You are on page 1of 16

Engineering Structures 215 (2020) 110709

Contents lists available at ScienceDirect

Engineering Structures
journal homepage: www.elsevier.com/locate/engstruct

Numerical investigations of structure-soil-structure interaction in buildings T


a,⁎ b
Chandrakanth Bolisetti , Andrew S. Whittaker
a
Idaho National Laboratory, Idaho Falls, ID 83404, USA
b
University at Buffalo, The State University of New York, North Campus, 212 Ketter Hall, Amherst, NY 14260, USA

A B S T R A C T

The influence of structure-soil-structure interaction (SSSI) in low- to medium-rise buildings is investigated through numerical simulations, and observations are
compared with those from previous studies that analyzed data from a set of centrifuge experiments of similar models. The buildings include a one-story moment-
resisting frame building on spread footings and a two-story shear wall building on a basemat. The numerical simulations are performed using the industry-standard,
frequency-domain, linear analysis code SASSI, and the time-domain nonlinear finite-element analysis code, LS-DYNA. In LS-DYNA the simulations are performed
with and without geometric nonlinearities (gapping, sliding and uplift) to understand their effects on SSSI. Three plan arrangements of the buildings are considered to
characterize the influence of relative location on SSSI: (1) an in-plane SSSI (iSSSI) arrangement, in which the two buildings are placed adjacent to each other along a
line parallel to the direction of ground shaking, (2) an anti-plane arrangement (aSSSI), in which the two buildings are placed adjacent to each other along a line
perpendicular to the direction of ground shaking, and (3) a combined in-plane-anti-plane (cSSSI) arrangement, in which two shear wall buildings are placed at two
adjacent sides of the frame building on footings. Results from the numerical simulations in SASSI and LS-DYNA show that SSSI has negligible effect on the global
spectral accelerations of the buildings in these arrangements. The numerical simulations agree with experimental observations in this regard. Numerical in-
vestigations into the SSSI response of the frame building on footings placed adjacent to a deep basement show that the presence of the deep basement reduces uplift in
the footings and results in smaller peak spectral accelerations at the roof, underscoring the potential importance of geometric nonlinearities (gapping, sliding and
uplift) in SSSI and foundation design.

1. Introduction analytical calculations of highly simplified structures founded on


idealized soil profiles. The seminal study by Luco and Contesse [16]
Most commercial and industrial buildings are constructed in clusters examined anti-plane interaction between two infinitely long shear walls
and not in isolation. During an earthquake, the dynamic response of one subjected to vertically incident SH waves of harmonic time dependence.
structure can affect the response of a neighboring structure, resulting in Wong and Trifunac [36] extended this study to an array of structures
structure-soil-structure interaction (SSSI). The mechanisms of SSSI in- with varying size and stiffness subjected to a shear wave incident at an
clude (1) changes in the input ground motion due to the presence, or arbitrary angle. Both of these analytical studies investigated the sig-
the shaking, of a neighboring structure (wave-based SSSI), and/or (2) a nificance of parameters such as separation distance, foundation size,
change in the foundation restraint due to the presence of an adjacent and stiffness of the structures on SSSI. Luco and Contesse [16] identified
structure. Wave-based SSSI can also manifest as ‘shielding’, where the factors that determine the degree of interaction between structures: (a)
deep basements of buildings, or structures such as tunnels, protect other relative foundation sizes, (b) distance between the structures, (c) mass
buildings from incoming surface waves or inclined body waves during of the superstructure relative to the mass of the soil excavated for the
an earthquake. foundation, (d) mass of the foundation relative to the mass of excavated
Design standards in the United States such as ASCE/SEI 7-16 [2] do soil, and (e) relative stiffness of the structures and the soil. They con-
not include rules or guidance for consideration of SSSI. Indeed, it is cluded that SSSI effects could be important for smaller and lighter
unknown if the effects of SSSI are sufficient to warrant consideration in structures located close to heavier structures. A similar conclusion was
either analysis and/or design. However, in the context of performance- drawn by Wong and Trifunac [36]. Both studies noted that the degree
based earthquake engineering and seismic probabilistic risk assessment, of interaction depends mainly on the type of wave interference (con-
it is important to understand SSSI phenomena and address them if structive or destructive) occurring between the scattered waves from
important. the foundations, which is a function of the spacing and arrangement of
Unlike soil-structure interaction (SSI), very few studies have been the foundations. Some studies (e.g., Matthees and Magiera [24], Lin
conducted on SSSI. Previous SSSI studies were mostly limited to et al. [14], Qian and Beskos [28], Padron et al. [27], Anderson et al. [1]


Corresponding author.
E-mail addresses: chandu774@gmail.com (C. Bolisetti), awhittak@buffalo.edu (A.S. Whittaker).

https://doi.org/10.1016/j.engstruct.2020.110709
Received 30 January 2019; Received in revised form 8 February 2020; Accepted 22 April 2020
Available online 30 April 2020
0141-0296/ © 2020 Elsevier Ltd. All rights reserved.
C. Bolisetti and A.S. Whittaker Engineering Structures 215 (2020) 110709

and Bolisetti and Whittaker [3]) have investigated SSSI between ad- subjected to ground motions of different intensities. These ground
jacent structures (buildings and nuclear power plants) using three-di- motions included both ordinary motions and pulse motions with for-
mensional numerical methods, and most of these studies concluded that ward directivity. The data gathered from these tests has been carefully
SSSI can affect structural response. A comprehensive list of SSSI studies curated in data reports [19,20,22,31].
can be found in Menglin et al. [25]. The study presented in this paper numerically simulates the re-
A key reason for the lack of study of SSSI is the lack of data from sponse of models of buildings that are similar to those tested in Test 3
either the field or laboratory experiments. Çelebi [7] analyzed data and Test 4 of the City Block project. However, there is no attempt to
recorded during earthquake shaking from two neighboring buildings to either 1) validate or calibrate soil-structure models, or 2) compare
examine SSSI. He employed spectral analysis techniques to establish numerical predictions and measured responses, for the reasons de-
correlations between the responses of the buildings to the 1987 Whit- scribed in Section 3. The two building models chosen for this study are:
tier-Narrows earthquake. The study examined ground motion on the 1) a one-story, steel moment-resisting frame structure founded on
soil surface away from the buildings (assumed as free field), and re- spread footings (hereafter referred to as MS1F_2), and 2) a two-story
sponse at the roof and basement levels, and concluded that (1) the shear wall structure on a basemat (hereafter referred to as MS2F). The
dynamic responses of the buildings affected both the adjacent basement beams and columns of MS1F_2 were fabricated using 1/2″ x 1/2″ x 1/
responses and the free-field responses, and (2) there was considerable 16″ A513 steel tubing; 6061 aluminum was used for the footings. High
SSSI between the structures at specific frequencies. The second con- lateral stiffness in the direction orthogonal to the shaking was achieved
clusion corroborates the findings of Murakami and Luco [26], who with 2″ deep and 1/8″ thick steel plates. The shear walls and basemat of
analytically investigated the interaction between an array of idealized MS2F were fabricated from aluminum 6061 plates, 1/2″ and 1″ thick,
shear-wall-like structures and found that the interaction between these respectively. Much additional information on the design and fabrication
structures was the most prominent at the so called Rayleigh fre- of these models can be found in Bolisetti and Whittaker [4] and the City
quencies, which are determined by the properties of the soil and sizes of Block project data report [31]. Both models were placed on dense, dry
the foundation [26]. More recent studies include two small-scale ex- Nevada sand at a relative density of 80%. The soil in the laminar box
periments in a geotechnical centrifuge. These studies included the NSF- was 29.5 m deep at the prototype scale (55:1) and had a density of
funded NEES City Block project [6], which aimed in part to investigate 16.66 kN/m3. Models MS1F_2 and MS2F had 1st mode fixed-base nat-
SSSI between buildings in dense urban regions of high seismic hazard. ural periods of 0.47 sec (2.13 Hz) and 0.13 sec (7.69 Hz), respectively,
Subsequently, Kirkwood and Dashti [12] investigated SSSI between at the prototype scale. Both models had a flexible-base, prototype-scale
buildings on soils vulnerable to liquefaction, using the same centrifuge. natural period close to 0.6 sec (1.67 Hz), which is equal to the first
There have been very few SSSI studies utilizing three-dimensional mode natural period of the soil profile. The similarity between the
nonlinear finite-element simulations, due, in part, to the numerical flexible-base natural periods of the models and the natural period of the
challenges and computational expense. Specifically, the effects of SSSI soil profile was expected to amplify SSI and SSSI effects in the building
on nonlinear foundation response, including gapping, sliding and up- models. A detailed description of the designs of MS1F_2 and MS2F and
lift1, have not been investigated. In this paper, the authors build on their experimental models are presented in Trombetta et al. [34] and
their previous studies [4,5,8] that benchmarked nonlinear SSI simula- Bolisetti and Whittaker [4]. Photographs of the building models used in
tions using the commercial finite element codes LS-DYNA [15] and the experiments are presented in Fig. 1.
ABAQUS [9], to numerically investigate SSSI in models of low-to- The two models were designed to have first-mode properties of low-
medium-rise buildings [32,34] tested in two of the six experiments that to medium-rise buildings at the prototype scale. To investigate the ef-
comprised the City Block project. The goals of the study reported here fect of the relative positioning of these buildings on their seismic re-
are twofold: (1) investigate if SSSI influences the global response of sponses, and to reproduce conditions similar to those in a dense urban
low- to medium-rise buildings, and verify that the numerical and ex- region, four different arrangements of the models were tested in Tests 3
perimental observations agree in this regard, and (2) identify, through and 4. These arrangements are illustrated in Fig. 2 and photographs of
comprehensive 3D nonlinear finite-element simulations, the mechan- the test setups are presented in Fig. 3. The arrangements were:
isms of SSSI that result in a considerable influence on the structural
response. These goals are achieved by analysis of numerical and la- 1) A solitary MS1F_2 model, far away from other structures and which
boratory experiments, and exploratory numerical simulations to in- served as a baseline in the examination of SSSI response
vestigate other aspects of SSSI that were not considered in the City 2) An in-plane SSSI (iSSSI) arrangement, in which the MS1F_2 and
Block project. MS2F models were placed adjacent to each other along a line par-
allel to the direction of ground shaking
3) An anti-plane arrangement (aSSSI), in which the two models were
2. Buildings and ground motions chosen for numerical analysis
placed adjacent to each other along a line perpendicular to the di-
rection of ground shaking
2.1. Description of the building specimen
4) A combined in-plane-anti-plane (cSSSI) arrangement, in which two
MS2F models were placed at two adjacent sides of the MS1F_2
The City Block project included tests of small-scale model buildings
model.
at the UC Davis geotechnical centrifuge facility. The first two experi-
ments, Test 1 and Test 2, involved low- to mid-rise buildings designed
The first two arrangements were tested together during Test 3 and
to have realistic, code-conforming, dynamic properties with nonlinear
the remaining two arrangements were tested during Test 4. During test
behavior at the global level. These buildings were placed far apart in
planning, examining SSSI effects in MS1F_2 was prioritized and the
Test 1 and adjacent to each other in Test 2, and comparisons of the
baseline case of MS2F was not tested. Structure-soil-structure interac-
structural responses from the two experiments provided insight into
tion in MS1F_2 was examined by comparing the structural response in
SSSI in such structures [23]. Test 3 and Test 4 involved buildings that
the SSSI arrangements tested during Test 3 and Test 4, to the response
were much stiffer than those in Test 1 and Test 2, and designed to
of the baseline model tested during Test 3. Model MS2F was designed to
exacerbate SSI and SSSI effects. The models in each experiment were
have the same flexible-base period as MS1F_2 but was significantly
heavier. Through this arrangement, MS2F was expected to ‘transmit’
1
Gapping is the separation of a foundation and the surrounding soil; sliding is energy into the adjacent MS1F_2 and amplify its response during
relative horizontal motion at the foundation-soil interface; and uplift is the ground shaking. This ‘transmitter-receiver’ hypothesis was evaluated in
complete loss of contact between the foundation and the surrounding soil. Test 3 and Test 4. Although the City Block project focused on SSSI

2
C. Bolisetti and A.S. Whittaker Engineering Structures 215 (2020) 110709

Fig. 1. Building models used in the City Block project.

Fig. 2. SSSI arrangements tested in Test 3 and Test 4 of the City Block project (arrows indicate direction of shaking) [adapted from Trombetta [33]].

Fig. 3. Experimental setups in the laminar box for Test 3 (left) and Test 4 (right) (shaking is parallel to the longer side of the laminar box).

effects on MS1F_2, the study presented here also investigates SSSI ef- was used for the numerical simulations of this study. The City Block
fects on MS2F through numerical simulations. ground motions were chosen through a deterministic and probabilistic
seismic hazard analysis performed by Mason et al. [18]. Those ground
motions were chosen to represent shaking that could be expected at a
2.2. Input ground motions site in downtown Los Angeles and include both ordinary and near-fault
motions with forward directivity. Four of these input ground motions,
A subset of the ground motions used in the City Block experiments

3
C. Bolisetti and A.S. Whittaker Engineering Structures 215 (2020) 110709

Table 1
Input ground motions used for numerical simulations.
Ground Motion ID Earthquake Station Typea Peak acceleration (g)

JOS_L 92 Landers Joshua Tree 090 Ord 0.06


LCN 92 Landers Lucerne 260 NF, FD 0.26
SCS_H 94 Northridge Sylmar Conv Sta. 052 NF, FD 0.58
PRI_H 95 Kobe Port Island-Mod-79 m NF, FD 0.64

a
Type of ground motion: Ord = ordinary, NF = near fault, FD = forward directivity.

3. Numerical modeling and simulations

3.1. Background

The numerical investigations presented in this paper of this study


seek to characterize (1) if simulations using the state-of-the-art nu-
merical tools also show insignificant SSSI effects in the subject build-
ings, (2) if nonlinear behavior of the soil and the foundation-soil in-
terface play a role in local or global SSSI, and (3) if restraint on a spread
footing from a neighboring foundation affects its hysteretic response.
However, the numerical models used in this study do not replicate the
centrifuge test models and are not calibrated with the data from the
experiments. They differ from the models tested in the centrifuge in the
following ways:

1) They are scaled up to the prototype scale from the centrifuge model
scale (1:55).
Fig. 4. Acceleration response spectra (5% damping) of the input ground mo- 2) The soil domain is assumed to be infinite and the boundary effects in
tions. the laminar box, which were not characterized, are not simulated.
3) Nonlinear soil properties, derived from modulus reduction and
denoted JOS_L, LCN, SCS_H, and PRI_H, represent a broad range of damping curves, also not available from the experiments, were de-
intensities and frequency content and are used for this study. Table 1 rived from Seed and Idriss [30].
describes these ground motions and Fig. 4 presents their acceleration 4) Each input ground motion is applied separately and not con-
response spectra at a damping ratio of 5%. The centrifuge models were secutively to the same model as in the experiments. In the experi-
subjected to uniaxial shaking applied at the base of the laminar box. In ments, the input ground motions were input consecutively to the
the numerical models, the same ground motions (recorded at the base model resulting in cumulative soil strain, cumulative structural da-
of the laminar box), converted to the prototype scale, were input at the mage, and changing soil properties due to consolidation. The nu-
base of the soil domain, as described in Section 3. merical analysis for each ground motion is performed assuming zero
initial strain in the soil and the structure.

2.3. Experimental responses of the buildings These deviations of the numerical models from the experimental
models, namely, the elimination of scale effects, boundary effects,
The data gathered from the centrifuge Tests 3 and 4 have been centrifuge-shaker interaction, and cumulative damage, offer major ad-
analyzed by the members of the City Block team [4,34,35] to in- vantages of simplicity in interpreting the responses, and an unbiased
vestigate SSI and SSSI effects. This analysis showed that SSI effects were insight into SSSI effects. Consequently, the numerical results presented
significant in the buildings of Test 3 and Test 4 and they exhibited in Section 4, and the experimental results, previously presented in
period elongations that were similar to the pre-test numerical predic- Section 2.3, cannot and should not be compared directly.
tions [34]. However, the analysis also showed that wave-based, global This study presents numerical SSI analyses performed for the ar-
SSSI effects had very little influence on the response of the buildings rangements presented in Fig. 2. Although the City Block experiments
and invalidated the ‘transmitter-receiver’ hypothesis. This can be seen did not include a standalone test of MS2F (i.e., no neighboring struc-
from the acceleration response spectra in Fig. 5 below, which presents tures), this case was analyzed numerically. The industry-standard SSI
the roof accelerations of MS1F_2 and MS2F from Test 3 and Test 4 analysis codes, SASSI [17] and LS-DYNA were used for the numerical
(Fig. 3) for the ground motions considered in this study. Note that the simulations. SASSI is a widely used code for SSI analysis and performs
anomalous responses of MS1F_2 in the aSSSI case, and MS2F in the linear analysis in the frequency domain. LS-DYNA is a commercial fi-
caSSSI case in this figure were found to be a consequence of faults in the nite-element program capable of nonlinear three-dimensional SSI ana-
assembly of the building models [32]. Additionally, the variability in lysis in the time domain. Soil-structure interaction analyses in LS-DYNA
the input ground motions and material properties between Test 3 and are performed using the direct method [4,5,8]. The direct method
Test 4 also contributed to small differences between the responses of (Fig. 7) involves simulating a large soil domain around the structure
each building models in various arrangements (see Bolisetti and Whit- and providing the ground motion input at the bottom of this soil do-
taker [4]). Excluding the anomalous data, and the small differences due main as either as an acceleration (or displacement or velocity) history
to material property and input ground motion variability, the roof re- for simulating a within-profile input, or as a shear force history for si-
sponses of both the structures in various arrangements are very similar, mulating an outcrop input. This method enables the simulation of
indicating that SSSI had almost no influence on the global responses of material nonlinearities in the soil and the structure, and geometric
these models. nonlinearities such as gapping, sliding, and uplift at the foundation-soil
interface. Unlike the experiments, the SSSI arrangements and baseline

4
C. Bolisetti and A.S. Whittaker Engineering Structures 215 (2020) 110709

a. MS1F_2

b. MS2F
Fig. 5. 5% damped acceleration response spectra at the roofs of MS1F_2 and MS2F calculated from experimental data.

Table 2 to the end nodes of the beams, and the roof masses are included at the
Specifications of the building model, MS1F_2a,b, in prototype scale. center nodes of the roof beams. The shear walls of MS2F are also
Plan Height (m) Nodal masses on Moment of Inertia (m4) modeled as beam elements with the corresponding cross-section prop-
dimensionsb each beamc (kg) erties. The masses of the shear walls are lumped at the end nodes of the
L × B (m) beams. The beams and columns of MS1F_2 are modeled using the ma-
Center End Beamsc Columns terial properties of A513 steel, consistent with the experimental models.
9.96 × 11.00 12.76 145,578 1820 0.0106 0.0141
Similarly, the footings of MS1F_2, and the columns of MS2F are mod-
Footing dimensions (m) 4.37 × 4.37 eled with the material properties of 6061 aluminum. The footings of
Footing thickness (m) 0.875 MS1F_2 and the basemat of MS2F are both modeled as rigid using solid
a
elements. The rigidity is modeled in SASSI by providing a large elastic
The area of the cross section, shear area, and torsional inertia are the same modulus, and in LS-DYNA using the material model, *MAT_RIGID, that
for all elements, and are calculated from the cross section of the experimental
simulates rigid-body dynamics. Table 2 and Table 3 provide the spe-
model.
b cifications of the numerical models of the two buildings. Fig. 6 presents
L is along the direction of shaking; see Fig. 2.
c
Only for the beams parallel to the direction of shaking. the SASSI and LS-DYNA numerical models of the cSSSI arrangement
(see Fig. 2).
Table 3
Specifications of the building model, MS2F, in prototype scale. 3.2. Soil domain modeling
Story heights (m) Moment of Inertiaa,b (m4)
The mechanical properties of the soil were not measured in the
Y-bending X-bending experiments and so representative soil properties for Nevada sand are
12.58, 12.58 70.63 0.51
used for numerical modeling. The weight density of the Nevada sand is
Mat dimensions (m) 13.80 × 13.80 16.66 kN/m3.
Mat thickness (m) 1.34 Fig. 8, from Bolisetti and Whittaker [4], presents the small-strain
a
shear moduli over the height of the soil column. The small-strain shear
The area of the cross section, shear area, and torsional inertia are the same modulus ranges between 31.9 MPa at the surface and 178.4 MPa at a
for all elements, and are calculated from the cross section of the experimental
depth of 29.5 m. The corresponding small-strain shear-wave velocities
model.
b are 137 m/sec and 325 m/sec, respectively. Based on the re-
X is the direction of shaking; Z is the vertical direction; see Fig. 2.
commendations of Appendix B of ASCE/SEI 4-16, and as reported in
Bolisetti and Whittaker [4] and Coleman et al. [8], the maximum layer
models (shown in Fig. 2 and Fig. 3) are each analyzed in a separate
thickness was calculated as the small-strain shear-wave velocity divided
numerical simulation. The beams and columns of MS1F_2 are modeled
by 10 times the maximum frequency of interest. Using a maximum
using beam elements. The masses of the beams and columns are lumped
frequency of 8 Hz, which is higher than the fixed-base frequency of

5
C. Bolisetti and A.S. Whittaker Engineering Structures 215 (2020) 110709

Fig. 6. Numerical models of the cSSSI arrangement in SASSI (left) and LS-DYNA (right).

Fig. 7. The direct method of SSI analysis [from Bolisetti and Whittaker [4]].

MS2F (MS2F responds predominantly in the rocking mode with SSI, see linear backbone curve to simulate soil hysteresis. This material model is
Trombetta et al. [34] and much higher than the flexible-base fre- based on a multi-yield surface plasticity approach in which the back-
quencies of both the structures (see Section 2.1), the maximum layer bone curve is internally converted to a set of elastic-perfectly-plastic
thickness is 1.7 m (= 137/(10 × 8) ) at the surface and 4.1 m ‘layers’ acting in parallel. Each of these layers follows the Von Mises
(= 325/(10 × 8) ) at a depth of 29.5 m. Here, the 29.5-m deep soil profile yield criterion and the stress in the material is calculated as the sum of
is modeled using 29 layers, as follows, from the surface: 2 of 0.67 m, stresses in the layers. This material is similar to the distributed-element
0.46 m, 2 of 1.0 m, 22 at 1.05 m, and 2 of 1.3 m. This layering scheme class of models proposed by Iwan [11] and simulates a hysteresis that
results in a maximum frequency of 20.4 Hz (= 137/(10 × 0.67) ) for the follows Masing rules [13]. A sample hysteresis loop simulated by this
uppermost elements and 25.0 Hz (= 325/(10 × 1.3) ) for the lowermost material model is illustrated in Fig. 10 below. The modulus-reduction
elements, which are much higher than those of the MS1F and MS2F. curve presented in Fig. 9 is converted to a stress–strain backbone curve
Inputs for SSI analysis in SASSI were established using the equiva- and used as input to this material model. A frequency-independent,
lent-linear site-response analysis code, SHAKE [29] using the modulus small-strain damping of 1% is provided using the *DAMPING_FREQU-
reduction and damping curves for dry sand from Seed and Idriss [30]. ENCY_RANGE formulation [10].
These curves are presented in Fig. 9. The lower bound modulus-re- The numerical models are subjected to unidirectional ground
duction curve and the upper bound damping curve, both of which, are shaking, similar to the centrifuge models. The soil domain of the LS-
available in the SHAKE database, were employed for these simulations DYNA model shown in the figure is 218 m × 118 m in plan (218 m in
to increase the nonlinearity of the soil. The strain-compatible shear the shaking direction) and 29.5 m deep. In contrast, the laminar box
modulus and damping profiles calculated from the SHAKE simulations used in the City Block tests corresponds to a soil domain that is 83 m ×
for the ground motions used in this study are presented in Fig. 8. A 28 m in plan and 29.5 m deep at the prototype scale. As illustrated in
detailed description of the free-field analyses (i.e., no structures) is Fig. 7, the lateral boundaries of the soil domain of the LS-DYNA model
provided in Bolisetti and Whittaker [4], and results are also presented are constrained such that the nodes at each elevation move together in
in Bolisetti et al. [5]. the horizontal and vertical directions. This boundary condition ensures
The soil material in LS-DYNA is modeled using the that the elements at the boundary undergo pure shear deformation
*MAT_HYSTERETIC_SOIL material model [15], which features a multi- under horizontal shaking and pure axial deformation under vertical

6
C. Bolisetti and A.S. Whittaker Engineering Structures 215 (2020) 110709

the foundation-soil interface cannot be modeled in SASSI. A detailed


description of the numerical models and their specifications are pre-
sented in Bolisetti and Whittaker [4].
The ground motions listed in Table 1 are imposed at the base of the
soil domain as acceleration histories, simulating a ‘within-profile’ input.
Since these ground motions were recorded at the base of the laminar
box during the centrifuge experiments, this maintains equivalence be-
tween the ground motion inputs in the experiments and the numerical
simulations. The input motions are imposed in the X direction (along
the longer dimension of the numerical model in the right panel of
Fig. 6), and the bottom surface of the model is restrained against dis-
placements in the Y and Z directions. The same boundary conditions
and ground motion input methods are adopted in the SASSI simula-
tions.
Prior to performing nonlinear SSSI analyses, the LS-DYNA site-re-
sponse and SSI models are benchmarked against SHAKE and SASSI,
respectively, for linear and nonlinear analyses. These benchmarking
results, not repeated here, are presented in Bolisetti et al. [5] and Bo-
lisetti and Whittaker [4].

4. Analysis results

The degree of SSSI in the response of each structure in the ar-


rangements of Fig. 2 is judged by comparing with its baseline response,
Fig. 8. Small-strain and strain-compatible shear moduli and damping ratios namely, with no adjacent structures. The structural responses are re-
used for SSI analyses using SASSI for simulation of Tests 3 and 4. ported using 5% damped acceleration response spectra at the roof level.
The numerical responses calculated using SASSI, LS-DYNA with tied
foundation, and LS-DYNA with separation allowed at the foundation,
shaking. These deformation patterns are necessary to simulate a free-
are presented for the ground motions from Table 1 for each structure.
field condition at the boundaries of the model, as described in Bolisetti
These different foundation constraints are simulated to characterize the
and Whittaker [4] and Coleman et al. [8]. The domain size of the LS-
possible significance of geometric nonlinearities at the soil-structure
DYNA models is chosen such that the boundary effects are minimized so
interface (i.e., gapping, sliding, and uplift) on SSSI. Structure-soil-
that an approximate infinite soil domain is simulated. The domain size
structure interaction effects in MS1F_2 are also examined using footing
is verified by comparing the response at the edges of the soil domain to
response data from the numerical simulations, noting that the scale
the response at the same elevation calculated from a free-field site-re-
used for the experiments (1:55) is too small to investigate the effect of
sponse analysis. This comparison is presented in Bolisetti and Whittaker
foundation constraint from the experimental data. Comparisons of the
[4] and Bolisetti et al. [5]. The LS-DYNA analyses are performed for two
numerical responses calculated using SASSI and LS-DYNA are not the
cases: 1) the foundation (footings or basemat) ‘tied’ to the surrounding
focus of this study and so the numerical code-based differences in the
soil with no separation at the foundation-soil interface, and 2) separa-
responses of the building models are not reported. The SSSI effects
tion allowed at the foundation-soil interface, leading to potential gap-
observed from the predictions of each code are examined and compared
ping and sliding. The former is modeled in LS-DYNA using the *CON-
below. The authors have performed another study examining numerical
STRAINED_EXTRA_NODES option, which constrains the soil nodes at
code-based differences for the same buildings, soil profiles, and input
the foundation-soil interface to move with the rigid foundation. The
ground motions. This study is published in Bolisetti et al. [5] and Bo-
latter is modeled using *CONTACT_AUTOMA-
lisetti and Whittaker [4]. These publications also present and examine
TIC_SURFACE_TO_SURFACE option, which simulates both gapping and
the free-field responses calculated using equivalent-linear and non-
a Coulomb-type sliding at the foundation-soil interface. Separation at
linear tools for the same soil profile and ground motions.

Fig. 9. Modulus reduction and damping curves (mean, upper bound and lower bound) adopted from Seed and Idriss [30].

7
C. Bolisetti and A.S. Whittaker Engineering Structures 215 (2020) 110709

Fig. 10. Hysteresis loop of the *MAT_HYSTERETIC_SOIL model for a given input strain and the corresponding backbone curve.

4.1. MS1F_2 rotations, when restrained by a deep basement of a neighboring


structure. Mason [21] concluded from Tests 3 and 4 that SSSI did not
Fig. 11 presents 5% damped acceleration response spectra at the have a significant effect on the displacements of the footings closest to
roof of MS1F_2 for the four ground motions considered in this study. the basemat.
This figure shows that the roof response of MS1F_2 is not affected by In this section, the local effects of SSSI on the footings of MS1F_2
SSSI. The same conclusion is drawn from examination of the numerical adjacent to the MS2F basemat are investigated numerically by ex-
results calculated using SASSI, LS-DYNA with a tied foundation-soil amining the moment-rotation responses of the footings. Only the
interface, and LS-DYNA with a separation-allowing interface. The nu- footing responses calculated from LS-DYNA (separation allowed) are
merical predictions also agree reasonably well with the experimental examined here, since this is the only model that simulates both 1) the
results, in regard to the degree of SSSI in MS1F_2. nonlinear hysteretic response of the footings, and 2) gapping, sliding
The shapes of the response spectra from the numerical simulations and uplift.
(Fig. 11) and the experiments (Fig. 5) are similar for the JOS_L ground Fig. 12, Fig. 13, Fig. 14 and Fig. 15 present the moment-rotation
motion. Both SASSI and LS-DYNA reasonably predict the flexible-base responses of two footings in MS1F_2 during the ground motions, JOS_L,
period for this ground motion. For the other ground motions the re- LCN, SCS_H and PRI_H, respectively, for the baseline case as well as the
sponse spectra calculated from SASSI and LS-DYNA (with a tied inter- SSSI arrangements illustrated in Fig. 2. The first row of plots in each
face) have a very different shape from the spectra calculated from the figure presents the responses of the footing that is situated farthest from
experimental data. The LS-DYNA analysis with the separation-allowed the adjacent structures (referred to as ‘far footing’ hereafter), and the
interface results in response spectra that are similar in shape to the second row presents the responses of the footing that is closest to the
experimental responses for SCS_H and PRI_H ground motions, and adjacent structures (referred to as ‘near footing’ hereafter). The location
better predict the period of peak spectral acceleration, indicating that of the footing in the plan view of each arrangement is identified by a
the simulation of gapping and sliding of the footings provided more solid black square in an inset in the top left corner of each plot.
reasonable predictions of the spectral shape and the period of peak The plots presented in the first row of each figure show that the
spectral acceleration. This is an expected result because the soil around moment-rotation behavior of the far footing is virtually identical for all
the footings is highly nonlinear and footing uplift is expected during the arrangements, indicating that it is barely affected by the presence of
stronger ground motions. The results in Fig. 11 also show considerable adjacent structures. However the moment-rotation response of the near
differences between the individual building responses from SASSI and footing is affected, although not considerably, for all ground motions.
LS-DYNA. These differences arise from the nonlinear foundation re- The plots corresponding to the iSSSI and cSSSI cases (second and fourth
sponse, as well as the nonlinear soil response, which cannot be simu- in each row) of the near footing show larger negative rotations on
lated in the linear analysis of SASSI. This is explained in detail in Bo- average (the plots are skewed slightly to the left), than the baseline and
lisetti et al. [5] and is not discussed here. aSSSI cases, for all four ground motions. Negative rotation denotes
footing rotation towards basemat (i.e., the side of the footing closest to
the basemat is moving downwards) in the iSSSI arrangement for the co-
4.1.1. Moment-rotation response of the footings of MS1F_2
The results presented in Fig. 11 show that SSSI has negligible effect ordinate system used here. This observation indicates that the footing,
when close to the basemat of the iSSSI and ciSSSI models of MS2F,
on the global response of MS1F_2 during Test 3 and Test 4. However,
SSSI may still affect the local response of the footing and supported undergoes more negative, asymmetric, rotation and rotates towards the
basemat when compared with its response in the baseline and aSSSI
columns due to restraint provided by adjacent substructures. Analysis of
data from Tests 1 and 2 by Mason [21] and Trombetta [33] showed that cases. However, this asymmetric rotation is not significant enough to
result in a consistent change in the peak footing moment or rotation in
the footings of MS1F_12 underwent smaller displacements and
the near footing in comparison with the far footing. As such, no con-
siderable change is noticed in the peak moments in the footings due to
2
MS1F_1 is an experimental model used in Test 1 and Test 2 and designed to the presence of the adjacent structure.
be representative of low- to mid-rise, steel moment-resisting frame buildings. Although the moment-rotation responses for JOS_L and LCN ground
This model has the design properties (natural period, yield drift ratio and yield
base shear ratio) of a three-story steel moment-frame building in the prototype
scale. It has a larger story height, roof mass and foundation mass compared to (footnote continued)
MS1F_2 and therefore has a fixed-base natural period of 1.1 sec. Note that natural period of 0.47 sec. The dimensions and properties of MS1F_1 are pro-
MS1F_2 was designed later to elicit significant SSI effects and has a fixed-base vided in Bolisetti and Whittaker [4].

8
C. Bolisetti and A.S. Whittaker Engineering Structures 215 (2020) 110709

Fig. 11. Acceleration response spectra at the roof of MS1F_2 calculated from the numerical simulations.

motions are almost symmetric, the responses for SCS_H and PRI_H of MS1F_2 showed negligible change due to SSSI, even for the cSSSI
ground motions result in negative slopes at large rotations, indicating arrangement, which comprised two MS2F models adjacent to the
negative stiffness. The negative stiffness is a result of partial uplift of the MS1F_2, indicating that MS2F did not perform the expected ‘transmit-
footings as the axial force in the corresponding columns decreases due ting’ function. This section examines if the presence of MS1F_2 affects
to increasing lateral force in the superstructure. The negative stiffness the response of MS2F through SSSI. Fig. 16 below presents the 5%
also results in increased spectral accelerations at the roof as seen in the damped acceleration response spectra at the roof of MS2F calculated
differences between the responses from the ‘separation allowed’ and from SASSI, LS-DYNA with tied foundation, and LS-DYNA with se-
‘tied’ cases of MS1F_2 in Fig. 11. paration allowed at the foundation, for the four ground motions, JOS_L,
LCN, SCS_H and PRI_H.
4.2. MS2F The numerical results presented in Fig. 16 show that the responses
of MS2F in the different arrangements are very similar for each ground
The shear wall structure, MS2F, was designed to act as a ‘trans- motion. Similar to MS1F_2, it can be concluded that the SSSI arrange-
mitter’ structure to affect the response of the adjacent MS1F_2 by ments illustrated in Fig. 2 have little to no effect on the global structural
transmitting energy through the soil. However, the numerical responses response of MS2F in any of the ground motions considered in this study.

9
C. Bolisetti and A.S. Whittaker Engineering Structures 215 (2020) 110709

Fig. 12. Moment-rotation response of the footings of MS1F_2 in various SSSI arrangements, calculated using LS-DYNA (separation allowed) analyses for the JOS_L
ground motion.

Fig. 13. Moment rotation-response of the footings of MS1F_2 in various SSSI arrangements, calculated using LS-DYNA (separation allowed) analyses for the LCN
ground motion.

The numerical simulations of MS2F agree well with results measured in spread footings and a deep basement, is investigated numerically below
the experiments. using nonlinear SSI analyses in LS-DYNA.
The two superstructure models, MS1F_2 and MS2F, are used for
analysis. However, the basemat of MS2F is replaced by a deep basement
5. Effect of basement restraint on MS1F_2
with plan dimensions 15 m × 17 m, and is embedded to a depth of
10.1 m. This new structure is denoted as MS2F_B. The basement walls
The numerical simulations and experimental observations presented
are 0.5 m thick and are numerically modeled using shell elements and a
in the previous sections showed that the presence of the adjacent MS2F
rigid material model. No changes are made to the footings of MS1F_2.
did not result in considerable changes in the response of MS1F_2. This
The effect of an adjacent basement on the response of MS1F_2 is in-
section investigates those cases in which, SSSI can affect the global
vestigated using a pseudo-static analysis involving horizontal forces
response of MS1F_2 through pseudo-static and dynamic SSI analyses in
applied at the roof of MS1F_2, and a dynamic analysis with the four
LS-DYNA.
ground motions used previously. These analyses are performed for two
Although the restraint from the basemat of MS2F did not con-
cases: 1) a baseline case with MS1F_2 alone, and 2) an in-plane SSSI
siderably affect the footing response of MS1F_2, it is possible that a
case with MS1F_2 placed adjacent to MS2F_B (hereafter referred to as
larger neighboring foundation such as a deeply embedded basement
the iSSSI_B case). The baseline case of MS1F_2 is identical to the pre-
would provide more restraint, resulting in a more significant change in
vious sections and its LS-DYNA numerical model presented in Section 3.
the near footing response. The effects of such a restraint, between

10
C. Bolisetti and A.S. Whittaker Engineering Structures 215 (2020) 110709

Fig. 14. Moment-rotation response of the footings of MS1F_2 in various SSSI arrangements, calculated using LS-DYNA (separation allowed) analyses for the SCS_H
ground motion.

Fig. 15. Moment-rotation response of the footings of MS1F_2 in various SSSI arrangements, calculated using LS-DYNA (separation allowed) analyses for the PRI_H
ground motion.

The LS-DYNA numerical model of the iSSSI_B case is illustrated in corresponding lumped masses. The peak value of each lateral force at
Fig. 17. The foundation-soil interface (both for footings and basement) each roof node is equal to the corresponding nodal lumped mass times
is modeled to allow for separation in both the pseudo-static and dy- acceleration due to gravity. No lateral forces are applied to the MS2F
namic analyses. The response of MS2F_B is not examined. superstructure. The forces are applied in the negative X direction,
pushing the superstructure and footings of MS1F_2 towards MS2F. This
5.1. Pseudo-static analyses action provides more restraint to the footings than loading in the op-
posite direction. The foundation-soil interface in both structures allows
Pseudo-static analyses are performed to examine the effect of for separation.
basement restraint on the lateral force–displacement response of Panels a and b of Fig. 19 present the moment-rotation response and
MS1F_2. Local SSSI effects due to footing restraint are examined using the vertical force–displacement response of one of the two MS1F_2
the moment-rotation response and vertical force–displacement response footings closest to the basement. The responses are presented for the
of the restrained (near) footings. Given that dynamic effects in this baseline and iSSSI cases. The figure shows that the slopes of the mo-
analysis are negligible, a smaller soil domain is built to reduce the ment-rotation and vertical force–displacement responses are greater for
computation time. The LS-DYNA finite-element model used for the the iSSSI_B case, for which the footing is restrained by a rigid basement,
pseudo-static analysis is presented in Fig. 18. Lateral force is applied to indicating an increase in both the rocking stiffness and vertical stiffness
the roof nodes of MS1F_2 and the forces are proportional to the of the restrained footings due to SSSI. As a result, the global lateral

11
C. Bolisetti and A.S. Whittaker Engineering Structures 215 (2020) 110709

Fig. 16. Acceleration response spectra at the roof of MS2F calculated from the numerical simulations.

stiffness of MS1F_2 is greater for the iSSSI_B case, as seen in panel c of acceleration response spectra. The SSSI effects are characterized by
the same figure. comparing the response of MS1F_2 in the baseline case to that in the
iSSSI_B case. Results are presented in Fig. 20.
Fig. 20 shows significant differences in the baseline and iSSSI re-
5.2. Dynamic analyses
sponses for the LCN, SCS_H and PRI_H ground motions. The differences
are primarily in the peak spectral accelerations, which are much
The pseudo-static analyses showed an increase in the rocking and
smaller for the iSSSI_B case than in the baseline case. Additionally, the
vertical stiffness of the near footings of MS1F_2 if restrained by a deep
period at peak spectral acceleration is slightly smaller in the iSSSI_B
basement, which led to an increase in the global lateral stiffness of the
case, indicating stiffening in the response. To further investigate these
structure. This section investigates the effects of this increased stiffness
differences, the moment-rotation responses of the restrained footing are
on the dynamic response of MS1F_2. Since MS2F_B is founded on a deep
calculated for the baseline and iSSSI_B cases and presented in Fig. 21.
basement, it is assumed that the SSSI effects on this structure are neg-
Two important observations can be made from this figure: 1) the
ligible; the response of MS2F_B is therefore not examined. The dynamic
average slope of the moment-rotation curves is higher for the iSSSI_B
analyses are performed for the four ground motions, JOS_L, LCN, SCS_H
case than in the baseline case, and 2) the slope of moment-rotation
and PRI_H, from Table 1. The structural response is reported as roof

12
C. Bolisetti and A.S. Whittaker Engineering Structures 215 (2020) 110709

for gapping between the soil and the footings.

5.3. Effect of basement restraint on MS1F_1

The presence of a basement significantly affects the response of


MS1F_2 as shown above. However, MS1F_2 is founded on footings with
smaller mass, designed to intensify rocking and sliding of the footings,
unlike MS1F_1, which was designed to be more representative of low- to
mid-rise SMRF buildings (see footnote 2). This section investigates the
effect of an adjacent basement on the response of MS1F_1.
The numerical models of the baseline and iSSSI_B cases of MS1F_1
are assembled by making modifications to the numerical models of
MS1F_2. No changes are made to the superstructure or the basement of
MS2F_B. The foundation-soil interfaces of both structures are modeled
to allow separation; the ground motions, JOS_L, LCN, SCS_H and PRI_H,
are used for the simulations. Fig. 22 presents acceleration response
spectra at the roof of MS1F_1. The figure shows that the responses are
almost identical for the baseline and iSSSI_B cases, unlike MS1F_2. The
figure also shows that the period of peak spectral acceleration (the
flexible-base natural period) of MS1F_1 is around 1.2 sec, which is very
close to the fixed-base period of 1.1 sec, indicating that even SSI effects
in this structure are minor. Therefore, the foundation flexibility has
little effect on the superstructure response of MS1F_1, and the basement
restraint introduced in the iSSSI_B case does not change the response of
Fig. 17. LS-DYNA numerical model of the iSSSI_B case with MS1F_2 on the right
and MS2F_B on the left (soil elements are not shown for clarity). the structure. The minor differences in the peak spectral acceleration in
the iSSSI_B case could be due to the modification of the input motion
due to the presence of the adjacent deep basement.

6. Summary and conclusions

This paper presents a study examining the SSSI effects in two


structures through numerical simulations. Qualitative statements are
made regarding the importance of SSSI from the numerical simulations,
and similarities and differences in trends are compared with those
drawn from analysis of the centrifuge data in previous studies. The
numerical simulations include frequency-domain linear SSI simulations
in SASSI and time-domain nonlinear SSI simulations in LS-DYNA. The
buildings considered are (1) a one-story, steel moment-resisting frame
on spread footings (MS1F_2) and (2) a two-story shear wall building on
a basemat (MS2F): representative of low- and medium-rise buildings.
These buildings are placed in three different arrangements and the SSSI
effects in these arrangements are evaluated by comparing the responses
of each building in the arrangement to the corresponding baseline case,
in which the building is stand-alone with no neighboring structures. To
gain further insight into the response of MS1F_2, pseudo-static and
Fig. 18. Isometric view of the LS-DYNA numerical model used for pseudo-static
analysis of the iSSSI_B case. dynamic numerical simulations are performed with the basemat of
MS2F replaced by a deep basement. These simulations are repeated
with MS1F_2 replaced by MS1F_1, which is more representative of real
curves of the iSSSI_B case for SCS_H and PRI_H ground motions is footing structures.
mostly positive with a small segment with negative slope at large po- The key conclusions from the experimental and numerical results
sitive rotations, unlike in the baseline case where the moment-rotation examined in this paper are:
curves for the same ground motions have large segments of negative
slope and show larger peak rotations. The former is consistent with the 1. Structure-soil-structure interaction, did not result in any significant
findings of the pseudo-static analyses, which showed stiffening in the change in the global structural responses of MS1F_2 and MS2F. The
moment-rotation behavior of the restrained footing. The negative slope numerical and experimental results agree in this regard. The foot-
segment in the moment-rotation response corresponds to partial uplift ings of MS1F_2 immediately adjacent to the basemat of MS2F have
(or gapping) of the footing, which results in a sudden drop in the mo- slightly larger peak rotations than those farther away but the peak
ment with gradual increase in the rotation and therefore negative footing moments remain almost the same, which is important for the
stiffness. Therefore the moment-rotation curves of Fig. 21 indicate that design of footings in closely spaced buildings.
there is significant gapping in the baseline case and almost no gapping 2. The ‘transmitter-receiver’ effect, which was hypothesized during the
in the iSSSI case for the SCS_H and PRI_H ground motions. Furthermore, design of Tests 3 and 4, was not observed since MS1F_2 and MS2F
the negative stiffness in the baseline case results in larger spectral ac- responded almost independently of each other both in the numerical
celerations in comparison with the iSSSI case as seen in Fig. 21. The simulations and the experiments.
reduction in gapping in the iSSSI_B case is due to the increased rocking 3. Of the different numerical models, only the LS-DYNA model that
stiffness, which results in smaller footing rotations. In the restrained permits separation at the foundation-soil interface predicts roof
footings of MS1F_2 the rotations are smaller than the threshold required acceleration response spectra that are similar in shape to the

13
C. Bolisetti and A.S. Whittaker Engineering Structures 215 (2020) 110709

Fig. 19. Pseudo-static response of the of the restrained MS1F_2 footings in the baseline and iSSSI_B cases.

Fig. 20. Response spectra of the roof acceleration of MS1F_2 from numerical simulations using LS-DYNA for four ground motions.

experimental results, indicating, albeit not conclusively, that simu- foundation restraint from the basement indicates that geometric
lating foundation nonlinearities, such as footing sliding and gap- nonlinearities can play a role in SSSI.
ping, will better predict structural response for surface- or near-
surface-mounted structures subjected to intense ground motions. In summary, the study presented in this paper shows that SSSI does
4. The presence of a deep basement adjacent to MS1F_2 considerably not affect the global response of surface-founded low- to medium-rise
affects its response by reducing peak roof spectral accelerations. buildings similar to the ones considered in the City Block project.
This reduction is due to the increase in footing stiffness (both in However, SSSI does influence peak footing response, especially in the
rotation and vertical translation) as shown by pseudo-static ana- presence of a deep basement, and this is only observed when geometric
lyses. The increase in footing stiffness reduces gapping and the as- nonlinearities are accounted for in the simulation. This outcome shows
sociated negative stiffness in the moment-rotation behavior in that simulation of geometric foundation nonlinearities (gapping, sliding
comparison with the baseline case. The sensitivity of gapping to the and uplift) should be accounted for in evaluating the effects of SSSI and

Fig. 21. Moment-rotation responses of the restrained footings of MS1F_2 in the baseline and iSSSI_B cases calculated for the four ground motions.

14
C. Bolisetti and A.S. Whittaker Engineering Structures 215 (2020) 110709

Fig. 22. Response spectra of the roof acceleration of MS1F_1 calculated using numerical simulations in LS-DYNA for four ground motions.

the design of footings and columns. 2018.


Given the sensitivity of gapping, sliding and uplift response to ad- [5] Bolisetti C, Whittaker AS, Coleman JL. Linear and Nonlinear Soil-Structure
Interaction Analysis for Buildings and Safety-Related Nuclear Structures. Soil Dyn
jacent foundations, future studies should further investigate the effect Earthq Eng 2018;107C:218–33. https://doi.org/10.1016/j.soildyn.2018.01.026.
of geometric nonlinearities on SSSI. Additionally, given the influence of [6] Bray J, Kutter BL, Hutchinson TC, Whittaker AS, Fiegel GL, Reitherman R. NEESR-
the deep basement on the response of the footing structure, further SG: Seismic Performance Assessment in Dense Urban Environments. Funded
Research Proposal to the US National Science Foundation: University of California,
studies involving the presence of large underground structures (like Berkeley, California; 2008.
tunnels and deep basements) adjacent to surface-founded structures are [7] Çelebi M. Seismic Responses of Two Adjacent Buildings. I: Data and Analyses. J
warranted. Struct Eng 1993;119(8):2461–76. https://doi.org/10.1061/(ASCE)0733-
9445(1993)119:8(2461).
[8] Coleman JL, Bolisetti C, Whittaker AS. Time-Domain Soil-Structure Interaction
CRediT authorship contribution statement Analysis of Nuclear Facilities. Nucl Eng Des 2016;298:264–70.
[9] Systèmes Dassault. Computer Program: ABAQUS - Finite Element Analysis Software.
Providence, Rhode Island: Dassault Systèmes; 2018.
Chandrakanth Bolisetti: Investigation, Writing - original draft,
[10] Huang Y, Sturt R, Willford M. A Damping Model for Nonlinear Dynamic Analysis
Conceptualization, Methodology. Andrew S. Whittaker: Providing Uniform Damping over a Frequency Range. Comput Struct
Conceptualization, Supervision, Writing - review & editing, Project 2019;212:101–9. https://doi.org/10.1016/j.compstruc.2018.10.016.
administration, Funding acquisition. [11] Iwan WD. On a Class of Models for the Yielding Behavior of Continuous and
Composite Systems. J Appl Mech 1967;34:612–7.
[12] Kirkwood P, Dashti S. A Centrifuge Study of Seismic Structure-Soil-Structure
Declaration of Competing Interest Interaction on Liquefiable Ground and Implications for Design in Dense Urban
Areas. Earthquake Spectra 2018. https://doi.org/10.1193/052417eqs095m.
[13] Kramer SL. Geotechnical Earthquake Engineering. Upper Saddle River, New Jersey:
The authors declare that they have no known competing financial Prentice Hall; 1996.
interests or personal relationships that could have appeared to influ- [14] Lin HT, Roesset JM, Tassoulas JL. Dynamic Interaction Between Adjacent
Foundations. Earthq Eng Struct Dyn 1987;15:323–43.
ence the work reported in this paper. [15] Livermore Software Technology Corporation (LSTC). LS-DYNA Keyword User's
Manual - Version R 7.0. California: Livermore; 2013.
Acknowledgments [16] Luco JE, Contesse L. Dynamic Structure-Soil-Structure Interaction. Bull Seismol Soc
Am 1973;63(4):1289–303.
[17] Lysmer J, Ostadan F, Chin C. Computer Program: SASSI2000 - A System for Analysis
This study was funded by the US National Science Foundation (NSF) of Soil-Structure Interaction. Berkeley, California: University of California; 1999.
under Grant No. CMMI-0830331, MCEER at the University at Buffalo, [18] Mason HB, Kutter BL, Bray JD, Wilson DW, Choy BY. Earthquake Motion Selection
and Calibration for Use in a Geotechnical Centrifuge. Proceedings: 7th International
State University of New York, and the US Department of Energy. The
Conference on Physical Modelling in Geotechnics, 28th June to 1st Campus Science
authors gratefully acknowledge this financial support. The opinions, City (Hönggerberg) July 2010. Zurich, Switzerland: ETH; 2010.
findings, conclusions and recommendations expressed in this paper are [19] Mason HB, Trombetta NW, Gille S, Lund J, Zupan J, Jones KC, et al. Seismic
those of the authors alone. The authors thank the NEES City Block team, Performance Assessment in Dense Urban Environments: Centrifuge Data Report for
HBM02 (Test 1). Davis, California: University of California at Davis Center for
especially, Dr. Nicholas Trombetta, Prof. Ben Mason, Prof. Zhiqiang Geotechnical Modeling; 2010.
Chen, Prof. Jonathan Bray, Prof. Tara Hutchinson and Prof. Bruce [20] Mason HB, Trombetta NW, Gille S, Lund J, Zupan J, Jones KC, et al. Seismic
Kutter, for their contributions to the centrifuge experiments, Dr. Performance Assessment in Dense Urban Environments: Centrifuge Data Report for
HBM03 (Test 2). Davis, California: University of California at Davis Center for
Farhang Ostadan for providing guidance in performing SSI analysis in Geotechnical Modeling; 2010.
SASSI, and Ibrahim Almufti, Michael Willford and Dr. Yuli Huang at [21] Mason HB. Seismic Performance Assessment in Dense Urban Environments. Ph.D.
Arup, San Francisco for providing insight into performing SSI analysis Dissertation, University of California, Berkeley, California; 2011.
[22] Mason HB, Trombetta NW, Gille S, Lund J, Zupan J, Jones KC, et al. Seismic
in LS-DYNA. Performance Assessment in Dense Urban Environments: Centrifuge Data Report for
HBM04 (Test 3). Davis, California: University of California at Davis Center for
References Geotechnical Modeling; 2011.
[23] Mason HB, Trombetta NW, Chen Z, Bray JD, Hutchinson TC, Kutter BL. Seismic
Soil–Foundation–Structure Interaction Observed in Geotechnical Centrifuge
[1] Anderson LM, Carey S, Amin J, Ostadan F. Effect of Separation Distance and Soil Experiments. Soil Dyn Earthq Eng 2013;48:162–74. https://doi.org/10.1016/j.
Parameters on the Structure-Soil-Structure Interaction Response of Adjacent Deeply soildyn.2013.01.014.
Embedded Structures. Transactions: 22nd International Conference in Structural [24] Matthees W, Magiera G. A Sensitivity Study of Seismic Structure-Soil-Structure
Mechanics in Reactor Technology (SMiRT-22), San Francisco, California. 2013. Interaction Problems for Nuclear Power Plants. Nucl Eng Des 1982;73:343–63.
[2] American Society of Civil Engineers (ASCE). (2016). Minimum Design Loads for [25] Menglin L, Huaifeng W, Xi C, Yongmei Z. Structure-Soil-Structure Interaction:
Buildings and Other Structures. ASCE/SEI 7-16, Reston, Virginia. Literature Review. Soil Dyn Earthq Eng 2011;31:1724–31.
[3] Bolisetti C, Whittaker AS. Seismic Structure-Soil-Structure Interaction in Nuclear [26] Murakami H, Luco JE. Seismic Response of a Periodic Array of Structures. J Eng
Power Plant Structures. Transactions: 21st International Conference in Structural Mech Divis, ASCE SM5 1977;103:965–76.
Mechanics in Reactor Technology (SMiRT-21), New Delhi, India; 2011. [27] Padron LA, Aznarez JJ, Maeso O. Dynamic Structure-Soil-Structure Interaction
[4] Bolisetti C, Whittaker AS. Site Response, Soil-Structure Interaction and Structure- between Nearby Piled Buildings under Seismic Excitation by BEM-FEM Model. Soil
Soil-Structure Interaction for Performance Assessment of Buildings and Nuclear Dyn Earthq Eng 2009;29:1084–96.
Structures. MCEER-15-0002, Multidisciplinary Center for Earthquake Engineering [28] Qian J, Beskos DE. Harmonic Wave Response of Two 3-D Rigid Surface
Research, University at Buffalo, The State University of New York, Buffalo, NY

15
C. Bolisetti and A.S. Whittaker Engineering Structures 215 (2020) 110709

Foundations. Soil Dyn Earthq Eng 1996;15:95–110. [33] Trombetta NW. Seismic Soil-Foundation Structure Interaction in Urban
[29] Schnabel PB, Lysmer J, Seed HB. Computer Program: SHAKE: A Computer Program Environments. Ph.D. Dissertation, University of California, San Diego, California;
for Earthquake Response Analysis of Horizontally Layered Sites. Berkeley, 2013.
California: University of California; 2012. [34] Trombetta NW, Mason HB, Hutchinson TC, Zupan JD, Bray JD, Kutter BL. Nonlinear
[30] Seed HB, Idriss M. Soil Moduli and Damping Factors for Dynamic Response Soil-Foundation-Structure and Structure-Soil-Structure Interaction: Centrifuge Test
Analysis. Earthquake Engineering Research Center: University of California, Observations. J Geotech Geoenviron Eng 2014;140(5):04013057. https://doi.org/
Berkeley, California; 1970. 10.1061/(ASCE)GT.1943-5606.0001074.
[31] Trombetta NW, Zupan J, Bolisetti C, Puangnak H, Jones KC, Tran JK, et al. Seismic [35] Trombetta NW, Mason HB, Hutchinson TC, Zupan JD, Bray JD, Kutter BL. Nonlinear
Performance Assessment in Dense Urban Environments: Centrifuge Data Report for Soil-Foundation-Structure and Structure-Soil-Structure Interaction: Engineering
NWT01 (Test 4). Davis, California: University of California at Davis Center for Demands. J Struct Eng 2015;141(7):04014177. https://doi.org/10.1061/(ASCE)ST.
Geotechnical Modeling; 2011. 1943-541X.0001127.
[32] Trombetta NW, Zupan J, Bolisetti C, Puangnak H, Jones KC, Tran JK, et al. Seismic [36] Wong HL, Trifunac MD. Two-Dimensional, Antiplane, Building-Soil-Building
Performance Assessment in Dense Urban Environments: Centrifuge data report for Interaction for Two or More Buildings and for Incident Plane SH Waves. Bull
NWT01. Davis, California: University of California at Davis Center for Geotechnical Seismol Soc Am 1975;65(6):1863–85.
Modeling; 2011.

16

You might also like