embedment used a time-averaged version of these velocities

over the embedment depth for use in Eqs. (3-2) and (3-3). The Kim and Stewart model

also uses a shear wave velocity for the evaluation of the a parameter, which is time-

averaged over the depth range zero to D+ BeA .

3.4.3 Results for Subject Buildings

Figures 3-4 to 3-8 show predicted transfer functions from the SASSI and NIST models

along with transfer function and coherence ordinates for each site considered. The dots

show transfer function ordinates with high coherence.

The predicted transfer function ordinates are similar from the SASSI and NIST

models for each considered structure. Further research is needed to evaluate the range of

foundation conditions where this similarity holds.

Figures 3-4 and 3-5 present results for the Sendai site for the two earthquakes.

The transfer and coherence functions from the two sets of earthquake recordings are

82
remarkably similar for the two events. For frequencies less than about 8 Hz where the

data have high coherence and hence relatively little effects of noise, the transfer functions

are nearly identical. The transfer functions from data fall below those from the two

models to a significant degree for frequencies beyond 2.5 Hz. This occurs despite the

slower velocities produced by the larger shear strains in the Tohoku earthquake.

In Figures 3-4 and 3-5 at frequencies above about 9 Hz, the coherence is low,

being near the range of 0.2-0.3 that is associated with pure noise for the level of

smoothing used in the calculations. Some peculiar transfer function ordinates are

observed in this range, including transfer function ordinates above unity, but those results

are not considered reliable due to the strong effects of noise. Investigation of piles with

incoherent inputs is beyond the ability of SASSI. The Sendai building was modeled with

piles for the coherent case, however, excluded for the incoherent case.

Figures 3-6 and 3-7 presents horizontal and rotational transfer functions from the

Onagawa site. The two models produce similar transfer functions for the horizontal

direction. The SASSI model produces higher rotational transfer functions than NIST,

which is likely due to stiffer materials below the foundation that are accounted for in the

SASSI model. The horizontal transfer functions (Figure 3-6) from these models over-

predict those from data for frequencies under about 7 Hz, where the coherence is

sufficiently high for the results to be meaningful. The rotational transfer functions (Figure

3-7) begin at zero, gradually rise up to about 15 Hz, but have a pronounced peak at about

3 Hz which is likely due to base rocking from inertial interaction. The NIST model

matches the data reasonably well whereas the SASSI model under-predicts.

83

Figure34 Sendai site horizontal transfer functions and coherence for the 2003 Off
Miyagiearthquake.

Figure35 Sendai site horizontal transfer functions and coherence for the 2011 Tohoku
earthquake.

84

Figure36 Onagawasitehorizontaltransferfunctionsandcoherenceforthe2011Tohoku
earthquake.

Figure37 Onagawa site rocking transfer functions and coherence for the 2011 Tohoku
earthquake.

85

Figure38 Nagoya site horizontal transfer functions and coherence for the 2004 Off Kii
Peninsulaearthquake.

Figure 3-8 presents horizontal transfer functions for the Nagoya site, which is of

special interest because of the large embedment and the relatively large foundation

horizontal dimensions, which are near the limit of the calibration range in the Kim and

Stewart study. The data from this site are a relatively high quality as indicated by high

coherence over a wide frequency range. The data are quite consistent with the model, and

the low transfer function ordinates (indicating substantial reductions of foundation

motions relative to free-field motions) are dominated by embedment effects.

3.5 Interpretation and Conclusions

It has become increasingly common in recent years for structural engineers to take

advantage of kinematic interaction effects to reduce foundation level ground motions

relative to those specified in the free-field. The guidelines that appear in NIST (2012) and

other documents account for the effects of base slab averaging and embedment. The base

slab averaging model is semi-empirical, being based on a theoretical formulation

calibrated for use in California.

86
 We examine data from several sites in Japan to investigate the effectiveness of the

NIST procedures, both relative to the data and relative to predictions from the finite

element code SASSI. For all three sites, predictions of horizontal transfer functions from

the SASSI code and the NIST procedures are similar over the frequency range of

principal interest (up to about 10 Hz). For horizontal motions, the model-based transfer

functions were too large (hence conservative) for the Sendai and Onagawa sites. For

kinematic base rocking, the model-based predications are accurate from the NIST model

but too low from the SASSI model, when compared to data. The model predictions are

fairly accurate relative to data at the Nagoya site, which is dominated by embedment

effects.

An interesting aspect of the data from the Sendai site is that it has experienced

numerous earthquakes of varying shaking amplitudes. Two earthquakes producing

moderate and very strong shaking are considered here, with remarkably similar results,

suggesting a lack of significant nonlinear effects on the kinematic ground motion

reductions for this site. Whether this finding is repeatable and transferable to other sites

requires further investigation.

Based on this work, we find generally satisfactory performance of the simplified

models for kinematic interaction presented in NIST (2012) using these data from Japan.

87
4 ASSESSMENT OF ALTERNATIVE SSI
MODELINGSTRATEGIES

4.1 Introduction

The analysis of building seismic response can utilize various approaches for modeling

soil-structure interaction (SSI) at the base of the building, including ignoring SSI effects

altogether. While some of these approaches are relatively simple, others require

significant effort to capture the linear or nonlinear SSI. What is not clear is whether more

complex and time-consuming approaches produce substantially different or more

accurate results than simpler approaches. This issue is examined while also

demonstrating the application of analysis tools presented in a recent guidelines document

(NIST, 2012) referred to here as the NIST report. The NIST recommendations for

kinematic effects have been previously discussed in Chapter 3. The analysis tools are

applied in the context of substructure analysis of the SSI problem, which requires

assumptions of equivalent-linear soil behaviour.

The approach taken in this work is to identify suitable instrumented buildings that

have recorded earthquakes, develop complete SSI models for substructure-based analysis

of seismic response (referred to as a baseline model), roughly calibrate the structural

88
elements of the baseline model to approximately match recordings, and then remove

components of the baseline model for additional analyses to evaluate the impact of those

components. This process is used to identify the critical components of an SSI model for

use in response history analysis, and develop recommendations based on those results as

well as prior results in the literature. Prior studies have taken a similar approach, but

utilized conventional structural software packages (ETABS and SAP) on different

buildings (Tileylioglu et al., 2010; Naeim et al., 2008) with slightly different SSI analysis

procedures.

4.2 Similar Studies in the Literature

Naeim et al. (2008) and Tileylioglu et al. (2010) utilized conventional structural

engineering software packages (ETABS and SAP) to perform response history analysis

including elements to capture SSI. The analyses were performed for two buildings (54-

and 6-story buildings in Los Angeles) with embedded foundations that had recorded

earthquake motions, although only the 54-story building results have been published. The

analyses were carried out initially with the model components show in Figure 4-1, where

kinematic base rocking was applied to the base of vertical foundation springs in addition

to depth-variable ground motions, which double-counts the effects of kinematic rotation.

Fortunately, those kinematic rocking effects were minor for the investigated structures

and hence do not affect significantly the findings.

89
Figure41 SchematicillustrationofsimulationusedinNaeimetal.(2008)andTileylioglu
etal.(2010).

The springs were elastic without a compression capacity limit and zero tension

capacity. Kinematic effects were evaluated for embedded foundations in a manner similar

to that described in Chapter 3 of the NIST report (with exception of kinematic base

rocking, which was applied in this work but is not recommended in Chapter 3 of the

NIST report except for rigid foundations for which a single horizontal input motion is

provided). Difficulties were encountered implementing multi-support excitation in the

computer programs, and the results that were obtained were considered reliable for

displacements but not for forces. Nonetheless, good matches to observed responses with

the model were obtained.

Elements of the model depicted in Figure 4-1 were then stripped from the

structural models to investigate their impact on the computed response. Factors found to

not significantly influence the results include consideration of multi-support excitation

along basement walls and application of the zero-tension condition on foundation springs.

Consideration of kinematic interaction effects was significant for the distribution of inter-

90
story drifts, particularly below the ground line. Consideration of foundation springs was

important for the building vibration periods and drift distributions.

Two approximations commonly used in practice were shown to provide poor

results: (1) fixing the structure at ground line with input consisting of free-field

translation (Model 1 in present study, as described in following section) and (2) fixing the

structure at the base level, applying free-field motions as input at the base level, and

using horizontal foundation springs along basement walls with their end condition fixed

to the free-field ground motion (Model 3 in present study). Fixing the structure at the

foundation level (i.e., not ignoring the below ground portion of the structure) and

applying free-field translations provided better results. A procedure was used in which

kinematic base rocking was applied to a rigid element connected to the foundation

springs (termed Model 2b by Naeim et al., 2008). That approach provided an excellent

match to MB results.

4.3 Model Types

4.3.1 Baseline Model (MB)

Similar to Naeim et al. (2008) and Tileylioglu et al. (2010) work, analyses of the subject

structures began by developing three-dimensional models of the buildings and foundation

systems, which are referred to as baseline models (MBs). MBs are not intended to be

the most accurate models that could be developed (e.g., a direct SSI analysis in a finite

element platform could potentially provide improved results), but instead represent the

implementation of procedures given in the NIST report. As shown in Figure 4-2, MBs

incorporate SSI in the vertical and horizontal directions, including rocking, with a series

91
of springs and dashpots representing site soil properties. Seismic demands imposed on

MBs include base translation as well as kinematic loading of basement walls (simulated

by displacement histories applied to the ends of horizontal springs attached to basement

walls). Using those seismic demands, MBs are calibrated to match the response

interpreted from recorded motions.

Figure 4-2 depicts the MB model, which is implemented using a substructure

approach. The springs and dashpots in the figure represent the frequency-dependent

foundation impedance for horizontal, vertical, and rocking modes of vibration. The

motion uFIM in Figure 4-2 represents the Foundation Input Motion, which is modified

from the free-field motion (ug) for the effects of kinematic interaction.

Figure42 Schematicillustrationofbaselinemodelconsideredinsimulations.

4.3.2 Simplified and Alternative Models

Once the MBs successfully match the recorded data, components of the seismic demand

and soil-foundation interaction elements are replaced, one or more at a time, to match

modeling approaches used in practice. The goal of simulations with these simpler models

is to assess the changes induced by each simplification on computed engineering demand

92
parameters (EDPs). As illustrated in Figure 4-3, four simplified models were considered,

as follow:

Model 1: Model the above-ground portion of the structure as fixed at the ground

surface and exclude subterranean and foundation components. Excite the base of the

structure with free-field ground motion ug. (Flexible structure, rigid basement and soil)

Model 2: Model above-ground and subterranean portions of structure but use no

horizontal foundation springs along basement walls and fix the base of the structure

against translation or rocking. Excite base of structure with free-field ground motion ug.

(Flexible structure and basement, rigid soil)

Model 3: Subterranean levels are modeled and horizontal and vertical soil springs

are included. Horizontal spring at base slab receives input motion (ug) but horizontal

springs at higher elevations are fixed at end with no input. Vertical springs are supported

on a fixed base. Applications of Model 3 in engineering practice have been

predominantly for pushover analysis, although its use here is for response history

analysis, which is expected to demonstrate poor performance.

Model 4 (bathtub): Subterranean levels are supported by horizontal and vertical

soil springs that are fixed at their ends to a rigid bathtub. The bathtub is excited with

the horizontal foundation input motion (uFIM) or free-field motion (ug). (Flexible

structure, basement and soil)

93
Figure43 Schematicillustrationofsimplifiedmodels1to4consideredinsimulations.

These simplified approaches were chosen based on consultations between the

ATC/NIST project team and structural and geotechnical engineering practitioners. The

purpose of those consultations was to document the state-of-practice in SSI analysis. The

outcome of that process is described in the NIST report. Referring to Figure 4-3, Models

1 and 2 are used most often, followed by Model 3. Model 4, which was introduced by

Naeim et al. (2008) and was evaluated in the present work, has only very recently begun

to see applications in practice.

The only differences between Model 4 and MB pertain to the manner in which

seismic demand is specified. The implementation of Model 4 with uFIM is identical to MB

94
except that the effects of kinematic loading of basement walls associated with depth-

variable displacement histories applied to the ends of horizontal foundation springs is

neglected. The implementation of Model 4 with ug takes this a step further by neglecting

kinematic interaction altogether by replacing the recorded motions at the base of the

building by equivalent free-field motions applied uniformly to the end of all horizontal

foundation springs. Input motions are generally specified in the engineering software as

accelerations. An exception is the multi-support excitation runs applied for the MB, in

which ground motions are specified as displacement histories that are obtained through

integration.

Table 4-1 Summary of the Properties of the MB and the Simplified Models Considered
Model Variations
No. 4
Parameters No. 1 No. 2 No. 3 (Bathtub) MB MB.1 MB.2
Structural Foundation
N/A Flexible Flexible Flexible Flexible Rigid Flexible
Elements
Input Motion ug ug ug 4a: uFIM 4b: ug uFIM uFIM uFIM
Depth-variable ground
No No No No Yes Yes Yes
motion
Spring tension N/A N/A Allowed Allowed Allowed Allowed Not Allowed

As shown in Table 4-1, variations on the MB (denoted MB.1 and MB.2) are

considered as follows:

1. The embedded portion of the building is assumed to be rigid (MB.1). The

specification of seismic demand is not modified. The objective here is to specifically

investigate the effects of flexibility in structural elements below ground line.

95
2. Change the manner in which soil flexibility is modeled (MB.2). Specifically, springs

are not allowed to develop tension (removal of no-tension interface elements) to

provide insight into the effects of nonlinearities in the foundation springs. More

sophisticated nonlinear springs were not considered because this effect proved to be

minor at the levels of demand imposed by the recorded ground motions.

As shown in Table 4-1, all modeling schemes but MB.2 allow tension in soil

springs. The rationale for this approach is that actual basement wall-soil contact prior to

earthquake shaking is characterized by a compressive normal stress that would fall in the

range of at rest to active earth pressures, whereas the springs have an initial condition of

zero force. Earthquake shaking will impose alternating cycles of increased and decreased

pressure relative to the initial state, which is well represented by a spring that can develop

tension provided the level of deformation does not lead to gap formation. Given the

modest demand levels imposed on the Sherman Oaks and Walnut Creek buildings, this

was anticipated to be a reasonable assumption, which was tested with Model MB.2 for

the Sherman Oaks building.

4.4 Buildings Considered

Only buildings with seismic instrumentation and available earthquake recordings in

California were considered. Desirable attributes of a building in this selection process

were as follows: (1) embedded foundations, so that kinematic effects associated with

embedment and depth-variable ground motions could be evaluated; (2) two vertical

sensors to record rocking at the foundation level; (3) relatively regular structural

configurations, so that the results obtained are not peculiar to an atypical building type;

96
and (4) structural configurations and site conditions that would tend to be conducive to

significant inertial SSI effects. No single building met all of these criteria. The buildings

that were selected are a 13-story building with two-levels of basement in Sherman Oaks,

California and a 10-story building without embedment in Walnut Creek, California,

shown in Figure 4-4. The Sherman Oaks building is a reinforced concrete moment frame

structure that was shaken strongly by the 1994 Northridge earthquake and weakly by

several other more distant events. The building is founded on grade beams and friction

pile foundations supported by alluvial sediments. The Walnut Creek building is a

reinforced concrete dual-system structure with perimeter moment frames and a central

core of shear walls. The shear walls are founded on a mat foundation with shallow

embedment, resting on weathered shale bedrock. The Sherman Oaks building satisfies the

criteria (1) and (3) above, while the Walnut Creek building satisfies (2), (3), and (4).

Figure44 Photographs of the example applications: 13story Sherman Oaks building

(left) and the 10story Walnut Creek (right). Photographs obtained from
CESMD.

97
 A parameter that is often used to identify conditions where inertial SSI effects are

likely to significantly affect structural response is h/(VsT), where h is approximately 2/3

of the building height (distance from base of foundation to centroid of first mode shape),

Vs is the time-averaged strain-compatible shear wave velocity of the foundation soil over

a representative depth, and T is the fixed-base first mode period for the structure. The

ratio h/(VsT) effectively represents a ratio of structure to soil stiffness, and inertial SSI

becomes more significant as this parameter increases, typically being important for

values of approximately 0.05 to 0.1 or greater (e.g., Veletsos, 1977; Stewart et al., 1999).

The h/(VsT) parameter for each case study is discussed in the following subsections.

4.5 Sherman Oaks Building

The building has 13 stories above the ground surface and a two-level basement. The

building is 50 m (164 ft) tall from the ground surface to the roof, with approximately 6.2

m (20.5 ft) of embedment. The height of the first floor is 7.0 m (23 ft), while all other

floors above are 3.6 m (11.75 ft). The 1st and 2nd subterranean levels are 3.5 m (11.5 ft)

and 2.7 m (9 ft) tall, respectively. The lateral loads are carried by moment-resisting

concrete frames that extend from the roof to the foundation, supplemented by perimeter

concrete walls at the subterranean levels. Vertical loads are carried by 11.4 cm (4.5 in)

concrete slabs supported by the concrete beams and columns in the moment frames. The

building was designed in 1964 and was later seismically rehabilitated with friction

dampers following the 1994 Northridge Earthquake.

The ratio of structure to soil stiffness for this building is h/(VsT) = 0.07, using

building height, soil shear wave velocity, and period values given in subsequent sections.

98
Because this ratio is < 0.1, strong inertial SSI effects in the form of period lengthening

and foundation damping are not expected. The building was nonetheless analyzed

because of the other attributes mentioned in Section 4.4 (common building type, regular

configuration, embedded foundation, multiple recordings) and the potential for

significant kinematic interaction effects on higher mode responses. Moreover, we sought

to investigate whether a building for which traditional first-mode SSI metrics indicate no

significant effect (i.e., period lengthening near unity, foundation damping near zero),

could in fact exhibit potentially significant impacts of SSI on the vertical distribution of

EDPs used in structural design (e.g., inter-story drifts, story shears, etc.).

4.5.1 Site Description

Figure 4-5 shows a plan view of the site. The ground surface elevation is approximately

216 m (709 ft) above mean sea level (AMSL). Geotechnical reports with boring and

geophysical logs for adjacent properties were obtained from MACTEC Engineering and

Consulting Inc. The reports contained 11 borings at a property south of the site and 36

borings at another property northwest of the site. Boring depths ranged from 15 to 38 m

(50 to 125 ft). The borings indicate relatively consistent soil conditions, consisting of

alluvial deposits composed of silt, with sand and clay overlying shale bedrock at depths

ranging from 21-27 m (70 to 90 ft). Figure 4-6 shows our interpretation of the soil

layering at the site based on the borehole data.

99

Figure45 Plan view of Sherman Oaks building site showing locations of borehole and
geophysicallogsusedforsitecharacterization

The geophysical data at the two neighboring sites was developed from downhole

seismic surveys; results from both are shown in Figure 4-6. Also shown is the profile

utilized for subsequent analysis, which gives weight to the conditions revealed by the

borings south of the site due to their relative proximity. Based on the MACTEC reports,

average moist unit weight is approximately 20 kN/m3 (124 lb/ft3). The water table was

measured at adjacent sites at an approximate depth of 12 m (40 ft) below the ground

surface.

100


Figure46 Subsurface characteristics based on adjacent site exploration at sites south

andnorthwest(NW)ofthelocalintersectionoftheShermanOaksbuilding:a)
shearwavevelocityprofileandb)materialprofile.1ft=0.3048m

To estimate strength properties of the foundation soils, drained direct shear test

results from samples retrieved at the south and NW properties were utilized. In the

shallow soils where most of the foundation-soil load transfer will occur, the soil is

unsaturated so drained shear strengths are used. For the range of surcharge pressures

appropriate for the soil profile across the pile depth of 16.2 m (53.1 ft) below ground

surface, the soil Mohr Coulomb strength parameters inferred from the available data in

the MACTEC reports are c = 20.1 kN/m2 (419.8 psf) and 30for 0 to 6.1 m (0 to

20 ft) and c = 12.9 kN/m2 (269.4 psf) and 38for6.1 to 16.8 m (20 to 55.1 ft).

101
4.5.2 Ground Motion Recordings

The Sherman Oaks building is designated as CSMIP Station No. 24322. It contains 15

accelerometers at the locations shown in Figure 4-7 and has been instrumented since

1977. The accelerometers are located on 5 levels: the 2nd subterranean, ground surface,

2nd, 8th and roof levels. Note that there is only one vertical sensor at the foundation level,

so base rocking effects cannot be measured. There is also no free-field ground instrument

in the vicinity of the site. The foundation and structural conditions illustrated in Figure 4-

7 are described in the next section.

Figure47 CSMIPStationNo.24322:ShermanOaks13Storycommercialbuilding,cross
sectionalviewandsensorlocationsketch

102
 The Center for Engineering Strong Motion Data (CESMD) has cataloged and

reviewed six sets of recorded motions at this building since 1977. The recorded

earthquake events and peak accelerations for the 2nd subterranean level, ground floor and

roof are presented in Table 4-2. Note that the accelerations follow an unusual pattern in

which the largest motions are at the ground floor (foundation and roof motions being

smaller).

The horizontal translations recorded at the base of the 2nd subterranean level of

the building were used as the foundation input motions (uFIM). The uFIM is the modified ug

response due to base slab averaging and embedment effects as discussed in Chapter 3.

Typically in practice, ug is known and the uFIM would be calculated for structural analysis

from transfer functions. In this case, uFIM was measured and ug was back-calculated by

removing the base slab averaging and embedment effects. Figure 4-8 shows the uFIM and

ug response spectra for the Northridge earthquake. The identified periods shown in Figure

4-8 were calculated as horizontal transfer functions of the power spectral density

functions of the foundation recording and roof recording from the Northridge earthquake.

Table 4-2 Summary of Recorded Earthquake Events and Peak Accelerations (g) at Three Levels
Earthquake 2nd Subterranean Ground Floor Roof
NS EW V NS EW V NS EW V
1987 Whittier 0.100 0.148 0.038 0.250 0.169 ---- 0.140 0.140 ----
1992 Landers 0.039 0.031 0.012 0.045 0.035 ---- 0.085 0.095 ----
1994 Northridge 0.453 0.227 0.181 0.886 0.392 ---- 0.467 0.257 ----
2007 Chatsworth 0.015 0.022 ------ 0.045 0.058 ---- 0.025 0.036 ----
2008 Chino Hills 0.049 0.037 0.010 0.073 0.045 ---- 0.082 0.067 ----
2010 Calexico 0.004 0.004 0.003 0.004 0.005 ---- 0.049 0.022 ----
*NS, EW, and V = north-south, east-west, and vertical, respectively.

103

Figure48 Response spectra of uFIM (recorded) and ug (inferred) for Sherman Oaks
building,1994Northridgeearthquake

An equivalent linear ground response analysis was performed in DeepSoil v4.0 to

estimate the depth-variable ground responses adjacent to the embedded portion of the

structure, taking the recording of uFIM as input at 6.7 m (22 ft) with an elastic half-space1.

Dynamic soil properties consisted of the Vs profile shown in Figure 4-6 and modulus

reduction and damping curves appropriate for the overburden pressures and soil types as

given by Darendeli (2001). DeepSoil does not update the shear modulus reduction of the

elastic half-space as performed for the soil column above during the analysis. Therefore,

the elastic half-space Vs parameter was updated through iterative runs until convergence

with the Vs of the deepest layer of the soil column. Figure 4-9 shows the computed

1
Input motions in ground response analyses can be specified as outcropping or within, the former requiring
an elastic halfspace and the latter a rigid base (Kwok et al., 2007). For the Sherman Oaks building, the
motion was recorded within the profile, but on a large foundation slab that could be interpreted as
representing an outcropping condition. Analyses were performed for both conditions, with the outcropping
results appearing more realistic.

104
variation of PGA and PGV with depth through the soil column. Note that PGA increases

significantly over the embedment depth (approximately 50%) whereas the change in

PGV is minor.

Figure49 Calculated variation of peak acceleration and velocity over the embedment
depthforShermanOaksbuilding,1994Northridgeearthquake

4.5.3 Foundation Conditions

We obtained construction drawings of the building from the California Strong Motion

Implementation Program (CSMIP). The drawings are heavily pixilated and difficult to

read, so our interpretation of structural details may not perfectly reflect the content of the

original drawings. This subsection focuses on the foundation conditions that were

captured for the MB. The foundation support with springs and dashpots is described in

105
4.5.4. Structural modeling is briefly discussed in Section 4.7, and further details are

provided in Appendix A of the NIST report.

Figure 4-10 shows that the foundation is composed of variously configured bored

pile groups connected with grade beams. The cast-in-place (bored) concrete piles are 51

cm (20 in) in diameter and extend to a depth of approximately 9.9 m (32.5 ft) below the

lower subterranean level. Typical spacing of piles (center to center) is 0.9 to 1.2 m (3 to 4

ft). The foundation is rectangular in shape with plan dimensions of 36.3 by 57.6 m (119

by 189 ft). The dimensions of the main structure above the ground surface are 21.9 by

57.6 m (72 by 189 ft), although there is a widened section at the first floor matching the

foundation dimension. Basement walls surround all embedded levels with the southern

wall supported on approximately 36 0.6 m (2 ft) bored piles. Spacing for these piles is

generally 1.8 m (6 ft) center to center with the exception of closer spacing near the

elevator shaft.

106


Figure410 ShermanOaksbuildingfoundationplan,basedonconstructiondrawings.1ft
=0.3048m

4.5.4 Development of Foundation Springs and Dashpots

In this section, the development of foundation springs and dashpots is described using the

input parameters given in Sections 4.5.1-4.5.3. The critical input parameters are

foundation dimensions, structural period, soil properties, and input motion amplitudes.

The analysis methodologies are described in Chapter 2 of the NIST report. Foundation-

soil resistance from both shallow and deep foundation elements are considered in our

analysis of springs and dashpots. Hence, it is assumed that the shallow foundation

elements (principally grade beams) remain in contact with the soil, which is justified by

107
the sandy nature of the foundation soils and the presence of friction (not end bearing)

piles. The following are emphasized in the present discussion:

Identification of effective profile shear wave velocity for shallow foundation elements

in consideration of foundation dimensions, overburden pressures from the structure,

and nonlinear effects.

The calculation of shallow foundation horizontal stiffness and damping and their

separation into contributions from the base slab (derived principally from slab-soil

frictional stresses) and passive pressures against the basement walls.

The calculation of vertical stiffness and damping for a bed of springs under the base

slab, emphasizing the proper representation of rotational impedance for shallow

foundation elements.

The distribution of shallow foundation springs and dashpots to the modeled structure

nodes.

The calculation of limiting spring forces (capacities).

The contributions of pile groups to the vertical and lateral stiffness of the foundation

system.

Effective Profile Velocity: As described in Section 2.2.2 of the NIST report,

effective profile velocity is calculated by averaging velocities (as ratio of profile depth to

shear wave travel time) across specified profile depths, zp. Table 4-3 shows the values of

zp for each foundation vibration mode along with the depth range used for the calculation

of effective velocity. The maximum depth considered is not equal to zp because of the

108
foundation embedment (D). The depth range is taken from the ground surface to zp+D for

each vibration mode. For the specific case of the horizontal base spring, a narrower depth

interval starting at the base of the foundation is used, as described further below.

Table 4-3 Effective Profile Depths and Velocities for Calculation of Static Stiffness. 1
ft=0.3048 m
Eq. for zp zp Depth Range Vs0b
Vibration Mode (from Eq. 2.46a (m) (m) (m/s)

Horizontal Translation (x & y), overall 17.8 0 to 24.0 254.2

Horizontal Translation (x & y), base spring 17.8 6.2 to 24.0 296.2
Vertical Translation (z) 17.8 0 to 24.0 254.2

Rocking along x-axis (xx) 14.0 0 to 20.2 243.0

Rocking along y-axis (yy) 22.6 0 to 28.8 277.7

aB=11 m (foundation half width in short direction), L=28.8 m (half-length). Foundation depth D=6.2 m.
bVs0 is the free-field, non-overburden corrected average shear wave velocity.

As described in Section 2.2.2 of the NIST report, free-field velocities Vs0 should

be increased to account for the increased overburden provided by structure weight. For

the Sherman Oaks building, the mass of the excavated soil for the basement is slightly

greater than the structural mass (using values from Section 4.5.3), so no overburden

correction is needed. The nonlinear shear modulus reduction factor is 0.5 per Table 4-1

and the Northridge earthquake peak acceleration of about 0.4g (Site Class D). No

modulus reduction is required for other events.

Horizontal Stiffness and Damping: Table 4-4 shows the calculation of horizontal

static stiffnesses and damping values. Horizontal static stiffnesses, including embedment

effects, are calculated based on the gross foundation dimensions of 21.9 m by 57.6 m. As

shown in Figure 4-10, the foundation plan for the structure includes pile-caps

interconnected with grade beams and a surficial slab. The use of gross foundation

109
dimensions to calculate static stiffnesses assumes that stress-transfer from pile caps and

grade beams occurs in stress bulbs that overlap at modest depth, so that the shallow

foundation components effectively act as a unified system. This assumption does not

require load transfer from the relatively thin foundation slab to the soil. The horizontal

stiffness has contributions from base shear and passive pressure against basement walls.

Because the base shear reaction mobilizes only soil below the embedment depth D, the

effective profile velocity for the base spring is taken from D to D+zp, as shown in Table

4-3. These horizontal base spring stiffnesses, which are given in Table 4-4, are then

subtracted from the overall horizontal stiffness to evaluate the contribution from passive

pressures against the basement walls.

Table 4-4 is organized to first present the reduced shear modulus used for the

spring calculations, which are derived from the velocities in Table 4-3 along with the

mass density (20 kN/m3/g) and the shear modulus reduction (0.5 for Northridge). Static

stiffnesses are then computed as the product of a surface stiffness (Table 2-1a) and

embedment factor (Table 2-1b); both values are given in Table 4-4. Dimensionless

frequency a0 is calculated, from which dynamic modifiers of stiffness (x, y) and

radiation damping coefficients (x, y) are computed. The damping ratio used for dashpot

calculations is the sum of the radiation damping coefficient and hysteretic damping,

which is evaluated from Darendeli (2001) as 0.088 for Northridge and 0.009 for other

events. The calculation of dynamically modified spring stiffness and dashpot coefficient

from these results is detailed in Table 4-4.

110


Table 4-4 Evaluation of the Impedance Functions For Sherman Oaks Building. Static Stiffness Results Shown are for Case of
Northridge Earthquake, Which Uses a Shear Modulus Reduction Factor of G/G0=0.5. Shear Moduli (G) and Static Stiffnesses
(Ksur) for Other Earthquakes are Doubled Relative to Those Shown In This Table (G/G0=1 for Those Events)
Table 2-1a
Table Eq. Table Table Table
G Ksur (kN/m)a 2-1b 2-7b 2-1a 2-1a 2-1b
Spring/Dashpot (MPa) or (kN.m/rad)b a0c sur emb Dynamic Stiffness Dashpot
Horizontal, overall ,
x-direction 65.2 6.48e6 1.45 0.13 1 0.076 0.114 , ,
2
,
, & ,
Horizontal, overall ,
y-direction 65.2 7.04e6 1.45 0.13 1 0.070 0.106 , , 2
,
, & ,

Horizontal, base spring ,

x-direction 88.6 8.81e6 N/A 0.11 1 0.065 N/A , ,
2
,
, & ,

Horizontal, base spring ,

y-direction 88.6 9.56e6 N/A 0.11 1 0.060 N/A , , 2
,
, & ,

Vertical, z 2
65.2 8.55e6 1.22 0.13 1 N/A 0.132 ,
&

Rocking about x-axis 2

58.0 1.05e9 1.74 0.14 1 N/A 0.010 ,
&

Rocking about y-axis 2

73.9 5.55e9 1.58 0.12 0.99 N/A 0.003 ,
&

a Translation static stiffnesses calculated using B=11 m, L/B=2.625, and =0.33.

b Rotational Static stiffnesses calculated using B=11 m, L/B=2.625, and =0.33.
c Calculated at frequency corresponding to first-mode period of structure, taken as T = 2.7 sec.

111

 Vertical Stiffness and Damping: Overall vertical and rotational stiffnesses are

computed using the profile velocities given in Table 4-3 with the results in Table 4-4.

Distributed vertical springs are provided over the base of the foundation, as shown in

Figure 4-10. The springs are provided in the form of a vertical stiffness intensity, which

is stiffness normalized by area (e.g., Equation 2-19 of the NIST report). The stiffness of

any individual spring is computed as the product of stiffness intensity and the tributary

area for the spring. As described in Section 2.2.3 of the NIST report, stiffness intensities

are adjusted near the edges of the foundation to match the overall rocking stiffness values

given in Table 4-4 (using Equation 2-20a-b of the NIST report). End length ratio Re = 0.4

was used for these calculations.

.
8,202 715
29,361 2,454
36,524 3,209
43,687 3,964
Figure411 Verticalspringanddashpotintensitiesandareasdistinguishedbycolor.Solid
lines represent tributary area boundaries in which stiffness and dashpot
values were distributed to the column node of each area. Solid circles
representcolumnnodes.1ft=0.3048m

112
 The vertical stiffness intensities in the gray central zone in Figure 4-11 are

computed from Equation 2-19 of the NIST report and reflect the vertical impedance. The

edge intensities in the blue and yellow zones are increased by factor Rk = 5.33 for the xx

direction and 3.58 for the yy direction. Stiffness intensities for corner zones were

evaluated as the average of the xx and yy intensities. Dashpot intensities are calculated

from stiffness intensities as c z = 2Rc k z , where Rc represents the reduction factor to

i i

capture rotational rocking (Equation 2-20c-d of the NIST report).

Pile Groups: Pile group horizontal and vertical stiffness and damping were

evaluated in accordance with Section 2.3 of the NIST report. As shown in Table 4-5,

active pile length La is evaluated separately for the two vibration modes and is longer for

vertical than horizontal (the calculation is iterative using results from Syngros, 2004).

Effective shear wave velocity is calculated over the depth interval D to D+La; shear

modulus is calculated from effective velocity and reduced for large strain effects

(reduction factor of 0.5 for Northridge; 1.0 otherwise). Static single pile stiffnesses are

calculated using Equation 2-21b and Table 2-3a of the NIST report. Dynamic group

factors are then evaluated from Figure 2-12 of the NIST report; results are given in Table

4-5 for 3x3 and 4x4 group sizes. Those results were applied to individual pile groups

(shown in Figure 4-3), with 3x3 results applied to groups with Npile=12 and 4x4 results

applied to groups with Npile13. Per Figure 2-12 of the NIST report, the stiffnesses of

individual groups are then calculated as the product of the group stiffness ratio,

static single pile stiffness, and Npile.

113
Table 4-5 Calculation of Pile Group Stiffnesses for Horizontal and Vertical Vibration.
Modulus Applies for Case of Northridge Earthquake with G/G0=0.5
Static SP Group
Active stiffness stiffness Group Radiation
Spring, length, La Vs0 G ratio, Damp., j
dashpot (m) (m/s) (MPa) Fig 2-11a Fig 2-11 a

Hor., x&y
4.1 180.1 63.4 237.3 0.47 0.2
(3x3 grp)
Vert., z (3x3) 9.9 183.7 67.6 859.3 0.43 0.25
Hor., x&y
4.1 180.1 63.4 237.3 0.36 0.05
(4x4 grp)
Vert., z (4x4) 9.9 183.7 67.6 859.3 0.30 0.29
Note: Static single pile (SP) stiffnesses calculated for Ep = 474.8 MN/m and d=0.6 m.
a Figure is in the NIST report (NIST, 2012)

Nodal Distribution: As shown in Figure 4-12, the horizontal springs and dashpots

were distributed across nodes on the basement walls (representing passive pressures) and

base slab (representing base shear). Vertical springs and dashpots were distributed across

nodes on the base slab. There are a total of 68 nodes on the foundation having springs and

dashpots. The basement walls nodes are positioned at the columns lines at the foundation

floor, 1st subterranean floor and ground surface floor, resulting in each wall containing

38 and 33 nodes, in the longitudinal and transverse directions, respectively. Based on

column locations, the foundation slab has 38 nodes, with the 22 perimeter nodes also

being used for the basement walls.

114


PileGroupContribution , ,

Translationinthe , , , ,

ydirection 48 16 16

Translationinthe , , , ,

xdirection 27 6 6

Translationinthe
zdirection Note: isnotaconstantvalue, isafunctionofNpilesand
isafunctionofintensitiesinFigure411.

Figure412 Distribution of nodes on foundation and calculation of individual spring

stiffnesses.

115
 The overall horizontal stiffness, combining the attributes from base shear, passive

pressure against basement walls and pile groups, was distributed to the foundation

perimeter (i.e., interior nodes have no horizontal springs). Half the overall horizontal

stiffness was distributed to each edge of the foundation for a given direction as shown in

Figure 4-12a. As described previously, the total stiffness is partitioned vertically into a

base stiffness and a wall stiffness for a given direction. Wall stiffnesses are equally

distributed to each basement wall node. Base stiffnesses are equally distributed along the

sides of the foundation. These distributions are shown in cross-sections in Figure 4-12b

and 4-12c with their respective calculations for each elevation. The vertical stiffness was

modeled at all 24 foundation nodes as the contribution of the aforementioned stiffness

intensity areas in Figure 4-11 and the columns respective vertical pile group stiffness

. Dashpots were distributed in the same manner as their respective springs as shown

in Figure 4-12.

Capacities of individual springs were evaluated using conventional vertical and

lateral pile group capacity analysis (e.g., Salgado, 2006). The pile group capacities were

assigned to nodes in the manner described above for stiffness. These capacities are

compared to spring force demands in the following section.

4.6 Walnut Creek Building

The Walnut Creek office building has 10 stories above the ground surface and no

subterranean levels. The building is 39.2 m (129 ft) tall from the ground surface to the

roof. The height of the first floor is 4.9 m (14 ft), while all other floors above are 3.8 m

(12.5 ft). Lateral loads are carried by an interior concrete shear wall core that is

116
embedded 3 m (10 ft) and an exterior precast and cast-in-place concrete frame. Vertical

loads are carried by 7 cm (2.75 in) lightweight concrete over 7 cm (2.75 in) precast panel

slabs supported by a precast, prestressed reinforced concrete beams. The building was

designed in 1970.

The ratio of structure to soil stiffness for this building is h/(VsT) = 0.11 using

effective building height, soil shear wave velocity, and period values given in subsequent

sections. Because this ratio is > 0.1, potentially significant inertial SSI effects in the form

of period lengthening and foundation damping are expected. Other motivating factors for

analyzing this building include the dual-system configuration of the structure and the

presence of two vertical instruments at the foundation level that can be used to infer

rocking of the core shear wall system.

4.6.1 Site Description

Figure 4-13 shows a plan view of the site. The ground surface elevation is approximately

51.8 m (170 ft) above mean sea level (AMSL). Portions of a geotechnical report,

prepared by Harding, Miller, Lawson & Associates (1970), were obtained for the

modeled structure and adjacent parking structure. A geotechnical report with seismic

refraction investigation of the subsurface soils, prepared by Raney Geotechnical (1983),

was obtained for the property adjacent to the aforementioned parking structure.

117


Figure413 Site plan of Walnut Creek office building showing locations of borings and
refractionlines

As shown in Figure 4-13, the reports contained six borings in and around the

ground surface footprint of the Walnut Creek building, five borings at the parking

structure, and eight borings at the adjacent property. Four seismic refraction surveys were

performed and the two closest to the modeled site were utilized for analysis of the soils

seismic properties. Boring depths ranged from 1.5 to 17.4 m (5 to 57 ft). The borings

indicate predominantly west dipping contacts of sandy clays and silts with variable

thickness of 0.6 to 5.5 m (2 to 18 ft) overlying siltstone and sandstone of the Orinda

Formation. Figure 4-14 shows the material profile based on the aforementioned reports at

the location of the interior core foundation.

The geophysical data at the neighboring site was developed from seismic

refraction surveys; interpreted shear wave velocity results for the modeled site are shown

118
in Figure 4-14. The shear wave velocities in the soils were estimated based on

correlations by Fumal and Tinsley (1985), while the shear wave velocity in the rock was

based on the seismic refraction data presented by Raney Geotechnical. Based on the

reports, average moist unit weight is approximately 18.1 kN/m3 (115 lb/ft3). The water

table depth was measured between 6.1 to 12.2 m (20 to 40 ft) below the ground surface.

Figure414 Subsurface characteristics based site and adjacent site explorations of the
WalnutCreekbuilding:a)shearwavevelocityprofileandb)materialprofile

4.6.2 Ground Motion Recordings

The Walnut Creek building is designated as CSMIP Station No. 58364. It contains 16

accelerometers at the locations shown in Figure 4-15 and has been instrumented since

1979. The accelerometers are located on four levels: ground surface, 3rd, 8th and roof

119
levels. There are two vertical sensors at the ground level, allowing the base rocking to be

measured. There is no free-field ground instrument in the vicinity of the site.

The Center for Engineering Strong Motion Data (CESMD) has cataloged and

reviewed five sets of recorded motions at this building since 1979. The recorded

earthquake events and peak accelerations for the ground floor and roof are presented in

Table 4-6.

Figure415 CSMIPStationNo.58364:WalnutCreek10Storycommercialbuilding,sensor
locationsketch.

120
Table 4-6 Summary of Recorded Earthquake Events and Peak Accelerations at Three
Levels
Earthquake Ground Floor Roof
NS EW V1 V2 NS EW V
Livermore 80A 0.030 0.033 0.022 0.023 0.116 0.108 ----
Livermore 80B 0.061 0.048 0.024 0.018 0.137 0.192 ----
1989 Loma Prieta 0.102 0.046 0.053 0.047 0.208 0.164 ----
2007 AlumRock 0.005 0.003 ------ ------ 0.018 0.015 ----
2008 Alamo 0.029 0.018 ------ ------ 0.057 0.034 ----
*NS, EW, and V = north-south, east-west, and vertical, respectively. V1=Chan. 12 (west) and V2=Chan. 13 (East)

The horizontal translations recorded at the ground level of the building were

assumed to incorporate the base slab averaging component of kinematic interaction, but

not the embedment component [the differential elevation between the ground and

foundation levels is 4.2 m (14 ft)]. Transfer functions derived using the embedment

procedures in Section 3.3.2 were used to modify the recorded translation to the

foundation input motion (uFIM). No free-field ground motions (ug) were recorded, so the

recordings from the ground level of the structure were divided by a transfer function for

base slab averaging (calculated using the procedures in Section 3.3.2) to infer ug. Figure

4-16 shows the uFIM and ug response spectra for the Loma Prieta earthquake.

121


Figure416 Response spectra of the uFIM and ug for Walnut Creek building, 1989 Loma
Prietaearthquake

4.6.3 Foundation Conditions

Construction drawings of the building were obtained from CSMIP. In this section, the

foundation conditions are described. The representation of the foundation support with

springs and dashpots is provided in Section 4.6.4. Structural modeling details are briefly

discussed in Section 4.7, with details provided in Appendix C of the NIST report.

Figure 4-17 provides the North-South elevation view of the Walnut Creek

building. This illustrates how the shear wall core extends below the ground level to a

foundation slab. Frame elements terminate at ground line, where they are supported on

spread footings or drilled shafts inter-connected with grade beams. The core wall extends

3 m (10 feet) below the ground level and the footing is 1.2 m (4 feet) thick, so the base of

the core wall footing is a total of 4.3 m (14 feet) below the ground level surface. This

figure also shows that the ground surface is not level and that the left columns are slightly

122
longer than the right columns; for simplicity, this is not represented in the structural

model and all columns are terminated at the same level (labeled as Ground in Figure 4-

13).

Figure 4-18 shows that the foundation includes spread footings, drilled shafts, and

mat elements. The mat foundation supporting the core walls has lateral dimensions of

11.6 by 25.6 m (38 by 84 ft), whereas the dimensions of the core walls are 7.3 by 21.3 m

(24 by 70 ft). Between the top of the mat and ground surface, the mat is backfilled with

concrete and sand. The distribution of spread footings and drilled shafts for support of

exterior columns is shown in Figure 4-18.

Figure417 OverallnorthsouthelevationviewoftheWalnutCreekbuilding(fromStewart
andStewart,1997).

123

Figure418 WalnutCreekbuildingfoundationplan(fromconstructiondrawings).

4.6.4 Development of Foundation Springs and Dashpots

In this section the development of foundation springs and dashpots is described using the

input parameters described in Sections 4.6.1-4.6.3. The critical input parameters are

foundation dimensions, structural period, soil properties, and input motion amplitudes.

The analysis methodologies are as described in Chapter 2 of the NIST report. The

following is emphasized below:

Identification of the potential effect of foundation element flexibility on foundation

impedance.

Identification of effective profile shear wave velocity in consideration of foundation

dimensions, overburden pressures from the structure, and nonlinear effects.

The calculation of the interior mat foundation horizontal stiffness and damping.

124
 The calculation of vertical stiffness and damping for a bed of springs under the base

slab, emphasizing the proper representation of rotational impedance for shallow

foundation elements.

The distribution of mat foundation springs and dashpots to the modeled structure

nodes.

The calculation of springs and dashpots for the footings and drilled shafts that support

the moment frame.

Foundation Flexibility: As described in Section 2.2.1 of the NIST report, standard

impedance function models apply for rigid foundations. The Walnut Creek building has a

central core consisting of effectively rigid shear walls supported by an underlying

reinforced concrete slab that is larger in plan dimension that the core. As discussed in

Section 2.2.3 of the NIST report, flexibility in the mat beyond the limits of the effectively

rigid core can reduce stiffness and radiation damping in the rocking mode (Iguchi and

Luco, 1982).

To evaluate impedance reduction relative to the rigid case (k/[k =0]), plots from

Iguchi and Luco (1982) that are similar to Figure 2-7 of the NIST report were utilized,

but for a ratio of core/foundation radius equal to 0.75. The results of this calculation are

shown in Table 4-7, which indicate that the foundation stiffness ratios are close to one,

indicating little deviation from the rigid foundation case. Therefore the full mat

dimensions of 11.6 by 25.6m (38 by 84 ft) were used for the analysis of foundation

impedance.

125
Table 4-7 Evaluation of Soil-to-Foundation Stiffness and Foundation Flexibility.
1 ft=0.3048 m

Core rc Slab
Vibration Mode Eq. for rf (m) rf (m) (Eq. 2-19)a ,

Vertical Translation (z) 7.0 9.7 0.72 127 0.9

Rocking along x-axis (xx) 4 9.3 12.0 0.78 188 0.8

Rocking along x-axis (yy) 4 5.5 8.1 0.68 58 0.8

Ef =22,894.7 MPa, tf = 1.22 m, f = 0.2, a0 = 0.13, and G = 347.5 MPa (based on Vs,avg = 14 m/s)
Note: k/(k, =0) assessed from Iguchi and Luco (1982) plots using rc/rf = 0.75 and = 100.
a Figures are in the NIST report (NIST, 2012)

Effective Profile Velocity: As described in Section 2.2.2 of the NIST report,

effective profile velocity is calculated by averaging velocities (as ratio of profile depth to

shear wave travel time) across specified profile depths, zp. Table 4-8 shows the values of

zp for each foundation vibration mode along with the depth range used for the calculation

of effective velocity. The maximum depth considered is not equal to zp because of the

foundation embedment (D). The depth range is taken from the ground surface to zp+D for

each vibration mode. For the specific case of the horizontal base spring, a narrower depth

interval starting at the base of the foundation is used, as described further below.

Table 4-8 Effective Profile Depths and Velocities for Calculation of Static Stiffness
Eq. for zp zp Depth Range Vs0b
Vibration Mode (from 2.46) (m) (m) (m/s)

Horizontal Translation (x & y) 8.6 4.3 to 12.9 434.3

Vertical Translation (z) 8.6 0 to 12.9 415.9

Rocking along x-axis (xx) 7.1 0 to 11.3 413.5

Rocking along y-axis (yy) 10.5 0 to 14.8 418.2

a B=5.8 m (foundation half width in short direction), L=12.8 m (half-length). Foundation depth D=4.3 m.
b Vs0 is the free-field, non-overburden corrected average shear wave velocity.

126
 As described in Section 2.2.2 of the NIST report, free-field velocities Vs0 should

be increased to account for the increased overburden provided by structure weight. The

overburden-corrected Vs,f is evaluated as described in Equation 2-16 of the NIST report.

The increase in vertical stress from the structural weight was evaluated at two-foot

intervals using the 2V:1H method (Holtz et al., 2011, Sec. 10.3). The velocity change was

calculated using Equation 2-16 of the NIST report and the adjusted velocities were

averaged over the depth ranges from Table 4-8 to provide the results presented in Table

4-9.

Table 4-9 Velocity Correction for Structural Overburden Weight

Z Vs0a (m/s) Vs,fb (m/s)
Horizontal Translation (x & y) 434.3 517.8
Vertical Translation (z) 415.9 463.7
Rocking along x-axis (xx) 413.5 460.8
Rocking along y-axis (yy) 418.2 465.5
aVs0 is the free-field, non-overburden corrected average shear wave velocity.
bVsf is the overburden corrected average shear wave velocity.

The nonlinear shear modulus reduction factor is 0.95 per Table 4-1 of the NIST

report and the Loma Prieta earthquake peak acceleration of about 0.1g (Site Class C). No

modulus reduction is required for other events.

Mat Foundation Horizontal Stiffness and Damping: Table 4-10 shows the

calculation of horizontal static stiffnesses and damping values. Based on the geometric

configuration of the building the horizontal stiffness values for the interior core were

modeled from the contribution of the base shear and neglected any passive pressure along

the embedded portion of the core walls. Because the base shear reaction mobilizes only

127
soil below the embedment depth D, the effective profile velocity for the horizontal base

spring is taken from D to D+zp, as shown in Table 4-9.

Table 4-10 is organized to first present the reduced shear modulus used for the

spring calculations, which are derived from the velocities in Table 4-9 along with the

mass density (18.1 kN/m3/g) and the shear modulus reduction (0.95 for Loma Prieta).

Static stiffnesses are then computed as the product of a surface stiffness (Table 2-1a) and

embedment factor (Table 2-1b); both values are given in Table 4-10. Dimensionless

frequency a0 is calculated, from which dynamic modifiers of stiffness (x, y) and

radiation damping coefficients (x, y) are computed. The damping ratio used for dashpot

calculations is the sum of the radiation damping coefficient and hysteretic damping,

which is evaluated from Menq (2003) as 0.01. The calculation of dynamically modified

spring stiffness and dashpot coefficient from these results is detailed in Table 4-10.

128
Table 4-10 Evaluation of the Impedance Functions for Walnut Creek Building. Static Stiffness Results Shown are for Case of Loma Prieta
Earthquake
Table 2-1a
Table Eq. Table Table Table
G Ksur (kN/m)a 2-1b 2-7b 2-2a 2-2a 2-2b
Spring/Dashpot (MPa) or (kN.m/rad)b a0c sur emb Dynamic Stiffness Dashpot
Horizontal, base spring ,
x-direction 469.4 2.24e7 N/A 0.11 1 0.057 N/A , ,
2
,
, & ,

Horizontal, base spring ,

y-direction 469.4 2.40e7 N/A 0.11 1 0.053 N/A , , 2
,
, & ,

Vertical, z 2
376.4 2.35e7 1.22 0.12 1 N/A 0.111 ,
&

Rocking about x-axis 2

371.6 8.49e8 1.70 0.12 1 N/A 0.005 ,
&

Rocking about y-axis

379.3 2.78e9 1.54 0.12 0.99 N/A 0.002 ,
2
&

a Translationstatic stiffnesses calculated using B=5.8 m, L/B=2.2, and =0.33.

b RotationalStatic stiffnesses calculated using B=5.8 m, L/B=2.2, and =0.33.
c Calculated at frequency corresponding to first-mode period of structure, taken as T = 0.66 sec

129

 Mat Foundation Vertical Stiffness and Damping: Overall vertical and

rotational stiffnesses are computed using the profile velocities given in Table 4-9 with

the results in Table 4-10. Distributed vertical springs are provided over the base of the

mat, as shown in Figure 4-19. The springs are provided in the form of a vertical

stiffness intensity, which is stiffness normalized by area (e.g., Equation 2-19 of the

NIST report). The stiffness of any individual spring is computed as the product of

stiffness intensity and the tributary area for the spring. As described in Section 2.2.3,

stiffness intensities are adjusted near the edges of the foundation to match the overall

rocking stiffness values given in Table 4-10 (using Equation 2-20a-b of the NIST

report). End length ratio Re = 0.4 was used for these calculations.

.
96 275
308 778
418 1,235
526 1,697
Figure419 Verticalspringanddashpotintensitiesandareasdistinguishedbycolor.
Solidlinesrepresenttributaryareaboundariesinwhichstiffnessand
dashpotvaluesweredistributedtoeacharea.1ft=0.3048m

The vertical stiffness intensities in the gray central zone in Figure 4-19 are

computed from Equation 2-19 of the NIST report and reflect the vertical impedance.

The edge intensities in the blue and yellow zones are increased by factor Rk = 5.48 for

130
the xx direction and 3.21 for the yy direction. Stiffness intensities for corner zones

were evaluated as the average of the xx and yy intensities. Dashpot intensities are

calculated from stiffness intensities as c z = 2Rc k z , where Rc represents the

i i

reduction factor to capture rotational rocking (Equation 2-20c-d of the NIST report).

Springs and Dashpots Representing Footings and Grade Beams: Single

valued springs and dashpots were provided for the translational modes (horizontal and

vertical) for the exterior spread footings and piles. For simplicity the spread footings

and belled pile heads were treated as individual shallow rectangular foundations with

dimensions of 3.05 m (10 ft) by 3.05 m (10 ft). The calculations were similar to those

for the interior mat, as described in Table 4-10; the same shear moduli G were used,

but foundation half-width (B) was taken as 1.52 m (5 ft). The horizontal springs (kx

and ky) and vertical springs (kz) were calculated as 3.91e6 kN/m and 4.97e6 kN/m,

respectively. The horizontal dashpots (cx and cy) and vertical dashpots (cz) were

calculated as 1.68e3 kN-s/m and 2.75e4 kN-s/m, respectively. The dimensionless

frequency utilized for these calculations was a0=0.02.

4.7 Baseline Model and Comparison to Recordings

Open System for Earthquake Engineering Simulation (OpenSees) software was used

to perform response history analyses of a three-dimensional (3-D) model of the

Sherman Oaks and Walnut Creek buildings. OpenSees is an open-source software

package that was developed at the University of California, Berkeley for simulations

of structural and geotechnical responses to earthquakes (OpenSees 2011). The

structural conditions of the buildings are described in the NIST report with reference

131
to their representation in OpenSees models. The development of foundation springs

and dashpots for both MBs is described in Sections 4.5.4 and 4.6.4 of this document.

The original OpenSees structural model for the Sherman Oaks building was

provided by Erol Kalkan of the CSMIP and the US Geological Survey (personal

communication, 2011). The original structural model was improved with inclusion of

the subterranean levels and foundation springs and dashpots for the MB, but

otherwise the initial structural properties were those from the Kalkan model. The

response history recording data from the 1994 Northridge, 1992 Landers and 1987

Whittier earthquakes were used for the Sherman Oaks building.

The Walnut Creek OpenSees model was generated as part of this research.

Because of the modest embedment of the Walnut Creek core wall foundation and the

lack of subterranean levels, multi-support excitation along embedded portion of the

foundation was not considered, causing the baseline model (MB) to match the bathtub

model (Model 4) in this case. The implementation of the bathtub model for this

structure is illustrated in Figure 4-20. The other models considered are fixed-base

models analogous to Models 1 and 2 from Figure 4-3 (implementation for Walnut

Creek building is shown in Figure 4-19). To expedite the analyses and the post-

processing, the model was reduced to two-dimensional (2-D). The 2-D model

includes all the elevations spanning the East-West direction.

132

Figure420 SchematicillustrationofmodelingapproachesformodelingWalnutCreek
building.

Displacement histories for both directions and multiple floor levels obtained

from the calibrated Sherman Oaks MB are compared to recordings in Figures 4-21 for

the Northridge earthquake. Figure 4-22 shows the peak displacements and

accelerations. The match in both horizontal directions at the foundation and ground

levels is excellent. Elsewhere over the height of the building, the quality of the match

is generally better in the longitudinal direction than transverse (e.g. Channel 1 fit at

roof level has a better match of phasing and amplitudes than Channel 2). The

calibration process began with expanding the original structural model by Kalkan to

include the sub-grade levels, where shear stiffness of the reinforced concrete walls

was of concern. To keep the calibration approach simple, a multiplier on the

theoretical shear modulus of the uncracked concrete (Gc) was taken as the calibration

parameter, with the goal of matching near-ground response. A multiplier of 0.25 was

used (which is a reasonable value to reflect the cracked stiffness properties of the

walls), however, the near-ground response was not highly sensitive to this parameter

over the range of 0.25-0.40 that was considered. The next step in the calibration was

to seek to match the building period through adjustments of the structural stiffness

133
and mass. The real stiffnesses of a reinforced concrete element is highly variable

(Haselton et. al 2011). Therefore the rebar stiffness (as a proxy for overall element

flexural stiffness) was modified by 0.7 from the original value, whereas the mass was

not modified.

134


Figure421 Comparison of recorded displacement histories at roof, 8th floor, ground
levelandbaselevelwiththosepredictedusingtheBaselineModel(MB).
Northridgeearthquake.

135

Figure422 Comparison of recorded peak displacements over the height (X and Y
direction)withthosepredictedusingthebaselinemodel(MB).Northridge
earthquake.

A comparison of the relative displacement and absolute accelerations of the

Walnut Creek MB and the measured data from the Loma Prieta earthquake at the

different floors are shown in Figures 4-23. The plots in this figure indicate that the

numerical model is a reasonable representation of the dynamic response of the

physical structure. The relative displacement and absolute acceleration profiles are

compared between the model and the recorded data in Figure 4-24. This model used a

factor of 0.3 between cracked and uncracked concrete stiffness. The stiffness

reduction of 30% was applied to the elastic modulus of the columns and shear wall

components (modeled as trusses as described in the NIST report). No further

calibration was performed and all other properties represented best estimates

developed prior to viewing analysis results and analysis-data comparisons.

136
 Because two vertical instruments are present at the ground floor level of the

Walnut Creek building, it is possible to compute base rocking as the difference

between the two verticals divided by their horizontal separation distance. As shown in

Figure 4-24, low frequency noise causes base rocking evaluated in this manner from

the CSMIP data to have unrealistic features. Accordingly, the vertical records were

filtered using high-pass Butterworth filters with a corner frequency of 0.4 Hz, which

was selected through examination of displacement histories with different corners as

shown in Figure 4-25. Following the recommendations of Bommer and Boore (2005),

the minimum filter corner that removes the low frequency noise features in the

rocking displacement history was selected.

Figure 4-26 compares the filtered base rocking from recordings to that

computed from simulations. The numerical model captures the observed rocking

reasonably well. In Figure 4-27, the rotation at the base of the numerical model was

multiplied by the vertical distance from the foundation-to-roof (39.2 m) to compute

the rocking contribution to the lateral roof displacement. The rocking-induced lateral

displacement accounts for approximately 10% of the total roof displacement.

137

Figure423 Comparison of recorded lateral acceleration and relative displacement
historiesatroof,8thfloor,3rdfloorandgroundlevelwiththosepredicted
usingtheBaselineModel(MB).LomaPrietaearthquake.

138


Figure424 Comparison of maximum floor relative displacement and absolute

acceleration between model and recorded response. Loma Prieta
earthquake.

139
Figure425 HighpassfilteringofWalnutCreekbuildingverticalmotionsatthebaselevel.Figureshowstheeffectofthecornerfrequencyon
displacementhistoriesandrockinghistoryfortheWalnutCreekbuilding.

140

Figure426 Comparison ofthe foundation rotationfrombase rocking ofthebaseline
modelandrecordeddatafortheWalnutCreekbuilding.

Figure427 Comparison of the baseline model roof displacement from base rocking
compared to the total roof displacement from the recorded data for the
WalnutCreekbuilding.

Based on the favorable comparisons of the response histories (Figure 4-23),

maximum accelerations and displacement (Figure 4-24) and contribution to roof

displacement from base rocking (Figures 4-27), the MBs provide a reasonable

141
approximation of the soil-structure system response and to provide a suitable basis for

model to model comparisons presented in the next section.

4.8 Model to Model Comparison

Table 4-11 shows the 1st mode periods for the alternative model configurations for the

Sherman Oaks and Walnut Creek structures. As expected, fixed base models have

shorter periods than flexible-base models, although the changes are relatively modest,

especially for the Sherman Oaks structure. Many of the changes in response that we

compute can be related to the change in period coupled with spectral shape near the

first mode period.

Table 4-11 Comparison of 1st Mode Periods for Alternative Model Configurations (for
both models)
Fundamental Period (sec)
Sherman Oaks Bldg., Walnut Creek Bldg., Loma
Model Type Norhtridge Earthquake Prieta Earthquake
MB (Baseline Model) 2.72 0.83
1 (fixed @ ground surface) 2.67 0.78
2 (fixed @ base) 2.71 0.78
3 (fixed horizontal spring) 2.65 N/A
4 (bathtub) 2.72 0.83
N/A = not applicable

4.8.1 Sherman Oaks Analysis Results

This section presents detailed Sherman Oaks model-to-model comparisons of the

displacement time series and peak EDPs of maximum displacements, inter-story drift,

story shear coefficients (Cs) and floor accelerations. Figures 4-28 through 4-32 show

the Sherman Oaks MB compared individually to each simplified and alternative

model discussed in Section 4.3.2.

142
 Figure 4-28 shows the bathtub model introduces changes in displacement

history response throughout the height that are negligible (e.g., less than 3.6% at roof

level). Changes in maximum inter-story drift and story shear force profiles are also

small both above and below ground level. The engineering demand parameter most

affected by model-to-model variations is peak floor accelerations, which are relatively

unaffected above ground line but are sensitive to the use of kinematic versus free-field

ground motions. This can be seen by examination of the results marked 4a and 4b

(i.e., use of uFIM in lieu of ug) in Figure 4-28. Similar trends are observed for MB.1

and MB.2 (Figure A-5 of the NIST report).

Ignoring the subterranean levels by assuming a rigid base at ground level

(Model 1) alters the vibration period of the building (Table 4-11). As a result,

displacement history responses are more out of phase than other configurations

considered as shown in Figure 4-29, although the differences are small. Differences in

above-ground story drifts, story shears, and peak floor accelerations are relatively

large in some cases (up to a 50% change).

143
Figure428 Comparison of displacement histories, drift ratios, and story shears
obtainedfromMBandModel4(bathtub)fortheShermanOaksbuilding.

144
Figure429 Comparison of displacement histories, drift ratios, and story shears
obtained from MB and Model 1 (fixed at ground line) for the Sherman
Oaksbuilding.

145
 Ignoring the embedment effect by running the structure down to a fixed base

at the foundation level (Model 2) provides good matches to MB in vibration periods

(Table 4-11), but drift, story shear, and peak floor acceleration profiles are

significantly changed in Model 2 relative to MB (Figure 4-30). The differences are

less than those for Model 1 and are generally larger below ground line than above.

The worst results were obtained with Model 3, in which upper-level

foundation springs are fixed against displacement and input motions are applied at the

base level with rocking allowed. Large differences in vibration periods, displacement

histories, and drift/shear distributions are encountered, as shown in Figure 4-31. This

model should not be used.

Figure 4-32 compares the MB results to those computed for models identical

to MB except for rigid foundation structural elements (MB.1) and gapping allowed at

the basement wall-soil interface (MB.2). These are not models used in practice, per

Chapter 6 of the NIST report, but were investigated to gain insight into the

importance of these effects in foundation modeling. The results show that allowing

geometric nonlinearities (gap formation) has no discernible impact on the results. This

supports the equivalent-linear foundation/soil modeling assumptions used in this

work. The use of a rigid foundation increases responses, likely due to period change.

In Figure 4-33, the results are synthesized from each of the considered

alternative practical configurations along with the MB as profiles of drift, inter-story

drift, and story shear for transverse response (y-direction). The Model 3 results are

clear outliers for each of the engineering demand parameters considered. Among the

other models, Model 4 is closest to MB, followed by Model 2, and then Model 1.

Differences in EDPs are generally greater below ground line than above.

146

Figure430 Comparison of displacement histories, drift ratios, and story shears
obtained from MB and Model 2 (fixed at base) for the Sherman Oaks
building.

147
Figure431 Comparison of displacement histories, drift ratios, and story shears
obtainedfromMBandModel3(fixedhorizontalsprings)fortheSherman
Oaksbuilding.

148
Figure432 Comparisonofdisplacementhistories,driftratios,andstoryshearsforMB,
MB.1(rigidfoundations),andMB.2(gaps)fortheShermanOaksbuilding.

149
Figure433 Comparison of peak displacement over height, drift ratios, and story
shears in transverse direction from all models of the Sherman Oaks
building.

150
4.8.2 Walnut Creek Analysis Results

A set of EDPs similar to those for the Sherman Oaks building were computed for the

Walnut Creek building. However, the buildings configuration only allowed for the

development of the two fixed-base modeling schemes (Model 1 and 2) for comparison

to the MB. The MB matches the bathtub model (Model 4), as described in Section

4.7.

Figure 4-34 compares roof accelerations and displacements for the three

models. The fixed-base models (Models 1 and 2) capture MB peak roof accelerations

and displacements reasonably well, but the phasing of the recorded and computed

roof motions is quite different. These phasing variations are likely due to differences

in higher mode responses. Note that Models 1 and 2 are nearly identical in this case

(unlike Sherman Oaks), because of the small embedment.

Figure 4-35 presents profiles of the peak EDPs of maximum displacement,

inter-story drift, and floor accelerations. There is a general trend towards over-

prediction of response by the fixed-base models (Models 1 and 2) relative to MB. The

differences are largest in peak floor accelerations, and likely result from differences in

higher-mode effects as a result of SSI.

151


Figure434 Comparison of the roof acceleration and displacement histories for the
reference model (MB,equivalent to Model 4)and Models 1 and2 of the
WalnutCreekbuilding.

152

Figure435 Comparison of profiles of maximum displacement, maximum interstory
drift,andpeakflooraccelerationforthereferencemodel(MB,equivalent
toModel4)andModels1and2oftheWalnutCreekbuilding.

153
4.9 Interpretation and Conclusions

Drawing upon the previous findings from Naeim et al (2008), Tileylioglu et al. (2010)

and the present work, several consistencies can be observed. First, MB-type models

have an encouraging ability to match observed responses with only modest tuning of

structural parameters (i.e., damping ratios, element masses or stiffnesses). In

particular, the SSI portion of the models appears to be performing well for each of the

investigated structures having recordings of foundation rocking (two from prior work

and Walnut Creek in the present work). These results suggest that the relatively

simple equivalent-linear spring/dashpot approach for SSI modeling presented in the

NIST report can provide satisfactory predictions of foundation response for the levels

of excitation in these case studies.

Second, there are varying levels of performance for some of the simplified

approaches used in practice. The worst-performing approach is Model 3, which

should not be used for response history analysis. Model 1 is the next least-desirable

option, with the level of difference from the MB being greatest for stiff structures or

deeply embedded structures. The errors are especially noticed in under-prediction of

building periods, variations in profiles of inter-story drifts, story shears and peak floor

accelerations, and phasing errors in response histories. The best-performing models

relative to MB are Model 2 and Model 4 (bathtub).

Our third general finding concerns the portions of a structure whose response

is significantly affected by the SSI modeling. For the buildings considered, above-

ground building responses, as measured for example by envelopes such as drift or

story shear profiles, were modestly to insignificantly affected by SSI. The largest

effects of SSI on above-ground enveloped responses were for the stiffest structures. In

154
contrast, below-ground enveloped responses (e.g. basement wall shear forces and

below-ground inter-story drifts) were much more sensitive to SSI. Both the kinematic

ground motion description and spring distributions contribute to these effects. This

sensitivity was observed for all structures, whether flexible or relatively stiff.

Considering all of the above, it is observed that none of the simplified

approaches either currently used or proposed for use in practice (Models 1-4 in Figure

4-3) can mimic the MB response over the full height of the structure for the range of

structural conditions that have been considered to date. For flexible structures with

modest embedment, any approach other than Model 3 would likely provide acceptable

results above the ground line. More generally, the bathtub approach (Model 4)

appears to be the most generally applicable and reliable among the simplifications

considered. However, even Model 4 does a poor job below the ground line because

the application of depth-variable ground motion or kinematic rocking can

significantly affect EDPs for deeply embedded structures (approximately two levels

or more). When it is desired to avoid the use of foundation springs, Model 2 has been

shown to provide reasonably good results, especially for embedded, flexible structures

and EDPs associated with the portion of the structure above the ground line.

155
5.0 FORCEDVIBRATIONTESTINGOFA
SHALLOWFOUNDATIONFIELD
TESTSTRUCTURE

5.1 Introduction

Following the devastating 1971 San Fernando earthquake, the engineering community

realized the need to instrument buildings with accelerometers to study their seismic

response. In 1972, the California Legislature established the California Strong Motion

Instrumentation Program (CSMIP) in which hundreds of buildings have been

instrumented with arrays of accelerometers. The US Geological Survey also

instruments buildings, including the influential ANSS program currently in operation.

Unfortunately, configurations of the arrays in a manner to enable the evaluation of

SSI is (in general) poorly understood and has not typically been a considered a high

priority. The evaluation of SSI effects on buildings requires the horizontal

accelerometers at the free-field, foundation, and roof and two vertical accelerometers

on the foundation to evaluate rocking response (Stewart and Fenves, 1998). There are

about 12 buildings in California that meet all of these criteria, although approximately

60 meet enough of the criteria that some SSI effects can be evaluated (Stewart et al.,

1999b). These structures have variable levels of inertial and kinematic interaction

effects, from strong to practically negligible. There is some indication that CSMIP

156
and USGS may improve their instrumentation arrays to include SSI-type sensor

implementation plans, and data from such structures is particularly useful for response

history analysis studies, investigations of period lengthening and foundation damping,

and investigations of kinematic interaction.

One issue that is not easily investigated with building seismic arrays is

inference of impedance functions. As explained by Tileylioglu et al. (2011), the

measurement of horizontal impedance requires relative displacements between the

foundation and free-field. Such measurements will be highly noisy in a seismic

setting, but can be effectively made in forced vibration testing of structures in the

field.

Forced vibration testing of a steel and reinforced concrete structure has been

performed to evaluate soil-structure interaction (SSI) systems at multiple George E.

Brown, Jr. Network for Earthquake Engineering Simulations (NEES) sites. This work

has been completed as part of a multi-institutional National Science Foundation

(NSF) grand challenge (NSF 06-504 Program) project called, Mitigation of Collapse

Risk in Vulnerable Concrete Buildings. This project has been documented in the

Network for Earthquake Engineering Simulations Research (NEESR) website, known

as NEEShub, as project NEES-2008-0637. The SSI test series consisted of three test

sites: University of California, Los Angeles Laboratory (LAB), Wildlife Liquefaction

Array (WLA), and Garner Valley Downhole Array (GVDA). This NSF program

presented a unique opportunity for research in that until this series of tests no single

structure had been tested with different types of underlying materials to evaluate their

influence to the structural response.

157
 The sequence of testing at the three sites investigates the response of SSI

systems involving varying degrees of base flexibility. The LAB represents an

effectively fixed-base condition with structural columns attached to the existing

laboratory strong-floor. The WLA site, with a soft clay layer overlying loose,

liquefiable sands, and the GVDA site, with firm lake-bed alluvium material,

represents flexible base conditions having variable levels of soil stiffness. The three

sites were chosen to represent a spectrum of SSI effects based on the foundation

conditions, with practically no SSI effect expected at the LAB to moderate effects at

the GVDA site, to a strong SSI influence at the WLA.

This chapter discusses the forced vibration testing performed at GVDA in

which the author was involved in the field testing. The forced vibration testing at

WLA and LAB was presented by Star (2011). This chapter first describes the field

test setup and instrumentation used at GVDA. Secondly, the forced vibration testing

schemes and data processing types are discussed. Then a review of the structural

response histories from the forced vibration testing is presented in Section 5.4. The

chapter finishes with some concluding remarks about the forced vibration testing

performed at GVDA.

5.2 Test Setup and Instrumentation

5.2.1 Garner Valley Downhole Array (GVDA) Test Site

The Garner Valley Downhole Array (GVDA) site has been well studied since 1989

and is well documented at the NEES@UCSB site

(http://nees.ucsb.edu/facilities/gvda). Figure 5-1 shows a plan view of the site. The

ground surface elevation is approximately 1,329 m (4,360 ft) above mean sea level

(AMSL). Documents indicate relatively consistent soil conditions, consisting of 18.3

158
m (60 ft) to 25.3 m (83 ft) of thick lake-bed alluvium overlying decomposed granite

that extends to about 91.4 m (300 ft) where weathered granodiorite bedrock is

encountered. Figure 5-2 shows our interpretation of the soil layering at the site based

on the NEES@UCSB data. This site was chosen based on its accessibility, existing

instrumentation and infrastructure, the availability of site data, and its distinct

geotechnical conditions relative to the other two NEES sites considered in prior work

(WLA and LAB).

Previously published geophysical data for the site has been developed from

Spectral-Analysis-of-Surface-Waves (SASW) measurements (Stokoe et al. 2004).

Locations of SASW tests are shown in Figure 5-1; the resulting interpretation of the

shear wave velocity (Vs) profile is shown in Figure 5-2. Based on the NEES@UCSB

data, the average moist unit weight of the soil is approximately 17.3 kN/m3 (110

lb/ft3) to 19.6 kN/m3 (125 lb/ft3). The water table is typically at approximately 3 m

(10 ft) below the ground surface.

159
Figure51 PlanviewoftheGVDAsiteshowingtheapproximatelocationsofthe
SASWtestlocations.

Figure52 SubsurfacecharacteristicsbasedonsiteexplorationoftheGVDAsite:a)
shearwavevelocityprofileandb)materialprofile.

160
 Crosshole seismic surveys were completed as part of this study to evaluate the

soil shear wave velocity within the upper 2 m (6.6 ft) of the soil column below the

structural footprint. The testing was performed on May 26, 2011 and the resulting soil

shear wave velocity profiles are shown in Figure 5-3. The survey was conducted in

two phases in an effort to evaluate the influence of the structural weight. Initial

crosshole testing was performed across the structural footprint with only the weight of

the foundation. Secondly, testing was performed following installation of the full

structure. In addition to the measured results, Figure 5-3 provides velocity profile

model names that will be used for analyses presented in Chapter 6.

Figure53 Soilshearwavevelocityprofilesbasedoncrossholeseismicsurveys
performedonMay26,2011.

Crosshole testing in which the weight of the structure is acting on the soil

between the geophones produced an average velocity of Vs = 193.5 m/s (634.8 ft/s),

which is referred to as Model 1. Testing with the foundation, but not the structure,

in place produced a slower velocity of Vs = 187.1 m/s (537.6 ft/s). Based on these

results, the scaling of Vs with overburden pressure can be evaluated as follows

(adapted from Tileylioglu et al., 2011):

161
 n/2
(V s )1 v 0 + ( v )1
= (5-1)
(V s )2 v 0 + ( v )2

where ( Vs )1 and ( Vs )2 are the overburden-corrected shear wave velocity for a

particular depth z, v0 is the effective stress from the soils self-weight at depth z, and

(v)1 and (v)2 are the increment of vertical stress at depth z from varying structural

weights (e.g., for combined foundation/structure and foundation alone, respectively).

The v quantities can be computed using classical Boussinesq-type stress

distribution theory (e.g., Fadum, 1948). The n value typically varies from

approximately 0.5 for granular soils to 1.0 for cohesive soils with plasticity index PI >

6.5 (Yamada et al., 2008). A site-specific n value of 0.6 was back-calculated for a

depth of 1.5 m (5 ft) below ground surface (BGS) from Eq. (5-1) using the

aforementioned crosshole velocities and conditions where v0 = 25 kN/m2, (v)1 =

10 kN/m2 and (v)2 = 6.3 kN/m2. The GVDA site-specific n value determined for

sandy and silty materials corresponds well to the Yasuda et al., 2008 recommended

range of n values.

The velocity that is expected at the site prior to foundation placement (referred

to as Model 2) was derived in a similar manner to Eq. (5-1):

n/ 2
+ v
V s V s0 v0 (5-2)
v 0

where Vso denotes the shear wave velocity measured in the free-field. The back-

calculated Vso profile is shown in Figure 5-2. The Vso = 175.5 m/s (575.8 ft/s) was

back-calculated for a depth of 1.5 m (5 ft) BGS from Eq. (5-2) using the site-specific

n = 0.6, Vs = 193.5 m/s (634.8 ft/s), v0 = 25 kN/m2 and v = 10 kN/m2.

162
5.2.2 Test Structures

The test structure that was used at all test sites for forced vibration testing is referred

to as Mini-Me. Mini-Me is a rectangular shaped steel moment frame with

removable cross-bracing consisting of reinforced concrete roof and foundation, as

shown in Figure 5-4. Unembedded reinforced concrete foundations were constructed

at each field tests site and were cast-in-place to ensure realistic foundation-soil

contact. The foundations were 0.61 m (2 ft) thick, 4.28 m (14 ft) long and 2.13 m (7

ft) wide and were designed to be nearly rigid. The foundation mat is composed of

nominal 27.6 MPa (4,000 psi) concrete reinforced with #5 rebar. Prior to construction

of the foundation, the pad was cleared and grubbed to remove surficial top soils and

organics. The foundation mat was constructed on site and had more than 28 days to

cure before testing. Based on the dimensions, the foundation has an estimated mass of

13,340 kg (29,409.4 lbs).

Nine 2.54 cm (1) diameter threaded rods were used to anchor the base plate

of each column of the super structure to the foundation. The anchors were pre-drilled

and secured using epoxy. The upper deck (roof) is composed of nominal 27.6 MPa

(4,000 psi) concrete reinforced with #7 rebar and has similar dimensions to the

foundation mat. The braced frame columns are constructed from four 30.5 cm x 30.5

cm x 1.3 cm (12x12x1/2) square hollow steel sections (HSS) that are 2.13 m (7 ft)

tall and have a mass of approximately 250 kg (551 lbs) each. The super structure was

constructed in September, 2008 and was transported from site-to-site. Removable

cross-bracing can be added to the structure in order to modify the stiffness of the

structure system. The braces are constructed from HSS 10.2 cm x 10.2 cm x 1.3 cm

(4 x 4x 1/2) steel sections. The total mass of the braces is approximately 800 kg

163
(1,763.7 lbs). The super structure and foundation together are 3 m (9.8 ft) tall. Braced

and unbraced conditions were investigated as part of this experiment. As of this

writing, the structure still exists and is present at GVDA.

Figure54 ConstructionplancrosssectionoftheteststructureMiniMe.

The GVDA site was previously improved with a similar type test structure for

evaluate of SSI effects, referred to here as Dr. Evil. Dr. Evil is a square shaped steel

moment frame with removable cross-bracing consisting of reinforced concrete roof

and foundation. Details on the dimensions and mass of this structure are provided in

Tileylioglu et al. (2011) and at NEES@UCSB. During the testing period of this

164
experiment Dr. Evil was unbraced. Dr. Evil is situated approximately 4.6 m (14 ft) to

the south of Mini-Me. The construction materials are similar to Mini-Me.

The construction drawings for Mini-Me are located in the NEEShub for this

project. The foundation geometry adopted for testing is shown in Figure 5-5.

Vibration modes for translation is represented by x, y and z, while rotation about

theses axis, respectively, are xx, yy and zz. Figure 5-6 shows a photograph of the test

structures at GVDA. With the exception of the foundation, the same Mini-Me super

structure (columns, bracing, and roof) was used at GVDA, WLA, and LAB.

Figure55 Sketchshowingthegeometryadaptedfortherectangularfoundation
(L B ) .

Figure56 PhotographshowingtheteststructureMiniMe(right)andthepreviously
constructedstructureDr.Evil(left)locatedatGVDA

165
5.2.3 Sensors

In an effort to capture the response of the structure during the forced vibration testing

at GVDA a total of 42 Episensor accelerometers were installed to monitor the

response of Mini-Me, Dr. Evil and the adjacent soil. Figure 5-7 shows the sensor

location plan along with a structure location plan. Mini-Me was instrumented with

eight Kinemetrics Episensor ES-T triaxial accelerometers. The accelerometers are

13.3 cm (5.2 in) diameter (cylinder) and 6.2 cm (2.4 in) high. The accelerometers

have a selected output range of +/- 20 V, for an amplitude range of +/- 4.0 g. The

bandwidth is DC to 200 Hz. The accelerometers have a calibrated sensitivity of 5

Volts per g. One accelerometer was bolted in place at each corner of the structure top

slab and foundation slab. Four Trans-Tek inc. model 0244-0000 DC_LVDTs (linear

variable differential transformers) were installed near the corners of Mini-Mes

foundation mat. The LVDTs have a measurement range of 5 cm (2 in). Dr. Evil was

instrumented with four triaxial accelerometers. Additional triaxial accelerometers

were installed in the soil near the structure. One was buried at the elevation of the

foundation base in the longitudinal direction relative to the orientation of the structure

(positive x-direction) and one in the direction transverse to the orientation of the

structure (positive y-direction). All accelerometers sampled at a rate of 200 Hz.

166
Figure57 Sketchshowingthestructureslocationandsensorarraysfortestingat
GVDA.

The shaker forces were prescribed via the control system, however, it has been

found that the prescribed input does not always equal the actual shaker response.

167
Therefore additional accelerometers were used to evaluate the input shaker force and

ensure proper calibration for analysis. Two shakers were used for this experiment,

which will be explained in detail in Section 5.2.4; a small linear shaker was

instrumented with a single accelerometer in the direction of shaking and a larger

eccentric mass shaker. During testing with the larger shaker two horizontal

accelerometers were installed parallel to the existing accelerometers at the southwest

corner of Mini-Me. Figure 5-7 shows the location of these two accelerometers in blue.

This was done in order to allow the correlation of data on two separate acquisition

systems.

The sensor implementation plan was prepared with support of NEES@UCLA.

Details pertaining to the Episensor accelerometers and LVDTs can be found at

http://nees.ucla.edu/equipment.html.

5.2.4 Forced Vibration Shakers

Mini-Me was subjected to forced vibration testing using two types of shakers. A small

shaker referred to as Atom-Ant, was used for x, y and torsion modal forced

vibration testing at the roof and top of foundation mat levels. A larger shaker referred

to as Mighty Mouse, was used for x and y modal forced vibration testing at the roof

level. The forced vibration testing was performed on June 22 through June 24, 2011

and August 1 through August 2, 2011.

Atom Ant is a small portable Acoustic Power System ELECTRO-SEIS long

stroke shaker model 400. The shaker characteristics are described here because it is

not NEES@UCLA equipment documented on the web site. Atom Ant is a 7.3 kg (161

lbs) shaker that has a maximum force of 444.8 N (100 lbf). The shaker was controlled

by an APS Dynamic Inc. dual mode power amplifier model 144 and a HP Agilent

168
33220A, 2 Mhz function / arbitrary waveform generator. Forcing frequencies were

controlled through the generator output by two methods: sweeps and steps. Sweeps

consist of producing a consistent scan of frequency output that was controlled

automatically for defined ranges. Sweeps frequencies ranged from 4 Hz to a

maximum of 54 Hz. Steps consisted of producing constant frequencies in short

sustained intervals at 0.5 Hz steps for the same frequency range as the sweeps. The

steps were produced for redundancy to present individual frequency point

comparisons to the sweep data.

Mighty Mouse is a large NEES 4600A eccentric mass shaker that consists of

two sets of circular weights. The eccentricity of the circular weights can be adjusted

from 0 to 110.7 N-m (980 lb-in.) and is intended for a frequency range of 0 to 20 Hz.

The force is proportional to the eccentricity. Forcing frequencies were controlled by

the Mobile Command Center using LabVIEW (described in Sec 5.2.5). For

consistency with the forced vibration testing performed with the Atom Ant shaker,

sweeps and steps were performed. Sweeps frequencies ranged from 0 to a maximum

of 19 Hz. Steps consisted of short sustained intervals at 0.2 Hz steps for the same

frequency range as the sweeps.

Atom Ant was on loan from Geovision Inc. during the testing phase and

Mighty Mouse is maintained and operated by NEES@UCLA.

5.2.5 Data Acquisition Setup

Data collection was performed with the Mobile Command Center operated by

NEES@UCLA. The sensors responses were collected by eight (8) Quanterra Q330

wireless data loggers. The Q330s have 6-channels, global positioning system (GPS)

receivers, and a dynamic range of 135 dB RMS S/N with a time stamp accuracy of

169
<0.1 ms. The Q330 data was then collected using Rockhound data acquisition

software in the Mobile Command Center located onsite. The sensor sampling rate was

200 Hz. The GVDA data acquisition channels have been described in Table 5-1 and

are presented in Figure 5-7. In this report, the data from this acquisition system is

referred to as structural response data.

During forced vibration testing using Mighty Mouse, the shaker pulse and

correlation accelerometer data was recorded using a DAQ which had a sampling rate

of 5000 Hz. The higher sampling rate is necessary in order to determine phasing of

the shaker force with adequate precision. The data acquisition (DA) channels used for

data collection pertaining to testing with the Mighty Mouse shaker is described in

Table 5-2. In this report, the data from this acquisition system is referred to as

Mighty Mouse shaker data. The Mighty Mouse shaker data includes two

accelerometers used for correlation of the two different data acquisition systems. The

accelerometer locations are presented in Figure 5-7 as blue arrows, parallel to

accelerometers designated as channels 15 and 16 in the structural response data. The

correlation method used is described in Section 5.3.3. Two tables have been presented

for Table 5-2 because channel 4 and 5 were switched from the June to August testing.

The prefix MM is used to distinguish the Mighty Mouse shaker data channels from

the Structural Response data channels, shown in column 2 of Table 5-2.

170
Table 5-1 GVDA Sensor Locations plan for all forced vibration testing located at Garner Valley Downhole Array site in June and August of 2011.
ChannelName Position Orientation CalibrationConstant
DAChannelNumber SensorLabel SensorType Comments PositionUnits CoordinateSpace SerialNumber SensorMeasurementUnits
Corrected/Converted X Y Z I J K Value Units
1 1 Time Position RawDatahasUnixtime 228 252 0 in 0 0 0 GlobalCoordinateSpace Time s 1 s
2 2 MMBottom_SW_Z Accelerometer 136.5 78 24 in 0 0 1 GlobalCoordinateSpace 1193 count 0.0000119 G/count
3 3 MMBottom_SW_Y Accelerometer 136.5 78 24 in 0 1 0 GlobalCoordinateSpace 1193 count 0.0000119 G/count
4 4 MMBottom_SW_X Accelerometer 136.5 78 24 in 1 0 0 GlobalCoordinateSpace 1193 count 0.0000119 G/count
5 5 MMBottom_NW_Z Accelerometer 136.5 6 24 in 0 0 1 GlobalCoordinateSpace 1700 count 0.0000119 G/count
6 6 MMBottom_NW_Y Accelerometer 136.5 6 24 in 0 1 0 GlobalCoordinateSpace 1700 count 0.0000119 G/count
7 7 MMBottom_NW_X Accelerometer 136.5 6 24 in 1 0 0 GlobalCoordinateSpace 1700 count 0.0000119 G/count
8 8 MMBottom_NE_Z Accelerometer 31.5 6 24 in 0 0 1 GlobalCoordinateSpace 1550 count 0.0000119 G/count
9 9 MMBottom_NE_Y Accelerometer 31.5 6 24 in 0 1 0 GlobalCoordinateSpace 1550 count 0.0000119 G/count
10 10 MMBottom_NE_X Accelerometer 31.5 6 24 in 1 0 0 GlobalCoordinateSpace 1550 count 0.0000119 G/count
11 11 MMBottom_SE_Z Accelerometer 31.5 78 24 in 0 0 1 GlobalCoordinateSpace 1701 count 0.0000119 G/count
12 12 MMBottom_SE_Y Accelerometer 31.5 78 24 in 0 1 0 GlobalCoordinateSpace 1701 count 0.0000119 G/count
13 13 MMBottom_SE_X Accelerometer 31.5 78 24 in 1 0 0 GlobalCoordinateSpace 1701 count 0.0000119 G/count
14 14 MMTop_SW_Z Accelerometer 151.5 67.5 118 in 0 0 1 GlobalCoordinateSpace 1444 count 0.0000119 G/count
15 15 MMTop_SW_Y Accelerometer 151.5 67.5 118 in 0 1 0 GlobalCoordinateSpace 1444 count 0.0000119 G/count
16 16 MMTop_SW_X Accelerometer 151.5 67.5 118 in 1 0 0 GlobalCoordinateSpace 1444 count 0.0000119 G/count
17 17 MMTop_NW_Z Accelerometer 151.5 16.5 118 in 0 0 1 GlobalCoordinateSpace 1443 count 0.0000119 G/count
18 18 MMTop_NW_Y Accelerometer 151.5 16.5 118 in 0 1 0 GlobalCoordinateSpace 1443 count 0.0000119 G/count
19 19 MMTop_NW_X Accelerometer 151.5 16.5 118 in 1 0 0 GlobalCoordinateSpace 1443 count 0.0000119 G/count
20 20 MMTop_NE_Z Accelerometer 16.5 16.5 118 in 0 0 1 GlobalCoordinateSpace 1190 count 0.0000119 G/count
21 21 MMTop_NE_Y Accelerometer 16.5 16.5 118 in 0 1 0 GlobalCoordinateSpace 1190 count 0.0000119 G/count
22 22 MMTop_NE_X Accelerometer 16.5 16.5 118 in 1 0 0 GlobalCoordinateSpace 1190 count 0.0000119 G/count
23 23 MMTop_SE_Z Accelerometer 16.5 67.5 118 in 0 0 1 GlobalCoordinateSpace 1883 count 0.0000119 G/count
24 24 MMTop_SE_Y Accelerometer 16.5 67.5 118 in 0 1 0 GlobalCoordinateSpace 1883 count 0.0000119 G/count
25 25 MMTop_SE_X Accelerometer 16.5 67.5 118 in 1 0 0 GlobalCoordinateSpace 1883 count 0.0000119 G/count
26 26 DEBottom_SW_Z Accelerometer 158.7 394 24 in 0 0 1 GlobalCoordinateSpace 1702 count 0.0000119 G/count
27 27 DEBottom_SW_Y Accelerometer 158.7 394 24 in 0 1 0 GlobalCoordinateSpace 1702 count 0.0000119 G/count
28 28 DEBottom_SW_X Accelerometer 158.7 394 24 in 1 0 0 GlobalCoordinateSpace 1702 count 0.0000119 G/count
29 29 DEBottom_NW_Z Accelerometer 158.7 270.5 24 in 0 0 1 GlobalCoordinateSpace 1192 count 0.0000119 G/count
30 30 DEBottom_NW_Y Accelerometer 158.7 270.5 24 in 0 1 0 GlobalCoordinateSpace 1192 count 0.0000119 G/count
31 31 DEBottom_NW_X Accelerometer 158.7 270.5 24 in 1 0 0 GlobalCoordinateSpace 1192 count 0.0000119 G/count
32 32 DEBottom_NE_Z Accelerometer 18 255.3 24 in 0 0 1 GlobalCoordinateSpace 3849 count 0.0000119 G/count
33 33 DEBottom_NE_Y Accelerometer 18 255.3 24 in 0 1 0 GlobalCoordinateSpace 3849 count 0.0000119 G/count
34 34 DEBottom_NE_X Accelerometer 18 255.3 24 in 1 0 0 GlobalCoordinateSpace 3849 count 0.0000119 G/count
35 35 DETop_W_Z Accelerometer 158.7 264 117 in 0 0 1 GlobalCoordinateSpace 1445 count 0.0000119 G/count
36 36 DETop_W_Y Accelerometer 158.7 264 117 in 0 1 0 GlobalCoordinateSpace 1445 count 0.0000119 G/count
37 37 DETop_W_X Accelerometer 158.7 264 117 in 1 0 0 GlobalCoordinateSpace 1445 count 0.0000119 G/count
38 38 FieldW_Z Accelerometer 336 42 1 in 0 0 1 GlobalCoordinateSpace 1442 count 0.0000119 G/count
39 39 FieldW_Y Accelerometer 336 42 1 in 0 1 0 GlobalCoordinateSpace 1442 count 0.0000119 G/count
40 40 FieldW_X Accelerometer 336 42 1 in 1 0 0 GlobalCoordinateSpace 1442 count 0.0000119 G/count
41 41 FieldS_Z Accelerometer 84 252 1 in 0 0 1 GlobalCoordinateSpace 1446 count 0.0000119 G/count
42 42 FieldS_Y Accelerometer 84 252 1 in 0 1 0 GlobalCoordinateSpace 1446 count 0.0000119 G/count
43 43 FieldS_X Accelerometer 84 252 1 in 1 0 0 GlobalCoordinateSpace 1446 count 0.0000119 G/count
44 44 SWcorner LVDT MultConv.Databy10 153 85 22 in 0 0 1 GlobalCoordinateSpace D29 count 3.08E07 in/count
45 45 NWcorner LVDT 153 1 22 in 0 0 1 GlobalCoordinateSpace D8 count 2.36E07 in/count
46 46 NEcorner LVDT MultConv.Databy10 15 1 22 in 0 0 1 GlobalCoordinateSpace D15 count 2.43E07 in/count
47 47 SEcorner LVDT MultConv.Databy10 15 85 22 in 0 0 1 GlobalCoordinateSpace D17 count 2.50E07 in/count
48 48 AAWeightAcc Accelerometer 42 84 118 in 1 1 0 GlobalCoordinateSpace 548 count 0.0000119 G/count
* MM: Mini-Me test structure; AA: Atom Ant Shaker; DE: Dr. Evil test structure; Bottom: sensor located on bottom (foundation) slab; Top: sensor located on top (roof) slab;
* SW, SE, NE, NW: Relative directional in "Sensor Label" and locatin for "Position," based on referenced north in Structures Location Plan (Figure 7);
*X, Y, Z: Sensor direction and location based on global coordinate system shown in the Structures Location Plan (Figure 7)
*G: Gravity; in: inches; s: seconds; LVDT: linear variable differential transformer; I, J, K: Represents orientation of sensor, respectively, in X, Y, and Z direction
*Note converted LVDT data (Chs. 44, 46, and 47) has an error that can be corrected by multipling processed channels by a factor of 10.

171
Table 5-2a Mighty Mouse Shaker Data plan for vibration testing using Mighty Mouse shaker located at Garner Valley Downhole Array site in June
of 2011.
ChannelName Position Orientation CalibrationConstant
DAChannelNumber SensorLabel SensorType Comments PositionUnits CoordinateSpace SerialNumber SensorMeasurementUnits
Corrected/Converted X Y Z I J K Value Units

1 MM1 Time Position 228 252 0 in 0 0 0 GlobalCoordinateSpace Time s 1 s

Correlationtochannel
MMTop_SW_
2 MM2 Accelerometer 16ofGVDAMiniMe 163 67.5 118 in 1 0 0 GlobalCoordinateSpace 551 count 0.0000119 G/count
X_Correlation
SensorLocations
Correlationtochannel
MMTop_SW_
3 MM3 Accelerometer 15ofGVDAMiniMe 151.5 82 118 in 0 1 0 GlobalCoordinateSpace 542 count 0.0000119 G/count
Y_Correlation
SensorLocations
MMControl
4 MM4 N/A 1PulsePerRevolution 228 252 0 in 0 0 0 GlobalCoordinateSpace Pulse1 V 1 V
Frequecy

5 MM5 ShakerPulses N/A 10PulsesPerRevolution 228 252 0 in 0 0 0 GlobalCoordinateSpace Pulse10 V 1 V

* MM: Mini-Me test structure; Top: sensor located on top (roof) slab;
* SW Relative southwest location based on referenced north in Structures Location Plan (Figure 5-7);
*X, Y, Z: Sensor direction in "Senosr Label" and location for "Position," based on global coordinate system shown in the Structures Location Plan (Figure 7)
*G: Gravity; V: volts; in: inches; s: seconds; N/A: not applicable; I, J, K: Represents orientation of sensor, respectively, in X, Y, and Z direction

Table 5-2b Mighty Mouse Shaker Data plan for vibration testing using Mighty Mouse shaker located at Garner Valley Downhole Array site in
August of 2011.
ChannelName Position Orientation CalibrationConstant
DAChannelNumber SensorLabel SensorType Comments PositionUnits CoordinateSpace SerialNumber SensorMeasurementUnits
Corrected/Converted X Y Z I J K Value Units

1 MM1 Time Position 228 252 0 in 0 0 0 GlobalCoordinateSpace Time s 1 s

Correlationtochannel
MMTop_SW_
2 MM2 Accelerometer 16ofGVDAMiniMe 163 67.5 118 in 1 0 0 GlobalCoordinateSpace 551 count 0.0000119 G/count
X_Correlation
SensorLocations
Correlationtochannel
MMTop_SW_
3 MM3 Accelerometer 15ofGVDAMiniMe 151.5 82 118 in 0 1 0 GlobalCoordinateSpace 542 count 0.0000119 G/count
Y_Correlation
SensorLocations

4 MM4 ShakerPulses N/A 10PulsesPerRevolution 228 252 0 in 0 0 0 GlobalCoordinateSpace Pulse10 V 1 V

MMControl
5 MM5 N/A 1PulsePerRevolution 228 252 0 in 0 0 0 GlobalCoordinateSpace Pulse1 V 1 V
Frequecy
* MM: Mini-Me test structure; Top: sensor located on top (roof) slab;
* SW Relative southwest location based on referenced north in Structures Location Plan (Figure 7);
*X, Y, Z: Sensor direction in "Senosr Label" and location for "Position," based on global coordinate system shown in the Structures Location Plan (Figure 5-7)
*G: Gravity; V: volts; in: inches; s: seconds; N/A: not applicable; I, J, K: Represents orientation of sensor, respectively, in X, Y, and Z direction

172
5.3 Trials and Data Processing Procedures

The experiment consists of four (4) trials distinguished by the type of loading (i.e.,

type of shaker) and structural stiffness (braced or unbraced condition) as follows:

Trial 1: Braced Structure, Atom Ant Shaker

Trial 2: Braced Structure, Mighty Mouse Shaker
Trial 3: Unbraced Structure, Atom Ant Shaker
Trial 4: Unbraced Structure, Mighty Mouse Shaker

The forced vibration testing where Mini-Me was braced was performed in

June 2011, while the unbraced condition was tested in August 2011.

Each trial consisted of forced vibration testing of Mini-Me in multiple modal

directions (x, y, and torsion). All data has been uploaded to the NEEShub. Each trial

contains unprocessed data, converted data and corrected data files.

This section begins with a discussion of the field test logs, which have labeling

corresponding to the data files located at the NEEShub. Section 5.3.2 describes the

different file types and the processes taken in preparation of the data files from

unprocessed data to corrected data. Section 5.3.3 describes how the Mighty Mouse

shaker and structural response data acquisition systems were synchronized. Then

Sections 5.3.4 through 5.3.6 describe the specific data file types that are located at the

NEEShub in detail.

5.3.1 Field Test Logs

Tables 5-3 through 5-6 show field testing logs for the forced vibration testing of Mini-

Me that were performed on June 22 through June 24, 2011 and August 1 through

August 2, 2011. Mini-Me was in a braced condition during the June testing, where

both Atom Ant and Mighty Mouse shakers were utilized. The August testing

173
consisted of forced vibration testing of Mini-Me in an unbraced condition, where both

Atom Ant and Mighty Mouse shakers were utilized.

The highlighted blue rows in Tables 5-3 through 5-6 represent the unprocessed

data files. These unprocessed data files contain several different forced vibration

testing runs. The underlying non-highlighted rows represent the subevents that were

determined by shaking direction, shaking force (or shaker type), shaking duration, and

shaker pulsing type (steeps or sweeps) as indicated in the loading description column.

Each subevent was categorized with a subevent identification number based on the

test location, trial, order location in the series of testing, and repetitions. The three test

locations have the following numbering 1=LAB, 2=WLA and 3=GVDA. The trial

number corresponds to the aforementioned list of bracing and shaking type condition.

The order number was based on the chronology of testing in a particular trial. If the

subevent labeling has a letter following the order number, this letter represents the

order of repetition of the type of forced vibration force, duration and pulsing type. For

example, in Table 5-3, Exp.3.1.1b represents experiment at GVDA, trial 1, first forced

vibration test of the trial, and b represents redundancy with this test being the second

of its type.

174
Table 5-3 GVDA forced vibration testing logs for testing on June 22, 2011 (Trial 1: Mini-Me in braced condition with forcing from Atom Ant
shaker).
Local Time Recorded in Field

Sweep Snipped
Recorded Data File Subevent File Snipped Unix
Subevent # Local Date Description Loading Description GMT First Sample Last Sample Frequency Subevent
Name Name time
Range Time Length

- 6/22/2011 20110622161102.GV3
1a " " Exp3.1.1a.txt Braced X AA sweep top 17:10 10:10:06 AM 4-54 Hz 50 s 1308762590
- 20110622171102.GV3
1b " Exp3.1.1b.txt Braced X AA sweep top 17:12 10:12:00 AM 4-54 Hz 50 s 1308762710
1c " Exp3.1.1c.txt Braced X AA sweep top 17:13 10:13:50 AM 4-54 Hz 50 s 1308762800
1d " Exp3.1.1d.txt Braced X AA sweep top 17:15 10:15:30 AM 4-54 Hz 200 s 1308762930
1e " Exp3.1.1e.txt Braced X AA sweep top 17:19 10:19:30 AM 4-54 Hz 200 s 1308763175
2 " Exp3.1.2.txt Braced X AA step 0.5 Hz top 17:24 10:24:00 AM 10:41:15 AM 4-54 Hz 1035 s 1308763440
3a " Exp3.1.3a.txt no rope on shaker X AA sweep top 17:48 10:48:30 AM 4-54 Hz 50 s 1308764900
3b " Exp3.1.3b.txt Braced X AA sweep top 17:49 10:49:30 AM 4-54 Hz 50 s 1308764960
- 20110622181102.GV3
4a " Exp3.1.4a.txt Braced Y AA torsion 42" sweep top 18:12 11:12:30 AM 4-54 Hz 50 s 1308766338
4b " Exp3.1.4b.txt Braced Y AA torsion 42" sweep top 18:14 11:14:00 AM 4-54 Hz 50 s 1308766428
4c " Exp3.1.4c.txt Braced Y AA torsion 42" sweep top 18:15 11:15:15 AM 4-54 Hz 50 s 1308766503
4d " Exp3.1.4d.txt Braced Y AA torsion 42" sweep top 18:22 ambient 11:22:15 AM 4-54 Hz 200 s 1308766925
4e " Exp3.1.4e.txt Braced Y AA torsion 42" sweep top 18:26 11:26:00 AM 4-54 Hz 200 s 1308767150
5 " Exp3.1.5.txt Braced Y AA torsion step 0.5 Hz top 18:30 11:30:30 AM 4-44 Hz 1000 s 1308767405
6a " Exp3.1.6a.txt Braced Y AA sweep top 18:59 11:59:45 AM 4-54 Hz 50 s 1308769173
6b " Exp3.1.6b.txt Braced Y AA sweep top 19:01 12:01:00 PM 4-54 Hz 50 s 1308769248
6c " Exp3.1.6c.txt Braced Y AA sweep top 19:02 12:02:15 PM 4-54 Hz 50 s 1308769323
6d " Exp3.1.6d.txt Braced Y AA sweep top 19:03 12:03:30 PM 4-54 Hz 200 s 1308769400
6e " Exp3.1.6e.txt Braced Y AA sweep top 19:07 12:07:30 PM 4-54 Hz 200 s 1308769640
- 20110622191102.GV3
7 " Exp3.1.7.txt Braced Y AA step 0.5 Hz top 19:12 12:12:30 PM 12:25:00 PM 4-44 Hz 800 s 1308769920
- 20110622201102.GV3
8a " Exp3.1.8a.txt Braced, Shaker untied X AA sweep bottom 20:21 1:21:15 PM 4-54 Hz 50 s 1308774063
8b " Exp3.1.8b.txt Braced, Shaker untied X AA sweep bottom 20:26 1:26:00 PM 4-54 Hz 50 s 1308774348
8c " Exp3.1.8c.txt Braced, Shaker untied X AA sweep bottom 20:27 1:27:30 PM 4-54 Hz 50 s 1308774438
8d " Exp3.1.8d.txt Braced, Shaker untied X AA sweep bottom 20:29 1:29:25 PM 4-54 Hz 50 s 1308774553
8e " Exp3.1.8e.txt Braced, Shaker untied X AA sweep bottom 20:31 1:31:50 PM 4-54 Hz 200 s 1308774705
8f " Exp3.1.8f.txt Braced, Shaker untied X AA sweep bottom 20:35 1:35:50 PM 4-54 Hz 200 s 1308774945
8g " Exp3.1.8g.txt Braced, Shaker untied X AA sweep bottom 20:39 1:39:50 PM 4-54 Hz 200 s 1308775185
- " Void Braced, Shaker untied X AA step .5 Hz bottom 20:46 1:46:30 PM Void 4-44 Hz N/A VOID
9 " Exp3.1.9.txt Braced, Shaker untied X AA step .5 Hz bottom 20:50 1:50:55 PM 2:09:45 PM 4-44 Hz 1100 s 1308775900
- 20110622211102.GV3
10a " Exp3.1.10a.txt Braced, Shaker untied X AA sweep bottom 21:11 2:11:30 PM 4-54 Hz 50 s 1308777078
10b " Exp3.1.10b.txt Braced, Shaker untied X AA sweep bottom 21:12 2:12:55 PM 4-54 Hz 200 s 1308777170
11a " Exp3.1.11a.txt Braced, Shaker untied Y AA sweep bottom 21:21 2:21:40 PM 4-54 Hz 50 s 1308777686
11b " Exp3.1.11b.txt Braced, Shaker untied Y AA sweep bottom 21:22 2:22:50 PM 4-54 Hz 50 s 1308777756
11c " Exp3.1.11c.txt Braced, Shaker untied Y AA sweep bottom 21:24 2:24:00 PM 4-54 Hz 50 s 1308777826
11d " Exp3.1.11d.txt Braced, Shaker untied Y AA sweep bottom 21:25 2:25:30 PM 2:28:50 PM 4-54 Hz 200 s 1308777918
11e " Exp3.1.11e.txt Braced, Shaker untied Y AA sweep bottom 21:29 2:29:30 PM 2:32:50 PM 4-54 Hz 200 s 1308778158
12 " Exp3.1.12.txt Braced, Shaker untied Y AA step .5 Hz bottom 21:34 2:34:30 PM 2:51:30 PM 4-44 Hz 1040 s 1308778456
13a " Exp3.1.13a.txt Braced, Shaker untied Y AA Torsion 42" sweep bottom 22:02 3:02:30 PM 4- 54 Hz 50 s 1308780138
13b " Exp3.1.13b.txt Braced, Shaker untied Y AA Torsion 42" sweep bottom 22:03 3:03:45 PM 4- 54 Hz 50 s 1308780213
13c " Exp3.1.13c.txt Braced, Shaker untied Y AA Torsion 42" sweep bottom 22:04 3:04:50 PM 4- 54 Hz 50 s 1308780278
13d " Exp3.1.13d.txt Braced, Shaker untied Y AA Torsion 42" sweep bottom 22:06 3:06:50 PM 4- 54 Hz 200 s 1308780398
13e " Exp3.1.13e.txt Braced, Shaker untied Y AA Torsion 42" sweep bottom 22:10 3:10:30 PM 4- 54 Hz 200 s 1308780618
- 20110622221102.GV3
13f " Exp3.1.13f.txt Braced, Shaker untied Y AA Torsion 42" sweep bottom 22:14 3:14:30 PM 4- 54 Hz 200 s 1308780858
14 " Exp3.1.14.txt Braced, Shaker untied Y AA Torsion 42" Step .5 Hz bottom 22:21 3:21:30 PM 3:38:00 PM 4-44 Hz 1000s 1308781278
* Blue highlight shows raw data file, in which snipped files below came from; X, Y and Torsion are direction of shaking; AA: Atom Ant shaker used; MM: Mighty Mouse shaker used;
* step: shaker increments stepped at a defined frequency (given directly after step label) was used; sweep: increment of shaker frequency was provided in across a defined frequency zone and then brought back to zero;
*Bottom: indicates shaker was located on the bottom concrete deck; Top: indicates shaker was located on the top concrete deck; % ecc 980: indicated percent eccentricity of the 980 lb weight used during forcing.

175
Table 5-4a GVDA forced vibration testing logs for testing on June 23 and 24, 2011 (Trial 2: Mini-Me in braced condition with forcing from Mighty
Mouse shaker).
Local Time Recorded in Field

Sweep Snipped
Recorded Data File Subevent File Snipped Unix
Subevent # Local Date Description Loading Description GMT First Sample Last Sample Frequency Subevent
Name Name time
Range Time Length

20110623163939.GV3
1a 6/23/2011 Exp3.2.1a.txt Braced Y 10% ecc, 980 Force, sweep 17:16 10:16:40 AM 10:23:00 AM 0-19 Hz 400 1308849445
1b 6/23/2011 Exp3.2.1b.txt Braced Y 10% ecc, 980 Force, sweep 17:25 10:25:00 AM 10:32:00 AM 0-19 Hz 400 1308849945
1c 6/23/2011 Exp3.2.1c.txt Braced Y 10% ecc, 980 Force, sweep 17:32 10:32:00 AM 10:39:00 AM 0-19 Hz 400 1308850378
20110623173939.GV3
2a 6/23/2011 Exp3.2.2a.txt Braced Y 25% ecc, 980 Force, sweep 17:56 10:56:00 AM 11:03:00 AM 0-19 Hz 400 1308851805
2b 6/23/2011 Exp3.2.2b.txt Braced Y 25% ecc, 980 Force, sweep 18:03 11:03:00 AM 11:10:00 AM 0-19 Hz 400 1308852255
2c 6/23/2011 Exp3.2.2c.txt Braced Y 25% ecc, 980 Force, sweep 18:11 11:11:00 AM 11:18:00 AM 0-19 Hz 400 1308852720
20110623183939.GV3
3 6/23/2011 Exp3.2.3.txt Braced Y 25% ecc, 980 Force, step 0.2 Hz 18:39 11:39:00 AM 11:46:00 AM 1-19 Hz 450 1308854440
Void 6/23/2011 Void Braced X 25% ecc, 980 Force, sweep 19:06 12:06:00 PM 12:13:00 PM Error Void Void
4a 6/23/2011 Exp3.2.4a.txt Braced X 25% ecc, 980 Force, sweep 19:17 12:17:00 PM 12:23:00 PM 1-19 Hz 400 1308856680
4b 6/23/2011 Exp3.2.4b.txt Braced X 25% ecc, 980 Force, sweep 19:25 12:25:00 PM 12:32:00 PM 1-19 Hz 400 1308857160
5 6/23/2011 Exp3.2.5.txt Braced X 25% ecc, 980 Force,step 0.2 Hz 19:33 12:33:00 PM 12:41:00 PM 1-19 Hz 420 1308857700
20110623193939.GV3
6a 6/23/2011 Exp3.2.6a.txt Braced X 10% ecc, 980 Force, sweep 20:03 1:03:00 PM 1:10:00 PM 1-19 Hz 400 1308859440
6b 6/23/2011 Exp3.2.6b.txt Braced X 10% ecc, 980 Force, sweep 20:11 1:11:00 PM 1:18:00 PM 1-19 Hz 400 1308859920
6c 6/23/2011 Exp3.2.6c.txt Braced X 10% ecc, 980 Force, sweep 20:18 1:18:00 PM 1:25:00 PM 1-19 Hz 400 1308860340
Void 6/23/2011 Void Braced X 10% ecc, 980 Force, step 0.2 Hz Error Void Void Error Void Void
Void 6/23/2011 Void Braced X 5% ecc, 980 Force, sweep Error Void Void Error Void Void
20110623220430.GV3
7a 6/23/2011 Exp3.2.7a.txt Braced X 5% ecc, 980 Force, sweep 22:09 3:09:00 PM 0-19 Hz 400 1308867000
7b 6/23/2011 Exp3.2.7b.txt Braced X 5% ecc, 980 Force, sweep 22:25 3:25:00 PM 0-19 Hz 400 1308867960
7c 6/23/2011 Exp3.2.7c.txt Braced X 5% ecc, 980 Force, sweep 22:33 3:33:00 PM 0-19 Hz 400 1308868440
8 6/23/2011 Exp3.2.8.txt Braced X 5% ecc, 980 Force, step 0.2 Hz 22:41 3:41:00 PM 0-19 Hz 420 1308868940
20110623230430.GV3
9a 6/23/2011 Exp3.2.9a.txt Braced Y 5% ecc, 980 Force, sweep 23:09 4:09:00 PM 4:18:00 PM 0-19 Hz 400 1308870600
9b 6/23/2011 Exp3.2.9b.txt Braced Y 5% ecc, 980 Force, sweep 23:19 4:19:00 PM 4:24:00 PM 0-19 Hz 400 1308871050
9c 6/23/2011 Exp3.2.9c.txt Braced Y 5% ecc, 980 Force, sweep 23:25 4:25:00 PM 0-19 Hz 400 1308871520
10 6/23/2011 Exp3.2.10.txt Braced Y 5% ecc, 980 Force, step 0.2 Hz 23:32 4:32:00 PM 0-19 Hz 450 1308872020
Void 6/23/2011 Void Braced Y 50% ecc, 980 Force, sweep 23:54 4:54:00 PM Error Void Void
20110624000037.GV3
Void 6/23/2011 Void Braced Y 50% ecc, 980 Force, sweep 0:24 5:24:00 PM Error Void Void
11 6/23/2011 Exp3.2.11.txt Braced Y 50% ecc, 980 Force, sweep 0:28 5:28:00 PM 5:36:00 PM 0-17.5 Hz 400 1308875340
Void 6/23/2011 Void Braced Y 50% ecc, 980 Force, sweep 0:42 5:42:00 PM Error Void Void
Void 6/23/2011 Void Braced Y 50% ecc, 980 Force, sweep 0:54 5:54:00 PM Error Void Void
20110624010037.GV3
Void 6/23/2011 Void Braced X 50% ecc, 980 Force, sweep 1:17 6:17:00 PM Error Void Void
20110624161014.GV3
20110624171014.GV3
12a 6/24/2011 Exp3.2.12a.txt Braced X 50% ecc, 980 Force, sweep 17:12 10:12:00 AM 0-16 Hz 360 1308935590
12b 6/24/2011 Exp3.2.12b.txt Braced X 50% ecc, 980 Force, sweep 17:26 10:26:00 AM 10:33:00 AM 0-16.5 Hz 360 1308936430
13 6/24/2011 Exp3.2.13.txt Braced X 50% ecc, 980 Force, step 0.2 Hz 17:36 10:36:00 AM 0-16.5 Hz 360 1308937100
14 6/24/2011 Exp3.2.14.txt Braced X 100% ecc, 980 Force, sweep 18:00 11:00:00 AM 11:08:00 AM 0-14 Hz 397 1308938483
20110624181014.GV3
15 6/24/2011 Exp3.2.15.txt Braced Y 100% ecc, 980 Force, sweep 18:53 11:53:00 AM 0-14 Hz 419 1308941641
20110624191014.GV3
16 6/24/2011 Exp3.2.16.txt Braced Y 100% ecc, 980 Force, step 0.2 Hz 19:22 12:22:00 PM 0-14 Hz 398 1308943442

* Blue highlight shows raw data file, in which snipped files below came from; X, Y and Torsion are direction of shaking; AA: Atom Ant shaker used; MM: Mighty Mouse shaker used;
* step: shaker increments stepped at a defined frequency (given directly after step label) was used; sweep: increment of shaker frequency was provided in across a defined frequency zone and then brought back to zero;
*Bottom: indicates shaker was located on the bottom concrete deck; Top: indicates shaker was located on the top concrete deck; % ecc 980: indicated percent eccentricity of the 980 lb weight used during forcing.

176
Table 5-4b Mighty Mouse shaker data logs for testing on June 23 and 24, 2011 (Trial 2: Mini-Me in braced condition with forcing from Mighty
Mouse shaker).
Cross-Correlation Data

Sweep Snipped First Snipped First Snipped

Recorded Data File Subevent File
Subevent # Local Date Description Loading Description Notes Frequency Subevent Snipped time Sample (Point Sample (Time
Name Name
Range Time Length Location) Location, sec)

GV_11_06_23_1015
1a 6/23/2011 Exp3.2.1a.txt Braced Y 10% ecc, 980 Force, sweep 0-19 Hz 418 112 1556 0.3112
1b 6/23/2011 Exp3.2.1b.txt Braced Y 10% ecc, 980 Force, sweep 0-19 Hz 398 612 1562 0.3124
1c 6/23/2011 Exp3.2.1c.txt Braced Y 10% ecc, 980 Force, sweep 0-19 Hz 424 1045 1567 0.3134
GV_11_06_23_1055
2a 6/23/2011 Exp3.2.2a.txt Braced Y 25% ecc, 980 Force, sweep 0-19 Hz 420 50 2231 0.4462
2b 6/23/2011 Exp3.2.2b.txt Braced Y 25% ecc, 980 Force, sweep 0-19 Hz 440 505 2237 0.4474
2c 6/23/2011 Exp3.2.2c.txt Braced Y 25% ecc, 980 Force, sweep 0-19 Hz 416 965 2239 0.4478
GV_11_06_23_1139
3 6/23/2011 Exp3.2.3.txt Braced Y 25% ecc, 980 Force, step 0.2 Hz 1-19 Hz 470 90 789 0.1578
Void 6/23/2011 Void Braced X 25% ecc, 980 Force, sweep Error Encountered Error Void Void
GV_11_06_23_1217
4a 6/23/2011 Exp3.2.4a.txt Braced X 25% ecc, 980 Force, sweep X-Acc Rev. for Corr. 1-19 Hz 427 30 1806 0.3612
4b 6/23/2011 Exp3.2.4b.txt Braced X 25% ecc, 980 Force, sweep X-Acc Rev. for Corr. 1-19 Hz 409 511 1809 0.3618
5 6/23/2011 Exp3.2.5.txt Braced X 25% ecc, 980 Force,step 0.2 Hz X-Acc Rev. for Corr. 1-19 Hz 452 1048 11813 2.3626
GV_11_06_23_1303
6a 6/23/2011 Exp3.2.6a.txt Braced X 10% ecc, 980 Force, sweep X-Acc Rev. for Corr. 1-19 Hz 406 54 9529 1.9058
6b 6/23/2011 Exp3.2.6b.txt Braced X 10% ecc, 980 Force, sweep X-Acc Rev. for Corr. 1-19 Hz 385 535 4532 0.9064
6c 6/23/2011 Exp3.2.6c.txt Braced X 10% ecc, 980 Force, sweep X-Acc Rev. for Corr. 1-19 Hz 427 953 14535 2.907
Void 6/23/2011 Void Braced X 10% ecc, 980 Force, step 0.2 Hz Error Encountered Error Void Void
Void 6/23/2011 Void Braced X 5% ecc, 980 Force, sweep Error Encountered Error Void Void
GV_11_06_23_1509
7a 6/23/2011 Exp3.2.7a.txt Braced X 5% ecc, 980 Force, sweep X-Acc Rev. for Corr. 0-19 Hz 436 32 2095 0.419
GV_11_06_23_1525
7b 6/23/2011 Exp3.2.7b.txt Braced X 5% ecc, 980 Force, sweep X-Acc Rev. for Corr. 0-19 Hz 435 25 11421 2.2842
7c 6/23/2011 Exp3.2.7c.txt Braced X 5% ecc, 980 Force, sweep X-Acc Rev. for Corr. 0-19 Hz 420 495 11424 2.2848
8 6/23/2011 Exp3.2.8.txt Braced X 5% ecc, 980 Force, step 0.2 Hz X-Acc Rev. for Corr. 0-19 Hz 459 1004 16424 3.2848
GV_11_06_23_1609
9a 6/23/2011 Exp3.2.9a.txt Braced Y 5% ecc, 980 Force, sweep 0-19 Hz 419 40 1226 0.2452
9b 6/23/2011 Exp3.2.9b.txt Braced Y 5% ecc, 980 Force, sweep 0-19 Hz 430 489 6223 1.2446
9c 6/23/2011 Exp3.2.9c.txt Braced Y 5% ecc, 980 Force, sweep 0-19 Hz 415 960 1218 0.2436
10 6/23/2011 Exp3.2.10.txt Braced Y 5% ecc, 980 Force, step 0.2 Hz 0-19 Hz 460 1460 1213 0.2426
Void 6/23/2011 Void Braced Y 50% ecc, 980 Force, sweep Error Encountered Error Void Void
Void 6/23/2011 Void Braced Y 50% ecc, 980 Force, sweep Error Encountered Error Void Void
GV_11_06_23_1728
11 6/23/2011 Exp3.2.11.txt Braced Y 50% ecc, 980 Force, sweep 0-17.5 Hz 408 48 1920 0.384
Void 6/23/2011 Void Braced Y 50% ecc, 980 Force, sweep Error Encountered Error Void Void
Void 6/23/2011 Void Braced Y 50% ecc, 980 Force, sweep Error Encountered Error Void Void
Void 6/23/2011 Void Braced X 50% ecc, 980 Force, sweep Error Encountered Error Void Void
GV_11_06_24_1011
12a 6/24/2011 Exp3.2.12a.txt Braced X 50% ecc, 980 Force, sweep X-Acc Rev. for Corr. 0-16 Hz 384 72 2406 0.4812
GV_11_06_24_1026
12b 6/24/2011 Exp3.2.12b.txt Braced X 50% ecc, 980 Force, sweep X-Acc Rev. for Corr. 0-16.5 Hz 408 62 3456 0.6912
13 6/24/2011 Exp3.2.13.txt Braced X 50% ecc, 980 Force, step 0.2 Hz X-Acc Rev. for Corr. 0-16.5 Hz 375 732 3465 0.693
GV_11_06_24_1100
14 6/24/2011 Exp3.2.14.txt Braced X 100% ecc, 980 Force, sweep X-Acc Rev. for Corr. 0-14 Hz 419 35 5980 1.196
GV_11_06_24_1153
15 6/24/2011 Exp3.2.15.txt Braced Y 100% ecc, 980 Force, sweep 0-14 Hz 411 43 3774 0.7548
GV_11_06_24_1222
16 6/24/2011 Exp3.2.16.txt Braced Y 100% ecc, 980 Force, step 0.2 Hz 0-14 Hz 389 65 3939 0.7878

* Blue highlight shows raw data file, in which snipped files below came from; X, Y and Torsion are direction of shaking; AA: Atom Ant shaker used; MM: Mighty Mouse shaker used;
* step: shaker increments stepped at a defined frequency (given directly after step label) was used; sweep: increment of shaker frequency was provided in across a defined frequency zone and then brought back to zero;
*Bottom: indicates shaker was located on the bottom concrete deck; Top: indicates shaker was located on the top concrete deck; % ecc 980: indicated percent eccentricity of the 980 lb weight used during forcing.

177
Table5-5 GVDA forced vibration testing logs for testing on August 1 and 2, 2011 (Trial 3: Mini-Me in unbraced condition with forcing from Atom
Ant shaker).
Local Time Recorded in Field

Sweep Snipped
Recorded Data File Subevent File Snipped Unix
Subevent # Local Date Description Loading Description GMT First Sample Last Sample Frequency Subevent
Name Name time
Range Time Length

20110801213133.GV4
1 8/1/2011 Exp3.3.1a.txt Not Braced X AA sweep top 22:55 2:55:30 PM 4-54 Hz 50 s 1312235714
2 8/1/2011 Exp3.3.1b.txt Not Braced, Med Amplitude X AA sweep top 22:57 2:57:00 PM 4-54 Hz 50 s 1312235805
3 8/1/2011 Exp3.3.1c.txt Not Braced, X AA sweep top 22:58 2:58:25 PM 4-54 Hz 50 s 1312235890
4 8/1/2011 Exp3.3.1d.txt Not Braced, X AA sweep top 23:00 3:00:00 PM 4-54 Hz 200 s 1312235985
5 8/1/2011 Exp3.3.1e.txt Not Braced, Med Amp X AA sweep top 23:05 3:05:30 PM 4-54 Hz 200 s 1312236318
6 8/1/2011 Exp3.3.2.txt Not Braced X AA step 0.5 Hz top, may have slipped @ 25Hz 23:10 3:10:15 PM 4-35 Hz 760 s 1312236580
7 8/1/2011 Exp3.3.2f.txt Not Braced X AA sweep top 23:27 3:27:45 PM 4-54 Hz 50 s 1312237652
20110801223133.GV4
8 8/1/2011 Exp3.3.3a.txt Not Braced Y AA torsion 47" top 23:43 3:43:25 PM 4-54 Hz 50 s 1312238592
9 8/1/2011 Exp3.3.3b.txt Not Braced Y AA torsion 47" top 23:44 3:44:40 PM 4-54 Hz 50 s 1312238667
10 8/1/2011 Exp3.3.3c.txt Not Braced Y AA torsion 47" top 23:46 3:46:20 PM 4-54 Hz 200 s 1312238767
11 8/1/2011 Exp3.3.3d.txt Not Braced Y AA torsion 47" top 23:50 3:50:13 PM 4-54 Hz 200 s 1312239001
12 8/1/2011 Exp3.3.4.txt Not Braced Y AA torsion step 1.0 Hz top 23:55 3:55:00 PM 4:06:20 PM 4-44 Hz 700 s 1312239278
13 8/1/2011 Exp3.3.5a.txt Not Braced Y AA, Sweep,Shaker Acc Direction Switch 0:22 4:22:15 PM 4-54 Hz 50 s 1312240925
14 8/1/2011 Exp3.3.5b.txt Not Braced Y AA, Sweep,Shaker Acc Direction Switch 0:24 4:24:00 PM 4-54 Hz 50 s 1312241028
15 8/1/2011 Exp3.3.5c.txt Not Braced Y AA, Sweep,Shaker Acc Direction Switch 0:29 4:29:45 PM 4-54 Hz 200 s 1312241373
20110801233133.GV4
16 8/1/2011 Exp3.3.5d.txt Not Braced Y AA, Sweep,Shaker Acc Direction Switch 0:33 4:33:25 PM 4-54 Hz 200 s 1312241593
17 8/1/2011 Exp3.3.6.txt Not Braced Y AA, 1 Hz Steps,Shaker Acc Direction Switch 0:38 4:38:00 PM 4:48:00 PM 4-44Hz 600 s 1312241868
20110802000036.GV4
18 8/1/2011 Exp3.3.7a.txt Not Braced X AA sweep bottom 1:09 5:09:50 PM 4-54 Hz 50 s 1312157380
19 8/1/2011 Exp3.3.7b.txt Not Braced X AA sweep bottom 1:11 5:11:15 PM 4-54 Hz 50 s 1312243863
20 8/1/2011 Exp3.3.7c.txt Not Braced X AA sweep bottom 1:12 5:12:25 PM 4-54 Hz 200 s 1312243933
21 8/1/2011 Exp3.3.7d.txt Not Braced X AA sweep bottom 1:16 5:16:10 PM 4-54 Hz 200 s 1312244158
22 8/1/2011 Exp3.3.8.txt Not Braced X AA step 1 Hz bottom 1:20 5:20:30 PM 5:27:35 PM 4-44 Hz 450 s 1312244418
23 8/1/2011 Exp3.3.9a.txt Not Braced Y AA sweep bottom 1:36 5:36:50 PM 4-54 Hz 50 s 1312245402
24 8/1/2011 Exp3.3.9b.txt Not Braced Y AA sweep bottom 1:38 5:38:04 PM 4-54 Hz 50 s 1312245470
25 8/1/2011 Exp3.3.9c.txt Not Braced Y AA sweep bottom 1:39 5:39:30 PM 4-54 Hz 200 s 1312245558
26 8/1/2011 Exp3.3.9d.txt Not Braced Y AA sweep bottom 1:43 5:43:25 PM 4-54 Hz 200 s 1312245792
27 8/1/2011 Exp3.3.10.txt Not Braced Y AA step 1 Hz bottom 1:47 5:47:20 PM 5:54:25 PM 4-44 Hz 450 s 1312246028
20110802150036.GV4
28 8/2/2011 Exp3.3.11a.txt Not Braced Y AA torsion 47" bottom 16:47 8:47:07 AM 4-54 Hz 50 s 1312300012
29 8/2/2011 Exp3.3.11b.txt Not Braced Y AA torsion 47" bottom 16:48 8:48:08 AM 4-54 Hz 50 s 1312300070
20110802161037.GV4
30 8/2/2011 Exp3.3.11c.txt Not Braced Y AA torsion 47" bottom 17:12 9:12:10 AM 4-54 Hz 200 s 1312300208
31 8/2/2011 Exp3.3.11d.txt Not Braced Y AA torsion 47" bottom 17:16 9:16:00 AM 4-54 Hz 200 s 1312300426
32 8/2/2011 Exp3.3.12.txt Not Braced Y AA torsion step 1.0 Hz bottom 17:20 9:20:00 AM 9:26:50 AM 4-44 Hz 420 s 1312300673

* Blue highlight shows raw data file, in which snipped files below came from; X, Y and Torsion are direction of shaking; AA: Atom Ant shaker used; MM: Mighty Mouse shaker used;
* step: shaker increments stepped at a defined frequency (given directly after step label) was used; sweep: increment of shaker frequency was provided in across a defined frequency zone and then brought back to zero;
*Bottom: indicates shaker was located on the bottom concrete deck; Top: indicates shaker was located on the top concrete deck; % ecc 980: indicated percent eccentricity of the 980 lb weight used during forcing.

178
Table 5-6a GVDA forced vibration testing logs for testing on August 2, 2011 (Trial 4: Mini-Me in unbraced condition with forcing from Mighty
Mouse shaker).
Local Time Recorded in Field

Sweep Snipped
Recorded Data File Subevent File Snipped Unix
Subevent # Local Date Description Loading Description GMT First Sample Last Sample Frequency Subevent
Name Name time
Range Time Length

20110802161037.GV4
1 8/2/2011 Exp3.4.1a.txt Not Braced Y 5% ecc, 980 Force, sweep 16:46 9:46:00 AM 9:54:00 AM 0-19 Hz 399 1312303611
2 8/2/2011 Exp3.4.1b.txt Not Braced Y 5% ecc, 980 Force, sweep 16:54 9:54:00 AM 10:02:00 AM 0-19 Hz 398 1312304072
3 8/2/2011 Exp3.4.2.txt Not Braced Y 5% ecc, 980 Force, step 0.2 Hz 17:05 10:05:00 AM 10:12:00 AM 0-19 Hz 400 1312304750
20110802171037.GV4
4 8/2/2011 Exp3.4.3a.txt Not Braced Y 10% ecc, 980 Force, sweep 17:23 10:23:00 AM 10:31:00 AM 0-19 Hz 390 1312305840
5 8/2/2011 Exp3.4.3b.txt Not Braced Y 10% ecc, 980 Force, sweep 17:31 10:31:00 AM 10:39:00 AM 0-19 Hz 400 1312306310
6 8/2/2011 Exp3.4.4.txt Not Braced Y 10% ecc, 980 Force, step 0.2 Hz 17:39 10:39:00 AM 10:47:00 AM 0-19 Hz 420 1312306850
7 8/2/2011 Exp3.4.5a.txt Not Braced Y 25% ecc, 980 Force, sweep 17:58 10:58:00 AM 11:06:00 AM 0-19 Hz 400 1312307920
8 8/2/2011 Exp3.4.5b.txt Not Braced Y 25% ecc, 980 Force, sweep 18:06 11:06:00 AM 11:14:00 AM 0-19 Hz 399 1312308371
20110802181037.GV4
9 8/2/2011 Exp3.4.6.txt Not Braced Y 25% ecc, 980 Force, step 0.2 Hz 18:14 11:14:00 AM 11:23:00 AM 0-19 Hz 417 1312308938
10 8/2/2011 Exp3.4.7a.txt Not Braced X 5% ecc, 980 Force, sweep 18:39 11:39:00 AM 11:48:00 AM 0-19 Hz 400 1312310425
11 8/2/2011 Exp3.4.7b.txt Not Braced X 5% ecc, 980 Force, sweep 18:48 11:48:00 AM 11:56:00 AM 0-19 Hz 400 1312310905
12 8/2/2011 Exp3.4.8.txt Not Braced X 5% ecc, 980 Force, step 0.2 Hz 18:58 11:58:00 AM 12:07:00 PM 0-19 Hz 419 1312311611
20110802191037.GV4
13 8/2/2011 Exp3.4.9a.txt Not Braced X 10% ecc, 980 Force, sweep 19:19 12:19:00 PM 12:27:00 PM 0-19 Hz 399 1312312781
14 8/2/2011 Exp3.4.9b.txt Not Braced X 10% ecc, 980 Force, sweep 19:27 12:27:00 PM 12:36:00 PM 0-19 Hz 400 1312313250
15 8/2/2011 Exp3.4.10.txt Not Braced X 10% ecc, 980 Force, step 0.2 Hz 19:41 12:41:00 PM 12:49:00 PM 0-19 Hz 440 1312314160
16 8/2/2011 Exp3.4.11a.txt Not Braced X 25% ecc, 980 Force, sweep 19:58 12:58:00 PM 1:06:00 PM 0-19 Hz 400 1312315140
17 8/2/2011 Exp3.4.11b.txt Not Braced X 25% ecc, 980 Force, sweep 20:06 1:06:00 PM 1:13:00 PM 0-19 Hz 400 1312315610
20110802201037.GV4
18 8/2/2011 Exp3.4.12.txt Not Braced X 25% ecc, 980 Force, step 0.2 Hz 20:14 1:14:00 PM 1:23:00 PM 0-19 Hz 440 1312316125
19 8/2/2011 Exp3.4.13a.txt Not Braced X 50% ecc, 980 Force, sweep 20:38 1:38:00 PM 1:46:00 PM 0-14 Hz 400 1312317520
20 8/2/2011 Exp3.4.13b.txt Not Braced X 50% ecc, 980 Force, sweep 20:46 1:46:00 PM 1:55:00 PM 0-14 Hz 396 1312318004
21 8/2/2011 Exp3.4.14.txt Not Braced X 50% ecc, 980 Force, step 0.2 Hz 20:54 1:54:00 PM 2:02:00 PM 0-14 Hz 417 1312318493
20110802211037.GV4
22 8/2/2011 Exp3.4.15a.txt Not Braced, step X 100 % ecc, east weight slipped to 95% ecc 21:19 2:19:00 PM 2:27:00 PM 0-14 Hz 391 1312320009
23 8/2/2011 Exp3.4.15b.txt Not Braced, step X 100 % ecc, east weight slipped to 95% ecc 21:45 2:45:00 PM 2:53:00 PM 0-14 Hz 393 1312321547
24 8/2/2011 Exp3.4.16.txt Not Braced, sweep 0.2 Hz X 100 % ecc, east weight slipped to 95% ecc 21:53 2:53:00 PM 3:01:00 PM 0-14 Hz 410 1312322040
20110802221037.GV4
25 8/2/2011 Exp3.4.17a.txt Not Braced, step Y-Dir, 50 % ecc, directional sensor off 60 deg 22:22 3:22:00 PM 3:29:00 PM 0-14 Hz 399 1312323751
26 8/2/2011 Exp3.4.17b.txt Not Braced, step Y-Dir, 50 % ecc, directional sensor off 60 deg 22:29 3:29:00 PM 3:36:00 PM 0-14 Hz 399 1312324201
27 8/2/2011 Exp3.4.18.txt Not Braced, sweep 0.2 Hz Y-Dir, 50 % ecc, directional sensor off 60 deg 22:37 3:37:00 PM 3:45:00 PM 0-14 Hz 419 1312324681
28 8/2/2011 Exp3.4.19a.txt Not Braced Y 100% ecc, 980 Force, sweep 23:00 4:00:00 PM 4:07:00 PM 0-12 Hz 399 1312326031
20110802231037.GV4
29 8/2/2011 Exp3.4.19b.txt Not Braced Y 100% ecc, 980 Force, sweep 23:07 4:07:00 PM 4:14:00 PM 0-12 Hz 400 1312326480
30 8/2/2011 Exp3.4.20.txt Not Braced Y 100% ecc, 980 Force, step 0.2 Hz 23:16 4:16:00 PM 0-12 Hz 419 1312326961

* Blue highlight shows raw data file, in which snipped files below came from; X, Y and Torsion are direction of shaking; AA: Atom Ant shaker used; MM: Mighty Mouse shaker used;
* step: shaker increments stepped at a defined frequency (given directly after step label) was used; sweep: increment of shaker frequency was provided in across a defined frequency zone and then brought back to zero;
*Bottom: indicates shaker was located on the bottom concrete deck; Top: indicates shaker was located on the top concrete deck; % ecc 980: indicated percent eccentricity of the 980 lb weight used during forcing.

179
Table 5-6b Mighty Mouse shaker data logs for testing on August 2, 2011 (Trial 4: Mini-Me in unbraced condition with forcing from Mighty Mouse
shaker).
Cross-Correlation Data

Sweep Snipped First Snipped First Snipped

Recorded Data File Subevent File Snipped Unix
Subevent # Local Date Description Loading Description Notes Frequency Subevent Sample (Point Sample (Time
Name Name time
Range Time Length Location) Location, sec)

GV_11_08_02_0945_1
1 8/2/2011 Exp3.4.1a.txt Not Braced Y 5% ecc, 980 Force, sweep Y-Acc Rev. for Corr. 0-19 Hz 411 49 4485 0.897
2 8/2/2011 Exp3.4.1b.txt Not Braced Y 5% ecc, 980 Force, sweep Y-Acc Rev. for Corr. 0-19 Hz 405 510 4486 0.8972
GV_11_08_02_1003_1
3 8/2/2011 Exp3.4.2.txt Not Braced Y 5% ecc, 980 Force, step 0.2 Hz Y-Acc Rev. for Corr. 0-19 Hz 410 138 2443 0.4886
GV_11_08_02_1023_1
4 8/2/2011 Exp3.4.3a.txt Not Braced Y 10% ecc, 980 Force, sweep Y-Acc Rev. for Corr. 0-19 Hz 428 32 15813 3.1626
5 8/2/2011 Exp3.4.3b.txt Not Braced Y 10% ecc, 980 Force, sweep Y-Acc Rev. for Corr. 0-19 Hz 416 504 5814 1.1628
6 8/2/2011 Exp3.4.4.txt Not Braced Y 10% ecc, 980 Force, step 0.2 Hz Y-Acc Rev. for Corr. 0-19 Hz 445 1044 5815 1.163
GV_11_08_02_1057_1
7 8/2/2011 Exp3.4.5a.txt Not Braced Y 25% ecc, 980 Force, sweep Y-Acc Rev. for Corr. 0-19 Hz 416 64 8560 1.712
8 8/2/2011 Exp3.4.5b.txt Not Braced Y 25% ecc, 980 Force, sweep Y-Acc Rev. for Corr. 0-19 Hz 434 516 3561 0.7122
9 8/2/2011 Exp3.4.6.txt Not Braced Y 25% ecc, 980 Force, step 0.2 Hz Y-Acc Rev. for Corr. 0-19 Hz 439 1083 3561 0.7122
GV_11_08_02_1139_1
10 8/2/2011 Exp3.4.7a.txt Not Braced X 5% ecc, 980 Force, sweep 0-19 Hz 426 34 3808 0.7616
11 8/2/2011 Exp3.4.7b.txt Not Braced X 5% ecc, 980 Force, sweep 0-19 Hz 421 513 8808 1.7616
GV_11_08_02_1158_1
12 8/2/2011 Exp3.4.8.txt Not Braced X 5% ecc, 980 Force, step 0.2 Hz 0-19 Hz 414 116 2466 0.4932
GV_11_08_02_1219_1
13 8/2/2011 Exp3.4.9a.txt Not Braced X 10% ecc, 980 Force, sweep 0-19 Hz 423 37 4762 0.9524
14 8/2/2011 Exp3.4.9b.txt Not Braced X 10% ecc, 980 Force, sweep 0-19 Hz 424 506 4762 0.9524
GV_11_08_02_1240_1
15 8/2/2011 Exp3.4.10.txt Not Braced X 10% ecc, 980 Force, step 0.2 Hz 0-19 Hz 455 98 1773 0.3546
16 8/2/2011 Exp3.4.11a.txt Not Braced X 25% ecc, 980 Force, sweep 0-19 Hz Missing Missing 0
17 8/2/2011 Exp3.4.11b.txt Not Braced X 25% ecc, 980 Force, sweep 0-19 Hz Missing Missing 0
18 8/2/2011 Exp3.4.12.txt Not Braced X 25% ecc, 980 Force, step 0.2 Hz 0-19 Hz Missing Missing 0
GV_11_08_02_1337_1
19 8/2/2011 Exp3.4.13a.txt Not Braced X 50% ecc, 980 Force, sweep 0-14 Hz 421 39 4802 0.9604
20 8/2/2011 Exp3.4.13b.txt Not Braced X 50% ecc, 980 Force, sweep 0-14 Hz 417 523 4802 0.9604
21 8/2/2011 Exp3.4.14.txt Not Braced X 50% ecc, 980 Force, step 0.2 Hz 0-14 Hz 412 1012 4802 0.9604
GV_11_08_02_1418_1
22 8/2/2011 Exp3.4.15a.txt Not Braced, step X 100 % ecc, east weight slipped to 95% ecc 0-14 Hz 441 68 6602 1.3204
GV_11_08_02_1444_1
23 8/2/2011 Exp3.4.15b.txt Not Braced, step X 100 % ecc, east weight slipped to 95% ecc 0-14 Hz 414 46 8859 1.7718
24 8/2/2011 Exp3.4.16.txt Not Braced, sweep 0.2 Hz X 100 % ecc, east weight slipped to 95% ecc 0-14 Hz 360 540 3858 0.7716
GV_11_08_02_1521_1
25 8/2/2011 Exp3.4.17a.txt Not Braced, step Y-Dir, 50 % ecc, directional sensor off 60 deg Y-Acc Rev. for Corr. 0-14 Hz 427 46 14 0.0028
26 8/2/2011 Exp3.4.17b.txt Not Braced, step Y-Dir, 50 % ecc, directional sensor off 60 deg Y-Acc Rev. for Corr. 0-14 Hz 423 497 11 0.0022
27 8/2/2011 Exp3.4.18.txt Not Braced, sweep 0.2 Hz Y-Dir, 50 % ecc, directional sensor off 60 deg Y-Acc Rev. for Corr. 0-14 Hz 418 977 6 0.0012
GV_11_08_02_1559_1
28 8/2/2011 Exp3.4.19a.txt Not Braced Y 100% ecc, 980 Force, sweep Y-Acc Rev. for Corr. 0-12 Hz 423 42 2981 0.5962
29 8/2/2011 Exp3.4.19b.txt Not Braced Y 100% ecc, 980 Force, sweep Y-Acc Rev. for Corr. 0-12 Hz 469 491 2976 0.5952
30 8/2/2011 Exp3.4.20.txt Not Braced Y 100% ecc, 980 Force, step 0.2 Hz Y-Acc Rev. for Corr. 0-12 Hz 408 972 2972 0.5944

* Blue highlight shows raw data file, in which snipped files below came from; X, Y and Torsion are direction of shaking; AA: Atom Ant shaker used; MM: Mighty Mouse shaker used;
* step: shaker increments stepped at a defined frequency (given directly after step label) was used; sweep: increment of shaker frequency was provided in across a defined frequency zone and then brought back to zero;
*Bottom: indicates shaker was located on the bottom concrete deck; Top: indicates shaker was located on the top concrete deck; % ecc 980: indicated percent eccentricity of the 980 lb weight used during forcing.

180
 Two data acquisition systems were used for Trial 2 (braced structure, Mighty

Mouse shaker) and Trial 4 (unbraced structure, Mighty Mouse shaker). The structural

response test logs, corresponding to the data acquisition system attributes described in

Table 5-1, are shown in Tables 5-4a and 5-6a. The Mighty Mouse shaker test logs,

corresponding to the data acquisition system described in Table 5-3, are described in

Tables 5-4b and 5-6b.

The last few columns of Tables 5-3 through 5-6 represent the location where the

subevents were cropped from the unprocessed data files. The last few columns of Tables

5-4b and 5-6b present additional information regarding cross-correlation analyses

performed to synchronize the structural response data files with the Mighty Mouse data

files. Details of the cross-correlation analysis are presented in Section 5.3.3.

5.3.2 Data Processing

Several types of data files were prepared as part of this experiment. The workflow and

development of the data files differ, dependent on the type of shaker used for forced

vibration testing. The different workflow schemes used for processing the data in this

study are shown in Figures 5-8 and 5-9. The solid rectangular boxes in Figures 5-8 and 5-

9 represent the different types of data files (with file name type indicated in brackets),

while the dashed rectangular boxes represent the work required to create the next type of

data file in the processing procedure.

5.3.2.1 Structural Response Data Processing

Table 5-1 shows the attributes of the structural response data acquisition system. The

attributes include the data acquisition (DA) channel, sensor type, sensor location, and

181
sensor orientation. Table 5-1 was used for the forced vibration testing with the small

shaker (Atom Ant) and large shaker (Mighty Mouse). Figure 5-8 shows the processing of

the structural response data from unprocessed data to corrected data. The unprocessed

data files contain multiple events as indicated by the non-highlighted rows in Tables 5-3

and 5-5. The unprocessed structural response data collected in the field was divided into

subevents based on vibration modes and duration of shaking. When the data was divided

into subevents the data was also converted to engineering units and was documented as

converted data files. Then the converted data files were baseline corrected. The baseline

correction consisted of determining the mean responses and removing them from the

records. The baseline corrected data has been saved as the corrected data to be used for

analysis. Shown graphically in Figure 5-8 is an example of the data processing using

channel 15 of the structural response data. The data files shown in Figure 5-8 from top to

bottom are the unprocessed data file: 20110802221037.txt; the converted data file:

Exp.3.4.19a_GVDA_08.02.2011_Unbraced_1312326030_MM_Y_400_

Sweep_100ecc_TopDeck.txt; and the corrected data file:

Exp.3.4.19a_GVDA_08.02.2011_

Unbraced_1312326030_MM_Y_400_Sweep_100ecc_TopDeck_MeanCorrected.t

xt.

182
Figure58 Flowchartdescribingtheworkflowandassociateddatafilesforthestructural
responsedatafiles.Thegraphicalexampleisforexperiment3.4.19aas
describedinTable56.Thesolidboxesintheflowchartdescribethedatafile
typeswithfilenamedescriptionsinbrackets,whilethedashedboxesdescribe
theefforttakentoachievetheunderlyingprocesseddatafile.

5.3.2.2 Mighty Mouse Shaker Data Processing

Table 5-2 shows the attributes of the Mighty Mouse shaker data acquisition system and

Table 5-1 shows the attributes of the structural response data acquisition. Attributes

shown in Tables 5-1 and 5-2 are the data acquisition (DA) channel, sensor type, sensor

location, and sensor orientation. Figure 5-9 shows at the left branch of the workflow how

183
the Mighty Mouse shaker unprocessed data was processed directly to corrected data file

types. The unprocessed data files contain multiple events as indicated by the non-

highlighted rows in Tables 5-4b and 5-6b. The unprocessed Mighty Mouse data files

were divided into subevents based on different shaker force intensities and modes of

vibration. During segregation of these subevents the data was converted from field counts

to engineering units along with a baseline correction (removal of mean response). This

data has been uploaded to the NEEShub as Mighty Mouse corrected data. However, the

two corrected data files needed to be combined at a common sampling frequency for

analysis. Therefore, the high frequency Mighty Mouse shaker data was decimated and

then synchronized with the low frequency structural response data. Synchronization was

performed using cross-correlation, as described in Section 5.3.3. The final corrected data

shown at the bottom of Figure 5-9 contains additional data including calculated shaker

forces, frequency, angle of rotating mass and response of parallel accelerometer used for

data acquisition correlation, which is described in Section 5.3.6. This final corrected data

that combines the two acquisition systems together is referred to as Synchronized

Structural Response and Mighty Mouse Shaker Data.

184
Figure59 FlowchartdescribingtheworkflowandassociateddatafilesfortheMighty
Mousedatafilesandcorrespondingstructuralresponsedatafiles(described
inFigure8).Thesolidboxesintheflowchartdescribethedatafiletypeswith
filenamedescriptionsinbrackets,whilethedashedboxesdescribetheeffort
takentoachievetheunderlyingprocesseddatafile.

185
5.3.3 Cross-Correlation of the Data Acquisition Systems

A cross-correlation analysis was performed to synchronize the two different data

acquisition systems used for testing the large Mighty Mouse shaker. As described in

Section 5.2.5, the Mighty Mouse shaker data was collected at a faster sampling rate

compared to the structural responses data (5000 Hz vs. 200 Hz). The Mighty Mouse

shaker DA system collected data using a relative time signature from start of recording

and required synchronization to the structural response data acquisition, which used a

GPS time signature. This synchronization was achieved by comparing adjacent parallel

accelerometers recording to the respective DA systems in the direction of shaking. The

accelerometers used for correlation were located on the roof at the southwest corner of

Mini-Me, as shown in Figure 5-7.

The cross-correlation analysis was performed using decimated corrected Mighty

Mouse shaker data and unaltered corrected structural response data as described in

Section 5.3.3. The cross-correlation analysis was performed by comparing an

approximate 20 sec structural response window at the middle of the response. This

window of comparison was selected to avoid artificial noise that exists near the ends of

the windows of time in which measured responses occur. The responses typically have an

approximate 50 or 400 sec duration. However, the 20 sec window was compared to a

more robust window encompassing most of the record and both sets were found to

produce equivalent record-to-record lags. Therefore the 20 sec window was used to

reduce computational effort. Two examples of the cross-correlation analysis are provided

in Figures 5-10 and 5-11.

186
 Figure 5-10 shows the cross-correlation analysis for experiment 3.4.1a (unbraced

condition with 5% eccentricity force in the y-direction). Figure 5-10a and 5-10b shows

the corrected data for the adjacent parallel accelerometers used for the cross-correlation

analysis. Since loading for this test was in the y-direction, structural response data

channel 15 and Mighty Mouse shaker data channel 2 (see Figure 5-7) were correlated.

Figure 5-10c shows the graphical result of the cross-correlation, where the peak

represents the time-lag (t) between accelerometers, found to be 0.897 sec. Figure 5-10c

shows the correlated accelerometer responses superimposed in a short time window from

342-345 sec. This short time window is used to present how the Mighty Mouse shaker

data files can be adjusted by t to synchronize with the structural response data.

Figure 5-11 shows the synchronized results for experiment 3.4.1a using the same

accelerometers. Figure 5-11a shows an unaltered structural response from Figure 5-10a,

while Figure 5-11b shows the corrected Mighty Mouse data. The Mighty Mouse data was

shifted by t and decimated to match the sampling frequency of the structural response

data. Figure 5-11c shows the final match of the data sets in a short time window of 342-

345 sec. The Mighty Mouse shaker acceleration time series contains visible levels of

noise as shown in Figure 5-10b. However, following a time shift of t and decimation to

the structural response sampling frequency, synchronization was achieved and the noise

was reduced as a result of the low-pass filter association with the decimation routine (see

Figure 5-11c).

187
Figure510 Experiment3.4.1acrosscorrelationsanalysiscomponents:(a)structural
responsedatachannel15accelerometer,(b)parallelMightyMouseshaker
datachannel2,(c)crosscorrelationgraphicalresult,and(d)shorttime
windowshowingrequiredtimeshiftforcorrelation.

188
Figure511 Experiment3.4.1aresultsfollowingapplicationoftimelag:(a)structural
responsedatachannel15accelerometer,(b)parallelMightyMouseshaker
datachannel2adjustedbyt=0.8907,and(c)shorttimewindowshowing
correlationofadjustedMightyMouseshakerdatatostructuralresponsedata.

Figure 5-12 shows the cross-correlation analysis for experiment 3.4.19a (unbraced

condition with 100% eccentricity force in the y-direction). Figure 5-12a and 5-12b shows

the corrected data for the adjacent parallel accelerometers used for the cross-correlation

analysis. Since loading for this test was in the y-direction, structural response data

channel 15 and Mighty Mouse shaker data channel 2 (see Figure 5-7) were correlated.

Figure 5-12c shows the graphical result of the cross-correlation, where the peak

represents the time-lag (t) between accelerometers, found to be 0.5962 sec. Figure 5-12c

shows the synchronized accelerograms superimposed for a short time window from 110-

113 sec.

189
 Figure 5-13 shows the synchronized results for experiment 3.4.19a using the same

accelerometers. Figure 5-13a shows an unaltered structural response from Figure 5-12a,

while Figure 5-13b shows the corrected Mighty Mouse data. The Mighty Mouse data was

shifted by t and decimated to match the sampling frequency of the structural response

data. Figure 5-13c shows the final match of the data sets in a short time window of 110-

113 sec.

Figure512 Experiment3.4.19acrosscorrelationsanalysiscomponents:(a)structural
responsedatachannel15accelerometer,(b)parallelMightyMouseshaker
datachannel2,(c)crosscorrelationgraphicalresult,and(d)shorttime
windowshowingrequiredtimeshiftforcorrelation.

190
Figure513 Experiment3.4.19aresultsfollowingapplicationoftimelag:(a)structural
responsedatachannel15accelerometer,(b)parallelMightyMouseshaker
datachannel2adjustedbyt=0.8907,and(c)shorttimewindowshowing
correlationofadjustedMightyMouseshakerdatatostructuralresponsedata.

5.3.4 Unprocessed Data Files

The unprocessed data files contain raw data as obtained from in the field from the two

DA systems described in Section 5.2.5. At NEEShub, the unprocessed data has been

stored in two folders: Structural Response Data and Mighty Mouse Shaker Data.

Applicable units for the raw data are Unix time or relative time, counts, and pulses. The

structural response unprocessed data and Mighty Mouse shaker unprocessed data files are

described in the following Sections.

5.3.4.1 Unprocessed Structural Response Data Files

Unix time and counts are the units that were used for recording the unprocessed data of

the structural response in the field. Unix time is a system describing time in sec from

191
Thursday, January 1, 1970 in the Greenwich Mean Time (GMT) zone. It should be noted

that the testing was performed at GVDA in California which is located in the Pacific

Standard Time (PST) zone that is eight hours behind GMT. The June testing was during

daylight savings and used Pacific Daylight Time (PDT) which is seven hours behind

GMT.

The unprocessed data is labeled according to the date and GMT

(YYYYMMDDGMT) when the recording began. For example, a test recorded with a

beginning GMP time of 6:39 pm and 39 seconds on June 23, 2011 would have a text file

name of 20110623183939.txt. The structural response unprocessed data files are listed

and highlighted blue in Tables 5-3, 5-4a, 5-5 and 5-6a. A brief summary of the

documented channels (columns) of data in the structural response data plan (see Table 5-

2) are listed below.

Channel 1: Time (Unix time)

Channels 2 through 43: Accelerometers (Counts)
Channels 44 through 47: LVDTs (Counts)
Channel 48: Accelerometer for Atom Ant shaker mass in direction of loading
(Counts)
5.3.4.2 Unprocessed Mighty Mouse Shaker Data Files

Relative time (sec) from the beginning of data recording, counts and pulses are the units

that were used for recording the unprocessed data for the Mighty Mouse shaker data.

During excitation with Mighty Mouse, the aforementioned channel 48 of the structural

response data files described in Section 5.3.4.1 was unused and slated with zeros.

The unprocessed Mighty Mouse shaker data is labeled according to the date and

PST/PDT (GV_YY_MM_DD_PST/PDT) when the recording begun. For example, a test

192
recorded with a beginning time using PST/PDT at 10:15 am (seconds are not included)

on June 23, 2011 would have a text file name of GV_11_06_23_1015.txt. A brief

summary of the documented channels (columns) of the Mighty Mouse shaker data are

listed below.

Channel 1: Time (sec)

Channels 2 and 3: Correlation accelerometers (volts)

Channels 4 and 5: Shaker pulses (10 pulses per revolution) and shaker control

frequency (1 pulse per revolution)

Channels 4 and 5 of the unprocessed Mighty Mouse shaker data were switched during the

different testing dates. The switch has been noted in Tables 5-2. During the June (braced

condition, Trial 2) testing, channel 4 is the shaker control frequency data and channel 5 is

the shaker pulses data. However, during the August (unbraced condition, Trial 4) testing,

channel 4 is the shaker pulses data and channel 5 is the shaker control frequency data.

5.3.5 Converted Data Files

Converted data was produced by separating the unprocessed data into subevents for each

sweep or step testing interval, and then converting the data to engineering units. First the

start times for sweep and step subevents were identified from the test logged PST/PDT

converted to Unix time. Then through visual inspection using Matlab graphical tools,

appropriate segments were cropped from the unprocessed data file to encompass the

duration of loading and response for each respective subevent. The data within the

subevent segments were then converted to engineering units.

193
5.3.5.1 Converted Structural Response Data Files

Following conversion, the units used for the accelerometers and LVDTs are provided in

Tables 5-1 and 5-2. Conversion was performed by first dividing 2 24 40 counts per volt.

Acceleration signals in volts were converted to g using the differential gain of 5 volts

per g. Displacement signals from LVDTs were converted to volts as done with the

acceleration signals and then converted to inches using the calibration of each LVDT

based on volts per in. It should be noted that three of the LVDT converted data files (Chs.

44, 46, and 47) located at the NEEShub contain an artificial error that can be corrected by

multiplying the response by a factor of 10. (Following processing it was found that these

LVDT conversion factors were off by a tenth). The converted acceleration and

displacement data channels are similar to the unprocessed data discussed in Section 5.3.4.

However, it is presented in Unix time, acceleration in g, and inches as shown below. The

time signature channel (column) of the converted data was not converted to standard sec

and remains in Unix time. The time signature was unchanged during this process so that

the location of each subevent is well documented during the cropping procedure.

Channel 1: Time (Unix time)

Channels 2 through 43: Accelerometers (g)

Channels 44 through 47: LVDTs (in)

Channel 48: Accelerometer for Atom Ant shaker mass in direction of loading (g)

When the structural response converted data was obtained for forced vibration

testing using Mighty Mouse shaker, channel 48 was slated with zeros. This was done for

both Trial 2 and Trial 4.

194
5.3.5.2 Converted Mighty Mouse Shaker Data Files

No converted Mighty Mouse shaker data files were created. As shown in Figure 5-8, the

Mighty Mouse Shaker data was converted in conjunction with corrective measures.

Therefore the conversion details of the Mighty Mouse data are described below in

Section 5.3.2.2.

5.3.6 Corrected Data Files

Three types of corrected data files were created: (1) structural response (2) Mighty

Mouse shaker (3) synchronized structural response and Mighty Mouse shaker. As shown

in Figures 5-8 and 5-9, the procedure used to obtain the corrected structural response data

was identical for all trials. However, the preparation of the corrected Mighty Mouse

shaker data consisted of performing both the conversion and time shift correction

simultaneously, as shown in Figure 5-9 on the left branch. The synchronized structural

response and Mighty Mouse shaker corrected data sets were created by the cross-

correlation procedure described in Section 5.3.3. The time signature of all subevent data

files were converted from Unix time to relative time (sec) from the beginning of

subevent.

5.3.6.1 Corrected Structural Response Data Files

The corrected structural response data (accelerations and displacements) were created by

applying mean removal to the converted data prepared in the manner discussed in Section

5.3.5.1. The mean was computed using Matlab built-in functions across the subevent time

window. This type of baseline correction just shifts the responses relative to the

horizontal axis. The corrected structural response data file signals have units identical to

195
those for the converted data, with the exception of time that is in sec relative to the

window start time (not Unix time).

5.3.6.2 Corrected Mighty Mouse Shaker Data Files

The corrected Mighty Mouse shaker data files were separated into subevents and

acceleration signals were converted to engineering units and baseline corrected. The

subevents start and end times were chosen to be consistent with the structural response

data files discussed in Section 5.3.6.1. The time signature was re-established to relative

time (in sec) from the beginning of each subevent. The conversion of the accelerometers

to gravity (g) units required division by the sensitivity factor of 5 volts/g. The pulsing

data is used to identify the beginning of the sine wave and no conversion was required.

The baseline correction consisted of mean removal, as before. The converted data fields

are similar to the unprocessed data discussed in Section 5.3.2 as shown below.

Channel 1: Time (sec)

Channels 2 and 3: Correlation accelerometers (g)

Channels 4 and 5: Shaker pulses and shaker control frequency

Channels 4 and 5 of the corrected Mighty Mouse shaker data were switched

between testing dates. The details of this switch are provided in Table 5-2. During the

June (braced condition, Trial 2) testing channel 4 is the shaker control frequency data and

channel 5 is the shaker pulses data. However, during the August (unbraced condition,

Trial 4) testing channel 4 is the shaker pulses data and channel 5 is the shaker control

frequency data. No conversion was performed to these channels.

196
5.3.6.3 Corrected Synchronized Structural Response and Mighty Mouse Shaker
Data Files

The forced vibration testing using Mighty Mouse for excitation required synchronizing

and merging of the two data acquisition file types as discussed in Section 5.3.3. Channels

2 and 3 of the Mighty Mouse shaker data were compared, respectively, to channels 16

and 15 of the structural response data. The signals used in the comparison have a

common direction of vibration. Based on the time shift from cross-correlation, t, the

beginning of the corrected Mighty Mouse shaker data was truncated to correlate with the

corrected structural response data. The cross-correlation t was determined as described

in Section 5.3.3 and presented in the last column of Tables 5-4b and 5-6b.

The correlated Mighty Mouse Shaker data was then evaluated for the shaker

frequency, angular position of the shakers rotating mass, and shaker harmonic force. The

shaker frequency and angular position of the rotating mass were determined directly from

the frequency of the shaker control frequency data. The shaker force was determined

based on

F = mr r 2 cos (5-2)

where mr is the rotating mass, is the angular frequency, is the angular position of the

rotating mass, and r is the radius from the center of rotation to the centroid of the rotating

mass. The shaker control frequency channel of the Mighty Mouse shaker data, as shown

in Table 5-2, records a single pulse per revolution of the forcing weight. This pulse is

recorded when the weight is aligned in the positive direction of loading. Therefore, an

197
angular position of = 0 represents when the weights are pointed in the positive

direction of shaking (see Figure 5-7).

The corrected synchronized structural response and Mighty Mouse shaker data

files are similar to the corrected structural response data files, however, they have been

supplemented with the shaker angular frequency, shaker force, angular position of the

shakers rotating mass, and the respective Mighty Mouse shaker data accelerometer that

was used for cross-correlation. The agreement of the cross-correlation for the different

acquisition systems can be obtained by comparing channel 52 with channel 15 or channel

16 dependent if the vibration mode is in the y or x direction, respectively. A summary of

the corrected synchronized structural response and Mighty Mouse shaker data files is

shown below.

Channel 1: Time (sec)

Channels 2 through 43: Accelerometers (g)

Channels 44 through 47: LVDTs (in)

Channel 48: Slated with zeros (column used for Atom Ant shaker mass

acceleration)

Channel 49: Shaker angular frequency

Channel 50: Shaker force (N)

Channel 51: Angular position (rad) of the shakers rotating mass

Channel 52: Mighty Mouse shaker data accelerometer (g) used for cross-

correlation (from either channel 2 [x-direction] or channel 3 [y-direction])

198
5.4 Structural Response of Test Structure

In this section, the processed data is visually inspected to detect obvious errors and to

provide insight into the structural response from forced vibration loading. While all

records were subject to visual inspection, only three loading conditions are discussed in

this section. These three tests represent a spectrum of the loading conditions applied to

Mini Me at GVDA. Table 5-7 describes the test experiments chosen for this discussion

along with the loading direction, shaker type, maximum force applied, and the intensity

descriptor. The intensity descriptors represent the amplitude of forces applied during a

particular test relative to other tests at GVDA (i.e., small, moderate, large). Figure 5-14

shows force histories in the time-domain for the tests referenced in this section. All three

tests described in this section had the shaker installed on the roof with the shaking

applied in the y-direction. Figures 5-15 through 5-17 show structural response histories

for the force histories in Figure 5-14.

Table 5-7 Test Experiment Used for Presentation of Structural Responses Due to Forced
Vibration Testing.
Loading Loading Max Forcec Force
a
Test Experiment (Figure Below) Direction Shakerb (kN) Intensityd
Exp. 3.3.5a (Figure 5-15) Y AA 0.2 Small
Exp. 3.4.1a (Figure 5-16) Y MM 7.1 Moderate
Exp. 3.4.19a (Figure 5-17) Y MM 61.7 Large
a See subevent file name from Tables 5-3 to 5-6. Structural response from experiment shown in bracketed Figure no.
b AA=Atom Ant and MM=Mighty Mouse.
c Maximum force applied from the shaker. See Figure 5-14 for shaker force history.

d Force intensity is relative to the test forces at GVDA.

Figure 5-15 shows the structural response of the unbraced structure with loading

from the Atom Ant shaker. The maximum force applied by the shaker during this

199
experiment was 201 N. This force level produces small shear stresses and shear strains in

the foundation soil that would not be expected to induce nonlinear response in the

foundation soil (discussed further in Chapter 6).

Figures 5-15a and Figure 5-15b show the horizontal structural responses near the

four corners of the foundation and roof, respectively. The acceleration trends in Figures

5-15a and 5-15b look fairly consistent from channel to channel in any particular

direction. An order of magnitude difference is observed between the maximum

accelerations in the direction of applied shaking force (y-direction) and the perpendicular

direction (x-direction). Comparing the responses shown in Figures 5-15a and 5-15b, it is

observed that the roof acceleration in any given direction is about an order of magnitude

greater than the foundation acceleration. These features are particularly evident at about

12 to 15 sec.

Figures 5-15c and 5-15d show the vertical response of the structure. Figure 5-15c

shows that the foundation and roof vertical accelerations are similar to each other. Figure

5-15d shows displacements of the foundation from LVDTs. The displacements on the

eastern side (channel nos. 46 and 47) are consistent with each other, having a maximum

displacement of approximately 0.001 cm (0.0004 in). The western side (channel nos. 44

and 45) shows more displacement than the eastern side with having a maximum

displacement of approximately 0.0013 cm (0.0006 in). It should be noted that Figure 5-

15d accounted for the correction of the artificial errors as a result of incorrect calibration

factors being used during data processing as discussed in Section 5.3.5.1. (Channel nos.

200
44, 46, and 47 were corrected by multiplication of the response history by 10 and then

converted to cm form in for presentation).

Figure 5-15e shows the horizontal foundation accelerations for the adjacent

structure (Dr. Evil), ground accelerations between the two structures, and foundation

accelerations from Mini-Me. The accelerometers on Dr. Evil require a baseline correction

that is time-dependent, opposed to the previously applied linear mean correction. In the

direction of shaking, accelerations on the foundation of the Dr. Evil structure are

approximately 5% of those on the Mini-Me foundation. Peak accelerations on the ground

between the two structures are approximately 15% of those on the Mini-Me foundation.

Figure 5-16 shows the response of the unbraced structure with loading from the

Mighty Mouse shaker (5% eccentricity). The maximum force applied by the shaker

during this experiment was 7.1 kN, which can be described as a moderate force. This

level of excitation could reasonably be expected to produce some nonlinearity in the

response of the foundation soils (discussed further in Chapter 6).

Figures 5-16a and Figure 5-16b show the horizontal structural responses near the

four corners of the foundation and roof, respectively. As before, the shaking is much

stronger in the direction of applied shaking (y-direction) than in the perpendicular x-

direction. A noteworthy feature in these plots is the different waveforms recorded by

channels 6 and 9 (Figure 5-16a) during the approximate time intervals of 75-95 sec and

200-300 sec. In the former interval (near 85 sec), at the scale of the plot shown in Figure

5-16a, the responses appear to be reversed from each other (i.e., simultaneous peaks in

Channel 6 and 9 of 0.04g and -0.04g). In the latter interval (between 200-300 sec), the

201
channel 9 signal exhibits a s plateau-type appearance while channel no. 6 lacks that

plateau feature and gradually reduces in amplitude with time. These same features can be

seen in the roof responses in Figure 5-16b. These distinct characteristics of the west and

east signals may result from torsional responses of the structure, perhaps arising from

loosening of bolts securing the columns to the top and bottom slabs (these bolts were

periodically re-tightened in the field during the large amplitude tests). As expected,

Figures 5-16a and 5-16b show that the roof accelerations are about an order magnitude

larger than foundation accelerations.

Figure 5-16c and 5-16d show the vertical response of the structure. Figure 5-16c

shows that the foundation and roof vertical accelerations are similar. Figure 5-16d shows

foundation displacements from LVDT recordings. Displacements on the northeast corner

(channel 46) were at the noise level so this instrument appears to have been disconnected.

Channels from the south side (nos. 44 and 47) are fairly consistent, having maximum

displacements around 0.013 cm (0.005 in), channel 44 at the southwestern corner has

slightly larger displacements. The northwest LVDT (channel 45) is fairly consistent with

channel 47. Figure 5-16d accounted for the aforementioned correction as applied for

Figure 5-15d.

Figure 5-16e shows the horizontal accelerations of the adjacent structure (Dr.

Evil), ground accelerations, and foundation accelerations from Mighty Mouse excitation

of Mini-Me. As before, the foundation accelerograms from Dr. Evil have baseline drift.

Compared to foundation accelerations on the excited structure, intermediate ground

202
accelerations are lower by a factor of 9 and Dr Evil foundation accelerations are lower by

a factor of 40.

Figure 5-17 shows the structural response of the unbraced structure with loading

from the Mighty Mouse shaker operating at maximum capacity that produced a peak

force of 61.7 kN. This can be described as a large force. A substantial nonlinear

response of the foundation soils is expected from this experiment (discussed further in

Chapter 6).

Figures 5-17a and Figure 5-17b show horizontal structural accelerations near the

four corners of the foundation and roof, respectively. A noteworthy feature in these plots

is the different waveforms recorded by channels 15 and 24 (Figure 5-16b) during the

approximate time intervals of 75-100 sec and 325-350 sec. At the scale of the plot shown

in Figure 5-16b, the responses appear to be reversed from each other (i.e., simultaneous

peaks in channels 15 and 24 of 0.25g and -0.25g). These same features can be seen in the

foundation responses in Figure 5-17a. These distinct characteristics of the west and east

signals may result from torsional responses of the structure, perhaps arising from

loosening of bolts securing the columns to the top and bottom slabs (these bolts were

periodically re-tightened in the field during the large amplitude tests). Structural

accelerations perpendicular to the loading direction in Figure 5-17a and Figure 5-17b are

about an order of magnitude lower in amplitude than those in the loading direction. The

perpendicular accelerations peak at the same time intervals as the different horizontal

responses noted previously, which supports the notion that these features are associated

with a torsional mode response.

203
 Figure 5-17c and 5-17d show the vertical response of the structure. Many of the

trends are similar to those noted at lower shaking levels: (1) similar amplitudes at roof

and foundation (Figure 5-17c), and (2) similar foundation displacements on the eastern

and western sides (Figure 5-17d). A new features not seen in previous tests is that the

LVDT displacements in Figure 5-17c are not symmetric, being much larger in the up

direction (positive) than down (negative direction), implying the formation of gapping.

Figure 5-17e shows the horizontal accelerations of the adjacent structure (Dr.

Evil), ground accelerations, and foundation accelerations from Mighty Mouse excitation

of Mini-Me. As before, the foundation accelerograms from Dr. Evil have baseline drift.

In the direction of shaking, accelerations on the foundation of the Dr. Evil structure are

approximately 5% of those on the Mini-Me foundation. Peak accelerations on the ground

between the two structures are approximately 14% of those on the Mini-Me foundation.

204
Figure514 Shakerforceloadinghistoriesthatrepresentthesmall,moderateandlarge
forcesusedduringforcedvibrationtestingatGVDA.Thetitlesoftheshaker
forcescorrespondtoTable57.

205


Figure515a ThehorizontalstructuralresponsesofthefoundationareshownforExp.3.3.5a(Unbracedcondition,AtomAntshaker
loadingontheroofintheydirection).Thelocalaxesalongwithchannelidentificationareshowninthemiddle,whilethe
structuralresponsesareshownintheperipheral.

206

Figure515b ThehorizontalstructuralresponsesoftheroofareshownforExp.3.3.5a(Unbracedcondition,AtomAntshakerloadingon
theroofintheydirection).Thelocalaxesalongwithchannelidentificationareshowninthemiddle,whilethestructural
responsesareshownintheperipheral.

207

Figure515c TheverticalstructuralresponsesareshownforExp.3.3.5a(Unbracedcondition,AtomAntshakerloadingontheroofinthey
direction).Thelocalaxesalongwithchannelidentificationareshowninthemiddle,whilethestructuralresponsesareshown
onatthesides.

208

Figure515d TheverticalstructuraldisplacementsbasedonLVDTrecordingsareshownforExp.3.3.5a(Unbracedcondition,AtomAnt
shakerloadingontheroofintheydirection).Thelocalaxesalongwithchannelidentificationareshowninthemiddle,while
thestructuralresponsesareshownonaboveandbelow.

209

Figure515e Thehorizontalstructuralresponsesoftheadjacentstructure(Dr.Evil)areshownincomparisontothefreefieldandMiniMe
responsesforExp.3.3.5a(Unbracedcondition,AtomAntshakerloadingontheroofintheydirection).ThelocalaxesofMini
Meisshownalongwithchannelidentificationsoftheresponses.Dr.Evilfoundationplanviewisshownatthetopleftandthe
crosssectionrelationshipofthestructuresandfreefieldisshownatthetopright.

210

Figure516a ThehorizontalstructuralresponsesofthefoundationareshownforExp.3.4.1a(Unbracedcondition,MightyMouseshaker
with5%eccentricityloadingontheroofintheydirection).Thelocalaxesalongwithchannelidentificationareshowninthe
middle,whilethestructuralresponsesareshownintheperipheral.

211

Figure516b ThehorizontalstructuralresponsesoftheroofareshownforExp.3.4.1a(Unbracedcondition,MightyMouseshakerwith5%
eccentricityloadingontheroofintheydirection).Thelocalaxesalongwithchannelidentificationareshowninthemiddle,
whilethestructuralresponsesareshownintheperipheral.

212

Figure516c TheverticalstructuralresponsesareshownforExp.3.4.1a(Unbracedcondition,MightyMouseshakerwith5%eccentricity
loadingontheroofintheydirection).Thelocalaxesalongwithchannelidentificationareshowninthemiddle,whilethe
structuralresponsesareshownonatthesides.

213

Figure516d TheverticalstructuraldisplacementsbasedonLVDTrecordingsareshownforExp.3.4.1a(Unbracedcondition,MightyMouse
shakerwith5%eccentricityloadingontheroofintheydirection).Thelocalaxesalongwithchannelidentificationareshown
inthemiddle,whilethestructuralresponsesareshownonaboveandbelow.

214

Figure516e Thehorizontalstructuralresponsesoftheadjacentstructure(Dr.Evil)areshownincomparisontothefreefieldandMiniMe
responsesforExp.3.4.1a(Unbracedcondition,MightyMouseshakerwith5%eccentricityloadingontheroofinthey
direction).ThelocalaxesofMiniMeisshownalongwithchannelidentificationsoftheresponses.Dr.Evilfoundationplan
viewisshownatthetopleftandthecrosssectionrelationshipofthestructuresandfreefieldisshownatthetopright.

215

Figure517a ThehorizontalstructuralresponsesofthefoundationareshownforExp.3.4.19a(Unbracedcondition,MightyMouseshaker
with100%eccentricityloadingontheroofintheydirection).Thelocalaxesalongwithchannelidentificationareshownin
themiddle,whilethestructuralresponsesareshownintheperipheral.

216

Figure517b ThehorizontalstructuralresponsesoftheroofareshownforExp.3.4.19a(Unbracedcondition,MightyMouseshakerwith
100%eccentricityloadingontheroofintheydirection)Thelocalaxesalongwithchannelidentificationareshowninthe
middle,whilethestructuralresponsesareshownintheperipheral.

217

Figure517c TheverticalstructuralresponsesareshownforExp.3.4.19a(Unbracedcondition,MightyMouseshakerwith100%
eccentricityloadingontheroofintheydirection).Thelocalaxesalongwithchannelidentificationareshowninthemiddle,
whilethestructuralresponsesareshownonatthesides.

218

Figure517d TheverticalstructuraldisplacementbasedonLVDTsareshownforExp.3.4.19a(Unbracedcondition,MightyMouseshaker
with100%eccentricityloadingontheroofintheydirection).Thelocalaxesalongwithchannelidentificationareshownin
themiddle,whilethestructuralresponsesareshownonatthesides.

219

Figure517e Thehorizontalstructuralresponsesoftheadjacentstructure(Dr.Evil)areshownincomparisontothefreefieldandMiniMe
responsesforExp.3.4.19a(Unbracedcondition,MightyMouseshakerwith100%eccentricityloadingontheroofinthey
direction).ThelocalaxesofMiniMeisshownalongwithchannelidentificationsoftheresponses.Dr.Evilfoundationplan
viewisshownatthetopleftandthecrosssectionrelationshipofthestructuresandfreefieldisshownatthetopright.

220

5.5 Conclusions

Forced vibration testing of a shallow foundation field test structure was successfully

performed at GVDA in June and August 2011. The structure was loaded cyclically by a

small linear mass shaker that imparted consistent force amplitudes over the frequency

range 0-50 Hz and by a large eccentric mass shaker that imparted frequency-dependent

forces over the frequency range 0-19 Hz. The test structure was well instrumented with

accelerometers and LVDTs to capture the structural response histories. The response

histories have been processed from the unprocessed data files to convert to engineering

units, convenient time formats, and to remove baseline errors. The data files created from

the forced vibration testing at GVDA have been archived at the NEEShub

(https://nees.org/warehouse/project/637).

This chapter provides a detailed description of GVDA soil conditions, test

structures, sensor arrays, data acquisition systems, shaker types and forces, field test logs,

and the data processing procedures used for the GVDA SSI force vibration testing.

Section 5.4 presented a review of the processed data files and the resultant

structural responses from the forced vibration loading at GVDA. Visual inspection

revealed several problems a disconnected LVDT (channel 46), baseline errors of

accelerations from the foundation of adjacent structure Dr. Evil, and conversion error of

three vertical foundation LVDTs (channels44, 46, and 47). The conversion error can be

corrected by multiplying the converted data channels by a factor of 10.

The observed responses indicate no significant torsional response for the lowest

shaking level, but asymmetric responses suggestive of a torsional response for the larger

221
shakers. Gapping was observed in the LVDT displacements applied by the strongest

shaking, which is consistent with visual observations from the field.

222
6 EVALUATIONOFSOILSTRUCTURE
INTERACTIONEFFECTSFROM
FORCEDVIBRATIONTESTING

6.1 Introduction

As described in Chapter 5, forced vibration testing of a well-instrumented test structure

was performed to evaluate soil-structure interaction (SSI) effects. The experiments were

performed at the Garner Valley Downhole Array (GVDA) site operated by

NEES@UCSB. Forced vibration testing of our test structure, hereafter referred to as

Mini-Me, was previously performed at the Wildlife Liquefaction Array (WLA) and at

the UCLA laboratory (Star, 2011). Mini-Me was the first structure to be tested with

varying foundation conditions (i.e. different underlying soils), so that we could examine

the impact of variable soil conditions on SSI effects. In addition, the GVDA site

conditions allowed for instrumentation of the adjacent, existing structure and a large

rectangular concrete slab during testing. The adjacent structure, hereafter referred to as

Dr. Evil, was instrumented for evaluation of structure-soil-structure interaction (SSSI)

from the vibration of Mini-Me. Also, an existing concrete slab in the parking lot adjacent

to the GVDA site was instrumented for evaluation of kinematic interaction effects.

223
6.2 Previous Forced Vibration Testing of Field Test Structures to Evaluate
Impedance Functions

Limited full-scale forced vibration testing studies have been performed with the

configuration of arrays necessary to evaluate SSI. As discussed in Section 5.1, the

evaluation of SSI effects on buildings requires horizontal accelerometers at the free-field,

foundation, and roof, and two vertical accelerometers on the foundation to evaluate

rocking response (Stewart and Fenves, 1998). In the case of forced vibration testing, the

free-field motion is not required since the external force is introduced directly to the

structures foundation or roof and the free-field motion is taken as zero. Table 6-1 shows

the attributes of previous forced vibration testing studies involving large-scale test

structures that have been instrumented to evaluate SSI. The studies listed in Table 6-1

investigated the responses of their respective structures to evaluate the real and imaginary

parts of the SSI system impedance functions and to compare them to theoretical models.

Recall from Section 2.3.1 that the soil-foundation stiffness is represented by only the real

part, whereas the damping is represents by a ratio of the real and imaginary parts (Eq. 2-

5).

224
Table 6-1 Summary of Field Forced Vibration Tests Used to Evaluate Foundation
Impedance Functions (NIST, 2012)
Excitation Results
Foundation Embed. 2 3 4 ~ 5 Impedance Freq. range
f1 (Hz) V s (m/s) Source Freq. (Hz) T T Reference
Dimensions (m)1 obtained (Hz)6
1.3
Shaker on k x, cx, modal freq. Lin and Jennings,
3 3m 0-1.5 17.5 305 7-70 (D=1.5);
ground k yy , c yy only 1984
1.5 (D = 0)
NS: 0.8-
NS: 2.16; Shaker on NS:1.06; k x, cx, NS:0.8-2.5; Luco et al., 1988;
25 25m 4-5.5 300 2.5; EW:
EW: 1.26 roof EW:1.1 k yy , c yy EW:0.8-1.75 Wong et al., 1988
0.8-1.75
1.31.3m; Shaker on k x, cx, Crouse et al.,
0 n/a 120; 75 10-60 n/a 0-60
1.2 1.1m fndn. k yy , c yy 1990
k x, cx,
Shaker on
diam = k yy , c yy , DeBarros and
5.2 9.37 300 roof & 2-20 2 5-14
10.8m k x-yy , c x Luco, 1995
fndn
yy , k z, c z
Shaker on k x, cx, Tileylioglu et al.
4.1 4.1m 0 6.0, 13 198 5-15 1.15, 1.29 5-15
roof k yy , c yy 2011

Braced Braced
NS: 31.2; NS: 3.06;
Shaker on
EW: 23.0; EW: 3.03; k x, cx,
4.3 2.1m 0 106; 98 roof & 4-45 4-45 Star 2011
Unbraced Unbraced k yy , c yy
fndn
NS: 11.5; NS: 1.35;
EW: 12.1 EW: 1.81
1
Foundation embedment depth; 2 Fundamental mode, fixed-base frequency; 3Vs=Shear wave velocity of soil; 4Frequency range
and 5Period lengthening. Symbols: D=embedment depth; n/a=not available; NS=North-South building axis;
EW=East-West building axis; diam=diameter; and fndn.=foundaiton

Lin and Jennings (1984) constructed a square 3 m (10 ft) single story structure

overlying 300 m (1,000 ft) of young river alluvial deposits with a Vs = 300 m/s (984 ft/s).

The water table was located at 76 m (250 ft) below the ground surface. An external

shaking force located 15 m (49 ft) from the structure was used to produce horizontal

incident SH-waves. This was done in an attempt to represent earthquake excitation.

Forced vibration was applied with the structure situated at three different embedment

depths. The impedance results for translation and rocking of the nonembedded structure

were found in good agreement with the analytical model of Veletsos and Wei (1971).

However, the embedment effects of the testing were found to be larger than the analytical

225
models (Beredugo and Novak, 1972, Elsabee and Morray, 1977, Luco et al., 1975, and

Permalee and Kudder, 1974).

A forced vibration test was performed at the roof of the Millikan Library Building

to evaluate horizontal and rocking impedance functions (Luco et al, 1988 and Wong et al,

1988). The structure is a nine-story reinforced concrete building with approximately 5.5

m (18 ft) of embedment overlying firm soil that has a Vs = 300 m/s (984 ft/s). For

modeling purposes, the foundation was considered rigid with an effective radius of 13.7

m (45ft). The lateral loads are resisted by a shear wall (NS direction) and a central core

(EW direction). Based on the different active systems, the embedment into the soil was

modeled as 4 m (13 ft) in the NS direction and 5.5 m (18 ft) in the EW direction. Figure

6-1 presents the Wong et al. (1988) experimental results versus the theoretical (Apsel and

Luco, 1987) model for rocking and horizontal responses in both the NS and EW

directions. The impedance functions in Figure 6-1 have been normalized by the

characteristic foundation length. The study was confined to a limited frequency range

(0.8-2.5 Hz), however, noteworthy trends were observed. The experimental rocking

impedances (Figures 6-1a and 6-1c) match the theoretical results in both directions fairly

well, with the exception of the underpredicted imaginary part of the EW direction. The

horizontal impedance model overpredicted the experimental results by as much as 50% in

the NS direction (Figure 6-1b).

Crouse et al (1990) performed forced vibration testing at two sites that had

different foundations. The first site (known as Cholame station) consisted of a square

1.3 m (4 ft) reinforced concrete foundation that had 200 cm (8 in) diameter piers at the

226
four corners. The Cholame station site has a Vs = 120 m/s (390 ft/s). The Cholame station

foundation was modeled as a rigid circular foundation with an equivalent radius of 0.73

m (2.4 ft). The concept of an equivalent radius was applied because of the impedance

functions considered in the study, which assume circular foundation geometry.. The

second site (known as Station 6) consisted of a rectangular 1.2 m x 1.1 m (3.9 ft x 3.6

ft) reinforced concrete foundation at ground surface. Station 6 site has a Vs = 75 m/s (250

ft/s). The Station 6 foundation was modeled as rigid. Both foundations overlay soil

deposits that are approximately 70 m (230 ft) of soft to moderately stiff Quaternary

alluvial deposits. Forced vibration was applied at the roof between the frequencies of 10

to 60 Hz. Three theoretical impedance models were evaluated for comparison at these

sites, as follows: 1) using the best estimate of Vs and damping properties: 2) correcting

for overburden pressure from the structure by increasing the Vs values of the upper 0.6 m

(2 ft) by 30%: 3) using Model 2 updated with a higher soil damping (8% instead of 1.5%)

for the upper 0.9 m (3 ft). These models of soil properties were applied with impedance

function solutions presented by Wong and Luco (1985) and Apsel and Luco (1987).

Figures 6-2 and 6-3 show normalized results of the models compared to the

experimentally derived impedance functions for the Cholame station and Station 6 sites,

respectively. Real parts have been normalized by Gr (for translation) and Gr3 (for

rocking), while the imaginary parts are normalized by a0, where r is the effective radius

and a0r = r Vs . For comparison, the soil and foundation parameters used for

normalization purposes are: G = 9.44 MPa (0.197 psf), r = 0.61 m (2 ft) and Vs = 73 m/s

(240 fps) as described by Crouse and Hushmand (1989).

227
 Crouse et al. (1990) showed that the impedance functions are frequency-

dependent at both sites (Figures 6-2 and 6-3). The Cholame station (Figure 6-2)

horizontal impedances are generally lower than the models, while the rocking

impedances are larger than expected. Station 9 (Figure 6-3) showed both the horizontal

and rocking impedances to be larger than expected. Crouse et al (1990) found a better

theoretical fit to the experimental data by including the overburden pressure and by

increasing the soil damping (Model 3). The predicted theoretical impedance functions

based on no modification to the measured Vs or soil damping (Model 1) procedure

typically had the worst comparison to the experimental data.

228
Figure61 Impedance functions derived from forced vibration testing of the Millikan
LibraryBuilding(Wongetal.,1988).Thefigureshowstheimpedancefunctions
in the northsouth direction for a) rocking and b) horizontal and in the east
westforc)rockingandd)horizontal.

229
Figure62 Impedance functions for Cholame station in the horizontal (ab) and rocking
(cd)modesofvibration.(Crouseetal.,1990).

Figure63 Impedance functions for Station 6 in the horizontal (ab) and rocking (cd)
modesofvibration.(Crouseetal.,1990).

230
 Further evaluation of the theoretical models was completed by forced vibration

testing of the one-quarter scale Hualien containment model (de Barros and Luco, 1995).

The Hualien model was constructed to evaluate the impedance functions for nuclear

power plants. The Hualien model is 16.13 m (52.9 ft) tall, with a circular 10.82 m (35.5

ft) diameter foundation embedded to a depth of 5.15 m (16.9 ft). The soils at the site

consist of sand and gravel that have been extensively studied by CRIEPI (Central

Research Institute of Electric Power Industry, Japan) and IES (Institute of Earth Sciences,

China). In an effort to validate the existing theoretical models evaluated in Crouse et al.

(1990) and Luco and Wong (1990, 1992), multiple models varying the Vs based on

existing conditions and structural overburden pressure were compared to the

experimental impedance functions of the Hualien model. Figure 6-4 shows the impedance

functions as stiffness and dashpot coefficients for the EW direction of the Hualien model.

For comparison purposes, model-based impedances computed using two equivalent Vs

values are also shown: 1) Vs = 317 m/s (1040 ft/s) for the soils below the foundation

(neglecting structural overburden) and 2) Vs = 382 m/s (1250 ft/s) by updating Model 1

Vs by accounting for excavation and structural overburden effects. de Barros and Luco

(1995) generally found the horizontal and rocking impedance functions were

overpredicted. A noteworthy exception, as seen previously in Figures 6-1 through 6-3, is

that the experimental rocking dashpot (or imaginary component) exceeded the theoretical

models.

Impedance functions from forced vibration testing of the Dr. Evil test structure at

GVDA have been previously studied by Tileylioglu et al. (2011). The details of the Dr.

231
Evil test structure and GVDA were described in Chapter 5. Figure 6-4 shows a

normalized comparison of the experimentally derived impedance functions (as stiffness

and damping ratios) and the Pais and Kausel (1988) theoretical model for rectangular

foundations (Table 2-1). The test structure was in an unbraced condition and was loaded

by a small linear mass shaker mounted to the bottom of the roof slab. Three suspension

logging and spectral analysis of surface waves (SASW) arrays were situated adjacent to

the structure and an averaged shear wave velocity of the soil was found to be Vs=198 m/s

(650 ft/s), with a 15 m/s (50 ft/s) range. The theoretical horizontal impedance functions

match well with only a slight overprediction, while the rocking damping was

underpredicted.

The Mini-Me test structure discussed in Section 5.2.2 was previously situated and

tested by forced vibrations at the WLA site (Star, 2011). The WLA site is maintained as

part by NEES@UCSB and is composed of softer materials than GVDA. The WLA site

soils are generally consistent with about a 2.5 m (0.76 ft) to 3.0 m (0.91 ft) thick layer of

silty and clay underlain by a granular layer that is 3.5 m (1.07 ft) to 4.0 m (1.22 ft) thick.

The granular layer is underlain by silt and clay. The forced vibration was applied with the

linear mass shaker and large Mighty Mouse shaker described in Section 5.2.5. The use of

the large shaker is unique in that the previous studies maintained small forces to avoid a

nonlinear response of the soil. Figures 6-6 and 6-7 show the experimental results of the

impedance functions (stiffness and damping ratios) compared to the theoretical Pais and

Kausel (1988) model using two Vs evaluations: 1) Vs = 106 m/s (347.8 ft/s) accounting

for structural overburden and 2) Vs = 98 m/s (321.7 ft/s) as determined from crosshole

232
seismic testing of the free-field material. Figure 6-6 shows the impedance functions

produced from small forces. As with the other studies, the model tends to overpredict the

stiffness slightly. Stable results for the impedance functions were found from about 4 Hz

to 20 Hz. Figure 6-7 shows the influence of large forces on the impedance functions. At

small frequencies the stiffness is similar to the small force tests, however as frequency

increases there is a significant stiffness decrease. The softening of the stiffness is shown

to correspond to the shearing of the soil and the induced nonlinearity. Figure 6-7 show

the estimated soil nonlinearity for each mode of vibration. The process of estimating the

soil strains and shear modulus reductions are explained in Section 6.6.2. The damping

component response shows a similar underprediction to the models.

Figure64 Normalized impedance functions for Hualien containment model in the
horizontal (ab) and rocking (cd) modes of vibration. (de Barros and Luco,
1995).

233

Figure65 NormalizedimpedancefunctionsfortestingofDr.EvilasGVDAforhorizontal
(top)androcking(bottom)modesofvibration.(Tileyliogluetal.,2011).

234

Figure66 Impedance functions for testing of MiniMe at WLA for horizontal (top) and
rocking(bottom)modes ofvibration usingsmallshaker. Exp2.1.7(Unbraced
condition, Atom Ant shaker loading on the roof in the xdirection). (Star,
2011).

235

Figure67 Impedance functions for testing of MiniMe at WLA for horizontal (top),
rocking(middle)modesofvibrationusingalargeshaker.Thebottomgraphic
shows a representation of the soil nonlinearity during testing. Exp. 2.2.224b
(Unbracedcondition,MightyMouseshaker(100%ecc.)loadingontheroofin
theydirection).(Star,2011).

236
6.3 Inversion Procedures for Evaluation of Impedance Ordinates from Forced
Vibration Test Data

Several procedures have been presented for evaluating impedance functions from forced

vibration testing of structures (Tileylioglu et al, 2011, de Barros and Luco, 1995, Crouse

et al, 1990, etc.). This study uses the Tileylioglu et al (2011) approach, in which an

above-ground single degree-of-freedom (SDOF) system was modeled to represent our

field test structure. The SDOF system shown in Figure 6-8 represents the soil-foundation-

structure system being excited by a shaker force (Fshaker).

Figure68 Sketch showing the soilstructure interaction system used for determination
ofimpedancefunctionsfromforcedvibrationtestingbasedonTileyliogluetal
(2011).

The Tileylioglu et al (2011) type model in Figure 6-8 shows the inclusion of the

frequency-dependent impedance functions for horizontal translation ( k x ), vertical

translation ( k z ) and rotation ( k yy ) where the overbar (-) represents a complex value. It

237
should be noted that the subscript notation assumes excitation in the x-z plane, however,

excitation in the y-z plane can also be represented by reversing the x and y subscripts.

Analysis of the impedance functions requires measurement of the foundation translation

(uf) and rotation ( f ) . The total roof displacement is:

u roof = u f + h f + u s (6-1)

Where h is the height of the center of the structural mass (ms) for the SDOF, and us is the

displacement of the top mass associated with the structural compliance as represented by

structural stiffness (ks) and damping (cs) coefficients. The equation of motion for the

system in Eq. (6-2) can be written as (Clough and Penzien, 1993):

M U&& + C U& + KU = F (6-2)

where U represents the horizontal displacements and rotations of each degree of freedom:

U = [u f f u s ]T . The vertical foundation and rotational superstructure

displacements were not included for the horizontal response. The mass (M), dashpot (C)

and stiffness (K) matrices given by Tileylioglu et al (2011) and Crouse et al. (2001) are:

m f + ms m f h f + ms h ms

M = m f h f + ms h I f + m f h f 2 + ms h 2 ms h (6-3)
ms ms h m s

kx k yx 0

K = k yx k yy 0 (6-4)
0 0 k s

238
 cx c yx 0

C = c yx c yy 0 (6-5)
0 0 c s

where mf is the foundation mass, ms is the superstructure mass, If is the mass moment of

inertia about the foundation, hf is the height of the foundation, and the subscripts x, y, xy

and yx represent the modes of vibration. As discussed before, the notation for excitation

in the x-z plane is shown with x representing vibration in the horizontal x-direction, yy

representing rotation about the y-axis and xy, and yx representing coupling terms between

the off-diagonal soil impedance terms. These coupling terms have been taken as zero for

this study because the foundations are not embedded.

The aforementioned SDOF system equations can be divided into three distinct

equations representing the foundation translation (Eq. 6-6), foundation rotation (Eq. 6-7)

and structural translation (Eq. 6-8).

ms (u&& f + h&&f + u&&s ) + m f (u&& f + h f &&f ) + c x u& f + k x u f = Fs (6-6)

ms h(u&& f + h&&f + u&&s ) + I f &&f + m f h f (u&& f + h f &&f ) + c yy& f + k yy f = hFs (6-7)

ms (u&& f + h&&f + u&&s ) + cs u& s + k s u s = Fs (6-8)

The unknown impedance functions are contained in Eqs. (6-6) and (6-7). Through

transformation to the frequency domain and presenting those equations as displacements,

the foundation translation and rotation can be written as Eqs (6-9) and (6-10),

respectively:

2 m s (u f + h f + u s ) 2 m f (u f + h f f ) + ic x u f + k x u f = Fs (6-9)

239
 2 m s h(u f + h f + u s ) 2 I f f 2 m f h f (u f + h f f ) + ic yy f + k yy f = hFs (6-10)

where the hat symbol (^) represents a frequency domain variable. Recalling the definition

of the impedance function from Eq. (2-3), then Eqs. (6-9a) and (6-10b) can be rewritten

as:

2 m s (u f + h f + u s ) 2 m f (u f + h f f ) + k x u f = Fs (6-11)

2 m s h(u f + h f + u s ) 2 I f f 2 m f h f (u f + h f f ) + k yy f = hFs (6-12)

Eqs. (6-11) and (6-12) can now be rearranged and expressed as a definition of the

impedance functions as:

Fs + 2 m s (u f + h f + u s ) + 2 m f (u f + h f f )
k x = (6-13)
u f

hFs + 2 m s h(u f + h f + u s ) + 2 I f f + 2 m f h f (u f + h f f )
k yy = (6-14)
f

Eqs. (6-13) and (6-14) represent the complex-valued impedance functions for the

modes of translation and rotation, respectively. As shown in Chapter 2, Eq. (2-5), a

damping ratio (j) can be derived for each mode based on the real and imaginary parts of

the complex-valued impedance function. Eq. (2-8) can be expressed using the complex-

valued impedance as Eq. (6-15).

imag ( k j )
j = (6-15)
2 * real ( k ) j

240
6.4 Forced Vibration and Inertial Responses of Test Structure at GVDA

Prior to analysis of the impedance functions, the input acceleration histories used in the

evaluation of the translational impedance (Eq. 6-13) and rotational impedance (Eq. 6-14)

are reviewed. The structural response histories used for the response characteristics have

been described in Section 5.4. This section will present the frequency-domain structural

response, forcing function and structural inertia components for the small, moderate, and

large amplitude force cases presented in Figure 5-14. Each force-response figure is

followed by load-deformation loops for the base shear versus horizontal displacement

and foundation moment versus rotation at varying frequencies. The loops are diagnostic

of the stiffness and damping produced from the forced vibration frequencies. For

comparison of the accelerations in the force-response figures, the rocking and torsion

responses are presented as linear acceleration quantities. The linear quantities are

calculated by multiplying the rotation of the rocking by the height (h) and the torsion by

the distance between accelerometers (L) where the subscripts represent the

accelerometer distance on the roof (r) and foundation (f).. The forces (F) are described

using subscript s for superstructure, f for foundation and secondary subscript to

indicate translation (y) or rocking (xx) mode of vibration.

The small amplitude forces and responses are shown in Figure 6-9. The shaker

force is the predominant force at small (4-7 Hz) and high (>13 Hz) frequencies, while the

translational inertia dominates around the resonant frequency. The foundation inertial

rocking component is greater than the translational foundation inertia. As expected, the

responses are greatest near the resonant frequency of the structure with the exception of

241
the torsional response. The torsion of the structure and foundation occur after the

resonance frequency is reached at about 17 Hz. It is noteworthy that the horizontal forces

produced from rocking are similar in trend and amplitude with the translational

responses. The loops in Figure 6-10 show a fairly consistent inclination across the

frequencies. This consistency indicates that the stiffness for both translation and rotation

are frequency-independent for the small force amplitude testing. However the loops are

not very smooth at small forces and might be presenting artificial results.

The moderate amplitude forces and responses are shown in Figure 6-11. The

shaker force becomes larger as frequency increases. The foundation and structural inertia

forces are relatively consistent from 5-12 Hz, with the dominant force of inertia is roof

translation associated with the foundation rotation. The greatest response was the

horizontal acceleration of the roof from the foundation rocking. The rocking response

was about twice as large as the horizontal structural response, which is the next dominant

response. There was a small amount of torsion displayed after the resonant frequency.

The loops in Figure 6-12 show that the loops inclination changes with frequency. This

change shows a frequency-dependency of the stiffness for both translation and rotation

during medium amplitude testing, unlike the case for the small amplitude testing. The

loops for the medium forcing test are found to be smoother than the small forcing test due

to the increase loading and deformations.

The medium amplitude forces and responses are shown in Figure 6-11. The

shaker force becomes larger as frequency increases. The foundation and structural inertia

forces are show a distinctive peak around 8 Hz and then the structural inertia force

242
follows a similar trend as the shaker force. The greatest responses are the horizontal

acceleration of the roof from the foundation rocking and the horizontal structural

response. These responses follow a similar amplitude and trend. There is significantly

more torsion compared to the small amplitude tests, where a distinctive peak is observed

at 16 Hz. The loops in Figure 6-12 show that the loops inclination changes with

frequency. This change shows a frequency-dependency of the stiffness for both

translation and rotation during medium amplitude testing. The loops for the medium

forcing test are found to be smoother than the small forcing test due to the increase

loading and deformations.

The large amplitude forces and responses are shown in Figure 6-13. The shaker

force becomes larger as frequency increases. The foundation and structural inertia forces

are relatively consistent from 5-12 Hz, with the dominant force of inertia is roof

translation associated with the foundation rotation. The greatest response was the

horizontal acceleration of the roof from the foundation rocking. The rocking response

was about twice as large as the horizontal structural response, which is the next dominant

response. There was a small amount of torsion displayed after the resonant frequency.

The loops in Figure 6-14 show that the loops inclination changes with frequency. This

change shows a frequency-dependency of the stiffness for both translation and rotation

during large amplitude testing, similar to the observed load-deformation of the medium

amplitude testing.

243
Figure69 Frequencydomainpresentationofthea)responsesandb)forcecomponents
ofExp.3.3.5a(Unbracedcondition,AtomAntshakerloadingontheroofinthe
ydirection).RefertoFigure68fordefinitions.

Figure610 Response correlation of the shear versus translation and moment versus
rotation ofExp.3.3.5a(Unbracedcondition,AtomAntshakerloadingonthe
roofintheydirection).

244
Figure611 Frequencydomainpresentationofthea)responsesandb)forcecomponents
ofExp.3.4.1a(Unbracedcondition,MightyMouseshaker(5%ecc.)loadingon
theroofintheydirection).RefertoFigure68fordefinitions.

Figure612 Response correlation of the shear versus translation and moment versus
rotation of Exp. 3.4.1a (Unbraced condition, Mighty Mouse shaker (5% ecc.)
loadingontheroofintheydirection).

245
Figure613 Frequencydomainpresentationofthea)responsesandb)forcecomponents
ofExp.3.4.19a(Unbracedcondition,MightyMouseshaker(100%ecc.)loading
ontheroofintheydirection).RefertoFigure68fordefinitions.

Figure614 Response correlation of the shear versus translation and moment versus
rotation of Exp. 3.4.19a (Unbraced condition, Mighty Mouse shaker (100%
ecc.)loadingontheroofintheydirection).

246
6.5 Impedance Functions from Forced Vibration Testing

As discussed in Chapter 5, two shakers were used to apply harmonic loads to Mini-Me. A

linear mass shaker was used to produce small-amplitude forces for which shear stresses

and shear strains would be expected to be in the linear-elastic range. An eccentric mass

shaker was used to produce relatively large-amplitude forces, for which nonlinear

response of the foundation soils is expected.

Within this section, foundation stiffness and damping inferred from test data are

compared to predictions from the impedance function model of Pais and Kausel (1988)

(Tables 2-1 and 2-2). Two representations of the subsurface soil conditions were

developed based on the Vs profiles discussed in Section 5.2.1: Model 1 refers to the use

of Vs = 193.5 m/s, which was directly measured from crosshole testing with the weight of

the structure acting on the soil; Model 2 refers to the use Vs0 = 175.5 m/s (575.8 ft/s),

which represents equivalent free-field velocities back-calculated using Eq. (5-2) with a

site-specific n-value (free-field velocities were not directly measured).

6.5.1 Small Amplitude Forced Vibration Testing

The small amplitude forced vibration testing consisted of shaking Mini-Me in both the

braced and unbraced stiffness conditions. Experimental impedance functions for a)

translation and b) rotation were back-calculated using the Tileylioglu et al. (2011) and

Crouse et al. (2001) method (Eqs. 6-15 and 6-16). The structural response acceleration

histories and shaker force were used as input parameters. The force history was

determined by multiplying the shaker mass (19.2 kg) by its acceleration history. When

the force was applied at the roof the shaker was secured in place with rope, whereas no

247
restraint was applied when the shaker was mounted on the foundation slab (see Figure 6-

15).

Figures 6-16 through 6-23 show the experimentally derived impedance functions

for the SSI system. Frequency smoothing was applied using three passes of an 11-point

hamming window based on methods described by Mikami et al. (2008). Repetitions of

each test (shaking direction and location) were processed and are shown in Figures 6-15

through 6-18. The experiment numbers associated with each figure can be referenced to

the testing details in Chapter 5 (Tables 5-3 and 5-5). A discussion about the interpretation

of the data is in Section 6.9.1.

Figure615 PhotographsshowingtheAtomAntshakerontheleft,securedinplacewith
ropeduringforcingtestsattheroofslabandontherightfreestandingduring
forcingtestatthefoundationslab.

248
Figure616 Inverted impedance function results from analysis of Exp. 3.3.5 (Unbraced
condition,AtomAntshakerloadingontheroofintheydirection).Theplots
show a) the translation stiffness, b) rocking stiffness, c) translation damping
ratio, and d) rocking damping ratio of the SSI system in the direction of
loading.

249
Figure617 Inverted impedance function results from analysis of Exp. 3.3.1 (Unbraced
condition,AtomAntshakerloadingontheroofinthexdirection).Theplots
show a) the translation stiffness, b) rocking stiffness, c) translation damping
ratio, and d) rocking damping ratio of the SSI system in the direction of
loading.

250
Figure618 Inverted impedance function results from analysis of Exp. 3.1.6 (Braced
condition,AtomAntshakerloadingontheroofintheydirection).Theplots
show a) the translation stiffness, b) rocking stiffness, c) translation damping
ratio, and d) rocking damping ratio of the SSI system in the direction of
loading.

251
Figure619 Inverted impedance function results from analysis of Exp. 3.1.1 (Braced
condition,AtomAntshakerloadingontheroofinthexdirection).Theplots
show a) the translation stiffness, b) rocking stiffness, c) translation damping
ratio, and d) rocking damping ratio of the SSI system in the direction of
loading.

252
Figure620 Inverted impedance function results from analysis of Exp. 3.3.9 (Unbraced
condition,AtomAntshakerloadingonthefoundationintheydirection).The
plots show a) the translation stiffness, b) rocking stiffness, c) translation
dampingratio,andd)rockingdampingratiooftheSSIsysteminthedirection
ofloading.Note:theshakerwasnotexternallymountedtothestructure.

253
Figure621 Inverted impedance function results from analysis of Exp. 3.3.7 (Unbraced
condition,AtomAntshakerloadingonthefoundationinthexdirection).The
plots show a) the translation stiffness, b) rocking stiffness, c) translation
dampingratio,andd)rockingdampingratiooftheSSIsysteminthedirection
ofloading.Note:theshakerwasnotexternallymountedtothestructure.

254
Figure622 Inverted impedance function results from analysis of Exp. 3.1.8 (Braced
condition,AtomAntshakerloadingonthefoundationinthexdirection).The
plots show a) the translation stiffness, b) rocking stiffness, c) translation
dampingratio,andd)rockingdampingratiooftheSSIsysteminthedirection
ofloading.Note:theshakerwasnotexternallymountedtothestructure.

255
Figure623 Inverted impedance function results from analysis of Exp. 3.1.11 (Braced
condition,AtomAntshakerloadingonthefoundationintheydirection).The
plots show a) the translation stiffness, b) rocking stiffness, c) translation
dampingratio,andd)rockingdampingratiooftheSSIsysteminthedirection
ofloading.Note:theshakerwasnotexternallymountedtothestructure.

256
6.5.2 Large Amplitude Forced Vibration Testing

The large amplitude forced vibration testing consisted of shaking Mini-Me in both the

braced and unbraced stiffness conditions. Experimental impedance functions for a)

translation and b) rotation were back-calculated using the Tileylioglu et al. (2011) and

Crouse et al. (2001) method (Eqs 6-15 and 6-16). The structural response acceleration

histories and shaker force were used as input parameters. The shaker force was

determined by methods described in Section 5.3.6.3, Eq. (5-2).

Figures 6-24 through 6-36 show the experimentally derived impedance functions

for the SSI system. The experiment numbers associated with each figure can be

referenced to the testing details in Chapter 5 (Tables 5-4 and 5-6). A discussion about the

interpretation of the data is in Section 6.9.2. Frequency smoothing was applied using 10

passes of an 11-point hamming window based on methods described by Mikami et al.

(2008). The peak shear strain () at the foundation-soil contact was approximated based

on the comparison of the foundation peak velocity (PGV) and soil shear wave velocity as:

PGV
= (6-16)
Vs

where Vs=193.5 m/s (Model 1 value) and PGV was derived from integration of the

average recorded accelerometer response of the foundation. The foundation response was

determined in the x-direction from channel nos. 3, 6, 9, and 12 and in the y-direction with

channel nos. 2, 5, 8, and 11. For each frequency, we calculated the soils shear modulus

reduction (G/G0) and damping (s) using Menqs model (2003) based on the estimated

peak shear strain. The modulus reduction is represented by the ratio of the shear modulus

257
(G0) at small strains and strain-dependent shear modulus (G) Figures 6-24 through 6-36

include plots at the bottom to represent the approximate soil strain, G/G0 and s over the

frequency ranges tested. The frequency-dependency of the strains is expected since the

peak velocities used in Eq. (-16) for evaluation are frequency-dependent.

258
Figure624 Inverted impedance function results from analysis of Exp. 3.4.1a (Unbraced
condition, Mighty Mouse shaker (5% ecc.) loading on the roof in the y
direction).Theplotsshowimpedancefunctionsinthea)translationalandb)
rocking modes of vibration. The bottom plots represent the soil nonlinearity
foreachmode.

259
Figure625 Inverted impedance function results from analysis of Exp. 3.4.3a (Unbraced
condition, Mighty Mouse shaker (10% ecc.) loading on the roof in the y
direction).Theplotsshowimpedancefunctionsinthea)translationalandb)
rocking modes of vibration. The bottom plots represent the soil nonlinearity
foreachmode.

260
Figure626 Inverted impedance function results from analysis of Exp. 3.4.5a (Unbraced
condition, Mighty Mouse shaker (25% ecc.) loading on the roof in the y
direction).Theplotsshowimpedancefunctionsinthea)translationalandb)
rocking modes of vibration. The bottom plots represent the soil nonlinearity
foreachmode.

261
Figure627 Inverted impedance function results from analysis of Exp. 3.4.17a (Unbraced
condition, Mighty Mouse shaker (50% ecc.) loading on the roof in the y
direction).Theplotsshowimpedancefunctionsinthea)translationalandb)
rocking modes of vibration. The bottom plots represent the soil nonlinearity
foreachmode.

262
Figure628 Inverted impedance function results from analysis of Exp. 3.4.19a (Unbraced
condition, Mighty Mouse shaker (100% ecc.) loading on the roof in the y
direction).Theplotsshowimpedancefunctionsinthea)translationalandb)
rocking modes of vibration. The bottom plots represent the soil nonlinearity
foreachmode.

263
Figure629 Inverted impedance function results from analysis of Exp. 3.4.7a (Unbraced
condition, Mighty Mouse shaker (5% ecc.) loading on the roof in the x
direction).Theplotsshowimpedancefunctionsinthea)translationalandb)
rocking modes of vibration. The bottom plots represent the soil nonlinearity
foreachmode.

264
Figure630 Inverted impedance function results from analysis of Exp. 3.4.9b (Unbraced
condition, Mighty Mouse shaker (10% ecc.) loading on the roof in the x
direction).Theplotsshowimpedancefunctionsinthea)translationalandb)
rocking modes of vibration. The bottom plots represent the soil nonlinearity
foreachmode.

265
Figure631 Inverted impedance function results from analysis of Exp. 3.4.13a (Unbraced
condition, Mighty Mouse shaker (50% ecc.) loading on the roof in the x
direction).Theplotsshowimpedancefunctionsinthea)translationalandb)
rocking modes of vibration. The bottom plots represent the soil nonlinearity
foreachmode.

266
Figure632 Inverted impedance function results from analysis of Exp. 3.2.7a (Braced
condition, Mighty Mouse shaker (5% ecc.) loading on the roof in the x
direction).Theplotsshowimpedancefunctionsinthea)translationalandb)
rocking modes of vibration. The bottom plots represent the soil nonlinearity
foreachmode.

267
Figure633 Inverted impedance function results from analysis of Exp. 3.2.6a (Braced
condition, Mighty Mouse shaker (10% ecc.) loading on the roof in the x
direction).Theplotsshowimpedancefunctionsinthea)translationalandb)
rocking modes of vibration. The bottom plots represent the soil nonlinearity
foreachmode.

268
Figure634 Inverted impedance function results from analysis of Exp. 3.2.4a (Braced
condition, Mighty Mouse shaker (25% ecc.) loading on the roof in the x
direction).Theplotsshowimpedancefunctionsinthea)translationalandb)
rocking modes of vibration. The bottom plots represent the soil nonlinearity
foreachmode.

269
Figure635 Inverted impedance function results from analysis of Exp. 3.2.1a (Braced
condition, Mighty Mouse shaker (10% ecc.) loading on the roof in the y
direction).Theplotsshowimpedancefunctionsinthea)translationalandb)
rocking modes of vibration. The bottom plots represent the soil nonlinearity
foreachmode.

270
Figure636 Inverted impedance function results from analysis of Exp. 3.2.2a (Braced
condition, Mighty Mouse shaker (25% ecc.) loading on the roof in the y
direction).Theplotsshowimpedancefunctionsinthea)translationalandb)
rocking modes of vibration. The bottom plots represent the soil nonlinearity
foreachmode.

271
6.6 Parametric System Identification of Forced Vibration Testing

6.6.1 System Identification Method

System identification refers to the process of computing the unknown properties of a

system given an input signal entering that system and the output signal that is modified

from the input as a result of system response. The system in this case is various

components of the SSI test structure, whereas the input and output signals are various

combinations of recordings from the forced vibration testing.

In this section, I utilize system identification to estimate the modal properties of

the SSI system using data generated from the forced vibration testing described in

Chapter 5. The modal properties of period and damping ratio are evaluated for two cases

of base fixity (fixed-base and flexible-base). Non-parametric and parametric methods can

be used to perform system identification. Non-parametric methods consist of evaluating

the complex-valued transmissibility function of the input-output motions of a system

without matching an underlying model (e.g., Ljung, 1987, Pandit, 1991). Parametric

system identification consists of evaluating the transfer function surface in the Laplace

domain by fitting an underlying model (Stewart and Fenves, 1998). A parametric analysis

is more quantitative in the sense that specific values of period and damping ratio are

recovered, and hence are generally preferred for studies of this type.

Parametric system identification analysis begins with an iterative process to

evaluate the time delay between input-output signals and the number of modes required

to best capture the system response. The iterative process uses least-squared techniques to

identify the model parameters that produce the smallest error between the model and

272
output signal. The total duration of the signal is used in this process (i.e., recursive

procedures are not considered). The model generated by the least squares regressions

describes a surface in the Laplace domain. That surface is referred to here as a transfer

function. As described by Stewart and Fenves (1998), the horizontal plane below the

surface has axes that can be interpreted in the form of frequency and damping ratio, with

the peaks (i..e., poles) of the surface defining modal properties. The intersection of the

surface with a vertical plane drawn through the frequency axis is the parametric estimate

of the so-called transmissibility function. That parametric estimate of the transmissibility

function can be compared to the non-parametric transmissibility functions evaluated

essentially as an output/input signal ratio, using procedures given in Mikami et al. (2008).

A key issue in the use of either parametric or non-parametric system identification

methods is what combinations of input-output signals should be used to recover specific

system properties of interest. Stewart and Fenves (1998) evaluated these motions for the

case of earthquake excitation and system properties corresponding to fixed-base

(representing structural flexibility and mass only), pseudo flexible-base (representing

structural mass and flexibility from the structure and foundation in rocking), and flexible-

base (representing structural mass and flexibility from the structure and foundation in

rocking and translation). For the case of forced vibration testing, the input-output pairs

for application to single degree-of-freedom structural systems have been evaluated by

Tileylioglu (2008) by solving equations of motion in the Laplace domain, with the results

in Table 6-2. Input-output pairs for more general application involving multi-degree of

freedom structures have yet to be identified.

273
Table 6-2 Input-Output Pairs To Evaluate Flexible-Base and Fixed-Base Modal
Parameters. (Tileylioglu, 2008)
Base Fixity Input Output

Fs
Flexible-Base u&& f + h&&f + u&&s
ms
Fs
Fixed-Base u&& f h&&f u&&s
ms
See Section 6.3 for definitions in table.

6.6.2 System Identification of Instrumented Field Test Structures

As described in Section 2.6.2, Stewart et al. (1999b) performed parametric system

identification of 58 buildings shaken by earthquakes to identify fixed-based, pseudo

flexibile-base, and flexible-base modal properties. The properties of greatest interest for

SSI applications are the first-mode period and damping ratio for the fixed-base ( T , i )

~
and flexible-base ( T , 0 ) conditions. Since the work of Stewart et al. (1999b), a number

of case studies of forced vibration have become available from which these same

properties can be identified. The available data comes from this study along with

Tileylioglu et al. (2011) for the GVDA Dr Evil structure and Star (2011) for the Mini Me

structure at the WLA site. The results of this work have been plotted and interpreted in

Section 2.6.2 for use in investigating period lengthening and foundation damping models.

This subsection describes the parametric system identification work that produced the

modal properties from the instrumented field test structures.

Table 6-3 shows the attributes of the instrumented field test structures (i.e., data

source, loading source, shaking direction, etc.), velocity reduction (Vs Vs0 ) due to

274
estimated shear strain (evaluated using procedures given in Section 6.5), parametric
~
system identified results, and relevant SSI parameters (h/(VsT), T T and f ). As shown

previously in Figure 2-14, the data occupies a previously sparsely populated region of

moderately high h/(VsT) = 0.14 to 0.45, and hence is a valuable addition to the database

of Stewart et al. (1999b).

The data contributed by Tileylioglu et al. (2011) for the Dr Evil structure come from

earthquake shaking and forced vibration testing (4 record sets). An additional 19 record

sets have been derived from forced vibration testing of Mini-Me at the GVDA (this

study, Chapter 5) and WLA (Star, 2011) sites. Data sets from both sites are available for

braced and unbraced structural conditions. The distinction in structural stiffness is

important in this case because the stiffer structure produces higher h/(VsT) and related SSI

effects. As shown in Table 6-3, similar forcing of Mini-Me (Exps. 3.1.1a and 3.3.5a)
~
showed larger results for the braced condition [h/(VsT)=0.22 and T T = 1.38 ] compared

~
to the unbraced condition [h/(VsT)=0.17 and T T = 1.31 ].

275
Table 6-3 System Identification of Instrumented Field Test Structures
SystemIdentification
Direction
Structure Structure Shake Vs0
DataSource
b
of d
h (m) B(m) L(m) Vs/Vs0
e
h/(VsT) se(%) ~ ~
&Site
a
Cond. Source
c
(m/s) Ti (sec) i ,i (%) Ti (sec) 0,i (%) Ti Ti f ,i (%)
Shaking

TEA(2011) Unbraced NA EQML=4.2 1.00 0.15 1.2 0.149 0.5 0.172 1.3 1.15 0.9
DE
@ TEA(2011) Unbraced NA EQMw=5.4 4.36 2.03 2.03 198.0 1.00 0.15 1.2 0.149 0.9 0.172 4.1 1.15 3.4
GVDA TEA(2011) Unbraced NA LM 1.00 0.14 1.2 0.152 1.5 0.166 1.7 1.09 0.4
TEA(2011) Braced NA LM 1.00 0.28 1.2 0.078 9.3 0.101 4.6 1.29 0.0
Exp.3.3.1a Unbraced X AA 1.00 0.17 1.2 0.089 0.9 0.101 3.1 1.13 2.4
Exp.3.4.9b Unbraced X MME10% 0.65 0.26 9.8 0.088 2.9 0.114 6.1 1.28 4.3
Exp.3.4.13a Unbraced X MME50% 0.50 0.29 13.7 0.101 2.1 0.127 7.7 1.25 6.4
Exp.3.4.15a Unbraced X MME100% Errorintest
Exp.3.3.5a Unbraced Y AA 1.00 0.17 1.2 0.087 1.7 0.114 3.4 1.31 2.4
Exp.3.4.3a Unbraced Y MME10% 0.73 0.23 7.0 0.088 2.3 0.141 9.1 1.61 8.2
Exp.3.4.17a Unbraced Y MME50% 0.57 0.27 12.1 0.096 0.4 0.169 8.8 1.76 8.7
MM
Exp.3.4.19a Unbraced Y MME100% 0.54 0.26 12.9 0.104 0.0 0.185 11.1 1.78 11.1
@ 2.87 1.07 2.13 193.5
Exp.3.1.1a Braced X AA 1.00 0.24 1.2 0.063 8.0 0.070 7.2 1.12 0.8
GVDA
Exp.3.2.6a Braced X MME10% 0.68 0.33 9.0 0.067 1.7 0.089 11.0 1.33 10.0
Exp.3.2.12a Braced X MME50% 0.49 0.45 13.9 0.067 7.2 0.109 11.8 1.63 9.1
Exp.3.2.14 Braced X MME100% 0.45 0.31 14.9 0.106 0.0 0.119 13.5 1.12 13.5
Exp.3.1.6a Braced Y AA 1.00 0.22 1.2 0.067 5.6 0.092 5.5 1.38 2.5
Exp.3.2.1a Braced Y MME10% 0.73 0.31 7.1 0.065 2.7 0.106 6.3 1.65 5.3
Exp.3.2.11 Braced Y MME50% 0.56 0.27 12.4 0.097 5.5 0.159 9.9 1.63 7.8
Exp.3.2.15 Braced Y MME100% 0.56 0.34 12.4 0.079 1.8 0.179 15.1 2.27 14.8
Exp.2.1.9a Unbraced X AA 1.00 0.33 2.0 0.082 1.4 0.118 9.5 1.44 8.8
MM
Exp.2.1.7a Unbraced Y AA 1.00 0.29 2.0 0.093 6.5 0.143 8.8 1.53 6.0
@ 2.87 1.07 2.13 106.0
Exp.2.1.17a Braced X AA 1.00 0.41 2.0 0.065 2.1 0.100 12.7 1.53 11.8
WLA
Exp.2.1.15a Braced Y AA 1.00 0.39 2.0 0.069 7.8 0.125 9.1 1.81 6.7
a
GVDAGarnerValleyDownholeArraySite;WLAWildlifeLiquefactionArraySite;DEDr.EvilStructure;MMMiniMeStructure(SeeChap.5)
b
TEA=Tileyliogluetal(2011);Datawith"Exp"nameswerecompiledaspartofaUCLAfieldtestingseries,informationcanbefoundathttps://nees.org/warehouse/project/637/andChapter5.
c
AA=AtomAntshaker;MME#%=MightyMouseshakerwith#%eccentricityforce;LMLinearmassshaker,similartoAA
d
Centerofmassheightoftheroofslab
e 1/2
BasedonMenq(2003)modelusing65%ofapproximatedstrainvalues(fondationvelocitypersoilshearwaveveolicty)atflexiblebasefrequency,Note:Vs /Vs0=(Gs /Gs0 ) andWLAsiteapproximateds =0.2

276
 The parametric studies using the measured responses of Mini-Me were performed

as part of this study. Nineteen data sets were evaluated using the input-output pairs in

Table 6-2 (hence, 19 values each of period lengthening and foundation damping result

from these computations). The iterative process described in Section 6.6.2 to identify

optimal values of time delay and number of modes was undertaken for each input-output

signal pair. For consistency, typically the same best fit delay and number of modes was

used for both the fixed-base and flexible-base cases. For each input-output pair, we check

results by comparing the transmissibility function from parametric system identification

with the corresponding non-parametric result. A small misfit between the fixed-base

cases occurred because of the accuracy that was applied to the delay and number of

modes. Greater precision could have been achieved by using different parameters for the

fixed-base and flexible-base cases at the cost of accuracy. I felt the relative accuracy
~
between analyses would better predict the SSI parameters (h/(VsT), T T and f ).

Figures 6-37 and 6-38 show example transmissibility functions from parametric

and non-parametric system identifications procedures using data from forced vibration

testing with the Atom Ant shaker at GVDA. Figure 6-37 shows the fixed-base and

flexible-base transmissibility functions for forced vibration testing of Mini-Me in an

unbraced condition for both a) shaking in the long (x) direction and b) shaking in the

short (y) direction. The peaks occur at modal frequencies, with the first distinct peak

representing first-mode response. Figure 6-38 shows similarly formatted results for Mini-

Me in a braced condition. As expected, the first-mode frequency is higher with bracing,

277
particularly for the fixed-base condition. The fixed-base peak is also broader, indicating

higher damping.

Figure637 Transmissibility functions from parametric and nonparametric system

identification using response history data from forced vibration testing at
GVDA with Atom Ant shaker. a) Exp. 3.3.1a (Unbraced condition, loading on
theroofinthexdirection)andb)Exp.3.3.5a(Unbracedcondition,loadingon
the roof in the ydirection). f1= firstmode frequency and 1= corresponding
firstmodedamping.

278
Figure638 Transmissibility functions from parametric and nonparametric system
identification using response history data from forced vibration testing at
GVDAwithAtomAntshaker.a)Exp.3.1.1a(Bracedcondition,loadingonthe
roof in the xdirection) and b) Exp. 3.1.6a (Braced condition, loading on the
roofintheydirection).f1=firstmodefrequencyand 1=correspondingfirst
modedamping.

The period lengthening and foundation damping computed from identified first-

mode parameters were compared to model predictions represented by closed-form

equations in Chapter 2. Input parameters for the foundation damping models, appropriate

soil shear wave include each of the variables in the wave parameter h/(VsT) along with

soil hysteretic damping (s), soil Poissons ratio , and mass ratio m. The soil shear wave

velocity (Vs) includes the effects of the weight of the structure and is taken as the time-

279
averaged value from the surface to the relevant zp depth as described in Section 2.6.1.

Peak shear strains for a given test were evaluated using procedures given in Section 6.5.

Since those are peak shear strain, they are reduced to an effective strain that is 65% of the

peak, and those effective strains were then used to evaluate soil hysteretic damping (s)

and shear modulus reduction (G/G0) values for each test using the generic models of

Menq (2003), as shown in Figures 6-24 through 6-36. No modulus reduction was applied

for the small linear mass shakers and the minimum s value was used (i.e., Dmin). The

shear modulus reduction can be equated to soil shear wave reduction as

Vs Vs 0 = G G 0 . The modulus reduction was used for both the determination of the

impedance function input G value used in the computation of period lengthening and

foundation damping and in the computation of the h/(VsT) value used for plotting

purposes. It should be noted the h values represent the height of the single-degree of

freedom mass height which is represented at the middle of the upper concrete deck.

Poissons ratio was chosen for analyses as =0.35 by Tileylioglu et al. (2011), =0.45 for

WLA and =0.33 for this study of GVDA. The average mass ratio (m=0.15) was used for

evaluation of the predicted period lengthening and foundation damping in Section 2.6.2.

6.7 Kinematic Interaction Analysis from Forced Vibration Testing at GVDA

During testing at GVDA, a preexisting non-embedded 12.2 m (40 ft) x 18.3 m (60 ft)

concrete slab was instrumented to evaluate kinematic soil-structure interaction effects

associated with forced vibration testing of the neighboring Mini Me structure. Recall

from Chapter 3, base slab averaging is a kinematic effect where motions on foundations

280
are reduced in amplitude relative to free-field motions. This section looks to gain insight

into the base-slab averaging phenomenon by investigating differences in the particle

motions between the ground motions and slab foundation motions at select frequencies of

excitation. Additionally, the vertical and horizontal transfer functions discussed in

Chapter 3 are examined.

Figure 6-39 shows the concrete slab location compared to the test structures

described in Chapter 5 and the sensor array. During selected forced vibration tests of

Mini-Me with the large eccentric mass shaker (described in Section 5.2.4) in the x and y

directions, a wireless data acquisition (DA) system was installed with four

accelerometers: two (sensors B and C) on the slab and two on the ground (sensor A

and D). A fifth sensor (E) was placed near the middle of the longitudinal (long)

direction of the slab on the ground surface and attached to the previously established

structural DA described in Section 5.2.5. Four experiments were performed while

accelerometers were situated on the concrete slab. The details of the experiments are

listed in Table 6-4.

Table 6-4 Test Experiments of Kinematic Interaciton Due to Forced Vibration Testing.
Loading Loading Eccentricityb Mini-Me
Test Experiment Direction Shakera (%) Stiffnessc
Exp. 3.5.1, shaking perpendicular to slab Y MM 100 Unbraced
Exp. 3.5.2, shaking perpendicular to slab Y MM 100 Unbraced
Exp. 3.5.3, shaking normal to slab X MM 100 Unbraced
Exp. 3.5.4, shaking normal to slab X MM 100 Unbraced
a MM=Mighty Mouse loading to Mini-Me test structure.
b Maximum force applied from the shaker.
c Stiffness is adjusted by bracings (i.e., either braced or unbraced condition).

281
 Forced vibration testing of a structure is not an ideal source for seismic waves,

since test structures are expected to generate surface waves, whereas seismic excitation

over the frequency range of typical interest in kinematic SSI problems is dominated more

by body waves. Nonetheless, the test was undertaken to see if useful results could

potentially be generated from a test configuration of this sort, which if successful is

potentially of substantial value. The wireless DA was provided by NEES@UCLA and the

details of the accelerometers and sampling rate are described in Section 5.2.3.

Figure 6-40a and Figure 6-40b show 3D depictions of the seismic wave particle

motions for Experiment 3.5.1 while the shaker was applying force frequencies of 5 Hz

and 10 Hz to the Mini-Me structure. The frequencies were determined from the

synchronized Mighty Mouse shaker and Structural Response DA systems in a similar

fashion as described in Section 5.3.6. All plotted scales in Figure 6-40 have been kept

consistent relative to each other, with the exception of accelerometers A and B in

Figure 6-40a that required a larger scale. The red lines in the figures represent the

projection of motion in any given plane. Figure 6-40 presents the following observations

of the particle motions:

Motions of A and B are dominated by vertical motions

The intensity of the vertical motion declines from north-south (i.e., A > E >

D)

The slab motions are altered from the ground surface motions. The motion

alteration is well observed in the comparison of motions of A and B in

Figure 6-38a, where the ground surface is dominated by an elliptical motion in

282
 the x-z plane (normal-vertical to shaking) and the slab motion resembles the

motion of the other slab accelerometer C, rather than the adjacent motion of

Due to the existence of the larger motions to the north, the wave passage is

compensated by the alteration of slab motion intensity. The northern motion

of the slab (B) is slightly reduced compared to the local ground motion

(A), while the southern motion of the slab (C) is larger than the local

ground motion (D)

Base slab averaging is quantified by transfer functions (Hu) as discussed in

Chapter 3. Figures 6-41 through 6-44 shows the observed vertical and horizontal Hu

along with predicted horizontal Hu using the NIST procedure in Equation 3-1. The

modeled used a Vs=266.9 m/s (875.7 ft/s) from SASW testing that was conducted

between the source of excitation (the Mini-Me location) and the concrete slab (Stokoe et

al., 2004). Figures 6-41 through 6-44 generally show a good match between the NIST

model predictions and observed data. The results are further discussed in Section 6.9.5

283
Figure639 Sketch showing the kinematic interaction test setup with an instrumented
concreteslabnearthelocationoftheforcedvibrationtestingofMiniMe.The
sensorIDsareshownalongwiththeirdistancefromthecenteroftheMiniMe
teststructure.

284
Figure640a 3Ddepictionofparticlemotionsateachaccelerometer(AE)at5Hzfrequency
excitation as shown in the bottom right corner (Mighty Mouse shaker
frequency).

285
Figure640b 3Ddepictionofparticlemotionateachaccelerometer(AE)at10Hzfrequency
excitation 10 Hz as shown in the bottom right corner (Mighty Mouse shaker
frequency).BasedonExp.3.5.1data.

286
Figure641 Experimentalverticalandhorizontaltransferfunctions(Hu)showninblueand
the predicted horizontal Hu shown in red. Seismic source was from the
harmonicforcedvibrationtestingofMiniMestructurewithshakingfromthe
MightyMouseshakerat100%ecc.intheydirection.BasedonExp.3.5.1data.

287
Figure642 Experimentalverticalandhorizontaltransferfunctions(Hu)showninblueand
the predicted horizontal Hu shown in red. Seismic source was from the
harmonicforcedvibrationtestingofMiniMestructurewithshakingfromthe
MightyMouseshakerat100%ecc.intheydirection.BasedonExp.3.5.2data.

288
Figure643 Experimentalverticalandhorizontaltransferfunctions(Hu)showninblueand
the predicted horizontal Hu shown in red. Seismic source was from the
harmonicforcedvibrationtestingofMiniMestructurewithshakingfromthe
MightyMouseshakerat100%ecc.intheydirection.BasedonExp.3.5.3data.

289
Figure644 Experimentalverticalandhorizontaltransferfunctions(Hu)showninblueand
the predicted horizontal Hu shown in red. Seismic source was from the
harmonicforcedvibrationtestingofMiniMestructurewithshakingfromthe
MightyMouseshakerat100%ecc.intheydirection.BasedonExp.3.5.4data.

290
6.8 Structure-Soil-Structure Interaction from Forced Vibration Testing

The forced vibration testing at GVDA described in Chapter 5 provides a unique

opportunity for investigation of the interaction of the forced vibrations from Mini-Me to

Dr. Evil through the soil medium. Luco and Contesse (1973) numerically investigated the

effects of SSSI from vertical incident SH waves. They found SSSI can be important at

low a0 values (< 2) and at the fixed-base first-mode frequency of the adjacent structure.

In this section, I evaluate SSSI occurring during forced vibration testing that is

principally emitting surface waves. The absence of the body waves allow for a direct

evaluate of the energy transmitted to Dr. Evil during testing, using transfer functions.

Transfer functions and coherence are calculated as described in Section 3.4.1 with Eqs.

(3-4) and (3-5). However, for these analyses the input is the foundation translation of

Mini-Me (uMM), while the output is the foundation translation of Dr. Evil (uDE).

To evaluate the varying interactions from different loading forces, transfer

functions were performed using the foundation response histories from the experiments

outlined in Section 5.4 (Table 5-7, Figure 5-14) that represent a spectrum of loading

conditions tested at GVDA. The sensor locations, foundation locations, and acceleration

records for the tests analyzed (referenced as small, medium and large amplitude force

tests) are shown in Section 5.4 (Figures 5-15e, 5-16e, 5-17e, respectively). All three tests

had the shaker installed on the roof with the shaking applied in the y-direction (towards

Dr. Evil). Both structures were in the unbraced condition during these tests.

Accelerometers situated closest together in the direction of shaking were used for

291
analysis (Dr. Evil, Ch. 30 and Mini-Me, Ch. 3). Figures 6-45 through 6-47 show the

transfer functions and coherence of the SSSI between the two structures at GVDA.

Figure 6-45 shows results from small amplitude force (Exp. 3.3.5a, Atom Ant

Shaker), which is not expected to produce nonlinearity in the response of the foundation

soils. The transfer function amplitude rises from 0.04 to about 0.1 from 6-25 Hz. A

distinctive peak is observed following the first-mode frequency (Section 6.6.2, f1 = 11.5

Hz) of Mini-Me.

Figure 6-46 shows results from medium amplitude force (Exp. 3.4.1a, Mighty

Mouse 5% ecc.), which is expected to produce nonlinearity in the response of the

foundation soils. Shaking for this test was limited to the frequency range of 4-18 Hz. The

transfer function shows a general amplitude trend at about 0.025 with a small peak to

0.08 at 5.5 Hz.

Figure 6-47 shows results from large amplitude force (Exp. 3.4.19a, Mighty

Mouse shaker full force) testing that was limited from 4-12 Hz to stay in the safe limits of

operation. As discussed in Section 5.4, the loading conditions caused gapping of the

foundation from rocking that was consistent with visual observations in the field. The

transfer function shows a decrease for low frequencies (3-6 Hz) that drops by an order of

magnitude (0.04 to 0.004) and then increases to 0.03 at 12 Hz. Figures 6-45 through 6-47

show that from small to large amplitude force testing the transfer functions decrease. The

nonlinearity produced in the response of the foundation soils tend to reduce the

interaction between the structures.

292
Figure645 Transfer function and coherence for Dr. Evil and MiniMe foundations. Exp.
3.3.5a(Unbracedstructure,shakinginydirectionwithAtomAnt).

Figure646 Transfer function and coherence for Dr. Evil and MiniMe foundations. Exp.
3.4.1a(Unbracedstructure,shakinginydirectionwithMightyMouse5%ecc).

293
Figure647 Transfer function and coherence for Dr. Evil and MiniMe foundations. Exp.
3.4.19a(Unbracedstructure,shakinginydirectionwith MightyMouse100%
ecc).

6.9 Interpretation of Soil-Structure Interaction Effects from Forced Vibration

Testing

6.9.1 Small Amplitude Forced Vibration Testing

Small amplitude harmonic forced vibration testing was performed to evaluate the

relatively linear range of the SSI system. A relatively uniform load across the range of 4-

50 Hz was applied by the linear mass shaker referred to as Atom Ant. As shown for

Experiment 3.3.5a in Figure 6-10, the response loop for translation show a frequency

independent demand on the soils.

The shaker location of applied force and structure stiffness (i.e. bracing condition)

introduced local complexities (i.e. near the bracing frequency). The Atom Ant shaker was

294
tied and secured in place during testing at the roof slab, whereas shaking applied at the

foundation slab only used the weight of the shaker itself. The lack of an external mount

(i.e., rope or bolts) during shaking at the foundation slab introduced noise into the data

obscuring the impedance results, as shown in Figures 6-20 to 6-23. Testing in the braced

condition resulted in additional noise at the resonance of the braces. As displayed in

Figures 6-18 and 6-19, obvious disturbances in our ability to read the impedance function

trends are seen at 15 Hz.

The experiments having the cleanest data sets at GVDA are those from the

structure in an unbraced condition with forcing at the roof. The testing was absent of the

bracing resonance effect and the shaker was secured in place so that no adverse

(artificial) displacements would occur during the test experiment. Therefore the

remainder of this section will discuss interpretations of the small amplitude forced

vibration testing shown in Figures 6-16 and 6-17.

The horizontal translational stiffnesses were observed as frequency independent

and stable from about 4-20 Hz. The observed translational stiffnesses are softer than

predicted based on direct crosshole testing of the soil after construction (Model 1). The

translational stiffnesses were closest to the predictions based on Model 2 velocities back-

calculated from field data, which excludes overburden pressures from the structure.

Shaking in the y-direction resulted in lower translational stiffnesses as compared to

shaking in the x-direction. This contradicts the models that predict the opposite trend (y-

direction stiffness x-direction stiffness). A 15% reduction of the overburden in-situ

shear wave velocity (Vs) would be required to match the experimental translational

295
stiffness in the y-direction, while the x-direction is matched well by Model 2 (which is a

10% reduction). One speculation is that the reduction from the expected stiffnesses

(Model 1) arises from localized shearing of the soil underlying the foundation. The

dynamic interaction of the concrete-soil contact could cause disturbances to the shear

stiffness of a thin layer of soil in the immediate vicinity of the foundation that are not

captured by the crosshole seismic testing at 1m (3.3 ft) depth.

The rotational stiffnesses and damping ratios showed a frequency-dependency (4-

20 Hz), where the rotational stiffnesses decreasesat higher frequencies. When shaking the

structure in the x-direction, the rotational stiffness matched the expected results of Model

1, whereas shaking y-direction produced a rotational stiffness that was softer than

expected and matched Model 2. The translational damping ratios were also relatively

well captured by Model 2. The predictive models underestimated the rotational damping

ratios compared to the observed results. The larger than predicted rotational damping is

suspected to arise from one or a combination of the following phenomena (listed in no

particular order):

1. Hyperelasticity in the underlying soils produced from the rocking mode of

vibration.

2. Small compliances (flexibility) of the foundation that introduce out-of-

phase radiation waves from the sides of the foundation. Such waves would

be destructive resulting in more than expected radiation damping.

3. Foundation-soil contact friction.

296
 4. A thin, soft unidentified layer under the foundation where the Vs is smaller

than the surrounding material.

5. Model-based radiation damping is limited because it applies for

homogeneous soil conditions in which upward and downward excitation

from a rocking slab effectively cancels out at large distances (St. Venants

principal, as applied to a dynamic problem). In contrast, the actual site

lacks homogeneity, so the perfect cancelation of the radiated waves does

not occur, increasing radiation damping relative to model predictions.

6.9.2 Large Amplitude Forced Vibration Testing

Large force harmonic excitation is expected to produce some nonlinearity in the response

of the foundation soils. The cyclic load-deformation loops were established for the

translation-shear and rocking-moment (as previously shown in Figures 6-10, 6-12, and 6-

14). Figures 6-48 and 6-49 show lines representing secant stiffnesses for particular loops

(lines and loops for a common frequency share a like color). Figure 6-48 and Figure 6-49

show the impedance and cyclic loops from testing in the y-direction from the moderate

force Experiment 3.4.1a (max force of 7.1 kN) and large force Experiment 3.4.19a (max

force of 61.7 kN), respectively. The decreasing trend (3-8 Hz) and then increasing trend

of the stiffnesses (8-15 Hz) of the foundation-soil system occurs for frequencies below

and above, respectively, the systems resonant frequency. Those trends are illustrated by

both the impedance functions and loop slopes. As a check of the impedance ordinates

evaluated from frequency-domain procedures (Section 6.5), we plot in Figures 6-48 and

297
Figure 6-49 discrete dots representing the slopes of the secant lines at selected

frequencies (the agreement is excellent).

Forced vibrations with the Mighty Mouse shaker were performed while the

structure was in braced and unbraced conditions. A series of tests were performed at

varying degrees of intensity conditioned in the field by the percent eccentricity of the

shaker. The shaker force was proportional to the frequency and mass eccentricity. This

resulted in the maximum forces near the maximum frequency of the test, with the mass

situated at 100% eccentricity being the maximum force tested at GVDA. The

approximated nonlinear nature of the soil below the foundation shows the same trend.

The shaker force introduces a frequency-dependency into these tests.

The translational and rotational stiffnesses generally follow the trend of the

approximated nonlinearity of the soils underlying the foundation. Low eccentricity (5-10

%) produce similar results as the low amplitude force testing at low frequencies (2-5 Hz).

As the frequency increases, the foundation soil becomes increasingly nonlinear as

indicated by lower foundation stiffnesses. As the eccentricity is increased (25-100%), the

stiffness impedance functions become much lower than the low-force testing results even

at low frequencies (2-5 Hz).

The structural condition (i.e. braced or unbraced) affects the overall trend of the

system stiffness because of its affect on the systems resonant frequency. As frequency

increases for the unbraced case (e.g. Figure 6-30), the demand increases causing a

reduction in the soil stiffness until the resonant frequency is reached. Then the soil

stiffness rebounds and strengthens, whereas the braced condition (e.g. Figure 6-34)

298
produced a gradual reduction over the frequencies tested (4-14 Hz). The resonance of the

bracing is not as pronounced in the impedance functions from large shaker as previously

seen in the low-force amplitude testing.

The existence of negative damping ratios of the large-force test results (e.g.

Figure 6-24) indicate that the results have been compromised. It is suspected that phasing

issues derived from either the Mighty Mouse data acquisition (DA) system or the

synchronization of the two DA systems resulted in irrational results. Many of the

damping ratios show negative results for low frequencies (2-6 Hz). The Mighty Mouse

shaker force was derived by converting a recorded pulse at the time when the weights

were aligned in the direction of loading. Sensitivity studies were performed to evaluate if

a phase shift of the shaker pulse would produce more realistic results. It was found that

improvement could be made at low frequencies, but at the cost of the accuracy of the

higher frequencies. Therefore, it is suspected that some of the damping ratio results are

realistic, but there are no means to classify the difference. Another possibility for the

negative damping ratios at low frequencies could be the small phase difference at low

frequencies and our ability in resolving through measurements of the very small numbers.

If the latter is true, then at the larger frequencies where phase differences increase, the

results would be considered more reliable.

299
Figure648 Hystereticresponsesecantstiffnesses(depictedbythedottedlines)andtheir
relationship to the loaddeformation loops for the system. Exp. 3.4.1a
(Unbraced structure, shaking in the ydirection with Mighty Mouse at 5%
eccentricity).

300
Figure649 Hystereticresponsesecantstiffnesses(depictedbythedottedlines)andtheir
relationship to the loaddeformation loops for the system. Exp. 3.4.19a
(Unbraced structure, shaking in the ydirection with Mighty Mouse at 100%
eccentricity).

6.9.3 Small Amplitude to Large Amplitude Comparisons

Figure 6-50 shows a comparison of the normalized experimental results for the small

(AA shaker), moderate (MM shaker @ 5% ecc.) and large (MM shaker @ maximum

capacity) representative force cases, as shown in Table 5-8. The stiffnesses are

normalized by G0B and G0B3 for translation and rocking, respectively, where G0 is the

301
small-strain maximum shear modulus of the soil. The model in Figure 6-50 results from

the normalization of Model 1 and Model 2. The frequency axis in Hz has been replaced

with dimensionless frequency a0 (Eq. 2-7b).

The small force produced results over the broadest range of frequencies (a0 = 0.2

1). As force amplitude increased, the frequency ranges narrow: a0 = 0.1 0.6 for

moderate forces and a0 = 0.05 0.4 for large forces. The large range produced by the

small force is a result of the small shakers force being relatively consistent and not

proportional to frequency. Therefore, the Atom Ant shaker could be applied to higher

frequencies (up to 45-50 Hz). In contrast, the Mighty Mouse shaker force is proportional

to frequency and has limiting conditions based on safety of the equipment. As loading

increases, observed foundation stiffnesses decrease, indicating an increase in strain

applied to the soils underlying the foundation. Based on the phasing issues discussed in

Section 6.9.2, the damping results at low frequencies appear to be in error. However, it is

suspected from sensitivity studies that the moderate force results are reliable for a0 = 0.3-

0.6. As forces increase to higher frequencies, the translational damping ratio matched the

Models well, while the rotational damping ratio was larger than expected. The sensitivity

studies involved incremental changes to the recorded angular position of the large shaker

rotating masses described in Eq. (5-2). As the angular position was changed by a few

degrees, resulting in an updated forcing function, the observed impedance functions

improved compared to the models at lower frequencies. However this was at the cost of

the observed impedance functions at higher frequencies.

302
Figure650 Normalized impedance function results from analysis of the MiniMe test
structure(unbracedcondition)loadedontheroofintheydirectionbyvarious
loading amplitudes. The plots show impedance functions in the a)
translational and b) rocking modes of vibration. The bottom plots represent
thesoilnonlinearityforeachmode.

303
6.9.4 GVDA Results and Previous Studies

Figure 6-51 shows a comparison of the normalized stiffness and damping ratios from

WLA and GVDA for shaking with Atom Ant at the roof in the y-direction. The

stiffnesses are normalized by the site-specific G0 and B (translation) or B3 (rocking).

Additionally all impedance functions are plotted against a0 rather than frequency in Hz.

Two models using Table 2.1 impedance functions have been plotted. To represent the

different sites the models use a =0.45 and =0.33 to represent the WLA and GVDA

sites, respectively. The models show a Poissons ratio effect in which stiffness increases

and damping decreases as Poissons ratio increases. With the exception of the

translational damping ratio, this Poissons effect is also evident in the observed data. It

should be noted that a prominent difference between the two testing schemes was that no

external support for the shaker was used at WLA. As a result, the GVDA testing

produced more stable results to higher frequencies ( a0 = 1) compared to WLA. The

deviation from the expected trend of the WLA results at a0 > 0.5 is likely due to slip of

the shaker on the top slab of Mini Me.

Previous case studies performed for evaluation of impedance functions from

forced vibration testing have been discussed in Section 6.2. The results of the previous

studies have been visually discritized and fit with spline smoothing. The fitted curves

were then normalized by their site-specific G0 and foundation half-width (B) raised to an

appropriate power (one for translation, three for rocking). Figure 6-52 shows normalized

results of the impedance functions for the previous studies described in Section 6.2 with

304
the omission of GVDA and WLA. As represented in Table 6-1, previous studies typically

had foundations with L/B = 1, whereas GVDA and WLA structure has L/B = 2. Therefore

for comparative reasons Figure 6-51 shows only the recent studies of Mini-Me with L/B

= 2, whereas Figure 6-52 shows previous studies with foundations of L/B = 1. This

presentation allows for a direct comparison of the stiffness and damping effects

investigated from forced vibration testing. For comparison a predicted model using Table

2-1 impedance functions with L/B = 1 is included in Figure 6-52. It should be noted that

the models use a consistent =0.33 even though some sites are situated on soils with

larger values of (clay sites). That would produce similar results as discussed in relation

to Figure 6-51

305
Figure651 Normalized impedance function results from analysis of the MiniMe test
structure(unbracedcondition)loadedontheroofintheydirectionwithAtom
AntshakeratGVDA(Exp.3.3.5a)andWLA(Exp.2.1.7a).Theplotsshowthea)
horizontal translational stiffness, b) rotational stiffness, c) horizontal
translational damping ratio and d) rotational damping ratios. Included are
modeledimpedancefunctionsbasedonTable2.1for=0.33and=0.33.

306
Figure652 Normalized impedance function results from previous forced vibration
studies. The plots show the a) horizontal translational stiffness, b) rotational
stiffness, c) horizontal translational damping ratio and d) rotational damping
ratios. Included are modeled impedance functions based on Table 2.1 for
L/B=1using=0.33andD=0.

307
6.9.5 Kinematic Interaction and Structure-Soil-Structure Interaction Interpretation

Kinematic interaction (Section 6.7) and structure-soil-structure interaction (SSSI)

(Section 6.8) effects were investigated from the data accumulated from the forced

vibration testing at GVDA. The forced vibrations of the structure produced principally

surface waves, , whereas seismic excitation over the frequency range of typical interest in

kinematic SSI problems is dominated more by body waves. Surface waves could be

expected for SSSI problems, however, the SSSI problem would typically be introduced

from body waves and then surface waves could be generated from both structures.

The particle motions discussed in Section 6.7 showed the phenomenon of base-

slab averaging. The averaging was only slight in that the motions were not the same at

both sides of the slab. General trends of the motions on the slab were similar whereas the

adjacent ground surface motion was distinctly different. The vertical transfer functions

were observed to be about one, while the horizontal transfer functions generally

decreased with frequency. The NIST horizontal transfer function model discussed in

Chapter 3, matched the experimental data well, with the exception of some peaks that

were greater than the model.

Three transfer functions are evaluated to investigate the relative motions of the

Mini-Me and Dr. Evil foundations from small, medium and large amplitude forced

vibrations. Only the small amplitude testing (Figure 6-45) showed a distinctive response

of Dr. Evils foundation near the fixed-base first-mode frequency of Mini-Me. However

as the loading and induced nonlinearity of the foundation soils increased, the interaction

of the systems decreased. Particularly this can be seen by comparing the approximated

308
strain and shear modulus reductions for the same test (Figure 6-28). The distinctive

reduction to the transfer function up to 6 Hz (Figure 6-47) correlates well to the increased

strain to the foundation soils. However the transfer function shows a rebound whereas the

strain rate is approximated as unchanged (no decrease) for the higher frequencies.

6.10 Conclusions

The forced vibration testing at GVDA produced data that was evaluated for SSI effects

(i.e., inertial interaction and kinematic interaction). The most useful data for comparison

of experimentally derived impedance functions to the predicted models was from testing

performed when Mini-Me was in an unbraced condition and was excited by the small

linear mass shaker (Atom Ant) at the roof. During these tests the shaker was secured in-

place from movement at higher frequencies. The large eccentric mass shaker (Mighty

Mouse) produced forces large enough to cause significant strains in the soils underlying

the foundation to behave nonlinear. Monitoring of the concrete slab and adjacent soil

during the forced vibration testing provided data for evaluation of kinematic interaction.

The modeled predictions of the impedance functions tended to overpredict the

stiffnesses, while the rotational damping ratio is well underpredicted. It was found that a

Vs reduction of about 10-15% from the in-situ Vs would be required to match the models

to the observed data. The experimental translational stiffness was found to be stiffer in

the x-direction compared to the y-direction, which is the opposite of the modeled results.

The particle motions generated adjacent and on the concrete slab showed an

averaging effect. The waves generated at either side of the slab were not identical in

motion or amplitude, but did exhibit different motions then the adjacent ground surface

309
accelerometer. The motions on the slab showed a tendency towards similar motions.

Interestingly, kinematic interaction effects observed on an adjacent slab excited

principally by surface waves were of a similar character to expectations from analytical

models for body wave excitation from earthquakes. If verified, these results could lead to

site- and foundation-specific test methods for evaluating kinematic interaction effects.

Structure-soil-structure interaction (SSSI) was briefly investigated from forced

vibration testing. Transfer functions showed at small amplitude forces Dr. Evil had a

small response at the fixed-base frequency of Mini-Me. The transfer functions showed

that nonlinearity significantly reduces the interaction between the foundations.

310
7 CONCLUSIONSAND
RECOMMENDATIONS

7.1 Scope of Research

The objective of this research was to evaluate the effects of soil-structure interaction

(SSI) through careful interpretation of available data from instrumented buildings and

recently performed forced vibration experiments on instrumented buildings and test

structures. Analyses of field performance data was undertaken to provide deeper insights

into SSI phenomena ranging from kinematic effects on foundation ground motions to

mobilized foundation stiffness and damping across a wide range of frequencies and

loading levels. These data were interpreted to evaluate strengths and limitations of

engineering analysis procedures for SSI.

Following the introduction of the SSI problem, this research began by studying

the contribution of foundation damping to the overall damping of a soil-foundation-

structure system (Chapter 2). An initial step commonly taken to evaluate the effect of SSI

on the seismic response of a particular structure is a representation through its impact on

the fundamental-mode period and damping ratio (i.e., response spectra). As part of this

research, two equally viable equations for foundation damping were derived and

311
compared to previous solutions and empirical data. Current seismic guidelines use

foundation damping solutions based on rigid circular foundation types on homogenous

half-space. Considering most foundations are neither circular nor situated on uniform

soils, procedures were presented to consider various site-specific conditions such as

rectangular footprint, embedment and non-uniform soil conditions. The two foundation

damping solutions derived as part of this work along with the aforementioned procedures

were then compared to empirical data.

Simple kinematic interaction models were compared to newly available Japanese

data and more robust finite element modeling results (i.e. SASSI) in Chapter 3. Limited

case studies are compared to modeled transfer functions; however, this work expands the

data inventory to buildings not located in the California where sites were used to develop

the simplified models. A subset of the Japanese buildings includes relatively stout pile

foundations that were not part of the earlier California data set. This work is part of a

long-term goal to re-generate the kinematic interaction models. Observations as to the

comparison of the models and Japanese data transfer functions were made.

Various approaches can be used to analyze seismic response to SSI at the base of

a building, including ignoring SSI effects altogether. Chapter 4 presents a study that

identifies suitable instrumented buildings that have recorded earthquakes, develops

complete SSI models for substructure-based analysis of seismic response (referred to as a

baseline model), roughly calibrates the structural elements of the baseline model to

approximately match recordings, and then removes components of the baseline model for

additional analyses to evaluate the impact of those components. This process was used to

312
identify the critical components of an SSI model for use in response history analysis, and

develop recommendations based on those results as well as prior results in the literature.

Forced vibration testing of a well-instrumented test structure was performed in

Garner Valley, CA to expand the data inventory for analysis of SSI effects. Chapter 5

presents the details of the field testing and associated data files created as part of the

work. In addition to monitoring of the test structure, instrumentation included monitoring

of a local concrete slab and adjacent structure during the forced vibration testing. The

additional monitoring was done to establish field data for evaluation of kinematic

interaction and structure-soil-structure (SSSI).

The output/input forced vibration testing data was used for evaluation of inertial

and kinematic interaction effects (Chapter 6). Horizontal impedance functions of the

system were evaluated from the recorded data and then compared to the analytical

solutions in Chapter 2. Additionally the output/input signals were analyzed using

parametric system identification procedures to establish fixed-base and flexible-base first

mode periods and damping ratios. The particle motions generated from the field test

structure during forced vibration testing were evaluated on and adjacent a local concrete

slab to investigate kinematic interaction. The particle motions were reviewed to gain

insight into the slab-ground interaction. Then transfer functions were computed and

compared to the simplified kinematic interaction model in Chapter 3. SSSI analysis was

performed by evaluating transfer functions of the foundation motions of Mini-Me and Dr.

Evil from three intensity-levels of forced vibration loading conditions.

313
7.2 Research Findings and Recommendations

7.2.1 Chapter 2; Foundation Damping

1. Foundation damping incorporates the combined effects of energy loss due to

waves propagating away from the vibrating foundation in translational and

rotational modes (radiation damping), as well as hysteretic action in the soil

(material damping). Foundation damping is sensitive to the height aspect ratios

(h/r & h/B) and foundation aspect ratio (L/B). Much more foundation damping is

produced when translation controls (i.e. h B 2 ) the SSI system as compared to

rocking (i.e. h r & h B > 2 ). As the foundation aspect ratio increases (L B > 1) ,

shaking in the long direction (x-direction) produces higher foundation damping

than does shaking in the short (y) direction. Rocking becomes less pronounced in

the long direction, in turn causing foundation translation to dominate, which

produces more foundation damping.

2. Foundation damping expressions (Eqs. 2-25 and 2-35) were derived by starting

from fundamental principles of equating the properties of an equivalent fixed-base

oscillator to that of a flexible-base oscillator. The derivations were performed by

both matching the imaginary component and considering the full complex valued

stiffness. These derivations lead to two equally viable solutions for foundation

damping.

3. Procedures were presented to consider various site-specific conditions such as

rectangular footprint, embedment, soil nohomogeneity and nonlinearity

314
 conditions. As a result of these recommendations, foundation damping can be

solved using closed-form solutions (Eqs. 2-25 and 2-35);

4. Empirical data was used to guide in the selection of a single solution for

foundation damping. It was found that the levels of damping from Eq. (2-25) are

higher than those from Eq. (2-35) for foundations of modest height aspect ratios.

The lower damping from Eq. (2-35) is generally more consistent with field case

histories. Eq. (2-35) appears to be the preferred formulation at the present time.

7.2.2 Chapter 3; Kinematic Soil-Structure Interaction Effects from Building and

Free-Field Seismic Arrays in Japan

1. A list of well-instrumented Japanese sites has been compiled and three were

evaluated for transfer functions between the foundation and free-field motions.

This work adds to the inventory of case studies looking specifically at simplified

kinematic interaction models. This work is limited in that it only reviewed three

sites for comparison of the horizontal transfer functions and one site for rocking

transfer function. However, this work does expand the data invitory to include

buildings with stout piles that were not considered in the data set used to develop

the models;

2. NIST-type kinematic models for horizontal transfer functions matched the finite

element modeling code SASSI results well for frequencies of principal interest

(up to about 10 Hz). The NIST-type model predicted kinematic base rocking well,

as opposed to the SASSI model that was too low compared to the data;

315
 3. Model-based horizontal transfer functions were too large (hence conservative) for

the Sendai and Onagawa sites. The model predictions are fairly accurate relative

to data at the Nagoya site, which is dominated by embedment effects;

4. For the Japanese buildings situated on piles at the ground surface, the NIST-type

model tends to underpredict the ground motion reduction from kinematic

interaction for frequencies larger than the structures resonant frequency. This is a

component of kinematic interaction that does not have an analytical model;

5. The NIST-type model worked well for the limited cases examined, even though

the base-slab averaging model is semi-empirical and developed based on western

United State sites. The models either matched the ground motion reduction or

underpredicted (hence conservative).

7.2.3 Chapter 4; Assessment of Alternative SSI Modeling Strategies

1. Baseline models (MB) type models can match observed responses with modest

adjustments of structural mass and damping;

2. An equivalent-linear spring and dashpot approach for SSI modeling can provide

satisfactory predictions of soil-foundation-building performance;

3. Four simplified models were considered for response history analysis. Model 1

includes the above ground portion of the structure as fixed at the ground surface

(excluding subterranean levels and foundation components), excited by the free-

field motion (ug). Model 2 includes the above ground and subterranean portions of

the structure, but ignores the embedment soils (no soil springs) and is fixed at the

base, excited by ug. Model 3 includes subterranean levels with horizontal and

316
 vertical soil springs, with the horizontal springs at the base are excited by ug and

the springs at higher elevations are fixed and the end with no input motion. Model

4 includes subterranean levels supported by horizontal and vertical soil springs

that are fixed at their ends to a rigid bathtub, excited by the foundation input

motion (uFIM) or ug.

4. Engineering demand parameters (EDPs) such as inter-story drifts, story shear

distributions, and floor accelerations were evaluated for each modeling scheme

and compared to the baseline model (MB).

5. Model 2 and Model 4 (bathtub) had the best performance for the above-ground

portion compared to the MB model;

6. Model 3 performed the worst compared to the above-ground portion of the MB

Model and should not be used;

7. Model 1 is the next least-desirable compared to Model 3, with the level of error

increasing with stiffer structures or deeply embedded structures;

8. None of the simplified modeling schemes (Models 1-4) were able to match the

MB response below the ground line.

7.2.4 Chapter 5; Forced Vibration Testing of a Test Structure

1. The response histories have been processed from the unprocessed data files to

convert to engineering units, convenient time formats, and to remove baseline

errors. Visual inspection revealed several problems: a disconnected LVDT (Ch.

46), baseline errors of accelerations from the foundation of adjacent structure Dr.

Evil, and a conversion error of three vertical foundation LVDTs (Chs. 44, 46, 47).

317
 The artificial LVDT conversion error can be corrected by multiplying the

response by a factor of 10. The data files created from the forced vibration testing

at GVDA have been archived at the NEEShub

(https://nees.org/warehouse/project/637);

2. The observed responses indicate no significant torsional response for the lowest

shaking level, but asymmetric responses during testing with the large shaker

suggest a torsional response;

3. Gapping was observed in the LVDT displacements applied by the strongest

shaking, which is consistent with visual observations from the field.

7.2.5 Chapter 6; Evaluation of Soil-Structure Interaction Effects from Forced

Vibration Testing

1. Impedance functions have been analyzed from the output/input responses of the

Mini-Me test structure. Comparison of the results from the different loading

conditions, structural condition (i.e., braced or unbraced), and shaker location

revealed varying degrees of noise in the data. Local complexities were produced

when testing in the braced condition due to the resonance of the braces. The large

shaker was affixed to the Mini-Me structure for all tests, whereas the small shaker

was externally mounted for testing on the roof, but not for testing on the

foundation. As a result noise was introduced into the recorded shaker response as

the shaker produced artificial translations not associated with the loading of the

structure. Testing in an unbraced condition with the small shaker externally

mounted on the roof produced the best reading of the impedance ordinates.

318
2. The data was analyzed to evaluate the stiffness and damping associated with the

foundation-soil interaction, revealing linear-elastic behavior at low forcing levels

characterized by smaller stiffnesses and both lower and higher damping than was

predicted by classical models. Modeling to account for the overburden of the

structure tended to overpredict the stiffness values. Shaking in the long direction

(y-direction) resulted in lower stiffnesses as compared to shaking in the short (x)

direction, which is opposite of the modeled trends. Whereas the translational

damping ratio was either matched well or slightly overpredicted by the models,

the rotational damping ratio was much larger than predicted;

3. The foundation soil strain () was approximated for the large amplitude force

vibration testing by comparing the foundation peak velocity (PGV) and soil shear

wave velocity ( = PGV Vs ) . Shear modulus reduction (G/G0) and damping (s)

was then evaluated using Menqs model (2003) at the corresponding strain. The

strains showed a frequency-dependency. A rapid increase in demand occurs at

low frequencies (0-7 Hz). At larger frequencies (7-14 Hz) the demand would

either steady at large strains for eccentricity tests ( 50%) or slightly relax. This

frequency-dependency of the strains was able to capture the observed behavior of

the foundation-soil system well. Similar trends were observed in the inverted

stiffness results.

4. At small frequencies (where small soil strains are expected) from testing with the

large shaker, the experimental stiffness values are smaller than the results from

the testing with the small shaker.

319
 5. Negative damping ratios of the large-force test results indicate that the results

have been compromised. It is suspected that the small phase difference at low

frequencies and our ability in resolving through measurements of the very small

numbers resulted in the negative damping ratios.

6. Kinematic interaction effects were investigated from the response data processed

for the local concrete slab and adjacent ground surface. Interestingly, the

kinematic interaction effects observed on the adjacent slab excited principally by

surface waves were of a similar character to expectations from analytical models

for body wave excitation from earthquakes. If verified, these results could lead to

site- and foundation-specific test methods for evaluating kinematic interaction

effects.

7. The Stewart et al. (1999b) data set from instrumented structures has been

expanded to include results from forced vibration tests. The new data occupies a

previously sparsely populated region of moderately high h/(VsT) = 0.14 to 0.45,

and hence is a valuable addition to the previous database.

8. Structure-soil-structure interaction (SSSI) was evaluated through use of transfer

functions of the foundation motions of Dr. Evil and Mini-Me. The transfer

functions showed that nonlinearity significantly reduces the interaction between

the foundations.

320
7.3 Recommendations for Future Research

The following are recommendations for future research:

1. Forced vibration testing showed a reduction of the stiffnesses as loading was

increased from test to test. Forced vibration testing that is expected to cause large

stress should include strain gauges in the soil to evaluate the soils strain

underneath the foundation.

2. Improve upon the current instrumentation programs, such as CSMIP and USGS,

to include SSI-type sensor plans. The objective is to identify buildings that are

moderately to highly susceptible to period lengthening, foundation damping and

kinematic interaction. An initial evaluation of a building for the inertial

interaction level of interest can be made based on h / Vs T > 0.1 . Buildings in areas

prone to high frequency content seismic events (such as east coast buildings)

should be identified for kinematic interaction arrays. SSI-type sensor plans

designed to capture SSI effects on buildings requires at least the horizontal

accelerometers at the free-field, foundation, and roof and two vertical

accelerometers on the foundation to evaluate rocking response (Stewart and

Fenves, 1998).

3. Kinematic interaction models (as considered in Chapter 3) currently do not

capture the reduction of ground motion at low frequencies for buildings situated

on piles at the ground surface. A verified model to capture the effect of piles

should be developed.

321
4. Kinematic interaction models should be evaluated with computer based software,

such as SASSI, to validate or update the input parameters. Currently the models

are based on a and a0. The a term was semi-empirically derived as a function of

Vs and a0r = r Vs . The implication is that both and r have an equal effect on

kinematic interaction. It should be determined if the frequency or geometry has a

greater effect to kinematics and possibly decoupled as an input parameter.

322
8 REFERENCES

Abrahamson, NA, 1991. Spatial variation of strong ground motion for use in soil-structure
interaction analysis, Proceedings of Fourth U.S. National Conference on Earthquake
Engineering, Vol.1, 317-326.
American Society of Civil Engineers, ASCE, 2010. Minimum Design Loads for Buildings and
Other Structures, Report ASCE/SEI 7-10, Reston, VA.
American Society of Civil Engineers, ASCE, 2006. Seismic rehabilitation of existing buildings,
Report ASCE/SEI 41-06, Reston, VA.
Ancheta, TD, JP Stewart, and NA Abramhamson, 2011. Engineering characterization of
earthquake ground motion coherency and amplitude variability, Proc. 4th International
Symposium on Effects of Surface Geology on Seismic Motion, IASPEI / IAEE, August
2326, 2011, University of California, Santa Barbara.
Applied Technology Council, ATC, 1996, Seismic evaluation and retrofit of concrete buildings,
Report ATC-40, Redwood City, CA.
Apsel, R.J. and Luco, J.E., 1987. Impedance functions for foundations embedded in a layered
medium: an integral equation approach, Earthquake Engineering and Structural
Dynamics, Vol. 15, No. 2, pp. 213-231.
Aviles, J. and Perez-Rocha, L.E., 1996. Evaluation of interaction effects on system period and the
system damping due to foundation embedment and layer depth, Soil Dyn. and
Eqk.Engrg., 15, 11-27.
ASCE, 2013, Seismic Evaluation and Retrofit of Existing Buildings, ASCE/SEI 41-13, American
Society of Civil Engineers, Reston, Virginia.
ASCE, 2010, Minimum Design Loads for Buildings and Other Structures, ASCE/SEI 7-10,
American Society of Civil Engineers, Reston, Virginia.
ASCE, 2006, Minimum Design Loads for Buildings and Other Structures, ASCE/SEI 7-05,
American Society of Civil Engineers, Reston, Virginia.

323
Bielak, J., 1971. Earthquake response of building-foundation systems, Ph.D. Dissertation,
California Institute of Technology, Pasadena.
Bielak, J., 1975. Dynamic behavior of structures with embedded foundations, J. Earthquake
Engng. Struct. Dyn, 3, 259-274.
Bielak J 1976. Modal analysis for building-soil interaction, J. Engrg. Mech, 102, 771-786.
Boore, D.M., and Boomer, J.J., 2005. Processing of strong-motion accelerograms: needs, options
and consequences, Soil Dyn. and Eqk.Engrg., 25, 93-115.
California Strong Motion Instrumentation Program (CSMIP), 2011. Accessed through website at
http://www.strongmotioncenter.org
Chubu Regional Bureau office, MLIT, 2003. PS logging at Nagoya Courthouse building
complex, unpublished report (in Japanese).
Chubu Regional Bureau office, MLIT, personal communication 2005. Portions of construction
drawings and the soil investigation (in Japanese).
Chopra, A.K. 2011. Dynamics of structures, theory and application sto earthquake engineering,
fourth edition, Prentice Hall, Upper Saddle River, NJ.
Chopra, A.K., and Goel, R.K., 1999. Capacity-demand-diagram methods based on inelastic
design spectrum, Earthquake Spectra, Vol. 15, No. 4, pp. 637-656.
Crouse, C.B., Hushmand, B., Luco, J.E., and Wong, H.L., 1990, Foundation impedance
functions: Theory versus experiment, Journal of Geotechnical Engineering, Vol. 116,
No. 3, pp. 432-449.
Darendeli, 2001. Development of a new family of normalized modulus reduction and material
damping curves, Ph.D. Dissertation, Univ. of Texas at Austin.
De Barros, F.C.P. and Luco, J.E., 1995, Identification of foundation impedance functions and
soil properties from vibration tests of the Hualien containment model, Soil Dynamics
and Earthquake Engineering, Vol. 14, pp. 229-248.
Fadum, R.E., 1948, Influence values for estimating stresses in elastic foundations, Proceedings,
2nd International Conference on Soil Mechanics and Foundation Engineering, Rotterdam,
Vol. 3, pp. 77-84.
Federal Emergency Management Agency (FEMA), 2003. NEHRP Recommended Provisions for
Seismic Regulations for New Buildings and Other Structures, FEMA 450, Federal
Emergency Management Agency, Washington, D.C.
Federal Emergency Management Agency (FEMA), 2005. Improvement of Inelastic Seismic
Analysis Procedures, FEMA 440, Federal Emergency Management Agency,
Washington,D.C.

324
Federal Emergency Management Agency (FEMA), 2009. NEHRP Recommended Seismic
Provisions for New Buildings and Other Structures, Document FEMA P-750, 2009
Edition. Building Seismic Safety Council, Washington DC.
Fumal, T.E. and Tinsley, J.C., 1985. Mapping shear wave velocities of near surface geologic
materials, in evaluating earthquake hazards in the Los Angeles region-an earth-science
perspective, U.S. Geol. Surv. Profess. Pap, 1360, 101-126.
Gazetas, G., 1991. Foundation vibrations, Foundation Engineering Handbook, 2nd Edition,
Chapter 15, H.-Y. Fang, ed., Chapman and Hall, New York, NY.
Ghiocel, DM, 2006. ACS SASSI Version 2.3 An advanced computational software for system
analysis for soil-structure interaction; User manual, Technical Report, Rochester, NY.
Harding, Miller, Lawson, and Associates, 1970. Foundation investigation Walnut Creek office
building for Dillingham Corporation, Walnut Creek, California, Report 3008,002.04, San
Francisco, CA.
Haselton, C.B., Leil, A.B., Deierlein, G.G., and Chou, J.H., 2011, Seismic Collapse Safety of
Reinforced Concrete Buildings: I, Assessment of Ductile Moment Frames, Journal of
Structural Engineering, Vol. 137, pp. 481-491.
Holtz, R.D., Kovacs, W.D., and Sheahan, T.C., 2010. An Introduction to Geotechnical
Engineering, 2nd Edition, Prentice Hall, Englewood Cliffs, NJ.
Idriss, IM, and JI Sun, 1991. Users manual for SHAKE91: A computer program for conducting
equivalent linear seismic response analyses of horizontally layered soil deposits, Center
for Geotechnical Modeling, Univ. of California, Davis.
International Building Code, IBC, 2012. 2012 International Building Code, prepared by
International Code Council (ICC), Washington, D.C.
Iguchi, M. and Luco, J.E., 1982. Vibration of flexible plate on viscoelastic medium, J.
Eng.Mech., 108(6), 1103-1120.
Kausel E, A Whitman, J Murray, and F Elsabee, 1978. The spring method for embedded
foundations, Nuclear Engrg. and Design, 48, 377-392.
Kim S and JP Stewart, 2003. Kinematic soil-structure interaction from strong motion recordings,
Journal of Geotechnical and Geoenvironmental Engineering, ASCE, 129(4), 323-335.
Kwok, A.O., Stewart, J.P., Hashash, Y.M.A., Matasovic, N., Pyke, R., Wang, Z., and Yang, Z.,
2007. Use of exact solutions of wave propagation problems to guide implementation of
nonlinear seismic ground response analysis procedures, J. Geotech. & Geoenv. Engrg.,
ASCE, 133 (11), 1385-1398.

325
LeRoy, Crandall and Associates, 1978. Report of supplementary studies proposed retaining walls
proposed Sherman Oaks Galleria, Ventura and Sepulveda Boulevards, Los Angeles,
California, for Kendall International, Report No. ADE-78044, provided by MACTEC,
Los Angeles, CA.
LeRoy, Crandall and Associates, 1982. Completion of exploration program proposed office
building and parking structure, Ventura and Sepulveda Boulevards, Los Angeles,
California, for McNeill Enterprises, Report No. ADE-81384, provided by MACTEC, Los
Angeles, CA.
Lin, A.N. and Jennings, P.C., 1984, Effect of embedment on foundation-soil impedances, J. of
Engrg. Mech., Vol. 110, No. 7, pp. 1060-1075.
Ljung, L., 1987, System Identification: Theory for the User, Prentice Hall, Englewood Cliffs, NJ.
Luco, J.E. and Westmann, R.A., 1971, Dynamic response of circular footings, J. of Engr. Mech.,
Vol. 97, No. 5, pp. 1381-1395.
Luco, J.E. and Contesse, L., 1973, Dynamic structure-soil-structure interaction, Bulletin of
Seismo. Soc. of America, Vol. 63, No. 4, pp. 1289-1303.
Luco, J.E., 1976, Vibrations of a rigid disc on a layered viscoelastic medium, Nuclear Engrg. and
Design, Vol. 36, pp. 325-340.
Luco, J.E., Trifunac, M.D., and Wong, H.L., 1988, Isolation of soil-structure interaction effects
by fullscale forced vibration tests, Eqk. Engrg. and Struc. Dyn., Vol. 16, No. 1, pp. 1-21.
Lysmer, J, M Tabatabaie-Raissi, F Tajirian, S Vahdani, and F Ostadan, 1981. SASSI A system
for analysis of soil-structure interaction, Report No. UCB/GT/81-02, Geotechnical
Engineering, University of California, Berkeley.
Lysmer, J, F Ostadan and C Chin, 1999. SASSI2000 Theoretical Manual, Geotechnical
Engineering Division, Civil Engineering Department, University of California, Berkeley.
Maravas, A., Mylonakis, G., and Karabalis, D.L., 2007. Dynamic characteristics of simple
structures on piles and footings, Proc. 4th Int. Con. on Eqk. Geotech. Engrg.,
Thessaloniki, Greece, June 25-28, Paper No. 1672.
Menq., F.-Y, 2003. Dynamic properties of sandy and gravelly soils, Ph.D. Dissertation, Univ. of
Texas at Austin.
Mikami, A, JP Stewart, F Ostadan, and CB Crouse, 2006. Representation of ground motion
incoherence for the analysis of kinematic soil-structure interaction, Proc. 8th U.S. Nat.
Conf. Eqk. Engrg, April 18-22, San Francisco, CA, Paper 1071.

326
Mikami, A, JP Stewart, and M Kamiyama, M.., 2008. Effects of time series analysis protocols on
transfer functions calculated from earthquake accelerograms, Soil Dyn. and Eqk.Engrg.,
28, 695-706.
Mylonakis, G., Nikolaou, S., and Gazetas, G., 2006. Footings under seismic loading: Analysis
and design issues with emphasis on bridge foundations, Soil Dyn. & Eqk. Engrg., 26,
824-853.
Naeim, F., Tileylioglu, S., Alimoradi, A., and Stewart, J.P., 2008. Impact of foundation modeling
on the accuracy of response history analysis of a tall building, Proc. SMIP2008 Seminar
on Utilization of Strong Motion Data, California Strong Motion Instrumentation
Program, Sacramento, CA, 19-55.
Network for Earthquake Engineering Simulations hub (NEEShub), 2012. Mitigation of collapse
risk in vulnerable concrete buildings, Project NEES-2008-0637, access online at
https://nees.org/warehouse/project/637.
Network for Earthquake Engineering Simulations at University of California, Los Angelse
(NEES@UCLA), 2012. Access online at http://nees.ucla.edu/equipment.html.
Network for Earthquake Engineering Simulations at University of California, Santa Barbara
(NEES@UCSB), 2012. Access online at http://nees.ucsb.edu/facilities/gvda.
National Institute of Standards and Technology, NIST (2012). Soil-Structure Interaction for
Building Structures, Report NIST/GCR 12-917-21, presented by NEHRP Consultants
Joint Venture, J.P. Stewart (project director), September.
Ostadan, F, 2005. Soil-structure interaction analysis including ground motion incoherency effects,
18th International Conference on Structural Mechanics in Reactor Technology (SMiRT
18), Beijing, China.
OYO Corporation, 2007. PS logging at Tokoku Institute of Technology campus in 2007,
unpublished report, September 2007 (in Japanese).
Pais, A. and Kausel, E., 1988. Approximate formulas for dynamic stiffnesses of rigid foundations,
Soil Dyn. & Eqk. Engrg., 7 (4), 213-227.
Pak, R.Y.S., and Guzina, B.B., 1998. Vertical vibration of a circular footing on a linear-wave-
velocity half-space, Geotechnique, 48 (2), 159-168.
Pandit, S.M., 1991. Modal and Spectrum Analysis. Wiley, New York
PEER, 2010. Tall Buildings Initiative, Guidelines for Performance-Based Seismic Design of Tall
Buildings, Report No. 2010/05, Pacific Earthquake Engineering Research Center, UC
Berkeley.

327
Powell, G.H., 2006. Static pushover methods explanation, comparison and implementation,
Modeling for Structural Analysis: Proc. 8th U.S. National Conference on Earthquake
Engineering, San Francisco, California.
Raney Geotechnical, 1983. Foundation investigation North Main Centre, North Main St. and
Principle Ave., Walnut Creek, CA, Repot 029-005, West Sacramento, CA.
Roesset, J.M., 1980. Stiffness and damping coefficients of foundations, Proc. ASCE Geotechnical
Engineering Division National Convention, Oct. 30, 1-30.
Salgado, R., 2008. The Engineering of Foundations, McGraw-Hill, New York.
Scott, R., 1981, Foundation Analysis, Prentice Hall, Englewood Cliffs, NJ.
Star. L., 2011. Seismic vulnerability of structures: Demand characteristics and field testing to
evaluate soil-structure interaction effects, Ph.D. Dissertation, Univ. of CA, Los Angeles.
Stewart, J.P, and Stewart, A.F, 1997. Analysis of soil-structure interaction effects on building
response from earthquake strong motion recordings at 58 sites, Report UCB/EERC-
97/01, University of Berkeley, CA.
Stewart, J.P. and Fenves, G.L., 1998. System identification for evaluating soil-structure
interaction effects in buildings from strong motion recordings, J. Eqk Engng. Struct.
Dyn., 27, 869-885.
Stewart, J. P., Seed, R. B., and Fenves, G. L., 1999a. Seismic soil-structure interaction in
buildings. I: Analytical aspects, J. Geotech. Geoenviron. Engrg. 125, 2637.
Stewart, J.P., Fenves, G.L., and Seed, R.B., 1999b. Seismic soil-structure interaction in buildings.
II: Empirical findings, J. Geotech. Geoenviron. Engrg. 125, 3848.
Stewart, J.P., Kim, S., Bielak, J., Dobry, R., and Power, M., 2003, Revisions to soil structure
interaction procedures in NEHRP design provisions, Earthquake Spectra, Vol. 19, No.
3, pp. 677-696.
Stokoe, K.H., Graves J.C., Graves M.T., Kurtulus, A., and Menq, F.Y., 2004. Data Report:
SASW measurements at the NEES Garner Valley Test Site, California., Access online at
http://nees.ucsb.edu/facilities/gvda.
Tileylioglu, S., 2008. Evaluation of soil-structure interaction effects from field performance data,
Ph.D. Dissertation, Univ. of California, Los Angeles.
Tileylioglu, S., Naeim, F., Alimoradi, A., and Stewart, J.P., 2010. Impact of foundation modeling
on the accuracy of response analysis for a tall building, Proc. 9th US National & 10th
Canadian Conf. on Earthquake Engrg.,, EERI and Canadian Assoc. for Earthquake Eng.,
July 25-29, 2010, Paper No. 1666.

328
Tileylioglu, S, JP Stewart, and RL Nigbor, 2011. Dynamic stiffness and damping of a shallow
foundation from forced vibration of a field test structure, J. Geotech. & Geoenv. Engrg.,
137 (4), 344-353.
Tohoku Electric Power Company, 2009. Seismic analysis of Onagawa NPS#1, Siryo 2-1 report
(in Japanese).
Tohoku Electric Power Company, 2009. Vibration characteristics of the ground site, Siryo 2-3
report (in Japanese).
Tohoku Electric Power Company, 2011. Accelerograms observed at Onagawa NPS during 2011
Great East Japan Earthquake, available on CD-ROM from Japan Association for
Earthquake Engineering.
Veletsos, A.S. and Wei, Y.T., 1971. Lateral and rocking vibrations of footings. J. Soil Mech. and
Found. Div., 97(9), 1227-1248.
Veletsos, AS, and B. Verbic, 1973. Vibration of viscoelastic foundations, J. Eqk. Engrg. Struct.
Dyn., 2, 87-102.
Veletsos, A.S. and Meek, J.W., 1974. Dynamic behavior of building-foundation systems, J. Eqk.
Engrg. and Str. Dyn., 3, 121138.
Veletsos, A. S., and Nair, V. V., 1975. Seismic interaction of structures on hysteretic foundations,
J. Struct. Engrg. 101, 109129.
Veletsos, A. S., 1977. Dynamics of Soil-Foundation Systems, in W.J. Hall (ed.) Structural and
Geotechnical Mechanics, Prentice-Hall, Inc., Englewood Cliffs, New Jersey, pp 333-361.
Veletsos, AS and AM Prasad, 1989. Seismic interaction of structures and soils: Stochastic
approach, J. Struct. Engrg., 115 (4), 935956.
Veletsos, AS, AM Prasad, and WH Wu, 1997. Transfer functions for rigid rectangular
foundations, J. Eqk. Engrg. Struct. Dyn., 26 (1), 5-17.
Vrettos, C., 1999. Vertical and rocking impedances for rigid rectangular foundations on soils with
bounded non-homogeneity, J. Eqk. Engrg. and Str. Dyn., 28, 1525-1540.
Wolf, J.P., 1985. Dynamic Soil-Structure Interaction, Prentice-Hall, Inc., Englewood Cliffs, New
Jersey.
Yamada, S., Hyodo, M., M., Orense, R.P., Dinesh, S.V., and Hyodo, T., 2008. Strain-dependent
dynamic properties of remolded sand-clay mixtures, J. Geotech. & Geoenv. Engrg., 137
(4), 344-353.
Zerva, A and V Zervas, 2002. Spatial variations of seismic ground motion an overview. Applied
Mechanics Reviews, 55, 271-297.

329